From d215f014b3526a9898d87bfb4b866287f0864cc9 Mon Sep 17 00:00:00 2001 From: Claudio Imbrenda Date: Fri, 28 Aug 2026 13:54:32 +0200 Subject: [PATCH 001/365] KVM: s390: Fix dirty marking in adapter_indicators_set*() When the indicator and/or summary bits are set in the guest, the accessed page was only marked dirty in KVM if the access was performed using the slow path; accesses through the new kvm_arch_set_irq_inatomic fast inject path would not mark the page as dirty. Fix by adding/moving the missing calls to mark_page_dirty(). Note that for the inatomic path set_page_dirty{,_lock}() is not needed as the page stays pinned; the unpin path correctly marks it as dirty. Opportunistically reorder the local variables to be in reverse Christmas tree order and refactor to use guard(). Fixes: 1e95e3bc6b05 ("KVM: s390: Enable adapter_indicators_set to use mapped pages") Signed-off-by: Claudio Imbrenda Message-ID: <20260828115439.145885-2-imbrenda@linux.ibm.com> --- arch/s390/kvm/s390/interrupt.c | 77 +++++++++++++++++----------------- 1 file changed, 39 insertions(+), 38 deletions(-) diff --git a/arch/s390/kvm/s390/interrupt.c b/arch/s390/kvm/s390/interrupt.c index 0381ae981703..49b4e233e791 100644 --- a/arch/s390/kvm/s390/interrupt.c +++ b/arch/s390/kvm/s390/interrupt.c @@ -2984,61 +2984,58 @@ static int adapter_indicators_set(struct kvm *kvm, struct s390_io_adapter *adapter, struct kvm_s390_adapter_int *adapter_int) { - unsigned long bit; - int summary_set, idx; struct s390_map_info *ind_info, *summary_info; - void *map; struct page *ind_page, *summary_page; - unsigned long flags; + unsigned long bit; + int summary_set; + void *map; ind_page = NULL; - spin_lock_irqsave(&adapter->maps_lock, flags); - ind_info = get_map_info(adapter, adapter_int->ind_addr); + scoped_guard(spinlock_irqsave, &adapter->maps_lock) { + ind_info = get_map_info(adapter, adapter_int->ind_addr); + if (ind_info) { + map = page_address(ind_info->page); + bit = get_ind_bit(ind_info->addr, adapter_int->ind_offset, adapter->swap); + set_bit(bit, map); + } + } if (!ind_info) { - spin_unlock_irqrestore(&adapter->maps_lock, flags); ind_page = pin_map_page(kvm, adapter_int->ind_addr, 0); if (!ind_page) return -1; - idx = srcu_read_lock(&kvm->srcu); map = page_address(ind_page); bit = get_ind_bit(adapter_int->ind_addr, adapter_int->ind_offset, adapter->swap); set_bit(bit, map); - mark_page_dirty(kvm, adapter_int->ind_gaddr >> PAGE_SHIFT); set_page_dirty_lock(ind_page); - srcu_read_unlock(&kvm->srcu, idx); unpin_user_page(ind_page); - } else { - map = page_address(ind_info->page); - bit = get_ind_bit(ind_info->addr, adapter_int->ind_offset, adapter->swap); - set_bit(bit, map); - spin_unlock_irqrestore(&adapter->maps_lock, flags); } + scoped_guard(srcu, &kvm->srcu) + mark_page_dirty(kvm, gpa_to_gfn(adapter_int->ind_gaddr)); - spin_lock_irqsave(&adapter->maps_lock, flags); - summary_info = get_map_info(adapter, adapter_int->summary_addr); + scoped_guard(spinlock_irqsave, &adapter->maps_lock) { + summary_info = get_map_info(adapter, adapter_int->summary_addr); + if (summary_info) { + map = page_address(summary_info->page); + bit = get_ind_bit(summary_info->addr, adapter_int->summary_offset, + adapter->swap); + summary_set = test_and_set_bit(bit, map); + } + } if (!summary_info) { - spin_unlock_irqrestore(&adapter->maps_lock, flags); summary_page = pin_map_page(kvm, adapter_int->summary_addr, 0); if (!summary_page) return -1; - idx = srcu_read_lock(&kvm->srcu); map = page_address(summary_page); bit = get_ind_bit(adapter_int->summary_addr, adapter_int->summary_offset, adapter->swap); summary_set = test_and_set_bit(bit, map); - mark_page_dirty(kvm, adapter_int->summary_gaddr >> PAGE_SHIFT); set_page_dirty_lock(summary_page); - srcu_read_unlock(&kvm->srcu, idx); unpin_user_page(summary_page); - } else { - map = page_address(summary_info->page); - bit = get_ind_bit(summary_info->addr, adapter_int->summary_offset, - adapter->swap); - summary_set = test_and_set_bit(bit, map); - spin_unlock_irqrestore(&adapter->maps_lock, flags); } + scoped_guard(srcu, &kvm->srcu) + mark_page_dirty(kvm, gpa_to_gfn(adapter_int->summary_gaddr)); return summary_set ? 0 : 1; } @@ -3048,26 +3045,29 @@ static int adapter_indicators_set_fast(struct kvm *kvm, struct kvm_s390_adapter_int *adapter_int, int setbit) { + struct s390_map_info *ind_info, *summary_info; unsigned long bit; int summary_set; - struct s390_map_info *ind_info, *summary_info; void *map; - spin_lock(&adapter->maps_lock); + guard(srcu)(&kvm->srcu); + guard(spinlock)(&adapter->maps_lock); + ind_info = get_map_info(adapter, adapter_int->ind_addr); - if (!ind_info) { - spin_unlock(&adapter->maps_lock); + if (!ind_info) return -EWOULDBLOCK; - } + map = page_address(ind_info->page); bit = get_ind_bit(ind_info->addr, adapter_int->ind_offset, adapter->swap); - if (setbit) + if (setbit) { set_bit(bit, map); - summary_info = get_map_info(adapter, adapter_int->summary_addr); - if (!summary_info) { - spin_unlock(&adapter->maps_lock); - return -EWOULDBLOCK; + mark_page_dirty(kvm, gpa_to_gfn(adapter_int->ind_gaddr)); } + + summary_info = get_map_info(adapter, adapter_int->summary_addr); + if (!summary_info) + return -EWOULDBLOCK; + map = page_address(summary_info->page); bit = get_ind_bit(summary_info->addr, adapter_int->summary_offset, adapter->swap); @@ -3077,7 +3077,8 @@ static int adapter_indicators_set_fast(struct kvm *kvm, summary_set = test_and_set_bit(bit, map); else summary_set = test_and_clear_bit(bit, map); - spin_unlock(&adapter->maps_lock); + mark_page_dirty(kvm, gpa_to_gfn(adapter_int->summary_gaddr)); + return summary_set ? 0 : 1; } From ae12d2f9c119639a142d92c4a37c30277e8576da Mon Sep 17 00:00:00 2001 From: Claudio Imbrenda Date: Fri, 28 Aug 2026 13:54:33 +0200 Subject: [PATCH 002/365] KVM: s390: Fix compile warning for kvm_s390_update_cmma_dirty() The parameter "old" should be marked as const, to prevent compile-time warnings. Fixes: d487a24041c2 ("KVM: s390: Prepare gmap for a second KVM implementation") Signed-off-by: Claudio Imbrenda Message-ID: <20260828115439.145885-3-imbrenda@linux.ibm.com> --- arch/s390/kvm/s390/s390.c | 2 +- arch/s390/kvm/s390/s390.h | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/arch/s390/kvm/s390/s390.c b/arch/s390/kvm/s390/s390.c index b0839e887221..8f7e09d7d049 100644 --- a/arch/s390/kvm/s390/s390.c +++ b/arch/s390/kvm/s390/s390.c @@ -5766,7 +5766,7 @@ static long cmma_d_count_pte(union pte *ptep, gfn_t gfn, gfn_t next, struct dat_ return 0; } -void kvm_s390_update_cmma_dirty(struct kvm *kvm, struct kvm_memory_slot *old) +void kvm_s390_update_cmma_dirty(struct kvm *kvm, const struct kvm_memory_slot *old) { const struct dat_walk_ops ops = { .pte_entry = cmma_d_count_pte, }; diff --git a/arch/s390/kvm/s390/s390.h b/arch/s390/kvm/s390/s390.h index d284a263ba70..aa0d1d062f8d 100644 --- a/arch/s390/kvm/s390/s390.h +++ b/arch/s390/kvm/s390/s390.h @@ -472,7 +472,7 @@ int __kvm_s390_mprotect_many(struct gmap *gmap, gpa_t gpa, u8 npages, unsigned i unsigned long bits); bool kvm_arch_setup_async_pf(struct kvm_vcpu *vcpu); -void kvm_s390_update_cmma_dirty(struct kvm *kvm, struct kvm_memory_slot *old); +void kvm_s390_update_cmma_dirty(struct kvm *kvm, const struct kvm_memory_slot *old); int kvm_s390_vm_stop_migration(struct kvm *kvm); From faff4c8ff3dbec6d71b89dd281a0d17c4478fa44 Mon Sep 17 00:00:00 2001 From: Claudio Imbrenda Date: Fri, 28 Aug 2026 13:54:34 +0200 Subject: [PATCH 003/365] KVM: s390: Fix _gaccess_shadow_fault() In some circumstances, it is possible that the page of nested guest memory that is being shadowed is not present at all in the parent guest gmap. dat_entry_walk() will not find any leaf entry and return with -ENOENT, which will erroneously be propagated all the way to userspace. Fix by manually calling gmap_link() on the memory of the nested guest that is being shadowed if the mapping was not already present. Fixes: e38c884df921 ("KVM: s390: Switch to new gmap") Signed-off-by: Claudio Imbrenda Message-ID: <20260828115439.145885-4-imbrenda@linux.ibm.com> --- arch/s390/kvm/s390/gaccess.c | 13 +++++++++++++ 1 file changed, 13 insertions(+) diff --git a/arch/s390/kvm/s390/gaccess.c b/arch/s390/kvm/s390/gaccess.c index e5c064f263df..405345ccc4f1 100644 --- a/arch/s390/kvm/s390/gaccess.c +++ b/arch/s390/kvm/s390/gaccess.c @@ -1589,12 +1589,25 @@ static inline int ___gaccess_shadow_fault(struct kvm_vcpu *vcpu, struct gmap *sg parent = READ_ONCE(sg->parent); if (!parent) return -EAGAIN; +retry: scoped_guard(spinlock, &parent->children_lock) { if (READ_ONCE(sg->parent) != parent) return -EAGAIN; sg->invalidated = false; rc = _gaccess_do_shadow(vcpu->arch.mc, sg, saddr, walk); } + if (rc == -ENOENT) { + struct kvm_memory_slot *slot; + struct guest_fault *entries; + + entries = get_entries(walk); + slot = kvm_vcpu_gfn_to_memslot(vcpu, entries[LEVEL_MEM].gfn); + if (!slot) + return PGM_ADDRESSING; + rc = gmap_link(vcpu->arch.mc, parent, entries + LEVEL_MEM, slot); + if (!rc) + goto retry; + } if (!rc) kvm_s390_release_faultin_array(vcpu->kvm, walk->raw_entries, false); return rc; From 00c0ae5e438615bde828a3b9f33912960f4e6511 Mon Sep 17 00:00:00 2001 From: Claudio Imbrenda Date: Fri, 28 Aug 2026 13:54:35 +0200 Subject: [PATCH 004/365] KVM: s390: Refactor dat_set_slot() Refactor dat_set_slot(), _dat_slot_pte(), _dat_slot_crste(). Now they only take a struct kvm_s390_mmu_cache as priv. For dat_delete_slot(), mc is NULL, as no allocations should take place. This is needed as a prerequisite to move gmap DAT table setup from kvm_arch_commit_memory_region() to kvm_arch_prepare_memory_region(). Signed-off-by: Claudio Imbrenda Message-ID: <20260828115439.145885-5-imbrenda@linux.ibm.com> --- arch/s390/kvm/gmap/dat.c | 29 +++++++++-------------------- arch/s390/kvm/gmap/dat.h | 10 ++++------ arch/s390/kvm/gmap/kvm_mmu.c | 4 ++-- 3 files changed, 15 insertions(+), 28 deletions(-) diff --git a/arch/s390/kvm/gmap/dat.c b/arch/s390/kvm/gmap/dat.c index 24547e39fab2..dcedd5479d82 100644 --- a/arch/s390/kvm/gmap/dat.c +++ b/arch/s390/kvm/gmap/dat.c @@ -846,19 +846,12 @@ long dat_reset_skeys(union asce asce, gfn_t start) } #endif /* KVM_S390_MANAGES_S390_GUEST */ -struct slot_priv { - unsigned long token; - struct kvm_s390_mmu_cache *mc; -}; - static long _dat_slot_pte(union pte *ptep, gfn_t gfn, gfn_t next, struct dat_walk *walk) { - struct slot_priv *p = walk->priv; - union crste dummy = { .val = p->token }; union pte new_pte, pte = READ_ONCE(*ptep); union pgste pgste; - new_pte = _PTE_TOK(dummy.tok.type, dummy.tok.par); + new_pte = walk->priv ? _PTE_EMPTY : _PTE_TOK(_DAT_TOKEN_PIC, PGM_ADDRESSING); /* Table entry already in the desired state. */ if (pte.val == new_pte.val) @@ -875,10 +868,9 @@ static long _dat_slot_pte(union pte *ptep, gfn_t gfn, gfn_t next, struct dat_wal static long _dat_slot_crste(union crste *crstep, gfn_t gfn, gfn_t next, struct dat_walk *walk) { union crste new_crste, crste = READ_ONCE(*crstep); - struct slot_priv *p = walk->priv; + struct kvm_s390_mmu_cache *mc = walk->priv; - new_crste.val = p->token; - new_crste.h.tt = crste.h.tt; + new_crste = mc ? _CRSTE_EMPTY(crste.h.tt) : _CRSTE_HOLE(crste.h.tt); /* Table entry already in the desired state. */ if (crste.val == new_crste.val) @@ -902,7 +894,10 @@ static long _dat_slot_crste(union crste *crstep, gfn_t gfn, gfn_t next, struct d if (!crste.h.fc && !crste.h.i) return 0; /* Split (install a lower level table), and handle things there. */ - return dat_split_crste(p->mc, crstep, gfn, walk->asce, false); + if (mc) + return dat_split_crste(mc, crstep, gfn, walk->asce, false); + /* A large page should never cross memslots boundaries */ + return -EINVAL; } static const struct dat_walk_ops dat_slot_ops = { @@ -910,16 +905,10 @@ static const struct dat_walk_ops dat_slot_ops = { .crste_ops = { _dat_slot_crste, _dat_slot_crste, _dat_slot_crste, _dat_slot_crste, }, }; -int dat_set_slot(struct kvm_s390_mmu_cache *mc, union asce asce, gfn_t start, gfn_t end, - u16 type, u16 param) +int dat_set_slot(struct kvm_s390_mmu_cache *mc, union asce asce, gfn_t start, gfn_t end) { - struct slot_priv priv = { - .token = _CRSTE_TOK(0, type, param).val, - .mc = mc, - }; - return _dat_walk_gfn_range(start, end, asce, &dat_slot_ops, - DAT_WALK_IGN_HOLES | DAT_WALK_ANY, &priv); + DAT_WALK_IGN_HOLES | DAT_WALK_ANY, mc); } static void pgste_set_unlock_multiple(union pte *first, int n, union pgste *pgstes) diff --git a/arch/s390/kvm/gmap/dat.h b/arch/s390/kvm/gmap/dat.h index e452c141b841..90389d47ba4e 100644 --- a/arch/s390/kvm/gmap/dat.h +++ b/arch/s390/kvm/gmap/dat.h @@ -547,8 +547,7 @@ long dat_reset_skeys(union asce asce, gfn_t start); unsigned long dat_get_ptval(struct page_table *table, struct ptval_param param); void dat_set_ptval(struct page_table *table, struct ptval_param param, unsigned long val); -int dat_set_slot(struct kvm_s390_mmu_cache *mc, union asce asce, gfn_t start, gfn_t end, - u16 type, u16 param); +int dat_set_slot(struct kvm_s390_mmu_cache *mc, union asce asce, gfn_t start, gfn_t end); #if KVM_S390_MANAGES_S390_GUEST int dat_set_prefix_notif_bit(union asce asce, gfn_t gfn); @@ -973,16 +972,15 @@ static inline int get_level(union crste *crstep, union pte *ptep) return ptep ? TABLE_TYPE_PAGE_TABLE : crstep->h.tt; } -static inline int dat_delete_slot(struct kvm_s390_mmu_cache *mc, union asce asce, gfn_t start, - unsigned long npages) +static inline int dat_delete_slot(union asce asce, gfn_t start, unsigned long npages) { - return dat_set_slot(mc, asce, start, start + npages, _DAT_TOKEN_PIC, PGM_ADDRESSING); + return dat_set_slot(NULL, asce, start, start + npages); } static inline int dat_create_slot(struct kvm_s390_mmu_cache *mc, union asce asce, gfn_t start, unsigned long npages) { - return dat_set_slot(mc, asce, start, start + npages, _DAT_TOKEN_NONE, 0); + return dat_set_slot(mc, asce, start, start + npages); } static inline bool crste_is_ucas(union crste crste) diff --git a/arch/s390/kvm/gmap/kvm_mmu.c b/arch/s390/kvm/gmap/kvm_mmu.c index b08b8229bb6f..4c8054e18490 100644 --- a/arch/s390/kvm/gmap/kvm_mmu.c +++ b/arch/s390/kvm/gmap/kvm_mmu.c @@ -111,10 +111,10 @@ void s390_kvm_mmu_commit_memory_region(struct kvm *kvm, kvm_s390_update_cmma_dirty(kvm, old); switch (change) { case KVM_MR_DELETE: - rc = dat_delete_slot(mc, kvm->arch.gmap->asce, old->base_gfn, old->npages); + rc = dat_delete_slot(kvm->arch.gmap->asce, old->base_gfn, old->npages); break; case KVM_MR_MOVE: - rc = dat_delete_slot(mc, kvm->arch.gmap->asce, old->base_gfn, old->npages); + rc = dat_delete_slot(kvm->arch.gmap->asce, old->base_gfn, old->npages); if (rc) break; fallthrough; From 19192a4043277af380f83d550a55b92eeeaac7c6 Mon Sep 17 00:00:00 2001 From: Claudio Imbrenda Date: Fri, 28 Aug 2026 13:54:36 +0200 Subject: [PATCH 005/365] KVM: s390: Move all code into s390_kvm_mmu_prepare_memory_region() Move all code from s390_kvm_mmu_commit_memory_region() into s390_kvm_mmu_prepare_memory_region(). This allows the function to fail gracefully if needed. The previous behaviour was to print a warning and continue execution with page tables inconsistent with the memslots. Fixes: e38c884df921 ("KVM: s390: Switch to new gmap") Signed-off-by: Claudio Imbrenda Message-ID: <20260828115439.145885-6-imbrenda@linux.ibm.com> --- arch/s390/kvm/gmap/kvm_mmu.c | 89 +++++++++++++++--------------------- arch/s390/kvm/gmap/kvm_mmu.h | 4 -- arch/s390/kvm/s390/s390.c | 1 - 3 files changed, 36 insertions(+), 58 deletions(-) diff --git a/arch/s390/kvm/gmap/kvm_mmu.c b/arch/s390/kvm/gmap/kvm_mmu.c index 4c8054e18490..c2ffb5e59ec6 100644 --- a/arch/s390/kvm/gmap/kvm_mmu.c +++ b/arch/s390/kvm/gmap/kvm_mmu.c @@ -47,6 +47,9 @@ int s390_kvm_mmu_prepare_memory_region(struct kvm *kvm, struct kvm_memory_slot *new, enum kvm_mr_change change) { + struct kvm_s390_mmu_cache *mc __free(kvm_s390_mmu_cache) = NULL; + int rc = 0; + if (kvm_is_ucontrol(kvm) && new && new->id < KVM_USER_MEM_SLOTS) return -EINVAL; @@ -61,6 +64,10 @@ int s390_kvm_mmu_prepare_memory_region(struct kvm *kvm, * and munmap() stuff in this slot after doing this call at any * time. */ + if (change != KVM_MR_MOVE && change != KVM_MR_CREATE) { + WARN(1, "Unknown KVM MR CHANGE: %d\n", change); + return -EINVAL; + } if (new->userspace_addr & ~PAGE_MASK) return -EINVAL; if ((new->base_gfn + new->npages) * PAGE_SIZE > kvm->arch.mem_limit) @@ -69,65 +76,41 @@ int s390_kvm_mmu_prepare_memory_region(struct kvm *kvm, return -EINVAL; } - if (!kvm_s390_is_migration_mode(kvm)) - return 0; - - /* - * Turn off migration mode when: - * - userspace creates a new memslot with dirty logging off, - * - userspace modifies an existing memslot (MOVE or FLAGS_ONLY) and - * dirty logging is turned off. - * Migration mode expects dirty page logging being enabled to store - * its dirty bitmap. - */ - if (change != KVM_MR_DELETE && - !(new->flags & KVM_MEM_LOG_DIRTY_PAGES)) - WARN(kvm_s390_vm_stop_migration(kvm), - "Failed to stop migration mode"); - - return 0; -} - -void s390_kvm_mmu_commit_memory_region(struct kvm *kvm, - struct kvm_memory_slot *old, - const struct kvm_memory_slot *new, - enum kvm_mr_change change) -{ - struct kvm_s390_mmu_cache *mc __free(kvm_s390_mmu_cache) = NULL; - int rc = 0; - - guard(mutex)(&kvm->slots_arch_lock); + if (kvm->arch.migration_mode) { + /* + * Turn off migration mode when: + * - userspace creates a new memslot with dirty logging off, + * - userspace modifies an existing memslot (MOVE or FLAGS_ONLY) + * and dirty logging is turned off. + * Migration mode expects dirty page logging being enabled to + * store its dirty bitmap. + */ + if (change != KVM_MR_DELETE && + !(new->flags & KVM_MEM_LOG_DIRTY_PAGES)) + WARN(kvm_s390_vm_stop_migration(kvm), + "Failed to stop migration mode"); + } if (change == KVM_MR_FLAGS_ONLY) - return; - - mc = kvm_s390_new_mmu_cache(); - if (!mc) { - rc = -ENOMEM; - goto out; + return 0; + if (change != KVM_MR_DELETE) { + /* Enough capacity to add a new memslot */ + mc = kvm_s390_new_mmu_cache(); + if (!mc) + return -ENOMEM; } - scoped_guard(write_lock, &kvm->mmu_lock) { kvm_s390_update_cmma_dirty(kvm, old); - switch (change) { - case KVM_MR_DELETE: + if (change == KVM_MR_DELETE || change == KVM_MR_MOVE) rc = dat_delete_slot(kvm->arch.gmap->asce, old->base_gfn, old->npages); - break; - case KVM_MR_MOVE: - rc = dat_delete_slot(kvm->arch.gmap->asce, old->base_gfn, old->npages); - if (rc) - break; - fallthrough; - case KVM_MR_CREATE: + if (!rc && (change == KVM_MR_MOVE || change == KVM_MR_CREATE)) rc = dat_create_slot(mc, kvm->arch.gmap->asce, new->base_gfn, new->npages); - break; - case KVM_MR_FLAGS_ONLY: - break; - default: - WARN(1, "Unknown KVM MR CHANGE: %d\n", change); - } } -out: - if (rc) - pr_warn("failed to commit memory region\n"); + /* + * Can only be triggered if dat_{create,delete}_slot() found an + * internal inconsistency or if the mmu cache ran out of memory; + * both should be impossible. + */ + KVM_BUG_ON(rc, kvm); + return rc; } diff --git a/arch/s390/kvm/gmap/kvm_mmu.h b/arch/s390/kvm/gmap/kvm_mmu.h index cdbd390bd33c..43cde61bae03 100644 --- a/arch/s390/kvm/gmap/kvm_mmu.h +++ b/arch/s390/kvm/gmap/kvm_mmu.h @@ -10,9 +10,5 @@ int s390_kvm_mmu_prepare_memory_region(struct kvm *kvm, const struct kvm_memory_slot *old, struct kvm_memory_slot *new, enum kvm_mr_change change); -void s390_kvm_mmu_commit_memory_region(struct kvm *kvm, - struct kvm_memory_slot *old, - const struct kvm_memory_slot *new, - enum kvm_mr_change change); #endif /* ARCH_KVM_GMAP_KVM_MMU_H */ diff --git a/arch/s390/kvm/s390/s390.c b/arch/s390/kvm/s390/s390.c index 8f7e09d7d049..eca4a4359ab2 100644 --- a/arch/s390/kvm/s390/s390.c +++ b/arch/s390/kvm/s390/s390.c @@ -5781,7 +5781,6 @@ void kvm_arch_commit_memory_region(struct kvm *kvm, struct kvm_memory_slot *old, const struct kvm_memory_slot *new, enum kvm_mr_change change) { - s390_kvm_mmu_commit_memory_region(kvm, old, new, change); } /** From f3a557067d57ce6ae98d485c8009b223e16f5f36 Mon Sep 17 00:00:00 2001 From: Claudio Imbrenda Date: Fri, 28 Aug 2026 13:54:37 +0200 Subject: [PATCH 006/365] KVM: s390: Add missing srcu in kvm_s390_set_irq_state() Like kvm_s390_inject_vcpu(), kvm_s390_set_irq_state() also needs the kvm->srcu or the slots lock when performing the Store status operation. Fix by taking kvm->srcu in kvm_s390_set_irq_state(). Fixes: ba5c1e9b6cee ("KVM: s390: interrupt subsystem, cpu timer, waitpsw") Fixes: 062e44a9319f ("KVM: s390: Use srcu in kvm_arch_vcpu_unlocked_ioctl()") Signed-off-by: Claudio Imbrenda Message-ID: <20260828115439.145885-7-imbrenda@linux.ibm.com> --- arch/s390/kvm/s390/interrupt.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/arch/s390/kvm/s390/interrupt.c b/arch/s390/kvm/s390/interrupt.c index 49b4e233e791..f892f4307883 100644 --- a/arch/s390/kvm/s390/interrupt.c +++ b/arch/s390/kvm/s390/interrupt.c @@ -3229,9 +3229,9 @@ int kvm_s390_set_irq_state(struct kvm_vcpu *vcpu, void __user *irqstate, int len break; } } - if (storestatus) { - n = kvm_s390_store_status_unloaded(vcpu, KVM_S390_STORE_STATUS_NOADDR); + scoped_guard(srcu, &vcpu->kvm->srcu) + n = kvm_s390_store_status_unloaded(vcpu, KVM_S390_STORE_STATUS_NOADDR); return r ? r : n; } From 27554b9505ddfc0aeab466aeb60929dfa17284c7 Mon Sep 17 00:00:00 2001 From: Claudio Imbrenda Date: Fri, 28 Aug 2026 13:54:38 +0200 Subject: [PATCH 007/365] KVM: s390: Fix potential races in dat skey functions When dat_cond_set_storage_key() finds a large page, it will conditionally set the storage key in absolute memory using large_crste_to_phys() to get the absolute address. There is a race window between dat_entry_walk() and large_crste_to_phys(): the large page could have been split concurrently, and large_crste_to_phys() might be called with a crste that does not designate a large page, leading to crashes. Similar issues were also present in dat_set_storage_key(). dat_get_storage_key() and dat_reset_reference_bit() did instead check for a potential concurrent splitting of the large page, but then handled it incorrectly. Fix by performing a READ_ONCE on the crste pointer, checking and using the result, instead of dereferencing the pointer again. In case a race is detacted, try dat_entry_walk() again. Fixes: 8e03e8316eb2 ("KVM: s390: KVM page table management functions: storage keys") Signed-off-by: Claudio Imbrenda Message-ID: <20260828115439.145885-8-imbrenda@linux.ibm.com> --- arch/s390/kvm/gmap/dat.c | 36 ++++++++++++++++++++++++++++-------- 1 file changed, 28 insertions(+), 8 deletions(-) diff --git a/arch/s390/kvm/gmap/dat.c b/arch/s390/kvm/gmap/dat.c index dcedd5479d82..ff80d02c9f56 100644 --- a/arch/s390/kvm/gmap/dat.c +++ b/arch/s390/kvm/gmap/dat.c @@ -621,17 +621,20 @@ int dat_get_storage_key(union asce asce, gfn_t gfn, union skey *skey) union pte *ptep; int rc; +again: skey->skey = 0; rc = dat_entry_walk(NULL, gfn, asce, DAT_WALK_ANY, TABLE_TYPE_PAGE_TABLE, &crstep, &ptep); if (rc) return rc; if (!ptep) { - union crste crste; + union crste crste = READ_ONCE(*crstep); - crste = READ_ONCE(*crstep); - if (!crste.h.fc || !crste.s.fc1.pr) + if (!crste_leaf(crste) && !crste.h.i) + goto again; + if (!crste.s.fc1.pr) return 0; + skey->skey = page_get_storage_key(large_crste_to_phys(crste, gfn)); return 0; } @@ -662,13 +665,20 @@ int dat_set_storage_key(struct kvm_s390_mmu_cache *mc, union asce asce, gfn_t gf union pte *ptep; int rc; +again: rc = dat_entry_walk(mc, gfn, asce, DAT_WALK_LEAF_ALLOC, TABLE_TYPE_PAGE_TABLE, &crstep, &ptep); if (rc) return rc; if (!ptep) { - page_set_storage_key(large_crste_to_phys(*crstep, gfn), skey.skey, !nq); + union crste crste = READ_ONCE(*crstep); + + /* A large page has been split concurrently, try again */ + if (!crste_leaf(crste)) + goto again; + + page_set_storage_key(large_crste_to_phys(crste, gfn), skey.skey, !nq); return 0; } @@ -718,15 +728,22 @@ int dat_cond_set_storage_key(struct kvm_s390_mmu_cache *mmc, union asce asce, gf union pte *ptep; int rc; +again: rc = dat_entry_walk(mmc, gfn, asce, DAT_WALK_LEAF_ALLOC, TABLE_TYPE_PAGE_TABLE, &crstep, &ptep); if (rc) return rc; if (!ptep) { + union crste crste = READ_ONCE(*crstep); + + /* A large page has been split concurrently, try again */ + if (!crste_leaf(crste)) + goto again; if (!oldkey) oldkey = &prev; - return page_cond_set_storage_key(large_crste_to_phys(*crstep, gfn), skey, oldkey, + + return page_cond_set_storage_key(large_crste_to_phys(crste, gfn), skey, oldkey, nq, mr, mc); } @@ -768,7 +785,7 @@ int dat_reset_reference_bit(union asce asce, gfn_t gfn, union skey *skey) int rc; skey->skey = 0; - +again: rc = dat_entry_walk(NULL, gfn, asce, DAT_WALK_ANY, TABLE_TYPE_PAGE_TABLE, &crstep, &ptep); if (rc) return rc; @@ -776,9 +793,12 @@ int dat_reset_reference_bit(union asce asce, gfn_t gfn, union skey *skey) if (!ptep) { union crste crste = READ_ONCE(*crstep); - if (!crste.h.fc || !crste.s.fc1.pr) + /* A large page has been split concurrently, try again */ + if (!crste_leaf(crste) && !crste.h.i) + goto again; + if (!crste.s.fc1.pr) return 0; - skey->skey = page_reset_referenced(large_crste_to_phys(*crstep, gfn)) << 1; + skey->skey = page_reset_referenced(large_crste_to_phys(crste, gfn)) << 1; return 0; } old = pgste_get_lock(ptep); From 4ca00a9154f998116fba9a32cce5bd938d228065 Mon Sep 17 00:00:00 2001 From: Claudio Imbrenda Date: Fri, 28 Aug 2026 13:54:39 +0200 Subject: [PATCH 008/365] KVM: s390: Fix race in _destroy_pages_crste() Use READ_ONCE() in _destroy_pages_crste() to read the crste, avoid dereferencing the pointer multiple times. Fixes: a2c17f9270cc ("KVM: s390: New gmap code") Signed-off-by: Claudio Imbrenda Message-ID: <20260828115439.145885-9-imbrenda@linux.ibm.com> --- arch/s390/kvm/gmap/gmap.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/arch/s390/kvm/gmap/gmap.c b/arch/s390/kvm/gmap/gmap.c index 4968330e9553..3f3fa864cc36 100644 --- a/arch/s390/kvm/gmap/gmap.c +++ b/arch/s390/kvm/gmap/gmap.c @@ -994,11 +994,13 @@ static long _destroy_pages_pte(union pte *ptep, gfn_t gfn, gfn_t next, struct da static long _destroy_pages_crste(union crste *crstep, gfn_t gfn, gfn_t next, struct dat_walk *walk) { phys_addr_t origin, cur, end; + union crste crste; - if (!crstep->h.fc || !crstep->s.fc1.pr) + crste = READ_ONCE(*crstep); + if (!crste.h.fc || !crste.s.fc1.pr) return 0; - origin = crste_origin_large(*crstep); + origin = crste_origin_large(crste); cur = ((max(gfn, walk->start) - gfn) << PAGE_SHIFT) + origin; end = ((min(next, walk->end) - gfn) << PAGE_SHIFT) + origin; for ( ; cur < end; cur += PAGE_SIZE) From 65e05ec252a9b79e75930d3c4dd42d8877db04c5 Mon Sep 17 00:00:00 2001 From: Anthony Krowiak Date: Tue, 18 Aug 2026 15:33:49 -0400 Subject: [PATCH 009/365] s390/vfio-ap: fix KVM GISC and page leak when queue removed from host config Three related problems exist in the handling of KVM interrupt and page resources when a queue is removed from the host's AP configuration while assigned to a mediated device (mdev). Problem 1: ~~~~~~~~~ AP_RESPONSE_Q_NOT_AVAIL not handled in vfio_ap_mdev_reset_queue() When the AP bus removes a queue device whose adapter or domain has been removed from the host's AP configuration, vfio_ap_mdev_remove_queue() is called. If the queue is still in the host's AP configuration at that point, it calls vfio_ap_mdev_reset_queue(), which issues a PQAP(ZAPQ). Since the adapter is already gone from the host configuration, ap_zapq() returns AP_RESPONSE_Q_NOT_AVAIL (0x01). This response code is not handled in vfio_ap_mdev_reset_queue()'s switch statement and falls through to the default case, which issues a WARN but does not call vfio_ap_free_aqic_resources(). As a result, if IRQ handling was enabled for the queue by the guest, the KVM GISC registration and the pinned guest page holding the notification indicator byte (NIB) are both leaked. This is fixed by adding AP_RESPONSE_Q_NOT_AVAIL to the same case as AP_RESPONSE_DECONFIGURED and AP_RESPONSE_CHECKSTOPPED in vfio_ap_mdev_reset_queue(). Like those response codes, Q_NOT_AVAIL indicates the queue is not operational and no further reset attempts are possible; the correct action is to free the IRQ resources immediately. Problem 2: ~~~~~~~~~ AP_RESPONSE_Q_NOT_AVAIL not handled in apq_status_check() In vfio_ap_mdev_reset_queue(), there are four cases that indicate a queue reset has not yet completed, in which case apq_reset_check() is queued to a work queue to verify completion of the reset operation. This function uses the PQAP(TAPQ) function to get the queue's status and calls apq_status_check() to verify whether the reset has completed, failed or needs to be executed again. As described in Problem #1 above, apq_reset_check() does not specifically check for AP_RESPONSE_Q_NOT_AVAIL, thereby potentially leaking KVM GISC registration and the pinned guest page holding the NIB. This is fixed by adding a case statement for AP_RESPONSE_Q_NOT_AVAIL to apq_status_check() and returning -ENODEV for that case. The caller, apq_reset_check() will then check for this return code and call vfio_ap_free_aqic_resources() to prevent the leak. Problem 3: ~~~~~~~~~ vfio_ap_free_aqic_resources() leaks saved_isc when kvm is NULL vfio_ap_free_aqic_resources() guards the call to kvm_s390_gisc_unregister() with: if (q->saved_isc != VFIO_AP_ISC_INVALID && !WARN_ON(!(q->matrix_mdev && q->matrix_mdev->kvm))) If matrix_mdev->kvm is NULL -- which can happen when vfio_ap_mdev_unset_kvm() has already run and cleared kvm before a subsequent cleanup path reaches this function -- the WARN_ON fires and the entire block is skipped. This leaves q->saved_isc set to a non-invalid value, creating a potential double-free on any subsequent call to this function. When kvm is NULL the KVM guest is already torn down, so kvm_s390_gisc_unregister() need not and cannot be called; however, q->saved_isc must always be cleared. Fix this by separating the kvm_s390_gisc_unregister() call from the q->saved_isc reset. The WARN_ON now guards only the genuinely impossible case of matrix_mdev being NULL. A NULL kvm is handled gracefully by skipping only the unregister call, and q->saved_isc = VFIO_AP_ISC_INVALID is set unconditionally whenever saved_isc was not already invalid. Additionally, add an else clause to the host-config check in vfio_ap_mdev_remove_queue() to call vfio_ap_free_aqic_resources() directly when the queue is not in the host's AP configuration. This serves as a backstop: when the AP bus fires the driver .remove callback after an adapter is removed from the host config, the queue is by definition no longer addressable, so vfio_ap_mdev_reset_queue() would always return Q_NOT_AVAIL. The else clause handles this case directly without the unnecessary ap_zapq() call, and ensures cleanup occurs even if kvm has already been set to NULL by a prior call to vfio_ap_mdev_unset_kvm(). Fixes: b9bd10c43456d ("s390/vfio-ap: do not reset queue removed from host config") Cc: stable@vger.kernel.org Signed-off-by: Anthony Krowiak Reviewed-by: Matthew Rosato Acked-by: Halil Pasic Signed-off-by: Claudio Imbrenda Message-ID: <20260818193349.1877940-2-akrowiak@linux.ibm.com> --- drivers/s390/crypto/vfio_ap_ops.c | 18 ++++++++++++++---- 1 file changed, 14 insertions(+), 4 deletions(-) diff --git a/drivers/s390/crypto/vfio_ap_ops.c b/drivers/s390/crypto/vfio_ap_ops.c index 940c0ff668be..4db878c18f41 100644 --- a/drivers/s390/crypto/vfio_ap_ops.c +++ b/drivers/s390/crypto/vfio_ap_ops.c @@ -277,9 +277,9 @@ static void vfio_ap_free_aqic_resources(struct vfio_ap_queue *q) { if (!q) return; - if (q->saved_isc != VFIO_AP_ISC_INVALID && - !WARN_ON(!(q->matrix_mdev && q->matrix_mdev->kvm))) { - kvm_s390_gisc_unregister(q->matrix_mdev->kvm, q->saved_isc); + if (q->saved_isc != VFIO_AP_ISC_INVALID) { + if (!WARN_ON(!q->matrix_mdev) && q->matrix_mdev->kvm) + kvm_s390_gisc_unregister(q->matrix_mdev->kvm, q->saved_isc); q->saved_isc = VFIO_AP_ISC_INVALID; } if (q->saved_iova && !WARN_ON(!q->matrix_mdev)) { @@ -1935,6 +1935,8 @@ static int apq_status_check(int apqn, struct ap_queue_status *status) * a value indicating a reset needs to be performed again. */ return -EAGAIN; + case AP_RESPONSE_Q_NOT_AVAIL: + return -ENODEV; default: WARN(true, "failed to verify reset of queue %02x.%04x: TAPQ rc=%u\n", @@ -1961,6 +1963,10 @@ static void apq_reset_check(struct work_struct *reset_work) ret = apq_status_check(q->apqn, &status); if (ret == -EIO) return; + if (ret == -ENODEV) { + vfio_ap_free_aqic_resources(q); + return; + } if (ret == -EBUSY) { pr_notice_ratelimited(WAIT_MSG, elapsed, AP_QID_CARD(q->apqn), @@ -2004,6 +2010,7 @@ static void vfio_ap_mdev_reset_queue(struct vfio_ap_queue *q) break; case AP_RESPONSE_DECONFIGURED: case AP_RESPONSE_CHECKSTOPPED: + case AP_RESPONSE_Q_NOT_AVAIL: vfio_ap_free_aqic_resources(q); break; default: @@ -2528,12 +2535,15 @@ void vfio_ap_mdev_remove_queue(struct ap_device *apdev) /* * If the queue is not in the host's AP configuration, then resetting * it will fail with response code 01, (APQN not valid); so, let's make - * sure it is in the host's config. + * sure it is in the host's config. If it is not, free the KVM GISC + * resources. */ if (test_bit_inv(apid, (unsigned long *)matrix_dev->info.apm) && test_bit_inv(apqi, (unsigned long *)matrix_dev->info.aqm)) { vfio_ap_mdev_reset_queue(q); flush_work(&q->reset_work); + } else { + vfio_ap_free_aqic_resources(q); } done: From d12ce6bce5ec5175c3581e01c71e7a5abb286d9b Mon Sep 17 00:00:00 2001 From: Steffen Eiden Date: Wed, 12 Aug 2026 17:55:19 +0200 Subject: [PATCH 010/365] s390/uv: Fix loop condition in uv_find_secrets Systems with more than 85 UV secrets got -ENOENT for any secret past the first page. Fix this by setting the start index at the beginning of the loop in uv_find_secret() and not at the end. First test if there are more secrets left by comparing start_idx with list->next_secret_idx, and then set the start index to the next secret index. Fixes: 7c9137af2042 ("s390/uv: Retrieve UV secrets support") Acked-by: Claudio Imbrenda Reviewed-by: Christoph Schlameuss Signed-off-by: Steffen Eiden Signed-off-by: Claudio Imbrenda Message-ID: <20260812-uv_secrets_fix-v3-1-a85bd29e0666@linux.ibm.com> --- arch/s390/kernel/uv.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/arch/s390/kernel/uv.c b/arch/s390/kernel/uv.c index dc14ebc0105b..52003c463fad 100644 --- a/arch/s390/kernel/uv.c +++ b/arch/s390/kernel/uv.c @@ -846,11 +846,14 @@ int uv_find_secret(const u8 secret_id[UV_SECRET_ID_LEN], struct uv_secret_list *list, struct uv_secret_list_item_hdr *secret) { - u16 start_idx = 0; + u16 start_idx; u16 list_rc; int ret; + list->next_secret_idx = 0; + do { + start_idx = list->next_secret_idx; uv_list_secrets(list, start_idx, &list_rc, NULL); if (list_rc != UVC_RC_EXECUTED && list_rc != UVC_RC_MORE_DATA) { if (list_rc == UVC_RC_INV_CMD) @@ -861,7 +864,6 @@ int uv_find_secret(const u8 secret_id[UV_SECRET_ID_LEN], ret = find_secret_in_page(secret_id, list, secret); if (ret == 0) return ret; - start_idx = list->next_secret_idx; } while (list_rc == UVC_RC_MORE_DATA && start_idx < list->next_secret_idx); return -ENOENT; From f47190b08b71e8482072978373ee88cb2dfbdaf4 Mon Sep 17 00:00:00 2001 From: Steffen Eiden Date: Wed, 12 Aug 2026 17:55:20 +0200 Subject: [PATCH 011/365] s390/uv: Prevent potential out-of-bounds read When the system has more than 85 secrets, the uv_secret_list struct array only holds up to 85 items per page, resulting in an out of bounds read in find_secret_in_page if the targeted secret is in the next page or not stored at all. Fix this by looping over the number of stored secrets which is the per sub-list count of stored secrets and not the overall count. Fixes: 7c9137af2042 ("s390/uv: Retrieve UV secrets support") Signed-off-by: Steffen Eiden Reviewed-by: Christoph Schlameuss Signed-off-by: Claudio Imbrenda Message-ID: <20260812-uv_secrets_fix-v3-2-a85bd29e0666@linux.ibm.com> --- arch/s390/kernel/uv.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/s390/kernel/uv.c b/arch/s390/kernel/uv.c index 52003c463fad..8ea9dd7704ff 100644 --- a/arch/s390/kernel/uv.c +++ b/arch/s390/kernel/uv.c @@ -825,7 +825,7 @@ static int find_secret_in_page(const u8 secret_id[UV_SECRET_ID_LEN], { u16 i; - for (i = 0; i < list->total_num_secrets; i++) { + for (i = 0; i < list->num_secr_stored; i++) { if (memcmp(secret_id, list->secrets[i].id, UV_SECRET_ID_LEN) == 0) { *secret = list->secrets[i].hdr; return 0; From b5965c4221165045e19736794ca5f48ae3d67142 Mon Sep 17 00:00:00 2001 From: Pei Xiao Date: Thu, 13 Aug 2026 10:48:12 +0800 Subject: [PATCH 012/365] mtd: block2mtd: Fix divide error when erase_size is zero MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The erase size is parsed from the "block2mtd" module parameter and can be set to zero. add_device() then evaluates if (size % erase_size) with a zero divisor, which triggers a divide error: divide error: 0000 [#1] PREEMPT SMP PTI RIP: 0010:add_device drivers/mtd/devices/block2mtd.c:296 [inline] RIP: 0010:block2mtd_setup2+0x592/0xda0 drivers/mtd/devices/block2mtd.c:459 Call Trace: block2mtd_setup+0x27/0xe0 drivers/mtd/devices/block2mtd.c:476 param_attr_store+0x214/0x310 kernel/params.c:589 module_attr_store+0x65/0x90 kernel/params.c:904 kernfs_fop_write_iter+0x3a4/0x540 fs/kernfs/file.c:345 ... Reject a zero erase size before performing the modulo operation so the existing "erasesize must be a divisor of device size" error path reports the invalid argument and frees the device. Fixes: ea6d833a3fdd ("mtd: block2mtd: check device size") Reported-by: syzbot+b320a4d5f65a61dbbf89@syzkaller.appspotmail.com Closes: https://lore.kernel.org/lkml/6a7b58ba.ac361c09.22ff0a.0045.GAE@google.com/ Suggested-by: Jörn Engel Cc: stable@vger.kernel.org Signed-off-by: Pei Xiao Signed-off-by: Miquel Raynal --- drivers/mtd/devices/block2mtd.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/mtd/devices/block2mtd.c b/drivers/mtd/devices/block2mtd.c index 03e80b2c4f5a..349fa07be314 100644 --- a/drivers/mtd/devices/block2mtd.c +++ b/drivers/mtd/devices/block2mtd.c @@ -293,7 +293,7 @@ static struct block2mtd_dev *add_device(char *devname, int erase_size, } size = bdev_nr_bytes(bdev); - if ((long)size % erase_size) { + if (!erase_size || (long)size % erase_size) { pr_err("erasesize must be a divisor of device size\n"); goto err_free_block2mtd; } From 63d6cace2c4a7f36c1cb44f1bb5e0f3ef19e5c7f Mon Sep 17 00:00:00 2001 From: Han Xu Date: Thu, 13 Aug 2026 08:07:24 -0500 Subject: [PATCH 013/365] mtd: spinand: Enable QE on all dies MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The QUAD ENABLE (QE) bit is stored in a per-die configuration register on some SPI-NAND devices. When a device contains multiple dies, updating the QE bit only on the currently selected die can leave the remaining dies operating in non-quad mode.   Iterate over all targets and update the QE setting on each die during initialization to ensure consistent quad I/O operation across the entire device. Tested on ISSI IS38SMW04G8B. Fixes: 7529df465248 ("mtd: nand: Add core infrastructure to support SPI NANDs") Cc: stable@vger.kernel.org Signed-off-by: Han Xu Signed-off-by: Miquel Raynal --- drivers/mtd/nand/spi/core.c | 22 ++++++++++++++++++++-- 1 file changed, 20 insertions(+), 2 deletions(-) diff --git a/drivers/mtd/nand/spi/core.c b/drivers/mtd/nand/spi/core.c index 8bf9301f25e7..bea352dd82d7 100644 --- a/drivers/mtd/nand/spi/core.c +++ b/drivers/mtd/nand/spi/core.c @@ -281,8 +281,26 @@ static int spinand_init_cfg_cache(struct spinand_device *spinand) static int spinand_init_quad_enable(struct spinand_device *spinand, bool enable) { - return spinand_upd_cfg(spinand, CFG_QUAD_ENABLE, - enable ? CFG_QUAD_ENABLE : 0); + struct nand_device *nand = spinand_to_nand(spinand); + unsigned int target; + int ret; + + /* + * QE is a per-die setting on some devices. Program each target + * individually when enabling or disabling quad I/O mode. + */ + for (target = 0; target < nand->memorg.ntargets; target++) { + ret = spinand_select_target(spinand, target); + if (ret) + return ret; + + ret = spinand_upd_cfg(spinand, CFG_QUAD_ENABLE, + enable ? CFG_QUAD_ENABLE : 0); + if (ret) + return ret; + } + + return 0; } static int spinand_ecc_enable(struct spinand_device *spinand, From 26300879cd8e4612dce66bb8f6ff7cd35fcf3e59 Mon Sep 17 00:00:00 2001 From: Karl Mehltretter Date: Sun, 16 Aug 2026 18:30:27 +0200 Subject: [PATCH 014/365] mtd: core: avoid double-free of OTP NVMEM device If factory OTP setup fails after the user OTP NVMEM device has been registered, mtd_otp_nvmem_add() unregisters the user device but leaves mtd->otp_user_nvmem set. On an error, the caller unregisters it again. For -EOPNOTSUPP, registration continues and normal teardown unregisters it again. Clear mtd->otp_user_nvmem after unregistering it. Fixes: e0489f6e221f ("mtd: core: fix error path for nvmem provider") Fixes: fe0b8213c012 ("mtd: core: Don't fail mtd_otp_nvmem_add() if OTP is unsupported") Assisted-by: Codex:gpt-5.6-sol Signed-off-by: Karl Mehltretter Signed-off-by: Miquel Raynal --- drivers/mtd/mtdcore.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/mtd/mtdcore.c b/drivers/mtd/mtdcore.c index 16629382a787..6158452be24d 100644 --- a/drivers/mtd/mtdcore.c +++ b/drivers/mtd/mtdcore.c @@ -1083,6 +1083,7 @@ static int mtd_otp_nvmem_add(struct mtd_info *mtd) err: nvmem_unregister(mtd->otp_user_nvmem); + mtd->otp_user_nvmem = NULL; /* Don't report error if OTP is not supported. */ if (err == -EOPNOTSUPP) return 0; From 12b31d1acd20b3185bb5f0748db0cf6874c8e9d1 Mon Sep 17 00:00:00 2001 From: Karl Mehltretter Date: Sun, 16 Aug 2026 18:30:58 +0200 Subject: [PATCH 015/365] mtd: core: call _get_device() with the master MTD __get_mtd_device() calls the master's _get_device() callback with mtd, which may be a partition. __put_mtd_device() passes the master. This breaks gluebi partitions because gluebi_get_device() uses container_of() and therefore requires the master MTD. Pass the master to _get_device(), matching _put_device(). Fixes: 46b5889cc2c5 ("mtd: implement proper partition handling") Assisted-by: Codex:gpt-5.6-sol Signed-off-by: Karl Mehltretter Signed-off-by: Miquel Raynal --- drivers/mtd/mtdcore.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/mtd/mtdcore.c b/drivers/mtd/mtdcore.c index 6158452be24d..491a27d0315f 100644 --- a/drivers/mtd/mtdcore.c +++ b/drivers/mtd/mtdcore.c @@ -1333,7 +1333,7 @@ int __get_mtd_device(struct mtd_info *mtd) int err; if (master->_get_device) { - err = master->_get_device(mtd); + err = master->_get_device(master); if (err) return err; } From 39b975ff208610301bc400460ec245caeffa4ee0 Mon Sep 17 00:00:00 2001 From: Karl Mehltretter Date: Mon, 24 Aug 2026 03:47:03 +0200 Subject: [PATCH 016/365] mtd: cfi_cmdset_0001: shrink do_write_buffer() stack frame arm32 allmodconfig fails to build with gcc: drivers/mtd/chips/cfi_cmdset_0001.c:1883:1: error: the frame size of 1296 bytes is larger than 1280 bytes [-Werror=frame-larger-than=] With MTD_MAP_BANK_WIDTH_32 a map_word is 32 bytes, and with KASAN_STACK every temporary gets its own redzoned slot. do_write_buffer() builds a map_word for each of the twelve commands it issues, and those temporaries dominate its frame. Issue the commands through a small noinline helper that takes the command as a plain value and builds the map_word in its own frame. The sequence of reads and writes to the chip is unchanged. The helper must not be inlined, and reusing one map_word local is not enough: clang cannot assume the callee does not observe the local, so it still returns each CMD() into a per-call-site temporary, and the frame grows (1760 -> 1824 bytes in cfi_intelext_writev(), into which clang inlines do_write_buffer()). Frame sizes with MTD_MAP_BANK_WIDTH_32 and KASAN_STACK: gcc 15.2 do_write_buffer() 1296 -> 520 clang 21 cfi_intelext_writev() 1760 -> 896 The helper itself takes 168 (gcc) / 160 (clang) bytes. Reported-by: Andy Shevchenko Closes: https://lore.kernel.org/all/20260408211407.2295175-1-andriy.shevchenko@linux.intel.com/ Suggested-by: Miquel Raynal Link: https://lore.kernel.org/all/87ik9cfm6g.fsf@bootlin.com/ Assisted-by: LLM Signed-off-by: Karl Mehltretter Signed-off-by: Miquel Raynal --- drivers/mtd/chips/cfi_cmdset_0001.c | 42 ++++++++++++++++++----------- 1 file changed, 26 insertions(+), 16 deletions(-) diff --git a/drivers/mtd/chips/cfi_cmdset_0001.c b/drivers/mtd/chips/cfi_cmdset_0001.c index 6049ba2d6bcb..60d82a1acb01 100644 --- a/drivers/mtd/chips/cfi_cmdset_0001.c +++ b/drivers/mtd/chips/cfi_cmdset_0001.c @@ -1716,12 +1716,24 @@ static int cfi_intelext_write_words (struct mtd_info *mtd, loff_t to , size_t le } +/* + * Keep noinline: inlined, the map_word temporaries put do_write_buffer() over + * the frame-size limit with MTD_MAP_BANK_WIDTH_32 and KASAN_STACK. + */ +static noinline void __xipram cfi_write_cmd(struct map_info *map, + unsigned long cmd, unsigned long adr) +{ + struct cfi_private *cfi = map->fldrv_priv; + + map_write(map, CMD(cmd), adr); +} + static int __xipram do_write_buffer(struct map_info *map, struct flchip *chip, unsigned long adr, const struct kvec **pvec, unsigned long *pvec_seek, int len) { struct cfi_private *cfi = map->fldrv_priv; - map_word status, write_cmd, datum; + map_word status, datum; unsigned long cmd_adr; int ret, wbufsize, word_gap, words; const struct kvec *vec; @@ -1740,9 +1752,6 @@ static int __xipram do_write_buffer(struct map_info *map, struct flchip *chip, if (is_LH28F640BF(cfi)) cmd_adr = adr; - /* Let's determine this according to the interleave only once */ - write_cmd = (cfi->cfiq->P_ID != P_ID_INTEL_PERFORMANCE) ? CMD(0xe8) : CMD(0xe9); - mutex_lock(&chip->mutex); ret = get_chip(map, chip, cmd_adr, FL_WRITING); if (ret) { @@ -1759,7 +1768,7 @@ static int __xipram do_write_buffer(struct map_info *map, struct flchip *chip, So we must check here and reset those bits if they're set. Otherwise we're just pissing in the wind */ if (chip->state != FL_STATUS) { - map_write(map, CMD(0x70), cmd_adr); + cfi_write_cmd(map, 0x70, cmd_adr); chip->state = FL_STATUS; } status = map_read(map, cmd_adr); @@ -1767,21 +1776,22 @@ static int __xipram do_write_buffer(struct map_info *map, struct flchip *chip, xip_enable(map, chip, cmd_adr); printk(KERN_WARNING "SR.4 or SR.5 bits set in buffer write (status %lx). Clearing.\n", status.x[0]); xip_disable(map, chip, cmd_adr); - map_write(map, CMD(0x50), cmd_adr); - map_write(map, CMD(0x70), cmd_adr); + cfi_write_cmd(map, 0x50, cmd_adr); + cfi_write_cmd(map, 0x70, cmd_adr); } chip->state = FL_WRITING_TO_BUFFER; - map_write(map, write_cmd, cmd_adr); + cfi_write_cmd(map, (cfi->cfiq->P_ID != P_ID_INTEL_PERFORMANCE) ? 0xe8 : 0xe9, + cmd_adr); ret = WAIT_TIMEOUT(map, chip, cmd_adr, 0, 0); if (ret) { /* Argh. Not ready for write to buffer */ map_word Xstatus = map_read(map, cmd_adr); - map_write(map, CMD(0x70), cmd_adr); + cfi_write_cmd(map, 0x70, cmd_adr); chip->state = FL_STATUS; status = map_read(map, cmd_adr); - map_write(map, CMD(0x50), cmd_adr); - map_write(map, CMD(0x70), cmd_adr); + cfi_write_cmd(map, 0x50, cmd_adr); + cfi_write_cmd(map, 0x70, cmd_adr); xip_enable(map, chip, cmd_adr); printk(KERN_ERR "%s: Chip not ready for buffer write. Xstatus = %lx, status = %lx\n", map->name, Xstatus.x[0], status.x[0]); @@ -1800,7 +1810,7 @@ static int __xipram do_write_buffer(struct map_info *map, struct flchip *chip, } /* Write length of data to come */ - map_write(map, CMD(words), cmd_adr ); + cfi_write_cmd(map, words, cmd_adr); /* Write data */ vec = *pvec; @@ -1837,7 +1847,7 @@ static int __xipram do_write_buffer(struct map_info *map, struct flchip *chip, *pvec_seek = vec_seek; /* GO GO GO */ - map_write(map, CMD(0xd0), cmd_adr); + cfi_write_cmd(map, 0xd0, cmd_adr); chip->state = FL_WRITING; ret = INVAL_CACHE_AND_WAIT(map, chip, cmd_adr, @@ -1845,7 +1855,7 @@ static int __xipram do_write_buffer(struct map_info *map, struct flchip *chip, chip->buffer_write_time, chip->buffer_write_time_max); if (ret) { - map_write(map, CMD(0x70), cmd_adr); + cfi_write_cmd(map, 0x70, cmd_adr); chip->state = FL_STATUS; xip_enable(map, chip, cmd_adr); printk(KERN_ERR "%s: buffer write error (status timeout)\n", map->name); @@ -1858,8 +1868,8 @@ static int __xipram do_write_buffer(struct map_info *map, struct flchip *chip, unsigned long chipstatus = MERGESTATUS(status); /* reset status */ - map_write(map, CMD(0x50), cmd_adr); - map_write(map, CMD(0x70), cmd_adr); + cfi_write_cmd(map, 0x50, cmd_adr); + cfi_write_cmd(map, 0x70, cmd_adr); xip_enable(map, chip, cmd_adr); if (chipstatus & 0x02) { From 200c32e3cd53132fbf99f1e2a5751f556755cf6d Mon Sep 17 00:00:00 2001 From: Menachem Adin Date: Thu, 27 Aug 2026 08:11:56 +0300 Subject: [PATCH 017/365] mtd: mtd_intel_dg: reset poll counter for each erase The non-posted erase polling counter is initialized only once per MTD erase request. Large requests therefore share the polling budget across all 4K erase commands and can fail with -ETIME even though no individual command timed out. Reset the counter for each 4K erase command so every operation gets the intended completion timeout. Cc: stable@vger.kernel.org Fixes: a1c940cbf505 ("drm/xe/nvm: add support for non-posted erase") Signed-off-by: Menachem Adin Signed-off-by: Alexander Usyskin Signed-off-by: Miquel Raynal --- drivers/mtd/devices/mtd_intel_dg.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/mtd/devices/mtd_intel_dg.c b/drivers/mtd/devices/mtd_intel_dg.c index a566e86eb5e3..55a945c1a67a 100644 --- a/drivers/mtd/devices/mtd_intel_dg.c +++ b/drivers/mtd/devices/mtd_intel_dg.c @@ -387,7 +387,7 @@ idg_erase(struct intel_dg_nvm *nvm, u8 region, loff_t from, u64 len, u64 *fail_a void __iomem *base2 = nvm->base2; void __iomem *base = nvm->base; const u32 block = 0x10; - u32 iter = 0; + u32 iter; u32 reg; u64 i; @@ -396,6 +396,7 @@ idg_erase(struct intel_dg_nvm *nvm, u8 region, loff_t from, u64 len, u64 *fail_a iowrite32(region << 24 | block, base + NVM_ERASE_REG); if (nvm->non_posted_erase) { /* Wait for Erase Done */ + iter = 0; reg = ioread32(base2 + NVM_DEBUG_REG); while (!(reg & NVM_NON_POSTED_ERASE_DONE) && ++iter < NVM_NON_POSTED_ERASE_DONE_ITER) { From b890e6163761ddb580cc74f43e15f8bbde15647e Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Nuno=20S=C3=A1?= Date: Mon, 31 Aug 2026 16:07:17 +0100 Subject: [PATCH 018/365] mtd: spinand: fix NULL pointer dereference with no ECC engine MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit When "nand-no-ecc-engine" is set in DT, nanddev_get_ecc_engine() takes the NAND_ECC_ENGINE_TYPE_NONE path and returns success while leaving nand->ecc.engine NULL. The SPI-NAND code nevertheless dereferences it unconditionally to test for a pipelined engine, so probing such a device oopses immediately. Rather than open-coding the test three times, add a nand_ecc_is_pipelined() helper to the NAND core that folds the NULL check into the integration comparison, and use it everywhere. Future callers then cannot reintroduce the problem. Fixes: f9d7c7265bcf ("mtd: spinand: Create direct mapping descriptors for ECC operations") Cc: stable@vger.kernel.org Signed-off-by: Nuno Sá Signed-off-by: Miquel Raynal --- drivers/mtd/nand/ecc.c | 7 +++++++ drivers/mtd/nand/spi/core.c | 8 +++----- include/linux/mtd/nand.h | 2 ++ 3 files changed, 12 insertions(+), 5 deletions(-) diff --git a/drivers/mtd/nand/ecc.c b/drivers/mtd/nand/ecc.c index 6ccdff3fc913..1d809cbd47e6 100644 --- a/drivers/mtd/nand/ecc.c +++ b/drivers/mtd/nand/ecc.c @@ -159,6 +159,13 @@ int nand_ecc_finish_io_req(struct nand_device *nand, } EXPORT_SYMBOL(nand_ecc_finish_io_req); +bool nand_ecc_is_pipelined(const struct nand_device *nand) +{ + return nand->ecc.engine && + nand->ecc.engine->integration == NAND_ECC_ENGINE_INTEGRATION_PIPELINED; +} +EXPORT_SYMBOL(nand_ecc_is_pipelined); + /* Define default OOB placement schemes for large and small page devices */ static int nand_ooblayout_ecc_sp(struct mtd_info *mtd, int section, struct mtd_oob_region *oobregion) diff --git a/drivers/mtd/nand/spi/core.c b/drivers/mtd/nand/spi/core.c index bea352dd82d7..4fb769581595 100644 --- a/drivers/mtd/nand/spi/core.c +++ b/drivers/mtd/nand/spi/core.c @@ -526,8 +526,7 @@ static int spinand_read_from_cache_op(struct spinand_device *spinand, else rdesc->info.op_tmpl = &rdesc->info.primary_op_tmpl; - if (nand->ecc.engine->integration == NAND_ECC_ENGINE_INTEGRATION_PIPELINED && - req->mode != MTD_OPS_RAW) + if (nand_ecc_is_pipelined(nand) && req->mode != MTD_OPS_RAW) rdesc->info.op_tmpl->data.ecc = true; else rdesc->info.op_tmpl->data.ecc = false; @@ -621,8 +620,7 @@ static int spinand_write_to_cache_op(struct spinand_device *spinand, wdesc = spinand->dirmaps[req->pos.plane].wdesc; - if (nand->ecc.engine->integration == NAND_ECC_ENGINE_INTEGRATION_PIPELINED && - req->mode != MTD_OPS_RAW) + if (nand_ecc_is_pipelined(nand) && req->mode != MTD_OPS_RAW) wdesc->info.op_tmpl->data.ecc = true; else wdesc->info.op_tmpl->data.ecc = false; @@ -1279,7 +1277,7 @@ static int spinand_create_dirmap(struct spinand_device *spinand, struct spi_mem_dirmap_desc *desc; bool enable_ecc = false, secondary_op = false; - if (nand->ecc.engine->integration == NAND_ECC_ENGINE_INTEGRATION_PIPELINED) + if (nand_ecc_is_pipelined(nand)) enable_ecc = true; if (spinand->cont_read_possible && spinand->op_templates->cont_read_cache) diff --git a/include/linux/mtd/nand.h b/include/linux/mtd/nand.h index 09c8c93e4dba..6936180b6ea5 100644 --- a/include/linux/mtd/nand.h +++ b/include/linux/mtd/nand.h @@ -305,6 +305,8 @@ int nand_ecc_prepare_io_req(struct nand_device *nand, struct nand_page_io_req *req); int nand_ecc_finish_io_req(struct nand_device *nand, struct nand_page_io_req *req); +bool nand_ecc_is_pipelined(const struct nand_device *nand); + bool nand_ecc_is_strong_enough(struct nand_device *nand); #if IS_REACHABLE(CONFIG_MTD_NAND_CORE) From 636fe10d8f3559210ff63108a208d39ea2e9c3bd Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Nuno=20S=C3=A1?= Date: Mon, 31 Aug 2026 16:07:18 +0100 Subject: [PATCH 019/365] mtd: spinand: fix zero oobavail when no ECC engine is used MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Commit 00c15b78b4b4 ("mtd: spinand: Allow the case where there is no ECC engine") made the OOB free bytes count conditional on having an ECC engine so that probing would not fail when none is requested. However mtd->oobavail is still assigned from ret just after that block, and ret is 0 there, so the device ends up advertising no available OOB bytes at all. mtd_oobavail() returns mtd->oobavail for MTD_OPS_AUTO_OOB, so a zero value makes every automatic OOB access fail with -EINVAL. JFFS2 fares worse: it keeps its cleanmarker in the OOB area on NAND and refuses to mount outright, reporting "inconsistent device description". No ECC engine also means no ooblayout was ever installed, which would make the count return -ENOTSUPP, so install the same fallback layout the on-die path already uses before counting unconditionally. Fixes: 00c15b78b4b4 ("mtd: spinand: Allow the case where there is no ECC engine") Cc: stable@vger.kernel.org Signed-off-by: Nuno Sá Signed-off-by: Miquel Raynal --- drivers/mtd/nand/spi/core.c | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-) diff --git a/drivers/mtd/nand/spi/core.c b/drivers/mtd/nand/spi/core.c index 4fb769581595..43bbe2629b85 100644 --- a/drivers/mtd/nand/spi/core.c +++ b/drivers/mtd/nand/spi/core.c @@ -1996,12 +1996,17 @@ static int spinand_init(struct spinand_device *spinand) goto err_cleanup_ecc_engine; } - if (nand->ecc.engine) { - ret = mtd_ooblayout_count_freebytes(mtd); - if (ret < 0) - goto err_cleanup_ecc_engine; + if (!nand->ecc.engine) { + if (spinand->eccinfo.ooblayout) + mtd_set_ooblayout(mtd, spinand->eccinfo.ooblayout); + else + mtd_set_ooblayout(mtd, &spinand_noecc_ooblayout); } + ret = mtd_ooblayout_count_freebytes(mtd); + if (ret < 0) + goto err_cleanup_ecc_engine; + mtd->oobavail = ret; /* Propagate ECC information to mtd_info */ From 37adc9c5789c76db3b1ee7aeec3999b8503020d0 Mon Sep 17 00:00:00 2001 From: Mehmet Fide Date: Tue, 1 Sep 2026 09:39:06 +0200 Subject: [PATCH 020/365] mtd: rawnand: vf610_nfc: fix reads on chips with more than 64 bytes of OOB The controller transfers 64 spare bytes per page and the driver only implements the matching 64-byte ECC layout, so attach_chip() shrinks mtd->oobsize when the chip provides more. That clamp does not survive: nand_scan_tail() runs nanddev_init() after ->attach_chip(), and it restores mtd->oobsize from the memory organization, which still holds the value detected from the chip. The driver then transfers writesize plus the chip's full OOB size, the hardware ECC parity ends up at a different offset than the layout the controller was set up for, and every ECC-protected read fails with -EBADMSG. Measured on a Colibri VF61 (MX30LF4G28AC, 2048-byte pages, 112 bytes of OOB): with the clamp lost, UBI cannot read the erase counter headers of the pages U-Boot has just written, and the on-flash bad block table written by an older kernel reads back with ECC errors, so the board does not boot. Kernels before commit a7ab085d7c16 ("mtd: rawnand: Initialize the nand_device object") are not affected because nothing overwrote the clamp there, which is why the same chip works with a v4.4 kernel and with U-Boot, whose copy of this driver has no memory organization to restore the value from. Edward Karpicz reported that the clamp no longer takes effect on this chip; see the link below. Instead of modifying the memory organization, keep the detected OOB size and give the driver its own mtd_ooblayout_ops: the same layout the NAND core uses for large pages, but computed on the first 64 OOB bytes instead of the whole OOB, so the ECC bytes stay where U-Boot and the old kernels put them. The data paths transfer writesize plus those 64 bytes, as the controller always has. Since mtd->oobsize now reports the chip's real spare size, fill the tail of oob_poi with 0xff after the 64 transferred bytes on ECC page reads: the core may copy the full mtd->oobsize from it, which would otherwise expose whatever the buffer held before. 0xff also matches what a raw read returns from flash, since the write path only ever programs the first 64 spare bytes. Reported-by: Edward Karpicz Link: https://community.toradex.com/t/colibri-vf50-vf61-on-the-current-bsp-mainline-u-boot-v2026-07-and-linux-6-18-lts/30735 Suggested-by: Miquel Raynal Fixes: a7ab085d7c16 ("mtd: rawnand: Initialize the nand_device object") Cc: stable@vger.kernel.org Signed-off-by: Mehmet Fide Signed-off-by: Miquel Raynal --- drivers/mtd/nand/raw/vf610_nfc.c | 72 ++++++++++++++++++++++++++------ 1 file changed, 60 insertions(+), 12 deletions(-) diff --git a/drivers/mtd/nand/raw/vf610_nfc.c b/drivers/mtd/nand/raw/vf610_nfc.c index 9940681810cf..1c3e7b167e53 100644 --- a/drivers/mtd/nand/raw/vf610_nfc.c +++ b/drivers/mtd/nand/raw/vf610_nfc.c @@ -505,6 +505,11 @@ static int vf610_nfc_exec_op(struct nand_chip *chip, check_only); } +static unsigned int vf610_nfc_spare_size(struct mtd_info *mtd) +{ + return min_t(unsigned int, mtd->oobsize, 64); +} + static inline int vf610_nfc_correct_data(struct nand_chip *chip, uint8_t *dat, uint8_t *oob, int page) { @@ -522,7 +527,7 @@ static inline int vf610_nfc_correct_data(struct nand_chip *chip, uint8_t *dat, return ecc_count; nfc->data_access = true; - nand_read_oob_op(&nfc->chip, page, 0, oob, mtd->oobsize); + nand_read_oob_op(&nfc->chip, page, 0, oob, vf610_nfc_spare_size(mtd)); nfc->data_access = false; /* @@ -530,7 +535,7 @@ static inline int vf610_nfc_correct_data(struct nand_chip *chip, uint8_t *dat, * at least less then half of the ECC strength. */ return nand_check_erased_ecc_chunk(dat, nfc->chip.ecc.size, oob, - mtd->oobsize, NULL, 0, + vf610_nfc_spare_size(mtd), NULL, 0, flips_threshold); } @@ -551,7 +556,7 @@ static int vf610_nfc_read_page(struct nand_chip *chip, uint8_t *buf, { struct vf610_nfc *nfc = chip_to_nfc(chip); struct mtd_info *mtd = nand_to_mtd(chip); - int trfr_sz = mtd->writesize + mtd->oobsize; + int trfr_sz = mtd->writesize + vf610_nfc_spare_size(mtd); u32 row = 0, cmd1 = 0, cmd2 = 0, code = 0; int stat; @@ -577,11 +582,16 @@ static int vf610_nfc_read_page(struct nand_chip *chip, uint8_t *buf, */ vf610_nfc_rd_from_sram(buf, nfc->regs + NFC_MAIN_AREA(0), mtd->writesize, false); - if (oob_required) + if (oob_required) { + unsigned int spare = vf610_nfc_spare_size(mtd); + vf610_nfc_rd_from_sram(chip->oob_poi, nfc->regs + NFC_MAIN_AREA(0) + mtd->writesize, - mtd->oobsize, false); + spare, false); + /* Not transferred, and never written: reads back erased */ + memset(chip->oob_poi + spare, 0xff, mtd->oobsize - spare); + } stat = vf610_nfc_correct_data(chip, buf, chip->oob_poi, page); @@ -599,7 +609,7 @@ static int vf610_nfc_write_page(struct nand_chip *chip, const uint8_t *buf, { struct vf610_nfc *nfc = chip_to_nfc(chip); struct mtd_info *mtd = nand_to_mtd(chip); - int trfr_sz = mtd->writesize + mtd->oobsize; + int trfr_sz = mtd->writesize + vf610_nfc_spare_size(mtd); u32 row = 0, cmd1 = 0, cmd2 = 0, code = 0; u8 status; int ret; @@ -740,6 +750,49 @@ static void vf610_nfc_init_controller(struct vf610_nfc *nfc) } } +/* + * With 64 byte OOB chips the core's large page layout matches what + * U-Boot uses, and on chips with more, U-Boot and older kernels clamped + * mtd->oobsize to 64. Modifying the OOB size is no longer possible, the + * actual chip geometry must be respected, so to avoid breaking those + * existing setups use our own layout: the core's large page one, + * computed over the first 64 spare bytes only. + */ +static int vf610_nfc_ooblayout_ecc(struct mtd_info *mtd, int section, + struct mtd_oob_region *oobregion) +{ + struct nand_device *nand = mtd_to_nanddev(mtd); + unsigned int total_ecc_bytes = nand->ecc.ctx.total; + + if (section || !total_ecc_bytes) + return -ERANGE; + + oobregion->length = total_ecc_bytes; + oobregion->offset = vf610_nfc_spare_size(mtd) - oobregion->length; + + return 0; +} + +static int vf610_nfc_ooblayout_free(struct mtd_info *mtd, int section, + struct mtd_oob_region *oobregion) +{ + struct nand_device *nand = mtd_to_nanddev(mtd); + unsigned int total_ecc_bytes = nand->ecc.ctx.total; + + if (section) + return -ERANGE; + + oobregion->length = vf610_nfc_spare_size(mtd) - total_ecc_bytes - 2; + oobregion->offset = 2; + + return 0; +} + +static const struct mtd_ooblayout_ops vf610_nfc_ooblayout_ops = { + .ecc = vf610_nfc_ooblayout_ecc, + .free = vf610_nfc_ooblayout_free, +}; + static int vf610_nfc_attach_chip(struct nand_chip *chip) { struct mtd_info *mtd = nand_to_mtd(chip); @@ -770,12 +823,7 @@ static int vf610_nfc_attach_chip(struct nand_chip *chip) return -ENXIO; } - /* Only 64 byte ECC layouts known */ - if (mtd->oobsize > 64) - mtd->oobsize = 64; - - /* Use default large page ECC layout defined in NAND core */ - mtd_set_ooblayout(mtd, nand_get_large_page_ooblayout()); + mtd_set_ooblayout(mtd, &vf610_nfc_ooblayout_ops); if (chip->ecc.strength == 32) { nfc->ecc_mode = ECC_60_BYTE; chip->ecc.bytes = 60; From 68fe2faf5c69d0e21f94cd695d28f0ba677d484f Mon Sep 17 00:00:00 2001 From: Mehmet Fide Date: Tue, 1 Sep 2026 09:39:07 +0200 Subject: [PATCH 021/365] mtd: rawnand: vf610_nfc: fix false bitflips on reads of erased pages When the ECC engine fails to decode a page, the driver re-reads the OOB area with the engine bypassed, but runs the erased-page check for the data area on the buffer left in the controller SRAM by the failed transfer. That buffer does not hold what is on the flash: the failing engine writes a bogus single-bit "correction" into it. In the 60-byte ECC mode the all-0xff content of an erased page always decodes to the same error location, so every erased page shows one stale zero bit at data offset 0x5FD, which the erased-page check then reports as a corrected bitflip. Edward Karpicz discovered this behaviour and identified the offset on a Colibri VF61; the analysis and the fix build on his finding. Measured with an instrumented driver on a Colibri VF50 (MX30LF1G18AC, 32-bit ECC): reading a 126 MiB partition with nanddump increased the corrected counter by 18035, exactly one per erased page, while raw reads of the same pages return clean 0xff. A v4.4 kernel on the VF61 (MX30LF4G28AC) accumulates the same false counts, so the behaviour follows the controller rather than the chip or the driver generation. Neither the Vybrid reference manual nor the published mask set errata (VFXXX_2N02G) document it. The 45-byte ECC mode is not affected. Restoring the known byte is not enough: on pages that fail to decode with content other than all-0xff the engine writes its correction wherever the syndrome points (measured at a different offset on such a page), so the check has to run on what the flash holds. Re-read the data area with the ECC engine bypassed, exactly as already done for the OOB area. The corrected counter then stays at zero on both boards. Reported-by: Edward Karpicz Link: https://community.toradex.com/t/colibri-vf50-vf61-on-the-current-bsp-mainline-u-boot-v2026-07-and-linux-6-18-lts/30735 Signed-off-by: Mehmet Fide Signed-off-by: Miquel Raynal --- drivers/mtd/nand/raw/vf610_nfc.c | 15 +++++++++++---- 1 file changed, 11 insertions(+), 4 deletions(-) diff --git a/drivers/mtd/nand/raw/vf610_nfc.c b/drivers/mtd/nand/raw/vf610_nfc.c index 1c3e7b167e53..1b9b370adfab 100644 --- a/drivers/mtd/nand/raw/vf610_nfc.c +++ b/drivers/mtd/nand/raw/vf610_nfc.c @@ -519,6 +519,7 @@ static inline int vf610_nfc_correct_data(struct nand_chip *chip, uint8_t *dat, u8 ecc_status; u8 ecc_count; int flips_threshold = nfc->chip.ecc.strength / 2; + int ret; ecc_status = vf610_nfc_read(nfc, ecc_status_off) & 0xff; ecc_count = ecc_status & ECC_STATUS_ERR_COUNT; @@ -526,15 +527,21 @@ static inline int vf610_nfc_correct_data(struct nand_chip *chip, uint8_t *dat, if (!(ecc_status & ECC_STATUS_MASK)) return ecc_count; + /* The failed decode leaves a bogus correction in SRAM; re-read without ECC */ nfc->data_access = true; - nand_read_oob_op(&nfc->chip, page, 0, oob, vf610_nfc_spare_size(mtd)); + ret = nand_read_page_op(&nfc->chip, page, 0, dat, mtd->writesize); + if (!ret) + ret = nand_read_oob_op(&nfc->chip, page, 0, oob, + vf610_nfc_spare_size(mtd)); nfc->data_access = false; + if (ret) + return ret; /* - * On an erased page, bit count (including OOB) should be zero or - * at least less then half of the ECC strength. + * Run the erased-page check with the driver's historic threshold + * of half the ECC strength. */ - return nand_check_erased_ecc_chunk(dat, nfc->chip.ecc.size, oob, + return nand_check_erased_ecc_chunk(dat, mtd->writesize, oob, vf610_nfc_spare_size(mtd), NULL, 0, flips_threshold); } From 21f027016b1290d13c30b198ae7a00e6b3d1d5a5 Mon Sep 17 00:00:00 2001 From: Runyu Xiao Date: Wed, 2 Sep 2026 15:05:42 +0800 Subject: [PATCH 022/365] mtd: rawnand: cadence: Initialize IRQ state before requesting IRQ The Cadence NAND interrupt handler uses both the IRQ lock and completion object. Registering the IRQ before initializing them leaves a window in which a pending interrupt can access uninitialized synchronization state. Initialize them before registering the handler. Fixes: ec4ba01e894d ("mtd: rawnand: Add new Cadence NAND driver to MTD subsystem") Cc: stable@vger.kernel.org Assisted-by: Codex:GPT-5 Signed-off-by: Runyu Xiao Signed-off-by: Miquel Raynal --- drivers/mtd/nand/raw/cadence-nand-controller.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/drivers/mtd/nand/raw/cadence-nand-controller.c b/drivers/mtd/nand/raw/cadence-nand-controller.c index d53b35a8b3cb..2085632631c0 100644 --- a/drivers/mtd/nand/raw/cadence-nand-controller.c +++ b/drivers/mtd/nand/raw/cadence-nand-controller.c @@ -3143,6 +3143,9 @@ static int cadence_nand_init(struct cdns_nand_ctrl *cdns_ctrl) goto free_buf_desc; } + spin_lock_init(&cdns_ctrl->irq_lock); + init_completion(&cdns_ctrl->complete); + if (devm_request_irq(cdns_ctrl->dev, cdns_ctrl->irq, cadence_nand_isr, IRQF_SHARED, "cadence-nand-controller", cdns_ctrl)) { @@ -3151,9 +3154,6 @@ static int cadence_nand_init(struct cdns_nand_ctrl *cdns_ctrl) goto free_buf; } - spin_lock_init(&cdns_ctrl->irq_lock); - init_completion(&cdns_ctrl->complete); - ret = cadence_nand_hw_init(cdns_ctrl); if (ret) goto disable_irq; From 86ad6489e38c76c06c0b4a4229391a794023b5e0 Mon Sep 17 00:00:00 2001 From: Julian Braha Date: Sun, 16 Aug 2026 15:22:07 +0100 Subject: [PATCH 023/365] ARC: cleanup dead ARC_CANT_LLSC option in Kconfig Currently, the ARC_CANT_LLSC config option can never be enabled. It is also not referenced anywhere else. So let's remove this dead code. Note that I have only compile-tested these changes, because I do not have the hardware to runtime test. This dead code was found by kconfirm, a static analysis tool for Kconfig. Signed-off-by: Julian Braha Signed-off-by: Vineet Gupta --- arch/arc/Kconfig | 4 ---- 1 file changed, 4 deletions(-) diff --git a/arch/arc/Kconfig b/arch/arc/Kconfig index 2ed7186c81c5..80e61175bf52 100644 --- a/arch/arc/Kconfig +++ b/arch/arc/Kconfig @@ -331,13 +331,9 @@ config ARC_FPU_SAVE_RESTORE for floating point exceptions and rounding modes. These are preserved across task context switch when enabled. -config ARC_CANT_LLSC - def_bool n - config ARC_HAS_LLSC bool "Insn: LLOCK/SCOND (efficient atomic ops)" default y - depends on !ARC_CANT_LLSC if ISA_ARCV2 From 04eb7b5e043fbccf83d27a9be5c3973c2dbb110f Mon Sep 17 00:00:00 2001 From: Anthony Iliopoulos Date: Thu, 20 Aug 2026 15:13:39 +0200 Subject: [PATCH 024/365] arc: remove unused profile.h includes None of the code in these files uses declarations provided by profile.h. Remove the unnecessary header includes across the various files to clean up dependencies and reduce include bloat. Signed-off-by: Anthony Iliopoulos Signed-off-by: Vineet Gupta --- arch/arc/kernel/smp.c | 1 - 1 file changed, 1 deletion(-) diff --git a/arch/arc/kernel/smp.c b/arch/arc/kernel/smp.c index 2d99dffed0ce..ee210a94ce45 100644 --- a/arch/arc/kernel/smp.c +++ b/arch/arc/kernel/smp.c @@ -12,7 +12,6 @@ #include #include #include -#include #include #include #include From d12c6a6f0c8fd5b5b41120f7be319f743dae3b4f Mon Sep 17 00:00:00 2001 From: blaze <1466528493@qq.com> Date: Wed, 9 Sep 2026 12:42:59 +0800 Subject: [PATCH 025/365] arc: kernel: Fix clk reference leak in show_cpuinfo() In show_cpuinfo(), clk_get() acquires a reference to the CPU clock, but clk_put() is never called to release it. Since show_cpuinfo() is the seq_file show callback for /proc/cpuinfo, this causes a clock reference count leak every time /proc/cpuinfo is read. The frequency is extracted via clk_get_rate() into a local variable, so the clock reference can be safely released immediately after. Fix this by calling clk_put(cpu_clk) right after clk_get_rate(). Fixes: 98f2892607c5 ("ARC: [cpuinfo] Add clk info in /proc/cpuinfo") Signed-off-by: blaze <1466528493@qq.com> Signed-off-by: Vineet Gupta --- arch/arc/kernel/setup.c | 1 + 1 file changed, 1 insertion(+) diff --git a/arch/arc/kernel/setup.c b/arch/arc/kernel/setup.c index 7b6a9beba9db..8bebd0e6fc2f 100644 --- a/arch/arc/kernel/setup.c +++ b/arch/arc/kernel/setup.c @@ -593,6 +593,7 @@ static int show_cpuinfo(struct seq_file *m, void *v) cpu_id); } else { freq = clk_get_rate(cpu_clk); + clk_put(cpu_clk); } if (freq) seq_printf(m, "CPU speed\t: %lu.%02lu Mhz\n", From d22c3e0088e85be8131f7a9283f759cdbb20726d Mon Sep 17 00:00:00 2001 From: Sven Schnelle Date: Wed, 9 Sep 2026 11:29:53 +0200 Subject: [PATCH 026/365] selftests/ftrace: Fix unique symbol check in kprobe_non_uniq_symbol.tc The current regex also matches symbols in modules, which makes the test fail on s390 where name_show is present only once in the kernel, but also multiple times in modules: 000001b1401cdc20 t name_show 000001b0c05e6c40 t name_show [mdev] 000001b0c0495f30 t name_show [i2c_core] Fix this by changing the regular expression to only match the function name. Link: https://lore.kernel.org/all/20260909092954.2200558-1-svens@linux.ibm.com/ Fixes: 03b80ff8023a ("selftests/ftrace: Add new test case which checks non unique symbol") Signed-off-by: Sven Schnelle Reviewed-by: Steven Rostedt Signed-off-by: Masami Hiramatsu (Google) --- .../selftests/ftrace/test.d/kprobe/kprobe_non_uniq_symbol.tc | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tools/testing/selftests/ftrace/test.d/kprobe/kprobe_non_uniq_symbol.tc b/tools/testing/selftests/ftrace/test.d/kprobe/kprobe_non_uniq_symbol.tc index bc9514428dba..07b1177c1634 100644 --- a/tools/testing/selftests/ftrace/test.d/kprobe/kprobe_non_uniq_symbol.tc +++ b/tools/testing/selftests/ftrace/test.d/kprobe/kprobe_non_uniq_symbol.tc @@ -6,7 +6,7 @@ SYMBOL='name_show' # We skip this test on kernel where SYMBOL is unique or does not exist. -if [ "$(grep -c -E "[[:alnum:]]+ t ${SYMBOL}" /proc/kallsyms)" -le '1' ]; then +if [ "$(grep -c -E "[[:alnum:]]+ t ${SYMBOL}$" /proc/kallsyms)" -le '1' ]; then exit_unsupported fi From 57a78ad2305f299bc3f933ed36b3d402df04784a Mon Sep 17 00:00:00 2001 From: David Howells Date: Wed, 9 Sep 2026 09:06:31 +0100 Subject: [PATCH 027/365] block: Fix start and length check added to iov_iter_extract_bvecs() Commit 14b007e17881 added an address check using iter_iov_addr() and a length check using iter_iov_len() to iov_iter_extract_bvecs(), but these cannot be used so and are unsafe in this circumstance as the functions have hardwired assumptions about the iterator type. They should only be used with ITER_UBUF or ITER_IOVEC-type iterators; they shouldn't be used with ITER_BVEC, ITER_KVEC, ITER_FOLIOQ, ITER_XARRAY or ITER_DISCARD iterators. This proves to be a problem for cachefiles as an iterator of type ITER_FOLIOQ is passed and iter_iov_addr() and iter_iov_len() both malfunction because iter->__iov in iter_iov() is not pointing to an iovec array. Fix this by using iov_iter_alignment() instead. Fixes: 14b007e17881 ("block: validate user space vectors during extraction") Signed-off-by: David Howells Link: https://patch.msgid.link/1667275.1788941191@warthog.procyon.org.uk Reviewed-by: Keith Busch Reviewed-by: Christoph Hellwig cc: Hannes Reinecke cc: Christoph Hellwig cc: Jens Axboe cc: Alexander Viro cc: Paulo Alcantara cc: netfs@lists.linux.dev cc: linux-block@vger.kernel.org cc: linux-fsdevel@vger.kernel.org Signed-off-by: Christian Brauner (Amutable) --- lib/iov_iter.c | 18 ++++++++++++++++-- 1 file changed, 16 insertions(+), 2 deletions(-) diff --git a/lib/iov_iter.c b/lib/iov_iter.c index 6665372ecf71..2072c04e99d0 100644 --- a/lib/iov_iter.c +++ b/lib/iov_iter.c @@ -1921,15 +1921,29 @@ ssize_t iov_iter_extract_bvecs(struct iov_iter *iter, struct bio_vec *bv, unsigned short max_vecs, unsigned mem_align_mask, iov_iter_extraction_t extraction_flags) { - unsigned long start = (unsigned long)iter_iov_addr(iter); unsigned short entries_left = max_vecs - *nr_vecs; unsigned short nr_pages, i = 0; size_t left, offset, len; struct page **pages; ssize_t size; - if ((start | iter_iov_len(iter)) & mem_align_mask) + /* + * DMA engines typically have both memory address and length alignment + * requirements, so check these against the alignment mask. For UBUF, + * IOVEC and KVEC, only the current segment will be extracted from; for + * everything else we might extract from multiple segments, so we need + * to check those too. + */ + if (likely(iter_is_ubuf(iter) || + iter_is_iovec(iter) || + iov_iter_is_kvec(iter))) { + unsigned long start = (unsigned long)iter_iov_addr(iter); + + if ((start | iter_iov_len(iter)) & mem_align_mask) + return -EINVAL; + } else if (iov_iter_alignment(iter) & mem_align_mask) { return -EINVAL; + } /* * Move page array up in the allocated memory for the bio vecs as far as From d59ac79915daa567c69aa93d458d41844676e92a Mon Sep 17 00:00:00 2001 From: Disha Goel Date: Fri, 3 Jul 2026 20:37:42 +0530 Subject: [PATCH 028/365] selftests/filesystems: fix missing and stale TARGETS entries filesystems/eventfd, filesystems/open_tree_ns and filesystems/xattr were never added to TARGETS when introduced. filesystems/openat2 was moved from selftests/openat2/ but the TARGETS entry was never updated, leaving a stale entry pointing at a directory that no longer exists. Fix this by adding the four missing subdirectories to TARGETS and removing the stale openat2 entry. Link: https://lore.kernel.org/20260703150742.58991-1-disgoel@linux.ibm.com Fixes: 7c37857fc23a ("selftests: add eventfd selftests") Fixes: b8f7622aa6e3 ("selftests/open_tree: add OPEN_TREE_NAMESPACE tests") Fixes: 7e28fef5d4db ("selftests/xattr: path-based AF_UNIX socket xattr tests") Fixes: fe087927046c ("selftests: move openat2 tests to selftests/filesystems/") Signed-off-by: Disha Goel Reviewed-by: Christian Brauner (Amutable) Cc: "Darrick J. Wong" Cc: Jan Kara Cc: Jeff Layton Cc: Shuah Khan Cc: Wen Yang Signed-off-by: Andrew Morton Link: https://patch.msgid.link/20260904183659.B81CD1F00A3D@smtp.kernel.org Signed-off-by: Christian Brauner (Amutable) --- tools/testing/selftests/Makefile | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/tools/testing/selftests/Makefile b/tools/testing/selftests/Makefile index 2d960626750e..330d061f6366 100644 --- a/tools/testing/selftests/Makefile +++ b/tools/testing/selftests/Makefile @@ -35,8 +35,11 @@ TARGETS += fchmodat2 TARGETS += filesystems TARGETS += filesystems/binderfs TARGETS += filesystems/epoll +TARGETS += filesystems/eventfd TARGETS += filesystems/failfs TARGETS += filesystems/fat +TARGETS += filesystems/openat2 +TARGETS += filesystems/open_tree_ns TARGETS += filesystems/overlayfs TARGETS += filesystems/statmount TARGETS += filesystems/mount-notify @@ -47,6 +50,7 @@ TARGETS += filesystems/empty_mntns TARGETS += filesystems/fsmount_ns TARGETS += filesystems/fscontext_ns TARGETS += filesystems/mntns_cleanup +TARGETS += filesystems/xattr TARGETS += firmware TARGETS += fpu TARGETS += ftrace @@ -103,7 +107,6 @@ TARGETS += prctl TARGETS += proc TARGETS += pstore TARGETS += ptrace -TARGETS += openat2 TARGETS += rdma TARGETS += resctrl TARGETS += riscv From 1abd643f3783ea8f8e273c18697ff0413aa92dc7 Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Wed, 9 Sep 2026 11:03:18 +0200 Subject: [PATCH 029/365] fs/ntfs3: use d_instantiate_new() in ntfs_create_inode() and murder syzbot's "WARNING in do_new_mount" saga ntfs_create_inode() creates a new inode via ntfs_new_inode(). It hashes it with insert_inode_locked() and so it's marked as I_NEW until unlock_new_inode(). ntfs 3 calls d_instantiate() in between though... Since the dentry was already hashed by the lookup before the create any path walk finds it without touching the parent's i_rwsem and so can lock the inode. If the inode is a directory unlock_new_inode() calls lockdep_annotate_inode_mutex_key() and marks i_rwsem with the i_mutex_dir_key class. That resets the count and the owner of a lock somebody else may already hold by now... syzbot has been spamming us with the same godforsaken bug "WARNING in do_new_mount" since 2023. I can't take it anymore so I went looking. Afaict, syzbot's executor chdirs into a freshly mounted ntfs3 image, creates a directory and then mounts some pseudofs on it. Everytime the mkdir() takes longer than syzbot waits mount() runs concurrently: mkdir("./sys") mount(NULL, "./sys", "sysfs") ntfs_create_inode() d_instantiate() user_path_at() finds the dentry do_lock_mount() inode_lock(inode) namespace_lock() unlock_new_inode() lockdep_annotate_inode_mutex_key() init_rwsem(&inode->i_rwsem) unlock_mount() inode_unlock(inode) The mount side then releases a lock that according to the rwsem nobody holds: DEBUG_RWSEMS_WARN_ON((rwsem_owner(sem) != current) && ...): count = 0x0, magic = 0xffff888043a854e8, owner = 0x0, curr 0xffff888000244880, list empty WARNING: CPU: 0 PID: 5346 at kernel/locking/rwsem.c:1368 __up_write Call Trace: inode_unlock include/linux/fs.h:877 [inline] unlock_mount fs/namespace.c:2892 [inline] do_new_mount_fc fs/namespace.c:3828 [inline] do_new_mount+0x777/0xa40 fs/namespace.c:3887 On PREEMPT_RT the same thing shows up as DEBUG_LOCKS_WARN_ON(rt_mutex_owner(lock) != current) WARNING: kernel/locking/rtmutex_common.h:193 at rt_mutex_slowunlock The up_write() underflows the reset count. A following inode_lock() on that directory then never returns. A path walk into the new directory racing with the mkdir() corrupts the lock the same way via inode_lock_shared() in lookup_slow(). Switch to d_instantiate_new() and drop the trailing unlock_new_inode(). All error paths bail out before that point with I_NEW still set and keep using discard_new_inode(). May we never see this fscking bug report again. Link: https://patch.msgid.link/20260909-work-ntfs3-d_instantiate_new-v1-1-2db697162ce8@kernel.org Fixes: 82cae269cfa9 ("fs/ntfs3: Add initialization of super block") Reviewed-by: Jan Kara Cc: stable@vger.kernel.org # v5.15+ Reported-by: syzbot+2a13ad6914e6fcec716c@syzkaller.appspotmail.com Closes: https://lore.kernel.org/6a9beced.a5e650b3.26d8a.000b.GAE@google.com Signed-off-by: Christian Brauner (Amutable) --- fs/ntfs3/inode.c | 7 ++----- 1 file changed, 2 insertions(+), 5 deletions(-) diff --git a/fs/ntfs3/inode.c b/fs/ntfs3/inode.c index 56b4f6469a28..4ac26c80bd34 100644 --- a/fs/ntfs3/inode.c +++ b/fs/ntfs3/inode.c @@ -1866,10 +1866,10 @@ int ntfs_create_inode(struct mnt_idmap *idmap, struct inode *dir, goto out6; /* - * Call 'd_instantiate' after inode->i_op is set + * Call 'd_instantiate_new' after inode->i_op is set * but before finish_open. */ - d_instantiate(dentry, inode); + d_instantiate_new(dentry, inode); /* Set original time. inode times (i_ctime) may be changed in ntfs_init_acl. */ inode_set_atime_to_ts(inode, ni->i_crtime); @@ -1917,9 +1917,6 @@ int ntfs_create_inode(struct mnt_idmap *idmap, struct inode *dir, if (!fnd) ni_unlock(dir_ni); - if (!err) - unlock_new_inode(inode); - return err; } From 79a71cc2568f4b5d42284da2aa26f3b4f47ce01b Mon Sep 17 00:00:00 2001 From: Jim Mattson Date: Wed, 2 Sep 2026 11:47:11 -0700 Subject: [PATCH 030/365] KVM: x86/pmu: Move Intel PMU global MSRs to intel_is_valid_msr() Commit c85cdc1cc1ea ("KVM: x86/pmu: Move handling PERF_GLOBAL_CTRL and friends to common x86") moved the existence check for the following Intel PMU MSRs to kvm_pmu_is_valid_msr(): - MSR_CORE_PERF_GLOBAL_STATUS - MSR_CORE_PERF_GLOBAL_CTRL - MSR_CORE_PERF_GLOBAL_OVF_CTRL That commit deemed these MSRs valid whenever pmu->version > 1. It intended to share the check with AMD PerfMonV2 because both vendor implementations require version 2 or greater for global PMU controls. However, as noted in the commit message, AMD uses different MSR indices for its global PMU registers. Commit 4a2771895ca6 ("KVM: x86/svm/pmu: Add AMD PerfMonV2 support") subsequently added AMD PerfMonV2 support and set pmu->version = 2. Because kvm_pmu_is_valid_msr() validated the Intel MSRs whenever pmu->version > 1, KVM incorrectly permitted AMD guests with PerfMonV2 to access these Intel MSRs without a #GP. Move the validation of these Intel MSRs to intel_is_valid_msr() and remove the common switch statement from kvm_pmu_is_valid_msr(). AMD already validates its own global PMU MSRs in amd_is_valid_msr(). Fixes: 4a2771895ca6 ("KVM: x86/svm/pmu: Add AMD PerfMonV2 support") Signed-off-by: Jim Mattson Reviewed-by: Like Xu Reviewed-by: Sandipan Das Link: https://patch.msgid.link/20260902184711.138538-1-jmattson@google.com Signed-off-by: Sean Christopherson --- arch/x86/kvm/pmu.c | 8 -------- arch/x86/kvm/vmx/pmu_intel.c | 3 +++ 2 files changed, 3 insertions(+), 8 deletions(-) diff --git a/arch/x86/kvm/pmu.c b/arch/x86/kvm/pmu.c index a7d60c8785cd..d2fd47ee5ec8 100644 --- a/arch/x86/kvm/pmu.c +++ b/arch/x86/kvm/pmu.c @@ -823,14 +823,6 @@ void kvm_pmu_deliver_pmi(struct kvm_vcpu *vcpu) bool kvm_pmu_is_valid_msr(struct kvm_vcpu *vcpu, u32 msr) { - switch (msr) { - case MSR_CORE_PERF_GLOBAL_STATUS: - case MSR_CORE_PERF_GLOBAL_CTRL: - case MSR_CORE_PERF_GLOBAL_OVF_CTRL: - return kvm_pmu_has_perf_global_ctrl(vcpu_to_pmu(vcpu)); - default: - break; - } return kvm_pmu_call(msr_idx_to_pmc)(vcpu, msr) || kvm_pmu_call(is_valid_msr)(vcpu, msr); } diff --git a/arch/x86/kvm/vmx/pmu_intel.c b/arch/x86/kvm/vmx/pmu_intel.c index bfa8612fb450..70a8c4816135 100644 --- a/arch/x86/kvm/vmx/pmu_intel.c +++ b/arch/x86/kvm/vmx/pmu_intel.c @@ -187,6 +187,9 @@ static bool intel_is_valid_msr(struct kvm_vcpu *vcpu, u32 msr) int ret; switch (msr) { + case MSR_CORE_PERF_GLOBAL_STATUS: + case MSR_CORE_PERF_GLOBAL_CTRL: + case MSR_CORE_PERF_GLOBAL_OVF_CTRL: case MSR_CORE_PERF_FIXED_CTR_CTRL: return kvm_pmu_has_perf_global_ctrl(pmu); case MSR_IA32_PEBS_ENABLE: From f13368e0acffd6d6289629705cb821d921104725 Mon Sep 17 00:00:00 2001 From: Sean Christopherson Date: Wed, 26 Aug 2026 09:14:30 -0700 Subject: [PATCH 031/365] KVM: selftests: Use __GLIBC__, not _GNU_SOURCE, to detect actual glibc Use __GLIBC__ in the hardware disable test to detect when selftests are being built/linked against glibc and thus pthread_attr_setaffinity_np() is (hopefully) available. As pointed out by Sashiko and Hisam, _GNU_SOURCE is effectively a "request" macro to enable functionality, whereas __GLIBC__ is an announcement of support and selftests' idiomatic way of guarding code that's specific to glibc. Fixes: 496779b54943 ("KVM: selftests: Pre-set threads affinity in hardware disable test when possible") Reported-by: Sashiko Bot Closes: https://lore.kernel.org/all/20260731201140.5AF0C1F00AC4@smtp.kernel.org Suggested-by: Hisam Mehboob Link: https://patch.msgid.link/20260826161430.714316-1-seanjc@google.com Signed-off-by: Sean Christopherson --- tools/testing/selftests/kvm/hardware_disable_test.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/tools/testing/selftests/kvm/hardware_disable_test.c b/tools/testing/selftests/kvm/hardware_disable_test.c index 43a36ef3ead8..1c20892d6782 100644 --- a/tools/testing/selftests/kvm/hardware_disable_test.c +++ b/tools/testing/selftests/kvm/hardware_disable_test.c @@ -37,7 +37,7 @@ static void *run_vcpu(void *arg) struct kvm_vcpu *vcpu = arg; struct kvm_run *run = vcpu->run; -#ifndef _GNU_SOURCE +#ifndef __GLIBC__ kvm_sched_setaffinity(0, sizeof(cpu_set_t), &threads_cpu_set); #endif @@ -51,7 +51,7 @@ static void *sleeping_thread(void *arg) { int fd; -#ifndef _GNU_SOURCE +#ifndef __GLIBC__ kvm_sched_setaffinity(0, sizeof(cpu_set_t), &threads_cpu_set); #endif @@ -71,7 +71,7 @@ static void run_test(u32 run) u32 i, j; TEST_ASSERT_EQ(pthread_attr_init(&attr), 0); -#ifdef _GNU_SOURCE +#ifdef __GLIBC__ TEST_ASSERT_EQ(pthread_attr_setaffinity_np(&attr, sizeof(cpu_set_t), &threads_cpu_set), 0); #endif From 8cd280282d1239bca68f8c3632ef1ac8556a396b Mon Sep 17 00:00:00 2001 From: Sean Christopherson Date: Thu, 6 Aug 2026 14:46:18 -0700 Subject: [PATCH 032/365] KVM: Never clear KVM_REQ_VM_DEAD from a vCPU's requests Use kvm_test_request() instead of kvm_check_request() when querying KVM_REQ_VM_DEAD, i.e. don't clear KVM_REQ_VM_DEAD, as the entire purpose of KVM_REQ_VM_DEAD is to prevent the vCPU from enterring the guest ever again, even if userspace insists on redoing KVM_RUN. Ensuring KVM_REQ_VM_DEAD is never cleared will allow relaxing KVM's rule that ioctls can't be invoked on dead VMs, to only disallow ioctls if the VM is bugged, i.e. if KVM hit a KVM_BUG_ON(). Opportunistically add compile-time assertions to guard against clearing KVM_REQ_VM_DEAD through the standard APIs. Reviewed-by: Kai Huang Acked-by: Marc Zyngier Link: https://patch.msgid.link/20260806214618.82180-1-seanjc@google.com Signed-off-by: Sean Christopherson --- arch/arm64/kvm/arm.c | 2 +- arch/x86/kvm/mmu/mmu.c | 2 +- arch/x86/kvm/vmx/tdx.c | 2 +- arch/x86/kvm/x86.c | 2 +- include/linux/kvm_host.h | 9 +++++++-- 5 files changed, 11 insertions(+), 6 deletions(-) diff --git a/arch/arm64/kvm/arm.c b/arch/arm64/kvm/arm.c index 8b080804bc90..62c81d6ae8ee 100644 --- a/arch/arm64/kvm/arm.c +++ b/arch/arm64/kvm/arm.c @@ -1129,7 +1129,7 @@ static int kvm_vcpu_suspend(struct kvm_vcpu *vcpu) static int check_vcpu_requests(struct kvm_vcpu *vcpu) { if (kvm_request_pending(vcpu)) { - if (kvm_check_request(KVM_REQ_VM_DEAD, vcpu)) + if (kvm_test_request(KVM_REQ_VM_DEAD, vcpu)) return -EIO; if (kvm_check_request(KVM_REQ_SLEEP, vcpu)) diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c index 064ecc33b926..8e62476e477b 100644 --- a/arch/x86/kvm/mmu/mmu.c +++ b/arch/x86/kvm/mmu/mmu.c @@ -5058,7 +5058,7 @@ static int kvm_tdp_page_prefault(struct kvm_vcpu *vcpu, gpa_t gpa, if (signal_pending(current)) return -EINTR; - if (kvm_check_request(KVM_REQ_VM_DEAD, vcpu)) + if (kvm_test_request(KVM_REQ_VM_DEAD, vcpu)) return -EIO; cond_resched(); diff --git a/arch/x86/kvm/vmx/tdx.c b/arch/x86/kvm/vmx/tdx.c index b272c20586a7..6c842e9191a5 100644 --- a/arch/x86/kvm/vmx/tdx.c +++ b/arch/x86/kvm/vmx/tdx.c @@ -1995,7 +1995,7 @@ static int tdx_handle_ept_violation(struct kvm_vcpu *vcpu) if (kvm_vcpu_has_events(vcpu) || signal_pending(current)) break; - if (kvm_check_request(KVM_REQ_VM_DEAD, vcpu)) { + if (kvm_test_request(KVM_REQ_VM_DEAD, vcpu)) { ret = -EIO; break; } diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c index 79468ddfe473..a137dc6dd8c6 100644 --- a/arch/x86/kvm/x86.c +++ b/arch/x86/kvm/x86.c @@ -8060,7 +8060,7 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu) bool req_immediate_exit = false; if (kvm_request_pending(vcpu)) { - if (kvm_check_request(KVM_REQ_VM_DEAD, vcpu)) { + if (kvm_test_request(KVM_REQ_VM_DEAD, vcpu)) { r = -EIO; goto out; } diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h index 03bfc92864b6..cf7fe835c4ad 100644 --- a/include/linux/kvm_host.h +++ b/include/linux/kvm_host.h @@ -2324,13 +2324,18 @@ static inline bool kvm_test_request(int req, struct kvm_vcpu *vcpu) return test_bit(req & KVM_REQUEST_MASK, (void *)&vcpu->requests); } -static inline void kvm_clear_request(int req, struct kvm_vcpu *vcpu) +static __always_inline void kvm_clear_request(int req, struct kvm_vcpu *vcpu) { + BUILD_BUG_ON(req == KVM_REQ_VM_DEAD); + clear_bit(req & KVM_REQUEST_MASK, (void *)&vcpu->requests); } -static inline bool kvm_check_request(int req, struct kvm_vcpu *vcpu) +static __always_inline bool kvm_check_request(int req, struct kvm_vcpu *vcpu) { + /* Once a VM is dead, it needs to stay dead. */ + BUILD_BUG_ON(req == KVM_REQ_VM_DEAD); + if (kvm_test_request(req, vcpu)) { kvm_clear_request(req, vcpu); From 77be3641f3e3a56e42a5ed889372ef395a933f3c Mon Sep 17 00:00:00 2001 From: Mikhail Gavrilov Date: Thu, 3 Sep 2026 16:51:35 +0500 Subject: [PATCH 033/365] debugfs: don't warn about uninitialized debugfs for an error parent Since commit c3a280ff728a ("debugfs: warn if file creation failed due to uninitialized debugfs") every boot with CONFIG_REF_TRACKER=y and CONFIG_DEBUG_FS=y prints two errors before the root filesystem is mounted: debugfs: Unable to create file 'net_refcnt@(____ptrval____)', debugfs is not initialized yet debugfs: Unable to create file 'net_notrefcnt@(____ptrval____)', debugfs is not initialized yet Nothing is actually wrong. Both files show up under /sys/kernel/debug/ref_tracker/ once the system is up. The kernel is reporting an error for a condition the caller has already accounted for. net_ns_init() runs directly from start_kernel(), before any initcall, and calls ref_tracker_dir_init() for init_net's two trackers. debugfs_init() is a core_initcall, so debugfs cannot possibly be up at that point. That is by design: ref_tracker_dir_debugfs() is documented as safe to call again later, and net/core/net_namespace.c has a late_initcall() that re-registers both directories once debugfs exists. ref_tracker also states that intent to debugfs. ref_tracker_debug_dir is initialised to ERR_PTR(-ENOENT) and only gets a real dentry in a late_initcall, so the early call hands debugfs_create_file() a parent that is already an error. debugfs_start_creating() honours that and returns the parent error, but only after the new pr_err() has fired. Move the IS_ERR(parent) check above the debugfs_initialized() test. A caller passing an error parent is propagating an earlier failure, which is the pattern debugfs documents and which the warning is not aimed at. A caller passing a valid or NULL parent too early - the case the warning was added for - still gets it. One behaviour change: an early caller with an error parent now gets PTR_ERR(parent) back instead of -ENOENT. All callers of these interfaces are documented to ignore the return value. Fixes: c3a280ff728a ("debugfs: warn if file creation failed due to uninitialized debugfs") Link: https://lore.kernel.org/all/6d1dc775f7d5e754d734907514534054f682bac5.1781171918.git.yk@y-koj.net/ Signed-off-by: Mikhail Gavrilov Tested-by: Yohei Kojima Link: https://patch.msgid.link/20260903115135.63210-1-mikhail.v.gavrilov@gmail.com Signed-off-by: Danilo Krummrich --- fs/debugfs/inode.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/fs/debugfs/inode.c b/fs/debugfs/inode.c index e054e62919ec..a4d08bd3743b 100644 --- a/fs/debugfs/inode.c +++ b/fs/debugfs/inode.c @@ -368,6 +368,9 @@ static struct dentry *debugfs_start_creating(const char *name, if (!debugfs_enabled) return ERR_PTR(-EPERM); + if (IS_ERR(parent)) + return parent; + if (!debugfs_initialized()) { pr_err("Unable to create file '%s', debugfs is not initialized yet\n", name); @@ -376,9 +379,6 @@ static struct dentry *debugfs_start_creating(const char *name, pr_debug("creating file '%s'\n", name); - if (IS_ERR(parent)) - return parent; - error = simple_pin_fs(&debug_fs_type, &debugfs_mount, &debugfs_mount_count); if (error) { From 44b8a0bf5f96e8393312fc34b956766882341f16 Mon Sep 17 00:00:00 2001 From: Runyu Xiao Date: Thu, 27 Aug 2026 16:26:56 +0800 Subject: [PATCH 034/365] mtd: spi-nor: core: Fix mutex leak in spi_nor_rww_start_exclusive() Use the same guard(mutex) pattern as the other RWW helpers so nor->lock is released on both the busy and successful return paths. Fixes: 03e7bb864d9a ("mtd: spi-nor: use scope-based mutex cleanup helpers") Cc: stable@vger.kernel.org Signed-off-by: Runyu Xiao Reviewed-by: Miquel Raynal Reviewed-by: Tudor Ambarus [mw: rephrased commit message] Signed-off-by: Michael Walle --- drivers/mtd/spi-nor/core.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/mtd/spi-nor/core.c b/drivers/mtd/spi-nor/core.c index ccf4396cdcd0..8bc117b46e02 100644 --- a/drivers/mtd/spi-nor/core.c +++ b/drivers/mtd/spi-nor/core.c @@ -1310,7 +1310,7 @@ static bool spi_nor_rww_start_exclusive(struct spi_nor *nor) { struct spi_nor_rww *rww = &nor->rww; - mutex_lock(&nor->lock); + guard(mutex)(&nor->lock); if (rww->ongoing_io || rww->ongoing_rd || rww->ongoing_pe) return false; From aaad136d56d91252517272b68cd533e5714698d5 Mon Sep 17 00:00:00 2001 From: Myeonghun Pak Date: Sat, 1 Aug 2026 01:35:50 +0900 Subject: [PATCH 035/365] RISC-V: KVM: Synchronize hrtimer callback during teardown The non-Sstc hrtimer callback clears next_set before its final uses of the enclosing vCPU. If teardown observes next_set as false while the callback is still running, kvm_riscv_vcpu_timer_cancel() skips hrtimer_cancel() and kvm_destroy_vcpus() can free the vCPU before the callback enters kvm_riscv_vcpu_set_interrupt(). A guest can arm the timer with SBI TIME and request shutdown with SBI legacy shutdown or SRST. A VMM that honors KVM_EXIT_SYSTEM_EVENT and destroys the VM supplies the teardown side of the race; no post-launch host ioctl is needed to arm or request teardown. On upstream master 62cc90241548, generic KASAN reported: BUG: KASAN: slab-use-after-free in do_raw_spin_lock Write of size 4 at addr ff60000005e58898 kvm_riscv_vcpu_set_interrupt kvm_riscv_vcpu_hrtimer_expired __hrtimer_run_queues hrtimer_interrupt The object was allocated by KVM_CREATE_VCPU and freed concurrently by: kvm_destroy_vcpus kvm_arch_destroy_vm kvm_destroy_vm __fput For deterministic validation, I added mdelay(1000) immediately after the existing next_set = false assignment. This only widens the existing post-clear callback window. A no-delay trace build naturally reached the callback-after-teardown-start/before-deinit ordering in 12 of 200 runs, but 1,500 stock-kernel stress iterations did not produce a KASAN report, so natural reproduction is timing-sensitive. Always invoke hrtimer_cancel() for an initialized timer. Preserve the existing -EINVAL result when the timer is no longer set, but only after synchronizing with a running callback. With this patch, hrtimer_cancel() blocked for the full widened callback window before vCPU destruction. KASAN reported no error in 100 fixed-and-widened runs or 200 fix-only timing-sweep runs. Fixes: 3a9f66cb25e1 ("RISC-V: KVM: Add timer functionality") Cc: stable@vger.kernel.org Assisted-by: OpenAI:GPT-5.6 Signed-off-by: Myeonghun Pak Reviewed-by: Anup Patel Link: https://lore.kernel.org/r/20260731163550.46991-1-mhun512@gmail.com Signed-off-by: Anup Patel --- arch/riscv/kvm/vcpu_timer.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/arch/riscv/kvm/vcpu_timer.c b/arch/riscv/kvm/vcpu_timer.c index ae53133c7ab0..a2cd277a4059 100644 --- a/arch/riscv/kvm/vcpu_timer.c +++ b/arch/riscv/kvm/vcpu_timer.c @@ -61,10 +61,13 @@ static enum hrtimer_restart kvm_riscv_vcpu_hrtimer_expired(struct hrtimer *h) static int kvm_riscv_vcpu_timer_cancel(struct kvm_vcpu_timer *t) { - if (!t->init_done || !t->next_set) + if (!t->init_done) return -EINVAL; hrtimer_cancel(&t->hrt); + + if (!t->next_set) + return -EINVAL; t->next_set = false; return 0; From 52c6b7d20d3e791a9e75aa2be2a467990154c7cd Mon Sep 17 00:00:00 2001 From: Yicong Yang Date: Tue, 4 Aug 2026 21:40:18 +0800 Subject: [PATCH 036/365] RISC-V: KVM: Fix the conversion between vsip and hvip Per AIA spec 1.0 Section 6.3.2, the interrupt numbers 13-63 shares same bit position between related VS shadow CSRs and hypervisor CSRs. So there's a shift only for SSI, STI and SEI interrupt. Currently the KVM always does a shift for all the interrupts (include LCOFI with number 13) when doing the conversion between vsip and hvip. Fix this by only doing shift the SSI, STI and SEI. Add wrappers for doing the conversion between vsip and hvip. Fixes: 16b0bde9a37c ("RISC-V: KVM: Add perf sampling support for guests") Signed-off-by: Yicong Yang Reviewed-by: Anup Patel Link: https://lore.kernel.org/r/20260804134018.85497-1-yang.yicong@picoheart.com Signed-off-by: Anup Patel --- arch/riscv/include/asm/csr.h | 20 ++++++++++++++++---- arch/riscv/kvm/vcpu.c | 3 +-- arch/riscv/kvm/vcpu_onereg.c | 8 +++----- tools/arch/riscv/include/asm/csr.h | 20 ++++++++++++++++---- 4 files changed, 36 insertions(+), 15 deletions(-) diff --git a/arch/riscv/include/asm/csr.h b/arch/riscv/include/asm/csr.h index 6c823361be86..621f84d1898f 100644 --- a/arch/riscv/include/asm/csr.h +++ b/arch/riscv/include/asm/csr.h @@ -188,12 +188,24 @@ #define HGATP_MODE_SHIFT HGATP32_MODE_SHIFT #endif -/* VSIP & HVIP relation */ +/* + * VSIP & HVIP relation + * + * The bit positions are same between VSIP and HVIP for interrupt + * numbers 13-63, where there's a shift for the SSI, STI and SEI. + */ #define VSIP_TO_HVIP_SHIFT (IRQ_VS_SOFT - IRQ_S_SOFT) -#define VSIP_VALID_MASK ((_AC(1, UL) << IRQ_S_SOFT) | \ +#define VSIP_BIAS_MASK ((_AC(1, UL) << IRQ_S_SOFT) | \ (_AC(1, UL) << IRQ_S_TIMER) | \ - (_AC(1, UL) << IRQ_S_EXT) | \ - (_AC(1, UL) << IRQ_PMU_OVF)) + (_AC(1, UL) << IRQ_S_EXT)) +#define VSIP_NO_BIAS_MASK (_AC(1, UL) << IRQ_PMU_OVF) +#define VSIP_VALID_MASK (VSIP_BIAS_MASK | VSIP_NO_BIAS_MASK) +#define vsip_to_hvip(_vsip) ((((_vsip) & VSIP_BIAS_MASK) << \ + VSIP_TO_HVIP_SHIFT) | \ + ((_vsip) & VSIP_NO_BIAS_MASK)) +#define hvip_to_vsip(_hvip) ((((_hvip) >> VSIP_TO_HVIP_SHIFT) & \ + VSIP_BIAS_MASK) | \ + ((_hvip) & VSIP_NO_BIAS_MASK)) /* AIA CSR bits */ #define TOPI_IID_SHIFT 16 diff --git a/arch/riscv/kvm/vcpu.c b/arch/riscv/kvm/vcpu.c index e062ca19f9d8..7f1636c44d41 100644 --- a/arch/riscv/kvm/vcpu.c +++ b/arch/riscv/kvm/vcpu.c @@ -491,8 +491,7 @@ bool kvm_riscv_vcpu_has_interrupts(struct kvm_vcpu *vcpu, u64 mask) bool ret; raw_spin_lock_irqsave(&vcpu->arch.irqs_pending_lock, flags); - ie = ((vcpu->arch.guest_csr.vsie & VSIP_VALID_MASK) - << VSIP_TO_HVIP_SHIFT) & (unsigned long)mask; + ie = vsip_to_hvip(vcpu->arch.guest_csr.vsie) & (unsigned long)mask; ie |= vcpu->arch.guest_csr.vsie & ~IRQ_LOCAL_MASK & (unsigned long)mask; ret = vcpu->arch.irqs_pending[0] & ie; diff --git a/arch/riscv/kvm/vcpu_onereg.c b/arch/riscv/kvm/vcpu_onereg.c index 841f2cf87168..f16e25098af1 100644 --- a/arch/riscv/kvm/vcpu_onereg.c +++ b/arch/riscv/kvm/vcpu_onereg.c @@ -272,7 +272,7 @@ static int kvm_riscv_vcpu_general_get_csr(struct kvm_vcpu *vcpu, if (reg_num == KVM_REG_RISCV_CSR_REG(sip)) { kvm_riscv_vcpu_flush_interrupts(vcpu); - *out_val = (csr->hvip >> VSIP_TO_HVIP_SHIFT) & VSIP_VALID_MASK; + *out_val = hvip_to_vsip(csr->hvip); *out_val |= csr->hvip & ~IRQ_LOCAL_MASK; } else *out_val = ((unsigned long *)csr)[reg_num]; @@ -293,10 +293,8 @@ static int kvm_riscv_vcpu_general_set_csr(struct kvm_vcpu *vcpu, reg_num = array_index_nospec(reg_num, regs_max); - if (reg_num == KVM_REG_RISCV_CSR_REG(sip)) { - reg_val &= VSIP_VALID_MASK; - reg_val <<= VSIP_TO_HVIP_SHIFT; - } + if (reg_num == KVM_REG_RISCV_CSR_REG(sip)) + reg_val = vsip_to_hvip(reg_val); ((unsigned long *)csr)[reg_num] = reg_val; diff --git a/tools/arch/riscv/include/asm/csr.h b/tools/arch/riscv/include/asm/csr.h index 21d8cee04638..8df64314d613 100644 --- a/tools/arch/riscv/include/asm/csr.h +++ b/tools/arch/riscv/include/asm/csr.h @@ -163,12 +163,24 @@ #define HGATP_MODE_SHIFT HGATP32_MODE_SHIFT #endif -/* VSIP & HVIP relation */ +/* + * VSIP & HVIP relation + * + * The bit positions are same between VSIP and HVIP for interrupt + * numbers 13-63, where there's a shift for the SSI, STI and SEI. + */ #define VSIP_TO_HVIP_SHIFT (IRQ_VS_SOFT - IRQ_S_SOFT) -#define VSIP_VALID_MASK ((_AC(1, UL) << IRQ_S_SOFT) | \ +#define VSIP_BIAS_MASK ((_AC(1, UL) << IRQ_S_SOFT) | \ (_AC(1, UL) << IRQ_S_TIMER) | \ - (_AC(1, UL) << IRQ_S_EXT) | \ - (_AC(1, UL) << IRQ_PMU_OVF)) + (_AC(1, UL) << IRQ_S_EXT)) +#define VSIP_NO_BIAS_MASK (_AC(1, UL) << IRQ_PMU_OVF) +#define VSIP_VALID_MASK (VSIP_BIAS_MASK | VSIP_NO_BIAS_MASK) +#define vsip_to_hvip(_vsip) ((((_vsip) & VSIP_BIAS_MASK) << \ + VSIP_TO_HVIP_SHIFT) | \ + ((_vsip) & VSIP_NO_BIAS_MASK)) +#define hvip_to_vsip(_hvip) ((((_hvip) >> VSIP_TO_HVIP_SHIFT) & \ + VSIP_BIAS_MASK) | \ + ((_hvip) & VSIP_NO_BIAS_MASK)) /* AIA CSR bits */ #define TOPI_IID_SHIFT 16 From 8ae12ccaec6ec74945d8c1ef39f2c1b8df779abc Mon Sep 17 00:00:00 2001 From: Xie Bo Date: Mon, 10 Aug 2026 09:21:15 +0800 Subject: [PATCH 037/365] RISC-V: KVM: Serialize IMSIC attributes with vCPU migration KVM device ioctls are not serialized against KVM_RUN. As a result, kvm_riscv_aia_imsic_rw_attr() can snapshot the physical CPU and HGEI of an IMSIC VS-file before a concurrent vCPU migration releases it. The HGEI can then be allocated to another vCPU before imsic_vsfile_rw() uses the stale tuple. A GET or SET attribute may consequently access the new owner's interrupt file. Serialize the entire IMSIC attribute operation with the target vCPU mutex. This prevents the VS-file from being migrated and recycled until the attribute access completes. Acquire the mutex killably so that the device ioctl remains interruptible while waiting for KVM_RUN to finish. Fixes: db8b7e97d613 ("RISC-V: KVM: Add in-kernel virtualization of AIA IMSIC") Cc: stable@vger.kernel.org Signed-off-by: Xie Bo Reviewed-by: Anup Patel Link: https://lore.kernel.org/r/20260810-imsic-attr-race-v2-1-00ed95ad321e@ultrarisc.com Signed-off-by: Anup Patel --- arch/riscv/kvm/aia_imsic.c | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) diff --git a/arch/riscv/kvm/aia_imsic.c b/arch/riscv/kvm/aia_imsic.c index c1af23e79ae0..f17dc559a709 100644 --- a/arch/riscv/kvm/aia_imsic.c +++ b/arch/riscv/kvm/aia_imsic.c @@ -965,9 +965,14 @@ int kvm_riscv_aia_imsic_rw_attr(struct kvm *kvm, unsigned long type, if (!vcpu) return -ENODEV; + if (mutex_lock_killable(&vcpu->mutex)) + return -EINTR; + imsic = vcpu->arch.aia_context.imsic_state; - if (!imsic) - return -ENODEV; + if (!imsic) { + rc = -ENODEV; + goto out_unlock; + } isel = KVM_DEV_RISCV_AIA_IMSIC_GET_ISEL(type); read_lock_irqsave(&imsic->vsfile_lock, flags); @@ -991,6 +996,8 @@ int kvm_riscv_aia_imsic_rw_attr(struct kvm *kvm, unsigned long type, rc = imsic_vsfile_rw(vsfile_hgei, vsfile_cpu, imsic->nr_eix, isel, write, val); +out_unlock: + mutex_unlock(&vcpu->mutex); return rc; } From ed54fdb460a6e81d5f8f38388d1f10b385dd4a58 Mon Sep 17 00:00:00 2001 From: Xie Bo Date: Mon, 10 Aug 2026 13:15:43 +0800 Subject: [PATCH 038/365] RISC-V: KVM: Release unused page after MMU invalidation If an MMU invalidation races with a G-stage fault, the fault handler skips installing the page but leaves ret set to zero. As a result, kvm_release_faultin_page() treats the page as used and can unnecessarily mark it dirty. Track the invalidation retry separately and release the page as unused, while preserving the existing return value so that the vCPU retries the fault. Fixes: 2ed90cb0938a ("KVM: RISC-V: Retry fault if vma_lookup() results become invalid") Cc: stable@vger.kernel.org Signed-off-by: Xie Bo Reviewed-by: Anup Patel Link: https://lore.kernel.org/r/20260810051544.3953925-2-xb@ultrarisc.com Signed-off-by: Anup Patel --- arch/riscv/kvm/mmu.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/arch/riscv/kvm/mmu.c b/arch/riscv/kvm/mmu.c index 6035b5ec9503..d189fd58d7bf 100644 --- a/arch/riscv/kvm/mmu.c +++ b/arch/riscv/kvm/mmu.c @@ -627,6 +627,7 @@ int kvm_riscv_mmu_map(struct kvm_vcpu *vcpu, struct kvm_memory_slot *memslot, int ret; kvm_pfn_t hfn; bool is_hugetlb; + bool unused = false; bool writable; unsigned int vma_pageshift; gfn_t gfn = gpa >> PAGE_SHIFT; @@ -719,8 +720,10 @@ int kvm_riscv_mmu_map(struct kvm_vcpu *vcpu, struct kvm_memory_slot *memslot, write_lock(&kvm->mmu_lock); - if (mmu_invalidate_retry(kvm, mmu_seq)) + if (mmu_invalidate_retry(kvm, mmu_seq)) { + unused = true; goto out_unlock; + } /* * Check if we are backed by a THP and thus use block mapping if @@ -743,7 +746,8 @@ int kvm_riscv_mmu_map(struct kvm_vcpu *vcpu, struct kvm_memory_slot *memslot, kvm_err("Failed to map in G-stage\n"); out_unlock: - kvm_release_faultin_page(kvm, page, ret && ret != -EEXIST, writable); + kvm_release_faultin_page(kvm, page, + unused || (ret && ret != -EEXIST), writable); write_unlock(&kvm->mmu_lock); return ret; } From f41fb17143df890855d3de980f8eb92dfd595817 Mon Sep 17 00:00:00 2001 From: Xie Bo Date: Mon, 10 Aug 2026 13:15:44 +0800 Subject: [PATCH 039/365] RISC-V: KVM: Propagate interrupted G-stage faults __kvm_faultin_pfn() reports an interrupted host page fault with KVM_PFN_ERR_SIGPENDING. RISC-V currently handles it as a generic error PFN and returns -EFAULT. Return -EINTR for the signal-pending sentinel so callers can distinguish an interrupted fault from an invalid userspace mapping. Do not log the expected interruption as a vCPU exit error. Fixes: 9d05c1fee837 ("RISC-V: KVM: Implement stage2 page table programming") Cc: stable@vger.kernel.org Signed-off-by: Xie Bo Reviewed-by: Anup Patel Link: https://lore.kernel.org/r/20260810051544.3953925-3-xb@ultrarisc.com Signed-off-by: Anup Patel --- arch/riscv/kvm/mmu.c | 2 ++ arch/riscv/kvm/vcpu_exit.c | 2 +- 2 files changed, 3 insertions(+), 1 deletion(-) diff --git a/arch/riscv/kvm/mmu.c b/arch/riscv/kvm/mmu.c index d189fd58d7bf..3e955d808743 100644 --- a/arch/riscv/kvm/mmu.c +++ b/arch/riscv/kvm/mmu.c @@ -708,6 +708,8 @@ int kvm_riscv_mmu_map(struct kvm_vcpu *vcpu, struct kvm_memory_slot *memslot, vma_pageshift, current); return 0; } + if (is_sigpending_pfn(hfn)) + return -EINTR; if (is_error_noslot_pfn(hfn)) return -EFAULT; diff --git a/arch/riscv/kvm/vcpu_exit.c b/arch/riscv/kvm/vcpu_exit.c index 88e0c369b354..8d36eb8abce1 100644 --- a/arch/riscv/kvm/vcpu_exit.c +++ b/arch/riscv/kvm/vcpu_exit.c @@ -283,7 +283,7 @@ int kvm_riscv_vcpu_exit(struct kvm_vcpu *vcpu, struct kvm_run *run, } /* Print details in-case of error */ - if (ret < 0) { + if (ret < 0 && ret != -EINTR) { kvm_err("VCPU exit error %d\n", ret); kvm_err("SEPC=0x%lx SSTATUS=0x%lx HSTATUS=0x%lx\n", vcpu->arch.guest_context.sepc, From b3d346838ec65fac7fd83f5dbcedd13cadfffddb Mon Sep 17 00:00:00 2001 From: Zongmin Zhou Date: Wed, 26 Aug 2026 15:50:09 +0800 Subject: [PATCH 040/365] KVM: riscv: Fix NACL hfence entry update order The SBI v3.0 specification (section 15.1.2) requires a nested HFENCE entry to be populated as follows: 1) find an unused entry with Config.Pending == 0 2) update the Page_Number and Page_Count words 3) update the Config word with Config.Pending set __kvm_riscv_nacl_hfence() writes the Config word first, so the SBI implementation (or NACL hardware) can observe a pending entry with pnum/pcount values left over from the previous use of that entry, resulting in incorrect TLB flush ranges. Write pnum and pcount first and the Config word last. Since the consumer is an external agent on coherent shared memory, use WRITE_ONCE() to stop the compiler from reordering the stores and smp_wmb() to make the parameter words globally visible before the Pending bit is set. Fixes: d466c19cead5 ("RISC-V: KVM: Add common nested acceleration support") Signed-off-by: Zongmin Zhou Reviewed-by: Anup Patel Link: https://lore.kernel.org/r/20260826075009.68952-1-min_halo@163.com Signed-off-by: Anup Patel --- arch/riscv/kvm/nacl.c | 20 ++++++++++++++++---- 1 file changed, 16 insertions(+), 4 deletions(-) diff --git a/arch/riscv/kvm/nacl.c b/arch/riscv/kvm/nacl.c index 9aff03c4f667..a5cda9a65156 100644 --- a/arch/riscv/kvm/nacl.c +++ b/arch/riscv/kvm/nacl.c @@ -42,12 +42,24 @@ void __kvm_riscv_nacl_hfence(void *shmem, } } - entp = shmem + SBI_NACL_SHMEM_HFENCE_ENTRY_CONFIG(i); - *entp = cpu_to_lelong(control); + /* + * Per SBI v3.0 section 15.1.2, the Page_Number and Page_Count + * words must be updated before the Config word with its Pending + * bit set. WRITE_ONCE() stops the compiler from reordering the + * stores and smp_wmb() makes the parameter words globally + * visible to the SBI implementation (or NACL hardware) before + * the Pending bit is set. + */ entp = shmem + SBI_NACL_SHMEM_HFENCE_ENTRY_PNUM(i); - *entp = cpu_to_lelong(page_num); + WRITE_ONCE(*entp, cpu_to_lelong(page_num)); entp = shmem + SBI_NACL_SHMEM_HFENCE_ENTRY_PCOUNT(i); - *entp = cpu_to_lelong(page_count); + WRITE_ONCE(*entp, cpu_to_lelong(page_count)); + + /* Ensure the parameter words are visible before the Pending bit */ + smp_wmb(); + + entp = shmem + SBI_NACL_SHMEM_HFENCE_ENTRY_CONFIG(i); + WRITE_ONCE(*entp, cpu_to_lelong(control)); } int kvm_riscv_nacl_enable(void) From b7749531a9b195f4fd7db92ca3cc49bda1a4dc8d Mon Sep 17 00:00:00 2001 From: Zongmin Zhou Date: Wed, 26 Aug 2026 14:41:17 +0800 Subject: [PATCH 041/365] RISC-V: KVM: Fix sdata leak and stale snapshot_addr in snapshot_set_shmem A guest may call SBI_PMU_SNAPSHOT_SET_SHMEM repeatedly. Each call overwrites kvpmu->sdata without freeing the old buffer (memory leak), and if a later kvm_vcpu_write_guest() fails, the error path frees sdata but leaves snapshot_addr stale. A subsequent SBI_PMU_COUNTER_START then passes the INVALID_GPA check and crashes the host with a NULL buffer in kvm_vcpu_read_guest(). Fix this by clearing the previously installed snapshot area before installing a new one, which keeps sdata and snapshot_addr consistent. The SBI spec suggests a single invocation but defines no error code for repeated calls, so KVM must tolerate them. Fixes: c2f41ddbcdd756 ("RISC-V: KVM: Implement SBI PMU Snapshot feature") Cc: stable@vger.kernel.org Signed-off-by: Zongmin Zhou Reviewed-by: Anup Patel Link: https://lore.kernel.org/r/20260826064117.58029-1-min_halo@163.com Signed-off-by: Anup Patel --- arch/riscv/kvm/vcpu_pmu.c | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/arch/riscv/kvm/vcpu_pmu.c b/arch/riscv/kvm/vcpu_pmu.c index 6ff741ee7803..cc0c138651bc 100644 --- a/arch/riscv/kvm/vcpu_pmu.c +++ b/arch/riscv/kvm/vcpu_pmu.c @@ -454,6 +454,14 @@ int kvm_riscv_vcpu_pmu_snapshot_set_shmem(struct kvm_vcpu *vcpu, unsigned long s } } + /* + * Clear any previously installed snapshot area to avoid leaking + * the old sdata and to keep sdata/snapshot_addr consistent if + * the re-install fails below. + */ + if (kvpmu->snapshot_addr != INVALID_GPA) + kvm_pmu_clear_snapshot_area(vcpu); + kvpmu->sdata = kzalloc(snapshot_area_size, GFP_ATOMIC | __GFP_ACCOUNT); if (!kvpmu->sdata) { sbiret = SBI_ERR_FAILURE; From 8b3fd1a8b305321171602bfa7c41212441cf69e4 Mon Sep 17 00:00:00 2001 From: SeungJu Cheon Date: Tue, 25 Aug 2026 17:37:17 +0900 Subject: [PATCH 042/365] RISC-V: KVM: Preserve firmware counter value across stop/start Firmware events accumulate in kvpmu->fw_event[].value while running, but counter stop only clears fw_event[].started without saving the value back to pmc->counter_val. A subsequent counter start without SBI_PMU_START_FLAG_SET_INIT_VALUE reloads the stale counter_val into fw_event[].value, losing all events counted so far. Save fw_event[].value into counter_val when actually stopping a running counter, and remove the now redundant synchronization from the snapshot path. Fixes: badc386869e2c ("RISC-V: KVM: Support firmware events") Signed-off-by: SeungJu Cheon Reviewed-by: Anup Patel Link: https://lore.kernel.org/r/20260825083719.643970-2-suunj1331@gmail.com Signed-off-by: Anup Patel --- arch/riscv/kvm/vcpu_pmu.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/arch/riscv/kvm/vcpu_pmu.c b/arch/riscv/kvm/vcpu_pmu.c index cc0c138651bc..5dfe2054d12b 100644 --- a/arch/riscv/kvm/vcpu_pmu.c +++ b/arch/riscv/kvm/vcpu_pmu.c @@ -683,10 +683,12 @@ int kvm_riscv_vcpu_pmu_ctr_stop(struct kvm_vcpu *vcpu, unsigned long ctr_base, goto out; } - if (!kvpmu->fw_event[fevent_code].started) + if (!kvpmu->fw_event[fevent_code].started) { sbiret = SBI_ERR_ALREADY_STOPPED; - - kvpmu->fw_event[fevent_code].started = false; + } else { + kvpmu->fw_event[fevent_code].started = false; + pmc->counter_val = kvpmu->fw_event[fevent_code].value; + } } else if (pmc->perf_event) { if (pmc->started) { /* Stop counting the counter */ @@ -704,9 +706,7 @@ int kvm_riscv_vcpu_pmu_ctr_stop(struct kvm_vcpu *vcpu, unsigned long ctr_base, } if (snap_flag_set && !sbiret) { - if (pmc->cinfo.type == SBI_PMU_CTR_TYPE_FW) - pmc->counter_val = kvpmu->fw_event[fevent_code].value; - else if (pmc->perf_event) + if (pmc->perf_event) pmc->counter_val += perf_event_read_value(pmc->perf_event, &enabled, &running); /* From 057dd2639ceae79adced5d8fe52c32d562edcb3a Mon Sep 17 00:00:00 2001 From: SeungJu Cheon Date: Tue, 25 Aug 2026 17:37:18 +0900 Subject: [PATCH 043/365] RISC-V: KVM: Report snapshot write failure to the guest If kvm_vcpu_write_guest() fails while updating the PMU snapshot area on counter stop, the guest may receive SBI_SUCCESS without the snapshot being updated, leaving stale data in shared memory. Return SBI_ERR_FAILURE when the snapshot write fails. Fixes: c2f41ddbcdd7 ("RISC-V: KVM: Implement SBI PMU Snapshot feature") Signed-off-by: SeungJu Cheon Reviewed-by: Anup Patel Link: https://lore.kernel.org/r/20260825083719.643970-3-suunj1331@gmail.com Signed-off-by: Anup Patel --- arch/riscv/kvm/vcpu_pmu.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/arch/riscv/kvm/vcpu_pmu.c b/arch/riscv/kvm/vcpu_pmu.c index 5dfe2054d12b..8be87f879473 100644 --- a/arch/riscv/kvm/vcpu_pmu.c +++ b/arch/riscv/kvm/vcpu_pmu.c @@ -735,9 +735,10 @@ int kvm_riscv_vcpu_pmu_ctr_stop(struct kvm_vcpu *vcpu, unsigned long ctr_base, } } - if (shmem_needs_update) - kvm_vcpu_write_guest(vcpu, kvpmu->snapshot_addr, kvpmu->sdata, - sizeof(struct riscv_pmu_snapshot_data)); + if (shmem_needs_update && + kvm_vcpu_write_guest(vcpu, kvpmu->snapshot_addr, kvpmu->sdata, + sizeof(struct riscv_pmu_snapshot_data))) + sbiret = SBI_ERR_FAILURE; out: retdata->err_val = sbiret; From c7e2cc38c56142cdab25e6f73602a8222bf9479b Mon Sep 17 00:00:00 2001 From: SeungJu Cheon Date: Tue, 25 Aug 2026 17:37:19 +0900 Subject: [PATCH 044/365] RISC-V: KVM: Fix perf-backed counter accounting across stop and read pmu_ctr_read() adds the event count returned by perf_event_read_value() to counter_val, which can accumulate the same count repeatedly across reads. kvm_riscv_vcpu_pmu_ctr_stop() also leaves counter_val stale by not folding the current event count into it. Make reads of perf-backed counters side-effect free, and use perf_event_pause() when stopping a counter to fold the current event count into counter_val while resetting it. This preserves the counter value across stop/start and lets the snapshot path use counter_val directly. Fixes: 0cb74b65d2e5 ("RISC-V: KVM: Implement perf support without sampling") Signed-off-by: SeungJu Cheon Reviewed-by: Anup Patel Link: https://lore.kernel.org/r/20260825083719.643970-4-suunj1331@gmail.com Signed-off-by: Anup Patel --- arch/riscv/kvm/vcpu_pmu.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) diff --git a/arch/riscv/kvm/vcpu_pmu.c b/arch/riscv/kvm/vcpu_pmu.c index 8be87f879473..e095d1ff439f 100644 --- a/arch/riscv/kvm/vcpu_pmu.c +++ b/arch/riscv/kvm/vcpu_pmu.c @@ -270,12 +270,13 @@ static int pmu_ctr_read(struct kvm_vcpu *vcpu, unsigned long cidx, return -EINVAL; pmc->counter_val = kvpmu->fw_event[fevent_code].value; + *out_val = pmc->counter_val; } else if (pmc->perf_event) { - pmc->counter_val += perf_event_read_value(pmc->perf_event, &enabled, &running); + *out_val = pmc->counter_val + + perf_event_read_value(pmc->perf_event, &enabled, &running); } else { return -EINVAL; } - *out_val = pmc->counter_val; return 0; } @@ -653,7 +654,6 @@ int kvm_riscv_vcpu_pmu_ctr_stop(struct kvm_vcpu *vcpu, unsigned long ctr_base, { struct kvm_pmu *kvpmu = vcpu_to_pmu(vcpu); int i, pmc_index, sbiret = 0; - u64 enabled, running; struct kvm_pmc *pmc; int fevent_code; bool snap_flag_set = flags & SBI_PMU_STOP_FLAG_TAKE_SNAPSHOT; @@ -691,8 +691,11 @@ int kvm_riscv_vcpu_pmu_ctr_stop(struct kvm_vcpu *vcpu, unsigned long ctr_base, } } else if (pmc->perf_event) { if (pmc->started) { - /* Stop counting the counter */ - perf_event_disable(pmc->perf_event); + /* + * Stop the counter and fold the live count into counter_val. + * Reset the event value to avoid redundant accumulation. + */ + pmc->counter_val += perf_event_pause(pmc->perf_event, true); pmc->started = false; } else { sbiret = SBI_ERR_ALREADY_STOPPED; @@ -706,9 +709,6 @@ int kvm_riscv_vcpu_pmu_ctr_stop(struct kvm_vcpu *vcpu, unsigned long ctr_base, } if (snap_flag_set && !sbiret) { - if (pmc->perf_event) - pmc->counter_val += perf_event_read_value(pmc->perf_event, - &enabled, &running); /* * The counter and overflow indices in the snapshot region are w.r.to * cbase. Modify the set bit in the counter mask instead of the pmc_index From 41e81f7e3ef96594fb840445343c0ee7723aa550 Mon Sep 17 00:00:00 2001 From: Tan Chi Date: Mon, 14 Sep 2026 11:11:46 +0800 Subject: [PATCH 045/365] RISC-V: KVM: Fix HSM hart status error propagation kvm_sbi_hsm_vcpu_get_status() returns SBI_ERR_INVALID_PARAM when the requested hart does not exist. However, the HART_STATUS case returns from the SBI handler without storing this error in retdata->err_val. As a result, a guest querying the status of a non-existent hart observes SBI_SUCCESS instead of SBI_ERR_INVALID_PARAM. Use the common SBI error handling path for HART_STATUS after saving a valid hart state in retdata->out_val. This preserves the returned error when kvm_sbi_hsm_vcpu_get_status() fails. Fixes: bae0dfd74e01 ("RISC-V: KVM: Modify SBI extension handler to return SBI error code") Cc: stable@vger.kernel.org Assisted-by: LLM Signed-off-by: Tan Chi Reviewed-by: Anup Patel Link: https://lore.kernel.org/r/20260914031146.446157-1-tanchi25@mails.ucas.ac.cn Signed-off-by: Anup Patel --- arch/riscv/kvm/vcpu_sbi_hsm.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/arch/riscv/kvm/vcpu_sbi_hsm.c b/arch/riscv/kvm/vcpu_sbi_hsm.c index f26207f84bab..06a15629c26b 100644 --- a/arch/riscv/kvm/vcpu_sbi_hsm.c +++ b/arch/riscv/kvm/vcpu_sbi_hsm.c @@ -95,9 +95,9 @@ static int kvm_sbi_ext_hsm_handler(struct kvm_vcpu *vcpu, struct kvm_run *run, ret = kvm_sbi_hsm_vcpu_get_status(vcpu); if (ret >= 0) { retdata->out_val = ret; - retdata->err_val = 0; + ret = 0; } - return 0; + break; case SBI_EXT_HSM_HART_SUSPEND: switch (lower_32_bits(cp->a0)) { case SBI_HSM_SUSPEND_RET_DEFAULT: From 8cd92f77ae4f5371a7d581f8324c24919670b304 Mon Sep 17 00:00:00 2001 From: Fuad Tabba Date: Fri, 21 Aug 2026 07:44:42 +0100 Subject: [PATCH 046/365] KVM: arm64: vgic-its: Free the caches when GITS_BASER changes A guest that disables the ITS and re-points or shrinks GITS_BASER with VALID still set keeps the devices and collections it mapped against the old table, as KVM frees them only when VALID is cleared. The contents of the table are IMPLEMENTATION DEFINED, so a write that gives GITS_BASER a different address or size may lose whatever the old value described. Free the list whenever the stored value changes, and drop the translation cache with it. The cache is not empty just because the ITS is disabled: its->enabled is written under the cmd_lock, while vgic_its_resolve_lpi() tests it under the its_lock, so an injection can still cache an entry after the ITS was disabled. Hence the invalidation inside the its_lock section. Test for a change rather than a write: its_restore_enable() rewrites GITS_BASER from its probe-time cache on resume, and KVM reports GITS_TYPER.HCC as 0, so nothing re-maps the boot CPU's collection afterwards. Fixes: 36d6961c2b481 ("KVM: arm/arm64: vgic-its: Free caches when GITS_BASER Valid bit is cleared") Suggested-by: Marc Zyngier Link: https://lore.kernel.org/all/87ecg9owwa.wl-maz@kernel.org/ Signed-off-by: Fuad Tabba Reviewed-by: Marc Zyngier Link: https://patch.msgid.link/20260821064445.615838-2-fuad.tabba@linux.dev Signed-off-by: Oliver Upton --- arch/arm64/kvm/vgic/vgic-its.c | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-) diff --git a/arch/arm64/kvm/vgic/vgic-its.c b/arch/arm64/kvm/vgic/vgic-its.c index 9e782a4fea7e..ab89b0138efd 100644 --- a/arch/arm64/kvm/vgic/vgic-its.c +++ b/arch/arm64/kvm/vgic/vgic-its.c @@ -1658,7 +1658,7 @@ static void vgic_mmio_write_its_baser(struct kvm *kvm, unsigned long val) { const struct vgic_its_abi *abi = vgic_its_get_abi(its); - u64 entry_size, table_type; + u64 old, entry_size, table_type; u64 reg, *regptr, clearbits = 0; /* When GITS_CTLR.Enable is 1, we ignore write accesses. */ @@ -1681,7 +1681,9 @@ static void vgic_mmio_write_its_baser(struct kvm *kvm, return; } - reg = update_64bit_reg(*regptr, addr & 7, len, val); + old = *regptr; + + reg = update_64bit_reg(old, addr & 7, len, val); reg &= ~GITS_BASER_RO_MASK; reg &= ~clearbits; @@ -1691,7 +1693,8 @@ static void vgic_mmio_write_its_baser(struct kvm *kvm, *regptr = reg; - if (!(reg & GITS_BASER_VALID)) { + /* The ITS driver rewrites an unchanged GITS_BASER on resume. */ + if (reg != old) { /* Take the its_lock to prevent a race with a save/restore */ mutex_lock(&its->its_lock); switch (table_type) { @@ -1702,6 +1705,8 @@ static void vgic_mmio_write_its_baser(struct kvm *kvm, vgic_its_free_collection_list(kvm, its); break; } + /* A concurrent injection may have cached a translation. */ + vgic_its_invalidate_cache(its); mutex_unlock(&its->its_lock); } } From 30908e7272479b6453745022abb9f2eb9c9933f7 Mon Sep 17 00:00:00 2001 From: Fuad Tabba Date: Fri, 21 Aug 2026 07:44:43 +0100 Subject: [PATCH 047/365] Revert "KVM: arm64: vgic-its: Don't save collections the table cannot hold" This reverts commit 9b10fb74e4b661543d188701bd4d024fc5c18f58. Freeing the collections when GITS_BASER changes removes the state this check rejected: vgic_its_cmd_handle_mapi(), vgic_its_cmd_handle_mapc() and vgic_its_restore_cte() all validate the ID against the current table before allocating, and the table can no longer change under the list. What remains is a collection whose entry is not backed by a memslot, which the write fails on anyway, so the check costs a save userspace should be able to issue reliably and buys nothing. The reverted commit credited the check with bounding the walk as well. It stays bounded without it: collection IDs are unique and each is below the table's capacity, so the list cannot be longer than the table. Suggested-by: Marc Zyngier Link: https://lore.kernel.org/all/87ecg9owwa.wl-maz@kernel.org/ Signed-off-by: Fuad Tabba Reviewed-by: Marc Zyngier Link: https://patch.msgid.link/20260821064445.615838-3-fuad.tabba@linux.dev Signed-off-by: Oliver Upton --- arch/arm64/kvm/vgic/vgic-its.c | 3 --- 1 file changed, 3 deletions(-) diff --git a/arch/arm64/kvm/vgic/vgic-its.c b/arch/arm64/kvm/vgic/vgic-its.c index ab89b0138efd..313bf9e802bf 100644 --- a/arch/arm64/kvm/vgic/vgic-its.c +++ b/arch/arm64/kvm/vgic/vgic-its.c @@ -2546,9 +2546,6 @@ static int vgic_its_save_collection_table(struct vgic_its *its) max_size = GITS_BASER_NR_PAGES(baser) * SZ_64K; list_for_each_entry(collection, &its->collection_list, coll_list) { - if (!vgic_its_check_id(its, baser, collection->collection_id, NULL)) - return -EINVAL; - ret = vgic_its_save_cte(its, collection, gpa); if (ret) return ret; From cc5d96036e01ac330d24b2f0c336d60f82ab4930 Mon Sep 17 00:00:00 2001 From: Fuad Tabba Date: Fri, 21 Aug 2026 07:44:44 +0100 Subject: [PATCH 048/365] KVM: arm64: vgic-its: Skip unreachable devices instead of failing the save vgic_its_save_device_tables() aborts with -EINVAL when a device's entry falls outside the device table, which a guest can arrange on its own: an indirect table lets it clear an L1 entry's valid bit without touching GITS_BASER. That fails a save userspace should be able to issue reliably. Skip the device instead, and point the saved DTE chain past it, as commit ad1e686e2378d ("KVM: arm64: vgic-its: Point saved ITEs at the next valid entry") does for ITEs. compute_next_devid_offset() takes the next device off the list whether or not it was saved, so the predecessor would otherwise point at an entry the save never wrote. Restore follows that offset while it stays inside the table being scanned: within an L2 block, or anywhere in a flat table. Both need userspace to remove a memslot under the table, since dropping an L1 entry takes the whole block with it and scan_its_table() stops at the block boundary. Fixes: 57a9a117154c9 ("KVM: arm64: vgic-its: Device table save/restore") Suggested-by: Marc Zyngier Link: https://lore.kernel.org/all/86bjaz5s6v.wl-maz@kernel.org/ Signed-off-by: Fuad Tabba Reviewed-by: Marc Zyngier Link: https://patch.msgid.link/20260821064445.615838-4-fuad.tabba@linux.dev Signed-off-by: Oliver Upton --- arch/arm64/kvm/vgic/vgic-its.c | 32 +++++++++++++++++++------------- 1 file changed, 19 insertions(+), 13 deletions(-) diff --git a/arch/arm64/kvm/vgic/vgic-its.c b/arch/arm64/kvm/vgic/vgic-its.c index 313bf9e802bf..0904ae850c35 100644 --- a/arch/arm64/kvm/vgic/vgic-its.c +++ b/arch/arm64/kvm/vgic/vgic-its.c @@ -2024,18 +2024,22 @@ static int vgic_its_attr_regs_access(struct kvm_device *dev, return ret; } -static u32 compute_next_devid_offset(struct list_head *h, +static u32 compute_next_devid_offset(struct vgic_its *its, u64 baser, struct its_device *dev) { - struct its_device *next; - u32 next_offset; + struct its_device *next = dev; - if (list_is_last(&dev->dev_list, h)) - return 0; - next = list_next_entry(dev, dev_list); - next_offset = next->device_id - dev->device_id; + /* + * Point at the next device vgic_its_save_device_tables() saves. It + * sorts device_list first, so the subtraction cannot underflow. + */ + list_for_each_entry_continue(next, &its->device_list, dev_list) { + if (vgic_its_check_id(its, baser, next->device_id, NULL)) + return min_t(u32, next->device_id - dev->device_id, + VITS_DTE_MAX_DEVID_OFFSET); + } - return min_t(u32, next_offset, VITS_DTE_MAX_DEVID_OFFSET); + return 0; } static u32 compute_next_eventid_offset(struct list_head *h, struct its_ite *ite) @@ -2276,17 +2280,18 @@ static int vgic_its_restore_itt(struct vgic_its *its, struct its_device *dev) * vgic_its_save_dte - Save a device table entry at a given GPA * * @its: ITS handle + * @baser: GITS_BASER the caller is saving against * @dev: ITS device * @ptr: GPA */ -static int vgic_its_save_dte(struct vgic_its *its, struct its_device *dev, - gpa_t ptr) +static int vgic_its_save_dte(struct vgic_its *its, u64 baser, + struct its_device *dev, gpa_t ptr) { u64 val, itt_addr_field; u32 next_offset; itt_addr_field = dev->itt_addr >> 8; - next_offset = compute_next_devid_offset(&its->device_list, dev); + next_offset = compute_next_devid_offset(its, baser, dev); val = (1ULL << KVM_ITS_DTE_VALID_SHIFT | ((u64)next_offset << KVM_ITS_DTE_NEXT_SHIFT) | (itt_addr_field << KVM_ITS_DTE_ITTADDR_SHIFT) | @@ -2385,15 +2390,16 @@ static int vgic_its_save_device_tables(struct vgic_its *its) int ret; gpa_t eaddr; + /* Don't fail a save that userspace must be able to issue. */ if (!vgic_its_check_id(its, baser, dev->device_id, &eaddr)) - return -EINVAL; + continue; ret = vgic_its_save_itt(its, dev); if (ret) return ret; - ret = vgic_its_save_dte(its, dev, eaddr); + ret = vgic_its_save_dte(its, baser, dev, eaddr); if (ret) return ret; } From 6f182db39fb00548c26d689163ceb2fe4a802793 Mon Sep 17 00:00:00 2001 From: Fuad Tabba Date: Fri, 21 Aug 2026 07:44:45 +0100 Subject: [PATCH 049/365] KVM: arm64: selftests: Add ITS table save tests Cover the two ways a guest can leave a table that KVM_DEV_ARM_ITS_SAVE_TABLES has to cope with: a GITS_BASER write that changes the table, which drops the collections it described, and a device whose L2 block the guest invalidated, which the save skips. Each case then resets and restores, which is what the save exists for. Both fail without the preceding patches. The first save returns -EINVAL, and the second saves a DTE pointing 8192 entries ahead, at an entry the save never wrote. Assisted-by: Antigravity:gemini-3.1-pro Signed-off-by: Fuad Tabba Reviewed-by: Marc Zyngier Link: https://patch.msgid.link/20260821064445.615838-5-fuad.tabba@linux.dev Signed-off-by: Oliver Upton --- tools/testing/selftests/kvm/Makefile.kvm | 1 + .../selftests/kvm/arm64/vgic_its_save.c | 441 ++++++++++++++++++ 2 files changed, 442 insertions(+) create mode 100644 tools/testing/selftests/kvm/arm64/vgic_its_save.c diff --git a/tools/testing/selftests/kvm/Makefile.kvm b/tools/testing/selftests/kvm/Makefile.kvm index 96bab7002d39..6a1482e3a286 100644 --- a/tools/testing/selftests/kvm/Makefile.kvm +++ b/tools/testing/selftests/kvm/Makefile.kvm @@ -189,6 +189,7 @@ TEST_GEN_PROGS_arm64 += arm64/stage2_block_transitions TEST_GEN_PROGS_arm64 += arm64/vcpu_width_config TEST_GEN_PROGS_arm64 += arm64/vgic_init TEST_GEN_PROGS_arm64 += arm64/vgic_irq +TEST_GEN_PROGS_arm64 += arm64/vgic_its_save TEST_GEN_PROGS_arm64 += arm64/vgic_lpi_stress TEST_GEN_PROGS_arm64 += arm64/vgic_v5 TEST_GEN_PROGS_arm64 += arm64/vpmu_counter_access diff --git a/tools/testing/selftests/kvm/arm64/vgic_its_save.c b/tools/testing/selftests/kvm/arm64/vgic_its_save.c new file mode 100644 index 000000000000..864da01539f3 --- /dev/null +++ b/tools/testing/selftests/kvm/arm64/vgic_its_save.c @@ -0,0 +1,441 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* + * vgic_its_save - KVM_DEV_ARM_ITS_SAVE_TABLES against tables a guest broke. + * + * Both cases are reachable by a guest on its own, and neither may fail a save + * that userspace has to be able to issue: + * + * - Changing GITS_BASER drops the collections it described, so the save + * writes nothing but the terminating invalid entry. + * - A device the device table can no longer address is skipped, and the saved + * DTE chain skips it too rather than pointing at an entry never written. + * + * Both cases then reset and restore, which is what the save exists for. + * + * Copyright (c) 2026 Google LLC + * Author: Fuad Tabba + */ + +#include +#include +#include +#include + +#include "kvm_util.h" +#include "gic.h" +#include "gic_v3.h" +#include "gic_v3_its.h" +#include "processor.h" +#include "ucall.h" +#include "vgic.h" + +#define TEST_MEMSLOT_INDEX 1 + +/* All three ITS table entry sizes are 8 bytes in ABI 0. */ +#define ESZ 8 +#define ENTRIES_PER_PAGE (SZ_64K / ESZ) + +/* CTE and DTE layout, mirroring KVM's KVM_ITS_* in arch/arm64/kvm/vgic/vgic.h */ +#define CTE_VALID_MASK BIT_ULL(63) +#define DTE_VALID_MASK BIT_ULL(63) +#define DTE_NEXT_SHIFT 49 +#define DTE_NEXT_MASK GENMASK_ULL(62, 49) + +/* L1 entry of an indirect table: valid bit plus a 64K aligned L2 address. */ +#define L1E_VALID_MASK BIT_ULL(63) +#define L1E_ADDR_MASK GENMASK_ULL(51, 16) + +#define GITS_BASER_PAGES_MASK GENMASK_ULL(7, 0) + +#define POISON 0xdeadbeefdeadbeefULL + +/* The collection table starts at two pages and is shrunk to one. */ +#define COLL_TBL_PAGES 2 +#define COLL_TBL_SZ (COLL_TBL_PAGES * SZ_64K) + +/* One more collection than the shrunken table can hold. */ +#define NR_COLLECTIONS (ENTRIES_PER_PAGE + 1) + +/* Two devices, one per L2 block of the indirect device table. */ +#define DEVICE_A_ID 0 +#define DEVICE_B_ID ENTRIES_PER_PAGE + +/* + * its_send_mapd_cmd() encodes ilog2(itt_size) - 1 as num_eventid_bits, and + * vgic_its_restore_itt() scans BIT_ULL(num_eventid_bits) * ESZ, so the size + * handed to MAPD has to match the ITT allocated for it. + */ +#define ITT_EVENTID_BITS 13 +#define ITT_MAPD_SIZE BIT_ULL(ITT_EVENTID_BITS + 1) +#define ITT_SZ (BIT_ULL(ITT_EVENTID_BITS) * ESZ) + +static struct kvm_vm *vm; +static struct kvm_vcpu *vcpu; +static int its_fd; +static gpa_t gpa_base; + +static struct test_data { + gpa_t device_table; + gpa_t collection_table; + gpa_t cmdq_base; + void *cmdq_base_va; + + gpa_t lpi_prop_table; + gpa_t lpi_pend_table; + + void *device_l1_va; + gpa_t device_l2[2]; + gpa_t itt_tables; +} test_data; + +static unsigned long its_baser_offset(unsigned int type) +{ + int i; + + for (i = 0; i < GITS_BASER_NR_REGS; i++) { + unsigned long offset = GITS_BASER + (i * sizeof(u64)); + u64 baser = readq_relaxed(GITS_BASE_GVA + offset); + + if (GITS_BASER_TYPE(baser) == type) + return offset; + } + + GUEST_FAIL("Couldn't find an ITS BASER of type %u", type); + return -1; +} + +static void its_set_enable(bool enable) +{ + u32 ctlr = readl_relaxed(GITS_BASE_GVA + GITS_CTLR); + + if (enable) + ctlr |= GITS_CTLR_ENABLE; + else + ctlr &= ~GITS_CTLR_ENABLE; + + writel_relaxed(ctlr, GITS_BASE_GVA + GITS_CTLR); +} + +/* + * Shrink the collection table to a single page, leaving VALID set. BASER + * writes are ignored while the ITS is enabled. + */ +static void guest_shrink_coll_table(void) +{ + unsigned long offset = its_baser_offset(GITS_BASER_TYPE_COLLECTION); + u64 baser; + + its_set_enable(false); + + baser = readq_relaxed(GITS_BASE_GVA + offset); + baser &= ~GITS_BASER_PAGES_MASK; + writeq_relaxed(baser, GITS_BASE_GVA + offset); +} + +static void guest_baser_change(void) +{ + u32 coll_id; + + gic_init(GIC_V3, 1); + gic_rdist_enable_lpis(test_data.lpi_prop_table, SZ_64K, + test_data.lpi_pend_table); + + its_init(test_data.collection_table, COLL_TBL_SZ, + test_data.device_table, SZ_64K, + test_data.cmdq_base, SZ_64K); + + for (coll_id = 0; coll_id < NR_COLLECTIONS; coll_id++) + its_send_mapc_cmd(test_data.cmdq_base_va, 0, coll_id, true); + + guest_shrink_coll_table(); + + GUEST_DONE(); +} + +/* Turn the already installed device table into an indirect one. */ +static void guest_make_device_table_indirect(void) +{ + unsigned long offset = its_baser_offset(GITS_BASER_TYPE_DEVICE); + u64 baser; + + its_set_enable(false); + + baser = readq_relaxed(GITS_BASE_GVA + offset); + writeq_relaxed(baser | GITS_BASER_INDIRECT, GITS_BASE_GVA + offset); + + its_set_enable(true); +} + +static void guest_unreachable_device(void) +{ + u64 *l1; + + gic_init(GIC_V3, 1); + gic_rdist_enable_lpis(test_data.lpi_prop_table, SZ_64K, + test_data.lpi_pend_table); + + its_init(test_data.collection_table, SZ_64K, + test_data.device_table, SZ_64K, + test_data.cmdq_base, SZ_64K); + + guest_make_device_table_indirect(); + + /* Both L2 blocks present, so both MAPDs are in range. */ + l1 = test_data.device_l1_va; + l1[0] = L1E_VALID_MASK | (test_data.device_l2[0] & L1E_ADDR_MASK); + l1[1] = L1E_VALID_MASK | (test_data.device_l2[1] & L1E_ADDR_MASK); + + its_send_mapd_cmd(test_data.cmdq_base_va, DEVICE_A_ID, + test_data.itt_tables, ITT_MAPD_SIZE, true); + its_send_mapd_cmd(test_data.cmdq_base_va, DEVICE_B_ID, + test_data.itt_tables + ITT_SZ, ITT_MAPD_SIZE, true); + + /* + * Drop the block holding device B. No ITS command and no GITS_BASER + * write is involved, so nothing tells KVM the device is now + * unreachable. + */ + l1[1] = 0; + + GUEST_DONE(); +} + +static void run_guest(void) +{ + struct ucall uc; + + vcpu_run(vcpu); + switch (get_ucall(vcpu, &uc)) { + case UCALL_DONE: + break; + case UCALL_ABORT: + REPORT_GUEST_ASSERT(uc); + break; + default: + TEST_FAIL("Unexpected ucall: %lu", uc.cmd); + } +} + +static int save_tables(void) +{ + return __kvm_device_attr_set(its_fd, KVM_DEV_ARM_VGIC_GRP_CTRL, + KVM_DEV_ARM_ITS_SAVE_TABLES, NULL); +} + +static u64 its_reg_get(unsigned long offset) +{ + u64 val; + + kvm_device_attr_get(its_fd, KVM_DEV_ARM_VGIC_GRP_ITS_REGS, offset, + &val); + return val; +} + +static void its_reg_set(unsigned long offset, u64 val) +{ + kvm_device_attr_set(its_fd, KVM_DEV_ARM_VGIC_GRP_ITS_REGS, offset, + &val); +} + +/* + * What a migration target does with the saved tables, in the order + * Documentation/virt/kvm/devices/arm-vgic-its.rst gives: the GITS_ registers + * first, then the tables. The reset in between clears GITS_BASER.Valid, + * which is why the registers have to be written back before the restore. + */ +static void reset_and_restore_tables(void) +{ + u64 baser[GITS_BASER_NR_REGS]; + int ret, i; + + for (i = 0; i < GITS_BASER_NR_REGS; i++) + baser[i] = its_reg_get(GITS_BASER + (i * sizeof(u64))); + + ret = __kvm_device_attr_set(its_fd, KVM_DEV_ARM_VGIC_GRP_CTRL, + KVM_DEV_ARM_ITS_CTRL_RESET, NULL); + TEST_ASSERT(!ret, "Expected the reset to succeed, got ret %d errno %d", + ret, errno); + + for (i = 0; i < GITS_BASER_NR_REGS; i++) + its_reg_set(GITS_BASER + (i * sizeof(u64)), baser[i]); + + ret = __kvm_device_attr_set(its_fd, KVM_DEV_ARM_VGIC_GRP_CTRL, + KVM_DEV_ARM_ITS_RESTORE_TABLES, NULL); + TEST_ASSERT(!ret, "Expected the restore to succeed, got ret %d errno %d", + ret, errno); +} + +static void poison_range(gpa_t base, size_t size) +{ + u64 *entry = addr_gpa2hva(vm, base); + size_t i; + + for (i = 0; i < size / ESZ; i++) + entry[i] = POISON; +} + +static void setup_memslot(size_t sz) +{ + size_t pages = sz / vm->page_size; + + gpa_base = ((vm_compute_max_gfn(vm) + 1) * vm->page_size) - sz; + vm_userspace_mem_region_add(vm, VM_MEM_SRC_ANONYMOUS, gpa_base, + TEST_MEMSLOT_INDEX, pages, 0); +} + +static gpa_t alloc_64k(size_t nr) +{ + size_t pages_per_64k = vm_calc_num_guest_pages(vm->mode, SZ_64K); + gpa_t gpa = vm_phy_pages_alloc(vm, nr * pages_per_64k, gpa_base, + TEST_MEMSLOT_INDEX); + + TEST_ASSERT(IS_ALIGNED(gpa, SZ_64K), + "Allocation at 0x%lx is not 64K aligned, GITS_BASER cannot address it", + gpa); + return gpa; +} + +static void map_to_guest(gpa_t gpa, size_t nr) +{ + size_t pages_per_64k = vm_calc_num_guest_pages(vm->mode, SZ_64K); + + virt_map(vm, gpa, gpa, nr * pages_per_64k); +} + +static void setup_common(void) +{ + test_data.cmdq_base = alloc_64k(1); + map_to_guest(test_data.cmdq_base, 1); + test_data.cmdq_base_va = (void *)test_data.cmdq_base; + + test_data.lpi_prop_table = alloc_64k(1); + test_data.lpi_pend_table = alloc_64k(1); +} + +static void teardown(void) +{ + close(its_fd); + kvm_vm_free(vm); + memset(&test_data, 0, sizeof(test_data)); +} + +/* + * A GITS_BASER write that changes the table drops the collections it + * described. The save then has an empty list, so it writes the terminating + * invalid entry and nothing else. + */ +static void test_baser_change_drops_collections(void) +{ + u64 *cte; + int ret, i; + + pr_info("Testing that a GITS_BASER change drops the collections\n"); + + vm = vm_create_with_one_vcpu(&vcpu, guest_baser_change); + setup_memslot((4 + COLL_TBL_PAGES) * SZ_64K); + its_fd = vgic_its_setup(vm); + + test_data.device_table = alloc_64k(1); + test_data.collection_table = alloc_64k(COLL_TBL_PAGES); + setup_common(); + + sync_global_to_guest(vm, test_data); + run_guest(); + + /* Anything KVM writes is then the only thing that changed. */ + poison_range(test_data.collection_table, COLL_TBL_SZ); + + ret = save_tables(); + TEST_ASSERT(!ret, "Expected the save to succeed, got %d errno %d", + ret, errno); + + cte = addr_gpa2hva(vm, test_data.collection_table); + + /* + * Finding the terminator at the head of the table is also what proves + * the reads below landed in the saved table rather than elsewhere. + */ + TEST_ASSERT(le64toh(cte[0]) == 0, + "CTE 0: expected the terminating invalid entry, got 0x%llx", + (unsigned long long)le64toh(cte[0])); + + for (i = 1; i < COLL_TBL_SZ / ESZ; i++) + TEST_ASSERT(cte[i] == POISON, + "CTE %d: expected it untouched, got 0x%llx", + i, (unsigned long long)cte[i]); + + reset_and_restore_tables(); + + teardown(); +} + +/* + * A device whose L2 block the guest dropped is skipped by the save, and the + * DTE chain skips it too: left alone, the surviving device would point at an + * entry the save never wrote. + */ +static void test_unreachable_device_skipped(void) +{ + u64 dte; + int ret; + + pr_info("Testing that an unreachable device is skipped by the save\n"); + + vm = vm_create_with_one_vcpu(&vcpu, guest_unreachable_device); + setup_memslot(9 * SZ_64K); + its_fd = vgic_its_setup(vm); + + test_data.device_table = alloc_64k(1); + test_data.collection_table = alloc_64k(1); + test_data.device_l2[0] = alloc_64k(1); + test_data.device_l2[1] = alloc_64k(1); + test_data.itt_tables = alloc_64k(2); + setup_common(); + + map_to_guest(test_data.device_table, 1); + test_data.device_l1_va = (void *)test_data.device_table; + + sync_global_to_guest(vm, test_data); + run_guest(); + + poison_range(test_data.device_l2[0], SZ_64K); + poison_range(test_data.device_l2[1], SZ_64K); + + ret = save_tables(); + TEST_ASSERT(!ret, "Expected the save to succeed, got %d errno %d", + ret, errno); + + dte = le64toh(*(u64 *)addr_gpa2hva(vm, test_data.device_l2[0])); + + /* Device A is still reachable, so it is saved. */ + TEST_ASSERT(dte & DTE_VALID_MASK, + "Device A: expected a valid DTE, got 0x%llx", + (unsigned long long)dte); + + /* + * Device B is the only device after it and was skipped, so nothing + * follows A in the saved chain. + */ + TEST_ASSERT(FIELD_GET(DTE_NEXT_MASK, dte) == 0, + "Device A: expected no next device, got offset %llu", + (unsigned long long)FIELD_GET(DTE_NEXT_MASK, dte)); + + /* And nothing was written into the block the guest dropped. */ + TEST_ASSERT(*(u64 *)addr_gpa2hva(vm, test_data.device_l2[1]) == POISON, + "Device B: expected its entry untouched"); + + reset_and_restore_tables(); + + teardown(); +} + +int main(void) +{ + TEST_REQUIRE(kvm_supports_vgic_v3()); + + test_baser_change_drops_collections(); + test_unreachable_device_skipped(); + + pr_info("All ok!\n"); + return 0; +} From 38b70fc453c3112f1a62583b89903ae41116cc27 Mon Sep 17 00:00:00 2001 From: "Lorenzo Stoakes (ARM)" Date: Tue, 1 Sep 2026 18:28:59 +0100 Subject: [PATCH 050/365] KVM: arm64: Fix spurious warning for benign stage 2 teardown race kvmtool was used to establish an L1 guest with 8 CPUs and 8 GiB of RAM, an L2 guest with 4 CPUs and 4 GiB of RAM and an L3 guest with 2 CPUs and 2 GiB of RAM, all of which was then exited. Under memory pressure in the L0 host warnings were observed due to migration triggered by compaction: WARNING: arch/arm64/kvm/mmu.c:336 at __unmap_stage2_range+0x64/0x80, CPU#5: kcompactd0/66 Which was, in turn, triggered by an MMU notifier for the host invalidation: mmu_notifier_invalidate_range_start() -> ... -> kvm_mmu_notifier_invalidate_range_start() -> kvm_mmu_unmap_gfn_range() -> kvm_unmap_gfn_range() -> kvm_nested_s2_unmap() -> kvm_stage2_unmap_range() -> __unmap_stage2_range() -> stage2_apply_range() <- -EINVAL, triggering a WARN_ON() Racing with L0's teardown of stage 2 page tables: exit_mm() -> mmput() -> __mmput() -> exit_mmap() -> mmu_notifier_release() -> ... -> kvm_mmu_notifier_release() -> kvm_flush_shadow_all() -> kvm_arch_flush_shadow_all() -> kvm_free_stage2_pgd() -> [ acquire kvm->mmu_lock for write ] -> mmu->pgt = NULL [ among other tasks ] -> [ release kvm->mmu_lock for write ] It turns out there is a benign race resulting in a spurious warning: Thread A - notify: migration | Thread B - notify: release -------------------------------|--------------------------------- < kvm->mmu_lock held > | stage2_apply_range() | get mmu->pgt, check !NULL | ... | kvm_arch_flush_shadow_all() cond_resched_rwlock_write(); | < contend, sleep kvm->mmu_lock > < drop kvm->mmu_lock > | < acquire kvm->mmu_lock> | ... | kvm_free_stage2_pgd() | mmu->pgt = NULL | < invalidate MMU > | ... | < release kvm->mmu_lock > [ scheduled ] | stage2_apply_range() | < loop to next > | get, mmu->pgt, check !NULL | is NULL, return -EINVAL | __unmap_stage2_range() | WARN_ON(-EINVAL) <--- entirely spurious - the race was handled correctly. Fix the spurious warning by updating stage2_apply_range() to no longer treat concurrent PGT teardown on lock release as an error - whether the walker is tearing down page tables or doing something else this is a legitimate reason to abort the operation without error. This keeps the warning in place for all other circumstances. In practice only __unmap_stage2_range() actually does anything with the error so this only impacts that. Fixes: ec14c272408a ("KVM: arm64: nv: Unmap/flush shadow stage 2 page tables") Cc: stable@vger.kernel.org Reviewed-by: Yuan Yao Reviewed-by: Marc Zyngier Signed-off-by: Lorenzo Stoakes (ARM) Link: https://patch.msgid.link/20260901-kvm-arm-nested-virt-fix-v3-1-b154676f7e4c@kernel.org Signed-off-by: Oliver Upton --- arch/arm64/kvm/mmu.c | 15 ++++++++++++--- 1 file changed, 12 insertions(+), 3 deletions(-) diff --git a/arch/arm64/kvm/mmu.c b/arch/arm64/kvm/mmu.c index 9ba86450fe4a..2d44cd6a5aed 100644 --- a/arch/arm64/kvm/mmu.c +++ b/arch/arm64/kvm/mmu.c @@ -59,27 +59,36 @@ static phys_addr_t stage2_range_addr_end(phys_addr_t addr, phys_addr_t end) * long will also starve other vCPUs. We have to also make sure that the page * tables are not freed while we released the lock. */ -static int stage2_apply_range(struct kvm_s2_mmu *mmu, phys_addr_t addr, +static int stage2_apply_range(struct kvm_s2_mmu *mmu, phys_addr_t start, phys_addr_t end, int (*fn)(struct kvm_pgtable *, u64, u64), bool resched) { struct kvm *kvm = kvm_s2_mmu_to_kvm(mmu); + bool lock_dropped = false; + phys_addr_t addr = start; int ret; u64 next; do { struct kvm_pgtable *pgt = mmu->pgt; + /* + * We may be raced on PGT teardown when we release the + * kvm->mmu_lock. That's fine as the PGT is legitimately no + * longer present. + */ if (!pgt) - return -EINVAL; + return lock_dropped ? 0 : -EINVAL; next = stage2_range_addr_end(addr, end); ret = fn(pgt, addr, next - addr); if (ret) break; - if (resched && next != end) + if (resched && next != end) { cond_resched_rwlock_write(&kvm->mmu_lock); + lock_dropped = true; + } } while (addr = next, addr != end); return ret; From 4c74e233cdedd11592775fae2a6243e67ca3f891 Mon Sep 17 00:00:00 2001 From: "Lorenzo Stoakes (ARM)" Date: Tue, 1 Sep 2026 18:29:00 +0100 Subject: [PATCH 051/365] KVM: arm64: nv: Fix null ptr deref on nested wp/unmap, teardown race Commit 7270cc9157f4 ("KVM: arm64: nv: Handle VNCR_EL2 invalidation from MMU notifiers") introduced VNCR_EL2 invalidation in both kvm_nested_s2_unmap() and kvm_nested_s2_wp(). However at the point of this being performed concurrent stage 2 teardown of a nested guest can cause kvm->arch.mmu.pgt to be set to NULL. This happens in kvm_flush_shadow_all() -> kvm_arch_flush_shadow_all() -> kvm_free_stage2_pgd() and is performed under the kvm->mmu_lock. Commit ec14c272408a ("KVM: arm64: nv: Unmap/flush shadow stage 2 page tables") introduced the teardown of the entire nested MMU range, which then invokes stage2_apply_range() with resched=true: mmu_notifier_invalidate_range_start() -> ... -> kvm_mmu_notifier_invalidate_range_start() -> kvm_mmu_unmap_gfn_range() -> kvm_unmap_gfn_range() -> kvm_nested_s2_unmap() -> kvm_stage2_unmap_range() -> __unmap_stage2_range() -> stage2_apply_range() This means that stage2_apply_range() can drop the kvm->mmu_lock and thus concurrent progress can be made in lockstep with kvm_arch_flush_shadow_all(). If kvm_arch_flush_shadow_all() advances ahead of stage2_apply_range() and completes its operation it guarantees a NULL pointer deref. Since kvm_free_stage2_pgd() is performed under the kvm->mmu_lock this will either be observed NULL or not and serialised against kvm_free_stage2_pgd(). Resolve the issue by abstracting the invalidation to a new function, kvm_invalidate_vncr_ipa_all(), and check that the pgt is non-NULL before dereferencing it. Fixes: 7270cc9157f4 ("KVM: arm64: nv: Handle VNCR_EL2 invalidation from MMU notifiers") Cc: stable@vger.kernel.org Reviewed-by: Marc Zyngier Signed-off-by: Lorenzo Stoakes (ARM) Tested-by: Jonathan Davies Link: https://patch.msgid.link/20260901-kvm-arm-nested-virt-fix-v3-2-b154676f7e4c@kernel.org Signed-off-by: Oliver Upton --- arch/arm64/kvm/nested.c | 15 +++++++++++++-- 1 file changed, 13 insertions(+), 2 deletions(-) diff --git a/arch/arm64/kvm/nested.c b/arch/arm64/kvm/nested.c index 3c4fc566eafc..a0808391a456 100644 --- a/arch/arm64/kvm/nested.c +++ b/arch/arm64/kvm/nested.c @@ -1260,6 +1260,17 @@ void kvm_handle_s1e2_tlbi(struct kvm_vcpu *vcpu, u32 inst, u64 val) invalidate_vncr_va(vcpu->kvm, &scope); } +static void kvm_invalidate_vncr_ipa_all(struct kvm *kvm) +{ + struct kvm_pgtable *pgt = kvm->arch.mmu.pgt; + + lockdep_assert_held_write(&kvm->mmu_lock); + + /* if the mmu lock was dropped, pgt teardown may have raced. */ + if (pgt) + kvm_invalidate_vncr_ipa(kvm, 0, BIT(pgt->ia_bits)); +} + void kvm_nested_s2_wp(struct kvm *kvm) { int i; @@ -1276,7 +1287,7 @@ void kvm_nested_s2_wp(struct kvm *kvm) kvm_stage2_wp_range(mmu, 0, kvm_phys_size(mmu)); } - kvm_invalidate_vncr_ipa(kvm, 0, BIT(kvm->arch.mmu.pgt->ia_bits)); + kvm_invalidate_vncr_ipa_all(kvm); } void kvm_nested_s2_unmap(struct kvm *kvm, bool may_block) @@ -1295,7 +1306,7 @@ void kvm_nested_s2_unmap(struct kvm *kvm, bool may_block) kvm_stage2_unmap_range(mmu, 0, kvm_phys_size(mmu), may_block); } - kvm_invalidate_vncr_ipa(kvm, 0, BIT(kvm->arch.mmu.pgt->ia_bits)); + kvm_invalidate_vncr_ipa_all(kvm); } void kvm_nested_s2_flush(struct kvm *kvm) From 2a2eb10795a1e495aebc7f829ccecb72c05b4fd9 Mon Sep 17 00:00:00 2001 From: Fuad Tabba Date: Tue, 25 Aug 2026 09:59:45 +0100 Subject: [PATCH 052/365] KVM: arm64: Validate the SVE vector length in pkvm_vcpu_init_sve() pkvm_vcpu_init_sve() clamps only the upper bound of the host-provided sve_max_vl, so an invalid vector length reaches sve_state_size_from_vl() and the WARN_ON() there, which is fatal at EL2. The existing !sve_state_size test rejects such a length, but only after the macro has run. Check sve_vl_valid() before deriving the state size. A valid length cannot yield a zero size, so the !sve_state_size test goes with it. Fixes: 5db1bef93342 ("KVM: arm64: Track SVE state in the hypervisor vcpu structure") Reported-by: Stefan Teodorescu Reviewed-by: Marc Zyngier Signed-off-by: Fuad Tabba Link: https://patch.msgid.link/20260825085948.1674721-2-fuad.tabba@linux.dev Signed-off-by: Oliver Upton --- arch/arm64/kvm/hyp/nvhe/pkvm.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/arch/arm64/kvm/hyp/nvhe/pkvm.c b/arch/arm64/kvm/hyp/nvhe/pkvm.c index 459bd9eb7e4b..627f13fc98d4 100644 --- a/arch/arm64/kvm/hyp/nvhe/pkvm.c +++ b/arch/arm64/kvm/hyp/nvhe/pkvm.c @@ -459,14 +459,15 @@ static int pkvm_vcpu_init_sve(struct pkvm_hyp_vcpu *hyp_vcpu, struct kvm_vcpu *h /* Limit guest vector length to the maximum supported by the host. */ sve_max_vl = min(READ_ONCE(host_vcpu->arch.sve_max_vl), kvm_host_sve_max_vl); - sve_state_size = sve_state_size_from_vl(sve_max_vl); sve_state = kern_hyp_va(READ_ONCE(host_vcpu->arch.sve_state)); - if (!sve_state || !sve_state_size) { + if (!sve_vl_valid(sve_max_vl) || !sve_state) { ret = -EINVAL; goto err; } + sve_state_size = sve_state_size_from_vl(sve_max_vl); + ret = hyp_pin_shared_mem(sve_state, sve_state + sve_state_size); if (ret) goto err; From a1b3c788ad31837e348075e93dbba3f447492776 Mon Sep 17 00:00:00 2001 From: Fuad Tabba Date: Tue, 25 Aug 2026 09:59:46 +0100 Subject: [PATCH 053/365] KVM: arm64: Do not clear VM-wide SVE feature on vCPU init failure pkvm_vcpu_init_sve() clears KVM_ARM_VCPU_SVE in kvm->arch.vcpu_features when it fails, but vcpu_has_sve() tests KVM_ARCH_FLAG_GUEST_HAS_SVE, which is left set. Later vCPUs on that VM then skip the SVE setup and register with a NULL sve_state, which the guest's first FP access hands to sve_load_state(). Return the error without touching vcpu_features. Fixes: 5db1bef93342 ("KVM: arm64: Track SVE state in the hypervisor vcpu structure") Signed-off-by: Fuad Tabba Link: https://patch.msgid.link/20260825085948.1674721-3-fuad.tabba@linux.dev Signed-off-by: Oliver Upton --- arch/arm64/kvm/hyp/nvhe/pkvm.c | 13 ++++--------- 1 file changed, 4 insertions(+), 9 deletions(-) diff --git a/arch/arm64/kvm/hyp/nvhe/pkvm.c b/arch/arm64/kvm/hyp/nvhe/pkvm.c index 627f13fc98d4..4857a11d4292 100644 --- a/arch/arm64/kvm/hyp/nvhe/pkvm.c +++ b/arch/arm64/kvm/hyp/nvhe/pkvm.c @@ -450,7 +450,7 @@ static int pkvm_vcpu_init_sve(struct pkvm_hyp_vcpu *hyp_vcpu, struct kvm_vcpu *h unsigned int sve_max_vl; size_t sve_state_size; void *sve_state; - int ret = 0; + int ret; if (!vcpu_has_feature(vcpu, KVM_ARM_VCPU_SVE)) { vcpu_clear_flag(vcpu, VCPU_SVE_FINALIZED); @@ -461,24 +461,19 @@ static int pkvm_vcpu_init_sve(struct pkvm_hyp_vcpu *hyp_vcpu, struct kvm_vcpu *h sve_max_vl = min(READ_ONCE(host_vcpu->arch.sve_max_vl), kvm_host_sve_max_vl); sve_state = kern_hyp_va(READ_ONCE(host_vcpu->arch.sve_state)); - if (!sve_vl_valid(sve_max_vl) || !sve_state) { - ret = -EINVAL; - goto err; - } + if (!sve_vl_valid(sve_max_vl) || !sve_state) + return -EINVAL; sve_state_size = sve_state_size_from_vl(sve_max_vl); ret = hyp_pin_shared_mem(sve_state, sve_state + sve_state_size); if (ret) - goto err; + return ret; vcpu->arch.sve_state = sve_state; vcpu->arch.sve_max_vl = sve_max_vl; return 0; -err: - clear_bit(KVM_ARM_VCPU_SVE, vcpu->kvm->arch.vcpu_features); - return ret; } static int vm_copy_id_regs(struct pkvm_hyp_vcpu *hyp_vcpu) From 0d62fbf34d8fe7a3ff56692d39fbb8e6e9bf15fb Mon Sep 17 00:00:00 2001 From: Fuad Tabba Date: Tue, 25 Aug 2026 09:59:47 +0100 Subject: [PATCH 054/365] KVM: arm64: Key unpin_host_sve_state() on the state it unpins unpin_host_sve_state() gates on the VM's SVE feature bit, but what it unpins is the state pkvm_vcpu_init_sve() pinned. A vCPU that completed init has sve_state set exactly when that bit is set, so the two agree. Gate on sve_state, which is what is being unpinned. Signed-off-by: Fuad Tabba Link: https://patch.msgid.link/20260825085948.1674721-4-fuad.tabba@linux.dev Signed-off-by: Oliver Upton --- arch/arm64/kvm/hyp/nvhe/pkvm.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/arch/arm64/kvm/hyp/nvhe/pkvm.c b/arch/arm64/kvm/hyp/nvhe/pkvm.c index 4857a11d4292..453d3fa4a420 100644 --- a/arch/arm64/kvm/hyp/nvhe/pkvm.c +++ b/arch/arm64/kvm/hyp/nvhe/pkvm.c @@ -398,10 +398,10 @@ static void unpin_host_sve_state(struct pkvm_hyp_vcpu *hyp_vcpu) { void *sve_state; - if (!vcpu_has_feature(&hyp_vcpu->vcpu, KVM_ARM_VCPU_SVE)) + sve_state = hyp_vcpu->vcpu.arch.sve_state; + if (!sve_state) return; - sve_state = hyp_vcpu->vcpu.arch.sve_state; hyp_unpin_shared_mem(sve_state, sve_state + vcpu_sve_state_size(&hyp_vcpu->vcpu)); } From 4f16c5fc8dc4c5596e3777ab9f449a54e3f85fd5 Mon Sep 17 00:00:00 2001 From: Fuad Tabba Date: Tue, 25 Aug 2026 09:59:48 +0100 Subject: [PATCH 055/365] KVM: arm64: Derive GUEST_HAS_SVE from the SVE feature bit at EL2 pkvm_init_features_from_host() takes KVM_ARCH_FLAG_GUEST_HAS_SVE and KVM_ARM_VCPU_SVE from the host separately, but pkvm_vcpu_init_sve() tests the bit while vcpu_has_sve() reads the flag. A host that sets the flag without the bit gets a vCPU with a NULL sve_state that the world switch loads the guest's SVE state from. Derive the flag from the bit, and drop the protected path's copy of the host's flag, which is dead code since protected VMs are not allowed SVE. Fixes: 41d6028e28bd ("KVM: arm64: Convert the SVE guest vcpu flag to a vm flag") Signed-off-by: Fuad Tabba Link: https://patch.msgid.link/20260825085948.1674721-5-fuad.tabba@linux.dev Signed-off-by: Oliver Upton --- arch/arm64/kvm/hyp/nvhe/pkvm.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) diff --git a/arch/arm64/kvm/hyp/nvhe/pkvm.c b/arch/arm64/kvm/hyp/nvhe/pkvm.c index 453d3fa4a420..bb3e0dc0676e 100644 --- a/arch/arm64/kvm/hyp/nvhe/pkvm.c +++ b/arch/arm64/kvm/hyp/nvhe/pkvm.c @@ -360,7 +360,7 @@ static void pkvm_init_features_from_host(struct pkvm_hyp_vm *hyp_vm, const struc if (test_bit(KVM_ARCH_FLAG_WRITABLE_IMP_ID_REGS, &host_arch_flags)) hyp_vm->kvm.arch.midr_el1 = host_kvm->arch.midr_el1; - return; + goto out; } if (kvm_pkvm_ext_allowed(kvm, KVM_CAP_ARM_MTE)) @@ -379,13 +379,14 @@ static void pkvm_init_features_from_host(struct pkvm_hyp_vm *hyp_vm, const struc if (kvm_pkvm_ext_allowed(kvm, KVM_CAP_ARM_PTRAUTH_GENERIC)) set_bit(KVM_ARM_VCPU_PTRAUTH_GENERIC, allowed_features); - if (kvm_pkvm_ext_allowed(kvm, KVM_CAP_ARM_SVE)) { + if (kvm_pkvm_ext_allowed(kvm, KVM_CAP_ARM_SVE)) set_bit(KVM_ARM_VCPU_SVE, allowed_features); - kvm->arch.flags |= host_arch_flags & BIT(KVM_ARCH_FLAG_GUEST_HAS_SVE); - } bitmap_and(kvm->arch.vcpu_features, host_kvm->arch.vcpu_features, allowed_features, KVM_VCPU_MAX_FEATURES); +out: + __assign_bit(KVM_ARCH_FLAG_GUEST_HAS_SVE, &kvm->arch.flags, + kvm_vcpu_has_feature(kvm, KVM_ARM_VCPU_SVE)); } static void unpin_host_vcpu(struct kvm_vcpu *host_vcpu) From 64dc6f1db7e620f2e9337bb181f305fb0561da79 Mon Sep 17 00:00:00 2001 From: Karl Mehltretter Date: Sat, 29 Aug 2026 07:48:55 +0200 Subject: [PATCH 056/365] KVM: arm64: Return -EINVAL for an empty SMCCC filter range at base 0 kvm_smccc_set_filter() only rejects a range if its inclusive end, base + nr_functions - 1, is below base. That catches an empty range (nr_functions == 0) at every nonzero base, but at base 0 the end wraps to U32_MAX and KVM tries to insert [0, U32_MAX], which overlaps the reserved Arm Architecture Calls ranges. KVM_ARM_VM_SMCCC_FILTER then returns -EEXIST instead of the -EINVAL that the smccc_filter selftest expects for an empty range. Reject a zero function count explicitly. Tested with a userspace reproducer on an arm64 VHE host under QEMU TCG: EEXIST before, EINVAL after. Fixes: 821d935c87bc ("KVM: arm64: Introduce support for userspace SMCCC filtering") Assisted-by: LLM Signed-off-by: Karl Mehltretter Reviewed-by: Steffen Eiden Reviewed-by: Fuad Tabba Tested-by: Fuad Tabba Link: https://patch.msgid.link/20260829054856.70549-2-kmehltretter@gmail.com Signed-off-by: Oliver Upton --- arch/arm64/kvm/hypercalls.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/arch/arm64/kvm/hypercalls.c b/arch/arm64/kvm/hypercalls.c index b11b8821c9fb..dfa25bb6f25d 100644 --- a/arch/arm64/kvm/hypercalls.c +++ b/arch/arm64/kvm/hypercalls.c @@ -185,7 +185,8 @@ static int kvm_smccc_set_filter(struct kvm *kvm, struct kvm_smccc_filter __user start = filter.base; end = start + filter.nr_functions - 1; - if (end < start || filter.action >= NR_SMCCC_FILTER_ACTIONS) + if (!filter.nr_functions || end < start || + filter.action >= NR_SMCCC_FILTER_ACTIONS) return -EINVAL; mutex_lock(&kvm->arch.config_lock); From 0a46eb5719fa57cc9d06025dcd8ba271643dee61 Mon Sep 17 00:00:00 2001 From: Karl Mehltretter Date: Sat, 29 Aug 2026 07:48:56 +0200 Subject: [PATCH 057/365] KVM: arm64: selftests: Test empty SMCCC filter range at base 0 test_invalid_nr_functions() only checks an empty range at PSCI_0_2_FN64_CPU_ON, which KVM's end < start check happens to catch. It never exercised base 0, where the inclusive end wraps to U32_MAX instead. Add the base 0 case. Without the preceding fix it fails with EEXIST. Assisted-by: LLM Signed-off-by: Karl Mehltretter Reviewed-by: Steffen Eiden Reviewed-by: Fuad Tabba Tested-by: Fuad Tabba Link: https://patch.msgid.link/20260829054856.70549-3-kmehltretter@gmail.com Signed-off-by: Oliver Upton --- tools/testing/selftests/kvm/arm64/smccc_filter.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/tools/testing/selftests/kvm/arm64/smccc_filter.c b/tools/testing/selftests/kvm/arm64/smccc_filter.c index 21e41880261b..a41ed3e016ba 100644 --- a/tools/testing/selftests/kvm/arm64/smccc_filter.c +++ b/tools/testing/selftests/kvm/arm64/smccc_filter.c @@ -140,6 +140,10 @@ static void test_invalid_nr_functions(void) TEST_ASSERT(r < 0 && errno == EINVAL, "Attempt to filter 0 functions should return EINVAL"); + r = __set_smccc_filter(vm, 0, 0, KVM_SMCCC_FILTER_DENY); + TEST_ASSERT(r < 0 && errno == EINVAL, + "Attempt to filter 0 functions at base 0 should return EINVAL"); + kvm_vm_free(vm); } From 3a8c562892b96f35bba1e00d5e455a15963bbb92 Mon Sep 17 00:00:00 2001 From: Fuad Tabba Date: Tue, 8 Sep 2026 12:07:10 +0100 Subject: [PATCH 058/365] KVM: arm64: Transfer the hyp stack pages out of the host stage-2 fix_host_ownership() walks only the linear-map alias of each memblock region, and the per-CPU hyp stack, mapped in the private VA range for its guard page, has none. Walk each stack's VA range with the same walker. Fixes: 1a919b17ef012 ("KVM: arm64: Add guard pages for pKVM (protected nVHE) hypervisor stack") Reported-by: Hiroyuki Katsura Cc: stable@vger.kernel.org Signed-off-by: Fuad Tabba Reviewed-by: Vincent Donnefort Tested-by: Vincent Donnefort Reviewed-by: Marc Zyngier Link: https://patch.msgid.link/20260908110713.1540304-2-fuad.tabba@linux.dev Signed-off-by: Oliver Upton --- arch/arm64/kvm/hyp/nvhe/setup.c | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/arch/arm64/kvm/hyp/nvhe/setup.c b/arch/arm64/kvm/hyp/nvhe/setup.c index 75b00c323310..362f2891cb32 100644 --- a/arch/arm64/kvm/hyp/nvhe/setup.c +++ b/arch/arm64/kvm/hyp/nvhe/setup.c @@ -269,6 +269,16 @@ static int fix_host_ownership(void) return ret; } + /* The stacks sit in the private VA range, not the linear map. */ + for (i = 0; i < hyp_nr_cpus; i++) { + struct kvm_nvhe_init_params *params = per_cpu_ptr(&kvm_init_params, i); + u64 start = params->stack_hyp_va - NVHE_STACK_SIZE; + + ret = kvm_pgtable_walk(&pkvm_pgtable, start, NVHE_STACK_SIZE, &walker); + if (ret) + return ret; + } + return 0; } From 5a8b505ede133fb30ca3b3a19d0db00c08237615 Mon Sep 17 00:00:00 2001 From: Fuad Tabba Date: Tue, 8 Sep 2026 12:07:11 +0100 Subject: [PATCH 059/365] KVM: arm64: Match hyp text by physical address in fix_host_ownership() On a non-hVHE host, fix_host_ownership_walker()'s test for PAGE_HYP_EXEC never matches: KVM_PGTABLE_PROT_UX is cleared at map time and only PX is reported on read-back. Hyp text is therefore donated rather than left read-only in the host stage-2, and the instruction dump in nvhe_hyp_panic_handler() reads a page the host has no access to. Match the text by physical address instead, in a helper a later patch reuses. A test on the permissions would leave any other executable mapping host-readable too. Fixes: 80cbfd7174f31 ("KVM: arm64: Honor UX/PX attributes for EL2 S1 mappings") Signed-off-by: Fuad Tabba Reviewed-by: Vincent Donnefort Tested-by: Vincent Donnefort Reviewed-by: Marc Zyngier Link: https://patch.msgid.link/20260908110713.1540304-3-fuad.tabba@linux.dev Signed-off-by: Oliver Upton --- arch/arm64/kvm/hyp/include/nvhe/mem_protect.h | 1 + arch/arm64/kvm/hyp/nvhe/mem_protect.c | 8 ++++++++ arch/arm64/kvm/hyp/nvhe/setup.c | 2 +- 3 files changed, 10 insertions(+), 1 deletion(-) diff --git a/arch/arm64/kvm/hyp/include/nvhe/mem_protect.h b/arch/arm64/kvm/hyp/include/nvhe/mem_protect.h index 29935c7da1de..cab27f7bd423 100644 --- a/arch/arm64/kvm/hyp/include/nvhe/mem_protect.h +++ b/arch/arm64/kvm/hyp/include/nvhe/mem_protect.h @@ -52,6 +52,7 @@ int __pkvm_host_test_clear_young_guest(u64 gfn, u64 nr_pages, bool mkold, struct int __pkvm_host_mkyoung_guest(u64 gfn, struct pkvm_hyp_vcpu *vcpu); bool addr_is_memory(phys_addr_t phys); +bool addr_is_hyp_text(phys_addr_t phys); int host_stage2_idmap_locked(phys_addr_t addr, u64 size, enum kvm_pgtable_prot prot); int host_stage2_set_owner_locked(phys_addr_t addr, u64 size, u8 owner_id); int kvm_host_prepare_stage2(void *pgt_pool_base); diff --git a/arch/arm64/kvm/hyp/nvhe/mem_protect.c b/arch/arm64/kvm/hyp/nvhe/mem_protect.c index 39aa8911f62c..d026f446bd8e 100644 --- a/arch/arm64/kvm/hyp/nvhe/mem_protect.c +++ b/arch/arm64/kvm/hyp/nvhe/mem_protect.c @@ -450,6 +450,14 @@ bool addr_is_memory(phys_addr_t phys) return !!find_mem_range(phys, &range); } +bool addr_is_hyp_text(phys_addr_t phys) +{ + phys_addr_t start = ALIGN_DOWN(__hyp_pa(__hyp_text_start), PAGE_SIZE); + phys_addr_t end = PAGE_ALIGN(__hyp_pa(__hyp_text_end)); + + return phys >= start && phys < end; +} + static bool is_in_mem_range(u64 addr, struct kvm_mem_range *range) { return range->start <= addr && addr < range->end; diff --git a/arch/arm64/kvm/hyp/nvhe/setup.c b/arch/arm64/kvm/hyp/nvhe/setup.c index 362f2891cb32..bb667cd7080b 100644 --- a/arch/arm64/kvm/hyp/nvhe/setup.c +++ b/arch/arm64/kvm/hyp/nvhe/setup.c @@ -217,7 +217,7 @@ static int fix_host_ownership_walker(const struct kvm_pgtable_visit_ctx *ctx, case PKVM_PAGE_OWNED: set_hyp_state(page, PKVM_PAGE_OWNED); /* hyp text is RO in the host stage-2 to be inspected on panic. */ - if (prot == PAGE_HYP_EXEC) { + if (addr_is_hyp_text(phys)) { set_host_state(page, PKVM_NOPAGE); return host_stage2_idmap_locked(phys, PAGE_SIZE, KVM_PGTABLE_PROT_R); } else { From 2245841401147b8022a5857061b870d82e595352 Mon Sep 17 00:00:00 2001 From: Fuad Tabba Date: Tue, 8 Sep 2026 12:07:12 +0100 Subject: [PATCH 060/365] KVM: arm64: Move the private VA allocation cursor to __io_map_next __io_map_base is the start of the private VA range only until the first allocation from it, after which it is the allocation cursor. Keep it as the start and move the cursor to __io_map_next, for the walk of the range the next patch adds. No functional change intended. Signed-off-by: Fuad Tabba Reviewed-by: Vincent Donnefort Tested-by: Vincent Donnefort Reviewed-by: Marc Zyngier Link: https://patch.msgid.link/20260908110713.1540304-4-fuad.tabba@linux.dev Signed-off-by: Oliver Upton --- arch/arm64/kvm/hyp/nvhe/mm.c | 20 +++++++++++--------- 1 file changed, 11 insertions(+), 9 deletions(-) diff --git a/arch/arm64/kvm/hyp/nvhe/mm.c b/arch/arm64/kvm/hyp/nvhe/mm.c index 3b0bee496bff..422ee57be956 100644 --- a/arch/arm64/kvm/hyp/nvhe/mm.c +++ b/arch/arm64/kvm/hyp/nvhe/mm.c @@ -25,6 +25,7 @@ struct memblock_region hyp_memory[HYP_MEMBLOCK_REGIONS]; unsigned int hyp_memblock_nr; static u64 __io_map_base; +static u64 __io_map_next; struct hyp_fixmap_slot { u64 addr; @@ -50,7 +51,7 @@ static int __pkvm_alloc_private_va_range(unsigned long start, size_t size) hyp_assert_lock_held(&pkvm_pgd_lock); - if (!start || start < __io_map_base) + if (!start || start < __io_map_next) return -EINVAL; /* The allocated size is always a multiple of PAGE_SIZE */ @@ -60,7 +61,7 @@ static int __pkvm_alloc_private_va_range(unsigned long start, size_t size) if (cur > __hyp_vmemmap) return -ENOMEM; - __io_map_base = cur; + __io_map_next = cur; return 0; } @@ -70,7 +71,7 @@ static int __pkvm_alloc_private_va_range(unsigned long start, size_t size) * @size: The size of the VA range to reserve. * @haddr: The hypervisor virtual start address of the allocation. * - * The private virtual address (VA) range is allocated above __io_map_base + * The private virtual address (VA) range is allocated above __io_map_next * and aligned based on the order of @size. * * Return: 0 on success or negative error code on failure. @@ -81,7 +82,7 @@ int pkvm_alloc_private_va_range(size_t size, unsigned long *haddr) int ret; hyp_spin_lock(&pkvm_pgd_lock); - addr = __io_map_base; + addr = __io_map_next; ret = __pkvm_alloc_private_va_range(addr, size); hyp_spin_unlock(&pkvm_pgd_lock); @@ -341,7 +342,7 @@ static int create_fixblock(void) return -EINVAL; hyp_spin_lock(&pkvm_pgd_lock); - addr = ALIGN(__io_map_base, PMD_SIZE); + addr = ALIGN(__io_map_next, PMD_SIZE); ret = __pkvm_alloc_private_va_range(addr, PMD_SIZE); if (ret) goto unlock; @@ -426,6 +427,7 @@ int hyp_create_idmap(u32 hyp_va_bits) */ __io_map_base = start & BIT(hyp_va_bits - 2); __io_map_base ^= BIT(hyp_va_bits - 2); + __io_map_next = __io_map_base; __hyp_vmemmap = __io_map_base | BIT(hyp_va_bits - 3); return __pkvm_create_mappings(start, end - start, start, PAGE_HYP_EXEC); @@ -433,19 +435,19 @@ int hyp_create_idmap(u32 hyp_va_bits) int pkvm_create_stack(phys_addr_t phys, unsigned long *haddr) { - unsigned long addr, prev_base; + unsigned long addr, prev_next; size_t size; int ret; hyp_spin_lock(&pkvm_pgd_lock); - prev_base = __io_map_base; + prev_next = __io_map_next; /* * Efficient stack verification using the NVHE_STACK_SHIFT bit implies * an alignment of our allocation on the order of the size. */ size = NVHE_STACK_SIZE * 2; - addr = ALIGN(__io_map_base, size); + addr = ALIGN(__io_map_next, size); ret = __pkvm_alloc_private_va_range(addr, size); if (!ret) { @@ -461,7 +463,7 @@ int pkvm_create_stack(phys_addr_t phys, unsigned long *haddr) ret = kvm_pgtable_hyp_map(&pkvm_pgtable, addr + NVHE_STACK_SIZE, NVHE_STACK_SIZE, phys, PAGE_HYP); if (ret) - __io_map_base = prev_base; + __io_map_next = prev_next; } hyp_spin_unlock(&pkvm_pgd_lock); From cfe80c3837f93202970b6d8f706f79c5c7396f17 Mon Sep 17 00:00:00 2001 From: Fuad Tabba Date: Tue, 8 Sep 2026 12:07:13 +0100 Subject: [PATCH 061/365] KVM: arm64: Check every private mapping is hyp-owned at pKVM init fix_host_ownership() transfers only what it walks, so a hyp mapping outside the linear map is not manipulated by the walk. Walk the quarter of the VA space holding the private range and the vmemmap once the transfer is done, and fail init unless every valid leaf is hyp-owned: in the vmemmap when the page is memory, and in the host stage-2, where hyp text may instead be mapped without write access. A leaf that is not memory has no vmemmap entry and is checked against the host stage-2 alone. Hyp text is matched by physical address, since the only executable mapping in the range is the Spectre-v3a vectors, whose VA is a private allocation, and an executable mapping of anything else must not be host-readable. The vmemmap can be block-mapped, so the walker checks each page of a leaf. Suggested-by: Will Deacon Signed-off-by: Fuad Tabba Reviewed-by: Vincent Donnefort Tested-by: Vincent Donnefort Reviewed-by: Marc Zyngier Link: https://patch.msgid.link/20260908110713.1540304-5-fuad.tabba@linux.dev Signed-off-by: Oliver Upton --- arch/arm64/kvm/hyp/include/nvhe/mem_protect.h | 1 + arch/arm64/kvm/hyp/include/nvhe/mm.h | 1 + arch/arm64/kvm/hyp/nvhe/mem_protect.c | 12 ++++ arch/arm64/kvm/hyp/nvhe/mm.c | 59 +++++++++++++++++++ arch/arm64/kvm/hyp/nvhe/setup.c | 4 ++ 5 files changed, 77 insertions(+) diff --git a/arch/arm64/kvm/hyp/include/nvhe/mem_protect.h b/arch/arm64/kvm/hyp/include/nvhe/mem_protect.h index cab27f7bd423..ec85a9547120 100644 --- a/arch/arm64/kvm/hyp/include/nvhe/mem_protect.h +++ b/arch/arm64/kvm/hyp/include/nvhe/mem_protect.h @@ -55,6 +55,7 @@ bool addr_is_memory(phys_addr_t phys); bool addr_is_hyp_text(phys_addr_t phys); int host_stage2_idmap_locked(phys_addr_t addr, u64 size, enum kvm_pgtable_prot prot); int host_stage2_set_owner_locked(phys_addr_t addr, u64 size, u8 owner_id); +bool host_stage2_pte_is_hyp_owned(kvm_pte_t pte); int kvm_host_prepare_stage2(void *pgt_pool_base); int kvm_guest_prepare_stage2(struct pkvm_hyp_vm *vm, void *pgd); void kvm_guest_destroy_stage2(struct pkvm_hyp_vm *vm); diff --git a/arch/arm64/kvm/hyp/include/nvhe/mm.h b/arch/arm64/kvm/hyp/include/nvhe/mm.h index 6e83ce35c2f2..31cae95ddb71 100644 --- a/arch/arm64/kvm/hyp/include/nvhe/mm.h +++ b/arch/arm64/kvm/hyp/include/nvhe/mm.h @@ -29,6 +29,7 @@ int __pkvm_create_private_mapping(phys_addr_t phys, size_t size, enum kvm_pgtable_prot prot, unsigned long *haddr); int pkvm_create_stack(phys_addr_t phys, unsigned long *haddr); +int pkvm_check_host_ownership(void); int pkvm_alloc_private_va_range(size_t size, unsigned long *haddr); #endif /* __KVM_HYP_MM_H */ diff --git a/arch/arm64/kvm/hyp/nvhe/mem_protect.c b/arch/arm64/kvm/hyp/nvhe/mem_protect.c index d026f446bd8e..a6a47c1e058b 100644 --- a/arch/arm64/kvm/hyp/nvhe/mem_protect.c +++ b/arch/arm64/kvm/hyp/nvhe/mem_protect.c @@ -641,6 +641,18 @@ int host_stage2_set_owner_locked(phys_addr_t addr, u64 size, u8 owner_id) return ret; } +bool host_stage2_pte_is_hyp_owned(kvm_pte_t pte) +{ + if (kvm_pte_valid(pte)) + return false; + + if (FIELD_GET(KVM_INVALID_PTE_TYPE_MASK, pte) != + KVM_HOST_INVALID_PTE_TYPE_DONATION) + return false; + + return FIELD_GET(KVM_HOST_DONATION_PTE_OWNER_MASK, pte) == PKVM_ID_HYP; +} + #define KVM_HOST_PTE_OWNER_GUEST_HANDLE_MASK GENMASK(15, 0) /* We need 40 bits for the GFN to cover a 52-bit IPA with 4k pages and LPA2 */ #define KVM_HOST_PTE_OWNER_GUEST_GFN_MASK GENMASK(55, 16) diff --git a/arch/arm64/kvm/hyp/nvhe/mm.c b/arch/arm64/kvm/hyp/nvhe/mm.c index 422ee57be956..29ab5ee9d57f 100644 --- a/arch/arm64/kvm/hyp/nvhe/mm.c +++ b/arch/arm64/kvm/hyp/nvhe/mm.c @@ -472,6 +472,65 @@ int pkvm_create_stack(phys_addr_t phys, unsigned long *haddr) return ret; } +static int check_page_ownership(phys_addr_t phys) +{ + kvm_pte_t pte; + bool host_ok; + int ret; + + if (addr_is_memory(phys)) { + struct hyp_page *page = hyp_phys_to_page(phys); + + if (get_hyp_state(page) != PKVM_PAGE_OWNED || + get_host_state(page) != PKVM_NOPAGE) + return -EPERM; + } + + ret = kvm_pgtable_get_leaf(&host_mmu.pgt, phys, &pte, NULL); + if (ret) + return ret; + + /* Hyp text may stay host-readable, see fix_host_ownership_walker(). */ + if (kvm_pte_valid(pte) && addr_is_hyp_text(phys)) + host_ok = !(kvm_pgtable_stage2_pte_prot(pte) & KVM_PGTABLE_PROT_W); + else + host_ok = host_stage2_pte_is_hyp_owned(pte); + + return host_ok ? 0 : -EPERM; +} + +static int check_host_ownership_walker(const struct kvm_pgtable_visit_ctx *ctx, + enum kvm_pgtable_walk_flags visit) +{ + phys_addr_t phys, end; + int ret; + + if (!kvm_pte_valid(ctx->old)) + return 0; + + phys = kvm_pte_to_phys(ctx->old); + end = phys + kvm_granule_size(ctx->level); + for (; phys < end; phys += PAGE_SIZE) { + ret = check_page_ownership(phys); + if (ret) + return ret; + } + + return 0; +} + +int pkvm_check_host_ownership(void) +{ + struct kvm_pgtable_walker walker = { + .cb = check_host_ownership_walker, + .flags = KVM_PGTABLE_WALK_LEAF, + }; + + /* The private range and the vmemmap share one quarter of the VA space. */ + return kvm_pgtable_walk(&pkvm_pgtable, __io_map_base, + BIT(pkvm_pgtable.ia_bits - 2), &walker); +} + static void *admit_host_page(void *arg) { struct kvm_hyp_memcache *host_mc = arg; diff --git a/arch/arm64/kvm/hyp/nvhe/setup.c b/arch/arm64/kvm/hyp/nvhe/setup.c index bb667cd7080b..45ac5f2ba4f7 100644 --- a/arch/arm64/kvm/hyp/nvhe/setup.c +++ b/arch/arm64/kvm/hyp/nvhe/setup.c @@ -334,6 +334,10 @@ void __noreturn __pkvm_init_finalise(void) if (ret) goto out; + ret = pkvm_check_host_ownership(); + if (ret) + goto out; + ret = hyp_ffa_init(ffa_proxy_pages); if (ret) goto out; From 33346f8960c7bb6a3b4e273b5cfe25c5a8be349f Mon Sep 17 00:00:00 2001 From: Marc Zyngier Date: Fri, 11 Sep 2026 17:22:02 +0100 Subject: [PATCH 062/365] KVM: arm64: nv: Fix life cycle of the nested_mmus array The nested_mmus array holds the shadow page tables that are used when a guest is running a nested context. These structures are allocated on VCPU_INIT for whole guest, which implies that they may have to be relocated as the array grows. Should a VCPU_INIT occur whilst a vcpu is actively running an L2 and that the allocation requires relocation, that vcpu will still be running with a pointer to the previous structure, which will have been freed. Fix this by turning the array of structures to an array of pointers, which is now allocated at VM creation, sized to the absolute maximum that KVM can handle. In turn, each VCPU_INIT contributes S2_MMU_PER_VCPU to the pool. No reallocation is ever performed, and the life cycle of each object is much clearer: - the nested_mmus array is allocated in kvm_init_nested(), and freed in kvm_arch_destroy_vm() - s2_mmu structures are allocated in kvm_vcpu_init_nested(), and freed on kvm_arch_flush_shadow_all() Finally, the freeing of vcpu->arch.vncr_array is made consistent rather than being done on some failure paths, but not others. Fixes: 4f128f8e1aaa ("KVM: arm64: nv: Support multiple nested Stage-2 mmu structures") Reported-by: Shen Yongchao Reported-by: Karl Mehltretter Suggested-by: Karl Mehltretter Acked-by: Lorenzo Stoakes (ARM) Link: https://lore.kernel.org/r/20260803224405.41468-1-kmehltretter@gmail.com Signed-off-by: Marc Zyngier Cc: stable@vger.kernel.org Reviewed-by: Wei-Lin Chang Link: https://patch.msgid.link/20260911162203.1919330-2-maz@kernel.org Signed-off-by: Oliver Upton --- arch/arm64/include/asm/kvm_host.h | 2 +- arch/arm64/include/asm/kvm_nested.h | 2 +- arch/arm64/kvm/arm.c | 8 ++- arch/arm64/kvm/nested.c | 93 +++++++++++++---------------- 4 files changed, 51 insertions(+), 54 deletions(-) diff --git a/arch/arm64/include/asm/kvm_host.h b/arch/arm64/include/asm/kvm_host.h index 27fe0cd5b2d7..cd9b9d2462f9 100644 --- a/arch/arm64/include/asm/kvm_host.h +++ b/arch/arm64/include/asm/kvm_host.h @@ -322,7 +322,7 @@ struct kvm_arch { * Stage 2 paging state for VMs with nested S2 using a virtual * VMID. */ - struct kvm_s2_mmu *nested_mmus; + struct kvm_s2_mmu **nested_mmus; size_t nested_mmus_size; int nested_mmus_next; diff --git a/arch/arm64/include/asm/kvm_nested.h b/arch/arm64/include/asm/kvm_nested.h index 1ed708335809..5b8edb2e8a87 100644 --- a/arch/arm64/include/asm/kvm_nested.h +++ b/arch/arm64/include/asm/kvm_nested.h @@ -66,7 +66,7 @@ static inline u64 translate_ttbr0_el2_to_ttbr0_el1(u64 ttbr0) extern bool forward_smc_trap(struct kvm_vcpu *vcpu); extern bool forward_debug_exception(struct kvm_vcpu *vcpu); -extern void kvm_init_nested(struct kvm *kvm); +extern int kvm_init_nested(struct kvm *kvm); extern int kvm_vcpu_init_nested(struct kvm_vcpu *vcpu); extern void kvm_init_nested_s2_mmu(struct kvm_s2_mmu *mmu); extern struct kvm_s2_mmu *lookup_s2_mmu(struct kvm_vcpu *vcpu); diff --git a/arch/arm64/kvm/arm.c b/arch/arm64/kvm/arm.c index 8b080804bc90..b53219e048bd 100644 --- a/arch/arm64/kvm/arm.c +++ b/arch/arm64/kvm/arm.c @@ -236,8 +236,6 @@ int kvm_arch_init_vm(struct kvm *kvm, unsigned long type) mutex_unlock(&kvm->lock); #endif - kvm_init_nested(kvm); - ret = kvm_share_hyp(kvm, kvm + 1); if (ret) return ret; @@ -252,6 +250,10 @@ int kvm_arch_init_vm(struct kvm *kvm, unsigned long type) if (ret) goto err_free_cpumask; + ret = kvm_init_nested(kvm); + if (ret) + goto err_uninit_mmu; + if (is_protected_kvm_enabled()) { /* * If any failures occur after this is successful, make sure to @@ -280,6 +282,7 @@ int kvm_arch_init_vm(struct kvm *kvm, unsigned long type) err_uninit_mmu: kvm_uninit_stage2_mmu(kvm); + kvfree(kvm->arch.nested_mmus); err_free_cpumask: free_cpumask_var(kvm->arch.supported_cpus); err_unshare_kvm: @@ -337,6 +340,7 @@ void kvm_arch_destroy_vm(struct kvm *kvm) kvm_unshare_hyp(kvm, kvm + 1); + kvfree(kvm->arch.nested_mmus); kvm_arm_teardown_hypercalls(kvm); } diff --git a/arch/arm64/kvm/nested.c b/arch/arm64/kvm/nested.c index a0808391a456..2571f177e654 100644 --- a/arch/arm64/kvm/nested.c +++ b/arch/arm64/kvm/nested.c @@ -45,11 +45,15 @@ struct vncr_tlb { */ #define S2_MMU_PER_VCPU 2 -void kvm_init_nested(struct kvm *kvm) +int kvm_init_nested(struct kvm *kvm) { - kvm->arch.nested_mmus = NULL; + kvm->arch.nested_mmus = kvmalloc_objs(struct kvm_s2_mmu *, + KVM_MAX_VCPUS * S2_MMU_PER_VCPU, + GFP_KERNEL_ACCOUNT); kvm->arch.nested_mmus_size = 0; atomic_set(&kvm->arch.vncr_tlb_count, 0); + + return kvm->arch.nested_mmus ? 0 : -ENOMEM; } static int init_nested_s2_mmu(struct kvm *kvm, struct kvm_s2_mmu *mmu) @@ -70,8 +74,9 @@ static int init_nested_s2_mmu(struct kvm *kvm, struct kvm_s2_mmu *mmu) int kvm_vcpu_init_nested(struct kvm_vcpu *vcpu) { struct kvm *kvm = vcpu->kvm; - struct kvm_s2_mmu *tmp; - int num_mmus, ret = 0; + int num_mmus; + + lockdep_assert_held(&kvm->arch.config_lock); if (test_bit(KVM_ARM_VCPU_HAS_EL2_E2H0, kvm->arch.vcpu_features) && !cpus_have_final_cap(ARM64_HAS_HCR_NV1)) @@ -84,51 +89,40 @@ int kvm_vcpu_init_nested(struct kvm_vcpu *vcpu) if (!vcpu->arch.ctxt.vncr_array) return -ENOMEM; - /* - * Let's treat memory allocation failures as benign: If we fail to - * allocate anything, return an error and keep the allocated array - * alive. Userspace may try to recover by initializing the vcpu - * again, and there is no reason to affect the whole VM for this. - */ num_mmus = atomic_read(&kvm->online_vcpus) * S2_MMU_PER_VCPU; if (num_mmus > kvm->arch.nested_mmus_size) { - tmp = kvzalloc_objs(*tmp, num_mmus, GFP_KERNEL_ACCOUNT); + struct kvm_s2_mmu *tmp; + int i, ret = 0; + + tmp = kvzalloc_objs(*tmp, S2_MMU_PER_VCPU, GFP_KERNEL_ACCOUNT); if (!tmp) - return -ENOMEM; + ret = -ENOMEM; - write_lock(&kvm->mmu_lock); - - if (kvm->arch.nested_mmus_size) { - memcpy(tmp, kvm->arch.nested_mmus, - size_mul(sizeof(*tmp), kvm->arch.nested_mmus_size)); - - for (int i = 0; i < kvm->arch.nested_mmus_size; i++) - tmp[i].pgt->mmu = &tmp[i]; + for (i = 0; !ret && i < S2_MMU_PER_VCPU; i++) { + ret = init_nested_s2_mmu(kvm, &tmp[i]); + if (ret) + break; } - swap(kvm->arch.nested_mmus, tmp); + if (ret) { + while (--i >= 0) + kvm_free_stage2_pgd(&tmp[i]); - write_unlock(&kvm->mmu_lock); + kvfree(tmp); + free_page((unsigned long)vcpu->arch.ctxt.vncr_array); + vcpu->arch.ctxt.vncr_array = NULL; + return ret; + } - kvfree(tmp); + guard(write_lock)(&kvm->mmu_lock); + + for (i = 0; i < S2_MMU_PER_VCPU; i++) + kvm->arch.nested_mmus[i + kvm->arch.nested_mmus_size] = &tmp[i]; + + kvm->arch.nested_mmus_size += S2_MMU_PER_VCPU; } - for (int i = kvm->arch.nested_mmus_size; !ret && i < num_mmus; i++) - ret = init_nested_s2_mmu(kvm, &kvm->arch.nested_mmus[i]); - - if (ret) { - for (int i = kvm->arch.nested_mmus_size; i < num_mmus; i++) - kvm_free_stage2_pgd(&kvm->arch.nested_mmus[i]); - - free_page((unsigned long)vcpu->arch.ctxt.vncr_array); - vcpu->arch.ctxt.vncr_array = NULL; - - return ret; - } - - kvm->arch.nested_mmus_size = num_mmus; - return 0; } @@ -742,7 +736,7 @@ void kvm_s2_mmu_iterate_by_vmid(struct kvm *kvm, u16 vmid, write_lock(&kvm->mmu_lock); for (int i = 0; i < kvm->arch.nested_mmus_size; i++) { - struct kvm_s2_mmu *mmu = &kvm->arch.nested_mmus[i]; + struct kvm_s2_mmu *mmu = kvm->arch.nested_mmus[i]; if (!kvm_s2_mmu_valid(mmu)) continue; @@ -784,7 +778,7 @@ struct kvm_s2_mmu *lookup_s2_mmu(struct kvm_vcpu *vcpu) * if S2 translation is disabled. */ for (int i = 0; i < kvm->arch.nested_mmus_size; i++) { - struct kvm_s2_mmu *mmu = &kvm->arch.nested_mmus[i]; + struct kvm_s2_mmu *mmu = kvm->arch.nested_mmus[i]; if (!kvm_s2_mmu_valid(mmu)) continue; @@ -823,7 +817,7 @@ static struct kvm_s2_mmu *get_s2_mmu_nested(struct kvm_vcpu *vcpu) for (i = kvm->arch.nested_mmus_next; i < (kvm->arch.nested_mmus_size + kvm->arch.nested_mmus_next); i++) { - s2_mmu = &kvm->arch.nested_mmus[i % kvm->arch.nested_mmus_size]; + s2_mmu = kvm->arch.nested_mmus[i % kvm->arch.nested_mmus_size]; if (atomic_read(&s2_mmu->refcnt) == 0) break; @@ -1281,7 +1275,7 @@ void kvm_nested_s2_wp(struct kvm *kvm) return; for (i = 0; i < kvm->arch.nested_mmus_size; i++) { - struct kvm_s2_mmu *mmu = &kvm->arch.nested_mmus[i]; + struct kvm_s2_mmu *mmu = kvm->arch.nested_mmus[i]; if (kvm_s2_mmu_valid(mmu)) kvm_stage2_wp_range(mmu, 0, kvm_phys_size(mmu)); @@ -1300,7 +1294,7 @@ void kvm_nested_s2_unmap(struct kvm *kvm, bool may_block) return; for (i = 0; i < kvm->arch.nested_mmus_size; i++) { - struct kvm_s2_mmu *mmu = &kvm->arch.nested_mmus[i]; + struct kvm_s2_mmu *mmu = kvm->arch.nested_mmus[i]; if (kvm_s2_mmu_valid(mmu)) kvm_stage2_unmap_range(mmu, 0, kvm_phys_size(mmu), may_block); @@ -1319,7 +1313,7 @@ void kvm_nested_s2_flush(struct kvm *kvm) return; for (i = 0; i < kvm->arch.nested_mmus_size; i++) { - struct kvm_s2_mmu *mmu = &kvm->arch.nested_mmus[i]; + struct kvm_s2_mmu *mmu = kvm->arch.nested_mmus[i]; if (kvm_s2_mmu_valid(mmu)) kvm_stage2_flush_range(mmu, 0, kvm_phys_size(mmu)); @@ -1328,16 +1322,15 @@ void kvm_nested_s2_flush(struct kvm *kvm) void kvm_arch_flush_shadow_all(struct kvm *kvm) { - int i; - - for (i = 0; i < kvm->arch.nested_mmus_size; i++) { - struct kvm_s2_mmu *mmu = &kvm->arch.nested_mmus[i]; + for (int i = kvm->arch.nested_mmus_size - 1; i >= 0; i--) { + struct kvm_s2_mmu *mmu = kvm->arch.nested_mmus[i]; if (!WARN_ON(atomic_read(&mmu->refcnt))) kvm_free_stage2_pgd(mmu); + + if ((i % S2_MMU_PER_VCPU) == 0) + kvfree(mmu); } - kvfree(kvm->arch.nested_mmus); - kvm->arch.nested_mmus = NULL; kvm->arch.nested_mmus_size = 0; kvm_uninit_stage2_mmu(kvm); } From e5843f4effaa2ffac3e789ecd4456403564961d4 Mon Sep 17 00:00:00 2001 From: Marc Zyngier Date: Fri, 11 Sep 2026 17:22:03 +0100 Subject: [PATCH 063/365] KVM: arm64: nv: Delay freeing of shadow S2 structures until VM destruction We free the shadow S2 structures from kvm_arch_flush_shadow_all(), which is a Bad Idea(tm). Freeing the page tables is fair game (this is what this callback is for), but freeing the container that could still be referenced by another part of the system is not great. Instead, grow separate destructors that gets called when we tear the VM down for good. From there, we can nuke both the individual MMUs as well as the global array that points to them, safe in the knowledge that the vcpus themselves have been destroyed already. Fixes: 4f128f8e1aaac ("KVM: arm64: nv: Support multiple nested Stage-2 mmu structures") Reviewed-by: Lorenzo Stoakes (ARM) Signed-off-by: Marc Zyngier Cc: stable@vger.kernel.org Reviewed-by: Wei-Lin Chang Link: https://patch.msgid.link/20260911162203.1919330-3-maz@kernel.org Signed-off-by: Oliver Upton --- arch/arm64/include/asm/kvm_nested.h | 1 + arch/arm64/kvm/arm.c | 4 ++-- arch/arm64/kvm/nested.c | 15 ++++++++++----- 3 files changed, 13 insertions(+), 7 deletions(-) diff --git a/arch/arm64/include/asm/kvm_nested.h b/arch/arm64/include/asm/kvm_nested.h index 5b8edb2e8a87..586026e85903 100644 --- a/arch/arm64/include/asm/kvm_nested.h +++ b/arch/arm64/include/asm/kvm_nested.h @@ -67,6 +67,7 @@ static inline u64 translate_ttbr0_el2_to_ttbr0_el1(u64 ttbr0) extern bool forward_smc_trap(struct kvm_vcpu *vcpu); extern bool forward_debug_exception(struct kvm_vcpu *vcpu); extern int kvm_init_nested(struct kvm *kvm); +extern void kvm_destroy_nested(struct kvm *kvm); extern int kvm_vcpu_init_nested(struct kvm_vcpu *vcpu); extern void kvm_init_nested_s2_mmu(struct kvm_s2_mmu *mmu); extern struct kvm_s2_mmu *lookup_s2_mmu(struct kvm_vcpu *vcpu); diff --git a/arch/arm64/kvm/arm.c b/arch/arm64/kvm/arm.c index b53219e048bd..eaf583b77193 100644 --- a/arch/arm64/kvm/arm.c +++ b/arch/arm64/kvm/arm.c @@ -282,7 +282,7 @@ int kvm_arch_init_vm(struct kvm *kvm, unsigned long type) err_uninit_mmu: kvm_uninit_stage2_mmu(kvm); - kvfree(kvm->arch.nested_mmus); + kvm_destroy_nested(kvm); err_free_cpumask: free_cpumask_var(kvm->arch.supported_cpus); err_unshare_kvm: @@ -340,7 +340,7 @@ void kvm_arch_destroy_vm(struct kvm *kvm) kvm_unshare_hyp(kvm, kvm + 1); - kvfree(kvm->arch.nested_mmus); + kvm_destroy_nested(kvm); kvm_arm_teardown_hypercalls(kvm); } diff --git a/arch/arm64/kvm/nested.c b/arch/arm64/kvm/nested.c index 2571f177e654..b191365d97cc 100644 --- a/arch/arm64/kvm/nested.c +++ b/arch/arm64/kvm/nested.c @@ -56,6 +56,15 @@ int kvm_init_nested(struct kvm *kvm) return kvm->arch.nested_mmus ? 0 : -ENOMEM; } +void kvm_destroy_nested(struct kvm *kvm) +{ + for (int i = 0; i < kvm->arch.nested_mmus_size; i+= S2_MMU_PER_VCPU) + kvfree(kvm->arch.nested_mmus[i]); + + kvm->arch.nested_mmus_size = 0; + kvfree(kvm->arch.nested_mmus); +} + static int init_nested_s2_mmu(struct kvm *kvm, struct kvm_s2_mmu *mmu) { /* @@ -1322,16 +1331,12 @@ void kvm_nested_s2_flush(struct kvm *kvm) void kvm_arch_flush_shadow_all(struct kvm *kvm) { - for (int i = kvm->arch.nested_mmus_size - 1; i >= 0; i--) { + for (int i = 0; i < kvm->arch.nested_mmus_size; i++) { struct kvm_s2_mmu *mmu = kvm->arch.nested_mmus[i]; if (!WARN_ON(atomic_read(&mmu->refcnt))) kvm_free_stage2_pgd(mmu); - - if ((i % S2_MMU_PER_VCPU) == 0) - kvfree(mmu); } - kvm->arch.nested_mmus_size = 0; kvm_uninit_stage2_mmu(kvm); } From 49d9d295d69d07e850cae35933ba8519e2915f26 Mon Sep 17 00:00:00 2001 From: Fuad Tabba Date: Mon, 14 Sep 2026 10:38:38 +0100 Subject: [PATCH 064/365] KVM: arm64: Don't WARN on an unknown VM ioctl in protected mode kvm_pkvm_ioctl_allowed() WARNs when kvm_get_cap_for_kvm_ioctl() doesn't find the ioctl number in vm_ioctl_caps[], and kvm_arch_vm_ioctl() calls it for every number the generic code doesn't handle, so ioctl(vm_fd, 0xdeadbeef) from userspace taints a pKVM host and panics it under panic_on_warn. The lookup is fed userspace input: return false, and userspace gets the -EINVAL kvm_arch_vm_ioctl() returns for that number on a host without pKVM. Fixes: b12b3b04f6ba0 ("KVM: arm64: Check whether a VM IOCTL is allowed in pKVM") Cc: stable@vger.kernel.org Signed-off-by: Fuad Tabba Reviewed-by: Suzuki K Poulose Link: https://patch.msgid.link/20260914093838.1082637-1-fuad.tabba@linux.dev Signed-off-by: Oliver Upton --- arch/arm64/include/asm/kvm_pkvm.h | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/arch/arm64/include/asm/kvm_pkvm.h b/arch/arm64/include/asm/kvm_pkvm.h index beea00e693a0..cad60569f061 100644 --- a/arch/arm64/include/asm/kvm_pkvm.h +++ b/arch/arm64/include/asm/kvm_pkvm.h @@ -65,8 +65,7 @@ static inline bool kvm_pkvm_ioctl_allowed(struct kvm *kvm, unsigned int ioctl) int r; r = kvm_get_cap_for_kvm_ioctl(ioctl, &ext); - - if (WARN_ON_ONCE(r < 0)) + if (r < 0) return false; return kvm_pkvm_ext_allowed(kvm, ext); From 96e6757cb0674acb86ee8b558ee6bffe0eec0bb0 Mon Sep 17 00:00:00 2001 From: Sebastian Ott Date: Mon, 14 Sep 2026 15:10:13 +0200 Subject: [PATCH 065/365] KVM: selftests: fix steal_time for arm64 with host page size > 4K Fix the following failure when running with 16K host page size: ==== Test Assertion Failure ==== lib/kvm_util.c:991: vm_adjust_num_guest_pages(vm->mode, npages) == npages pid=873 tid=873 errno=0 - Success 1 0x0000000000405a27: vm_mem_add at kvm_util.c:991 2 0x000000000040241f: check_steal_time_uapi at steal_time.c:223 (discriminator 7) 3 (inlined by) main at steal_time.c:539 (discriminator 7) 4 0x00007fff8b57af3b: ?? ??:0 5 0x00007fff8b57b007: ?? ??:0 6 0x0000000000402b6f: _start at ??:? Number of guest pages is not compatible with the host. Try npages=4 Fixes: fc240715fc50 ("KVM: selftests: arm64: Fix steal_time test after UAPI refactoring") Reported-by: Zenghui Yu Link: https://lore.kernel.org/kvmarm/7575a845-a542-4b16-b512-aec3126f97f3@linux.dev/T/#u Signed-off-by: Sebastian Ott Reviewed-by: Zenghui Yu (Huawei) Link: https://patch.msgid.link/20260914131013.60334-1-sebott@redhat.com Signed-off-by: Oliver Upton --- tools/testing/selftests/kvm/steal_time.c | 30 ++++++++++++++++-------- 1 file changed, 20 insertions(+), 10 deletions(-) diff --git a/tools/testing/selftests/kvm/steal_time.c b/tools/testing/selftests/kvm/steal_time.c index bc3c62b72c58..785d19f9ee8f 100644 --- a/tools/testing/selftests/kvm/steal_time.c +++ b/tools/testing/selftests/kvm/steal_time.c @@ -27,6 +27,9 @@ static void *st_gva[NR_VCPUS]; static u64 guest_stolen_time[NR_VCPUS]; +static struct kvm_vm *vm_create_steal_time(u32 nr_vcpus, void *guest_code, + struct kvm_vcpu *vcpus[]); + #if defined(__x86_64__) /* steal_time must have 64-byte alignment */ @@ -210,17 +213,14 @@ static void check_steal_time_uapi(void) u64 st_ipa; int ret; - vm = vm_create_with_one_vcpu(&vcpu, NULL); - struct kvm_device_attr dev = { .group = KVM_ARM_VCPU_PVTIME_CTRL, .attr = KVM_ARM_VCPU_PVTIME_IPA, .addr = (u64)&st_ipa, }; + vm = vm_create_steal_time(1, NULL, &vcpu); vcpu_ioctl(vcpu, KVM_HAS_DEVICE_ATTR, &dev); - vm_userspace_mem_region_add(vm, VM_MEM_SRC_ANONYMOUS, ST_GPA_BASE, 1, 1, 0); - virt_map(vm, ST_GPA_BASE, ST_GPA_BASE, 1); st_ipa = (ulong)ST_GPA_BASE | 1; ret = __vcpu_ioctl(vcpu, KVM_SET_DEVICE_ATTR, &dev); @@ -500,13 +500,27 @@ static void run_vcpu(struct kvm_vcpu *vcpu) } } +static struct kvm_vm *vm_create_steal_time(u32 nr_vcpus, void *guest_code, + struct kvm_vcpu *vcpus[]) +{ + unsigned int gpages; + struct kvm_vm *vm; + + /* Create a VM and an identity mapped memslot for the steal time structure */ + vm = vm_create_with_vcpus(nr_vcpus, guest_code, vcpus); + gpages = vm_calc_num_guest_pages(VM_MODE_DEFAULT, STEAL_TIME_SIZE * nr_vcpus); + vm_userspace_mem_region_add(vm, VM_MEM_SRC_ANONYMOUS, ST_GPA_BASE, 1, gpages, 0); + virt_map(vm, ST_GPA_BASE, ST_GPA_BASE, gpages); + + return vm; +} + int main(int ac, char **av) { struct kvm_vcpu *vcpus[NR_VCPUS]; struct kvm_vm *vm; pthread_t thread; cpu_set_t cpuset; - unsigned int gpages; long stolen_time; long run_delay; bool verbose; @@ -517,11 +531,7 @@ int main(int ac, char **av) /* Set CPU affinity so we can force preemption of the VCPU */ cpu = pin_self_to_any_cpu(); - /* Create a VM and an identity mapped memslot for the steal time structure */ - vm = vm_create_with_vcpus(NR_VCPUS, guest_code, vcpus); - gpages = vm_calc_num_guest_pages(VM_MODE_DEFAULT, STEAL_TIME_SIZE * NR_VCPUS); - vm_userspace_mem_region_add(vm, VM_MEM_SRC_ANONYMOUS, ST_GPA_BASE, 1, gpages, 0); - virt_map(vm, ST_GPA_BASE, ST_GPA_BASE, gpages); + vm = vm_create_steal_time(NR_VCPUS, guest_code, vcpus); ksft_print_header(); TEST_REQUIRE(is_steal_time_supported(vcpus[0])); From 089e4f3c4862ba3f29dff2361caa8084879194fd Mon Sep 17 00:00:00 2001 From: Mark Brown Date: Tue, 1 Sep 2026 22:47:00 +0100 Subject: [PATCH 066/365] KVM: arm64: Fix FGT mapping for HFGITR_EL2.nGCSEPP The encoding to trap mapping currently maps a FGT on OP_GCSPOPX to HFGITR_EL2.nGCSEPP but as per DDI0601 2026-06 this FGT controls trapping of GCSPUSHX and GCSPOPCX, and not the separate GCSPOPX instruction. Update the mapping to reflect the architecture. Fixes: 863ac38984a82 ("KVM: arm64: Add missing HFGITR_EL2 FGT entries to nested virt") Reviewed-by: Leonardo Bras Signed-off-by: Mark Brown Reviewed-by: Lorenzo Stoakes (ARM) Link: https://patch.msgid.link/20260901-arm64-gcs-v20-2-f31750bdfadb@kernel.org Signed-off-by: Oliver Upton --- arch/arm64/kvm/emulate-nested.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/arm64/kvm/emulate-nested.c b/arch/arm64/kvm/emulate-nested.c index 625604019fb3..3806ff0920fe 100644 --- a/arch/arm64/kvm/emulate-nested.c +++ b/arch/arm64/kvm/emulate-nested.c @@ -1445,7 +1445,7 @@ static const struct encoding_to_trap_config encoding_to_fgt[] __initconst = { SR_FGT(OP_AT_S1E1A, HFGITR, ATS1E1A, 1), SR_FGT(OP_COSP_RCTX, HFGITR, COSPRCTX, 1), SR_FGT(OP_GCSPUSHX, HFGITR, nGCSEPP, 0), - SR_FGT(OP_GCSPOPX, HFGITR, nGCSEPP, 0), + SR_FGT(OP_GCSPOPCX, HFGITR, nGCSEPP, 0), SR_FGT(OP_GCSPUSHM, HFGITR, nGCSPUSHM_EL1, 0), SR_FGT(OP_BRB_IALL, HFGITR, nBRBIALL, 0), SR_FGT(OP_BRB_INJ, HFGITR, nBRBINJ, 0), From 7de9a6fb44eae4f05e68c805d58b9c618815adfa Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Tue, 15 Sep 2026 12:34:44 -1000 Subject: [PATCH 067/365] sched_ext: Wait for SCX_OPSS_DISPATCHING before reenqueueing a task ebf1ccff79c4 ("sched_ext: Fix ops.dequeue() semantics") moved the final ops_state store in scx_dispatch_enqueue() after the DSQ unlock so that the custody update and ops.dequeue() precede it. A task can thus be found on a DSQ while still SCX_OPSS_DISPATCHING. The dequeue and core-sched pick paths wait for the state to clear in ops_dequeue() but the reenqueue paths don't. A reenqueue in that window runs ops.enqueue() and sets SCX_OPSS_QUEUED before the dispatch has completed. The dispatcher's final store then overwrites it with SCX_OPSS_NONE and finish_dispatch() drops every later dispatch of the task. Wait for SCX_OPSS_DISPATCHING to clear before dequeueing a task for reenqueue, the same way ops_dequeue() does. Fixes: ebf1ccff79c4 ("sched_ext: Fix ops.dequeue() semantics") Cc: stable@vger.kernel.org # v7.1+ Signed-off-by: Tejun Heo Reviewed-by: Andrea Righi --- kernel/sched/ext/ext.c | 13 +++++++++++++ kernel/sched/ext/internal.h | 1 + kernel/sched/ext/sub.c | 1 + 3 files changed, 15 insertions(+) diff --git a/kernel/sched/ext/ext.c b/kernel/sched/ext/ext.c index 70b711c4de6e..725b890fcab9 100644 --- a/kernel/sched/ext/ext.c +++ b/kernel/sched/ext/ext.c @@ -4404,6 +4404,17 @@ static bool local_task_should_reenq(struct rq *rq, struct task_struct *p, return *reenq_flags & SCX_REENQ_ANY; } +/* + * The dispatcher stores the final ops_state after dropping the DSQ lock, so @p + * can be found on a DSQ while still %SCX_OPSS_DISPATCHING. Reenqueueing @p + * before that store lands would have it clobber the new %SCX_OPSS_QUEUED. + */ +void scx_reenq_wait_dispatching(struct task_struct *p) +{ + if (unlikely(atomic_long_read_acquire(&p->scx.ops_state) == SCX_OPSS_DISPATCHING)) + wait_ops_state(p, SCX_OPSS_DISPATCHING); +} + static u32 reenq_local(struct scx_sched *sch, struct rq *rq, u64 reenq_flags) { LIST_HEAD(tasks); @@ -4447,6 +4458,7 @@ static u32 reenq_local(struct scx_sched *sch, struct rq *rq, u64 reenq_flags) if (!local_task_should_reenq(rq, p, &reenq_flags, &reason)) continue; + scx_reenq_wait_dispatching(p); scx_dispatch_dequeue(rq, p); if (WARN_ON_ONCE(p->scx.flags & SCX_TASK_REENQ_REASON_MASK)) @@ -4570,6 +4582,7 @@ static void reenq_user(struct rq *rq, struct scx_dispatch_q *dsq, u64 reenq_flag } /* @p is on @dsq, its rq and @dsq are locked */ + scx_reenq_wait_dispatching(p); dispatch_dequeue_locked(p, dsq); raw_spin_unlock(&dsq->lock); diff --git a/kernel/sched/ext/internal.h b/kernel/sched/ext/internal.h index 3464e0f113c1..115c96fbf932 100644 --- a/kernel/sched/ext/internal.h +++ b/kernel/sched/ext/internal.h @@ -2078,6 +2078,7 @@ void scx_kick_cpu(struct scx_sched *sch, s32 cpu, u64 flags); u64 __scx_bpf_now(struct rq *rq); void schedule_dsq_reenq(struct scx_sched *sch, struct scx_dispatch_q *dsq, u64 reenq_flags, struct rq *locked_rq); +void scx_reenq_wait_dispatching(struct task_struct *p); int __scx_init_task(struct scx_sched *sch, struct task_struct *p, struct cgroup *cgrp, bool fork); void scx_enable_task(struct scx_sched *sch, struct task_struct *p); diff --git a/kernel/sched/ext/sub.c b/kernel/sched/ext/sub.c index f7aeb1488566..a30c965b175d 100644 --- a/kernel/sched/ext/sub.c +++ b/kernel/sched/ext/sub.c @@ -801,6 +801,7 @@ void scx_reenq_reject(struct rq *rq) if (WARN_ON_ONCE(p->migration_pending)) continue; + scx_reenq_wait_dispatching(p); scx_dispatch_dequeue(rq, p); if (WARN_ON_ONCE(p->scx.flags & SCX_TASK_REENQ_REASON_MASK)) From b52d695d062095327b944acf7daabbc816ab319b Mon Sep 17 00:00:00 2001 From: Stanley Jhu Date: Sat, 12 Sep 2026 21:16:25 +0800 Subject: [PATCH 068/365] scsi: ufs: core: Keep internal commands dispatchable during error handling Commit 08b12cda6c44 ("scsi: ufs: core: Switch to scsi_get_internal_cmd()") switched UFS internal commands to allocate requests on hba->host->pseudo_sdev->request_queue, which shares the host tagset with regular LUNs. During error recovery, ufshcd_err_handling_prepare() calls blk_mq_quiesce_tagset(&hba->host->tag_set), marking all queues in the tagset as quiesced, including pseudo_sdev->request_queue. When ufshcd_verify_dev_init() subsequently issues internal commands (e.g. NOP OUT UPIU) via blk_execute_rq(), blk_mq_run_hw_queue() skips running the quiesced queue, resulting in an unrecoverable circular wait deadlock. Keep quiescing the tagset and unquiesce the pseudo SCSI device on top of that, so internal commands stay dispatchable while the logical units remain quiesced. Re-quiesce the pseudo device before unquiescing the tagset so that quiesce_depth stays balanced. Clock scaling and ufshcd_pause_command_processing() are unaffected: they keep quiescing the whole tagset, internal commands included. Fixes: 08b12cda6c44 ("scsi: ufs: core: Switch to scsi_get_internal_cmd()") Cc: stable@vger.kernel.org Link: https://lore.kernel.org/all/6f78c4bd-a70b-402d-abfd-599091b67674@acm.org/ Signed-off-by: Stanley Jhu Reviewed-by: Bart Van Assche Link: https://patch.msgid.link/20260912131625.2301486-1-stanleyjhu@google.com Signed-off-by: Martin K. Petersen (Oracle) --- drivers/ufs/core/ufshcd.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c index 2ba244cf40ac..54f4e7d7de02 100644 --- a/drivers/ufs/core/ufshcd.c +++ b/drivers/ufs/core/ufshcd.c @@ -6817,11 +6817,17 @@ static void ufshcd_err_handling_prepare(struct ufs_hba *hba) } /* Wait for ongoing ufshcd_queuecommand() calls to finish. */ blk_mq_quiesce_tagset(&hba->host->tag_set); + /* + * Internal commands are submitted on the pseudo SCSI device. Let them + * through so that the error handler can recover the link. + */ + blk_mq_unquiesce_queue(hba->host->pseudo_sdev->request_queue); cancel_work_sync(&hba->eeh_work); } static void ufshcd_err_handling_unprepare(struct ufs_hba *hba) { + blk_mq_quiesce_queue_nowait(hba->host->pseudo_sdev->request_queue); blk_mq_unquiesce_tagset(&hba->host->tag_set); ufshcd_release(hba); if (ufshcd_is_clkscaling_supported(hba)) From c9ee6511332687ea714ad8ab86a53cb837d86eea Mon Sep 17 00:00:00 2001 From: Geert Uytterhoeven Date: Mon, 14 Sep 2026 16:00:01 +0200 Subject: [PATCH 069/365] scsi: ufs: pltfrm: Add quirk for R-Car S4 lacking lanes-per-direction Since commit e72323f3b09f ("scsi: ufs: core: Configure only active lanes during link"), the following error is observed on R-Car S4: ufshcd-renesas e6860000.ufs: Tx lane mismatch [config,reported] [2,1] ufshcd-renesas e6860000.ufs: link startup failed -67 ufshcd-renesas e6860000.ufs: error -ENOLINK: Initialization failed with error -67 ufshcd-renesas e6860000.ufs: probe with driver ufshcd-renesas failed with error -67 R-Car S4 has one UFS lane per direction, as described in section 152.1 of its hardware manual. Without lanes-per-direction, the UFS platform driver defaults to two lanes. Previously, the core used PA_CONNECTEDRXDATALANES and PA_CONNECTEDTXDATALANES to configure the link without checking them against lanes-per-direction, so the missing property did not prevent initialization. While fixing the R-Car S4 DTS is the proper solution, doing only that would still break backwards compatibility with existing DTBs. Hence add a quirk to let lanes-per-direction default to one on R-Car S4. Fixes: e72323f3b09f9c89 ("scsi: ufs: core: Configure only active lanes during link") Reported-by: Koichiro Den Closes: https://lore.kernel.org/20260911073058.253000-1-den@valinux.co.jp Cc: stable@vger.kernel.org # 7.2+ Signed-off-by: Geert Uytterhoeven Link: https://patch.msgid.link/ae0cc2bd764e6dfffce99db3d8b44a55887c508c.1789394185.git.geert+renesas@glider.be Signed-off-by: Martin K. Petersen (Oracle) --- drivers/ufs/host/ufshcd-pltfrm.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/drivers/ufs/host/ufshcd-pltfrm.c b/drivers/ufs/host/ufshcd-pltfrm.c index 5ac7afe75934..169dbc1a75fe 100644 --- a/drivers/ufs/host/ufshcd-pltfrm.c +++ b/drivers/ufs/host/ufshcd-pltfrm.c @@ -206,7 +206,11 @@ static void ufshcd_init_lanes_per_dir(struct ufs_hba *hba) dev_dbg(hba->dev, "%s: failed to read lanes-per-direction, ret=%d\n", __func__, ret); - hba->lanes_per_direction = UFSHCD_DEFAULT_LANES_PER_DIRECTION; + /* Old R-Car S4 DTBs lack "lanes-per-direction = <1>" */ + if (of_device_is_compatible(dev->of_node, "renesas,r8a779f0-ufs")) + hba->lanes_per_direction = 1; + else + hba->lanes_per_direction = UFSHCD_DEFAULT_LANES_PER_DIRECTION; } } From bce07e2f37b5e4a427d36fd6b1c14067b27591db Mon Sep 17 00:00:00 2001 From: Yehyeong Lee Date: Sat, 1 Aug 2026 22:36:35 +0900 Subject: [PATCH 070/365] scsi: libiscsi_tcp: Check the data direction of a Data-In PDU The Data-In branch of iscsi_tcp_hdr_dissect() resolves the ITT to a task and copies the PDU's data segment into that command's scatterlist without asking whether the command was reading. iscsi_tcp_r2t_rsp() in the same file does ask, and rejects an R2T for a command that is not DMA_TO_DEVICE. A target that answers a WRITE command's ITT with a Data-In therefore has the initiator write target-supplied bytes into the pages that write was about to send. Those are the caller's own pinned pages for an O_DIRECT write, and page cache pages for a buffered one. Observed against a test target that emits one 512-byte Data-In naming a 128 KB write's ITT, after the R2T for that write. With O_DIRECT the caller's buffer ends up holding 512 bytes of the target's data while pwrite() returns 131072. Buffered is quieter: pwrite() and fsync() both succeed, nothing is logged, and reading those blocks back returns the target's bytes out of the page cache without a command going on the wire. Check the direction before using the scatterlist, the way the R2T path already does. Cc: stable@vger.kernel.org Signed-off-by: Yehyeong Lee Reviewed-by: Mike Christie Link: https://patch.msgid.link/20260801133635.1986706-1-yhlee@isslab.korea.ac.kr Fixes: a081c13e39b5 ("[SCSI] iscsi_tcp: split module into lib and lld") Signed-off-by: Martin K. Petersen (Oracle) --- drivers/scsi/libiscsi_tcp.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/drivers/scsi/libiscsi_tcp.c b/drivers/scsi/libiscsi_tcp.c index 7223bb18b048..d35f93451ee9 100644 --- a/drivers/scsi/libiscsi_tcp.c +++ b/drivers/scsi/libiscsi_tcp.c @@ -480,6 +480,9 @@ static int iscsi_tcp_data_in(struct iscsi_conn *conn, struct iscsi_task *task) int datasn = be32_to_cpu(rhdr->datasn); unsigned total_in_length = task->sc->sdb.length; + if (task->sc->sc_data_direction != DMA_FROM_DEVICE) + return ISCSI_ERR_PROTO; + /* * lib iscsi will update this in the completion handling if there * is status. From f06a44e235ef188689ba23ffc72e9e89b10951a9 Mon Sep 17 00:00:00 2001 From: "Ewan D. Milne" Date: Tue, 15 Sep 2026 13:38:00 -0400 Subject: [PATCH 071/365] scsi: devinfo: Add BLIST_SKIP_IO_HINTS for EMC Symmetrix EMC Symmetrix returns an error on MODE SENSE for page 0Ah subpage 05h because it does not implement the SBC-5 I/O hints. These commands began to be sent as a result of commit 4f53138fffc2 ("scsi: sd: Translate data lifetime information"). Add BLIST_SKIP_IO_HINTS to avoid sending these commands because in large configurations the failed commands are displacing other useful information in internal error logs. Signed-off-by: Ewan D. Milne Link: https://patch.msgid.link/20260915173800.39117-1-emilne@redhat.com Signed-off-by: Martin K. Petersen (Oracle) --- drivers/scsi/scsi_devinfo.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/scsi/scsi_devinfo.c b/drivers/scsi/scsi_devinfo.c index 15ffbe93ac72..88a911b35c94 100644 --- a/drivers/scsi/scsi_devinfo.c +++ b/drivers/scsi/scsi_devinfo.c @@ -161,7 +161,7 @@ static struct { {"DGC", "DISK", NULL, BLIST_SPARSELUN}, /* EMC CLARiiON, no storage on LUN 0 */ {"EMC", "Invista", "*", BLIST_SPARSELUN | BLIST_LARGELUN}, {"EMC", "SYMMETRIX", NULL, BLIST_SPARSELUN | BLIST_LARGELUN | - BLIST_REPORTLUN2 | BLIST_RETRY_ITF}, + BLIST_REPORTLUN2 | BLIST_RETRY_ITF | BLIST_SKIP_IO_HINTS}, {"EMULEX", "MD21/S2 ESDI", NULL, BLIST_SINGLELUN}, {"easyRAID", "16P", NULL, BLIST_NOREPORTLUN}, {"easyRAID", "X6P", NULL, BLIST_NOREPORTLUN}, From 278210c60c6f6958bd2eeaa2120c862683b83d09 Mon Sep 17 00:00:00 2001 From: Arnd Bergmann Date: Tue, 15 Sep 2026 22:20:59 +0200 Subject: [PATCH 072/365] scsi: leapraid: Avoid -Wformat-security warning When extra warnings are enabled, the alloc_ordered_workqueue() function cannot be called with a variable name for the format string: drivers/scsi/leapraid/leapraid_os.c: In function 'leapraid_probe': drivers/scsi/leapraid/leapraid_os.c:2062:58: error: format not a string literal and no format arguments [-Werror=format-security] 2062 | alloc_ordered_workqueue(adapter->fw_evt_s.fw_evt_name, 0); | ~~~~~~~~~~~~~~~~~^~~~~~~~~~~~ As the string is only assembled before the call and not used anywhere else, just fold the earlier snprintf() into the alloc_ordered_workqueue() call. Fixes: 5597088c9e79 ("scsi: leapraid: Add new SCSI driver") Signed-off-by: Arnd Bergmann Reviewed-by: Damien Le Moal Link: https://patch.msgid.link/20260915202134.3534708-1-arnd@kernel.org Signed-off-by: Martin K. Petersen (Oracle) --- drivers/scsi/leapraid/leapraid_func.h | 2 -- drivers/scsi/leapraid/leapraid_os.c | 8 +++----- 2 files changed, 3 insertions(+), 7 deletions(-) diff --git a/drivers/scsi/leapraid/leapraid_func.h b/drivers/scsi/leapraid/leapraid_func.h index 4c0b9ca728d8..e8a0815bf95f 100644 --- a/drivers/scsi/leapraid/leapraid_func.h +++ b/drivers/scsi/leapraid/leapraid_func.h @@ -554,7 +554,6 @@ struct leapraid_fw_evt_work { /** * struct leapraid_fw_evt_struct - Firmware event handling structure * - * @fw_evt_name: Name of the firmware event. * @fw_evt_thread: Workqueue used for processing firmware events. * @fw_evt_lock: Spinlock protecting access to the firmware event list. * @fw_evt_list: Linked list of pending firmware events. @@ -565,7 +564,6 @@ struct leapraid_fw_evt_work { */ struct leapraid_fw_evt_struct { u32 leapraid_evt_masks[4]; - char fw_evt_name[48]; struct workqueue_struct *fw_evt_thread; spinlock_t fw_evt_lock; /* protects firmware event */ struct list_head fw_evt_list; diff --git a/drivers/scsi/leapraid/leapraid_os.c b/drivers/scsi/leapraid/leapraid_os.c index ee3242779dfd..507f11862276 100644 --- a/drivers/scsi/leapraid/leapraid_os.c +++ b/drivers/scsi/leapraid/leapraid_os.c @@ -2054,12 +2054,10 @@ static int leapraid_probe(struct pci_dev *pdev, const struct pci_device_id *id) shost->transportt = leapraid_transport_template; shost->unique_id = adapter->adapter_attr.id; - snprintf(adapter->fw_evt_s.fw_evt_name, - sizeof(adapter->fw_evt_s.fw_evt_name), - "fw_event_%s%d", LEAPRAID_DRIVER_NAME, - adapter->adapter_attr.id); adapter->fw_evt_s.fw_evt_thread = - alloc_ordered_workqueue(adapter->fw_evt_s.fw_evt_name, 0); + alloc_ordered_workqueue("fw_event_%s%d", 0, + LEAPRAID_DRIVER_NAME, + adapter->adapter_attr.id); if (!adapter->fw_evt_s.fw_evt_thread) { dev_err(&adapter->pdev->dev, "%s: Failed to create fw event workqueue\n", __func__); From 1f7745fb3580152ca902ef181b605f33cabfb1d0 Mon Sep 17 00:00:00 2001 From: Ran Hongyun Date: Mon, 13 Jul 2026 19:55:25 +0800 Subject: [PATCH 073/365] squashfs: Add dictionary size range check to prevent shift-out-of-bounds When an abnormal SquashFS image (COMP_OPTS flag is 1 but dictionary size is 0) is mounted, and performs shift operations using dictionarysize, the shift exponent is -1, causing a shift-out-of-bounds. Detail as below: squashfs_comp_opts(msblk, buffer, length) squashfs_xz_comp_opts() if (comp_opts) n = ffs(opts->dict_size) - 1;<----opts->dict_size=0, n=-1 if (opts->dict_size != (1 << n) && opts->dict_size != (1 << n) + (1 << (n + 1))) <----shift-out-of-bounds Fix it by adding a dictionary size range check before the shift operation. Fixes: ff750311d30a ("Squashfs: add compression options support to xz decompressor") Signed-off-by: Ran Hongyun Link: https://patch.msgid.link/20260713115525.2661734-1-ranhongyun1@huawei.com Reviewed-by: Phillip Lougher Reviewed-by: Zhihao Cheng Signed-off-by: Christian Brauner (Amutable) --- fs/squashfs/xz_wrapper.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/fs/squashfs/xz_wrapper.c b/fs/squashfs/xz_wrapper.c index 0a4ff3ec9c8c..6610af241449 100644 --- a/fs/squashfs/xz_wrapper.c +++ b/fs/squashfs/xz_wrapper.c @@ -57,10 +57,10 @@ static void *squashfs_xz_comp_opts(struct squashfs_sb_info *msblk, opts->dict_size = le32_to_cpu(comp_opts->dictionary_size); - /* the dictionary size should be 2^n or 2^n+2^(n+1) */ + /* the dictionary size should be positive and 2^n or 2^n+2^(n+1) */ n = ffs(opts->dict_size) - 1; - if (opts->dict_size != (1 << n) && opts->dict_size != (1 << n) + - (1 << (n + 1))) { + if (opts->dict_size <= 0 || (opts->dict_size != (1 << n) && + opts->dict_size != (1 << n) + (1 << (n + 1)))) { err = -EIO; goto out; } From 7c431d61b69a3fd0784c20aa4cd0b8fb501b5653 Mon Sep 17 00:00:00 2001 From: ZHOU Jiaxiang Date: Wed, 16 Sep 2026 21:58:21 +0800 Subject: [PATCH 074/365] scsi: block: Fix zones_cond out-of-bounds write on zone report blk_revalidate_disk_zones() sizes the zones_cond array from the disk capacity and zone size, but the index used by blk_revalidate_zone_cond() comes from the device-driven report_zones() walk and is never checked against the array size. A device reporting more zones than fit the array makes blk_zone_set_cond() write out of bounds. One way to reach this is a zone count exceeding 32 bits: both blk_revalidate_zone_args.nr_zones and struct zoned_disk_info.nr_zones are unsigned int, so a disk advertising more than UINT_MAX zones (e.g. 2^32 + 1024 zones of one 512-byte logical block) gets its zone count truncated to a small value, undersizing the array while the report walk keeps counting upward. Check the index against the array size before storing the zone condition, and refuse to revalidate when the zone count does not fit 32 bits. Fixes: 6e945ffb6555 ("block: use zone condition to determine conventional zones") Signed-off-by: ZHOU Jiaxiang Reviewed-by: Damien Le Moal Link: https://patch.msgid.link/7815D1B293A8F55E+20260916135822.32584-2-me@fxti.xyz Signed-off-by: Martin K. Petersen (Oracle) --- block/blk-zoned.c | 15 +++++++++++++-- 1 file changed, 13 insertions(+), 2 deletions(-) diff --git a/block/blk-zoned.c b/block/blk-zoned.c index a5afb842bf35..475aa16bc41a 100644 --- a/block/blk-zoned.c +++ b/block/blk-zoned.c @@ -2018,12 +2018,17 @@ static int disk_revalidate_zone_resources(struct gendisk *disk, struct blk_revalidate_zone_args *args) { struct queue_limits *lim = &disk->queue->limits; + unsigned long long nr_zones; unsigned int pool_size; int ret = 0; args->disk = disk; - args->nr_zones = - DIV_ROUND_UP_ULL(get_capacity(disk), lim->chunk_sectors); + nr_zones = DIV_ROUND_UP_ULL(get_capacity(disk), lim->chunk_sectors); + if (nr_zones > UINT_MAX) { + pr_warn("%s: Too many zones (%llu)\n", disk->disk_name, nr_zones); + return -EINVAL; + } + args->nr_zones = nr_zones; /* Cached zone conditions: 1 byte per zone */ args->zones_cond = kzalloc(args->nr_zones, GFP_NOIO); @@ -2131,6 +2136,12 @@ static int blk_revalidate_zone_cond(struct blk_zone *zone, unsigned int idx, { enum blk_zone_cond cond = zone->cond; + if (idx >= args->nr_zones) { + pr_warn("%s: Zone report index %u exceeds zone count %u\n", + args->disk->disk_name, idx, args->nr_zones); + return -EINVAL; + } + /* Check that the zone condition is consistent with the zone type. */ switch (cond) { case BLK_ZONE_COND_NOT_WP: From b6ec0f79745967c751c85df373062c8d15e45fc4 Mon Sep 17 00:00:00 2001 From: ZHOU Jiaxiang Date: Wed, 16 Sep 2026 21:58:22 +0800 Subject: [PATCH 075/365] scsi: sd_zbc: Reject disks with too many zones sd_zbc_read_zones() computes the number of zones with 64-bit arithmetic and stores the result in the unsigned int nr_zones field of struct zoned_disk_info, silently truncating counts that exceed 32 bits. The truncated count is later used to size per-zone resources, while the device may still report more zones than fit. Moreover, sd_zbc_report_zones() counts the reported zones with a signed int zone_idx, which overflows past INT_MAX. Reject devices reporting more than INT_MAX zones at scan time; such a device is not realistic for any medium that exists today, and accepting it produces inconsistent zone bookkeeping. Fixes: 89d947561077 ("sd: Implement support for ZBC devices") Signed-off-by: ZHOU Jiaxiang Reviewed-by: Damien Le Moal Link: https://patch.msgid.link/C41798AB5AA6BF2B+20260916135822.32584-3-me@fxti.xyz Signed-off-by: Martin K. Petersen (Oracle) --- drivers/scsi/sd_zbc.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/drivers/scsi/sd_zbc.c b/drivers/scsi/sd_zbc.c index 56e455fb5add..456beaf2e769 100644 --- a/drivers/scsi/sd_zbc.c +++ b/drivers/scsi/sd_zbc.c @@ -589,7 +589,7 @@ int sd_zbc_revalidate_zones(struct scsi_disk *sdkp) int sd_zbc_read_zones(struct scsi_disk *sdkp, struct queue_limits *lim, u8 buf[SD_BUF_SIZE]) { - unsigned int nr_zones; + u64 nr_zones; u32 zone_blocks = 0; int ret; @@ -621,6 +621,12 @@ int sd_zbc_read_zones(struct scsi_disk *sdkp, struct queue_limits *lim, goto err; nr_zones = round_up(sdkp->capacity, zone_blocks) >> ilog2(zone_blocks); + if (nr_zones > INT_MAX) { + sd_printk(KERN_ERR, sdkp, "Too many zones (%llu)\n", + nr_zones); + ret = -EINVAL; + goto err; + } sdkp->early_zone_info.nr_zones = nr_zones; sdkp->early_zone_info.zone_blocks = zone_blocks; From c774ec8f0a5d02a06d34c27f5a7de7e333b91265 Mon Sep 17 00:00:00 2001 From: Eva Kurchatova Date: Wed, 16 Sep 2026 23:44:31 +0300 Subject: [PATCH 076/365] selftests: cgroup: give the O_TMPFILE open in get_temp_fd() a mode O_TMPFILE, like O_CREAT, needs the third argument. Without it glibc refuses the call at compile time as soon as fortification is on: In function 'open', inlined from 'get_temp_fd' at test_memcontrol.c:33:9: /usr/include/bits/fcntl2.h:52:11: error: call to '__open_missing_mode' declared with attribute error: open with O_CREAT or O_TMPFILE in second argument needs 3 arguments The fortify checks take effect only once the compiler optimises, and cgroup/Makefile builds with "-Wall -pthread" alone, so this goes unnoticed in a plain build. Building the tests with the flags distributions commonly use, -O2 -D_FORTIFY_SOURCE=3, loses test_memcontrol entirely. Fixes: 84092dbcf901 ("selftests: cgroup: add memory controller self-tests") Signed-off-by: Eva Kurchatova Signed-off-by: Tejun Heo --- tools/testing/selftests/cgroup/test_memcontrol.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tools/testing/selftests/cgroup/test_memcontrol.c b/tools/testing/selftests/cgroup/test_memcontrol.c index 3a84d068fbf3..0ed82347044e 100644 --- a/tools/testing/selftests/cgroup/test_memcontrol.c +++ b/tools/testing/selftests/cgroup/test_memcontrol.c @@ -30,7 +30,7 @@ static int page_size; int get_temp_fd(void) { - return open(".", O_TMPFILE | O_RDWR | O_EXCL); + return open(".", O_TMPFILE | O_RDWR | O_EXCL, 0600); } int alloc_pagecache(int fd, size_t size) From 42d1221d321e55afc7bba9109a77aaf5a817c8a3 Mon Sep 17 00:00:00 2001 From: Bart Van Assche Date: Mon, 31 Aug 2026 12:27:20 -0700 Subject: [PATCH 077/365] scsi: megaraid_sas: Protect megasas_get_ctrl_info() in megasas_resume() Protect the megasas_get_ctrl_info() call in megasas_resume() with instance->reset_mutex using scoped_guard(). megasas_get_ctrl_info() may release and reacquire instance->reset_mutex. Hence, calling this function without holding instance->reset_mutex is not safe. Fixes: c3b10a55abc9 ("scsi: megaraid_sas: Update controller info during resume") Cc: Kashyap Desai Cc: Sumit Saxena Cc: Shivasharan S Cc: Chandrakanth patil Signed-off-by: Bart Van Assche Link: https://patch.msgid.link/f06b5ee432b21cf293f0663e15b64f75a84b9fd5.1788204406.git.bvanassche@acm.org Signed-off-by: Martin K. Petersen (Oracle) --- drivers/scsi/megaraid/megaraid_sas_base.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/drivers/scsi/megaraid/megaraid_sas_base.c b/drivers/scsi/megaraid/megaraid_sas_base.c index b95f187297ae..4f9a53769966 100644 --- a/drivers/scsi/megaraid/megaraid_sas_base.c +++ b/drivers/scsi/megaraid/megaraid_sas_base.c @@ -7886,7 +7886,9 @@ megasas_resume(struct device *dev) goto fail_init_mfi; } - if (megasas_get_ctrl_info(instance) != DCMD_SUCCESS) + scoped_guard(mutex, &instance->reset_mutex) + rval = megasas_get_ctrl_info(instance); + if (rval != DCMD_SUCCESS) goto fail_init_mfi; tasklet_init(&instance->isr_tasklet, instance->instancet->tasklet, From 2f91fc9a96cdab6c03d436246056552e04677636 Mon Sep 17 00:00:00 2001 From: Hemanth Selam Date: Mon, 7 Sep 2026 12:08:56 +0530 Subject: [PATCH 078/365] s390: Fix typos in comments Fix typos in comments, reported by scripts/checkpatch.pl using the misspelling list in scripts/spelling.txt. Only touches comments, no code changes. Assisted-by: Cursor:claude-opus-5 Signed-off-by: Hemanth Selam Signed-off-by: Heiko Carstens --- drivers/s390/block/dasd_3990_erp.c | 6 +++--- drivers/s390/block/dasd_eckd.c | 4 ++-- drivers/s390/char/raw3270.c | 2 +- drivers/s390/char/vmlogrdr.c | 2 +- 4 files changed, 7 insertions(+), 7 deletions(-) diff --git a/drivers/s390/block/dasd_3990_erp.c b/drivers/s390/block/dasd_3990_erp.c index 736459477c19..121cb9ae5c67 100644 --- a/drivers/s390/block/dasd_3990_erp.c +++ b/drivers/s390/block/dasd_3990_erp.c @@ -103,7 +103,7 @@ dasd_3990_erp_int_req(struct dasd_ccw_req * erp) /* first time set initial retry counter and erp_function */ /* and retry once without blocking queue */ - /* (this enables easier enqueing of the cqr) */ + /* (this enables easier enqueuing of the cqr) */ if (erp->function != dasd_3990_erp_int_req) { erp->retries = 256; @@ -302,7 +302,7 @@ dasd_3990_erp_action_4(struct dasd_ccw_req * erp, char *sense) /* first time set initial retry counter and erp_function */ /* and retry once without waiting for state change pending */ - /* interrupt (this enables easier enqueing of the cqr) */ + /* interrupt (this enables easier enqueuing of the cqr) */ if (erp->function != dasd_3990_erp_action_4) { DBF_DEV_EVENT(DBF_INFO, device, "%s", @@ -1078,7 +1078,7 @@ dasd_3990_erp_bus_out(struct dasd_ccw_req * erp) /* first time set initial retry counter and erp_function */ /* and retry once without blocking queue */ - /* (this enables easier enqueing of the cqr) */ + /* (this enables easier enqueuing of the cqr) */ if (erp->function != dasd_3990_erp_bus_out) { erp->retries = 256; erp->function = dasd_3990_erp_bus_out; diff --git a/drivers/s390/block/dasd_eckd.c b/drivers/s390/block/dasd_eckd.c index 8d976dd58a6c..9d1b11020b4e 100644 --- a/drivers/s390/block/dasd_eckd.c +++ b/drivers/s390/block/dasd_eckd.c @@ -2060,7 +2060,7 @@ dasd_eckd_psf_ssc(struct dasd_device *device, int enable_pav, } /* - * Valide storage server of current device. + * Valid storage server of current device. */ static int dasd_eckd_validate_server(struct dasd_device *device, unsigned long flags) @@ -5672,7 +5672,7 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_raw(struct dasd_device *startdev, char *dst; /* - * raw track access needs to be mutiple of 64k and on 64k boundary + * raw track access needs to be multiple of 64k and on 64k boundary * For read requests we can fix an incorrect alignment by padding * the request with dummy pages. */ diff --git a/drivers/s390/char/raw3270.c b/drivers/s390/char/raw3270.c index aa9c4d81225c..6861bf27d15f 100644 --- a/drivers/s390/char/raw3270.c +++ b/drivers/s390/char/raw3270.c @@ -420,7 +420,7 @@ struct raw3270_ua { /* Query Reply structure for Usable Area */ char flags0; char flags1; short w; /* Width of usable area */ - short h; /* Heigth of usavle area */ + short h; /* Height of usavle area */ char units; /* 0x00:in; 0x01:mm */ int xr; int yr; diff --git a/drivers/s390/char/vmlogrdr.c b/drivers/s390/char/vmlogrdr.c index 383e7e2bd69f..005735a89920 100644 --- a/drivers/s390/char/vmlogrdr.c +++ b/drivers/s390/char/vmlogrdr.c @@ -444,7 +444,7 @@ static int vmlogrdr_receive_data(struct vmlogrdr_priv_t *priv) spin_unlock_bh(&priv->priv_lock); /* An rc of 5 indicates that the record was bigger than * the buffer, which is OK for us. A 9 indicates that the - * record was purged befor we could receive it. + * record was purged before we could receive it. */ if (rc == 5) rc = 0; From 4525a911049543c23885a540a788d13be318a486 Mon Sep 17 00:00:00 2001 From: Karl Mehltretter Date: Mon, 7 Sep 2026 07:58:47 +0200 Subject: [PATCH 079/365] s390/pci/docs: Fix sriov_numvfs attribute name The attribute is sriov_numvfs (drivers/pci/iov.c); the document names it sriov_numvf, which does not exist. Use sriov_numvfs. Fixes: de267a7c71ba ("s390/pci: Documentation for zPCI") Assisted-by: LLM Signed-off-by: Karl Mehltretter Reviewed-by: Randy Dunlap Signed-off-by: Heiko Carstens --- Documentation/arch/s390/pci.rst | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/Documentation/arch/s390/pci.rst b/Documentation/arch/s390/pci.rst index 80f4ba193159..565434626fb5 100644 --- a/Documentation/arch/s390/pci.rst +++ b/Documentation/arch/s390/pci.rst @@ -67,7 +67,7 @@ Entries specific to zPCI functions and entries that hold zPCI information. A physical function that currently supports a virtual function cannot be powered off until all virtual functions are removed with: - echo 0 > /sys/bus/pci/devices/DDDD:BB:dd.f/sriov_numvf + echo 0 > /sys/bus/pci/devices/DDDD:BB:dd.f/sriov_numvfs * /sys/bus/pci/devices/DDDD:BB:dd.f/: From 29d9e5835d89223aa913dcf7b942cc1c148bdd25 Mon Sep 17 00:00:00 2001 From: Vineeth Vijayan Date: Thu, 10 Sep 2026 11:32:01 +0200 Subject: [PATCH 080/365] s390/cio: Fix cio_update_schib() to not cache invalid schib When pmcw.dnv is 0, the contents of all SCHIB fields are unpredictable. Zero sch->schib in that case to prevent subsequent code from making decisions based on unpredictable data. Reported-by: William Bezenah Signed-off-by: Vineeth Vijayan Reviewed-by: Peter Oberparleiter Fixes: 8c58a229688c ("s390/cio: Do not unregister the subchannel based on DNV") Signed-off-by: Heiko Carstens --- drivers/s390/cio/cio.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-) diff --git a/drivers/s390/cio/cio.c b/drivers/s390/cio/cio.c index 70dc8cc76594..e1c62eb60cca 100644 --- a/drivers/s390/cio/cio.c +++ b/drivers/s390/cio/cio.c @@ -453,7 +453,8 @@ EXPORT_SYMBOL_GPL(cio_commit_config); /** * cio_update_schib - Perform stsch and update schib if subchannel is valid. * @sch: subchannel on which to perform stsch - * Return zero on success, -ENODEV otherwise. + * Return zero on success, -ENODEV if the subchannel is not operational, + * -EACCES if the subchannel has no valid device. */ int cio_update_schib(struct subchannel *sch) { @@ -462,10 +463,12 @@ int cio_update_schib(struct subchannel *sch) if (stsch(sch->schid, &schib)) return -ENODEV; - memcpy(&sch->schib, &schib, sizeof(schib)); - - if (!css_sch_is_valid(&schib)) + if (!css_sch_is_valid(&schib)) { + memset(&sch->schib, 0, sizeof(sch->schib)); return -EACCES; + } + + memcpy(&sch->schib, &schib, sizeof(schib)); return 0; } From f6f2985eabdb2bfdc82ce90a1ea3ec53ba795f34 Mon Sep 17 00:00:00 2001 From: Vineeth Vijayan Date: Thu, 10 Sep 2026 11:32:02 +0200 Subject: [PATCH 081/365] s390/cio: Check pmcw.dnv before pmcw.ena in I/O entry points The device number valid (dnv) bit in the PMCW must be checked before acting on any other PMCW fields for IO-type subchannels. A subchannel with dnv=0 has no valid device number associated, making it meaningless to evaluate the enabled (ena) state or issue any I/O instruction against it. Reported-by: William Bezenah Signed-off-by: Vineeth Vijayan Reviewed-by: Peter Oberparleiter Fixes: 8c58a229688c ("s390/cio: Do not unregister the subchannel based on DNV") Signed-off-by: Heiko Carstens --- drivers/s390/cio/device_ops.c | 12 ++++++++++++ 1 file changed, 12 insertions(+) diff --git a/drivers/s390/cio/device_ops.c b/drivers/s390/cio/device_ops.c index 61c07b4a0fe8..c1ba4a19368f 100644 --- a/drivers/s390/cio/device_ops.c +++ b/drivers/s390/cio/device_ops.c @@ -142,6 +142,8 @@ int ccw_device_clear(struct ccw_device *cdev, unsigned long intparm) if (!cdev || !cdev->dev.parent) return -ENODEV; sch = to_subchannel(cdev->dev.parent); + if (!sch->schib.pmcw.dnv) + return -ENODEV; if (!sch->schib.pmcw.ena) return -EINVAL; if (cdev->private->state == DEV_STATE_NOT_OPER) @@ -198,6 +200,8 @@ int ccw_device_start_timeout_key(struct ccw_device *cdev, struct ccw1 *cpa, if (!cdev || !cdev->dev.parent) return -ENODEV; sch = to_subchannel(cdev->dev.parent); + if (!sch->schib.pmcw.dnv) + return -ENODEV; if (!sch->schib.pmcw.ena) return -EINVAL; if (cdev->private->state == DEV_STATE_NOT_OPER) @@ -379,6 +383,8 @@ int ccw_device_halt(struct ccw_device *cdev, unsigned long intparm) if (!cdev || !cdev->dev.parent) return -ENODEV; sch = to_subchannel(cdev->dev.parent); + if (!sch->schib.pmcw.dnv) + return -ENODEV; if (!sch->schib.pmcw.ena) return -EINVAL; if (cdev->private->state == DEV_STATE_NOT_OPER) @@ -413,6 +419,8 @@ int ccw_device_resume(struct ccw_device *cdev) if (!cdev || !cdev->dev.parent) return -ENODEV; sch = to_subchannel(cdev->dev.parent); + if (!sch->schib.pmcw.dnv) + return -ENODEV; if (!sch->schib.pmcw.ena) return -EINVAL; if (cdev->private->state == DEV_STATE_NOT_OPER) @@ -548,6 +556,8 @@ int ccw_device_tm_start_timeout_key(struct ccw_device *cdev, struct tcw *tcw, int rc; sch = to_subchannel(cdev->dev.parent); + if (!sch->schib.pmcw.dnv) + return -ENODEV; if (!sch->schib.pmcw.ena) return -EINVAL; if (cdev->private->state == DEV_STATE_VERIFY) { @@ -694,6 +704,8 @@ int ccw_device_tm_intrg(struct ccw_device *cdev) { struct subchannel *sch = to_subchannel(cdev->dev.parent); + if (!sch->schib.pmcw.dnv) + return -ENODEV; if (!sch->schib.pmcw.ena) return -EINVAL; if (cdev->private->state != DEV_STATE_ONLINE) From 9590f4d83880dfb5a81906e48e72779248fbe8f0 Mon Sep 17 00:00:00 2001 From: Vineeth Vijayan Date: Thu, 10 Sep 2026 11:32:03 +0200 Subject: [PATCH 082/365] s390/cio: Guard PMCW field accesses with dnv check When PMCW.DNV is 0, no I/O device is associated with the subchannel. However, several code paths access PMCW fields directly from the cached sch->schib without first invoking the update helper. Add explicit DNV validation before accessing PMCW fields from the cached SCHIB to avoid using invalid data. Reported-by: William Bezenah Signed-off-by: Vineeth Vijayan Reviewed-by: Peter Oberparleiter Fixes: 8c58a229688c ("s390/cio: Do not unregister the subchannel based on DNV") Signed-off-by: Heiko Carstens --- drivers/s390/cio/chp.c | 3 +++ drivers/s390/cio/device.c | 9 +++++---- drivers/s390/cio/device_fsm.c | 3 +++ drivers/s390/cio/device_ops.c | 9 +++++++++ drivers/s390/cio/vfio_ccw_fsm.c | 2 +- 5 files changed, 21 insertions(+), 5 deletions(-) diff --git a/drivers/s390/cio/chp.c b/drivers/s390/cio/chp.c index c890f21a82ce..eaf0527bff6c 100644 --- a/drivers/s390/cio/chp.c +++ b/drivers/s390/cio/chp.c @@ -78,6 +78,9 @@ u8 chp_get_sch_opm(struct subchannel *sch) int opm; int i; + if (!sch->schib.pmcw.dnv) + return 0; + opm = 0; chp_id_init(&chpid); for (i = 0; i < 8; i++) { diff --git a/drivers/s390/cio/device.c b/drivers/s390/cio/device.c index fb591118ecb2..68dd4a62975d 100644 --- a/drivers/s390/cio/device.c +++ b/drivers/s390/cio/device.c @@ -922,7 +922,7 @@ static int ccw_device_move_to_sch(struct ccw_device *cdev, if (!sch_is_pseudo_sch(old_sch)) { spin_lock_irq(&old_sch->lock); - old_enabled = old_sch->schib.pmcw.ena; + old_enabled = old_sch->schib.pmcw.dnv && old_sch->schib.pmcw.ena; rc = 0; if (old_enabled) rc = cio_disable_subchannel(old_sch); @@ -941,7 +941,7 @@ static int ccw_device_move_to_sch(struct ccw_device *cdev, CIO_MSG_EVENT(0, "device_move(0.%x.%04x,0.%x.%04x)=%d\n", cdev->private->dev_id.ssid, cdev->private->dev_id.devno, sch->schid.ssid, - sch->schib.pmcw.dev, rc); + sch->schid.sch_no, rc); if (old_enabled) { /* Try to re-enable the old subchannel. */ spin_lock_irq(&old_sch->lock); @@ -1207,7 +1207,7 @@ static void io_subchannel_quiesce(struct subchannel *sch) cdev = sch_get_cdev(sch); if (cio_is_console(sch->schid)) goto out_unlock; - if (!sch->schib.pmcw.ena) + if (!sch->schib.pmcw.dnv || !sch->schib.pmcw.ena) goto out_unlock; ret = cio_disable_subchannel(sch); if (ret != -EBUSY) @@ -1254,7 +1254,8 @@ static int recovery_check(struct device *dev, void *data) switch (cdev->private->state) { case DEV_STATE_ONLINE: sch = to_subchannel(cdev->dev.parent); - if ((sch->schib.pmcw.pam & sch->opm) == sch->vpm) + if (sch->schib.pmcw.dnv && + (sch->schib.pmcw.pam & sch->opm) == sch->vpm) break; fallthrough; case DEV_STATE_DISCONNECTED: diff --git a/drivers/s390/cio/device_fsm.c b/drivers/s390/cio/device_fsm.c index ab419d40a8a7..b5686c25c83c 100644 --- a/drivers/s390/cio/device_fsm.c +++ b/drivers/s390/cio/device_fsm.c @@ -170,6 +170,9 @@ __recover_lost_chpids(struct subchannel *sch, int old_lpm) int mask, i; struct chp_id chpid; + if (!sch->schib.pmcw.dnv) + return; + chp_id_init(&chpid); for (i = 0; i<8; i++) { mask = 0x80 >> i; diff --git a/drivers/s390/cio/device_ops.c b/drivers/s390/cio/device_ops.c index c1ba4a19368f..f2f7f8cba410 100644 --- a/drivers/s390/cio/device_ops.c +++ b/drivers/s390/cio/device_ops.c @@ -490,6 +490,8 @@ struct channel_path_desc_fmt0 *ccw_device_get_chp_desc(struct ccw_device *cdev, struct chp_id chpid; sch = to_subchannel(cdev->dev.parent); + if (!sch->schib.pmcw.dnv) + return NULL; chp_id_init(&chpid); chpid.id = sch->schib.pmcw.chpid[chp_idx]; return chp_get_chp_desc(chpid); @@ -510,6 +512,8 @@ u8 *ccw_device_get_util_str(struct ccw_device *cdev, int chp_idx) struct chp_id chpid; u8 *util_str; + if (!sch->schib.pmcw.dnv) + return NULL; chp_id_init(&chpid); chpid.id = sch->schib.pmcw.chpid[chp_idx]; chp = chpid_to_chp(chpid); @@ -662,6 +666,9 @@ int ccw_device_get_mdc(struct ccw_device *cdev, u8 mask) struct chp_id chpid; int mdc = 0, i; + if (!sch->schib.pmcw.dnv) + return 0; + /* Adjust requested path mask to excluded varied off paths. */ if (mask) mask &= sch->lpm; @@ -798,6 +805,8 @@ int ccw_device_get_chpid(struct ccw_device *cdev, int chp_idx, u8 *chpid) if ((chp_idx < 0) || (chp_idx > 7)) return -EINVAL; + if (!sch->schib.pmcw.dnv) + return -ENODEV; mask = 0x80 >> chp_idx; if (!(sch->schib.pmcw.pim & mask)) return -ENODEV; diff --git a/drivers/s390/cio/vfio_ccw_fsm.c b/drivers/s390/cio/vfio_ccw_fsm.c index 5fd94e9d5c61..9a000b0231d6 100644 --- a/drivers/s390/cio/vfio_ccw_fsm.c +++ b/drivers/s390/cio/vfio_ccw_fsm.c @@ -399,7 +399,7 @@ static void fsm_close(struct vfio_ccw_private *private, spin_lock_irq(&sch->lock); - if (!sch->schib.pmcw.ena) + if (!sch->schib.pmcw.dnv || !sch->schib.pmcw.ena) goto err_unlock; ret = cio_disable_subchannel(sch); From fc3ae66514ca5e87251f79044236e3e8babd24a3 Mon Sep 17 00:00:00 2001 From: David Howells Date: Mon, 14 Sep 2026 16:20:27 +0100 Subject: [PATCH 083/365] netfs: Fix netfs_read_gaps() to use separate sink folios Fix netfs_read_gaps() to use separate folios rather than re-using a single sink folio to discard the unwanted data so that cifs checksum checking sees all the data that was fetched. Fixes: 7f84a7b9892d ("netfs: Make netfs_read_folio() handle streaming-write pages") Reported-by: Frank Sorenson Closes: https://lore.kernel.org/r/a385053c-1c4a-4060-a3bb-befa007ddb33@redhat.com/ Signed-off-by: David Howells Link: https://patch.msgid.link/3228134.1789399227@warthog.procyon.org.uk Tested-by: Frank Sorenson Reviewed-by: Paulo Alcantara cc: Paulo Alcantara cc: Namjae Jeon cc: netfs@lists.linux.dev cc: linux-cifs@vger.kernel.org cc: linux-fsdevel@vger.kernel.org Signed-off-by: Christian Brauner (Amutable) --- fs/netfs/buffered_read.c | 34 +++++++++++++++++++--------------- 1 file changed, 19 insertions(+), 15 deletions(-) diff --git a/fs/netfs/buffered_read.c b/fs/netfs/buffered_read.c index 424df70a5c30..105194de6e13 100644 --- a/fs/netfs/buffered_read.c +++ b/fs/netfs/buffered_read.c @@ -482,15 +482,14 @@ static int netfs_read_gaps(struct file *file, struct folio *folio) struct netfs_group *group = netfs_folio_group(folio); struct netfs_folio *finfo = netfs_folio_info(folio); struct netfs_inode *ctx = netfs_inode(mapping->host); - struct folio *sink = NULL; - struct bio_vec *bvec; + struct bio_vec *bvec = NULL; unsigned int from = finfo->dirty_offset; unsigned int to = from + finfo->dirty_len; - unsigned int off = 0, i = 0; + unsigned int off = 0; size_t flen = folio_size(folio); size_t nr_bvec = flen / PAGE_SIZE + 2; size_t part; - int ret; + int ret, i = 0, sink_from = -1, sink_to = -1; _enter("%lx", folio->index); @@ -515,24 +514,23 @@ static int netfs_read_gaps(struct file *file, struct folio *folio) if (!bvec) goto discard; - sink = folio_alloc(GFP_KERNEL, 0); - if (!sink) { - kfree(bvec); - goto discard; - } - trace_netfs_folio(folio, netfs_folio_trace_read_gaps); - rreq->direct_bv = bvec; - rreq->direct_bv_count = nr_bvec; if (from > 0) { bvec_set_folio(&bvec[i++], folio, from, 0); off = from; } + sink_from = i; while (off < to) { + struct folio *sink = folio_alloc(GFP_KERNEL, 0); + + if (!sink) + goto discard; part = min_t(size_t, to - off, PAGE_SIZE); - bvec_set_folio(&bvec[i++], sink, part, 0); + bvec_set_folio(&bvec[i], sink, part, 0); off += part; + sink_to = i; + i++; } if (to < flen) bvec_set_folio(&bvec[i++], folio, flen - to, to); @@ -553,8 +551,10 @@ static int netfs_read_gaps(struct file *file, struct folio *folio) folio_mark_uptodate(folio); } - if (sink) - folio_put(sink); + if (sink_to >= 0) + for (; sink_from <= sink_to; sink_from++) + folio_put(bvec_folio(&bvec[sink_from])); + kfree(bvec); folio_unlock(folio); netfs_put_request(rreq, netfs_rreq_trace_put_return); return ret < 0 ? ret : 0; @@ -563,6 +563,10 @@ static int netfs_read_gaps(struct file *file, struct folio *folio) netfs_put_failed_request(rreq); alloc_error: folio_unlock(folio); + if (sink_to >= 0) + for (; sink_from <= sink_to; sink_from++) + folio_put(bvec_folio(&bvec[sink_from])); + kfree(bvec); return ret; } From e9d810279f84b30738f7790c0ed15f8dd5b9024a Mon Sep 17 00:00:00 2001 From: Wentao Liang Date: Wed, 16 Sep 2026 09:47:01 +0000 Subject: [PATCH 084/365] gpio: arizona: Fix runtime PM leak in arizona_gpio_direction_out() Switching a persistent GPIO line from input to output acquires a runtime PM reference on the parent device, but if the subsequent regmap_update_bits() fails the reference is never dropped and no later direction_in() can balance it since the direction was never changed. Drop the reference on the update failure path. Fixes: 27a49ed17e22 ("gpio: arizona: Add support for GPIOs that need to be maintained") Cc: stable@vger.kernel.org Signed-off-by: Wentao Liang Reviewed-by: Charles Keepax Link: https://patch.msgid.link/20260916094701.2007509-1-vulab@iscas.ac.cn Signed-off-by: Bartosz Golaszewski --- drivers/gpio/gpio-arizona.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/drivers/gpio/gpio-arizona.c b/drivers/gpio/gpio-arizona.c index a7e98d395d8e..88cf013f0d90 100644 --- a/drivers/gpio/gpio-arizona.c +++ b/drivers/gpio/gpio-arizona.c @@ -115,8 +115,12 @@ static int arizona_gpio_direction_out(struct gpio_chip *chip, if (value) value = ARIZONA_GPN_LVL; - return regmap_update_bits(arizona->regmap, ARIZONA_GPIO1_CTRL + offset, - ARIZONA_GPN_DIR | ARIZONA_GPN_LVL, value); + ret = regmap_update_bits(arizona->regmap, ARIZONA_GPIO1_CTRL + offset, + ARIZONA_GPN_DIR | ARIZONA_GPN_LVL, value); + if (ret < 0 && (val & ARIZONA_GPN_DIR) && persistent) + pm_runtime_put_autosuspend(chip->parent); + + return ret; } static int arizona_gpio_set(struct gpio_chip *chip, unsigned int offset, From 02af7eac17bc62a72335c1fc8c4af4471654f6cc Mon Sep 17 00:00:00 2001 From: Rosen Penev Date: Mon, 14 Sep 2026 17:44:06 -0700 Subject: [PATCH 085/365] gpio: mvebu: keep resume masks within the irqchip cache mvebu_gpio_resume() writes the edge/level mask registers saved at suspend time straight back to hardware, bypassing the irqchip's mask_cache_priv. genirq skips mask_irq() for a line it already considers masked, so restoring a bit in hardware that genirq thinks is still masked leaves that line unmasked behind genirq's back. An asserted level line then has nobody to ack it, and the moment interrupts are re-enabled the chained handler storms, hanging resume. AND the restored mask values with the matching irqchip mask cache so only lines genirq currently considers unmasked are unmasked again. Read the caches under gc->lock to keep them consistent with the mask/unmask handlers. Tested on Helios4 (armhf): 5 suspend cycles woken by magic packet, no hang; mvebu_gpio_resume() returns in 6 usecs. Assisted-by: LLM Signed-off-by: Rosen Penev Link: https://patch.msgid.link/20260915004406.115230-1-rosenp@gmail.com Signed-off-by: Bartosz Golaszewski --- drivers/gpio/gpio-mvebu.c | 33 +++++++++++++++++++++++++++------ 1 file changed, 27 insertions(+), 6 deletions(-) diff --git a/drivers/gpio/gpio-mvebu.c b/drivers/gpio/gpio-mvebu.c index 93b8a08b04b9..a2796240fe14 100644 --- a/drivers/gpio/gpio-mvebu.c +++ b/drivers/gpio/gpio-mvebu.c @@ -1034,6 +1034,8 @@ static int mvebu_gpio_suspend(struct platform_device *pdev, pm_message_t state) static int mvebu_gpio_resume(struct platform_device *pdev) { struct mvebu_gpio_chip *mvchip = platform_get_drvdata(pdev); + u32 edge_cache = ~0U, level_cache = ~0U; + unsigned long flags; int i; regmap_write(mvchip->regs, GPIO_OUT_OFF + mvchip->offset, @@ -1045,32 +1047,51 @@ static int mvebu_gpio_resume(struct platform_device *pdev) regmap_write(mvchip->regs, GPIO_IN_POL_OFF + mvchip->offset, mvchip->in_pol_reg); + /* + * genirq skips mask_irq() for a line it already considers masked, so + * unmasking one behind its back leaves an asserted level line that + * nobody masks. Restore only bits the irqchip cache still has set. + * + * Snapshot the caches under the raw spinlock, but release it before + * the regmap writes below: regmap_write() takes a sleepable lock on + * PREEMPT_RT. + */ + if (mvchip->domain) { + struct irq_chip_generic *gc; + + gc = irq_get_domain_generic_chip(mvchip->domain, 0); + raw_spin_lock_irqsave(&gc->lock, flags); + level_cache = gc->chip_types[0].mask_cache_priv; + edge_cache = gc->chip_types[1].mask_cache_priv; + raw_spin_unlock_irqrestore(&gc->lock, flags); + } + switch (mvchip->soc_variant) { case MVEBU_GPIO_SOC_VARIANT_ORION: case MVEBU_GPIO_SOC_VARIANT_A8K: regmap_write(mvchip->regs, GPIO_EDGE_MASK_OFF + mvchip->offset, - mvchip->edge_mask_regs[0]); + mvchip->edge_mask_regs[0] & edge_cache); regmap_write(mvchip->regs, GPIO_LEVEL_MASK_OFF + mvchip->offset, - mvchip->level_mask_regs[0]); + mvchip->level_mask_regs[0] & level_cache); break; case MVEBU_GPIO_SOC_VARIANT_MV78200: for (i = 0; i < 2; i++) { regmap_write(mvchip->regs, GPIO_EDGE_MASK_MV78200_OFF(i), - mvchip->edge_mask_regs[i]); + mvchip->edge_mask_regs[i] & edge_cache); regmap_write(mvchip->regs, GPIO_LEVEL_MASK_MV78200_OFF(i), - mvchip->level_mask_regs[i]); + mvchip->level_mask_regs[i] & level_cache); } break; case MVEBU_GPIO_SOC_VARIANT_ARMADAXP: for (i = 0; i < 4; i++) { regmap_write(mvchip->regs, GPIO_EDGE_MASK_ARMADAXP_OFF(i), - mvchip->edge_mask_regs[i]); + mvchip->edge_mask_regs[i] & edge_cache); regmap_write(mvchip->regs, GPIO_LEVEL_MASK_ARMADAXP_OFF(i), - mvchip->level_mask_regs[i]); + mvchip->level_mask_regs[i] & level_cache); } break; default: From c51e89c5b5db3e1927738fad7cd18b66dde68aed Mon Sep 17 00:00:00 2001 From: David Howells Date: Tue, 15 Sep 2026 17:21:59 +0100 Subject: [PATCH 086/365] netfs, afs: Fix symlink reading Fix the reading of symlinks from the cache in afs by making netfslib trim the amount read down to i_size. The problem is that afs sets the size of the iterator to the size of the buffer (PAGE_SIZE) so that the cache can round the read size up to the cache's DIO size. Note that this also impacts the reading of AFS mountpoints as they're just stored as symlinks with an odd file mode. Link: https://patch.msgid.link/3912795.1789489319@warthog.procyon.org.uk Fixes: c0410adf3da6 ("afs: Fix the locking used by afs_get_link()") Reviewed-by: Paulo Alcantara cc: Paulo Alcantara cc: Marc Dionne cc: linux-afs@lists.infradead.org cc: netfs@lists.linux.dev cc: linux-fsdevel@vger.kernel.org Signed-off-by: Christian Brauner (Amutable) --- fs/netfs/read_collect.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/fs/netfs/read_collect.c b/fs/netfs/read_collect.c index 5cf22087d243..a94197ef0181 100644 --- a/fs/netfs/read_collect.c +++ b/fs/netfs/read_collect.c @@ -435,6 +435,11 @@ static void netfs_rreq_assess_single(struct netfs_io_request *rreq) netfs_single_mark_inode_dirty(rreq->inode); } + /* To do DIO, the cache has to round the size up, so we need to undo + * the rounding. + */ + rreq->transferred = min(rreq->transferred, rreq->i_size); + if (rreq->iocb) { rreq->iocb->ki_pos += rreq->transferred; if (rreq->iocb->ki_complete) { From 7459c021874246c196f397686e100702f059b9e7 Mon Sep 17 00:00:00 2001 From: Julian Sun Date: Tue, 15 Sep 2026 12:49:12 +0800 Subject: [PATCH 087/365] fs: avoid repeated scans in evict_inodes() We observed hung tasks when users attempted to unmount a filesystem after its disk had been removed while still in use. During device removal, fs_bdev_mark_dead() calls evict_inodes() while holding s_umount. Each time evict_inodes() drops s_inode_list_lock to reschedule, it restarts the walk from the head of s_inodes. With many referenced inodes at the head of the list, these restarts repeatedly scan the same inodes without reclaiming them. This can keep s_umount held for a long time, blocking concurrent umount attempts and triggering hung-task reports. Keep the current inode, already marked I_FREEING, out of the disposal batch until s_inode_list_lock is reacquired. Resume the walk from this inode and dispose of it in a later batch or at the end of the walk. The zero-refcount and state checks under i_lock allow this walker to claim the inode by setting I_FREEING and removing it from the LRU. Other reclaimers skip the inode, leaving this walker responsible for eviction. Only evict() removes it from s_inodes, so keeping it out of the disposal batch ensures that it remains on the list while the lock is dropped. After reacquiring the lock, reading its current next pointer accounts for concurrent removal of following inodes. The existing inode lifetime rules prohibit acquiring a reference to an inode marked I_FREEING or I_WILL_FREE. __iget() requires its caller to hold i_lock and establish that taking a reference is valid. Inode lookup and igrab() check these flags under i_lock when acquiring a reference from zero. ihold() requires an existing reference, which would keep i_count nonzero and prevent this walker from claiming the inode. These rules already allow iput_final() and the inode shrinker to release i_lock after setting I_FREEING and before eviction completes. A temporary __iget() reference would also keep the inode on the list, but its release must preserve last-reference handling. Another user can acquire a reference, update lazy timestamps and drop its reference while the pin is held. If the pin becomes the last reference, dropping it with atomic_dec_and_test() and evicting directly bypasses iput()'s lazytime handling and can lose those timestamp updates. Releasing the pin with iput() preserves that handling, but does not guarantee eviction. fs_bdev_mark_dead() runs with SB_ACTIVE set, so iput() may retain the inode in cache, whereas evict_inodes() must evict eligible zero-reference inodes. The inode may also have been freed when iput() returns, so the walker cannot then use it to force eviction. Using I_FREEING preserves the existing eviction behavior without introducing an additional last-reference transition. The xfstests auto group passed on ext4 and XFS with known unrelated failures excluded. No new issues were observed, and the previously reproducible hung task no longer occurs with this patch. Fixes: ac05fbb40062 ("inode: don't softlockup when evicting inodes") Signed-off-by: Julian Sun Link: https://patch.msgid.link/20260915044912.3183440-1-sunjunchao@bytedance.com Reviewed-by: Jan Kara Signed-off-by: Christian Brauner (Amutable) --- fs/inode.c | 11 +++++------ 1 file changed, 5 insertions(+), 6 deletions(-) diff --git a/fs/inode.c b/fs/inode.c index ba7da39be4a3..a9d37be390a1 100644 --- a/fs/inode.c +++ b/fs/inode.c @@ -880,7 +880,6 @@ void evict_inodes(struct super_block *sb) struct inode *inode; LIST_HEAD(dispose); -again: spin_lock(&sb->s_inode_list_lock); list_for_each_entry(inode, &sb->s_inodes, i_sb_list) { if (icount_read_once(inode)) @@ -899,19 +898,19 @@ void evict_inodes(struct super_block *sb) inode_state_set(inode, I_FREEING); inode_lru_list_del(inode); spin_unlock(&inode->i_lock); - list_add(&inode->i_lru, &dispose); /* - * We can have a ton of inodes to evict at unmount time given - * enough memory, check to see if we need to go to sleep for a - * bit so we don't livelock. + * Keep this inode out of dispose so it stays on s_inodes while + * the list lock is dropped. I_FREEING prevents new references + * and leaves eviction to us, so we can resume the walk from it. */ if (need_resched()) { spin_unlock(&sb->s_inode_list_lock); cond_resched(); dispose_list(&dispose); - goto again; + spin_lock(&sb->s_inode_list_lock); } + list_add(&inode->i_lru, &dispose); } spin_unlock(&sb->s_inode_list_lock); From f6988c90671e83db79df1b7b9d6fdb0e5947fd84 Mon Sep 17 00:00:00 2001 From: "Patrick Lu (Anthropic)" Date: Fri, 11 Sep 2026 18:49:49 +0000 Subject: [PATCH 088/365] writeback: bound cleanup_offline_cgwb() rescans by rotating scanned inodes cleanup_offline_cgwb() prepares at most WB_MAX_INODES_PER_ISW inodes per call and is called again until the dying wb is drained, but every call walks wb->b_attached and then wb->b_dirty_time from the same end. Inodes already prepared (they stay on the list with I_WB_SWITCH set until the switch worker runs) and inodes that cannot be switched (I_FREEING, I_WILL_FREE, !SB_ACTIVE, DAX, already on the target wb) stay where they are, so each pass rescans a growing run of them under wb->list_lock and a full drain is quadratic in the number of inodes on the list. With ~17M inodes attached to one dying cgwb we saw this end in soft lockups, with CPUs reported stuck for 21-48s. Walk both lists from the oldest end and move every scanned inode to the newest end, so the next pass starts where the previous one stopped and the drain becomes linear. b_attached is unordered, so nobody sees the reorder there. b_dirty_time is ordered by dirtied_when, but the oldest unscanned inode stays at the end move_expired_inodes() picks from, sync takes the whole list regardless of order, and prepared inodes leave the list as soon as the switch work runs and get a new dirtied_time_when on the new wb anyway, so the only inodes left out of order are the ones that can never switch (DAX), and only on the dying wb. Fixes: c22d70a162d3 ("writeback, cgroup: release dying cgwbs by switching attached inodes") Cc: stable@vger.kernel.org Acked-by: Tejun Heo Acked-by: Roman Gushchin Signed-off-by: Patrick Lu (Anthropic) Link: https://patch.msgid.link/20260911-wb-cgwb-rotate-v2-1-a9ab253a1295@gmail.com Reviewed-by: Jan Kara Signed-off-by: Christian Brauner (Amutable) --- fs/fs-writeback.c | 25 ++++++++++++++++++++----- 1 file changed, 20 insertions(+), 5 deletions(-) diff --git a/fs/fs-writeback.c b/fs/fs-writeback.c index e744f9f9d43f..ea3eb40bf828 100644 --- a/fs/fs-writeback.c +++ b/fs/fs-writeback.c @@ -727,19 +727,34 @@ static bool isw_prepare_wbs_switch(struct bdi_writeback *new_wb, struct inode_switch_wbs_context *isw, struct list_head *list, int *nr) { - struct inode *inode; + struct inode *inode, *tmp; + LIST_HEAD(scanned); + bool full = false; + + /* + * Walk from the oldest end and move scanned inodes to the newest + * end, so the next scan resumes at unscanned inodes instead of + * re-walking an ever-growing run of prepared and skipped ones. + * For b_dirty_time this keeps the oldest unscanned inode at the + * end move_expired_inodes() picks from; b_attached is unordered. + */ + list_for_each_entry_safe_reverse(inode, tmp, list, i_io_list) { + list_move(&inode->i_io_list, &scanned); - list_for_each_entry(inode, list, i_io_list) { if (!inode_prepare_wbs_switch(inode, new_wb)) continue; isw->inodes[*nr] = inode; (*nr)++; - if (*nr >= WB_MAX_INODES_PER_ISW - 1) - return true; + if (*nr >= WB_MAX_INODES_PER_ISW - 1) { + full = true; + break; + } } - return false; + list_splice(&scanned, list); + + return full; } /** From 1fca688e9443003e33cf30453e7a7560367656c9 Mon Sep 17 00:00:00 2001 From: shechenglong Date: Mon, 7 Sep 2026 11:51:47 +0800 Subject: [PATCH 089/365] drm/client: fix restore of partially initialized client I got a null-ptr-deref report when closing a DRM file descriptor: WARNING: drivers/gpu/drm/drm_atomic.c:2031 at __drm_atomic_helper_set_config+0x18e/0x1b0 [drm] Call Trace: drm_client_modeset_commit_atomic+0x16b/0x220 [drm] drm_client_modeset_commit_locked+0x56/0x160 [drm] drm_client_modeset_commit+0x21/0x40 [drm] __drm_fb_helper_restore_fbdev_mode_unlocked.part.0+0x7b/0x80 drm_fbdev_client_restore+0xe/0x20 [drm_client_lib] drm_client_dev_restore+0x9f/0xc0 [drm] drm_release+0xc5/0xe0 [drm] The warning is followed by a NULL pointer dereference: BUG: kernel NULL pointer dereference, address: 0000000000000008 RIP: __drm_fb_helper_restore_fbdev_mode_unlocked.part.0+0x41/0x80 [drm_kms_helper] Call Trace: drm_fbdev_client_restore+0xe/0x20 [drm_client_lib] drm_client_dev_restore+0x9f/0xc0 [drm] drm_release+0xc5/0xe0 [drm] __fput+0xdc/0x2b0 __x64_sys_close+0x39/0x80 do_syscall_64+0x8d/0x460 entry_SYSCALL_64_after_hwframe+0x76/0x7e drm_client_register() adds the DRM client to the device client list before invoking the initial hotplug callback. If the hotplug callback fails, the client remains registered. For the fbdev client, a failure during drm_fb_helper_initial_config() causes the partially initialized fbdev helper to be cleaned up. drm_fb_helper_fini() releases fb_helper->info and leaves it NULL. The fbdev client therefore remains registered even though there is no fully initialized framebuffer device. Later, when userspace closes the DRM file descriptor, drm_release() can invoke the restore callbacks of registered DRM clients: drm_release() drm_client_dev_restore() drm_fbdev_client_restore() drm_fb_helper_restore_fbdev_mode_unlocked() drm_fbdev_client_restore() currently restores the fbdev state unconditionally. For a partially initialized fbdev client this can submit an incomplete modeset state and subsequently access fbdev state which has not been initialized, resulting in the warning and NULL pointer dereference above. drm_fbdev_client_unregister() already uses fb_helper->info to distinguish a fully probed framebuffer device from a partially initialized client. Use the same condition in drm_fbdev_client_restore() and skip restore if no framebuffer device has been successfully initialized. Signed-off-by: shechenglong Reviewed-by: Thomas Zimmermann Fixes: 5d08c44e47b9 ("drm/fbdev: Add memory-agnostic fbdev client") Signed-off-by: Thomas Zimmermann Cc: # v6.13+ Link: https://patch.msgid.link/20260907035147.1339-1-shechenglong@xfusion.com --- drivers/gpu/drm/clients/drm_fbdev_client.c | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/drivers/gpu/drm/clients/drm_fbdev_client.c b/drivers/gpu/drm/clients/drm_fbdev_client.c index 91d196a397cf..1c16bc1084c4 100644 --- a/drivers/gpu/drm/clients/drm_fbdev_client.c +++ b/drivers/gpu/drm/clients/drm_fbdev_client.c @@ -42,6 +42,14 @@ static int drm_fbdev_client_restore(struct drm_client_dev *client, bool force) { struct drm_fb_helper *fb_helper = drm_fb_helper_from_client(client); + /* + * The client is registered before the initial fbdev probe. + * If probing failed, the client remains registered but there + * is no valid fbdev framebuffer to restore. + */ + if (!fb_helper->info || !fb_helper->fb) + return 0; + drm_fb_helper_restore_fbdev_mode_unlocked(fb_helper, force); return 0; From df5cdc2c832ca4e8a6d774596b9005558761a403 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Wed, 16 Sep 2026 12:00:24 -1000 Subject: [PATCH 090/365] sched_ext: Derive SCX_RQ_IN_WAKEUP from the core enqueue flags schedule_deferred_locked() skips scheduling a deferred action while SCX_RQ_IN_WAKEUP is set and relies on the task_woken_scx() call that follows a wakeup enqueue to run it. enqueue_task_scx() sets the flag from the merged enqueue flags, which include the flags stashed for a remote activation. move_remote_task_to_local_dsq() thus sets SCX_RQ_IN_WAKEUP on the destination rq when the moved task was woken up, although no task_woken_scx() follows that activation. An IMMED insert into a busy destination requests a local reenqueue during that enqueue. The request gets linked but not scheduled and stays pending until an unrelated wakeup or preemption on that CPU runs the deferred actions. The IMMED task sits behind the running task in the meantime. If nothing runs them before the scheduler is disabled, the request outlives the scheduler and points into its freed per-cpu area, which the next scheduler dereferences from run_deferred(). Test the core enqueue flags for the wakeup bit. Only the core's wakeup path is followed by task_woken_scx(). Fixes: 57ccf5ccdc56 ("sched_ext: Fix enqueue_task_scx() truncation of upper enqueue flags") Cc: stable@vger.kernel.org # v7.1+ Reported-by: Andrea Righi Link: https://lore.kernel.org/all/20260916145807.3250167-1-arighi@nvidia.com/ Signed-off-by: Tejun Heo Reviewed-by: Andrea Righi --- kernel/sched/ext/ext.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/kernel/sched/ext/ext.c b/kernel/sched/ext/ext.c index 725b890fcab9..e207ccd17164 100644 --- a/kernel/sched/ext/ext.c +++ b/kernel/sched/ext/ext.c @@ -2141,7 +2141,12 @@ static void enqueue_task_scx(struct rq *rq, struct task_struct *p, int core_enq_ int sticky_cpu = p->scx.sticky_cpu; u64 enq_flags = core_enq_flags | rq->scx.remote_activate_enq_flags; - if (enq_flags & ENQUEUE_WAKEUP) + /* + * SCX_RQ_IN_WAKEUP promises a task_woken_scx() call once this enqueue + * returns. Only the core's wakeup path delivers one. The flags stashed + * for a remote activation may carry the wakeup bit without it. + */ + if (core_enq_flags & ENQUEUE_WAKEUP) rq->scx.flags |= SCX_RQ_IN_WAKEUP; /* From 67b4411538c8341692548429d43256f25be99f7a Mon Sep 17 00:00:00 2001 From: Wentao Liang Date: Wed, 16 Sep 2026 18:00:36 +0000 Subject: [PATCH 091/365] drm/nouveau: Fix bridge reference leak in nv1a_ram_new() pci_get_domain_bus_and_slot() takes a reference to the PCI device, which is never released once the memory size has been read from its config space. Drop the reference before returning. Fixes: 2fa6d6cdaf283c05 ("drm/nouveau: deprecate pci_get_bus_and_slot()") Cc: stable@vger.kernel.org Signed-off-by: Wentao Liang Reviewed-by: Lyude Paul Signed-off-by: Lyude Paul Link: https://patch.msgid.link/20260916180036.2090118-1-vulab@iscas.ac.cn --- drivers/gpu/drm/nouveau/nvkm/subdev/fb/ramnv1a.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ramnv1a.c b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ramnv1a.c index 18241c6ba5fa..4d52a158f320 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ramnv1a.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ramnv1a.c @@ -51,6 +51,8 @@ nv1a_ram_new(struct nvkm_fb *fb, struct nvkm_ram **pram) mib = ((mem >> 4) & 127) + 1; } + pci_dev_put(bridge); + return nvkm_ram_new_(&nv04_ram_func, fb, NVKM_RAM_TYPE_STOLEN, mib * 1024 * 1024, pram); } From cb86607ada73185f321baa7f9d94a08a3a40bbf5 Mon Sep 17 00:00:00 2001 From: fangqiurong Date: Thu, 17 Sep 2026 15:52:41 +0800 Subject: [PATCH 092/365] sched_ext: Don't run ops.dequeue() with a DSQ lock held ops.dequeue() is invoked with the source user DSQ's lock still held on the consume and move paths (scx_consume_dispatch_q(), move_task_between_dsqs()). A BPF scheduler which locks the source user DSQ from ops.dequeue() - e.g. by iterating it with bpf_iter_scx_dsq - self-deadlocks. ops.dequeue() can only call the "any" kfuncs and none of them can lock a builtin DSQ, so the global and bypass paths can't deadlock; however, all DSQ locks share one lockdep class, so iterating any user DSQ from ops.dequeue() on those paths trips the recursion check. Move the invocation after the DSQ unlock on all three paths. SCX_TASK_IN_CUSTODY is cleared under the lock serializing the transfer so that the callback is invoked exactly once. Fixes: ebf1ccff79c4 ("sched_ext: Fix ops.dequeue() semantics") Cc: stable@vger.kernel.org # v7.1+ Acked-by: Andrea Righi Signed-off-by: fangqiurong Signed-off-by: Tejun Heo --- kernel/sched/ext/ext.c | 44 ++++++++++++++++++------------------- kernel/sched/ext/internal.h | 3 +-- kernel/sched/ext/sub.c | 10 ++++----- 3 files changed, 28 insertions(+), 29 deletions(-) diff --git a/kernel/sched/ext/ext.c b/kernel/sched/ext/ext.c index e207ccd17164..f568fd9973f6 100644 --- a/kernel/sched/ext/ext.c +++ b/kernel/sched/ext/ext.c @@ -1499,27 +1499,22 @@ static inline bool task_scx_migrating(struct task_struct *p) return p->scx.sticky_cpu >= 0; } -/* - * Call ops.dequeue() if the task is in BPF custody and not migrating. - * Clears %SCX_TASK_IN_CUSTODY when the callback is invoked. - */ -static void call_task_dequeue(struct scx_sched *sch, struct rq *rq, - struct task_struct *p, u64 deq_flags) +/* Must be called under the lock serializing @p's custody transfers. */ +static bool task_leave_custody(struct task_struct *p) { if (!(p->scx.flags & SCX_TASK_IN_CUSTODY) || task_scx_migrating(p)) - return; - - if (SCX_HAS_OP(sch, dequeue)) - SCX_CALL_OP_TASK(sch, dequeue, rq, p, deq_flags); + return false; p->scx.flags &= ~SCX_TASK_IN_CUSTODY; + return true; } static void rq_owned_post_enq(struct scx_sched *sch, struct rq *rq, struct scx_dispatch_q *dsq, struct task_struct *p, u64 enq_flags) { - call_task_dequeue(sch, rq, p, 0); + if (task_leave_custody(p) && SCX_HAS_OP(sch, dequeue)) + SCX_CALL_OP_TASK(sch, dequeue, rq, p, 0); /* * Only local inserts get the wakeup treatment below. Rejects kick the @@ -1705,20 +1700,28 @@ static void scx_dispatch_enqueue(struct scx_sched *sch, struct rq *rq, if (is_rq_owned) { rq_owned_post_enq(sch, rq, dsq, p, enq_flags); } else { + bool call_dequeue = false; + /* * Global and bypass DSQs are terminal - the task leaves the - * scheduler's custody, so ops.dequeue() fires here. It can run + * scheduler's custody, so ops.dequeue() fires. It can run * without @p's rq lock (finish_dispatch() passes the dispatch * rq); that's safe because dequeue_task_scx() waits on * SCX_OPSS_DISPATCHING (see the ops_state note above) and so * can't race it. A non-terminal DSQ keeps the task in custody. + * The custody transfer happens under @dsq->lock so that later + * consumers see the flag clear; the callback runs after + * @dsq->lock is dropped because it may lock a DSQ itself. */ if (dsq->id == SCX_DSQ_GLOBAL || dsq->id == SCX_DSQ_BYPASS) - call_task_dequeue(sch, rq, p, 0); + call_dequeue = task_leave_custody(p); else p->scx.flags |= SCX_TASK_IN_CUSTODY; raw_spin_unlock(&dsq->lock); + + if (call_dequeue && SCX_HAS_OP(sch, dequeue)) + SCX_CALL_OP_TASK(sch, dequeue, rq, p, 0); } /* @@ -2215,7 +2218,7 @@ static void ops_dequeue(struct rq *rq, struct task_struct *p, u64 deq_flags) /* * A queued task must always be in BPF scheduler's custody. If * SCX_TASK_IN_CUSTODY is clear, finish_dispatch() on another - * CPU has already passed call_task_dequeue() (which clears the + * CPU has already passed task_leave_custody() (which clears the * flag), but has not yet written SCX_OPSS_NONE. That final * store does not require this rq's lock, so retrying with * cpu_relax() is bounded: we will observe NONE (or DISPATCHING, @@ -2263,7 +2266,8 @@ static void ops_dequeue(struct rq *rq, struct task_struct *p, u64 deq_flags) * NONE but the task may still have %SCX_TASK_IN_CUSTODY set until * it is enqueued on the destination. */ - call_task_dequeue(sch, rq, p, deq_flags); + if (task_leave_custody(p) && SCX_HAS_OP(sch, dequeue)) + SCX_CALL_OP_TASK(sch, dequeue, rq, p, deq_flags); } static bool dequeue_task_scx(struct rq *rq, struct task_struct *p, int core_deq_flags) @@ -2379,14 +2383,10 @@ static void wakeup_preempt_scx(struct rq *rq, struct task_struct *p, int wake_fl } void scx_move_local_task_to_local_dsq(struct scx_sched *sch, struct task_struct *p, - u64 enq_flags, struct scx_dispatch_q *src_dsq, - struct rq *dst_rq) + u64 enq_flags, struct rq *dst_rq) { struct scx_dispatch_q *dst_dsq = scx_resolve_local_dsq(sch, dst_rq, p, &enq_flags); - /* @p is on @dst_rq, an rq-owned @src_dsq is covered by the rq lock */ - if (!dsq_is_rq_owned(src_dsq)) - lockdep_assert_held(&src_dsq->lock); lockdep_assert_rq_held(dst_rq); WARN_ON_ONCE(p->scx.holding_cpu >= 0); @@ -2634,8 +2634,8 @@ static struct rq *move_task_between_dsqs(struct scx_sched *sch, /* @p is going from a non-local DSQ to a local DSQ */ if (src_rq == dst_rq) { scx_task_unlink_from_dsq(p, src_dsq); - scx_move_local_task_to_local_dsq(sch, p, enq_flags, src_dsq, dst_rq); raw_spin_unlock(&src_dsq->lock); + scx_move_local_task_to_local_dsq(sch, p, enq_flags, dst_rq); } else { raw_spin_unlock(&src_dsq->lock); move_remote_task_to_local_dsq(sch, p, enq_flags, src_rq, dst_rq); @@ -2685,8 +2685,8 @@ bool scx_consume_dispatch_q(struct scx_sched *sch, struct rq *rq, if (rq == task_rq) { scx_task_unlink_from_dsq(p, dsq); - scx_move_local_task_to_local_dsq(sch, p, enq_flags, dsq, rq); raw_spin_unlock(&dsq->lock); + scx_move_local_task_to_local_dsq(sch, p, enq_flags, rq); return true; } diff --git a/kernel/sched/ext/internal.h b/kernel/sched/ext/internal.h index 115c96fbf932..d150de10a5c9 100644 --- a/kernel/sched/ext/internal.h +++ b/kernel/sched/ext/internal.h @@ -2065,8 +2065,7 @@ void scx_dispatch_dequeue(struct rq *rq, struct task_struct *p); void scx_do_enqueue_task(struct rq *rq, struct task_struct *p, u64 enq_flags, int sticky_cpu); void scx_move_local_task_to_local_dsq(struct scx_sched *sch, struct task_struct *p, - u64 enq_flags, struct scx_dispatch_q *src_dsq, - struct rq *dst_rq); + u64 enq_flags, struct rq *dst_rq); bool scx_consume_dispatch_q(struct scx_sched *sch, struct rq *rq, struct scx_dispatch_q *dsq, u64 enq_flags); bool scx_consume_global_dsq(struct scx_sched *sch, struct rq *rq); diff --git a/kernel/sched/ext/sub.c b/kernel/sched/ext/sub.c index a30c965b175d..3036190b0662 100644 --- a/kernel/sched/ext/sub.c +++ b/kernel/sched/ext/sub.c @@ -555,8 +555,8 @@ static void scx_rescue_timerfn(struct timer_list *timer) scx.dsq_list.node); scx_task_unlink_from_dsq(p, &rq->scx.rescue.dsq); scx_rescue_admit(rq, p, slice); - scx_move_local_task_to_local_dsq(scx_task_sched(p), p, SCX_ENQ_IGNORE_CAPS, - &rq->scx.rescue.dsq, rq); + scx_move_local_task_to_local_dsq(scx_task_sched(p), p, + SCX_ENQ_IGNORE_CAPS, rq); if (sched_class_above(&ext_sched_class, rq->curr->sched_class)) resched_curr(rq); } else if (p->scx.dsq && rq->scx.rescue.budget > 2 * scx_rescue_quantum_ns) { @@ -572,7 +572,7 @@ static void scx_rescue_timerfn(struct timer_list *timer) scx_task_unlink_from_dsq(p, &rq->scx.local_dsq); scx_move_local_task_to_local_dsq(scx_task_sched(p), p, SCX_ENQ_HEAD | SCX_ENQ_PREEMPT | SCX_ENQ_IGNORE_CAPS, - &rq->scx.local_dsq, rq); + rq); } out_arm: scx_rescue_timer_arm(rq); @@ -596,8 +596,8 @@ void scx_rescue_flush(struct rq *rq) /* and flush out all pending ones */ list_for_each_entry_safe(p, n, &rq->scx.rescue.dsq.list, scx.dsq_list.node) { scx_task_unlink_from_dsq(p, &rq->scx.rescue.dsq); - scx_move_local_task_to_local_dsq(scx_task_sched(p), p, SCX_ENQ_IGNORE_CAPS, - &rq->scx.rescue.dsq, rq); + scx_move_local_task_to_local_dsq(scx_task_sched(p), p, + SCX_ENQ_IGNORE_CAPS, rq); } timer_delete(&rq->scx.rescue.timer); From 9ec7ba20c97d92fa59b351832aeeb934b3b16177 Mon Sep 17 00:00:00 2001 From: fangqiurong Date: Thu, 17 Sep 2026 15:52:42 +0800 Subject: [PATCH 093/365] selftests/sched_ext: Test that ops.dequeue() can iterate the consumed DSQ Add a scheduler whose ops.dequeue() iterates the source user DSQ with bpf_iter_scx_dsq. The iteration takes the DSQ's raw spinlock; on a kernel that runs ops.dequeue() while the consume path still holds that lock, the first task consumed self-deadlocks the CPU with IRQs disabled. The watchdog cannot recover from that state, so on an unfixed kernel this test wedges the system instead of failing cleanly. On a fixed kernel the scheduler runs clean and the test passes. Signed-off-by: fangqiurong Signed-off-by: Tejun Heo --- tools/testing/selftests/sched_ext/Makefile | 1 + .../selftests/sched_ext/dequeue_iter.bpf.c | 73 +++++++++++++++++ .../selftests/sched_ext/dequeue_iter.c | 79 +++++++++++++++++++ 3 files changed, 153 insertions(+) create mode 100644 tools/testing/selftests/sched_ext/dequeue_iter.bpf.c create mode 100644 tools/testing/selftests/sched_ext/dequeue_iter.c diff --git a/tools/testing/selftests/sched_ext/Makefile b/tools/testing/selftests/sched_ext/Makefile index 3cfe90e0f34f..49897727f535 100644 --- a/tools/testing/selftests/sched_ext/Makefile +++ b/tools/testing/selftests/sched_ext/Makefile @@ -164,6 +164,7 @@ all_test_bpfprogs := $(foreach prog,$(wildcard *.bpf.c),$(INCLUDE_DIR)/$(patsubs auto-test-targets := \ create_dsq \ dequeue \ + dequeue_iter \ enq_last_no_enq_fails \ ddsp_bogus_dsq_fail \ ddsp_vtimelocal_fail \ diff --git a/tools/testing/selftests/sched_ext/dequeue_iter.bpf.c b/tools/testing/selftests/sched_ext/dequeue_iter.bpf.c new file mode 100644 index 000000000000..76c2c71a90b6 --- /dev/null +++ b/tools/testing/selftests/sched_ext/dequeue_iter.bpf.c @@ -0,0 +1,73 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * ops.dequeue() of this scheduler iterates the user DSQ it consumes + * tasks from with bpf_iter_scx_dsq, which takes the DSQ lock. + * On a kernel that still runs ops.dequeue() with that lock held, the + * iteration self-deadlocks the CPU - this test wedges the system on + * unfixed kernels instead of failing cleanly. + * + * Copyright (c) 2026 fangqiurong + */ + +#include + +char _license[] SEC("license") = "GPL"; + +UEI_DEFINE(uei); + +#define TEST_DSQ_ID 1000 + +u64 dq_count; + +s32 BPF_STRUCT_OPS_SLEEPABLE(dequeue_iter_init) +{ + return scx_bpf_create_dsq(TEST_DSQ_ID, -1); +} + +s32 BPF_STRUCT_OPS(dequeue_iter_select_cpu, struct task_struct *p, + s32 prev_cpu, u64 wake_flags) +{ + return prev_cpu; +} + +void BPF_STRUCT_OPS(dequeue_iter_enqueue, struct task_struct *p, u64 enq_flags) +{ + scx_bpf_dsq_insert(p, TEST_DSQ_ID, SCX_SLICE_DFL, enq_flags); +} + +void BPF_STRUCT_OPS(dequeue_iter_dispatch, s32 cpu, struct task_struct *task) +{ + scx_bpf_dsq_move_to_local(TEST_DSQ_ID, 0); +} + +void BPF_STRUCT_OPS(dequeue_iter_dequeue, struct task_struct *p, u64 deq_flags) +{ + struct bpf_iter_scx_dsq it; + struct task_struct *t; + + if (!bpf_iter_scx_dsq_new(&it, TEST_DSQ_ID, 0)) { + while ((t = bpf_iter_scx_dsq_next(&it))) + ; + } + bpf_iter_scx_dsq_destroy(&it); + + __sync_fetch_and_add(&dq_count, 1); +} + +void BPF_STRUCT_OPS(dequeue_iter_exit, struct scx_exit_info *ei) +{ + UEI_RECORD(uei, ei); + scx_bpf_destroy_dsq(TEST_DSQ_ID); +} + +SEC(".struct_ops.link") +struct sched_ext_ops dequeue_iter_ops = { + .init = (void *)dequeue_iter_init, + .select_cpu = (void *)dequeue_iter_select_cpu, + .enqueue = (void *)dequeue_iter_enqueue, + .dispatch = (void *)dequeue_iter_dispatch, + .dequeue = (void *)dequeue_iter_dequeue, + .exit = (void *)dequeue_iter_exit, + .timeout_ms = 1000U, + .name = "dequeue_iter", +}; diff --git a/tools/testing/selftests/sched_ext/dequeue_iter.c b/tools/testing/selftests/sched_ext/dequeue_iter.c new file mode 100644 index 000000000000..f60711bf3b39 --- /dev/null +++ b/tools/testing/selftests/sched_ext/dequeue_iter.c @@ -0,0 +1,79 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Copyright (c) 2026 fangqiurong + */ +#include +#include +#include +#include +#include "dequeue_iter.bpf.skel.h" +#include "scx_test.h" + +#define DQ_TARGET 10 +#define DQ_DEADLINE_MS 3000 + +static unsigned long long now_ms(void) +{ + struct timespec ts; + + clock_gettime(CLOCK_MONOTONIC, &ts); + + return ts.tv_sec * 1000ULL + ts.tv_nsec / 1000000; +} + +static enum scx_test_status setup(void **ctx) +{ + struct dequeue_iter *skel; + + skel = dequeue_iter__open(); + SCX_FAIL_IF(!skel, "Failed to open"); + SCX_ENUM_INIT(skel); + SCX_FAIL_IF(dequeue_iter__load(skel), "Failed to load skel"); + + *ctx = skel; + + return SCX_TEST_PASS; +} + +static enum scx_test_status run(void *ctx) +{ + struct dequeue_iter *skel = ctx; + struct bpf_link *link; + unsigned long long end; + + link = bpf_map__attach_struct_ops(skel->maps.dequeue_iter_ops); + SCX_FAIL_IF(!link, "Failed to attach scheduler"); + + end = now_ms() + DQ_DEADLINE_MS; + while (skel->bss->dq_count < DQ_TARGET && !UEI_EXITED(skel, uei) && + now_ms() < end) + usleep(100); + + bpf_link__destroy(link); + + SCX_EQ(skel->data->uei.kind, EXIT_KIND(SCX_EXIT_UNREG)); + + if (skel->bss->dq_count < DQ_TARGET) { + SCX_ERR("ops.dequeue() fired only %llu times", + (unsigned long long)skel->bss->dq_count); + return SCX_TEST_FAIL; + } + + return SCX_TEST_PASS; +} + +static void cleanup(void *ctx) +{ + struct dequeue_iter *skel = ctx; + + dequeue_iter__destroy(skel); +} + +struct scx_test dequeue_iter = { + .name = "dequeue_iter", + .description = "Verify ops.dequeue() can iterate its source user DSQ", + .setup = setup, + .run = run, + .cleanup = cleanup, +}; +REGISTER_SCX_TEST(&dequeue_iter) From 5ea72f7b7139b123713a7983448f910bc4514d9e Mon Sep 17 00:00:00 2001 From: Wentao Liang Date: Wed, 16 Sep 2026 18:02:02 +0000 Subject: [PATCH 094/365] drm/nouveau: Fix gem reference leak in validate_init() On the ttm_bo_reserve() failure and "vma not found" error paths, the loop breaks without adding the looked-up object to any validate list, so the reference taken by drm_gem_object_lookup() is never released; validate_fini() only walks the spliced lists. Drop the reference before breaking out on both paths. Fixes: 19ca10d82e33bcfe ("drm/nouveau/gem: lookup VMAs for buffers referenced by pushbuf ioctl") Cc: stable@vger.kernel.org Signed-off-by: Wentao Liang Reviewed-by: Lyude Paul Signed-off-by: Lyude Paul Link: https://patch.msgid.link/20260916180202.2090231-1-vulab@iscas.ac.cn --- drivers/gpu/drm/nouveau/nouveau_gem.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/gpu/drm/nouveau/nouveau_gem.c b/drivers/gpu/drm/nouveau/nouveau_gem.c index 0b7123b163e4..51188be57221 100644 --- a/drivers/gpu/drm/nouveau/nouveau_gem.c +++ b/drivers/gpu/drm/nouveau/nouveau_gem.c @@ -522,6 +522,7 @@ validate_init(struct nouveau_channel *chan, struct drm_file *file_priv, if (unlikely(ret)) { if (ret != -ERESTARTSYS) NV_PRINTK(err, cli, "fail reserve\n"); + drm_gem_object_put(gem); break; } } @@ -531,6 +532,7 @@ validate_init(struct nouveau_channel *chan, struct drm_file *file_priv, struct nouveau_vma *vma = nouveau_vma_find(nvbo, vmm); if (!vma) { NV_PRINTK(err, cli, "vma not found!\n"); + drm_gem_object_put(gem); ret = -EINVAL; break; } From 1e04611d3735543bd80a67d9d13dc13f503746fb Mon Sep 17 00:00:00 2001 From: Wentao Liang Date: Wed, 16 Sep 2026 18:03:42 +0000 Subject: [PATCH 095/365] drm/nouveau: Fix runtime PM leak in nouveau_connector_detect() If nvif_outp_edid_get() fails, nouveau_connector_detect() returns early without dropping the runtime PM reference taken at the start of the function, keeping the device powered on until the next successful detect. Balance the reference on the error path like the other exit paths do. Fixes: 0cd7e0718139 ("drm/nouveau/disp: add output method to fetch edid") Cc: stable@vger.kernel.org Signed-off-by: Wentao Liang Reviewed-by: Lyude Paul Signed-off-by: Lyude Paul Link: https://patch.msgid.link/20260916180342.2090360-1-vulab@iscas.ac.cn --- drivers/gpu/drm/nouveau/nouveau_connector.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/nouveau/nouveau_connector.c b/drivers/gpu/drm/nouveau/nouveau_connector.c index b0b0ad9a0c24..4cfc9c7c2ae0 100644 --- a/drivers/gpu/drm/nouveau/nouveau_connector.c +++ b/drivers/gpu/drm/nouveau/nouveau_connector.c @@ -600,8 +600,11 @@ nouveau_connector_detect(struct drm_connector *connector, bool force) new_edid = drm_get_edid(connector, nv_encoder->i2c); } else { ret = nvif_outp_edid_get(&nv_encoder->outp, (u8 **)&new_edid); - if (ret < 0) + if (ret < 0) { + pm_runtime_mark_last_busy(dev->dev); + pm_runtime_put_autosuspend(dev->dev); return connector_status_disconnected; + } } nouveau_connector_set_edid(nv_connector, new_edid); From 97077ac87afe9e91ec074ef0be64454e7ccbf344 Mon Sep 17 00:00:00 2001 From: Peiyang He Date: Wed, 16 Sep 2026 18:31:38 +0800 Subject: [PATCH 096/365] drm/nouveau: fix double-free in nvif_vmm_dtor On failure, nouveau_cli_init() calls nouveau_cli_fini() to tear the client down. Then, nouveau_drm_open() also enters into its cleanup path and calls nouveau_cli_fini() AGAIN. nouveau_cli_fini() calls nouveau_vmm_fini(): void nouveau_vmm_fini(struct nouveau_vmm *vmm) { nouveau_svmm_fini(&vmm->svmm); nvif_vmm_dtor(&vmm->vmm); vmm->cli = NULL; } Inside nvif_vmm_dtor(), vmm->page is freed unconditionally: void nvif_vmm_dtor(struct nvif_vmm *vmm) { kfree(vmm->page); nvif_object_dtor(&vmm->object); } vmm->page is never cleared after being freed, so the second call of nvif_vmm_dtor() will cause a double-free. Found by fuzzing the nouveau driver with a modified Syzkaller: BUG: KASAN: double-free in nvif_vmm_dtor+0x31/0x50 drivers/gpu/drm/nouveau/nvif/vmm.c:194 Free of addr ffff888010fcdc30 by task syz.0.173/2567 CPU: 1 UID: 0 PID: 2567 Comm: syz.0.173 Not tainted 7.2.0 #24 PREEMPT(lazy) Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.3-debian-1.16.3-2 04/01/2014 Call Trace: __dump_stack lib/dump_stack.c:94 [inline] dump_stack_lvl+0x95/0xe0 lib/dump_stack.c:120 print_address_description mm/kasan/report.c:378 [inline] print_report+0xcb/0x5a0 mm/kasan/report.c:482 kasan_report_invalid_free+0xaa/0xd0 mm/kasan/report.c:557 check_slab_allocation+0xe4/0x110 mm/kasan/common.c:235 kasan_slab_pre_free include/linux/kasan.h:199 [inline] slab_free_hook mm/slub.c:2622 [inline] slab_free mm/slub.c:6377 [inline] kfree+0x192/0x590 mm/slub.c:6692 nvif_vmm_dtor+0x31/0x50 drivers/gpu/drm/nouveau/nvif/vmm.c:194 nouveau_vmm_fini+0x16/0x50 drivers/gpu/drm/nouveau/nouveau_vmm.c:127 nouveau_cli_fini+0x10e/0x210 drivers/gpu/drm/nouveau/nouveau_drm.c:225 nouveau_drm_open+0x24e/0x740 drivers/gpu/drm/nouveau/nouveau_drm.c:1255 drm_file_alloc+0x5f2/0xad0 drivers/gpu/drm/drm_file.c:176 drm_open_helper+0x1d7/0x4a0 drivers/gpu/drm/drm_file.c:335 drm_open+0x190/0x3d0 drivers/gpu/drm/drm_file.c:388 drm_stub_open+0x1f2/0x460 drivers/gpu/drm/drm_drv.c:1211 chrdev_open+0x21c/0x660 fs/char_dev.c:411 do_dentry_open+0x59d/0x12b0 fs/open.c:947 vfs_open+0x82/0x390 fs/open.c:1052 do_open fs/namei.c:4700 [inline] path_openat+0x2345/0x3420 fs/namei.c:4863 do_file_open+0x207/0x460 fs/namei.c:4892 do_sys_openat2+0xd1/0x1d0 fs/open.c:1368 do_sys_open fs/open.c:1374 [inline] __do_sys_openat fs/open.c:1390 [inline] __se_sys_openat fs/open.c:1385 [inline] __x64_sys_openat+0x144/0x200 fs/open.c:1385 do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline] do_syscall_64+0x115/0x690 arch/x86/entry/syscall_64.c:94 entry_SYSCALL_64_after_hwframe+0x77/0x7f RIP: 0033:0x7fc6d687594d Code: ff c3 66 2e 0f 1f 84 00 00 00 00 00 90 f3 0f 1e fa 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 b0 ff ff ff f7 d8 64 89 01 48 RSP: 002b:00007fc6d5295008 EFLAGS: 00000246 ORIG_RAX: 0000000000000101 RAX: ffffffffffffffda RBX: 00007fc6d6b06180 RCX: 00007fc6d687594d RDX: 0000000000022501 RSI: 0000200000000000 RDI: ffffffffffffff9c RBP: 00007fc6d691c303 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000 R13: 00007fc6d6b06218 R14: 00007fc6d6b06180 R15: 00007ffd9451d760 Allocated by task 2567 on cpu 1 at 163.593900s: kasan_save_stack+0x24/0x50 mm/kasan/common.c:57 kasan_save_track+0x17/0x60 mm/kasan/common.c:78 poison_kmalloc_redzone mm/kasan/common.c:398 [inline] __kasan_kmalloc+0xaa/0xb0 mm/kasan/common.c:415 kasan_kmalloc include/linux/kasan.h:263 [inline] __do_kmalloc_node mm/slub.c:5334 [inline] __kmalloc_noprof+0x304/0x7c0 mm/slub.c:5359 _kmalloc_noprof include/linux/slab.h:992 [inline] nvif_vmm_ctor+0x3c0/0x7e0 drivers/gpu/drm/nouveau/nvif/vmm.c:237 nouveau_vmm_init+0x40/0x90 drivers/gpu/drm/nouveau/nouveau_vmm.c:134 nouveau_cli_init+0x7b9/0xe10 drivers/gpu/drm/nouveau/nouveau_drm.c:293 nouveau_drm_open+0x236/0x740 drivers/gpu/drm/nouveau/nouveau_drm.c:1243 drm_file_alloc+0x5f2/0xad0 drivers/gpu/drm/drm_file.c:176 drm_open_helper+0x1d7/0x4a0 drivers/gpu/drm/drm_file.c:335 drm_open+0x190/0x3d0 drivers/gpu/drm/drm_file.c:388 drm_stub_open+0x1f2/0x460 drivers/gpu/drm/drm_drv.c:1211 chrdev_open+0x21c/0x660 fs/char_dev.c:411 do_dentry_open+0x59d/0x12b0 fs/open.c:947 vfs_open+0x82/0x390 fs/open.c:1052 do_open fs/namei.c:4700 [inline] path_openat+0x2345/0x3420 fs/namei.c:4863 do_file_open+0x207/0x460 fs/namei.c:4892 do_sys_openat2+0xd1/0x1d0 fs/open.c:1368 do_sys_open fs/open.c:1374 [inline] __do_sys_openat fs/open.c:1390 [inline] __se_sys_openat fs/open.c:1385 [inline] __x64_sys_openat+0x144/0x200 fs/open.c:1385 do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline] do_syscall_64+0x115/0x690 arch/x86/entry/syscall_64.c:94 entry_SYSCALL_64_after_hwframe+0x77/0x7f Freed by task 2567 on cpu 1 at 163.601355s: kasan_save_stack+0x24/0x50 mm/kasan/common.c:57 kasan_save_track+0x17/0x60 mm/kasan/common.c:78 kasan_save_free_info+0x3b/0x60 mm/kasan/generic.c:584 poison_slab_object mm/kasan/common.c:253 [inline] __kasan_slab_free+0x61/0x80 mm/kasan/common.c:285 kasan_slab_free include/linux/kasan.h:235 [inline] slab_free_hook mm/slub.c:2677 [inline] slab_free mm/slub.c:6377 [inline] kfree+0x383/0x590 mm/slub.c:6692 nvif_vmm_dtor+0x31/0x50 drivers/gpu/drm/nouveau/nvif/vmm.c:194 nouveau_vmm_fini+0x16/0x50 drivers/gpu/drm/nouveau/nouveau_vmm.c:127 nouveau_cli_fini+0x10e/0x210 drivers/gpu/drm/nouveau/nouveau_drm.c:225 nouveau_cli_init+0x593/0xe10 drivers/gpu/drm/nouveau/nouveau_drm.c:324 nouveau_drm_open+0x236/0x740 drivers/gpu/drm/nouveau/nouveau_drm.c:1243 drm_file_alloc+0x5f2/0xad0 drivers/gpu/drm/drm_file.c:176 drm_open_helper+0x1d7/0x4a0 drivers/gpu/drm/drm_file.c:335 drm_open+0x190/0x3d0 drivers/gpu/drm/drm_file.c:388 drm_stub_open+0x1f2/0x460 drivers/gpu/drm/drm_drv.c:1211 chrdev_open+0x21c/0x660 fs/char_dev.c:411 do_dentry_open+0x59d/0x12b0 fs/open.c:947 vfs_open+0x82/0x390 fs/open.c:1052 do_open fs/namei.c:4700 [inline] path_openat+0x2345/0x3420 fs/namei.c:4863 do_file_open+0x207/0x460 fs/namei.c:4892 do_sys_openat2+0xd1/0x1d0 fs/open.c:1368 do_sys_open fs/open.c:1374 [inline] __do_sys_openat fs/open.c:1390 [inline] __se_sys_openat fs/open.c:1385 [inline] __x64_sys_openat+0x144/0x200 fs/open.c:1385 do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline] do_syscall_64+0x115/0x690 arch/x86/entry/syscall_64.c:94 entry_SYSCALL_64_after_hwframe+0x77/0x7f The buggy address belongs to the object at ffff888010fcdc30 which belongs to the cache kmalloc-16 of size 16 The buggy address is located 0 bytes inside of 16-byte region [ffff888010fcdc30, ffff888010fcdc40) The buggy address belongs to the physical page: page: refcount:0 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x10fcd flags: 0x100000000000000(node=0|zone=1) page_type: f5(slab) raw: 0100000000000000 ffff88800d441640 dead000000000100 dead000000000122 raw: 0000000000000000 0000000000550055 00000000f5000000 0000000000000000 page dumped because: kasan: bad access detected Memory state around the buggy address: ffff888010fcdb00: fc fc 00 04 fc fc fc fc fa fb fc fc fc fc fa fb ffff888010fcdb80: fc fc fc fc fa fb fc fc fc fc 00 07 fc fc fc fc >ffff888010fcdc00: fa fb fc fc fc fc fa fb fc fc fc fc fa fb fc fc ^ ffff888010fcdc80: fc fc fa fb fc fc fc fc 00 04 fc fc fc fc fa fb ffff888010fcdd00: fc fc fc fc 00 00 fc fc fc fc fa fb fc fc fc fc Fix by removing the redundant teardown in nouveau_drm_open(), since nouveau_cli_init() already does the cleanup work. Also clear vmm->page after its freeing. Cc: stable@vger.kernel.org Fixes: 20d8a88e557a ("drm/nouveau: tidy up the client init/fini interfaces") Signed-off-by: Peiyang He Assisted-by: LLM Reviewed-by: Lyude Paul Signed-off-by: Lyude Paul Link: https://patch.msgid.link/03BA723D9E5FF725+20260916103138.2651605-1-peiyang_he@smail.nju.edu.cn --- drivers/gpu/drm/nouveau/nouveau_drm.c | 4 +--- drivers/gpu/drm/nouveau/nvif/vmm.c | 1 + 2 files changed, 2 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/nouveau/nouveau_drm.c b/drivers/gpu/drm/nouveau/nouveau_drm.c index 4d1ad718e09b..b4dfaf70762a 100644 --- a/drivers/gpu/drm/nouveau/nouveau_drm.c +++ b/drivers/gpu/drm/nouveau/nouveau_drm.c @@ -1250,10 +1250,8 @@ nouveau_drm_open(struct drm_device *dev, struct drm_file *fpriv) mutex_unlock(&drm->clients_lock); done: - if (ret && cli) { - nouveau_cli_fini(cli); + if (ret && cli) kfree(cli); - } pm_runtime_mark_last_busy(dev->dev); pm_runtime_put_autosuspend(dev->dev); diff --git a/drivers/gpu/drm/nouveau/nvif/vmm.c b/drivers/gpu/drm/nouveau/nvif/vmm.c index 65c3e883b119..579af70766f2 100644 --- a/drivers/gpu/drm/nouveau/nvif/vmm.c +++ b/drivers/gpu/drm/nouveau/nvif/vmm.c @@ -192,6 +192,7 @@ void nvif_vmm_dtor(struct nvif_vmm *vmm) { kfree(vmm->page); + vmm->page = NULL; nvif_object_dtor(&vmm->object); } From 64ca4cdd1031206424e6455f0ae4fb0560d8f46a Mon Sep 17 00:00:00 2001 From: Junrui Luo Date: Mon, 17 Aug 2026 14:50:40 +0800 Subject: [PATCH 097/365] drm/nouveau/dmem: pin VRAM for the whole registered range Commit c32287471077 ("gpu/drm/nouveau: enable THP support for GPU memory migration") grew the device-private region that nouveau_dmem_chunk_alloc() registers from DMEM_CHUNK_SIZE to DMEM_CHUNK_SIZE * NR_CHUNKS, but left the VRAM buffer object backing that region at DMEM_CHUNK_SIZE. nouveau_dmem_page_addr() returns chunk->bo->offset plus the page's offset within the registered region, so every page past the first chunk resolves to VRAM outside the buffer object. Size the buffer object to the region it backs. Fixes: c32287471077 ("gpu/drm/nouveau: enable THP support for GPU memory migration") Reported-by: Yuhao Jiang Assisted-by: LLM Cc: stable@vger.kernel.org Signed-off-by: Junrui Luo Reviewed-by: Lyude Paul Signed-off-by: Lyude Paul Link: https://patch.msgid.link/20260817-nouveau-fixes-v1-1-f518d0c735f3@outlook.com --- drivers/gpu/drm/nouveau/nouveau_dmem.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/nouveau/nouveau_dmem.c b/drivers/gpu/drm/nouveau/nouveau_dmem.c index ad4570c50be7..e74d7bb975a8 100644 --- a/drivers/gpu/drm/nouveau/nouveau_dmem.c +++ b/drivers/gpu/drm/nouveau/nouveau_dmem.c @@ -339,8 +339,8 @@ nouveau_dmem_chunk_alloc(struct nouveau_drm *drm, struct page **ppage, chunk->pagemap.ops = &nouveau_dmem_pagemap_ops; chunk->pagemap.owner = drm->dev; - ret = nouveau_bo_new_pin(&drm->client, NOUVEAU_GEM_DOMAIN_VRAM, DMEM_CHUNK_SIZE, - &chunk->bo); + ret = nouveau_bo_new_pin(&drm->client, NOUVEAU_GEM_DOMAIN_VRAM, + DMEM_CHUNK_SIZE * NR_CHUNKS, &chunk->bo); if (ret) goto out_release; From cb4c7603678ccef4c52b38159f2aaad867586dbc Mon Sep 17 00:00:00 2001 From: Lyude Paul Date: Thu, 17 Sep 2026 14:54:25 -0400 Subject: [PATCH 098/365] drm/nouveau/gsp/r570: Add support for INTERNAL_GCX_ENTRY_PREREQUISITE OpenRM's runtime PM handling looks a bit different then nouveau's, one part in particular that differs from us: OpenRM actually consults GSP to ask whether the GPU should be allowed to enter Gc6 and/or GcOff before runtime suspending the GPU. In the event the card isn't ready, runtime suspend is simply delayed for a few seconds before retrying. Implement the command used for querying GSP about this, NV2080_CTRL_CMD_INTERNAL_GCX_ENTRY_PREREQUISITE, and check to ensure that the GPU is ready for runtime suspend in nouveau_pmops_runtime_suspend() using this query. If the GPU can't be runtime suspended, update the last busy counter of the device and then return -EBUSY from nouveau_pmops_runtime_suspend() - essentially delaying the runtime suspend process by whatever autosuspend_delay_ms is set to. Signed-off-by: Lyude Paul Reviewed-by: Dave Airlie Link: https://patch.msgid.link/20260917185916.1089621-2-lyude@redhat.com --- drivers/gpu/drm/nouveau/include/nvif/cl0080.h | 10 ++++++ drivers/gpu/drm/nouveau/include/nvif/device.h | 1 + .../gpu/drm/nouveau/include/nvkm/subdev/gsp.h | 2 ++ drivers/gpu/drm/nouveau/nouveau_drm.c | 13 +++++++ drivers/gpu/drm/nouveau/nvif/device.c | 13 +++++++ .../gpu/drm/nouveau/nvkm/engine/device/user.c | 35 +++++++++++++++++++ .../gpu/drm/nouveau/nvkm/subdev/gsp/base.c | 10 ++++++ .../gpu/drm/nouveau/nvkm/subdev/gsp/priv.h | 1 + .../drm/nouveau/nvkm/subdev/gsp/rm/r570/gsp.c | 28 +++++++++++++++ .../nvkm/subdev/gsp/rm/r570/nvrm/gsp.h | 7 ++++ .../gpu/drm/nouveau/nvkm/subdev/gsp/rm/rm.h | 2 ++ 11 files changed, 122 insertions(+) diff --git a/drivers/gpu/drm/nouveau/include/nvif/cl0080.h b/drivers/gpu/drm/nouveau/include/nvif/cl0080.h index ea8267e0d8da..9e639df1da46 100644 --- a/drivers/gpu/drm/nouveau/include/nvif/cl0080.h +++ b/drivers/gpu/drm/nouveau/include/nvif/cl0080.h @@ -4,6 +4,7 @@ #define NV_DEVICE_V0_INFO 0x00 #define NV_DEVICE_V0_TIME 0x01 +#define NV_DEVICE_V0_GCX_READY 0x02 struct nv_device_info_v0 { __u8 version; @@ -55,6 +56,15 @@ struct nv_device_time_v0 { __u64 time; }; +#define NV_DEVICE_GC6_READY BIT(0) +#define NV_DEVICE_GCOFF_READY BIT(1) + +struct nv_device_gcx_ready_v0 { + __u8 version; + __u8 pad01[6]; + __u8 ready; +}; + #define NV_DEVICE_INFO_UNIT (0xffffffffULL << 32) #define NV_DEVICE_INFO(n) ((n) | (0x00000000ULL << 32)) #define NV_DEVICE_HOST(n) ((n) | (0x00000001ULL << 32)) diff --git a/drivers/gpu/drm/nouveau/include/nvif/device.h b/drivers/gpu/drm/nouveau/include/nvif/device.h index 7877a2a79da9..ce2fadcb05d9 100644 --- a/drivers/gpu/drm/nouveau/include/nvif/device.h +++ b/drivers/gpu/drm/nouveau/include/nvif/device.h @@ -22,4 +22,5 @@ int nvif_device_ctor(struct nvif_client *, const char *name, struct nvif_device void nvif_device_dtor(struct nvif_device *); int nvif_device_map(struct nvif_device *); u64 nvif_device_time(struct nvif_device *); +int nvif_device_gcx_ready(struct nvif_device *); #endif diff --git a/drivers/gpu/drm/nouveau/include/nvkm/subdev/gsp.h b/drivers/gpu/drm/nouveau/include/nvkm/subdev/gsp.h index 64fed208e4cf..cd10c3705369 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/subdev/gsp.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/subdev/gsp.h @@ -495,6 +495,8 @@ nvkm_gsp_event_dtor(struct nvkm_gsp_event *event) int nvkm_gsp_intr_stall(struct nvkm_gsp *, enum nvkm_subdev_type, int); int nvkm_gsp_intr_nonstall(struct nvkm_gsp *, enum nvkm_subdev_type, int); +int nvkm_gsp_gcx_ready(struct nvkm_gsp *gsp); + int gv100_gsp_new(struct nvkm_device *, enum nvkm_subdev_type, int, struct nvkm_gsp **); int tu102_gsp_new(struct nvkm_device *, enum nvkm_subdev_type, int, struct nvkm_gsp **); int tu116_gsp_new(struct nvkm_device *, enum nvkm_subdev_type, int, struct nvkm_gsp **); diff --git a/drivers/gpu/drm/nouveau/nouveau_drm.c b/drivers/gpu/drm/nouveau/nouveau_drm.c index b4dfaf70762a..d0eb29583cb1 100644 --- a/drivers/gpu/drm/nouveau/nouveau_drm.c +++ b/drivers/gpu/drm/nouveau/nouveau_drm.c @@ -1148,6 +1148,7 @@ nouveau_pmops_runtime_suspend(struct device *dev) { struct pci_dev *pdev = to_pci_dev(dev); struct nouveau_drm *drm = pci_get_drvdata(pdev); + struct nvif_device *nvif = &drm->client.device; int ret; if (!nouveau_pmops_runtime()) { @@ -1155,6 +1156,18 @@ nouveau_pmops_runtime_suspend(struct device *dev) return -EBUSY; } + // Check if the GPU itself is ready for runtime suspend, otherwise mark as busy and check + // again in a bit. + ret = nvif_device_gcx_ready(nvif); + if (ret < 0) { + NV_ERROR(drm, "Failed to query GCX readiness (returned %d)\n", ret); + return -EBUSY; + } else if (!(ret & NV_DEVICE_GCOFF_READY)) { + NV_DEBUG(drm, "GPU isn't ready for suspend yet, delaying...\n"); + pm_runtime_mark_last_busy(dev); + return -EBUSY; + } + nouveau_switcheroo_optimus_dsm(); ret = nouveau_do_suspend(drm, true); pci_save_state(pdev); diff --git a/drivers/gpu/drm/nouveau/nvif/device.c b/drivers/gpu/drm/nouveau/nvif/device.c index 24880931039f..1be9fbe6cb70 100644 --- a/drivers/gpu/drm/nouveau/nvif/device.c +++ b/drivers/gpu/drm/nouveau/nvif/device.c @@ -38,6 +38,19 @@ nvif_device_time(struct nvif_device *device) return device->user.func->time(&device->user); } +int +nvif_device_gcx_ready(struct nvif_device *device) +{ + struct nv_device_gcx_ready_v0 args = {}; + int ret; + + ret = nvif_object_mthd(&device->object, NV_DEVICE_V0_GCX_READY, &args, sizeof(args)); + if (ret) + return ret; + + return args.ready; +} + int nvif_device_map(struct nvif_device *device) { diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/device/user.c b/drivers/gpu/drm/nouveau/nvkm/engine/device/user.c index 23d11d8221cb..f78e6b9b4292 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/device/user.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/device/user.c @@ -27,6 +27,7 @@ #include #include +#include #include #include @@ -189,6 +190,38 @@ nvkm_udevice_time(struct nvkm_udevice *udev, void *data, u32 size) return ret; } +static int +nvkm_udevice_gcx_ready(struct nvkm_udevice *udev, void *data, u32 size) +{ + struct nvkm_object *object = &udev->object; + struct nvkm_device *device = udev->device; + struct nvkm_gsp *gsp = device->gsp; + union { + struct nv_device_gcx_ready_v0 v0; + } *args = data; + int ret = -ENOSYS; + + if (!gsp) { + args->v0.ready = NV_DEVICE_GC6_READY | NV_DEVICE_GCOFF_READY; + return 0; + } + + nvif_ioctl(object, "device gcx ready size %d\n", size); + ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, false); + if (!ret) { + nvif_ioctl(object, "device gcx ready vers %d\n", args->v0.version); + + ret = nvkm_gsp_gcx_ready(gsp); + if (ret < 0) + return ret; + + args->v0.ready = ret; + ret = 0; + } + + return ret; +} + static int nvkm_udevice_mthd(struct nvkm_object *object, u32 mthd, void *data, u32 size) { @@ -199,6 +232,8 @@ nvkm_udevice_mthd(struct nvkm_object *object, u32 mthd, void *data, u32 size) return nvkm_udevice_info(udev, data, size); case NV_DEVICE_V0_TIME: return nvkm_udevice_time(udev, data, size); + case NV_DEVICE_V0_GCX_READY: + return nvkm_udevice_gcx_ready(udev, data, size); default: break; } diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/base.c b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/base.c index 9ba1316831e7..e475d0e8fa7b 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/base.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/base.c @@ -20,6 +20,7 @@ * OTHER DEALINGS IN THE SOFTWARE. */ #include "priv.h" +#include int nvkm_gsp_intr_nonstall(struct nvkm_gsp *gsp, enum nvkm_subdev_type type, int inst) @@ -47,6 +48,15 @@ nvkm_gsp_intr_stall(struct nvkm_gsp *gsp, enum nvkm_subdev_type type, int inst) return -ENOENT; } +int +nvkm_gsp_gcx_ready(struct nvkm_gsp *gsp) +{ + if (!gsp->rm->api->gsp->gcx_ready) + return NV_DEVICE_GC6_READY | NV_DEVICE_GCOFF_READY; + + return gsp->rm->api->gsp->gcx_ready(gsp); +} + static int nvkm_gsp_fini(struct nvkm_subdev *subdev, enum nvkm_suspend_state suspend) { diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/priv.h b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/priv.h index 71b7203bef50..b07797813b04 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/priv.h +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/priv.h @@ -93,6 +93,7 @@ void r535_gsp_dtor(struct nvkm_gsp *); int r535_gsp_oneinit(struct nvkm_gsp *); int r535_gsp_init(struct nvkm_gsp *); int r535_gsp_fini(struct nvkm_gsp *, enum nvkm_suspend_state suspend); +int r535_gsp_gcx_ready(struct nvkm_gsp *gsp); int nvkm_gsp_new_(const struct nvkm_gsp_fwif *, struct nvkm_device *, enum nvkm_subdev_type, int, struct nvkm_gsp **); diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/gsp.c b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/gsp.c index b45781cd0dfd..89b801c1e609 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/gsp.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/gsp.c @@ -10,6 +10,7 @@ #include "nvrm/gsp.h" #include "nvrm/rpcfn.h" #include "nvrm/msgfn.h" +#include "nvif/cl0080.h" #include #include @@ -215,6 +216,32 @@ r570_gsp_set_rmargs(struct nvkm_gsp *gsp, bool resume) args->bDmemStack = 1; } +int +r570_gsp_gcx_ready(struct nvkm_gsp *gsp) +{ + NV2080_CTRL_INTERNAL_GCX_ENTRY_PREREQUISITE_PARAMS *ctrl; + int ret = 0; + + ctrl = nvkm_gsp_rm_ctrl_rd(&gsp->internal.device.subdevice, + NV2080_CTRL_CMD_INTERNAL_GCX_ENTRY_PREREQUISITE, + sizeof(*ctrl)); + if (IS_ERR(ctrl)) + return PTR_ERR(ctrl); + + if (ctrl->bIsGC6Satisfied) + ret |= NV_DEVICE_GC6_READY; + if (ctrl->bIsGCOFFSatisfied) + ret |= NV_DEVICE_GCOFF_READY; + + nvkm_debug(&gsp->subdev, + "GCX ready status: GC6=%s GCOFF=%s\n", + str_yes_no(ctrl->bIsGC6Satisfied), str_yes_no(ctrl->bIsGCOFFSatisfied)); + + nvkm_gsp_rm_ctrl_done(&gsp->internal.device.subdevice, ctrl); + return ret; +} + + const struct nvkm_rm_api_gsp r570_gsp = { .set_rmargs = r570_gsp_set_rmargs, @@ -223,4 +250,5 @@ r570_gsp = { .xlat_mc_engine_idx = r570_gsp_xlat_mc_engine_idx, .drop_post_nocat_record = r570_gsp_drop_post_nocat_record, .sr_data_size = r570_gsp_sr_data_size, + .gcx_ready = r570_gsp_gcx_ready, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/nvrm/gsp.h b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/nvrm/gsp.h index c458569af9d7..2814629fddd2 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/nvrm/gsp.h +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/nvrm/gsp.h @@ -639,4 +639,11 @@ typedef struct GSP_FMC_BOOT_PARAMS } GSP_FMC_BOOT_PARAMS; #define GSP_FW_HEAP_PARAM_BASE_RM_SIZE_GH100 (14 << 20) // Hopper+ + +#define NV2080_CTRL_CMD_INTERNAL_GCX_ENTRY_PREREQUISITE (0x2080a7d7) + +typedef struct NV2080_CTRL_INTERNAL_GCX_ENTRY_PREREQUISITE_PARAMS { + NvBool bIsGC6Satisfied; + NvBool bIsGCOFFSatisfied; +} NV2080_CTRL_INTERNAL_GCX_ENTRY_PREREQUISITE_PARAMS; #endif diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/rm.h b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/rm.h index e9ac47d86b69..86970129ad96 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/rm.h +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/rm.h @@ -40,6 +40,7 @@ struct nvkm_rm_api { void (*drop_send_user_shared_data)(struct nvkm_gsp *); void (*drop_post_nocat_record)(struct nvkm_gsp *); u32 (*sr_data_size)(struct nvkm_gsp *); + int (*gcx_ready)(struct nvkm_gsp *gsp); } *gsp; const struct nvkm_rm_api_rpc { @@ -174,6 +175,7 @@ int r535_gr_chan_new(struct nvkm_gr *, struct nvkm_chan *, const struct nvkm_ocl int r535_gr_promote_ctx(struct r535_gr *, bool golden, struct nvkm_vmm *, struct nvkm_memory **pctxbuf_mem, struct nvkm_vma **pctxbuf_vma, struct nvkm_gsp_object *chan); +int r570_gsp_gcx_ready(struct nvkm_gsp *gsp); extern const struct nvkm_rm_api_engine r535_nvdec; extern const struct nvkm_rm_api_engine r535_nvenc; extern const struct nvkm_rm_api_engine r535_nvjpg; From 3217000f0b7e4f67e5b386d5b3491ec1b9b584b8 Mon Sep 17 00:00:00 2001 From: Lyude Paul Date: Thu, 17 Sep 2026 14:54:26 -0400 Subject: [PATCH 099/365] drm/nouveau/gsp/r535: Add support for MEMSYS_GET_STATIC_CONFIG This is a GSP structure describing various characteristics of the memory management system that GSP provides. Start by fetching it during driver load, but don't do anything with the information we get from it just yet. Signed-off-by: Lyude Paul Reviewed-by: Dave Airlie Link: https://patch.msgid.link/20260917185916.1089621-3-lyude@redhat.com --- .../gpu/drm/nouveau/include/nvkm/subdev/gsp.h | 4 ++ .../drm/nouveau/nvkm/subdev/gsp/rm/r535/gsp.c | 17 +++++++ .../nvkm/subdev/gsp/rm/r535/nvrm/gsp.h | 45 +++++++++++++++++++ .../drm/nouveau/nvkm/subdev/gsp/rm/r570/gsp.c | 8 ++++ .../gpu/drm/nouveau/nvkm/subdev/gsp/rm/rm.h | 1 + 5 files changed, 75 insertions(+) diff --git a/drivers/gpu/drm/nouveau/include/nvkm/subdev/gsp.h b/drivers/gpu/drm/nouveau/include/nvkm/subdev/gsp.h index cd10c3705369..ed5c6e0e68d3 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/subdev/gsp.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/subdev/gsp.h @@ -156,6 +156,10 @@ struct nvkm_gsp { struct sg_table fbsr; } sr; + struct { + bool use_raw_mode_comptagline_alloc; + } memsys; + struct { struct nvkm_gsp_mem mem; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r535/gsp.c b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r535/gsp.c index 94925f1590ea..63aa30f94747 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r535/gsp.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r535/gsp.c @@ -1782,6 +1782,23 @@ r535_gsp_fini(struct nvkm_gsp *gsp, enum nvkm_suspend_state suspend) return 0; } +int +r535_gsp_get_static_memsys_info(struct nvkm_gsp *gsp) +{ + NV2080_CTRL_INTERNAL_MEMSYS_GET_STATIC_CONFIG_PARAMS *ctrl; + + ctrl = nvkm_gsp_rm_ctrl_rd(&gsp->internal.device.subdevice, + NV2080_CTRL_CMD_INTERNAL_MEMSYS_GET_STATIC_CONFIG, + sizeof(*ctrl)); + if (IS_ERR(ctrl)) + return PTR_ERR(ctrl); + + gsp->memsys.use_raw_mode_comptagline_alloc = ctrl->bUseRawModeComptaglineAllocation; + + nvkm_gsp_rm_ctrl_done(&gsp->internal.device.subdevice, ctrl); + return 0; +} + int r535_gsp_init(struct nvkm_gsp *gsp) { diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r535/nvrm/gsp.h b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r535/nvrm/gsp.h index b6683a5bf870..7b10b7548c57 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r535/nvrm/gsp.h +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r535/nvrm/gsp.h @@ -782,6 +782,51 @@ typedef struct NV2080_CTRL_INTERNAL_INTR_GET_KERNEL_TABLE_PARAMS { #define GSP_FW_HEAP_PARAM_CLIENT_ALLOC_SIZE ((48 << 10) * 2048) // Support 2048 channels +typedef struct NV2080_CTRL_INTERNAL_MEMSYS_GET_STATIC_CONFIG_PARAMS { + /*! Determines if RM should use 1 to 1 Comptagline allocation policy */ + NvBool bOneToOneComptagLineAllocation; + + /*! Determines if RM should use 1 to 4 Comptagline allocation policy */ + NvBool bUseOneToFourComptagLineAllocation; + + /*! Determines if RM should use raw Comptagline allocation policy */ + NvBool bUseRawModeComptaglineAllocation; + + /*! Has COMPBIT_BACKING_SIZE been overridden to zero (i.e. disabled)? */ + NvBool bDisableCompbitBacking; + + /*! Determine if we need to disable post L2 compression */ + NvBool bDisablePostL2Compression; + + /*! Is ECC DRAM feature supported? */ + NvBool bEnabledEccFBPA; + + NvBool bL2PreFill; + + /*! L2 cache size */ + NV_DECLARE_ALIGNED(NvU64 l2CacheSize, 8); + + /*! Indicate whether fpba is present or not */ + NvBool bFbpaPresent; + + /*! Size covered by one comptag */ + NvU32 comprPageSize; + + /*! log32(comprPageSize) */ + NvU32 comprPageShift; + + /*! RAM type */ + NvU32 ramType; + + /*! LTC count */ + NvU32 ltcCount; + + /*! LTS per LTC count */ + NvU32 ltsPerLtcCount; +} NV2080_CTRL_INTERNAL_MEMSYS_GET_STATIC_CONFIG_PARAMS; + +#define NV2080_CTRL_CMD_INTERNAL_MEMSYS_GET_STATIC_CONFIG (0x20800a1c) /* finn: Evaluated from "(FINN_NV20_SUBDEVICE_0_INTERNAL_INTERFACE_ID << 8) | NV2080_CTRL_INTERNAL_MEMSYS_GET_STATIC_CONFIG_PARAMS_MESSAGE_ID" */ + typedef union rpc_message_rpc_union_field_v03_00 { NvU32 spare; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/gsp.c b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/gsp.c index 89b801c1e609..ea38a94211f4 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/gsp.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/gsp.c @@ -138,6 +138,14 @@ r570_gsp_get_static_info(struct nvkm_gsp *gsp) } } + ret = r535_gsp_get_static_memsys_info(gsp); + if (ret) { + nvkm_error(&gsp->subdev, "Retrieving static memsys info failed\n"); + return ret; + } + nvkm_debug(&gsp->subdev, "memsys: Use raw mode for comptag allocations? %s\n", + str_yes_no(gsp->memsys.use_raw_mode_comptagline_alloc)); + return 0; } diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/rm.h b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/rm.h index 86970129ad96..17480d4e527a 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/rm.h +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/rm.h @@ -176,6 +176,7 @@ int r535_gr_promote_ctx(struct r535_gr *, bool golden, struct nvkm_vmm *, struct nvkm_memory **pctxbuf_mem, struct nvkm_vma **pctxbuf_vma, struct nvkm_gsp_object *chan); int r570_gsp_gcx_ready(struct nvkm_gsp *gsp); +int r535_gsp_get_static_memsys_info(struct nvkm_gsp *gsp); extern const struct nvkm_rm_api_engine r535_nvdec; extern const struct nvkm_rm_api_engine r535_nvenc; extern const struct nvkm_rm_api_engine r535_nvjpg; From c7ef611a43bb3ab7e7738349a860418336d507db Mon Sep 17 00:00:00 2001 From: Lyude Paul Date: Thu, 17 Sep 2026 14:54:27 -0400 Subject: [PATCH 100/365] drm/nouveau/gsp/r570: Add comp mode workaround from issue #3172217 One of the things that OpenRM does right before initiating fbsr is apply a special workaround (nvidia issue #3172217) which temporarily disables raw compression mode on the GPU. It is later re-enabled after resuming with fbsr completes. Since we don't currently save the compbit backing with fbsr, this shouldn't currently make any functional difference in the suspend/resume process. But it will be required for implementing support for saving and restoring compbit backings from the GPU. Signed-off-by: Lyude Paul Reviewed-by: Dave Airlie Link: https://patch.msgid.link/20260917185916.1089621-4-lyude@redhat.com --- .../nouveau/nvkm/subdev/gsp/rm/r570/fbsr.c | 48 +++++++++++++++++++ .../nvkm/subdev/gsp/rm/r570/nvrm/fbsr.h | 6 +++ 2 files changed, 54 insertions(+) diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/fbsr.c b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/fbsr.c index af5aa5065c3d..f73d9b29e891 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/fbsr.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/fbsr.c @@ -26,6 +26,35 @@ r570_fbsr_suspend_channels(struct nvkm_gsp *gsp, bool suspend) return nvkm_gsp_rm_ctrl_wr(&gsp->internal.device.subdevice, ctrl); } +static int +r570_memsys_enable_raw_comp_mode(struct nvkm_gsp *gsp, bool enable) +{ + NV2080_CTRL_INTERNAL_MEMSYS_PROGRAM_RAW_COMPRESSION_MODE_PARAMS *ctrl; + int ret; + + ctrl = nvkm_gsp_rm_ctrl_get(&gsp->internal.device.subdevice, + NV2080_CTRL_CMD_INTERNAL_MEMSYS_PROGRAM_RAW_COMPRESSION_MODE, + sizeof(*ctrl)); + if (IS_ERR(ctrl)) + return PTR_ERR(ctrl); + + ctrl->bRawMode = enable; + + ret = nvkm_gsp_rm_ctrl_wr(&gsp->internal.device.subdevice, ctrl); + if (!ret) + nvkm_debug(&gsp->subdev, "memsys: Raw compression mode %s\n", + str_enabled_disabled(enable)); + + return ret; +} + +static bool +r570_need_raw_comp_war(struct nvkm_gsp *gsp, struct nvkm_device *device) +{ + return (device->card_type == GA100 || device->card_type == AD100) && + gsp->memsys.use_raw_mode_comptagline_alloc; +} + static void r570_fbsr_resume(struct nvkm_gsp *gsp) { @@ -33,6 +62,7 @@ r570_fbsr_resume(struct nvkm_gsp *gsp) struct nvkm_instmem *imem = device->imem; struct nvkm_instobj *iobj; struct nvkm_vmm *vmm; + int ret; /* Restore BAR2 page tables via BAR0 window, and re-enable BAR2. */ list_for_each_entry(iobj, &imem->boot, head) { @@ -54,6 +84,13 @@ r570_fbsr_resume(struct nvkm_gsp *gsp) vmm = nvkm_bar_bar1_vmm(device); vmm->func->flush(vmm, 0); + /* Re-enable raw mode if it was previously disabled */ + if (r570_need_raw_comp_war(gsp, device)) { + ret = r570_memsys_enable_raw_comp_mode(gsp, true); + if (ret) + nvkm_error(&gsp->subdev, "Failed to re-enable raw comp mode\n"); + } + /* Resume channel scheduling. */ r570_fbsr_suspend_channels(device->gsp, false); @@ -104,6 +141,17 @@ r570_fbsr_suspend(struct nvkm_gsp *gsp) /* Stop channel scheduling. */ r570_fbsr_suspend_channels(gsp, true); + /* Temporarily disable raw mode to prevent FBSR restore operations from corrupting + * compressed surfaces. Required for ampere and ada. + * + * Nvidia bug #3172217 + */ + if (r570_need_raw_comp_war(gsp, device)) { + ret = r570_memsys_enable_raw_comp_mode(gsp, false); + if (ret) + return ret; + } + /* Save BAR2 allocations to system memory. */ list_for_each_entry(iobj, &imem->list, head) { if (iobj->preserve) { diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/nvrm/fbsr.h b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/nvrm/fbsr.h index 8af432375f7a..9050a8274b27 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/nvrm/fbsr.h +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/nvrm/fbsr.h @@ -16,4 +16,10 @@ typedef struct NV2080_CTRL_INTERNAL_FBSR_INIT_PARAMS { NV_DECLARE_ALIGNED(NvU64 sysmemAddrOfSuspendResumeData, 8); } NV2080_CTRL_INTERNAL_FBSR_INIT_PARAMS; +#define NV2080_CTRL_CMD_INTERNAL_MEMSYS_PROGRAM_RAW_COMPRESSION_MODE (0x20800a6f) /* finn: Evaluated from "(FINN_NV20_SUBDEVICE_0_INTERNAL_INTERFACE_ID << 8) | NV2080_CTRL_INTERNAL_MEMSYS_PROGRAM_RAW_COMPRESSION_MODE_PARAMS_MESSAGE_ID" */ + +typedef struct NV2080_CTRL_INTERNAL_MEMSYS_PROGRAM_RAW_COMPRESSION_MODE_PARAMS { + NvBool bRawMode; +} NV2080_CTRL_INTERNAL_MEMSYS_PROGRAM_RAW_COMPRESSION_MODE_PARAMS; + #endif From 82f4394bbe223fda560153257e5cf21bdb12b6a3 Mon Sep 17 00:00:00 2001 From: Lyude Paul Date: Thu, 17 Sep 2026 14:54:28 -0400 Subject: [PATCH 101/365] drm/nouveau/gsp/r570: Start saving comptag backing stores One of the portions of OpenRM's fbsr process that we never implemented is the saving and restoring of comptag backing stores. This isn't strictly necessary for fbsr to work (as long as we don't specify bEnteringGcOff = 1), but implementing it brings us much closer to matching OpenRM's fbsr process - which means we can rely on things being well tested on Nvidia's side. Now that we have the required driver workarounds in place and fetch the required information from GSP's memsys on driver load, let's implement support for this by fetching the required space for the compbit backing stores and adding it to the amount of memory that we allocate for fbsr. With this, we should be able to safely enable bEnteringGcOff in fbsr. Signed-off-by: Lyude Paul Reviewed-by: Dave Airlie Link: https://patch.msgid.link/20260917185916.1089621-5-lyude@redhat.com --- .../nouveau/nvkm/subdev/gsp/rm/r570/fbsr.c | 27 ++++++++++++++++++- .../nvkm/subdev/gsp/rm/r570/nvrm/fbsr.h | 23 ++++++++++++++++ 2 files changed, 49 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/fbsr.c b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/fbsr.c index f73d9b29e891..a93a39ef7c69 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/fbsr.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/fbsr.c @@ -26,6 +26,23 @@ r570_fbsr_suspend_channels(struct nvkm_gsp *gsp, bool suspend) return nvkm_gsp_rm_ctrl_wr(&gsp->internal.device.subdevice, ctrl); } +static int +r570_fb_get_compbit_store_size(struct nvkm_gsp *gsp, u64 *size) +{ + NV0080_CTRL_FB_GET_COMPBIT_STORE_INFO_PARAMS *ctrl; + + ctrl = nvkm_gsp_rm_ctrl_rd(&gsp->internal.device.object, + NV0080_CTRL_CMD_FB_GET_COMPBIT_STORE_INFO, + sizeof(*ctrl)); + if (IS_ERR(ctrl)) + return PTR_ERR(ctrl); + + *size = ctrl->Size; + + nvkm_gsp_rm_ctrl_done(&gsp->internal.device.object, ctrl); + return 0; +} + static int r570_memsys_enable_raw_comp_mode(struct nvkm_gsp *gsp, bool enable) { @@ -135,7 +152,7 @@ r570_fbsr_suspend(struct nvkm_gsp *gsp) struct nvkm_device *device = subdev->device; struct nvkm_instmem *imem = device->imem; struct nvkm_instobj *iobj; - u64 size; + u64 size, compbit_store_size; int ret; /* Stop channel scheduling. */ @@ -152,6 +169,12 @@ r570_fbsr_suspend(struct nvkm_gsp *gsp) return ret; } + ret = r570_fb_get_compbit_store_size(gsp, &compbit_store_size); + if (ret < 0) + return ret; + nvkm_debug(&gsp->subdev, "fbsr: Compbit backing store size: 0x%llx bytes\n", + compbit_store_size); + /* Save BAR2 allocations to system memory. */ list_for_each_entry(iobj, &imem->list, head) { if (iobj->preserve) { @@ -174,6 +197,8 @@ r570_fbsr_suspend(struct nvkm_gsp *gsp) size = gsp->fb.heap.size; size += gsp->fb.rsvd_size; size += gsp->fb.bios.vga_workspace.size; + size += compbit_store_size; + nvkm_debug(subdev, "fbsr: size: 0x%llx bytes\n", size); ret = nvkm_gsp_sg(device, size, &gsp->sr.fbsr); diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/nvrm/fbsr.h b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/nvrm/fbsr.h index 9050a8274b27..cb3e448415b4 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/nvrm/fbsr.h +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/nvrm/fbsr.h @@ -16,6 +16,29 @@ typedef struct NV2080_CTRL_INTERNAL_FBSR_INIT_PARAMS { NV_DECLARE_ALIGNED(NvU64 sysmemAddrOfSuspendResumeData, 8); } NV2080_CTRL_INTERNAL_FBSR_INIT_PARAMS; +#define NV0080_CTRL_CMD_FB_GET_COMPBIT_STORE_INFO (0x801306) /* finn: Evaluated from "(FINN_NV01_DEVICE_0_FB_INTERFACE_ID << 8) | NV0080_CTRL_FB_GET_COMPBIT_STORE_INFO_PARAMS_MESSAGE_ID" */ + +typedef struct NV0080_CTRL_FB_GET_COMPBIT_STORE_INFO_PARAMS { + NV_DECLARE_ALIGNED(NvU64 Size, 8); + NV_DECLARE_ALIGNED(NvU64 Address, 8); + NvU32 AddressSpace; + NvU32 MaxCompbitLine; + NvU32 comptagsPerCacheLine; + NvU32 cacheLineSize; + NvU32 cacheLineSizePerSlice; + NvU32 cacheLineFetchAlignment; + NV_DECLARE_ALIGNED(NvU64 backingStoreBase, 8); + NvU32 gobsPerComptagPerSlice; + NvU32 backingStoreCbcBase; + NvU32 comptaglineAllocationPolicy; + NV_DECLARE_ALIGNED(NvU64 privRegionStartOffset, 8); + NvU32 cbcCoveragePerSlice; +} NV0080_CTRL_FB_GET_COMPBIT_STORE_INFO_PARAMS; + +#define NV0080_CTRL_CMD_FB_GET_COMPBIT_STORE_INFO_ADDRESS_SPACE_UNKNOWN 0 // ADDR_UNKNOWN +#define NV0080_CTRL_CMD_FB_GET_COMPBIT_STORE_INFO_ADDRESS_SPACE_SYSMEM 1 // ADDR_SYSMEM +#define NV0080_CTRL_CMD_FB_GET_COMPBIT_STORE_INFO_ADDRESS_SPACE_FBMEM 2 // ADDR_FBMEM + #define NV2080_CTRL_CMD_INTERNAL_MEMSYS_PROGRAM_RAW_COMPRESSION_MODE (0x20800a6f) /* finn: Evaluated from "(FINN_NV20_SUBDEVICE_0_INTERNAL_INTERFACE_ID << 8) | NV2080_CTRL_INTERNAL_MEMSYS_PROGRAM_RAW_COMPRESSION_MODE_PARAMS_MESSAGE_ID" */ typedef struct NV2080_CTRL_INTERNAL_MEMSYS_PROGRAM_RAW_COMPRESSION_MODE_PARAMS { From adb87c20081e5ed5b6eef1270fc08b28bd77e865 Mon Sep 17 00:00:00 2001 From: Lyude Paul Date: Thu, 17 Sep 2026 14:54:29 -0400 Subject: [PATCH 102/365] drm/nouveau/gsp/r570: Enable Gcoff in fbsr again Now that we're properly saving the compbit backing stores on fbsr init, we can start setting bEnteringGcOff = 1 again without things breaking, which brings us closer to following the exact same code-paths OpenRM does for fbsr. Signed-off-by: Lyude Paul Reviewed-by: Dave Airlie Link: https://patch.msgid.link/20260917185916.1089621-6-lyude@redhat.com --- drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/fbsr.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/fbsr.c b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/fbsr.c index a93a39ef7c69..469e7eed1d6f 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/fbsr.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/rm/r570/fbsr.c @@ -135,7 +135,7 @@ r570_fbsr_init(struct nvkm_gsp *gsp, struct sg_table *sgt, u64 size) ctrl->hClient = gsp->internal.client.object.handle; ctrl->hSysMem = memlist.handle; ctrl->sysmemAddrOfSuspendResumeData = gsp->sr.meta.addr; - ctrl->bEnteringGcoffState = 0; + ctrl->bEnteringGcoffState = 1; ret = nvkm_gsp_rm_ctrl_wr(&gsp->internal.device.subdevice, ctrl); if (ret) From 3359a372efb6d585c97019ee1b7f1874442bcebe Mon Sep 17 00:00:00 2001 From: Peiyang He Date: Mon, 7 Sep 2026 13:22:04 +0800 Subject: [PATCH 103/365] drm/nouveau/uvmm: fix UAF in nouveau_uvmm_sm when BO is in TTM_PL_SYSTEM nouveau_uvmm_sm() calls op_map(), which passes bo->resource through nouveau_mem() to nouveau_uvma_map(). nouveau_uvmm_vmm_map() then reads mem->mem.type. But this is only valid when bo->resource is backed by struct nouveau_mem, as is the case for VRAM and TT resources. If the BO is left in TTM_PL_SYSTEM, bo->resource is only a struct ttm_resource. Treating it as struct nouveau_mem makes the mem->mem.type read past the end of the resource, causing a KASAN: slab-use-after-free Read in nouveau_uvmm_sm report: BUG: KASAN: slab-use-after-free in nouveau_uvmm_vmm_map drivers/gpu/drm/nouveau/nouveau_uvmm.c:152 [inline] BUG: KASAN: slab-use-after-free in nouveau_uvma_map drivers/gpu/drm/nouveau/nouveau_uvmm.c:199 [inline] BUG: KASAN: slab-use-after-free in op_map drivers/gpu/drm/nouveau/nouveau_uvmm.c:849 [inline] BUG: KASAN: slab-use-after-free in nouveau_uvmm_sm.constprop.0+0x6ab/0x900 drivers/gpu/drm/nouveau/nouveau_uvmm.c:903 Read of size 1 at addr ffff888127d3e3a0 by task kworker/0:1/11 CPU: 0 UID: 0 PID: 11 Comm: kworker/0:1 Not tainted 7.2.0 #5 PREEMPT(lazy) Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.3-debian-1.16.3-2 04/01/2014 Workqueue: nouveau_sched_wq_2224 drm_sched_run_job_work Call Trace: __dump_stack lib/dump_stack.c:94 [inline] dump_stack_lvl+0x95/0xe0 lib/dump_stack.c:120 print_address_description mm/kasan/report.c:378 [inline] print_report+0xcb/0x5a0 mm/kasan/report.c:482 kasan_report+0xca/0x100 mm/kasan/report.c:595 nouveau_uvmm_vmm_map drivers/gpu/drm/nouveau/nouveau_uvmm.c:152 [inline] nouveau_uvma_map drivers/gpu/drm/nouveau/nouveau_uvmm.c:199 [inline] op_map drivers/gpu/drm/nouveau/nouveau_uvmm.c:849 [inline] nouveau_uvmm_sm.constprop.0+0x6ab/0x900 drivers/gpu/drm/nouveau/nouveau_uvmm.c:903 nouveau_uvmm_sm_unmap drivers/gpu/drm/nouveau/nouveau_uvmm.c:932 [inline] nouveau_uvmm_bind_job_run+0xd6/0x250 drivers/gpu/drm/nouveau/nouveau_uvmm.c:1532 nouveau_job_run drivers/gpu/drm/nouveau/nouveau_sched.c:350 [inline] nouveau_sched_run_job+0x62/0xd0 drivers/gpu/drm/nouveau/nouveau_sched.c:364 drm_sched_run_job_work+0x356/0xa10 drivers/gpu/drm/scheduler/sched_main.c:1061 process_one_work+0x8a5/0x1900 kernel/workqueue.c:3322 process_scheduled_works kernel/workqueue.c:3405 [inline] worker_thread+0x5dd/0xd80 kernel/workqueue.c:3486 kthread+0x31d/0x420 kernel/kthread.c:436 ret_from_fork+0x662/0x940 arch/x86/kernel/process.c:158 ret_from_fork_asm+0x1a/0x30 arch/x86/entry/entry_64.S:245 Allocated by task 2224 on cpu 0 at 66.550027s: kasan_save_stack+0x24/0x50 mm/kasan/common.c:57 kasan_save_track+0x17/0x60 mm/kasan/common.c:78 poison_kmalloc_redzone mm/kasan/common.c:398 [inline] __kasan_kmalloc+0xaa/0xb0 mm/kasan/common.c:415 kasan_kmalloc include/linux/kasan.h:263 [inline] __do_kmalloc_node mm/slub.c:5334 [inline] __kmalloc_noprof+0x304/0x7c0 mm/slub.c:5359 _kmalloc_noprof include/linux/slab.h:992 [inline] dma_resv_list_alloc+0x27/0x90 drivers/dma-buf/dma-resv.c:106 dma_resv_reserve_fences+0x60e/0xa30 drivers/dma-buf/dma-resv.c:205 ttm_bo_alloc_resource+0x12c/0xbd0 drivers/gpu/drm/ttm/ttm_bo.c:721 ttm_bo_validate+0x1bc/0x4a0 drivers/gpu/drm/ttm/ttm_bo.c:856 ttm_bo_init_reserved+0x2c3/0x570 drivers/gpu/drm/ttm/ttm_bo.c:970 nouveau_bo_init+0x159/0x2c0 drivers/gpu/drm/nouveau/nouveau_bo.c:359 nouveau_gem_new+0x234/0x5f0 drivers/gpu/drm/nouveau/nouveau_gem.c:272 nouveau_gem_ioctl_new+0x1eb/0x420 drivers/gpu/drm/nouveau/nouveau_gem.c:352 drm_ioctl_kernel+0x192/0x350 drivers/gpu/drm/drm_ioctl.c:817 drm_ioctl+0x4f8/0xb40 drivers/gpu/drm/drm_ioctl.c:914 nouveau_drm_ioctl+0xea/0x2c0 drivers/gpu/drm/nouveau/nouveau_drm.c:1338 vfs_ioctl fs/ioctl.c:51 [inline] __do_sys_ioctl fs/ioctl.c:597 [inline] __se_sys_ioctl fs/ioctl.c:583 [inline] __x64_sys_ioctl+0x180/0x1d0 fs/ioctl.c:583 do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline] do_syscall_64+0x115/0x690 arch/x86/entry/syscall_64.c:94 entry_SYSCALL_64_after_hwframe+0x77/0x7f Freed by task 2223 on cpu 0 at 66.554063s: kasan_save_stack+0x24/0x50 mm/kasan/common.c:57 kasan_save_track+0x17/0x60 mm/kasan/common.c:78 kasan_save_free_info+0x3b/0x60 mm/kasan/generic.c:584 poison_slab_object mm/kasan/common.c:253 [inline] __kasan_slab_free+0x61/0x80 mm/kasan/common.c:285 kasan_slab_free include/linux/kasan.h:235 [inline] slab_free_hook mm/slub.c:2677 [inline] __rcu_free_sheaf_prepare+0xb6/0x2e0 mm/slub.c:2928 rcu_free_sheaf+0x1b/0x120 mm/slub.c:5978 rcu_do_batch kernel/rcu/tree.c:2645 [inline] rcu_core+0x521/0x1490 kernel/rcu/tree.c:2897 handle_softirqs+0x1b1/0x8a0 kernel/softirq.c:622 __do_softirq kernel/softirq.c:656 [inline] invoke_softirq kernel/softirq.c:496 [inline] __irq_exit_rcu+0x137/0x1c0 kernel/softirq.c:735 irq_exit_rcu+0x9/0x20 kernel/softirq.c:752 instr_sysvec_apic_timer_interrupt arch/x86/kernel/apic/apic.c:1062 [inline] sysvec_apic_timer_interrupt+0x70/0x80 arch/x86/kernel/apic/apic.c:1062 asm_sysvec_apic_timer_interrupt+0x1a/0x20 arch/x86/include/asm/idtentry.h:674 The buggy address belongs to the object at ffff888127d3e380 which belongs to the cache kmalloc-96 of size 96 The buggy address is located 32 bytes inside of freed 96-byte region [ffff888127d3e380, ffff888127d3e3e0) The buggy address belongs to the physical page: page: refcount:0 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x127d3e flags: 0x200000000000000(node=0|zone=2) page_type: f5(slab) raw: 0200000000000000 ffff888100041280 dead000000000122 0000000000000000 raw: 0000000000000000 0000000000200020 00000000f5000000 0000000000000000 page dumped because: kasan: bad access detected Memory state around the buggy address: ffff888127d3e280: 00 00 00 00 00 00 00 00 00 fc fc fc fc fc fc fc ffff888127d3e300: 00 00 00 00 00 00 00 00 00 00 00 fc fc fc fc fc >ffff888127d3e380: fa fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc ^ ffff888127d3e400: fa fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc ffff888127d3e480: 00 00 00 00 00 00 00 00 00 00 fc fc fc fc fc fc Fix by resetting the placement to the BO's valid domains before calling nouveau_bo_validate(), matching the handling in nouveau_uvmm_bo_validate(), so map jobs do not run for SYSTEM resources; Reject BO that cannot reside in VRAM or GART; Also skip op_map() when the GPUVA has been invalidated, matching the handling in the unmap and remap paths. Found when fuzzing the nouveau driver with a modified Syzkaller. Fixes: b88baab82871 ("drm/nouveau: implement new VM_BIND uAPI") Cc: stable@vger.kernel.org Signed-off-by: Peiyang He Assisted-by: Codex:gpt-5.5 Reviewed-by: Lyude Paul Signed-off-by: Lyude Paul Link: https://patch.msgid.link/0D77BEC410CE0129+20260907052204.1431488-1-peiyang_he@smail.nju.edu.cn --- drivers/gpu/drm/nouveau/nouveau_uvmm.c | 13 +++++++++++-- 1 file changed, 11 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/nouveau/nouveau_uvmm.c b/drivers/gpu/drm/nouveau/nouveau_uvmm.c index fc125fd44a9b..2026fe6b48c6 100644 --- a/drivers/gpu/drm/nouveau/nouveau_uvmm.c +++ b/drivers/gpu/drm/nouveau/nouveau_uvmm.c @@ -846,6 +846,9 @@ op_map(struct nouveau_uvma *uvma) { struct nouveau_bo *nvbo = nouveau_gem_object(uvma->va.gem.obj); + if (drm_gpuva_invalidated(&uvma->va)) + return; + nouveau_uvma_map(uvma, nouveau_mem(nvbo->bo.resource)); } @@ -1232,6 +1235,7 @@ bind_lock_validate(struct nouveau_job *job, struct drm_exec *exec, drm_gpuva_for_each_op(va_op, op->ops) { struct drm_gem_object *obj = op_gem_obj(va_op); + struct nouveau_bo *nvbo; if (unlikely(!obj)) continue; @@ -1246,8 +1250,13 @@ bind_lock_validate(struct nouveau_job *job, struct drm_exec *exec, if (va_op->op == DRM_GPUVA_OP_UNMAP) continue; - ret = nouveau_bo_validate(nouveau_gem_object(obj), - true, false); + nvbo = nouveau_gem_object(obj); + if (!(nvbo->valid_domains & + (NOUVEAU_GEM_DOMAIN_VRAM | NOUVEAU_GEM_DOMAIN_GART))) + return -EINVAL; + + nouveau_bo_placement_set(nvbo, nvbo->valid_domains, 0); + ret = nouveau_bo_validate(nvbo, true, false); if (ret) return ret; } From f7eae6d8d768fabd6b59779ca7da79e02c74e113 Mon Sep 17 00:00:00 2001 From: "Jonghyuk Kim(MalHyuk)" Date: Wed, 2 Sep 2026 10:27:16 +0900 Subject: [PATCH 104/365] drm/nouveau: RCU-free the scheduler-containing nouveau_sched struct nouveau_sched embeds a struct drm_gpu_scheduler (base). nouveau_sched_destroy() calls nouveau_sched_fini() (which does drm_sched_fini(&sched->base)) and then frees the object with plain kfree(sched). drm_sched_fence_get_timeline_name() returns fence->sched->name, and the scheduler fence keeps a .release callback so it is not ops-detached on signalling. A finished fence exported to userspace via drm_syncobj / sync_file therefore keeps pointing at &sched->base after nouveau_sched_destroy(), and a later get_timeline_name() -- reachable unprivileged through SYNC_IOC_FILE_INFO -- dereferences freed memory (KASAN slab-use-after-free read). Per the dma-fence lifetime contract the exporter must keep the data backing a signalled fence alive for an RCU grace period. Free the scheduler-containing object with kfree_rcu() instead of kfree(). Fixes: 5f03a507b29e ("drm/nouveau: implement 1:1 scheduler - entity relationship") Cc: stable@vger.kernel.org Signed-off-by: Jonghyuk Kim(MalHyuk) Reviewed-by: Lyude Paul Signed-off-by: Lyude Paul Link: https://patch.msgid.link/20260902012717.880724-1-malhyuk97@gmail.com --- drivers/gpu/drm/nouveau/nouveau_sched.c | 2 +- drivers/gpu/drm/nouveau/nouveau_sched.h | 1 + 2 files changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/nouveau/nouveau_sched.c b/drivers/gpu/drm/nouveau/nouveau_sched.c index 8b9f935afe09..b3f02c490ecb 100644 --- a/drivers/gpu/drm/nouveau/nouveau_sched.c +++ b/drivers/gpu/drm/nouveau/nouveau_sched.c @@ -517,7 +517,7 @@ nouveau_sched_destroy(struct nouveau_sched **psched) struct nouveau_sched *sched = *psched; nouveau_sched_fini(sched); - kfree(sched); + kfree_rcu(sched, rcu); *psched = NULL; } diff --git a/drivers/gpu/drm/nouveau/nouveau_sched.h b/drivers/gpu/drm/nouveau/nouveau_sched.h index 20cd1da8db73..51ce8dcf6285 100644 --- a/drivers/gpu/drm/nouveau/nouveau_sched.h +++ b/drivers/gpu/drm/nouveau/nouveau_sched.h @@ -98,6 +98,7 @@ void nouveau_job_free(struct nouveau_job *job); struct nouveau_sched { struct drm_gpu_scheduler base; + struct rcu_head rcu; struct drm_sched_entity entity; struct workqueue_struct *wq; struct mutex mutex; From fefd9480ec361969f1a836df46326a1801062c26 Mon Sep 17 00:00:00 2001 From: Guangshuo Li Date: Sat, 8 Aug 2026 21:41:37 +0800 Subject: [PATCH 105/365] drm/nouveau: fix autosuspend cleanup during teardown nouveau_drm_device_init() calls pm_runtime_use_autosuspend(), but nouveau_drm_device_fini() does not call the matching pm_runtime_dont_use_autosuspend(). If the autosuspend delay is set to a negative value while autosuspend is enabled, the runtime PM core increments usage_count to prevent runtime suspend. Without calling pm_runtime_dont_use_autosuspend() during teardown, this reference is not dropped and usage_count remains unbalanced. The documentation for pm_runtime_use_autosuspend() also notes that it is important to undo it with pm_runtime_dont_use_autosuspend() at driver exit time, unless runtime PM was initially enabled with devm_pm_runtime_enable(). Add the missing pm_runtime_dont_use_autosuspend() call to the common device teardown path. This issue was found by manual code inspection. Fixes: 5addcf0a5f0f ("nouveau: add runtime PM support (v0.9)") Cc: stable@vger.kernel.org Signed-off-by: Guangshuo Li Reviewed-by: Lyude Paul Signed-off-by: Lyude Paul Link: https://patch.msgid.link/20260808134137.2864847-1-lgs201920130244@gmail.com --- drivers/gpu/drm/nouveau/nouveau_drm.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/gpu/drm/nouveau/nouveau_drm.c b/drivers/gpu/drm/nouveau/nouveau_drm.c index d0eb29583cb1..2c7077a49888 100644 --- a/drivers/gpu/drm/nouveau/nouveau_drm.c +++ b/drivers/gpu/drm/nouveau/nouveau_drm.c @@ -585,6 +585,7 @@ nouveau_drm_device_fini(struct nouveau_drm *drm) if (nouveau_pmops_runtime()) { pm_runtime_get_sync(dev->dev); pm_runtime_forbid(dev->dev); + pm_runtime_dont_use_autosuspend(dev->dev); } nouveau_led_fini(dev); From 1d653a183973f5283a3db5a38cd5e195eb152244 Mon Sep 17 00:00:00 2001 From: David Carlier Date: Thu, 17 Sep 2026 22:24:07 +0100 Subject: [PATCH 106/365] fprobe: Terminate the fgraph_data list when the reservation is not filled fprobe_fgraph_entry() reserves shadow stack space for every fprobe with an exit handler, but only fills it for those whose entry handler returns 0. fgraph_reserve_data() does not clear the area, so fprobe_return() parses the unused tail as headers left over from an earlier call, and an exit handler can run twice or despite its entry handler asking to skip it. Write a zero word after the last entry to terminate the walk. A zeroed slot does not decode to a NULL fprobe on the arches that encode the header into one unsigned long, since arch_decode_fprobe_header_fp() ORs in FPROBE_HEADER_MSB_PATTERN, so make read_fprobe_header() return NULL for a zeroed slot. Link: https://lore.kernel.org/all/20260917212407.384468-1-devnexen@gmail.com/ Fixes: e0a384434ae1 ("tracing: fprobe: do not zero out unused fgraph_data") Cc: stable@vger.kernel.org Suggested-by: Masami Hiramatsu (Google) Signed-off-by: David Carlier Signed-off-by: Masami Hiramatsu (Google) --- kernel/trace/fprobe.c | 15 +++++++++++++++ 1 file changed, 15 insertions(+) diff --git a/kernel/trace/fprobe.c b/kernel/trace/fprobe.c index 1e9b00997ff2..9f2d98181779 100644 --- a/kernel/trace/fprobe.c +++ b/kernel/trace/fprobe.c @@ -171,6 +171,11 @@ static inline bool write_fprobe_header(unsigned long *stack, static inline void read_fprobe_header(unsigned long *stack, struct fprobe **fp, unsigned int *size_words) { + if (!*stack) { + *fp = NULL; + *size_words = 0; + return; + } *fp = arch_decode_fprobe_header_fp(*stack); *size_words = arch_decode_fprobe_header_size(*stack); } @@ -203,6 +208,12 @@ static inline void read_fprobe_header(unsigned long *stack, { struct __fprobe_header *fph = (struct __fprobe_header *)stack; + if (!*stack) { + *fp = NULL; + *size_words = 0; + return; + } + *fp = fph->fp; *size_words = fph->size_words; } @@ -635,6 +646,10 @@ static int fprobe_fgraph_entry(struct ftrace_graph_ent *trace, struct fgraph_ops } } + /* Terminate the list, fgraph_reserve_data() does not clear it. */ + if (used && used < reserved_words) + fgraph_data[used] = 0; + /* If any exit_handler is set, data must be used. */ return used != 0; } From 5179241521401ef364294128bb43cdbce7252457 Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Thu, 17 Sep 2026 13:20:25 +0200 Subject: [PATCH 107/365] fs: don't create the private nullfs mount under namespace_sem init_mount_tree() mounts the mutable rootfs on top of nullfs via LOCK_MOUNT_EXACT(). That declares a pinned mountpoint with a cleanup attribute in the scope of the whole function so the nullfs root inode lock and namespace_sem are only dropped when init_mount_tree() returns. This became a problem when the private nullfs instance for kthreads was added. kern_mount() allocates a new superblock and alloc_super() takes the new s_umount with SINGLE_DEPTH_NESTING and then shrinker_mutex via shrinker_alloc(). Doing that with namespace_sem held teaches lockdep the dependency namespace_sem -> s_umount/1 -> shrinker_mutex With CONFIG_SHRINKER_DEBUG shrinker_debugfs_rename() takes the debugfs directory inode lock under shrinker_mutex every time a block device is mounted and lock_mount_exact() takes namespace_sem under the inode lock of the mountpoint for every mount. So mounting anything on debugfs, e.g. the tracefs automount on /sys/kernel/debug/tracing, closes the cycle: WARNING: possible circular locking dependency detected 7.3.0-rc3+ #17 Not tainted ------------------------------------------------------ rasdaemon/4449 is trying to acquire lock: (namespace_sem){++++}-{4:4}, at: lock_mount_exact+0x4c/0x308 but task is already holding lock: (&sb->s_type->i_mutex_key#17){++++}-{4:4}, at: lock_mount_exact+0x3c/0x308 which lock already depends on the new lock. ... Chain exists of: namespace_sem --> shrinker_mutex --> &sb->s_type->i_mutex_key#17 This can't actually deadlock. init_mount_tree() runs single-threaded during early boot before any other task exists and nothing allocates a superblock under namespace_sem after that. But lockdep can't know that and disables itself for the rest of the boot. Move mounting the rootfs on top of nullfs into a helper so the locks are dropped when it returns. Fixes: 32750c77e811 ("fs: start all kthreads in nullfs") Reported-by: Zenghui Yu Closes: https://lore.kernel.org/15174353-3f4a-a1ca-5bd1-ea2a4c77828e@huawei.com Link: https://patch.msgid.link/20260917-atemtechnik-bleichen-befassen-9a57db01baf0@brauner Signed-off-by: Christian Brauner (Amutable) --- fs/namespace.c | 25 ++++++++++++++++--------- 1 file changed, 16 insertions(+), 9 deletions(-) diff --git a/fs/namespace.c b/fs/namespace.c index ae5dc64f8b45..5e41021eaa63 100644 --- a/fs/namespace.c +++ b/fs/namespace.c @@ -6184,6 +6184,21 @@ struct mnt_namespace init_mnt_ns = { .poll = __WAIT_QUEUE_HEAD_INITIALIZER(init_mnt_ns.poll), }; +static void __init mount_rootfs_on_nullfs(struct vfsmount *mnt, + struct vfsmount *nullfs_mnt) +{ + struct path root = { + .mnt = nullfs_mnt, + .dentry = nullfs_mnt->mnt_root, + }; + + LOCK_MOUNT_EXACT(mp, &root); + if (unlikely(IS_ERR(mp.parent))) + panic("VFS: Failed to mount rootfs on nullfs"); + scoped_guard(mount_writer) + attach_mnt(real_mount(mnt), mp.parent, mp.mp); +} + static void __init init_mount_tree(void) { struct vfsmount *mnt, *nullfs_mnt; @@ -6215,15 +6230,7 @@ static void __init init_mount_tree(void) mnt_root = real_mount(nullfs_mnt); init_mnt_ns.root = mnt_root; - /* Mount mutable rootfs on top of nullfs. */ - root.mnt = nullfs_mnt; - root.dentry = nullfs_mnt->mnt_root; - - LOCK_MOUNT_EXACT(mp, &root); - if (unlikely(IS_ERR(mp.parent))) - panic("VFS: Failed to mount rootfs on nullfs"); - scoped_guard(mount_writer) - attach_mnt(real_mount(mnt), mp.parent, mp.mp); + mount_rootfs_on_nullfs(mnt, nullfs_mnt); pr_info("VFS: Finished mounting rootfs on nullfs\n"); From d54a489c8c4b627775445d54a6cbd32f209bda8f Mon Sep 17 00:00:00 2001 From: Frank Wunderlich Date: Thu, 17 Sep 2026 17:37:11 +0200 Subject: [PATCH 108/365] gpiolib: use of_node_name if line-name is missing Until v7.0, GPIO hogs inherited the DT node name when no line-name property was specified. This was implemented as a fallback in of_parse_own_gpio(). Commit d1d564ec4992 ("gpio: move hogs into GPIO core") moved hog parsing into the GPIO core and removed this fallback. Consequently, GPIO hogs without a line-name property are now displayed with a ? in /sys/kernel/debug/gpio. Restore the old fallback. Fixes: d1d564ec4992 ("gpio: move hogs into GPIO core") Signed-off-by: Frank Wunderlich Reviewed-by: Andy Shevchenko Link: https://patch.msgid.link/20260917153712.134367-1-linux@fw-web.de Signed-off-by: Bartosz Golaszewski --- drivers/gpio/gpiolib.c | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/drivers/gpio/gpiolib.c b/drivers/gpio/gpiolib.c index ef8ccaf17c9c..28f7265c5d56 100644 --- a/drivers/gpio/gpiolib.c +++ b/drivers/gpio/gpiolib.c @@ -1029,6 +1029,13 @@ int gpiochip_add_hog(struct gpio_chip *gc, struct fwnode_handle *fwnode) ret = of_gpiochip_get_lflags(gc, &gpiospec, &lflags); if (ret) return ret; + + /* + * If no line-name property is present, fall back to the OF + * node name as in the previous implementation. + */ + if (!name) + name = to_of_node(fwnode)->name; } else { /* * GPIO_ACTIVE_LOW is currently the only lookup flag From 4f948b5949d2e418b4506752e7c514fe91bd408b Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Fri, 18 Sep 2026 11:19:59 +0200 Subject: [PATCH 109/365] binfmt_misc: fix OOB read in bpf_binprm_select_interp() bpf_binprm_select_interp() checks the name its load program passes with strnlen(name, name__sz) and then hands the same buffer to binfmt_misc_find_interp(), which compares it with an unbounded strcmp(). The buffer can be a BPF map value that another CPU rewrites between the two reads. If the terminating NUL is overwritten in that window, strcmp() reads past the name__sz bytes the verifier checked. That is an out-of-bounds read of up to 31 bytes of whatever follows the checked name__sz bytes. The verifier checks the name and name__sz pair with BPF_READ | BPF_WRITE, so a writable array map value is an accepted argument. bpf(BPF_MAP_UPDATE_ELEM) on an array map copies the new value over the old one in place and takes no lock. The NUL that strnlen() finds can be overwritten before strcmp() reads the buffer again: CPU0 CPU1 bpf_binprm_select_interp() strnlen(name, name__sz) finds the NUL inside name__sz bpf(BPF_MAP_UPDATE_ELEM) array_map_update_elem() copy_map_value() overwrites the NUL binfmt_misc_find_interp() strcmp(interp->name, name) reads past name__sz strnlen() proves that a NUL lies inside name__sz only at the moment it runs. The map update on CPU1 takes no lock, so it can store over the NUL right after. The lookup on CPU0 then walks the live buffer again, once per bound interpreter: fs/binfmt_misc.c:binfmt_misc_find_interp list_for_each_entry(interp, interps, list) if (!strcmp(interp->name, name)) return interp; strcmp() stops at the first mismatch or at the end of interp->name. bm_entry_add_interp() caps a bound name at BINFMT_MISC_INTERP_NAME_MAX (32) bytes, so strcmp() reads at most 33 bytes of name. The smallest name__sz the kfunc accepts is 2, which leaves up to 31 bytes read beyond the checked extent. The handler's own load program has to pass a writable map value, and something has to store into it while the kfunc runs. The window between strnlen() and strcmp() is short, but with a BPF_F_MMAPABLE array the store is a plain user space write into the mapped value, so a loop can hit it without a single bpf() call. Copy the name into a stack buffer of BINFMT_MISC_INTERP_NAME_MAX + 1 bytes, terminate it, and look up the copy. The memcpy() length is below name__sz, so the copy stays inside the extent the verifier checked, and the BPF buffer is not read again afterwards. Return -ENOENT first for a name longer than BINFMT_MISC_INTERP_NAME_MAX. bm_entry_add_interp() rejects a longer name, and the only other binding site attaches the empty name. No entry can bind such a name, so that lookup already ended in -ENOENT and no result changes. Check the first byte of the copy and return -EINVAL if it is NUL, as the existing "!len" test does for an empty name. Only an 'F' entry binds the empty name and a 'B' entry cannot carry 'F', so without that check a racing store of NUL to byte 0 would look up a name no entry binds and end in -ENOENT rather than -EINVAL. A NUL stored further into the name only shortens it to another name the program could have passed anyway. binfmt_misc_find_interp() itself is left alone: entry_attach_interpreter() calls it with a kernel string, and this kfunc now calls it with a private copy. Fixes: 6ec7c96bee30 ("binfmt_misc: let a 'B' entry bind its interpreters") Signed-off-by: Chris Mason Link: https://patch.msgid.link/20260918-work-binfmt_misc-fixes-v1-1-647b24bc1c46@kernel.org Signed-off-by: Christian Brauner (Amutable) --- fs/binfmt_misc_bpf.c | 15 ++++++++++++++- 1 file changed, 14 insertions(+), 1 deletion(-) diff --git a/fs/binfmt_misc_bpf.c b/fs/binfmt_misc_bpf.c index 91576ff05911..ce1bc78e8511 100644 --- a/fs/binfmt_misc_bpf.c +++ b/fs/binfmt_misc_bpf.c @@ -176,6 +176,7 @@ __bpf_kfunc int bpf_binprm_select_interp(struct linux_binprm *bprm, const char *name, size_t name__sz) { const struct binfmt_misc_interp *interp; + char buf[BINFMT_MISC_INTERP_NAME_MAX + 1]; size_t len; char *path; @@ -184,8 +185,20 @@ __bpf_kfunc int bpf_binprm_select_interp(struct linux_binprm *bprm, len = strnlen(name, name__sz); if (len == name__sz || !len) return -EINVAL; + /* No entry binds a longer name, so it cannot be found. */ + if (len > BINFMT_MISC_INTERP_NAME_MAX) + return -ENOENT; - interp = binfmt_misc_find_interp(bprm->bpf_interps, name); + /* + * The program may pass memory that is written to while this runs, + * so look the name up in a private copy and check that instead. + */ + memcpy(buf, name, len); + buf[len] = '\0'; + if (!buf[0]) + return -EINVAL; + + interp = binfmt_misc_find_interp(bprm->bpf_interps, buf); if (!interp) return -ENOENT; From 1970fc4ecb52dabce2e57f9be721964b936a052d Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Fri, 18 Sep 2026 11:20:00 +0200 Subject: [PATCH 110/365] binfmt_misc: fix racy checks in bpf set_interp kfuncs bpf_binprm_set_interp() tests path[0] != '/' on the buffer its load program passes and then reads the same buffer again to copy it with kmemdup_nul(). The buffer can be a BPF map value that another CPU rewrites between the two reads. If byte 0 is overwritten in that window, the kfunc stages a relative or empty interpreter path. The staged path is not checked again, so open_exec() resolves a relative path against the working directory of the task doing the exec. bpf_binprm_set_interp_arg() has the same pattern for its "!len" test and can stage an empty argument, which the interpreter then receives as an empty argv entry. The verifier checks the path and path__sz pair with BPF_READ | BPF_WRITE, so a writable array map value is an accepted argument. bpf(BPF_MAP_UPDATE_ELEM) on an array map copies the new value over the old one in place and takes no lock. Both kfuncs are KF_SLEEPABLE and allocate with GFP_KERNEL between the test and the copy, so the task can sleep inside the window: load program bpf(BPF_MAP_UPDATE_ELEM) bpf_binprm_set_interp() strnlen(path, path__sz) path[0] != '/' is false kmemdup_nul(path, len, GFP_KERNEL) allocation may sleep array_map_update_elem() copy_map_value() rewrites byte 0 copy reads path again bm_bpf_stage_selection() The test in the load program's column proves what byte 0 held only at the moment the test ran. The map update takes no lock, so it can store to byte 0 right after. kmemdup_nul() then copies the rewritten bytes, and bm_bpf_stage_selection() publishes them as bprm->bpf_interp. The staged path is not checked again on its way to open_exec(): load_misc_binary() entry_select_interpreter() returns bprm->bpf_interp unchanged build_interp_argv() copy_string_kernel() copies it as argv[0] bprm_change_interp() kstrdup() entry_open_interpreter() open_exec() unless a bound file is staged or the entry is an 'F' entry None of these functions tests the first byte, and load_misc_binary() hands the pointer to nothing else. In bpf_binprm_set_interp_arg(), strnlen() finds a non-zero len, a NUL is then stored to byte 0, and build_interp_argv() later copies the empty bprm->bpf_interp_arg with copy_string_kernel(). The handler's own load program has to pass a writable map value, and something has to store into it while the kfunc runs. The allocation can sleep inside the window, and with a BPF_F_MMAPABLE array the store is a plain user space write into the mapped value, so a loop can hit it without a single bpf() call. Check the private copy in both kfuncs, so that the string that gets staged is the string that was checked. bpf_binprm_select_interp() already looks its name up in a private copy for the same reason. The remaining tests work on path__sz, arg__sz or the local len, and the copy length is len, so the copy stays inside the extent the verifier checked. Results of bpf_binprm_set_interp() with the check on the copy: - A NUL stored to byte 0 fails interp[0] != '/' and gets -EINVAL. - For len == 0, kmemdup_nul() returns an empty string, so an empty path still gets -EINVAL. - A NUL stored further into the string only shortens it to another absolute path, or another non-empty argument, that the program could have passed anyway. - A path that both lacks the leading '/' and is PATH_MAX or longer now gets -ENAMETOOLONG instead of -EINVAL. - A path that is empty or lacks the leading '/' is now rejected after the copy rather than before it, so such a call makes an allocation and returns -ENOMEM instead of -EINVAL if that allocation fails. bpf_binprm_set_interp_arg() still rejects an empty argument before allocating, so its results are unchanged apart from the raced case fixed here. Both new checks run before the previously staged string is freed or replaced. A failing call frees only its own allocation and leaves the earlier selection in place, as the -ENOMEM path already does. Fixes: b4bfe2f6b011 ("binfmt_misc: add binfmt_misc_ops bpf struct_ops") Signed-off-by: Chris Mason Link: https://patch.msgid.link/20260918-work-binfmt_misc-fixes-v1-2-647b24bc1c46@kernel.org Signed-off-by: Christian Brauner (Amutable) --- fs/binfmt_misc_bpf.c | 20 ++++++++++++++++++-- 1 file changed, 18 insertions(+), 2 deletions(-) diff --git a/fs/binfmt_misc_bpf.c b/fs/binfmt_misc_bpf.c index ce1bc78e8511..a3e26e8a4027 100644 --- a/fs/binfmt_misc_bpf.c +++ b/fs/binfmt_misc_bpf.c @@ -141,8 +141,6 @@ __bpf_kfunc int bpf_binprm_set_interp(struct linux_binprm *bprm, len = strnlen(path, path__sz); if (len == path__sz) return -EINVAL; - if (path[0] != '/') - return -EINVAL; if (len >= PATH_MAX) return -ENAMETOOLONG; @@ -150,6 +148,15 @@ __bpf_kfunc int bpf_binprm_set_interp(struct linux_binprm *bprm, if (!interp) return -ENOMEM; + /* + * The program may pass memory that is written to while this runs, + * so check the private copy and not the buffer it was made from. + */ + if (interp[0] != '/') { + kfree(interp); + return -EINVAL; + } + bm_bpf_stage_selection(bprm, interp, NULL); return 0; } @@ -241,6 +248,15 @@ __bpf_kfunc int bpf_binprm_set_interp_arg(struct linux_binprm *bprm, if (!val) return -ENOMEM; + /* + * The program may pass memory that is written to while this runs, + * so check the private copy and not the buffer it was made from. + */ + if (!val[0]) { + kfree(val); + return -EINVAL; + } + kfree(bprm->bpf_interp_arg); bprm->bpf_interp_arg = val; return 0; From aff09d9e37e02dc60bde79035ac15b136d602259 Mon Sep 17 00:00:00 2001 From: Dan Carpenter Date: Fri, 18 Sep 2026 15:16:17 +0200 Subject: [PATCH 111/365] drm/nouveau/clk: fix list cursor use after loop in nvkm_clk_ustate_update If list_for_each_entry() exits without hitting a break then "pstate" is not a valid pstate pointer. Introduce a "found" variable instead. The check is reachable from userspace: nvkm_clk_ustate_update() takes the pstate id straight from the 'pstate' debugfs file, so requesting an id that is not in clk->states - or any id at all when the perf tables are broken and the list is empty - makes the pstate->pstate != req test dereference the list head cast to a struct nvkm_pstate, which is an out-of-bounds read. Fixes: 7c8565220697 ("drm/nouveau/clk: implement power state and engine clock control in core") Signed-off-by: Dan Carpenter [Francesco: rebased on drm-misc-next, expanded the commit message] Signed-off-by: Francesco Magazzu Reviewed-by: Lyude Paul Signed-off-by: Lyude Paul Link: https://patch.msgid.link/20260918131620.405133-2-postadelmaga@gmail.com --- drivers/gpu/drm/nouveau/nvkm/subdev/clk/base.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/clk/base.c b/drivers/gpu/drm/nouveau/nvkm/subdev/clk/base.c index 572e63846315..5da82db71dde 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/clk/base.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/clk/base.c @@ -473,6 +473,7 @@ static int nvkm_clk_ustate_update(struct nvkm_clk *clk, int req) { struct nvkm_pstate *pstate; + bool found = false; int i = 0; if (!clk->allow_reclock) @@ -480,12 +481,14 @@ nvkm_clk_ustate_update(struct nvkm_clk *clk, int req) if (req != -1 && req != -2) { list_for_each_entry(pstate, &clk->states, head) { - if (pstate->pstate == req) + if (pstate->pstate == req) { + found = true; break; + } i++; } - if (pstate->pstate != req) + if (!found) return -EINVAL; req = i; } From 866dbd17c3d5f599b9d93ea9bd5b3d6859ff8350 Mon Sep 17 00:00:00 2001 From: Francesco Magazzu Date: Fri, 18 Sep 2026 15:16:18 +0200 Subject: [PATCH 112/365] drm/nouveau/clk: don't use the pstate cursor after the loop nvkm_pstate_prog() walks clk->states looking for the entry at index 'pstatei' and then keeps using the list_for_each_entry cursor after the loop. This is not triggerable today: every caller clamps the index against clk->state_nr before calling, so the loop always breaks on a real entry. It is safe by virtue of what the callers happen to do, not by anything the function itself checks. Should a caller ever pass an index that is not on the list, the cursor would point at the list head rather than at a pstate, and the pstate->base.domain[] and pstate->fanspeed accesses that follow would read past it. Rather than leave that trap in place for the next caller, track whether the entry was found and return -EINVAL if it was not. No functional change. Signed-off-by: Francesco Magazzu Reviewed-by: Lyude Paul Signed-off-by: Lyude Paul Link: https://patch.msgid.link/20260918131620.405133-3-postadelmaga@gmail.com --- drivers/gpu/drm/nouveau/nvkm/subdev/clk/base.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/clk/base.c b/drivers/gpu/drm/nouveau/nvkm/subdev/clk/base.c index 5da82db71dde..a43246ae681f 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/clk/base.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/clk/base.c @@ -270,13 +270,19 @@ nvkm_pstate_prog(struct nvkm_clk *clk, int pstatei) struct nvkm_fb *fb = subdev->device->fb; struct nvkm_pci *pci = subdev->device->pci; struct nvkm_pstate *pstate; + bool found = false; int ret, idx = 0; list_for_each_entry(pstate, &clk->states, head) { - if (idx++ == pstatei) + if (idx++ == pstatei) { + found = true; break; + } } + if (!found) + return -EINVAL; + nvkm_debug(subdev, "setting performance state %d\n", pstatei); clk->pstate = pstatei; From 7ca7b8b5f2cc89ee843c2eab3272aac5aafb7b73 Mon Sep 17 00:00:00 2001 From: Francesco Magazzu Date: Fri, 18 Sep 2026 15:16:19 +0200 Subject: [PATCH 113/365] drm/nouveau/device: don't use the pstate cursor after the loop nvkm_control_mthd_pstate_attr() looks up the pstate at the index supplied by userspace by walking clk->states, and then keeps using the list_for_each_entry cursor after the loop. This is not triggerable today: the function already rejects args->v0.state >= clk->state_nr before the loop, and clk->state_nr is kept in sync with the number of entries on clk->states, so the lookup always breaks on a real entry. Should the loop ever run to completion, the cursor would point at the list head rather than at a pstate, and the pstate->base.domain[] read and the walk of pstate->list that follow would read past it. Rather than leave that trap in place, track whether the entry was found and return -EINVAL if it was not, like the other lookup failures in this function. No functional change. Signed-off-by: Francesco Magazzu Reviewed-by: Lyude Paul Signed-off-by: Lyude Paul Link: https://patch.msgid.link/20260918131620.405133-4-postadelmaga@gmail.com --- drivers/gpu/drm/nouveau/nvkm/engine/device/ctrl.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/device/ctrl.c b/drivers/gpu/drm/nouveau/nvkm/engine/device/ctrl.c index f2e9a06263ce..28702741a88b 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/device/ctrl.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/device/ctrl.c @@ -74,6 +74,7 @@ nvkm_control_mthd_pstate_attr(struct nvkm_control *ctrl, void *data, u32 size) const struct nvkm_domain *domain; struct nvkm_pstate *pstate; struct nvkm_cstate *cstate; + bool found = false; int i = 0, j = -1; u32 lo, hi; int ret = -ENOSYS; @@ -104,10 +105,15 @@ nvkm_control_mthd_pstate_attr(struct nvkm_control *ctrl, void *data, u32 size) if (args->v0.state != NVIF_CONTROL_PSTATE_ATTR_V0_STATE_CURRENT) { list_for_each_entry(pstate, &clk->states, head) { - if (i++ == args->v0.state) + if (i++ == args->v0.state) { + found = true; break; + } } + if (!found) + return -EINVAL; + lo = pstate->base.domain[domain->name]; hi = lo; list_for_each_entry(cstate, &pstate->list, head) { From e5cccdafc855cd5f96f4b51d38114a0360b075d7 Mon Sep 17 00:00:00 2001 From: Francesco Magazzu Date: Fri, 18 Sep 2026 15:16:20 +0200 Subject: [PATCH 114/365] drm/nouveau/clk: don't clobber reclock status when restoring volt/fan nvkm_cstate_prog() reuses 'ret' for the voltage and fan-speed restore calls it makes after reprogramming the clocks. Those calls almost always succeed, so the status of the reclock itself is overwritten and the function reports success even when clk->func->calc() or clk->func->prog() failed. The converse is also true: a successful reclock is reported as an error if the final restore call fails, even though that failure is only logged and otherwise ignored. The only consumer of the return value is the error message in nvkm_pstate_work(), so in practice a failing reclock is simply never reported. Nothing else changes, but a function that returns success on failure is a trap for the next caller. Keep the calc/prog status in 'ret' and use a separate local for the restore calls. Fixes: 3eca809b3c05 ("drm/nouveau/clk: cosmetic changes") Signed-off-by: Francesco Magazzu Reviewed-by: Lyude Paul Signed-off-by: Lyude Paul Link: https://patch.msgid.link/20260918131620.405133-5-postadelmaga@gmail.com --- drivers/gpu/drm/nouveau/nvkm/subdev/clk/base.c | 16 +++++++++------- 1 file changed, 9 insertions(+), 7 deletions(-) diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/clk/base.c b/drivers/gpu/drm/nouveau/nvkm/subdev/clk/base.c index a43246ae681f..1cb83edc78dc 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/clk/base.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/clk/base.c @@ -199,16 +199,18 @@ nvkm_cstate_prog(struct nvkm_clk *clk, struct nvkm_pstate *pstate, int cstatei) } if (volt) { - ret = nvkm_volt_set_id(volt, cstate->voltage, - pstate->base.voltage, clk->temp, -1); - if (ret && ret != -ENODEV) - nvkm_error(subdev, "failed to lower voltage: %d\n", ret); + int err = nvkm_volt_set_id(volt, cstate->voltage, + pstate->base.voltage, clk->temp, -1); + + if (err && err != -ENODEV) + nvkm_error(subdev, "failed to lower voltage: %d\n", err); } if (therm) { - ret = nvkm_therm_cstate(therm, pstate->fanspeed, -1); - if (ret && ret != -ENODEV) - nvkm_error(subdev, "failed to lower fan speed: %d\n", ret); + int err = nvkm_therm_cstate(therm, pstate->fanspeed, -1); + + if (err && err != -ENODEV) + nvkm_error(subdev, "failed to lower fan speed: %d\n", err); } return ret; From 6a6870d3077faa501ca97760057ddca22b68418d Mon Sep 17 00:00:00 2001 From: Peiyang He Date: Fri, 18 Sep 2026 10:53:12 +0800 Subject: [PATCH 115/365] drm/nouveau: don't bump pin count on failed re-pin in nouveau_bo_pin_locked() nouveau_bo_pin_locked() checks whether an already pinned BO is in a memory domain compatible with a new pin request. When the domains are incompatible, it sets -EBUSY but still calls ttm_bo_pin() before returning. Callers treat a failed nouveau_bo_pin() as not having acquired a new pin, so the extra pin count is never decreased by a matching unpin. This triggers the warning in ttm_bo_release(): WARN_ON_ONCE(bo->pin_count); Found when fuzzing the nouveau driver with a modified Syzkaller: WARNING: drivers/gpu/drm/ttm/ttm_bo.c:256 at ttm_bo_release+0x827/0x9e0 drivers/gpu/drm/ttm/ttm_bo.c:256, CPU#1: syz.3.24/2212 Modules linked in: CPU: 1 UID: 0 PID: 2212 Comm: syz.3.24 Not tainted 7.2.0 #24 PREEMPT(lazy) nouveau 0000:01:00.0: gsp:msg fn:103 len:0x40/0x20 res:0x19 resp:0x19 Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.3-debian-1.16.3-2 04/01/2014 RIP: 0010:ttm_bo_release+0x827/0x9e0 drivers/gpu/drm/ttm/ttm_bo.c:256 Code: 02 00 0f 85 51 01 00 00 48 8b 7b 08 e8 d2 20 01 00 e9 80 fd ff ff e8 d8 15 c0 fe 90 0f 0b 90 e9 e1 f8 ff ff e8 ca 15 c0 fe 90 <0f> 0b 90 e9 a4 f8 ff ff e8 bc 15 c0 fe be 03 00 00 00 4c 89 e7 e8 msg: 00000000: 05 00 d0 c1 04 00 f0 f1 01 30 00 00 2d 90 00 00 .........0..-... RSP: 0018:ffffc9000f5cf710 EFLAGS: 00010293 RAX: 0000000000000000 RBX: ffff888018e5d2a8 RCX: ffffffff82bb1b36 RDX: ffff888017b68000 RSI: 0000000000000004 RDI: ffff888018e5d2a8 msg: 00000010: 19 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ RBP: ffff88801261c720 R08: 0000000000000001 R09: ffffed10031cba55 R10: ffff888018e5d2ab R11: 00000000000000f3 R12: ffff888018e5d290 R13: ffff888018e5d2d4 R14: ffff88801b219c18 R15: dffffc0000000000 FS: 0000000000000000(0000) GS:ffff8880e0f6f000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000001b31223ffc CR3: 0000000028e00005 CR4: 0000000000770ef0 PKRU: 80000000 Call Trace: kref_put include/linux/kref.h:65 [inline] ttm_bo_put drivers/gpu/drm/ttm/ttm_bo.c:325 [inline] ttm_bo_fini+0x55/0x80 drivers/gpu/drm/ttm/ttm_bo.c:330 nouveau_gem_object_del+0xb2/0x1b0 drivers/gpu/drm/nouveau/nouveau_gem.c:90 drm_gem_object_free+0x5f/0x90 drivers/gpu/drm/drm_gem.c:1165 kref_put include/linux/kref.h:65 [inline] __drm_gem_object_put include/drm/drm_gem.h:562 [inline] drm_gem_object_put include/drm/drm_gem.h:575 [inline] nouveau_abi16_chan_fini.constprop.0+0x44f/0x5a0 drivers/gpu/drm/nouveau/nouveau_abi16.c:195 nouveau 0000:01:00.0: syz.2.23[2209]: Unknown handle 0x00000000 nouveau_abi16_fini+0x1d0/0x340 drivers/gpu/drm/nouveau/nouveau_abi16.c:225 nouveau_drm_postclose+0x18b/0x3e0 drivers/gpu/drm/nouveau/nouveau_drm.c:1284 nouveau 0000:01:00.0: syz.2.23[2209]: validate_init drm_file_free.part.0+0x6d6/0xb60 drivers/gpu/drm/drm_file.c:267 drm_file_free drivers/gpu/drm/drm_file.c:237 [inline] drm_close_helper.isra.0+0x11a/0x160 drivers/gpu/drm/drm_file.c:290 drm_release+0x1ab/0x330 drivers/gpu/drm/drm_file.c:438 __fput+0x39c/0xa60 fs/file_table.c:512 nouveau 0000:01:00.0: syz.2.23[2209]: validate: -2 task_work_run+0x15a/0x230 kernel/task_work.c:233 exit_task_work include/linux/task_work.h:40 [inline] do_exit+0x82b/0x25a0 kernel/exit.c:1009 do_group_exit+0xc2/0x280 kernel/exit.c:1152 get_signal+0x1d6e/0x1f30 kernel/signal.c:3046 arch_do_signal_or_restart+0x7d/0x6e0 arch/x86/kernel/signal.c:337 __exit_to_user_mode_loop kernel/entry/common.c:66 [inline] exit_to_user_mode_loop+0xdf/0x440 kernel/entry/common.c:101 __exit_to_user_mode_prepare include/linux/irq-entry-common.h:207 [inline] syscall_exit_to_user_mode_prepare include/linux/irq-entry-common.h:230 [inline] syscall_exit_to_user_mode include/linux/entry-common.h:318 [inline] do_syscall_64+0x4f8/0x690 arch/x86/entry/syscall_64.c:100 entry_SYSCALL_64_after_hwframe+0x77/0x7f RIP: 0033:0x7f12bac8594d Code: Unable to access opcode bytes at 0x7f12bac85923. RSP: 002b:00007f12b96e70d8 EFLAGS: 00000246 ORIG_RAX: 00000000000000ca RAX: 0000000000000001 RBX: 00007f12baf15fa8 RCX: 00007f12bac8594d RDX: 00000000000f4240 RSI: 0000000000000081 RDI: 00007f12baf15fac RBP: 00007f12baf15fa0 R08: 00007f12baee8000 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000 R13: 00007f12baf16038 R14: 0000000000000006 R15: 00007ffe2ed394b0 irq event stamp: 47867 hardirqs last enabled at (47883): [] __up_console_sem+0x66/0x70 kernel/printk/printk.c:347 hardirqs last disabled at (47892): [] __up_console_sem+0x4b/0x70 kernel/printk/printk.c:345 softirqs last enabled at (47880): [] __do_softirq kernel/softirq.c:656 [inline] softirqs last enabled at (47880): [] invoke_softirq kernel/softirq.c:496 [inline] softirqs last enabled at (47880): [] __irq_exit_rcu+0x137/0x1c0 kernel/softirq.c:735 softirqs last disabled at (47875): [] __do_softirq kernel/softirq.c:656 [inline] softirqs last disabled at (47875): [] invoke_softirq kernel/softirq.c:496 [inline] softirqs last disabled at (47875): [] __irq_exit_rcu+0x137/0x1c0 kernel/softirq.c:735 Fix by calling ttm_bo_pin() only when the existing placement is compatible with the new pin request. This matches the correct behavior in other DRM drivers such as amdgpu_bo_pin() in amdgpu. Cc: stable@vger.kernel.org Fixes: ad76b3f7c7a0 ("drm/nouveau: teach nouveau_bo_pin() how to force a contig vram allocation") Signed-off-by: Peiyang He Assisted-by: LLM Reviewed-by: Lyude Paul Signed-off-by: Lyude Paul Link: https://patch.msgid.link/EACEF2F4E098413F+20260918025312.2814889-1-peiyang_he@smail.nju.edu.cn --- drivers/gpu/drm/nouveau/nouveau_bo.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c b/drivers/gpu/drm/nouveau/nouveau_bo.c index 0e8de6d4b36f..6dcb92575eb4 100644 --- a/drivers/gpu/drm/nouveau/nouveau_bo.c +++ b/drivers/gpu/drm/nouveau/nouveau_bo.c @@ -578,8 +578,9 @@ int nouveau_bo_pin_locked(struct nouveau_bo *nvbo, uint32_t domain, bool contig) "0x%08x vs 0x%08x\n", bo, bo->resource->mem_type, domain); ret = -EBUSY; + } else { + ttm_bo_pin(&nvbo->bo); } - ttm_bo_pin(&nvbo->bo); goto out; } From 1717fcc5be575d4768279148ae9465a8b13d4339 Mon Sep 17 00:00:00 2001 From: Giuseppe Ranieri Date: Thu, 17 Sep 2026 21:51:14 +0000 Subject: [PATCH 116/365] drm/nouveau/disp: don't reject HDMI config on cards without SCDC nv50_hdmi_enable() passes the sink's SCDC capability from its EDID straight through to nvif_outp_hdmi(). On pre-Maxwell-2 cards there is no hdmi->scdc callback, so nvkm_uoutp_mthd_hdmi() rejects the whole configuration with -EINVAL, and nv50_hdmi_enable() returns before hdmi->ctrl() runs and before the AVI and VSI infoframes are sent. The result on such a card driving an SCDC-capable HDMI 2.0 sink is that HDMI audio silently stops working. Video is unaffected, and nothing is logged, which makes the failure hard to attribute. SCDC is optional, and the hdmi->scdc() call further down is already guarded against a missing callback. Requesting it on a card that cannot do it need not invalidate the rest of the HDMI configuration, so drop that term from the condition and let the existing guard skip SCDC alone. Fixes: 6c6abab20b99 ("drm/nouveau/disp: add output hdmi config method") Signed-off-by: Giuseppe Ranieri Co-Authored-By: Tano Dzhinski Signed-off-by: Tano Dzhinski Tested-by: Tano Dzhinski Reviewed-by: Lyude Paul Signed-off-by: Lyude Paul Link: https://patch.msgid.link/20260917215114.1136715-1-tano.dzhinski@gmail.com --- drivers/gpu/drm/nouveau/nvkm/engine/disp/uoutp.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/uoutp.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/uoutp.c index 377d0e0cef84..9887b3898505 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/uoutp.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/uoutp.c @@ -253,8 +253,7 @@ nvkm_uoutp_mthd_hdmi(struct nvkm_outp *outp, void *argv, u32 argc) if (!ior->func->hdmi || args->v0.max_ac_packet > 0x1f || - args->v0.rekey > 0x7f || - (args->v0.scdc && !ior->func->hdmi->scdc)) + args->v0.rekey > 0x7f) return -EINVAL; if (!args->v0.enable) { From d58384c22739848efe14b34e9586e4f1242f33c0 Mon Sep 17 00:00:00 2001 From: Liz Fong-Jones Date: Fri, 18 Sep 2026 03:56:33 +0000 Subject: [PATCH 117/365] PCI: Fix BAR resize for devices on a root bus MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit pci_do_resource_release_and_resize() releases device BARs that share a bridge window with the BAR being resized, but when the device sits directly on a root bus (pdev->bus->self == NULL) it then skips resource assignment entirely and returns success, leaving the BARs it just released unassigned (IORESOURCE_UNSET). Skipping pbus_reassign_bridge_resources() is correct in that case -- there is no bridge window to adjust -- but the device BARs still have to be reassigned. Before the BAR release was consolidated into the PCI core, this case worked for amdgpu because the driver released the BARs itself and then called pci_assign_unassigned_bus_resources() unconditionally after the resize, which assigns unassigned device BARs also on a root bus. Commit db92e3fef53e ("drm/amdgpu: Remove driver side BAR release before resize") removed that call, so nothing assigns the released BARs anymore. This breaks amdgpu completely on the SolidRun HoneyComb LX2K (NXP LX2160A, arm64, ACPI), where ACPI doesn't expose the Root Port so the GPU endpoint appears directly on a "root bus" of its segment: amdgpu 0004:01:00.0: BAR 0 [mem 0xa400000000-0xa40fffffff 64bit pref]: releasing amdgpu 0004:01:00.0: BAR 2 [mem 0xa410000000-0xa4101fffff 64bit pref]: releasing amdgpu 0004:01:00.0: sw_init of IP block failed -19 amdgpu 0004:01:00.0: amdgpu_device_ip_init failed amdgpu 0004:01:00.0: Fatal error during GPU init No error is logged because the resize path reports success; amdgpu then finds BAR 0 IORESOURCE_UNSET and bails out with -ENODEV. When there is no upstream bridge, call pci_bus_assign_resources() on the root bus to place the BARs released above, using the same alignment-sorted algorithm as normal enumeration instead of a manual per-BAR loop. This also walks the rest of the hierarchy under the root bus, as pci_assign_unassigned_bus_resources() used to for amdgpu before commit db92e3fef53e ("drm/amdgpu: Remove driver side BAR release before resize") removed that call -- the core-side fix that commit asked for ("such a problem should be fixed inside pci_resize_resource() instead"). pci_bus_assign_resources() returns void, so failure is detected by checking whether the released BARs are still assigned afterward; if not, roll back as in the bridged case. This is stricter than the bridged path -- it fails on any unplaced resource, not just required ones -- since a root bus typically has one shared window, and failing loudly seemed better than leaving something silently unassigned. The root bus path also had a locking bug that any fix here necessarily touches: the old "goto out" jumped to up_read(&pci_bus_sem) without a matching down_read() (as does the "goto restore" taken when pci_dev_res_add_to_list() fails in the release loop). Take pci_bus_sem before the BAR release loop so every path through the function holds it exactly once. Fixes: 337b1b566db0 ("PCI: Fix restoring BARs on BAR resize rollback path") Link: https://bugs.launchpad.net/ubuntu/+source/linux-hwe-7.0/+bug/2159596 Suggested-by: Ilpo Järvinen Assisted-by: Claude:claude-fable-5 checkpatch Assisted-by: Claude:claude-sonnet-5 Signed-off-by: Liz Fong-Jones [bhelgaas: commit log] Signed-off-by: Bjorn Helgaas Reviewed-by: Ilpo Järvinen Cc: stable@vger.kernel.org Link: https://patch.msgid.link/20260918035633.566823-1-lizf@honeycomb.io --- drivers/pci/setup-bus.c | 23 +++++++++++++++++------ 1 file changed, 17 insertions(+), 6 deletions(-) diff --git a/drivers/pci/setup-bus.c b/drivers/pci/setup-bus.c index e8c94aa1d3c1..ed16ef7c26fa 100644 --- a/drivers/pci/setup-bus.c +++ b/drivers/pci/setup-bus.c @@ -2380,6 +2380,7 @@ int pci_do_resource_release_and_resize(struct pci_dev *pdev, int resno, int size struct resource *res = pci_resource_n(pdev, resno); struct pci_dev_resource *dev_res; struct pci_bus *bus = pdev->bus; + struct pci_dev *bridge = pci_upstream_bridge(pdev); struct resource *b_win, *r; LIST_HEAD(saved); unsigned int i; @@ -2397,6 +2398,8 @@ int pci_do_resource_release_and_resize(struct pci_dev *pdev, int resno, int size if (ret) return ret; + down_read(&pci_bus_sem); + pci_dev_for_each_resource(pdev, r, i) { if (i >= PCI_BRIDGE_RESOURCES) break; @@ -2415,13 +2418,21 @@ int pci_do_resource_release_and_resize(struct pci_dev *pdev, int resno, int size pci_resize_resource_set_size(pdev, resno, size); - if (!bus->self) - goto out; + if (bridge) { + ret = pbus_reassign_bridge_resources(bus, res, &saved); + if (ret) + goto restore; + } else { + /* No bridge window to adjust; let the core reassign the bus. */ + pci_bus_assign_resources(bus); - down_read(&pci_bus_sem); - ret = pbus_reassign_bridge_resources(bus, res, &saved); - if (ret) - goto restore; + list_for_each_entry(dev_res, &saved, list) { + if (!resource_assigned(dev_res->res)) { + ret = -ENOSPC; + goto restore; + } + } + } out: up_read(&pci_bus_sem); From 8805840aad73df7146778be243a196d48b4f6430 Mon Sep 17 00:00:00 2001 From: Angel J Date: Fri, 18 Sep 2026 14:55:40 -0500 Subject: [PATCH 118/365] PCI: of_property: Omit bus properties without a subordinate bus A bridge (a device with a Type 1 header) may not have a secondary bus allocated (pdev->subordinate), e.g., if there are no available bus numbers or the bridge secondary/subordinate bus numbers are not writable. The dynamic OF helpers of_pci_prop_bus_range() and of_pci_prop_intr_map() dereference pdev->subordinate without checking it. When CONFIG_PCI_DYNAMIC_OF_NODES is enabled, this can cause a NULL pointer dereference and early boot hang. Generate 'bus-range' and 'interrupt-map' properties only when a subordinate bus exists. Keep the node and its remaining properties for bridges without one. The problem was latent since 407d1a51921e ("PCI: Create device tree node for bridge"), but wasn't reachable until 1f340724419e ("PCI: of: Create device tree PCI host bridge node"), which appeared in v6.15. Before 1f340724419e, of_pci_make_dev_node() returned early because the parent OF node was missing. Fixes: 407d1a51921e ("PCI: Create device tree node for bridge") Signed-off-by: Angel J [bhelgaas: move pdev->subordinate test to callees, commit log] Signed-off-by: Bjorn Helgaas Cc: stable@vger.kernel.org # v6.6+ Link: https://patch.msgid.link/20260918195540.GA1187209@bhelgaas --- drivers/pci/of_property.c | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) diff --git a/drivers/pci/of_property.c b/drivers/pci/of_property.c index 75a358f73e69..1500740cc55d 100644 --- a/drivers/pci/of_property.c +++ b/drivers/pci/of_property.c @@ -95,9 +95,13 @@ static int of_pci_prop_bus_range(struct pci_dev *pdev, struct of_changeset *ocs, struct device_node *np) { - u32 bus_range[] = { pdev->subordinate->busn_res.start, - pdev->subordinate->busn_res.end }; + u32 bus_range[2]; + if (!pdev->subordinate) + return 0; + + bus_range[0] = pdev->subordinate->busn_res.start; + bus_range[1] = pdev->subordinate->busn_res.end; return of_changeset_add_prop_u32_array(ocs, np, "bus-range", bus_range, ARRAY_SIZE(bus_range)); } @@ -220,6 +224,9 @@ static int of_pci_prop_intr_map(struct pci_dev *pdev, struct of_changeset *ocs, int ret; u8 pin; + if (!pdev->subordinate) + return 0; + pnode = pci_device_to_OF_node(pdev->bus->self); if (!pnode) pnode = pci_bus_to_OF_node(pdev->bus); From 3bd46666cfe51e2bd333fb46a0234694ca594230 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Fri, 18 Sep 2026 14:28:37 -1000 Subject: [PATCH 119/365] sched_ext: Pass the initial cmask to cid-form ops.enable() The cid-form API has an obvious hole. A task's cid mask is only visible through ops.set_cmask(), which fires on affinity changes and class switches but not when a task enters a scheduler through fork, sub-sched enable or re-home, and there is no p->cpus_ptr equivalent to fall back on. Schedulers work around it by seeding the mask in ops.init_task() from p->cpus_ptr cid by cid, which is subtly wrong: on sub-sched enable and re-home, an affinity change between init_task() and enable() is delivered to the sched the task is still on, and nothing corrects the new sched's copy afterwards. Fix it by adding struct scx_enable_args to cid-form ops.enable() carrying the task's cmask, built in the per-cpu scratch under the rq lock as the task enters the scheduler, and calling set_cmask() with the same mask right after enable(), ahead of set_weight(). A scheduler can then track affinity in set_cmask() alone, and scx_qmap drops its init_task() seed. set_cmask() no longer fires for a cid-form task before it is enabled, and the class-switch republish in switching_to_scx() is limited to the cpu form. This changes the cid-form ops.enable() signature, which is fine as the cid-form API is still considered unreleased. An args struct rather than a bare cmask argument leaves room for more initial state without another signature change, and the cmask travels as a plain arena address because BTF can't type arena struct members yet. v2: The cmask travels as a u64 arena address, cmask_arena_addr, instead of a kernel-typed pointer, with the typing limitation and the planned typed alias documented (Sashiko review). v3: The initial set_cmask() is delivered before set_weight() so that the mask is in place when weight-dependent state is derived (Andrea Righi). Selftest added. Signed-off-by: Tejun Heo Reviewed-by: Andrea Righi --- kernel/sched/ext/ext.c | 92 +++++++++++++++++++++++----------- kernel/sched/ext/internal.h | 56 ++++++++++++++++++--- tools/sched_ext/scx_qmap.bpf.c | 3 -- 3 files changed, 110 insertions(+), 41 deletions(-) diff --git a/kernel/sched/ext/ext.c b/kernel/sched/ext/ext.c index f568fd9973f6..ad391a8cbd05 100644 --- a/kernel/sched/ext/ext.c +++ b/kernel/sched/ext/ext.c @@ -447,37 +447,45 @@ static void switch_rq_lock(struct rq *from, struct rq *to) DEFINE_STATIC_KEY_FALSE(__scx_is_cid_type); /** - * scx_call_op_set_cpumask - invoke ops.set_cpumask / ops_cid.set_cmask for @task + * scx_fill_cmask_scratch - Build this cpu's arena cmask from @cpumask + * @sch: scx_sched whose scratch to fill + * @cpumask: cpus to translate into cids + * + * The scratch lives in BPF-writable arena memory and its header can't be + * trusted, so it is rewritten from kernel geometry rather than read. Caller + * must hold an rq lock so this cpu is the sole kernel writer for as long as the + * returned address is in use. + */ +static struct scx_cmask *scx_fill_cmask_scratch(struct scx_sched *sch, + const struct cpumask *cpumask) +{ + struct scx_cmask *kern_va = *this_cpu_ptr(sch->set_cmask_scratch); + struct scx_cmask_ref ref; + + scx_cmask_ref_init_kern(sch, kern_va, 0, num_possible_cpus(), &ref); + scx_cmask_ref_from_cpumask(&ref, cpumask); + return kern_va; +} + +/** + * scx_call_op_set_cpumask - Invoke the set_cpumask or set_cmask op for @task * @sch: scx_sched being invoked * @rq: rq to update as the currently-locked rq, or NULL * @task: task whose affinity is changing * @cpumask: new cpumask * - * For cid-form schedulers, translate @cpumask to a cmask via the per-cpu - * scratch in cid.c and dispatch through the ops_cid union view. Caller - * must hold @rq's rq lock so this_cpu_ptr is stable across the call. + * For cid-form schedulers, translate @cpumask to a cmask in the per-cpu scratch + * and dispatch through the ops_cid union view. Caller must hold @rq's rq lock. */ static inline void scx_call_op_set_cpumask(struct scx_sched *sch, struct rq *rq, struct task_struct *task, const struct cpumask *cpumask) { - if (scx_is_cid_type()) { - struct scx_cmask *kern_va = *this_cpu_ptr(sch->set_cmask_scratch); - struct scx_cmask_ref ref; - - /* - * Build the per-cpu arena cmask from kernel geometry via @ref, - * never reading its BPF-writable header. set_cmask()'s __arena - * argument takes the kernel address and the struct_ops - * trampoline rebases it into BPF's arena pointer form. The rq - * lock makes this cpu the sole kernel writer. - */ - scx_cmask_ref_init_kern(sch, kern_va, 0, num_possible_cpus(), &ref); - scx_cmask_ref_from_cpumask(&ref, cpumask); - SCX_CALL_CID_OP_TASK(sch, set_cmask, rq, task, kern_va); - } else { + if (scx_is_cid_type()) + SCX_CALL_CID_OP_TASK(sch, set_cmask, rq, task, + scx_fill_cmask_scratch(sch, cpumask)); + else SCX_CALL_OP_TASK(sch, set_cpumask, rq, task, cpumask); - } } enum scx_dsq_iter_flags { @@ -3634,8 +3642,12 @@ static void set_cpus_allowed_scx(struct task_struct *p, * * Fine-grained memory write control is enforced by BPF making the const * designation pointless. Cast it away when calling the operation. + * + * The cid form receives the initial mask when the task is enabled and + * hears about changes only afterwards, see struct scx_enable_args. */ - if (SCX_HAS_OP(sch, set_cpumask)) + if (SCX_HAS_OP(sch, set_cpumask) && + (!scx_is_cid_type() || scx_get_task_state(p) == SCX_TASK_ENABLED)) scx_call_op_set_cpumask(sch, task_rq(p), p, (struct cpumask *)p->cpus_ptr); } @@ -3944,8 +3956,27 @@ static void __scx_enable_task(struct scx_sched *sch, struct task_struct *p) p->scx.weight = sched_weight_to_cgroup(weight); - if (SCX_HAS_OP(sch, enable)) - SCX_CALL_OP_TASK(sch, enable, rq, p); + if (SCX_HAS_OP(sch, enable)) { + if (scx_is_cid_type()) { + struct scx_cmask *cmask = scx_fill_cmask_scratch(sch, p->cpus_ptr); + struct scx_enable_args args = { + .cmask_arena_addr = scx_kaddr_to_arena(sch, cmask), + }; + + SCX_CALL_CID_OP_TASK(sch, enable, rq, p, &args); + } else { + SCX_CALL_OP_TASK(sch, enable, rq, p); + } + } + + /* + * The initial mask also goes out through set_cmask() so a scheduler can + * track affinity there alone, and before set_weight() so that the mask + * is in place when weight-dependent state is derived, see struct + * scx_enable_args. + */ + if (scx_is_cid_type() && SCX_HAS_OP(sch, set_cpumask)) + scx_call_op_set_cpumask(sch, rq, p, p->cpus_ptr); if (SCX_HAS_OP(sch, set_weight)) SCX_CALL_OP_TASK(sch, set_weight, rq, p, p->scx.weight); @@ -4288,9 +4319,10 @@ static void switching_to_scx(struct rq *rq, struct task_struct *p) /* * set_cpus_allowed_scx() is not called while @p is associated with a - * different scheduler class. Keep the BPF scheduler up-to-date. + * different scheduler class. Keep the BPF scheduler up-to-date. The cid + * form gets its mask from scx_enable_task(). */ - if (SCX_HAS_OP(sch, set_cpumask)) + if (!scx_is_cid_type() && SCX_HAS_OP(sch, set_cpumask)) scx_call_op_set_cpumask(sch, rq, p, (struct cpumask *)p->cpus_ptr); } @@ -8374,10 +8406,11 @@ static struct bpf_struct_ops bpf_sched_ext_ops = { /* * cid-form cfi stubs. Stubs whose signatures match the cpu-form (param types * identical, only param names differ across structs) are reused. Some need - * fresh stubs, set_cmask due to an argument type difference and the sub-sched - * notifiers because no cpu-form stub exists to reuse. + * fresh stubs, set_cmask and enable due to argument differences and the + * sub-sched notifiers because no cpu-form stub exists to reuse. */ static void sched_ext_ops_cid__set_cmask(struct task_struct *p, const struct scx_cmask *cmask__arena) {} +static void sched_ext_ops_cid__enable(struct task_struct *p, struct scx_enable_args *args) {} static void sched_ext_ops__sub_caps_updated(const struct scx_cmask *cmask__arena, u64 caps) {} static void sched_ext_ops__sub_ecaps_updated(s32 cid, u64 before, u64 after) {} @@ -8398,7 +8431,7 @@ static struct sched_ext_ops_cid __bpf_ops_sched_ext_ops_cid = { .update_idle = sched_ext_ops__update_idle, .init_task = sched_ext_ops__init_task, .exit_task = sched_ext_ops__exit_task, - .enable = sched_ext_ops__enable, + .enable = sched_ext_ops_cid__enable, .disable = sched_ext_ops__disable, #ifdef CONFIG_EXT_GROUP_SCHED .cpuctl_init = sched_ext_ops__cgroup_init, @@ -10421,8 +10454,7 @@ __bpf_kfunc const void *scx_bpf_online_cmask(const struct bpf_prog_aux *aux) if (unlikely(!online)) return NULL; - /* BPF rebases by the low 32 bits, like __arena callback args */ - return (void *)((unsigned long)online - sch->arena_kern_base); + return (void *)scx_kaddr_to_arena(sch, online); } /** diff --git a/kernel/sched/ext/internal.h b/kernel/sched/ext/internal.h index d150de10a5c9..1df8f583b0ec 100644 --- a/kernel/sched/ext/internal.h +++ b/kernel/sched/ext/internal.h @@ -250,6 +250,31 @@ struct scx_exit_task_args { bool cancelled; }; +/** + * struct scx_enable_args - Argument container for cid-form ops.enable() + * @cmask_arena_addr: BPF arena address of the cmask of cids the task may run on + * + * @cmask_arena_addr is the task's affinity as it enters the scheduler. + * set_cmask() delivers the same mask right after enable(), before set_weight() + * and the first enqueue, then every affinity change afterwards, and is never + * called before enable(). A scheduler may therefore track affinity in + * set_cmask() alone. + * + * The kernel builds the mask in the scheduler arena from its own geometry, so + * the header is valid regardless of what the scheduler last wrote there. The + * memory is per-cpu scratch reused once the callback returns: copy the bits + * out, don't keep the address. The set_cmask() argument follows the same rules. + * + * The address is a plain value rather than a typed pointer because BTF can't + * mark a struct member as an arena pointer yet and a pointer member would reach + * the program typed as a kernel pointer. Cast it to struct scx_cmask __arena * + * before use. Once arena members can be typed, a typed alias will join this + * field in an anonymous union at the same offset. + */ +struct scx_enable_args { + u64 cmask_arena_addr; +}; + /* argument container for ops.cgroup_init() */ struct scx_cgroup_init_args { /* the weight of the cgroup [1..10000] */ @@ -1037,6 +1062,7 @@ struct sched_ext_ops { * - dispatch -> dispatch (cpu arg is now cid) * - update_idle -> update_idle (cpu arg is now cid) * - set_cpumask -> set_cmask (cmask instead of cpumask) + * - enable -> enable (takes struct scx_enable_args) * - cpu_online -> cid_online * - cpu_offline -> cid_offline * - dump_cpu -> dump_cid @@ -1070,7 +1096,7 @@ struct sched_ext_ops_cid { struct scx_init_task_args *args); void (*exit_task)(struct task_struct *p, struct scx_exit_task_args *args); - void (*enable)(struct task_struct *p); + void (*enable)(struct task_struct *p, struct scx_enable_args *args); void (*disable)(struct task_struct *p); void (*dump)(struct scx_dump_ctx *ctx); void (*dump_cid)(struct scx_dump_ctx *ctx, s32 cid, bool idle); @@ -1533,7 +1559,8 @@ struct scx_sched { * by BUILD_BUG_ON in scx_init()). The anonymous union lets the kernel * access either view of the same storage without function-pointer * casts: use .ops for cpu-form and shared fields, .ops_cid for the - * cid-renamed callbacks (set_cmask, select_cid, cid_online, ...). + * callbacks whose cid-form signature differs (set_cmask, enable, + * select_cid, cid_online, ...). */ union { struct sched_ext_ops ops; @@ -1556,9 +1583,9 @@ struct scx_sched { uintptr_t arena_kern_base; /* - * Per-CPU arena cmask used by scx_call_op_set_cpumask() to hand a cmask - * to ops_cid.set_cmask(). The kernel writes through the stored kern_va - * and passes it to the callback's __arena argument. + * Per-CPU arena cmask the kernel fills from a task's cpumask and hands + * to ops_cid.enable() and ops_cid.set_cmask(). The stored pointers are + * the kernel addresses. */ struct scx_cmask * __percpu *set_cmask_scratch; struct scx_cmask *online_cmask; @@ -1669,6 +1696,19 @@ static inline void *scx_arena_to_kaddr(struct scx_sched *sch, const void *bpf_pt return (void *)(sch->arena_kern_base + (u32)(uintptr_t)bpf_ptr); } +/** + * scx_kaddr_to_arena - Translate a kernel arena address to the BPF form + * @sch: scheduler whose arena hosts @kaddr + * @kaddr: kernel address inside @sch's arena + * + * __arena callback arguments need no translation. Addresses handed to BPF any + * other way, such as struct fields and kfunc return values, go through this. + */ +static inline uintptr_t scx_kaddr_to_arena(struct scx_sched *sch, const void *kaddr) +{ + return (uintptr_t)kaddr - sch->arena_kern_base; +} + enum scx_wake_flags { /* expose select WF_* flags as enums */ SCX_WAKE_FORK = WF_FORK, @@ -2302,9 +2342,9 @@ do { \ } while (0) /* - * Dispatch a task op through the cid-form ops_cid table. Only set_cmask() needs - * this: it takes an arena cmask address instead of a cpumask, so it cannot be - * invoked via its cpu-form set_cpumask() slot. + * Dispatch a task op through the cid-form ops_cid table, for the ops whose + * cid-form signature differs from the cpu-form slot: set_cmask() takes an arena + * cmask instead of a cpumask and enable() takes scx_enable_args. */ #define SCX_CALL_CID_OP_TASK(sch, op, locked_rq, task, args...) \ __SCX_CALL_OP_TASK(sch, ops_cid, op, locked_rq, task, ##args) diff --git a/tools/sched_ext/scx_qmap.bpf.c b/tools/sched_ext/scx_qmap.bpf.c index 67b7c01cae55..2f3653199842 100644 --- a/tools/sched_ext/scx_qmap.bpf.c +++ b/tools/sched_ext/scx_qmap.bpf.c @@ -961,9 +961,6 @@ s32 BPF_STRUCT_OPS_SLEEPABLE(qmap_init_task, struct task_struct *p, taskc->highpri = false; taskc->core_sched_seq = 0; cmask_init(&taskc->cpus_allowed, 0, scx_bpf_nr_cids()); - bpf_rcu_read_lock(); - cmask_from_cpumask(&taskc->cpus_allowed, p->cpus_ptr); - bpf_rcu_read_unlock(); v = bpf_task_storage_get(&task_ctx_stor, p, NULL, BPF_LOCAL_STORAGE_GET_F_CREATE); From d781d1b78acf547bafeb1592e8ba4020dce515c6 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Sat, 19 Sep 2026 04:08:20 -1000 Subject: [PATCH 120/365] selftests/sched_ext: Check the cmask cid-form ops.enable() receives cid-form ops.enable() now hands the task's cmask to the scheduler and set_cmask() repeats it right after. Add a cid-form selftest that checks both against p->cpus_ptr, that they match each other, that the initial set_cmask() lands before set_weight() and before the task first becomes runnable, and that set_cmask() never precedes enable(), across class-switch enables, fork-path enables and live affinity changes. v2: Mismatch details returned through a caller-local struct instead of globals, alloc_words validated in the header check, loop bounded by nr_cids directly (Andrea Righi). Signed-off-by: Tejun Heo --- tools/testing/selftests/sched_ext/Makefile | 1 + .../selftests/sched_ext/enable_cmask.bpf.c | 217 ++++++++++++++++++ .../selftests/sched_ext/enable_cmask.c | 138 +++++++++++ 3 files changed, 356 insertions(+) create mode 100644 tools/testing/selftests/sched_ext/enable_cmask.bpf.c create mode 100644 tools/testing/selftests/sched_ext/enable_cmask.c diff --git a/tools/testing/selftests/sched_ext/Makefile b/tools/testing/selftests/sched_ext/Makefile index 49897727f535..4e06d0baaeec 100644 --- a/tools/testing/selftests/sched_ext/Makefile +++ b/tools/testing/selftests/sched_ext/Makefile @@ -169,6 +169,7 @@ auto-test-targets := \ ddsp_bogus_dsq_fail \ ddsp_vtimelocal_fail \ dsp_local_on \ + enable_cmask \ enq_select_cpu \ exit \ hotplug \ diff --git a/tools/testing/selftests/sched_ext/enable_cmask.bpf.c b/tools/testing/selftests/sched_ext/enable_cmask.bpf.c new file mode 100644 index 000000000000..0068f3c7ab3c --- /dev/null +++ b/tools/testing/selftests/sched_ext/enable_cmask.bpf.c @@ -0,0 +1,217 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * A cid-form scheduler checking the cmask cid-form ops.enable() receives: the + * header, every cid bit against p->cpus_ptr, and that set_cmask() follows with + * the same mask before set_weight() and before the task first becomes runnable, + * and never runs before enable(). + * + * Copyright (c) 2026 Tejun Heo + */ +#include + +char _license[] SEC("license") = "GPL"; + +struct { + __uint(type, BPF_MAP_TYPE_ARENA); + __uint(map_flags, BPF_F_MMAPABLE); + __uint(max_entries, 1 << 16); +} arena SEC(".maps"); + +struct task_ctx { + u64 enable_fp; /* fingerprint of the mask enable() received */ + bool enabled; + bool pending; /* enable() ran, the initial set_cmask() hasn't */ +}; + +struct { + __uint(type, BPF_MAP_TYPE_TASK_STORAGE); + __uint(map_flags, BPF_F_NO_PREALLOC); + __type(key, int); + __type(value, struct task_ctx); +} task_ctx_stor SEC(".maps"); + +/* details of a cid bit mismatch, filled by check_mask() */ +struct mask_mismatch { + s32 cid; + bool want; + bool got; +}; + +u64 nr_enable, nr_initial_set_cmask, nr_set_cmask, nr_set_weight; + +UEI_DEFINE(uei); + +static struct task_ctx *lookup_task_ctx(struct task_struct *p) +{ + struct task_ctx *tctx; + + tctx = bpf_task_storage_get(&task_ctx_stor, p, 0, 0); + if (!tctx) + scx_bpf_error("task_ctx lookup failed for %s[%d]", p->comm, p->pid); + return tctx; +} + +/* + * Verify @m's header and every cid bit against @p's cpumask and fingerprint the + * bits into @fp. Return 0 on success, -EINVAL on a bad header, -ENOENT on a cid + * without a cpu and -EIO on a bit mismatch with the details in @mm. + */ +static int check_mask(struct task_struct *p, const struct scx_cmask __arena *m, u64 *fp, + struct mask_mismatch *mm) +{ + u32 nr_cids = scx_bpf_nr_cids(); + u64 h = 0; + s32 cid; + + if (m->base || m->nr_cids != nr_cids || m->alloc_words != CMASK_NR_WORDS(nr_cids)) + return -EINVAL; + + bpf_for(cid, 0, nr_cids) { + bool want, got; + s32 cpu; + + cpu = scx_bpf_cid_to_cpu(cid); + if (cpu < 0) + return -ENOENT; + want = bpf_cpumask_test_cpu(cpu, p->cpus_ptr); + got = cmask_test(cid, m); + if (want != got) { + mm->cid = cid; + mm->want = want; + mm->got = got; + return -EIO; + } + h = h * 31 + got; + } + + *fp = h; + return 0; +} + +s32 BPF_STRUCT_OPS_SLEEPABLE(enable_cmask_init_task, struct task_struct *p, + struct scx_init_task_args *args) +{ + if (!bpf_task_storage_get(&task_ctx_stor, p, 0, BPF_LOCAL_STORAGE_GET_F_CREATE)) + return -ENOMEM; + return 0; +} + +void BPF_STRUCT_OPS(enable_cmask_enable, struct task_struct *p, struct scx_enable_args *args) +{ + struct scx_cmask __arena *m = (struct scx_cmask __arena *)args->cmask_arena_addr; + struct mask_mismatch mm = {}; + struct task_ctx *tctx; + int ret; + + asm volatile("" :: "r"(&arena)); + tctx = lookup_task_ctx(p); + if (!tctx) + return; + + __sync_fetch_and_add(&nr_enable, 1); + if (tctx->enabled || tctx->pending) { + scx_bpf_error("enable: %s[%d] enabled twice", p->comm, p->pid); + return; + } + + ret = check_mask(p, m, &tctx->enable_fp, &mm); + if (ret) { + scx_bpf_error("enable: %s[%d] cmask check failed %d cid=%d want=%d got=%d", + p->comm, p->pid, ret, mm.cid, mm.want, mm.got); + return; + } + tctx->enabled = true; + tctx->pending = true; +} + +void BPF_STRUCT_OPS(enable_cmask_set_cmask, struct task_struct *p, + struct scx_cmask __arena *m) +{ + struct mask_mismatch mm = {}; + struct task_ctx *tctx; + u64 fp; + int ret; + + asm volatile("" :: "r"(&arena)); + tctx = lookup_task_ctx(p); + if (!tctx) + return; + + __sync_fetch_and_add(&nr_set_cmask, 1); + if (!tctx->enabled) { + scx_bpf_error("set_cmask: %s[%d] not enabled", p->comm, p->pid); + return; + } + + ret = check_mask(p, m, &fp, &mm); + if (ret) { + scx_bpf_error("set_cmask: %s[%d] cmask check failed %d cid=%d want=%d got=%d", + p->comm, p->pid, ret, mm.cid, mm.want, mm.got); + return; + } + + if (tctx->pending) { + if (fp != tctx->enable_fp) { + scx_bpf_error("set_cmask: %s[%d] initial mask differs from enable()", + p->comm, p->pid); + return; + } + tctx->pending = false; + __sync_fetch_and_add(&nr_initial_set_cmask, 1); + } +} + +void BPF_STRUCT_OPS(enable_cmask_set_weight, struct task_struct *p, u32 weight) +{ + struct task_ctx *tctx; + + tctx = lookup_task_ctx(p); + if (!tctx) + return; + + __sync_fetch_and_add(&nr_set_weight, 1); + if (tctx->pending) + scx_bpf_error("set_weight: %s[%d] before the initial set_cmask()", p->comm, + p->pid); +} + +void BPF_STRUCT_OPS(enable_cmask_runnable, struct task_struct *p, u64 enq_flags) +{ + struct task_ctx *tctx; + + tctx = lookup_task_ctx(p); + if (!tctx) + return; + + if (tctx->pending) + scx_bpf_error("runnable: %s[%d] before the initial set_cmask()", p->comm, + p->pid); +} + +void BPF_STRUCT_OPS(enable_cmask_disable, struct task_struct *p) +{ + struct task_ctx *tctx; + + tctx = lookup_task_ctx(p); + if (!tctx) + return; + + tctx->enabled = false; + tctx->pending = false; +} + +void BPF_STRUCT_OPS(enable_cmask_exit, struct scx_exit_info *ei) +{ + UEI_RECORD(uei, ei); +} + +SCX_OPS_CID_DEFINE(enable_cmask_ops, + .init_task = (void *)enable_cmask_init_task, + .enable = (void *)enable_cmask_enable, + .set_cmask = (void *)enable_cmask_set_cmask, + .set_weight = (void *)enable_cmask_set_weight, + .runnable = (void *)enable_cmask_runnable, + .disable = (void *)enable_cmask_disable, + .exit = (void *)enable_cmask_exit, + .flags = SCX_OPS_SWITCH_PARTIAL, + .name = "enable_cmask"); diff --git a/tools/testing/selftests/sched_ext/enable_cmask.c b/tools/testing/selftests/sched_ext/enable_cmask.c new file mode 100644 index 000000000000..556bcad4431d --- /dev/null +++ b/tools/testing/selftests/sched_ext/enable_cmask.c @@ -0,0 +1,138 @@ +// SPDX-License-Identifier: GPL-2.0 +/* Copyright (c) 2026 Tejun Heo */ +#define _GNU_SOURCE +#include +#include +#include +#include +#include +#include +#include +#include +#include "enable_cmask.bpf.skel.h" +#include "scx_test.h" + +#define SCHED_EXT 7 +#define NR_CHILDREN 8 +#define MAX_CPUS 1024 + +static int cpus[MAX_CPUS]; +static int nr_cpus; + +static void spin_ms(int ms) +{ + struct timespec start, now; + + clock_gettime(CLOCK_MONOTONIC, &start); + do { + clock_gettime(CLOCK_MONOTONIC, &now); + } while ((now.tv_sec - start.tv_sec) * 1000 + + (now.tv_nsec - start.tv_nsec) / 1000000 < ms); +} + +static int pin(pid_t pid, int idx) +{ + cpu_set_t set; + + CPU_ZERO(&set); + CPU_SET(cpus[idx % nr_cpus], &set); + return sched_setaffinity(pid, sizeof(set), &set); +} + +/* + * Pin, switch to SCHED_EXT for a class-switch enable, fork a grandchild that + * inherits the policy for a fork-path enable, then change affinity a few times + * while running for set_cmask() on live tasks. + */ +static int child(int idx) +{ + struct sched_param param = {}; + int i, status; + pid_t pid; + + if (pin(0, idx) || sched_setscheduler(0, SCHED_EXT, ¶m)) + return 1; + + pid = fork(); + if (pid < 0) + return 1; + if (!pid) { + spin_ms(20); + return 0; + } + + for (i = 1; i <= 4; i++) { + if (pin(0, idx + i)) + return 1; + spin_ms(5); + } + + return waitpid(pid, &status, 0) == pid && !status ? 0 : 1; +} + +static enum scx_test_status run(void *ctx) +{ + struct enable_cmask *skel; + struct bpf_link *link; + pid_t pids[NR_CHILDREN]; + cpu_set_t set; + int i, status, failed = 0; + + if (!__COMPAT_struct_has_field("scx_enable_args", "cmask_arena_addr")) + return SCX_TEST_SKIP; + + SCX_FAIL_IF(sched_getaffinity(0, sizeof(set), &set), "Failed to read affinity"); + for (i = 0; i < MAX_CPUS && i < CPU_SETSIZE; i++) + if (CPU_ISSET(i, &set)) + cpus[nr_cpus++] = i; + if (nr_cpus < 2) + return SCX_TEST_SKIP; + + skel = enable_cmask__open(); + SCX_FAIL_IF(!skel, "Failed to open"); + SCX_ENUM_INIT(skel); + SCX_FAIL_IF(enable_cmask__load(skel), "Failed to load skel"); + + link = bpf_map__attach_struct_ops(skel->maps.enable_cmask_ops); + SCX_FAIL_IF(!link, "Failed to attach struct_ops"); + + for (i = 0; i < NR_CHILDREN; i++) { + pids[i] = fork(); + SCX_FAIL_IF(pids[i] < 0, "Failed to fork"); + if (!pids[i]) + exit(child(i)); + } + + /* affinity changes from the outside race with the children's own */ + for (i = 0; i < NR_CHILDREN; i++) + pin(pids[i], i + NR_CHILDREN); + + for (i = 0; i < NR_CHILDREN; i++) { + if (waitpid(pids[i], &status, 0) != pids[i] || status) + failed++; + } + + bpf_link__destroy(link); + + SCX_EQ(skel->data->uei.kind, EXIT_KIND(SCX_EXIT_UNREG)); + SCX_EQ(failed, 0); + SCX_GE(skel->bss->nr_enable, 2 * NR_CHILDREN); + SCX_EQ(skel->bss->nr_initial_set_cmask, skel->bss->nr_enable); + SCX_GT(skel->bss->nr_set_cmask, skel->bss->nr_initial_set_cmask); + SCX_GE(skel->bss->nr_set_weight, skel->bss->nr_enable); + printf("enable=%lu initial_set_cmask=%lu set_cmask=%lu set_weight=%lu\n", + (unsigned long)skel->bss->nr_enable, + (unsigned long)skel->bss->nr_initial_set_cmask, + (unsigned long)skel->bss->nr_set_cmask, + (unsigned long)skel->bss->nr_set_weight); + + enable_cmask__destroy(skel); + return SCX_TEST_PASS; +} + +struct scx_test enable_cmask = { + .name = "enable_cmask", + .description = "Check the cid-form ops.enable() cmask and the set_cmask() after it", + .run = run, +}; +REGISTER_SCX_TEST(&enable_cmask) From 6b1bca1b1ab77f60a62087337bfe6e2f0efb9e6d Mon Sep 17 00:00:00 2001 From: Karl Mehltretter Date: Mon, 10 Aug 2026 02:56:16 +0200 Subject: [PATCH 121/365] KVM: arm64: Fix AArch32 DBGBXVR handling The consolidation of the breakpoint and watchpoint register accessors switched DBGBXVR from trap_bvr() to trap_dbg_wb_reg(). The latter selects backing storage with demux_wb_reg(), which only handles Op2 values 4 through 7. Since DBGBXVR uses Op2 1, an AArch32 guest access hits KVM_BUG_ON() and marks the VM dead. DBGBXVR aliases DBGBVR_EL1[63:32], and its AA32(HI) descriptor already selects the upper half. Map Op2 1 to dbg_bvr[] alongside Op2 4, restoring the pre-regression behavior. Fixes: 3ce9f3357e9e ("KVM: arm64: Fold DBGxVR/DBGxCR accessors into common set") Cc: stable@vger.kernel.org Assisted-by: Claude:claude-opus-5 Signed-off-by: Karl Mehltretter Reviewed-by: Marc Zyngier Link: https://patch.msgid.link/20260810005616.13227-1-kmehltretter@gmail.com Signed-off-by: Oliver Upton --- arch/arm64/kvm/sys_regs.c | 1 + 1 file changed, 1 insertion(+) diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c index 44aae52c473d..a2f4e769a428 100644 --- a/arch/arm64/kvm/sys_regs.c +++ b/arch/arm64/kvm/sys_regs.c @@ -926,6 +926,7 @@ static u64 *demux_wb_reg(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd) struct kvm_guest_debug_arch *dbg = &vcpu->arch.vcpu_debug_state; switch (rd->Op2) { + case 0b001: case 0b100: return &dbg->dbg_bvr[rd->CRm]; case 0b101: From 8f6f8a48399f82f4a83f7b9f25b9707e0062a4f9 Mon Sep 17 00:00:00 2001 From: Adarsh Das Date: Sat, 19 Sep 2026 16:27:32 +0530 Subject: [PATCH 122/365] smb: client: delete compound mids on send failure before unlock When sending a compound request fails, smb_send_rqst() kicks off a reconnect. compound_send_recv() still has those mids on pending_mid_q, but it unlocks the server without removing them first. During reconnect, cifs_abort_connection() walks pending_mid_q and runs each mid callback. With no response yet, those callbacks return credits and drop in_flight. Then compound_send_recv()'s send-error path returns the same credits again. in_flight ends up decremented twice and smb2_add_credits() WARNs. syzbot hits this during SMB2_negotiate when the socket send fails. cifs_call_async() already calls delete_mid() before unlock on send failure. Do the same for compound chains and set cancelled_mid[] so the out: path does not delete them again. Fixes: ee258d79159a ("CIFS: Move credit processing to mid callbacks for SMB3") Reported-by: syzbot+eeb58d2197d88720a228@syzkaller.appspotmail.com Closes: https://lore.kernel.org/r/6a727d22.40259c87.584f4.04ce.GAE@google.com Tested-by: syzbot+eeb58d2197d88720a228@syzkaller.appspotmail.com Assisted-by: LLM Cc: stable@vger.kernel.org Signed-off-by: Adarsh Das Signed-off-by: Paulo Alcantara --- fs/smb/client/transport.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/fs/smb/client/transport.c b/fs/smb/client/transport.c index e266859818a4..7df5b3447aea 100644 --- a/fs/smb/client/transport.c +++ b/fs/smb/client/transport.c @@ -965,6 +965,10 @@ compound_send_recv(const unsigned int xid, struct cifs_ses *ses, if (rc < 0) { revert_current_mid(server, num_rqst); server->sequence_number -= 2; + for (i = 0; i < num_rqst; i++) { + delete_mid(server, mid[i]); + cancelled_mid[i] = true; + } } cifs_server_unlock(server); From b74aad23d99b279bb34d135795f39a6d8ecdc075 Mon Sep 17 00:00:00 2001 From: Peiyang He Date: Tue, 8 Sep 2026 20:13:23 +0800 Subject: [PATCH 123/365] drm/virtio: fix memory leak of fence event on execbuffer failure virtio_gpu_execbuffer_ioctl() reserves a DRM event with drm_event_reserve_init() when VIRTGPU_EXECBUF_RING_IDX selects a ring that userspace has enabled polling for. virtio_gpu_init_submit() does this before the BO handles, the command buffer, the syncobj arrays and the in-fence are processed, so every later error path runs with the event already pending, including plain argument validation failures such as an invalid bo_handle or an in-syncobj that carries no fence. On those paths, virtio_gpu_cleanup_submit() drops the out-fence without cancelling the event. The fence is freed without ever having been emitted, taking the only driver-side pointer to the event with it. Closing the DRM file does not help. drm_events_release() unlinks pending events but deliberately leaves the freeing to the driver's later drm_send_event(), which never runs for an orphaned event, so the allocation is leaked permanently. Found when fuzzing the virtio driver with Syzkaller: BUG: memory leak unreferenced object 0xffff88802c176e80 (size 96): comm "syz.1.367", pid 10561, jiffies 4294960122 hex dump (first 32 bytes): 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ c8 6e 17 2c 80 88 ff ff 00 00 00 00 00 00 00 00 .n.,............ backtrace (crc e1973c6b): kmemleak_alloc_recursive include/linux/kmemleak.h:44 [inline] slab_post_alloc_hook mm/slub.c:4597 [inline] slab_alloc_node mm/slub.c:4917 [inline] __kmalloc_cache_noprof+0x49d/0x6f0 mm/slub.c:5485 _kmalloc_noprof include/linux/slab.h:988 [inline] _kzalloc_noprof include/linux/slab.h:1309 [inline] virtio_gpu_fence_event_create drivers/gpu/drm/virtio/virtgpu_submit.c:282 [inline] virtio_gpu_init_submit drivers/gpu/drm/virtio/virtgpu_submit.c:398 [inline] virtio_gpu_execbuffer_ioctl+0xbbf/0x1aa0 drivers/gpu/drm/virtio/virtgpu_submit.c:505 drm_ioctl_kernel+0x1f4/0x3e0 drivers/gpu/drm/drm_ioctl.c:817 drm_ioctl+0x5f4/0xc70 drivers/gpu/drm/drm_ioctl.c:914 vfs_ioctl fs/ioctl.c:51 [inline] __do_sys_ioctl fs/ioctl.c:597 [inline] __se_sys_ioctl fs/ioctl.c:583 [inline] __x64_sys_ioctl+0x18e/0x210 fs/ioctl.c:583 do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline] do_syscall_64+0x116/0x800 arch/x86/entry/syscall_64.c:94 entry_SYSCALL_64_after_hwframe+0x77/0x7f Fix by cancelling and freeing the DRM event on the execbuffer error path before dropping the fence. Clear the fence's event pointer after cancellation so it does not retain a dangling pointer. Fixes: cd7f5ca33585 ("drm/virtio: implement context init: add virtio_gpu_fence_event") Cc: stable@vger.kernel.org Signed-off-by: Peiyang He Assisted-by: Codex:gpt-5.6-luna Signed-off-by: Dmitry Osipenko Link: https://patch.msgid.link/D320EAB5680C1411+20260908121323.2405044-1-peiyang_he@smail.nju.edu.cn --- drivers/gpu/drm/virtio/virtgpu_submit.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/drivers/gpu/drm/virtio/virtgpu_submit.c b/drivers/gpu/drm/virtio/virtgpu_submit.c index 32cb1e4aa425..734b1e976a75 100644 --- a/drivers/gpu/drm/virtio/virtgpu_submit.c +++ b/drivers/gpu/drm/virtio/virtgpu_submit.c @@ -538,6 +538,10 @@ int virtio_gpu_execbuffer_ioctl(struct drm_device *dev, void *data, virtio_gpu_process_post_deps(&submit); virtio_gpu_complete_submit(&submit); cleanup: + if (ret && submit.out_fence && submit.out_fence->e) { + drm_event_cancel_free(dev, &submit.out_fence->e->base); + submit.out_fence->e = NULL; + } virtio_gpu_cleanup_submit(&submit); return ret; From 36570ef2244cc4d7563b1f0157bc0f032498638c Mon Sep 17 00:00:00 2001 From: Junrui Luo Date: Tue, 15 Sep 2026 15:39:10 +0800 Subject: [PATCH 124/365] drm/virtio: fix object leak when drm_gem_handle_create() fails virtio_gpu_gem_create() owns the reference taken by virtio_gpu_object_create(). On the drm_gem_handle_create() error path it calls drm_gem_object_release() instead of dropping that reference. drm_gem_object_release() is the inverse of drm_gem_object_init() and does not touch the reference count or call obj->funcs->free(), so it is only correct as the last step of a destructor, as in virtio_gpu_cleanup_object(). Using it here leaves the bo at refcount 1 with no remaining reference, so virtio_gpu_free_object() never runs and the shmem pages, sg table and virtio_gpu_object are leaked. Since virtio_gpu_object_create() has already set bo->created, VIRTIO_GPU_CMD_RESOURCE_UNREF is not queued either, leaking the host-side resource and the resource id. drm_gem_handle_create_tail() drops the handle reference on all of its internal error paths, so the caller only has to drop its own. Use drm_gem_object_put(), matching the success path below. Fixes: dc5698e80cf7 ("Add virtio gpu driver.") Reported-by: Yuhao Jiang Assisted-by: Claude:claude-opus-5 Signed-off-by: Junrui Luo Signed-off-by: Dmitry Osipenko Link: https://patch.msgid.link/20260915-fixes-v2-1-a0d799e4db66@outlook.com --- drivers/gpu/drm/virtio/virtgpu_gem.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/virtio/virtgpu_gem.c b/drivers/gpu/drm/virtio/virtgpu_gem.c index 66c3f6f74e9c..d2f0b8a3f172 100644 --- a/drivers/gpu/drm/virtio/virtgpu_gem.c +++ b/drivers/gpu/drm/virtio/virtgpu_gem.c @@ -45,7 +45,7 @@ static int virtio_gpu_gem_create(struct drm_file *file, ret = drm_gem_handle_create(file, &obj->base.base, &handle); if (ret) { - drm_gem_object_release(&obj->base.base); + drm_gem_object_put(&obj->base.base); return ret; } From 477bc3068fc3777b9d8ffd79e265b0dfdf2d3a6b Mon Sep 17 00:00:00 2001 From: Junrui Luo Date: Tue, 15 Sep 2026 15:39:11 +0800 Subject: [PATCH 125/365] drm/virtio: fix object leak in virtio_gpu_resource_create_ioctl() virtio_gpu_resource_create_ioctl() calls drm_gem_object_release() on the drm_gem_handle_create() error path instead of dropping the reference it owns, so obj->funcs->free() never runs and the virtio_gpu_object, its pages and sg table, the resource id and the host-side resource are leaked. Use drm_gem_object_put() instead. Fixes: 62fb7a5e1096 ("virtio-gpu: add 3d/virgl support") Assisted-by: Claude:claude-opus-5 Signed-off-by: Junrui Luo Signed-off-by: Dmitry Osipenko Link: https://patch.msgid.link/20260915-fixes-v2-2-a0d799e4db66@outlook.com --- drivers/gpu/drm/virtio/virtgpu_ioctl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/virtio/virtgpu_ioctl.c b/drivers/gpu/drm/virtio/virtgpu_ioctl.c index 3d8e4ccdb7c1..d16f07abb266 100644 --- a/drivers/gpu/drm/virtio/virtgpu_ioctl.c +++ b/drivers/gpu/drm/virtio/virtgpu_ioctl.c @@ -185,7 +185,7 @@ static int virtio_gpu_resource_create_ioctl(struct drm_device *dev, void *data, ret = drm_gem_handle_create(file, obj, &handle); if (ret) { - drm_gem_object_release(obj); + drm_gem_object_put(obj); return ret; } From 24b6d5c7641412c9ebef0d4c8b888d49a0e6b880 Mon Sep 17 00:00:00 2001 From: Junrui Luo Date: Tue, 15 Sep 2026 15:39:12 +0800 Subject: [PATCH 126/365] drm/virtio: fix object leaks in virtio_gpu_resource_create_blob_ioctl() virtio_gpu_resource_create_blob_ioctl() calls drm_gem_object_release() on both the virtio_gpu_resource_assign_uuid() and drm_gem_handle_create() error paths instead of dropping the reference it owns, so obj->funcs->free() never runs and the virtio_gpu_object, the resource id and the host-side resource are leaked. Use drm_gem_object_put() instead. Fixes: 897b4d1acaf5 ("drm/virtio: implement blob resources: resource create blob ioctl") Assisted-by: Claude:claude-opus-5 Signed-off-by: Junrui Luo Signed-off-by: Dmitry Osipenko Link: https://patch.msgid.link/20260915-fixes-v2-3-a0d799e4db66@outlook.com --- drivers/gpu/drm/virtio/virtgpu_ioctl.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/virtio/virtgpu_ioctl.c b/drivers/gpu/drm/virtio/virtgpu_ioctl.c index d16f07abb266..fcdb07a37972 100644 --- a/drivers/gpu/drm/virtio/virtgpu_ioctl.c +++ b/drivers/gpu/drm/virtio/virtgpu_ioctl.c @@ -557,14 +557,14 @@ static int virtio_gpu_resource_create_blob_ioctl(struct drm_device *dev, if (params.blob_flags & VIRTGPU_BLOB_FLAG_USE_CROSS_DEVICE) { ret = virtio_gpu_resource_assign_uuid(vgdev, bo); if (ret) { - drm_gem_object_release(obj); + drm_gem_object_put(obj); return ret; } } ret = drm_gem_handle_create(file, obj, &handle); if (ret) { - drm_gem_object_release(obj); + drm_gem_object_put(obj); return ret; } From 036d28db1818af2f9d80db771f5405da84d7732d Mon Sep 17 00:00:00 2001 From: Junrui Luo Date: Tue, 15 Sep 2026 15:39:13 +0800 Subject: [PATCH 127/365] drm/virtio: release the GEM object on virtio_gpu_vram_create() errors virtio_gpu_vram_create() frees the object with a bare kfree(vram) on both error paths after drm_gem_private_object_init() has run, and on the second one after drm_gem_create_mmap_offset() has linked obj->vma_node into the device's VMA offset manager. The freed object stays in that interval tree, so a later lookup or insertion walks freed memory, and the dma_resv and gpuva lock are never destroyed. Call drm_gem_object_release() before kfree() on both paths. Fixes: 16845c5d5409 ("drm/virtio: implement blob resources: implement vram object") Assisted-by: Claude:claude-opus-5 Signed-off-by: Junrui Luo Signed-off-by: Dmitry Osipenko Link: https://patch.msgid.link/20260915-fixes-v2-4-a0d799e4db66@outlook.com --- drivers/gpu/drm/virtio/virtgpu_vram.c | 17 +++++++++-------- 1 file changed, 9 insertions(+), 8 deletions(-) diff --git a/drivers/gpu/drm/virtio/virtgpu_vram.c b/drivers/gpu/drm/virtio/virtgpu_vram.c index 5b4a3ab81cd5..01241ce4d07c 100644 --- a/drivers/gpu/drm/virtio/virtgpu_vram.c +++ b/drivers/gpu/drm/virtio/virtgpu_vram.c @@ -215,16 +215,12 @@ int virtio_gpu_vram_create(struct virtio_gpu_device *vgdev, /* Create fake offset */ ret = drm_gem_create_mmap_offset(obj); - if (ret) { - kfree(vram); - return ret; - } + if (ret) + goto err_release_obj; ret = virtio_gpu_resource_id_get(vgdev, &vram->base.hw_res_handle); - if (ret) { - kfree(vram); - return ret; - } + if (ret) + goto err_release_obj; virtio_gpu_cmd_resource_create_blob(vgdev, &vram->base, params, NULL, 0); @@ -240,6 +236,11 @@ int virtio_gpu_vram_create(struct virtio_gpu_device *vgdev, *bo_ptr = &vram->base; return 0; + +err_release_obj: + drm_gem_object_release(obj); + kfree(vram); + return ret; } void virtio_gpu_vram_map_deferred(struct virtio_gpu_object_vram *vram) From 1e3b08de63274d0b009e99ef51cd6a9c0c6bf08c Mon Sep 17 00:00:00 2001 From: Dmitry Osipenko Date: Fri, 11 Sep 2026 17:42:03 +0300 Subject: [PATCH 128/365] Revert "drm/virtio: Allow importing prime buffers when 3D is enabled" Guest userspace may import udmabuf to vrend. Vrend doesn't support guest blobs, and thus, further 3d operations with the imported blob are failing. Typical scenario of the problem shown with mouse cursor RGBA image imported into virtio-gpu, which previously was rejected by virtio-gpu driver. Revert enabling guest blobs importing into vrend to fix the regression. Link: https://gitlab.freedesktop.org/virgl/virglrenderer/-/work_items/674 Fixes: df4dc947c46b ("drm/virtio: Allow importing prime buffers when 3D is enabled") Signed-off-by: Dmitry Osipenko Reviewed-by: Val Packett Link: https://patch.msgid.link/20260911144204.2089401-1-dmitry.osipenko@collabora.com --- drivers/gpu/drm/virtio/virtgpu_prime.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/virtio/virtgpu_prime.c b/drivers/gpu/drm/virtio/virtgpu_prime.c index 149e6bcb5878..ebf471044d06 100644 --- a/drivers/gpu/drm/virtio/virtgpu_prime.c +++ b/drivers/gpu/drm/virtio/virtgpu_prime.c @@ -349,7 +349,7 @@ struct drm_gem_object *virtgpu_gem_prime_import(struct drm_device *dev, } } - if (!vgdev->has_resource_blob) + if (!vgdev->has_resource_blob || vgdev->has_virgl_3d) return drm_gem_prime_import(dev, buf); bo = kzalloc_obj(*bo); From 846b3c64fe3e77d9db20a7e3e62dbbb637c773e1 Mon Sep 17 00:00:00 2001 From: Peiyang He Date: Wed, 9 Sep 2026 17:11:14 +0800 Subject: [PATCH 129/365] drm/virtio: fix NULL pointer dereference on fence allocation failure virtio_gpu_fence_alloc() can fail due to memory pressure and return NULL, but its caller like virtio_gpu_init_submit() never checks it. Later, virtio_gpu_init_submit() passes the NULL fence to virtio_gpu_fence_event_create(), which unconditionally dereferences it. Found when fuzzing the virtio driver with Syzkaller: Oops: general protection fault, probably for non-canonical address 0xdffffc0000000012: 0000 [#1] SMP KASAN NOPTI KASAN: null-ptr-deref in range [0x0000000000000090-0x0000000000000097] CPU: 1 UID: 0 PID: 9991 Comm: syz.0.121 Not tainted 7.2.0 #4 PREEMPT(full) Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.17.0-0-gb52ca86e094d-prebuilt.qemu.org 04/01/2014 RIP: 0010:virtio_gpu_fence_event_create drivers/gpu/drm/virtio/virtgpu_submit.c:295 [inline] RIP: 0010:virtio_gpu_init_submit drivers/gpu/drm/virtio/virtgpu_submit.c:398 [inline] RIP: 0010:virtio_gpu_execbuffer_ioctl+0xc78/0x1aa0 drivers/gpu/drm/virtio/virtgpu_submit.c:505 Code: 85 ed 0f 85 21 09 00 00 e8 05 5a c9 fb 48 8b 44 24 10 48 8d b8 90 00 00 00 48 b8 00 00 00 00 00 fc ff df 48 89 fa 48 c1 ea 03 <80> 3c 02 00 0f 85 9a 0d 00 00 48 8b 44 24 10 4c 89 b0 90 00 00 00 RSP: 0018:ffffc900039dfad0 EFLAGS: 00010216 RAX: dffffc0000000000 RBX: ffffc900039dfdd8 RCX: ffffffff85f6fd3d RDX: 0000000000000012 RSI: ffffffff85f6fd4b RDI: 0000000000000090 RBP: 0000000000000000 R0virtio_gpu_virgl_process_cmd: ctrl 0x102, error 0x1203 R10: 0000000000000000 R11: 0000000000000000 R12: ffff8880132c4000 R13: 0000000000000000 R14: ffff888073b6c700 R15: 000000000000003b FS: 00007fab480b96c0(0000) GS:ffff8880eb6e9000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007effbf5e55a8 CR3: 0000000048d19000 CR4: 0000000000350ef0 Call Trace: drm_ioctl_kernel+0x1f4/0x3e0 drivers/gpu/drm/drm_ioctl.c:817 drm_ioctl+0x5f4/0xc70 drivers/gpu/drm/drm_ioctl.c:914 vfs_ioctl fs/ioctl.c:51 [inline] __do_sys_ioctl fs/ioctl.c:597 [inline] __se_sys_ioctl fs/ioctl.c:583 [inline] __x64_sys_ioctl+0x18e/0x210 fs/ioctl.c:583 do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline] do_syscall_64+0x116/0x800 arch/x86/entry/syscall_64.c:94 entry_SYSCALL_64_after_hwframe+0x77/0x7f RIP: 0033:0x7fab471a82bd Code: ff c3 66 2e 0f 1f 84 00 00 00 00 00 90 f3 0f 1e fa 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 b0 ff ff ff f7 d8 64 89 01 48 RSP: 002b:00007fab480b9018 EFLAGS: 00000246 ORIG_RAX: 0000000000000010 RAX: ffffffffffffffda RBX: 00007fab47435fa0 RCX: 00007fab471a82bd RDX: 00002000000000c0 RSI: 00000000c0406442 RDI: 0000000000000003 RBP: 00007fab480b9080 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000001 R13: 00007fab47436038 R14: 00007fab47435fa0 R15: 00007ffe85ab0740 Modules linked in: ---[ end trace 0000000000000000 ]--- RIP: 0010:virtio_gpu_fence_event_create drivers/gpu/drm/virtio/virtgpu_submit.c:295 [inline] RIP: 0010:virtio_gpu_init_submit drivers/gpu/drm/virtio/virtgpu_submit.c:398 [inline] RIP: 0010:virtio_gpu_execbuffer_ioctl+0xc78/0x1aa0 drivers/gpu/drm/virtio/virtgpu_submit.c:505 Code: 85 ed 0f 85 21 09 00 00 e8 05 5a c9 fb 48 8b 44 24 10 48 8d b8 90 00 00 00 48 b8 00 00 00 00 00 fc ff df 48 89 fa 48 c1 ea 03 <80> 3c 02 00 0f 85 9a 0d 00 00 48 8b 44 24 10 4c 89 b0 90 00 00 00 RSP: 0018:ffffc900039dfad0 EFLAGS: 00010216 RAX: dffffc0000000000 RBX: ffffc900039dfdd8 RCX: ffffffff85f6fd3d RDX: 0000000000000012 RSI: ffffffff85f6fd4b RDI: 0000000000000090 RBP: 0000000000000000 R08: 0000000000000005 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000000 R12: ffff8880132c4000 R13: 0000000000000000 R14: ffff888073b6c700 R15: 000000000000003b FS: 00007fab480b96c0(0000) GS:ffff888098ae9000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007f24c3759000 CR3: 0000000048d19000 CR4: 0000000000350ef0 ---------------- Code disassembly (best guess): 0: 85 ed test %ebp,%ebp 2: 0f 85 21 09 00 00 jne 0x929 8: e8 05 5a c9 fb call 0xfbc95a12 d: 48 8b 44 24 10 mov 0x10(%rsp),%rax 12: 48 8d b8 90 00 00 00 lea 0x90(%rax),%rdi 19: 48 b8 00 00 00 00 00 movabs $0xdffffc0000000000,%rax 20: fc ff df 23: 48 89 fa mov %rdi,%rdx 26: 48 c1 ea 03 shr $0x3,%rdx * 2a: 80 3c 02 00 cmpb $0x0,(%rdx,%rax,1) <-- trapping instruction 2e: 0f 85 9a 0d 00 00 jne 0xdce 34: 48 8b 44 24 10 mov 0x10(%rsp),%rax 39: 4c 89 b0 90 00 00 00 mov %r14,0x90(%rax) Fix by checking virtio_gpu_fence_alloc() in virtio_gpu_init_submit() and returning -ENOMEM before any later code can dereference the NULL fence. Fixes: 70d1ace56db6 ("drm/virtio: Conditionally allocate virtio_gpu_fence") Cc: stable@vger.kernel.org Signed-off-by: Peiyang He Assisted-by: Codex:gpt-5.5 Signed-off-by: Dmitry Osipenko Link: https://patch.msgid.link/00EFE4BA92889B14+20260909091114.2622550-1-peiyang_he@smail.nju.edu.cn --- drivers/gpu/drm/virtio/virtgpu_submit.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/virtio/virtgpu_submit.c b/drivers/gpu/drm/virtio/virtgpu_submit.c index 734b1e976a75..3d35326dd904 100644 --- a/drivers/gpu/drm/virtio/virtgpu_submit.c +++ b/drivers/gpu/drm/virtio/virtgpu_submit.c @@ -389,10 +389,13 @@ static int virtio_gpu_init_submit(struct virtio_gpu_submit *submit, if ((exbuf->flags & VIRTGPU_EXECBUF_FENCE_FD_OUT) || exbuf->num_out_syncobjs || exbuf->num_bo_handles || - drm_fence_event) + drm_fence_event) { out_fence = virtio_gpu_fence_alloc(vgdev, fence_ctx, ring_idx); - else + if (!out_fence) + return -ENOMEM; + } else { out_fence = NULL; + } if (drm_fence_event) { err = virtio_gpu_fence_event_create(dev, file, out_fence, ring_idx); From 6947b78df4d25bd1e86b26870b614ea54c625b1e Mon Sep 17 00:00:00 2001 From: Shixiong Ou Date: Fri, 28 Aug 2026 17:01:23 +0800 Subject: [PATCH 130/365] drm/virtio: Add pixel blend mode property to cursor plane The cursor plane exposes a format with an alpha channel (DRM_FORMAT_ARGB8888) without a pixel blend mode property. Since commit 860e748bddcc ("drm: ensure blend mode supported if pixel format with alpha exposed") this triggers a warning during drm_mode_config_validate(): [ 0.649020] ------------[ cut here ]------------ [ 0.649040] [PLANE:36:plane-1] pixel format with alpha exposed but blend mode not setup [ 0.649081] WARNING: drivers/gpu/drm/drm_mode_config.c:872 at drm_mode_config_validate ...... [ 0.649761] Call trace: [ 0.649764] drm_mode_config_validate+0x398/0x558 [drm] (P) [ 0.649912] drm_dev_register+0x1cc/0x2a0 [drm] [ 0.650058] virtio_gpu_probe+0xd4/0x1c0 [virtio_gpu] [ 0.650088] virtio_dev_probe+0x1c8/0x310 ...... [ 0.650261] ---[ end trace 0000000000000000 ]--- Create the property with the only supported blend mode, DRM_MODE_BLEND_PREMULTI, which is also the property's default and matches what userspace had to assume before the property existed. Fixes: 860e748bddcc ("drm: ensure blend mode supported if pixel format with alpha exposed") Reported-by: Ye Liu Signed-off-by: Shixiong Ou Signed-off-by: Dmitry Osipenko Link: https://patch.msgid.link/20260828090123.784944-1-oushixiong1025@163.com --- drivers/gpu/drm/virtio/virtgpu_plane.c | 13 +++++++++++++ 1 file changed, 13 insertions(+) diff --git a/drivers/gpu/drm/virtio/virtgpu_plane.c b/drivers/gpu/drm/virtio/virtgpu_plane.c index 640815af4098..b422eba42a5f 100644 --- a/drivers/gpu/drm/virtio/virtgpu_plane.c +++ b/drivers/gpu/drm/virtio/virtgpu_plane.c @@ -589,6 +589,7 @@ struct drm_plane *virtio_gpu_plane_init(struct virtio_gpu_device *vgdev, struct drm_plane *plane; const uint32_t *formats; int nformats; + int ret; if (type == DRM_PLANE_TYPE_CURSOR) { formats = virtio_gpu_cursor_formats; @@ -614,5 +615,17 @@ struct drm_plane *virtio_gpu_plane_init(struct virtio_gpu_device *vgdev, drm_plane_create_blend_mode_property(plane, BIT(DRM_MODE_BLEND_PREMULTI)); + if (type == DRM_PLANE_TYPE_CURSOR) { + /* + * The cursor plane exposes a format with an alpha channel, + * which requires a blend mode property. The host blends + * premultiplied alpha, matching the property's default. + */ + ret = drm_plane_create_blend_mode_property(plane, + BIT(DRM_MODE_BLEND_PREMULTI)); + if (ret) + return ERR_PTR(ret); + } + return plane; } From 598c1c3e895590f845e04455d5580ea28ffde666 Mon Sep 17 00:00:00 2001 From: Benjamin Leggett Date: Fri, 14 Aug 2026 18:20:11 -0400 Subject: [PATCH 131/365] drm/virtio: sync shmem backing on guest-bound transfers virtio_gpu_cmd_transfer_to_host_{2d,3d}() sync the shmem backing for the device before the transfer, but nothing syncs for the CPU when a transfer runs the other way. That breaks two ways. Where the DMA layer bounces, the device writes into the bounce buffer while the guest keeps reading the original pages. Where DMA is not coherent, the device writes memory while the CPU keeps stale cache lines, because nothing reaches arch_sync_dma_for_cpu(). Either way DRM_IOCTL_VIRTGPU_TRANSFER_FROM_HOST hands back stale data. Sashiko originally found this in https://lore.kernel.org/dri-devel/20260806231002.27B4D1F000E9@smtp.kernel.org but the suggestion there to fix this with dma_sync_sgtable_for_cpu() isn't a sufficient fix, for two reasons. - The transfer is asynchronous. virtio_gpu_cmd_transfer_from_host_3d() only queues the command, so a sync there would run before the device had written anything. It belongs on completion, and ahead of any fence signalling. A waiter woken by the fence would otherwise race the sync and read the backing pages regardless. It needs its own pass over the reclaim list rather than a step inside the existing one, because virtio_gpu_fence_event_process() also signals every earlier fence in the same context, so any entry in that loop may signal an earlier entry's fence. - The transfer is also partial, carrying an offset, a level and a box. Where the mapping bounces, a sync for the CPU copies the whole mapping back, so unless the mapping is primed first the regions the device did not write come back holding whatever the bounce buffer contained, discarding data the guest owned. So the fix: Prime the mapping before queueing, tag the vbuffer, and sync for the CPU on completion before the fence is signalled. A second transfer must not snapshot the mapping while an earlier one is still in flight, or the snapshot would predate whatever the CPU wrote once the earlier fence signalled and the later sync would discard it. To mitigate this, wait for outstanding fences under the reservation before priming. Neither sync copies anything unless the mapping genuinely bounces: swiotlb_sync_single_for_cpu() and its Xen counterpart look the address up in the bounce pool and return early when it is absent. On a platform with non-coherent DMA they still perform the necessary cache maintenance. The range cannot be narrowed to the box, since for a non-blob resource virtio_gpu_transfer_from_host_ioctl() rejects a caller-supplied stride and layer_stride, leaving the layout to the host and the guest with no way to work out which bytes the device writes. A host3d guest blob does carry both, so its extent could be bounded, but the sync is left whole there too rather than special-cased: priming makes the untouched regions round-trip unchanged either way. Behaviour changes worth noting: - TRANSFER_FROM_HOST can now block, where before it returned as soon as the command was queued. Repeated readbacks of one resource serialise, and a readback can wait behind an earlier queued command that touched it, since virtio_gpu_array_add_fence() tags uploads, execbufs and plane flushes alike with DMA_RESV_USAGE_WRITE. -ERESTARTSYS was already possible here via dma_resv_lock_interruptible(). - A CPU write racing an in-flight transfer to the same resource is now lost, where before it survived and the transfer was lost instead. Priming captures the pages as of queueing, so a write landing before completion is overwritten by the sync. - TRANSFER_TO_HOST can also block now, but only while a guest-bound transfer on the same resource is outstanding, which happens only for callers that issue both without waiting. - Where a batch of completions contains a guest-bound transfer, the sync pass delays fence signalling for the whole batch. Only bounced pages are copied and the swiotlb pool bounds it. A batch with no such transfer is unaffected. Tested under QEMU on x86 with swiotlb=force and virtio-vga-gl iommu_platform=on, which forces both preconditions required to hit the original bug. Fixes: a3b815f09bb8 ("drm/virtio: add iommu support.") Reported-by: Sashiko AI review Closes: https://lore.kernel.org/dri-devel/20260806231002.27B4D1F000E9@smtp.kernel.org/ Signed-off-by: Benjamin Leggett Signed-off-by: Dmitry Osipenko Link: https://patch.msgid.link/20260814-virtgpu-from-host-sync-v4-1-64dd736b1779@edera.io --- drivers/gpu/drm/virtio/virtgpu_drv.h | 5 +++ drivers/gpu/drm/virtio/virtgpu_ioctl.c | 43 ++++++++++++++++++++++++ drivers/gpu/drm/virtio/virtgpu_vq.c | 46 ++++++++++++++++++++++++++ 3 files changed, 94 insertions(+) diff --git a/drivers/gpu/drm/virtio/virtgpu_drv.h b/drivers/gpu/drm/virtio/virtgpu_drv.h index 9df4c7117341..f3bbbe4468bf 100644 --- a/drivers/gpu/drm/virtio/virtgpu_drv.h +++ b/drivers/gpu/drm/virtio/virtgpu_drv.h @@ -114,6 +114,8 @@ struct virtio_gpu_object { bool dumb; bool created; bool attached; + /* a guest-bound transfer is queued and its mapping not yet synced */ + bool from_host_pending; bool host3d_blob, guest_blob; uint32_t blob_mem, blob_flags; @@ -196,6 +198,9 @@ struct virtio_gpu_vbuffer { struct list_head list; uint32_t seqno; + + /* guest-bound transfer whose shmem backing needs a CPU sync */ + bool sync_for_cpu; }; struct virtio_gpu_output { diff --git a/drivers/gpu/drm/virtio/virtgpu_ioctl.c b/drivers/gpu/drm/virtio/virtgpu_ioctl.c index fcdb07a37972..81e70a12b356 100644 --- a/drivers/gpu/drm/virtio/virtgpu_ioctl.c +++ b/drivers/gpu/drm/virtio/virtgpu_ioctl.c @@ -261,6 +261,27 @@ static int virtio_gpu_transfer_from_host_ioctl(struct drm_device *dev, if (ret != 0) goto err_put_free; + if (virtio_gpu_is_shmem(bo) && virtio_gpu_use_dma_api(vgdev->vdev)) { + /* + * The sync on completion restores the whole mapping, so an + * earlier transfer has to be done before this one snapshots it. + * Otherwise the snapshot predates anything the CPU wrote once + * that transfer's fence signalled, and the later sync would + * discard it. Nothing can add a fence behind our back here, + * since doing so takes the reservation we already hold. + * This writes the pages, so it waits as a writer does. READ + * usage covers existing readers. + */ + long wait = dma_resv_wait_timeout(objs->objs[0]->resv, + DMA_RESV_USAGE_READ, true, + MAX_SCHEDULE_TIMEOUT); + + if (wait < 0) { + ret = wait; + goto err_unlock; + } + } + fence = virtio_gpu_fence_alloc(vgdev, vgdev->fence_drv.context, 0); if (!fence) { ret = -ENOMEM; @@ -320,6 +341,28 @@ static int virtio_gpu_transfer_to_host_ioctl(struct drm_device *dev, void *data, if (ret != 0) goto err_put_free; + /* + * A transfer the other way may have queued without yet syncing + * its mapping. Pushing the guest pages into it now would + * discard what the device wrote there, so wait for that sync: + * it runs before the fence it belongs to is signalled. The + * flag is only set under this reservation, so it cannot appear + * behind our back, and the acquire pairs with the release in + * that sync, so finding it clear means the pages it wrote are + * visible here too. + */ + if (smp_load_acquire(&bo->from_host_pending)) { + long wait = dma_resv_wait_timeout(objs->objs[0]->resv, + DMA_RESV_USAGE_WRITE, + true, + MAX_SCHEDULE_TIMEOUT); + + if (wait < 0) { + ret = wait; + goto err_unlock; + } + } + ret = -ENOMEM; fence = virtio_gpu_fence_alloc(vgdev, vgdev->fence_drv.context, 0); diff --git a/drivers/gpu/drm/virtio/virtgpu_vq.c b/drivers/gpu/drm/virtio/virtgpu_vq.c index c02c03c10d92..d99fb9e326e8 100644 --- a/drivers/gpu/drm/virtio/virtgpu_vq.c +++ b/drivers/gpu/drm/virtio/virtgpu_vq.c @@ -256,6 +256,33 @@ void virtio_gpu_dequeue_ctrl_func(struct work_struct *work) } while (!virtqueue_enable_cb(vgdev->ctrlq.vq)); spin_unlock(&vgdev->ctrlq.qlock); + /* + * Sync guest-bound transfers before signalling anything, so that a + * waiter cannot read the backing pages while what the device wrote is + * still in a bounce buffer. This cannot be folded into the loop below: + * virtio_gpu_fence_event_process() also signals every earlier fence in + * the same context, so any entry there may signal this entry's fence. + */ + list_for_each_entry(entry, &reclaim_list, list) { + if (entry->sync_for_cpu) { + struct virtio_gpu_object *bo = + gem_to_virtio_gpu_obj(entry->objs->objs[0]); + + dma_sync_sgtable_for_cpu(vgdev->vdev->dev.parent, + bo->base.sgt, DMA_FROM_DEVICE); + /* + * Release, so a transfer the other way that skips its + * wait on the strength of this cannot go on to read + * the backing pages before the sync above is visible. + * Nothing orders the two otherwise: where the mapping + * bounces on a coherent device the sync is a plain + * copy, and dma_direct_sync_sg_for_cpu() emits its + * barrier only for the non-coherent case. + */ + smp_store_release(&bo->from_host_pending, false); + } + } + list_for_each_entry(entry, &reclaim_list, list) { resp = (struct virtio_gpu_ctrl_hdr *)entry->resp_buf; @@ -1278,12 +1305,31 @@ void virtio_gpu_cmd_transfer_from_host_3d(struct virtio_gpu_device *vgdev, struct virtio_gpu_object *bo = gem_to_virtio_gpu_obj(objs->objs[0]); struct virtio_gpu_transfer_host_3d *cmd_p; struct virtio_gpu_vbuffer *vbuf; + bool use_dma_api = virtio_gpu_use_dma_api(vgdev->vdev); cmd_p = virtio_gpu_alloc_cmd(vgdev, &vbuf, sizeof(*cmd_p)); memset(cmd_p, 0, sizeof(*cmd_p)); vbuf->objs = objs; + if (virtio_gpu_is_shmem(bo) && use_dma_api) { + /* + * The device writes only the requested box, so prime the + * mapping with the current contents: otherwise the sync on + * completion would hand back whatever a bounce buffer held for + * the regions the device does not touch. + */ + dma_sync_sgtable_for_device(vgdev->vdev->dev.parent, + bo->base.sgt, DMA_TO_DEVICE); + vbuf->sync_for_cpu = true; + /* + * Set under the reservation the caller holds, so a transfer + * the other way cannot miss it and push the guest pages into + * the mapping while the device still owns it. + */ + WRITE_ONCE(bo->from_host_pending, true); + } + cmd_p->hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_TRANSFER_FROM_HOST_3D); cmd_p->hdr.ctx_id = cpu_to_le32(ctx_id); cmd_p->resource_id = cpu_to_le32(bo->hw_res_handle); From ada667890773e033d2f40dc94176e3beb930b516 Mon Sep 17 00:00:00 2001 From: Li Youhong Date: Fri, 4 Sep 2026 09:49:58 +0800 Subject: [PATCH 132/365] drm/bridge: samsung-dsim: fix TE GPIO lifetime for host attach When the Exynos DSI driver was generalized into samsung-dsim, the TE GPIO acquisition was switched from gpiod_get_optional() to devm_gpiod_get_optional() while keeping the matching gpiod_put() calls. That combination is wrong for a managed descriptor. However, dropping the puts and keeping the managed get is also wrong: samsung_dsim_register_te_irq() runs from the DSI host attach callback, and host detach/reattach can happen without destroying the device that owns the managed action. A second attach would then request the GPIO again without having released it. Switch back to a non-managed gpiod_get_optional() and keep the explicit gpiod_put() on the request_irq() error path and in samsung_dsim_unregister_te_irq(). Fixes: e7447128ca4a ("drm: bridge: Generalize Exynos-DSI driver into a Samsung DSIM bridge") Suggested-by: Luca Ceresoli Signed-off-by: Li Youhong Reviewed-by: Luca Ceresoli Tested-by: Luca Ceresoli Link: https://patch.msgid.link/20260904014958.1572918-1-dayou5941@163.com [Luca: remove unnecessary comment] Signed-off-by: Luca Ceresoli --- drivers/gpu/drm/bridge/samsung-dsim.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/bridge/samsung-dsim.c b/drivers/gpu/drm/bridge/samsung-dsim.c index e2fc69fc51b6..4694241f4d22 100644 --- a/drivers/gpu/drm/bridge/samsung-dsim.c +++ b/drivers/gpu/drm/bridge/samsung-dsim.c @@ -1862,7 +1862,7 @@ static int samsung_dsim_register_te_irq(struct samsung_dsim *dsi, struct device int te_gpio_irq; int ret; - dsi->te_gpio = devm_gpiod_get_optional(dev, "te", GPIOD_IN); + dsi->te_gpio = gpiod_get_optional(dev, "te", GPIOD_IN); if (!dsi->te_gpio) return 0; else if (IS_ERR(dsi->te_gpio)) From e6bae5034ef4a61f3f062b18140d4f58c8b9a149 Mon Sep 17 00:00:00 2001 From: Niklas Cassel Date: Fri, 18 Sep 2026 14:40:32 +0200 Subject: [PATCH 133/365] ata: libata-core: Extend Samsung LPM quirk to AMD controllers A Samsung SSD 870 QVO 8TB connected to an AMD 600 Series chipset SATA controller is reported to time out on STANDBY IMMEDIATE during system suspend with med_power_with_dipm enabled. The command completes when using max_performance instead. The existing Samsung LPM quirk only matches ATI controllers, leaving AMD controllers unaffected. Rename it to ATA_QUIRK_NO_LPM_ON_ATI_AND_AMD and extend the vendor check to AMD for the same Samsung SSD model patterns. Keep LPM behavior unchanged for other controller vendors, including Intel. Leave ATA_QUIRK_NO_NCQ_ON_ATI restricted to ATI, since the reported AMD issue concerns LPM rather than NCQ. Link: https://bugzilla.kernel.org/show_bug.cgi?id=221986 Reviewed-by: Damien Le Moal Reviewed-by: Mario Limonciello (AMD) > --- Link: https://lore.kernel.org/r/20260918124030.1962773-5-cassel@kernel.org Signed-off-by: Niklas Cassel --- drivers/ata/libata-core.c | 15 ++++++++------- include/linux/libata.h | 4 ++-- 2 files changed, 10 insertions(+), 9 deletions(-) diff --git a/drivers/ata/libata-core.c b/drivers/ata/libata-core.c index f482c0a6d7e9..92233fb25051 100644 --- a/drivers/ata/libata-core.c +++ b/drivers/ata/libata-core.c @@ -2972,9 +2972,10 @@ static void ata_dev_config_lpm(struct ata_device *dev) (dev->id[ATA_ID_SATA_CAPABILITY] & 0xe) == 0x2) dev->quirks |= ATA_QUIRK_NOLPM; - /* ATI specific quirk */ - if ((dev->quirks & ATA_QUIRK_NO_LPM_ON_ATI) && - ata_dev_check_adapter(dev, PCI_VENDOR_ID_ATI)) + /* ATI and AMD specific quirk */ + if ((dev->quirks & ATA_QUIRK_NO_LPM_ON_ATI_AND_AMD) && + (ata_dev_check_adapter(dev, PCI_VENDOR_ID_ATI) || + ata_dev_check_adapter(dev, PCI_VENDOR_ID_AMD))) dev->quirks |= ATA_QUIRK_NOLPM; } @@ -4136,7 +4137,7 @@ static const char * const ata_quirk_names[] = { [__ATA_QUIRK_MAX_SEC] = "maxsec", [__ATA_QUIRK_MAX_TRIM_128M] = "maxtrim128m", [__ATA_QUIRK_NO_NCQ_ON_ATI] = "noncqonati", - [__ATA_QUIRK_NO_LPM_ON_ATI] = "nolpmonati", + [__ATA_QUIRK_NO_LPM_ON_ATI_AND_AMD] = "nolpmonatiandamd", [__ATA_QUIRK_NO_ID_DEV_LOG] = "noiddevlog", [__ATA_QUIRK_NO_LOG_DIR] = "nologdir", [__ATA_QUIRK_NO_FUA] = "nofua", @@ -4420,15 +4421,15 @@ static const struct ata_dev_quirks_entry __ata_dev_quirks[] = { { "Samsung SSD 860*", NULL, ATA_QUIRK_NO_NCQ_TRIM | ATA_QUIRK_ZERO_AFTER_TRIM | ATA_QUIRK_NO_NCQ_ON_ATI | - ATA_QUIRK_NO_LPM_ON_ATI }, + ATA_QUIRK_NO_LPM_ON_ATI_AND_AMD }, { "Samsung SSD 870*", NULL, ATA_QUIRK_NO_NCQ_TRIM | ATA_QUIRK_ZERO_AFTER_TRIM | ATA_QUIRK_NO_NCQ_ON_ATI | - ATA_QUIRK_NO_LPM_ON_ATI }, + ATA_QUIRK_NO_LPM_ON_ATI_AND_AMD }, { "SAMSUNG*MZ7LH*", NULL, ATA_QUIRK_NO_NCQ_TRIM | ATA_QUIRK_ZERO_AFTER_TRIM | ATA_QUIRK_NO_NCQ_ON_ATI | - ATA_QUIRK_NO_LPM_ON_ATI }, + ATA_QUIRK_NO_LPM_ON_ATI_AND_AMD }, { "FCCT*M500*", NULL, ATA_QUIRK_NO_NCQ_TRIM | ATA_QUIRK_ZERO_AFTER_TRIM }, diff --git a/include/linux/libata.h b/include/linux/libata.h index 313e96173b19..48bde275968d 100644 --- a/include/linux/libata.h +++ b/include/linux/libata.h @@ -76,7 +76,7 @@ enum ata_quirks { __ATA_QUIRK_MAX_SEC, /* Limit max sectors */ __ATA_QUIRK_MAX_TRIM_128M, /* Limit max trim size to 128M */ __ATA_QUIRK_NO_NCQ_ON_ATI, /* Disable NCQ on ATI chipset */ - __ATA_QUIRK_NO_LPM_ON_ATI, /* Disable LPM on ATI chipset */ + __ATA_QUIRK_NO_LPM_ON_ATI_AND_AMD, /* Disable LPM on ATI and AMD chipsets */ __ATA_QUIRK_NO_ID_DEV_LOG, /* Identify device log missing */ __ATA_QUIRK_NO_LOG_DIR, /* Do not read log directory */ __ATA_QUIRK_NO_FUA, /* Do not use FUA */ @@ -115,7 +115,7 @@ enum { ATA_QUIRK_MAX_SEC = BIT_ULL(__ATA_QUIRK_MAX_SEC), ATA_QUIRK_MAX_TRIM_128M = BIT_ULL(__ATA_QUIRK_MAX_TRIM_128M), ATA_QUIRK_NO_NCQ_ON_ATI = BIT_ULL(__ATA_QUIRK_NO_NCQ_ON_ATI), - ATA_QUIRK_NO_LPM_ON_ATI = BIT_ULL(__ATA_QUIRK_NO_LPM_ON_ATI), + ATA_QUIRK_NO_LPM_ON_ATI_AND_AMD = BIT_ULL(__ATA_QUIRK_NO_LPM_ON_ATI_AND_AMD), ATA_QUIRK_NO_ID_DEV_LOG = BIT_ULL(__ATA_QUIRK_NO_ID_DEV_LOG), ATA_QUIRK_NO_LOG_DIR = BIT_ULL(__ATA_QUIRK_NO_LOG_DIR), ATA_QUIRK_NO_FUA = BIT_ULL(__ATA_QUIRK_NO_FUA), From 88a0474d92ba102fff860db3cae9da87a0964fbf Mon Sep 17 00:00:00 2001 From: Niklas Cassel Date: Fri, 18 Sep 2026 14:40:33 +0200 Subject: [PATCH 134/365] ata: libata: Correct libata.force parameter documentation Align the documented libata.force options with their implementation. The force table accepts PIO modes 0 through 6, not mode 7, and ncqati controls NCQ generally rather than only queued TRIM. The max_sec_1024 and max_sec_lba48 options only set transfer size limits. Remove the misleading claim that they can also clear them. Reviewed-by: Damien Le Moal Link: https://lore.kernel.org/r/20260918124030.1962773-6-cassel@kernel.org Signed-off-by: Niklas Cassel --- Documentation/admin-guide/kernel-parameters.txt | 10 ++++------ 1 file changed, 4 insertions(+), 6 deletions(-) diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt index 33cd30996e47..24459f7ff810 100644 --- a/Documentation/admin-guide/kernel-parameters.txt +++ b/Documentation/admin-guide/kernel-parameters.txt @@ -3491,7 +3491,7 @@ Kernel parameters * SATA link speed limit: 1.5Gbps or 3.0Gbps. - * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7]. + * Transfer mode: pio[0-6], mwdma[0-4] and udma[0-7]. udma[/][16,25,33,44,66,100,133] notation is also allowed. @@ -3509,7 +3509,7 @@ Kernel parameters * [no]ncqtrim: Enable or disable queued DSM TRIM. - * [no]ncqati: Enable or disable NCQ trim on ATI chipset. + * [no]ncqati: Enable or disable NCQ on ATI chipsets. * [no]trim: Enable or disable (unqueued) TRIM. @@ -3540,11 +3540,9 @@ Kernel parameters * max_sec_128: Set transfer size limit to 128 sectors. - * max_sec_1024: Set or clear transfer size limit to - 1024 sectors. + * max_sec_1024: Set transfer size limit to 1024 sectors. - * max_sec_lba48: Set or clear transfer size limit to - 65535 sectors. + * max_sec_lba48: Set transfer size limit to 65535 sectors. * external: Mark port as external (hotplug-capable). From 437f3727b4fd715266c296435c116288f82666fd Mon Sep 17 00:00:00 2001 From: Zhiling Zou Date: Mon, 3 Aug 2026 11:28:22 +0800 Subject: [PATCH 135/365] wifi: mac80211: count only matching reservations in reserved switch ieee80211_vif_use_reserved_switch() validates a replacement by counting assigned links on ctx->replace_ctx and treating reservations too broadly. A link can still be assigned to the old channel context while holding a reservation that belongs to a different replacement operation. That can make the current replacement appear complete too early. If another link finalizes first, the function can free the WILL_BE_REPLACED chanctx while the unmatched link and driver still use it, leading to a use-after-free in later beacon generation. Require an in-place reservation to point at the current replacement context, and require the reserved and assigned contexts to be the matching replacement pair, before counting or moving the link. Fixes: 5bcae31d9cb1 ("mac80211: implement multi-vif in-place reservations") Cc: stable@vger.kernel.org Reported-by: Vega Signed-off-by: Zhiling Zou Link: https://patch.msgid.link/4b6c4b8648ab1920c38a40be32c0b1a2797b4ac6.1785726007.git.zhilinz@nebusec.ai Signed-off-by: Johannes Berg --- net/mac80211/chan.c | 30 +++++++++++++++++++++++------- 1 file changed, 23 insertions(+), 7 deletions(-) diff --git a/net/mac80211/chan.c b/net/mac80211/chan.c index 75bb204ad743..17b935448f13 100644 --- a/net/mac80211/chan.c +++ b/net/mac80211/chan.c @@ -1772,7 +1772,8 @@ ieee80211_link_use_reserved_assign(struct ieee80211_link_data *link) } static bool -ieee80211_link_has_in_place_reservation(struct ieee80211_link_data *link) +ieee80211_link_has_in_place_reservation(struct ieee80211_link_data *link, + struct ieee80211_chanctx *ctx) { struct ieee80211_sub_if_data *sdata = link->sdata; struct ieee80211_chanctx *old_ctx, *new_ctx; @@ -1782,6 +1783,9 @@ ieee80211_link_has_in_place_reservation(struct ieee80211_link_data *link) new_ctx = link->reserved_chanctx; old_ctx = ieee80211_link_get_chanctx(link); + if (new_ctx != ctx) + return false; + if (!old_ctx) return false; @@ -1794,6 +1798,12 @@ ieee80211_link_has_in_place_reservation(struct ieee80211_link_data *link) if (new_ctx->replace_state != IEEE80211_CHANCTX_REPLACES_OTHER) return false; + if (new_ctx->replace_ctx != old_ctx) + return false; + + if (old_ctx->replace_ctx != new_ctx) + return false; + return true; } @@ -1823,7 +1833,8 @@ static int ieee80211_chsw_switch_vifs(struct ieee80211_local *local, } for_each_chanctx_user_reserved(local, ctx, &iter) { - if (!ieee80211_link_has_in_place_reservation(iter.link)) + if (!ieee80211_link_has_in_place_reservation(iter.link, + ctx)) continue; old_ctx = ieee80211_link_get_chanctx(iter.link); @@ -1925,7 +1936,9 @@ static int ieee80211_vif_use_reserved_switch(struct ieee80211_local *local) for_each_chanctx_user_assigned(local, ctx->replace_ctx, &iter) { n_assigned++; - if (iter.link && iter.link->reserved_chanctx) { + if (iter.link && + ieee80211_link_has_in_place_reservation(iter.link, + ctx)) { n_reserved++; if (iter.link->reserved_ready) n_ready++; @@ -1950,7 +1963,8 @@ static int ieee80211_vif_use_reserved_switch(struct ieee80211_local *local) use_reserved: ctx->conf.radar_enabled = false; for_each_chanctx_user_reserved(local, ctx, &iter) { - if (ieee80211_link_has_in_place_reservation(iter.link) && + if (ieee80211_link_has_in_place_reservation(iter.link, + ctx) && !iter.link->reserved_ready) return -EAGAIN; @@ -1991,7 +2005,8 @@ static int ieee80211_vif_use_reserved_switch(struct ieee80211_local *local) } for_each_chanctx_user_reserved(local, ctx, &iter) { - if (!ieee80211_link_has_in_place_reservation(iter.link)) + if (!ieee80211_link_has_in_place_reservation(iter.link, + ctx)) continue; ieee80211_chan_bw_change(local, @@ -2040,7 +2055,7 @@ static int ieee80211_vif_use_reserved_switch(struct ieee80211_local *local) struct ieee80211_bss_conf *link_conf = link->conf; u64 changed = 0; - if (!ieee80211_link_has_in_place_reservation(link)) + if (!ieee80211_link_has_in_place_reservation(link, ctx)) continue; rcu_assign_pointer(link_conf->chanctx_conf, @@ -2091,7 +2106,8 @@ static int ieee80211_vif_use_reserved_switch(struct ieee80211_local *local) for_each_chanctx_user_reserved(local, ctx, &iter) { struct ieee80211_link_data *link = iter.link; - if (WARN_ON(ieee80211_link_has_in_place_reservation(link))) + if (WARN_ON(ieee80211_link_has_in_place_reservation(link, + ctx))) continue; if (!link->reserved_ready) From 3412e9a800786d1dec4a58367db1cba275e7ef11 Mon Sep 17 00:00:00 2001 From: Zhiling Zou Date: Mon, 3 Aug 2026 12:35:10 +0800 Subject: [PATCH 136/365] wifi: mac80211: keep paired old chanctx alive An in-place replacement keeps reciprocal replace_ctx pointers between the old WILL_BE_REPLACED chanctx and the new REPLACES_OTHER chanctx. However, the early free paths only consider assigned and reserved link users when deciding whether to free the old chanctx. That lets the old chanctx be freed too early while the replacement partner still points back to it, either when the last assigned link is released or when that link successfully reassigns to another existing context. Later unreserve and switch-finalization paths can then follow a stale replace_ctx pointer. Keep a paired old chanctx alive until the replacement pair has been torn down. Apply the same check to both __ieee80211_link_release_channel() and ieee80211_link_use_reserved_reassign(). If the replacement is later abandoned, ieee80211_link_unreserve_chanctx() already tears down the pairing before freeing the old chanctx once no users remain. Fixes: 5bcae31d9cb1 ("mac80211: implement multi-vif in-place reservations") Cc: stable@vger.kernel.org Reported-by: Vega Signed-off-by: Zhiling Zou Link: https://patch.msgid.link/b9b582b0db3814f641648b8886bb50f9d68f93ac.1785730815.git.zhilinz@nebusec.ai Signed-off-by: Johannes Berg --- net/mac80211/chan.c | 20 ++++++++++++++++++-- 1 file changed, 18 insertions(+), 2 deletions(-) diff --git a/net/mac80211/chan.c b/net/mac80211/chan.c index 17b935448f13..9cba79507767 100644 --- a/net/mac80211/chan.c +++ b/net/mac80211/chan.c @@ -231,6 +231,13 @@ int ieee80211_chanctx_refcount(struct ieee80211_local *local, return num; } +static bool +ieee80211_chanctx_has_replace_partner(struct ieee80211_chanctx *ctx) +{ + return ctx->replace_state == IEEE80211_CHANCTX_WILL_BE_REPLACED && + ctx->replace_ctx; +} + static int ieee80211_num_chanctx(struct ieee80211_local *local, int radio_idx) { struct ieee80211_chanctx *ctx; @@ -1405,6 +1412,7 @@ void ieee80211_link_unreserve_chanctx(struct ieee80211_link_data *link) { struct ieee80211_sub_if_data *sdata = link->sdata; struct ieee80211_chanctx *ctx = link->reserved_chanctx; + struct ieee80211_chanctx *old_ctx = NULL; lockdep_assert_wiphy(sdata->local->hw.wiphy); @@ -1418,6 +1426,7 @@ void ieee80211_link_unreserve_chanctx(struct ieee80211_link_data *link) if (WARN_ON(!ctx->replace_ctx)) return; + old_ctx = ctx->replace_ctx; WARN_ON(ctx->replace_ctx->replace_state != IEEE80211_CHANCTX_WILL_BE_REPLACED); WARN_ON(ctx->replace_ctx->replace_ctx != ctx); @@ -1428,6 +1437,11 @@ void ieee80211_link_unreserve_chanctx(struct ieee80211_link_data *link) list_del_rcu(&ctx->list); kfree_rcu(ctx, rcu_head); + + if (ieee80211_chanctx_refcount(sdata->local, + old_ctx) == 0) + ieee80211_free_chanctx(sdata->local, old_ctx, + false); } else { ieee80211_free_chanctx(sdata->local, ctx, false); } @@ -1707,7 +1721,8 @@ ieee80211_link_use_reserved_reassign(struct ieee80211_link_data *link) ieee80211_check_fast_xmit_iface(sdata); - if (ieee80211_chanctx_refcount(local, old_ctx) == 0) + if (ieee80211_chanctx_refcount(local, old_ctx) == 0 && + !ieee80211_chanctx_has_replace_partner(old_ctx)) ieee80211_free_chanctx(local, old_ctx, false); ieee80211_recalc_chanctx_min_def(local, new_ctx); @@ -2192,7 +2207,8 @@ void __ieee80211_link_release_channel(struct ieee80211_link_data *link, } ieee80211_assign_link_chanctx(link, NULL, false); - if (ieee80211_chanctx_refcount(local, ctx) == 0) + if (ieee80211_chanctx_refcount(local, ctx) == 0 && + !ieee80211_chanctx_has_replace_partner(ctx)) ieee80211_free_chanctx(local, ctx, skip_idle_recalc); link->radar_required = false; From 7c1611780823b76219cb1478db53dfd890c4a9a2 Mon Sep 17 00:00:00 2001 From: Tianchu Chen Date: Fri, 4 Sep 2026 15:23:34 +0000 Subject: [PATCH 137/365] wifi: cw1200: fix TX padding OOB read leaking heap data to device The TX path rounds the transfer size up to the bus alignment (align_size(), up to 511 bytes for SDIO, 1 byte for SPI) but sends the padded length straight from the frame buffer, which has no room reserved for the padding. cw1200_data_write() therefore reads past the end of the frame and the bus transfer carries those bytes to the device. A bogus device can collect up to 511 bytes of kernel heap data per TX frame. The RX path does not have this problem: it allocates the skb with the aligned length (alloc_len) before reading. Compute the exact padding with the bus align_size() callback, make sure queued TX frames have tailroom for it, and zero the padding bytes before the transfer so nothing past the frame is exposed. This is not expected to change driver behavior on most cases: frame contents and transfer sizes are unchanged, only the content of the padding bytes (previously undefined stale memory, now zero), which the device never reads since the WSM header carries the real frame length. The skb expansion can only fail under memory exhaustion, in which case the frame is dropped like on the existing error paths of cw1200_tx(). Discovered by Atuin - Automated Vulnerability Discovery Engine. Fixes: a910e4a94f69 ("cw1200: add driver for the ST-E CW1100 & CW1200 WLAN chipsets") Cc: stable@vger.kernel.org Assisted-by: LLM Signed-off-by: Tianchu Chen Link: https://patch.msgid.link/7982b4a7e64f74e2068790ec389ca680a081f730@linux.dev Signed-off-by: Johannes Berg --- drivers/net/wireless/st/cw1200/bh.c | 8 +++++++- drivers/net/wireless/st/cw1200/txrx.c | 14 ++++++++++++++ 2 files changed, 21 insertions(+), 1 deletion(-) diff --git a/drivers/net/wireless/st/cw1200/bh.c b/drivers/net/wireless/st/cw1200/bh.c index b034bab4b489..9367a981f1de 100644 --- a/drivers/net/wireless/st/cw1200/bh.c +++ b/drivers/net/wireless/st/cw1200/bh.c @@ -344,6 +344,7 @@ static int cw1200_bh_tx_helper(struct cw1200_common *priv, int *tx_burst) { size_t tx_len; + size_t aligned_len; u8 *data; int ret; struct wsm_hdr *wsm; @@ -376,9 +377,14 @@ static int cw1200_bh_tx_helper(struct cw1200_common *priv, atomic_inc(&priv->bh_tx); - tx_len = priv->hwbus_ops->align_size( + aligned_len = priv->hwbus_ops->align_size( priv->hwbus_priv, tx_len); + /* Zero the bus alignment padding so no stale data is sent. */ + if (aligned_len > tx_len) + memset(data + tx_len, 0, aligned_len - tx_len); + tx_len = aligned_len; + /* Check if not exceeding CW1200 capabilities */ if (WARN_ON_ONCE(tx_len > EFFECTIVE_BUF_SIZE)) pr_debug("Write aligned len: %zu\n", tx_len); diff --git a/drivers/net/wireless/st/cw1200/txrx.c b/drivers/net/wireless/st/cw1200/txrx.c index 084d52b11f5b..e13695667a00 100644 --- a/drivers/net/wireless/st/cw1200/txrx.c +++ b/drivers/net/wireless/st/cw1200/txrx.c @@ -14,6 +14,7 @@ #include "wsm.h" #include "bh.h" #include "sta.h" +#include "hwbus.h" #include "debug.h" #define CW1200_INVALID_RATE_ID (0xFF) @@ -716,6 +717,7 @@ void cw1200_tx(struct ieee80211_hw *dev, struct ieee80211_sta *sta; struct wsm_tx *wsm; bool tid_update = false; + size_t pad_len; u8 flags = 0; int ret; @@ -764,6 +766,18 @@ void cw1200_tx(struct ieee80211_hw *dev, sta = t.sta; + /* + * The bus transfer is padded up to the bus alignment and the + * padding is transferred from the frame buffer. Make sure the + * buffer has room for the padding so the device never receives + * data past the end of the frame. + */ + pad_len = priv->hwbus_ops->align_size(priv->hwbus_priv, skb->len) - + skb->len; + if (skb_tailroom(skb) < pad_len && + pskb_expand_head(skb, 0, pad_len, GFP_ATOMIC)) + goto drop; + spin_lock_bh(&priv->ps_state_lock); { tid_update = cw1200_tx_h_pm_state(priv, &t); From ce67ebe9d8629776ce38e6995ae72fb903c828db Mon Sep 17 00:00:00 2001 From: Aamir Ahmed Date: Sat, 12 Sep 2026 14:47:55 +0100 Subject: [PATCH 138/365] wifi: wfx: validate MIB read length before copying to caller wfx_hif_read_mib() copies reply->length bytes from the firmware response into the caller-provided buffer without checking that reply->length does not exceed val_len. A firmware response with a length field larger than expected overruns the caller buffer and reads past the end of the reply allocation. Replace the dead -ENOMEM check with an active validation that reply->length fits within val_len before the memcpy. Fixes: f95a29d40782 ("staging: wfx: add HIF commands helpers") Assisted-by: LLM Signed-off-by: Aamir Ahmed Link: https://patch.msgid.link/AS8P251MB000155034FC391A2A54B9803C8BD2@AS8P251MB0001.EURP251.PROD.OUTLOOK.COM Signed-off-by: Johannes Berg --- drivers/net/wireless/silabs/wfx/hif_tx.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/drivers/net/wireless/silabs/wfx/hif_tx.c b/drivers/net/wireless/silabs/wfx/hif_tx.c index 9f403d275cb1..9efb108bd877 100644 --- a/drivers/net/wireless/silabs/wfx/hif_tx.c +++ b/drivers/net/wireless/silabs/wfx/hif_tx.c @@ -207,9 +207,11 @@ int wfx_hif_read_mib(struct wfx_dev *wdev, int vif_id, u16 mib_id, void *val, si dev_warn(wdev->dev, "%s: confirmation mismatch request\n", __func__); ret = -EIO; } - if (ret == -ENOMEM) + if (!ret && le16_to_cpu(reply->length) > val_len) { dev_err(wdev->dev, "buffer is too small to receive %s (%zu < %d)\n", wfx_get_mib_name(mib_id), val_len, le16_to_cpu(reply->length)); + ret = -EIO; + } if (!ret) memcpy(val, &reply->mib_data, le16_to_cpu(reply->length)); else From f8ebe286394c0bbb4a6dbacb182a34ab6376922d Mon Sep 17 00:00:00 2001 From: Guangshuo Li Date: Wed, 16 Sep 2026 15:37:36 +0800 Subject: [PATCH 139/365] wifi: iwlegacy: 3945: free EEPROM data on remove il3945_pci_probe() calls il_eeprom_init(), which allocates memory for the EEPROM data. The probe failure paths release this allocation with il_eeprom_free(), but the remove path does not free it after a successful probe. As a result, the EEPROM data is leaked when the PCI device is removed. Free the EEPROM data in il3945_pci_remove() before releasing the mac80211 hardware structure. This issue was found by manual code inspection. Fixes: 073d3f5f1b3b ("iwlwifi: changing EEPROM layout handling") Cc: stable@vger.kernel.org Signed-off-by: Guangshuo Li Acked-by: Stanislaw Gruszka Link: https://patch.msgid.link/20260916073736.2952028-1-lgs201920130244@gmail.com Signed-off-by: Johannes Berg --- drivers/net/wireless/intel/iwlegacy/3945-mac.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/net/wireless/intel/iwlegacy/3945-mac.c b/drivers/net/wireless/intel/iwlegacy/3945-mac.c index cbaf250626c5..f7ca182d1494 100644 --- a/drivers/net/wireless/intel/iwlegacy/3945-mac.c +++ b/drivers/net/wireless/intel/iwlegacy/3945-mac.c @@ -3807,6 +3807,7 @@ il3945_pci_remove(struct pci_dev *pdev) il_free_channel_map(il); il_free_geos(il); + il_eeprom_free(il); kfree(il->scan_cmd); dev_kfree_skb(il->beacon_skb); ieee80211_free_hw(il->hw); From ea4debcd8016f73c5dee3a29250a3d7977f015ef Mon Sep 17 00:00:00 2001 From: Sk Anirban Date: Wed, 9 Sep 2026 17:19:32 +0530 Subject: [PATCH 140/365] drm/xe/gt_throttle: Report power brake as a throttle reason on CRI CRI defines bit 5 of the perf limit reasons register as a power brake (PWRBRK) indicator. Add PWRBRK_MASK and a reason_pwrbrk sysfs attribute for CRI in place of reason_ratl. Signed-off-by: Sk Anirban Fixes: 8578e6d0546c ("drm/xe/gt_throttle: Drop individual show functions") Reviewed-by: Raag Jadav Signed-off-by: Matthew Brost Link: https://patch.msgid.link/20260909114931.1039331-2-sk.anirban@intel.com (cherry picked from commit e199c851c0ab608a0ca89e7be1756a1461b41a2c) Signed-off-by: Rodrigo Vivi --- drivers/gpu/drm/xe/regs/xe_gt_regs.h | 1 + drivers/gpu/drm/xe/xe_gt_throttle.c | 5 +++-- 2 files changed, 4 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/xe/regs/xe_gt_regs.h b/drivers/gpu/drm/xe/regs/xe_gt_regs.h index 08251c7a1a4b..247a736a54aa 100644 --- a/drivers/gpu/drm/xe/regs/xe_gt_regs.h +++ b/drivers/gpu/drm/xe/regs/xe_gt_regs.h @@ -651,6 +651,7 @@ #define MEM_THERMAL_MASK REG_BIT(2) #define VR_THERMAL_MASK REG_BIT(3) #define ICCMAX_MASK REG_BIT(4) +#define PWRBRK_MASK REG_BIT(5) #define SOC_AVG_THERMAL_MASK REG_BIT(6) #define FASTVMODE_MASK REG_BIT(7) #define PSYS_PL1_MASK REG_BIT(12) diff --git a/drivers/gpu/drm/xe/xe_gt_throttle.c b/drivers/gpu/drm/xe/xe_gt_throttle.c index 1e7e3a31aa69..c0af5484611d 100644 --- a/drivers/gpu/drm/xe/xe_gt_throttle.c +++ b/drivers/gpu/drm/xe/xe_gt_throttle.c @@ -39,7 +39,7 @@ * - ``reason_mem_thermal``: Memory thermal * - ``reason_vr_thermal``: VR thermal * - ``reason_iccmax``: ICCMAX - * - ``reason_ratl``: RATL thermal algorithm + * - ``reason_pwrbrk``: Power brake * - ``reason_soc_avg_thermal``: SoC average temp * - ``reason_fastvmode``: VR is hitting FastVMode * - ``reason_psys_pl1``: PSYS PL1 @@ -200,6 +200,7 @@ static THROTTLE_ATTR_RO(reason_psys_pl1, PSYS_PL1_MASK); static THROTTLE_ATTR_RO(reason_psys_pl2, PSYS_PL2_MASK); static THROTTLE_ATTR_RO(reason_p0_freq, P0_FREQ_MASK); static THROTTLE_ATTR_RO(reason_psys_crit, PSYS_CRIT_MASK); +static THROTTLE_ATTR_RO(reason_pwrbrk, PWRBRK_MASK); static struct attribute *cri_throttle_attrs[] = { /* Common */ @@ -209,12 +210,12 @@ static struct attribute *cri_throttle_attrs[] = { &attr_reason_pl2.attr.attr, &attr_reason_pl4.attr.attr, &attr_reason_prochot.attr.attr, - &attr_reason_ratl.attr.attr, /* CRI */ &attr_reason_vr_thermal.attr.attr, &attr_reason_soc_thermal.attr.attr, &attr_reason_mem_thermal.attr.attr, &attr_reason_iccmax.attr.attr, + &attr_reason_pwrbrk.attr.attr, &attr_reason_soc_avg_thermal.attr.attr, &attr_reason_fastvmode.attr.attr, &attr_reason_psys_pl1.attr.attr, From c7a925c84704ec431598f9411dd89c0e16ae34ae Mon Sep 17 00:00:00 2001 From: Shuicheng Lin Date: Mon, 14 Sep 2026 21:53:18 +0000 Subject: [PATCH 141/365] drm/xe/tlb_inval: Treat wedged-device invalidations as complete A TLB invalidation issued on a wedged device fails with -ENOTRECOVERABLE. xe_tlb_inval_issue() squashes only -ECANCELED, so the error reaches callers that treat it as unexpected and WARN, tainting the kernel on a wedge that was deliberately caused: ggtt_invalidate_gt_tlb() drivers/gpu/drm/xe/xe_ggtt.c xe_svm_invalidate() drivers/gpu/drm/xe/xe_svm.c xe_bo_trigger_rebind() drivers/gpu/drm/xe/xe_bo.c xe_vma_userptr_do_inval() drivers/gpu/drm/xe/xe_userptr.c igt@xe_exec_reset@gt-reset-fault-injection hits the GGTT one, turning an otherwise passing run into an abort: *ERROR* SIGID=102 FATAL (-EIO) WEDGED: Device declared wedged! Tile0: GT1: Failed to invalidate GGTT (-ENOTRECOVERABLE) WARNING: drivers/gpu/drm/xe/xe_ggtt.c:588 at ggtt_invalidate_gt_tlb Workqueue: xe-guc-destroy-wq __guc_exec_queue_destroy_async [xe] ggtt_node_remove+0xe3/0x100 [xe] xe_ggtt_remove_bo+0x89/0x2c0 [xe] xe_ttm_bo_destroy+0xcb/0x330 [xe] ... xe_lrc_destroy+0x74/0x90 [xe] xe_exec_queue_fini+0x2d/0x60 [xe] -ECANCELED and -ENOTRECOVERABLE mean the same thing at this layer: the message was dropped rather than delivered, and the fence has already been signalled before the error is returned, so there is nothing left to wait for. Squash both. A wedged device is only recovered by a fresh initialisation, so the error return in xe_bo_trigger_rebind() becomes unreachable by design. v2: fix all invalidation paths. (Sashiko) Fixes: 50fa9acac26f ("drm/xe/guc: distinguish wedged from recoverable cancellation") Assisted-by: Claude:claude-opus-5 Cc: Sk Anirban Cc: Matthew Brost Signed-off-by: Shuicheng Lin Reviewed-by: Matthew Brost Signed-off-by: Matthew Brost Link: https://patch.msgid.link/20260914215318.200603-1-shuicheng.lin@intel.com (cherry picked from commit af14e3705345cb57c53b237169873233052a5c16) Signed-off-by: Rodrigo Vivi --- drivers/gpu/drm/xe/xe_tlb_inval.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/xe/xe_tlb_inval.c b/drivers/gpu/drm/xe/xe_tlb_inval.c index bbd21d393062..67b205bf1291 100644 --- a/drivers/gpu/drm/xe/xe_tlb_inval.c +++ b/drivers/gpu/drm/xe/xe_tlb_inval.c @@ -280,7 +280,8 @@ static void xe_tlb_inval_fence_prep(struct xe_tlb_inval_fence *fence) xe_tlb_inval_fence_signal_unlocked((__fence)); \ mutex_unlock(&(__tlb_inval)->seqno_lock); \ \ - __ret == -ECANCELED ? 0 : __ret; \ + /* Undelivered: fence already signalled, report done */ \ + (__ret == -ECANCELED || __ret == -ENOTRECOVERABLE) ? 0 : __ret; \ }) /** From be1df8badae513e01d9575398438716cfae18655 Mon Sep 17 00:00:00 2001 From: Matthew Brost Date: Thu, 17 Sep 2026 13:31:58 -0700 Subject: [PATCH 142/365] drm/xe: Keep walking on SVM eviction failure The desired behavior for SVM eviction failures, which can occur due to various uncontrollable races, is for TTM to continue walking the LRU list and look for another eviction candidate. This is expressed by returning -ENOSPC from the ->move() callback. Adjust the SVM eviction failure path because of races in ->move() to return -ENOSPC so that TTM continues searching for another buffer to evict. Fixes: 3ca608dc7561 ("drm/xe: Basic SVM BO eviction") Cc: stable@vger.kernel.org Signed-off-by: Matthew Brost Reviewed-by: Himal Prasad Ghimiray Link: https://patch.msgid.link/20260917203158.292823-1-matthew.brost@intel.com Signed-off-by: Rodrigo Vivi (cherry picked from commit 36a86c23588b8f57c9d20feb4cf5a2ab27e3baba) Signed-off-by: Rodrigo Vivi --- drivers/gpu/drm/xe/xe_bo.c | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/drivers/gpu/drm/xe/xe_bo.c b/drivers/gpu/drm/xe/xe_bo.c index dde309821237..b4921a627ff3 100644 --- a/drivers/gpu/drm/xe/xe_bo.c +++ b/drivers/gpu/drm/xe/xe_bo.c @@ -1037,6 +1037,13 @@ static int xe_bo_move(struct ttm_buffer_object *ttm_bo, bool evict, } else { drm_dbg(&xe->drm, "Evict system allocator BO failed=%pe\n", ERR_PTR(ret)); + /* + * The semantic we want upon SVM eviction failure + * because of racing access is keep walking for + * eviction, which is -ENOSPC. + */ + if (ret == -EBUSY) + ret = -ENOSPC; } goto out; From 24a22fb3c731474b68e986af6804db450fb88617 Mon Sep 17 00:00:00 2001 From: Matthew Auld Date: Fri, 18 Sep 2026 14:10:35 +0100 Subject: [PATCH 143/365] drm/xe/vm: nuke PTs only after unlinking contested VMAs MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit In xe_vm_close_and_put(), external-BO VMAs are queued on the contested list for deferred destruction via xe_vma_destroy_unlocked(). However, xe_vm_pt_destroy() was previously invoked before processing contested VMAs, destroying vm->pt_root while those VMAs were still linked to their respective buffer objects (vm_bo->list.gpuva). If a concurrent thread evicts one of those shared buffer objects, xe_bo_trigger_rebind() holding only bo->resv walks the BO's VMAs and, in fault mode, calls xe_vm_invalidate_vma() -> xe_pt_zap_ptes(). Because vm->pt_root[tile->id] is already NULL, dereferencing pt->level causes a NULL ptr deref. Fix this by deferring xe_vm_free_scratch() and xe_vm_pt_destroy() until after all contested VMAs have been unlinked and destroyed. User is reporting hitting a NULL ptr deref in xe_pt_zap_ptes(), which could be explained by this race. Assisted-by: LLM Fixes: b06d47be7c83 ("drm/xe: Port Xe to GPUVA") Link: https://gitlab.freedesktop.org/drm/xe/kernel/-/work_items/9290 Signed-off-by: Matthew Auld Cc: Thomas Hellström Cc: Matthew Brost Cc: # v6.12+ Reviewed-by: Thomas Hellström Reviewed-by: Matthew Brost Link: https://patch.msgid.link/20260918131034.598078-2-matthew.auld@intel.com (cherry picked from commit c2863648959489767f08892fd6e90577d2ea0b6a) Signed-off-by: Rodrigo Vivi --- drivers/gpu/drm/xe/xe_vm.c | 21 +++++++++------------ 1 file changed, 9 insertions(+), 12 deletions(-) diff --git a/drivers/gpu/drm/xe/xe_vm.c b/drivers/gpu/drm/xe/xe_vm.c index 23952ad8951e..ef20e205a734 100644 --- a/drivers/gpu/drm/xe/xe_vm.c +++ b/drivers/gpu/drm/xe/xe_vm.c @@ -1947,21 +1947,13 @@ void xe_vm_close_and_put(struct xe_vm *vm) vma->gpuva.flags |= XE_VMA_DESTROYED; } - /* - * All vm operations will add shared fences to resv. - * The only exception is eviction for a shared object, - * but even so, the unbind when evicted would still - * install a fence to resv. Hence it's safe to - * destroy the pagetables immediately. - */ - xe_vm_free_scratch(vm); - xe_vm_pt_destroy(vm); xe_vm_unlock(vm); /* - * VM is now dead, cannot re-add nodes to vm->vmas if it's NULL - * Since we hold a refcount to the bo, we can remove and free - * the members safely without locking. + * Unlink and destroy all contested external-BO VMAs before destroying + * the page tables. Otherwise, concurrent eviction holding only bo->resv + * can walk the BO's VMAs and attempt to invalidate/zap page tables that + * have already been freed. */ list_for_each_entry_safe(vma, next_vma, &contested, combined_links.destroy) { @@ -1969,6 +1961,11 @@ void xe_vm_close_and_put(struct xe_vm *vm) xe_vma_destroy_unlocked(vma); } + xe_vm_lock(vm, false); + xe_vm_free_scratch(vm); + xe_vm_pt_destroy(vm); + xe_vm_unlock(vm); + xe_svm_fini(vm); up_write(&vm->lock); From 7325a44474f20c1393a1c114c6811abce6ddfaf4 Mon Sep 17 00:00:00 2001 From: Zhao Li Date: Wed, 16 Sep 2026 19:36:48 +0800 Subject: [PATCH 144/365] wifi: mac80211: release mesh path quota on failed additions Repeated failed or duplicate mesh path additions can exhaust MESH_MAX_MPATHS while the path table holds fewer paths, which stops new paths to reachable destinations from being created. mesh_path_add() reserves a slot before allocating and inserting a path. If allocation fails, the function returns without releasing the slot. If the rhashtable insertion reports an error or an existing destination, the function discards its candidate but retains the reservation. No new path is installed in any of these cases. Release the reservation whenever the candidate is not installed. Fixes: ae76eef027f7 ("mac80211: return new mpath from mesh_path_add()") Fixes: 60854fd94573 ("mac80211: mesh: convert path table to rhashtable") Cc: stable@vger.kernel.org Assisted-by: LLM sparse kasan Signed-off-by: Zhao Li Link: https://patch.msgid.link/20260916113649.28315-2-enderaoelyther@gmail.com Signed-off-by: Johannes Berg --- net/mac80211/mesh_pathtbl.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/net/mac80211/mesh_pathtbl.c b/net/mac80211/mesh_pathtbl.c index 03171cf00855..770fc16b439b 100644 --- a/net/mac80211/mesh_pathtbl.c +++ b/net/mac80211/mesh_pathtbl.c @@ -694,8 +694,10 @@ struct mesh_path *mesh_path_add(struct ieee80211_sub_if_data *sdata, return ERR_PTR(-ENOSPC); new_mpath = mesh_path_new(sdata, dst, GFP_ATOMIC); - if (!new_mpath) + if (!new_mpath) { + atomic_dec(&sdata->u.mesh.mpaths); return ERR_PTR(-ENOMEM); + } tbl = &sdata->u.mesh.mesh_paths; spin_lock_bh(&tbl->walk_lock); @@ -708,6 +710,7 @@ struct mesh_path *mesh_path_add(struct ieee80211_sub_if_data *sdata, if (mpath) { kfree(new_mpath); + atomic_dec(&sdata->u.mesh.mpaths); if (IS_ERR(mpath)) return mpath; From a5f35d4b112af1415ff06bd501c8cc26cc08aa8b Mon Sep 17 00:00:00 2001 From: Zhao Li Date: Wed, 16 Sep 2026 19:36:49 +0800 Subject: [PATCH 145/365] wifi: mac80211: account proxy paths against the mesh path limit Proxy-path churn can drive the interface path counter negative and let later mesh path additions exceed MESH_MAX_MPATHS. mesh_path_free_rcu() decrements the interface mpaths counter for entries from both the mesh and proxy path tables, but mpp_path_add() never reserves a slot in that counter. Removing or expiring a proxy path therefore returns a slot that was never charged. mesh_path_add() uses the same counter to enforce MESH_MAX_MPATHS, so once it falls below the number of installed paths the limit no longer holds. Reserve the shared quota before allocating a proxy path and release it on every unsuccessful addition. Mesh and proxy paths now share one 1024-entry budget, so mpp_path_add() can return -ENOSPC at that limit. Fixes: ece1a2e7e860 ("mac80211: Remove mesh paths when an interface is removed") Cc: stable@vger.kernel.org Assisted-by: LLM sparse kasan Signed-off-by: Zhao Li Link: https://patch.msgid.link/20260916113649.28315-3-enderaoelyther@gmail.com Signed-off-by: Johannes Berg --- net/mac80211/mesh_pathtbl.c | 13 ++++++++++--- 1 file changed, 10 insertions(+), 3 deletions(-) diff --git a/net/mac80211/mesh_pathtbl.c b/net/mac80211/mesh_pathtbl.c index 770fc16b439b..1c55b14c2ac0 100644 --- a/net/mac80211/mesh_pathtbl.c +++ b/net/mac80211/mesh_pathtbl.c @@ -736,10 +736,15 @@ int mpp_path_add(struct ieee80211_sub_if_data *sdata, if (is_multicast_ether_addr(dst)) return -EOPNOTSUPP; + if (!atomic_add_unless(&sdata->u.mesh.mpaths, 1, MESH_MAX_MPATHS)) + return -ENOSPC; + new_mpath = mesh_path_new(sdata, dst, GFP_ATOMIC); - if (!new_mpath) + if (!new_mpath) { + atomic_dec(&sdata->u.mesh.mpaths); return -ENOMEM; + } memcpy(new_mpath->mpp, mpp, ETH_ALEN); tbl = &sdata->u.mesh.mpp_paths; @@ -752,10 +757,12 @@ int mpp_path_add(struct ieee80211_sub_if_data *sdata, hlist_add_head_rcu(&new_mpath->walk_list, &tbl->walk_head); spin_unlock_bh(&tbl->walk_lock); - if (ret) + if (ret) { kfree(new_mpath); - else + atomic_dec(&sdata->u.mesh.mpaths); + } else { mesh_fast_tx_flush_addr(sdata, dst); + } sdata->u.mesh.mpp_paths_generation++; return ret; From 67a9e80c3ca20c9044cd6ee72cd6278dd19b5023 Mon Sep 17 00:00:00 2001 From: Zhao Li Date: Wed, 16 Sep 2026 19:37:35 +0800 Subject: [PATCH 146/365] wifi: mac80211: drain PS delivery work during station teardown Removing an AP or mesh station can notify the driver and wake TXQs after the station has already been removed from the driver, and can update num_sta_ps and the power-save queues after station cleanup has purged them. sta_deliver_ps_frames() tests sta->dead only before it starts delivery. A worker that passes that test can be delayed while teardown sets sta->dead and removes the station from the driver; when it resumes it still runs the full delivery. The existing cancel_work_sync() sits near the end of __cleanup_single_sta(), after the final driver state transition and after PS accounting and queue purging, so it drains the worker only once the state it has to protect has already changed. Before commit d34ba2168a3c ("mac80211: don't delay station destruction") cleanup was deferred through the same ordered workqueue as the delivery work, which kept the two ordered. Cancel the work after setting sta->dead and before moving the driver state, and move the existing cancellation to the start of __cleanup_single_sta(). The first drain closes the driver-lifetime window; driver callbacks during the remaining state transitions can still queue drv_deliver_wk through ieee80211_sta_block_awake(), so retain the second drain at the start of __cleanup_single_sta(). That drain also covers insertion failures, which reach cleanup without setting sta->dead. Fixes: d34ba2168a3c ("mac80211: don't delay station destruction") Cc: stable@vger.kernel.org Assisted-by: LLM sparse smatch coccinelle kasan Signed-off-by: Zhao Li Link: https://patch.msgid.link/20260916113735.31029-1-enderaoelyther@gmail.com Signed-off-by: Johannes Berg --- net/mac80211/sta_info.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/net/mac80211/sta_info.c b/net/mac80211/sta_info.c index fdf00cbf49d8..e81cd155197b 100644 --- a/net/mac80211/sta_info.c +++ b/net/mac80211/sta_info.c @@ -137,6 +137,8 @@ static void __cleanup_single_sta(struct sta_info *sta) struct ieee80211_local *local = sdata->local; struct ps_data *ps; + cancel_work_sync(&sta->drv_deliver_wk); + if (test_sta_flag(sta, WLAN_STA_PS_STA) || test_sta_flag(sta, WLAN_STA_PS_DRIVER) || test_sta_flag(sta, WLAN_STA_PS_DELIVER)) { @@ -166,8 +168,6 @@ static void __cleanup_single_sta(struct sta_info *sta) if (ieee80211_vif_is_mesh(&sdata->vif)) mesh_sta_cleanup(sta); - cancel_work_sync(&sta->drv_deliver_wk); - /* * Destroy aggregation state here. It would be nice to wait for the * driver to finish aggregation stop and then clean up, but for now @@ -1582,6 +1582,8 @@ static void __sta_info_destroy_part2(struct sta_info *sta, bool recalc) sta->dead = true; + cancel_work_sync(&sta->drv_deliver_wk); + local->num_sta--; local->sta_generation++; From a8b8881d0c1aff544fae190469d86db7b07f43a9 Mon Sep 17 00:00:00 2001 From: Wentao Liang Date: Thu, 17 Sep 2026 12:23:25 +0000 Subject: [PATCH 147/365] wifi: wlcore: Fix runtime PM leak in wlcore_remove() pm_runtime_get_sync() keeps the runtime PM usage counter elevated even when it returns an error, so the reference taken at the beginning of wlcore_remove() has to be dropped on every exit path. When the device was not initialized the function returned early without doing so, leaking the reference. Release it with pm_runtime_put_noidle() as done on the other early error paths of this driver. Fixes: fa2648a34e73f ("wlcore: Add support for runtime PM") Cc: stable@vger.kernel.org Signed-off-by: Wentao Liang Link: https://patch.msgid.link/20260917122325.2150910-1-vulab@iscas.ac.cn Signed-off-by: Johannes Berg --- drivers/net/wireless/ti/wlcore/main.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/drivers/net/wireless/ti/wlcore/main.c b/drivers/net/wireless/ti/wlcore/main.c index edf6ca23c6c3..26aabf4ea127 100644 --- a/drivers/net/wireless/ti/wlcore/main.c +++ b/drivers/net/wireless/ti/wlcore/main.c @@ -6809,8 +6809,10 @@ void wlcore_remove(struct platform_device *pdev) if (pdev_data->family && pdev_data->family->nvs_name) wait_for_completion(&wl->nvs_loading_complete); - if (!wl->initialized) + if (!wl->initialized) { + pm_runtime_put_noidle(wl->dev); return; + } if (wl->wakeirq >= 0) { dev_pm_clear_wake_irq(wl->dev); From 5e4f3509793b4db2c0835340224743db680e861f Mon Sep 17 00:00:00 2001 From: Yuchao Zhang Date: Fri, 18 Sep 2026 11:40:04 +0800 Subject: [PATCH 148/365] wifi: mac80211: drop oversized fragments to avoid extra_len overflow In ieee80211_rx_h_defragment(), fragment payloads are accumulated into entry->extra_len as subsequent fragments arrive: entry->extra_len += rx->skb->len; When the final fragment arrives, the head fragment is dequeued, and pskb_expand_head() is called with entry->extra_len to ensure sufficient tailroom for all queued fragments before copying: if (skb_tailroom(rx->skb) < entry->extra_len) { if (unlikely(pskb_expand_head(rx->skb, 0, entry->extra_len, GFP_ATOMIC))) { ... } } while ((skb = __skb_dequeue(&entry->skb_list))) { skb_put_data(rx->skb, skb->data, skb->len); dev_kfree_skb(skb); } In commit 69f132236827 ("mac80211: shrink struct ieee80211_fragment_entry"), entry->extra_len was narrowed from unsigned int to u16 in order to reduce structure memory footprint. However, an IEEE 802.11 frame sequence can contain up to 16 fragments (frag numbers 0..15). If a sequence of large fragments arrives (e.g. from a malicious peer or rogue AP), the sum of fragment lengths can exceed 65535 bytes (for example, 15 fragments of 4400 bytes total 66000 bytes). Because extra_len is a u16, this addition overflows and wraps around modulo 65536 (e.g. 66000 wraps to 464). Consequently, pskb_expand_head() allocates only the truncated amount of tailroom (or is skipped entirely if the head fragment already has >= 464 bytes of tailroom). When skb_put_data() subsequently iterates through the queued skb list, it appends the full payload into the undersized buffer, causing skb_put() to trigger skb_over_panic() and crash the kernel in softirq context. A legitimate MSDU in IEEE 802.11 is at most 2304 bytes (or up to 7935/11454 bytes for A-MSDU, which is not fragmented), well below U16_MAX. Fix this without increasing the size of struct ieee80211_fragment_entry by checking whether adding the incoming fragment length would exceed U16_MAX. If so, purge the queued fragments and drop the frame. Fixes: 69f132236827 ("mac80211: shrink struct ieee80211_fragment_entry") Signed-off-by: Yuchao Zhang Link: https://patch.msgid.link/20260918034004.85078-2-ndaugoing@gmail.com [assign a separate drop reason] Signed-off-by: Johannes Berg --- net/mac80211/drop.h | 1 + net/mac80211/rx.c | 6 ++++++ 2 files changed, 7 insertions(+) diff --git a/net/mac80211/drop.h b/net/mac80211/drop.h index f06a8aa905c5..f433ee17cce5 100644 --- a/net/mac80211/drop.h +++ b/net/mac80211/drop.h @@ -108,6 +108,7 @@ typedef unsigned int __bitwise ieee80211_rx_result; R(RX_DROP_U_UNHANDLED_PREQ) \ R(RX_DROP_U_UNHANDLED_MGMT_STYPE) \ R(RX_DROP_U_NO_LINK) \ + R(RX_DROP_U_DEFRAG_OVERFLOW) \ /* this line for the trailing \ - add before this */ /* having two enums allows for checking ieee80211_rx_result use with sparse */ diff --git a/net/mac80211/rx.c b/net/mac80211/rx.c index 5e26be8e27d8..7ead509304eb 100644 --- a/net/mac80211/rx.c +++ b/net/mac80211/rx.c @@ -2499,6 +2499,12 @@ ieee80211_rx_h_defragment(struct ieee80211_rx_data *rx) } skb_pull(rx->skb, ieee80211_hdrlen(fc)); + if (unlikely((u32)entry->extra_len + rx->skb->len > U16_MAX)) { + I802_DEBUG_INC(rx->local->rx_handlers_drop_defrag); + __skb_queue_purge(&entry->skb_list); + return RX_DROP_U_DEFRAG_OVERFLOW; + } + __skb_queue_tail(&entry->skb_list, rx->skb); entry->last_frag = frag; entry->extra_len += rx->skb->len; From ee3aadc43ea69d98f666431e787137ac0badbb85 Mon Sep 17 00:00:00 2001 From: Zihan Xi Date: Fri, 18 Sep 2026 05:26:51 +0000 Subject: [PATCH 149/365] wifi: mac80211: fix mesh fast xmit path deletion UAF mesh_fast_tx_cache() stores raw mesh_path pointers. Path deletion flushes the cache and then frees the path with kfree_rcu(), but a lookup that already holds the path can insert a new cache entry after the flush. The cache then points at freed memory. Set MESH_PATH_DELETED before flushing, and skip inserting a cache entry if the path or MPP path is already deleted. Check this under the cache walk lock so it is ordered with the flush. Use WRITE_ONCE() for the deletion flag updates because the cache check reads flags without taking state_lock. Fixes: d5edb9ae8d56 ("wifi: mac80211: mesh fast xmit support") Cc: stable@vger.kernel.org Reported-by: Vega Assisted-by: LLM Co-developed-by: Luxing Yin Signed-off-by: Luxing Yin Signed-off-by: Zihan Xi Link: https://patch.msgid.link/1daa7a98199fe6965c2da7c42717073a6fd39a0b.1789615568.git.zihanx@nebusec.ai Signed-off-by: Johannes Berg --- net/mac80211/mesh_pathtbl.c | 12 +++++++++++- 1 file changed, 11 insertions(+), 1 deletion(-) diff --git a/net/mac80211/mesh_pathtbl.c b/net/mac80211/mesh_pathtbl.c index 1c55b14c2ac0..07a21dfa8674 100644 --- a/net/mac80211/mesh_pathtbl.c +++ b/net/mac80211/mesh_pathtbl.c @@ -577,6 +577,12 @@ void mesh_fast_tx_cache(struct ieee80211_sub_if_data *sdata, goto unlock_sta; spin_lock(&cache->walk_lock); + if ((READ_ONCE(mpath->flags) & MESH_PATH_DELETED) || + (mppath && (READ_ONCE(mppath->flags) & MESH_PATH_DELETED))) { + kfree(entry); + goto unlock_cache; + } + prev = rhashtable_lookup_get_insert_fast(&cache->rht, &entry->rhash, fast_tx_rht_params); @@ -808,7 +814,8 @@ static void mesh_path_free_rcu(struct mesh_table *tbl, struct ieee80211_sub_if_data *sdata = mpath->sdata; spin_lock_bh(&mpath->state_lock); - mpath->flags |= MESH_PATH_RESOLVING | MESH_PATH_DELETED; + WRITE_ONCE(mpath->flags, + mpath->flags | MESH_PATH_RESOLVING | MESH_PATH_DELETED); mesh_gate_del(tbl, mpath); spin_unlock_bh(&mpath->state_lock); timer_shutdown_sync(&mpath->timer); @@ -822,6 +829,9 @@ static void __mesh_path_del(struct mesh_table *tbl, struct mesh_path *mpath) { hlist_del_rcu(&mpath->walk_list); rhashtable_remove_fast(&tbl->rhead, &mpath->rhash, mesh_rht_params); + spin_lock_bh(&mpath->state_lock); + WRITE_ONCE(mpath->flags, mpath->flags | MESH_PATH_DELETED); + spin_unlock_bh(&mpath->state_lock); if (tbl == &mpath->sdata->u.mesh.mpp_paths) mesh_fast_tx_flush_addr(mpath->sdata, mpath->dst); else From db7b86bd97b380ece8fa39cfdd7502a9fd35e56f Mon Sep 17 00:00:00 2001 From: Heyang Tan Date: Fri, 18 Sep 2026 16:33:49 +0800 Subject: [PATCH 150/365] wifi: p54: validate firmware record lengths The p54 firmware parser accesses component ID and component version records without checking that their lengths cover the fields being read. A malformed record can therefore make the parser read past the record. Validate the component ID and component version record lengths before accessing their data. The descriptor record is also checked against the shortest descriptor length found in the supported firmware images. Reject descriptor values that underflow the RX offset or produce an empty RX range before updating the driver state. Assisted-by: LLM Codex Signed-off-by: Heyang Tan Tested-by: Christian Lamparter Link: https://patch.msgid.link/20260918083349.891-1-thy15333007817@163.com Signed-off-by: Johannes Berg --- drivers/net/wireless/intersil/p54/fwio.c | 42 +++++++++++++++++++++--- 1 file changed, 37 insertions(+), 5 deletions(-) diff --git a/drivers/net/wireless/intersil/p54/fwio.c b/drivers/net/wireless/intersil/p54/fwio.c index a3d9053f043c..f028a7bfa734 100644 --- a/drivers/net/wireless/intersil/p54/fwio.c +++ b/drivers/net/wireless/intersil/p54/fwio.c @@ -52,6 +52,13 @@ int p54_parse_firmware(struct ieee80211_hw *dev, const struct firmware *fw) u32 code = le32_to_cpu(bootrec->code); switch (code) { case BR_CODE_COMPONENT_ID: + if (len < sizeof(struct bootrec_comp_id) / + sizeof(*bootrec->data)) { + wiphy_err(priv->hw->wiphy, + "firmware component ID is too short\n"); + return -EINVAL; + } + priv->fw_interface = be32_to_cpup((__be32 *) bootrec->data); switch (priv->fw_interface) { @@ -72,16 +79,41 @@ int p54_parse_firmware(struct ieee80211_hw *dev, const struct firmware *fw) } break; case BR_CODE_COMPONENT_VERSION: + if (len < DIV_ROUND_UP(sizeof(struct bootrec_comp_ver), + sizeof(*bootrec->data))) { + wiphy_err(priv->hw->wiphy, + "firmware component version is too short\n"); + return -EINVAL; + } + /* 24 bytes should be enough for all firmwares */ - if (strnlen((unsigned char *) bootrec->data, 24) < 24) - fw_version = (unsigned char *) bootrec->data; + if (strnlen((unsigned char *)bootrec->data, + sizeof(struct bootrec_comp_ver)) < + sizeof(struct bootrec_comp_ver)) + fw_version = (unsigned char *)bootrec->data; break; case BR_CODE_DESCR: { struct bootrec_desc *desc = (struct bootrec_desc *)bootrec->data; - priv->rx_start = le32_to_cpu(desc->rx_start); - /* FIXME add sanity checking */ - priv->rx_end = le32_to_cpu(desc->rx_end) - 0x3500; + u32 rx_start, rx_end; + + /* 0xa is the shortest descriptor in supported firmware. */ + if (len < 0xa) { + wiphy_err(priv->hw->wiphy, + "firmware descriptor is too short\n"); + return -EINVAL; + } + + rx_start = le32_to_cpu(desc->rx_start); + rx_end = le32_to_cpu(desc->rx_end); + if (rx_end < 0x3500 || rx_end - 0x3500 <= rx_start) { + wiphy_err(priv->hw->wiphy, + "firmware descriptor has invalid RX range\n"); + return -EINVAL; + } + + priv->rx_start = rx_start; + priv->rx_end = rx_end - 0x3500; priv->headroom = desc->headroom; priv->tailroom = desc->tailroom; priv->privacy_caps = desc->privacy_caps; From 15fb9e3cea55fbc378261ce35a32b0f8828f0000 Mon Sep 17 00:00:00 2001 From: Yuqi Xu Date: Sat, 19 Sep 2026 16:46:20 +0800 Subject: [PATCH 151/365] wifi: mac80211: minstrel_ht: validate fixed rate index The fixed_rate_idx debugfs file is a plain u32 attribute, so any value can be written to it. minstrel_ht_update_stats() then copies the value into mi->max_tp_rate[] and mi->max_prob_rate, and minstrel_ht_set_rate() uses MI_RATE_GROUP()/MI_RATE_IDX() to index minstrel_mcs_groups[] (42 entries) and mi->groups[].rates[] (10 entries) with it. Writing e.g. 65535 selects group 4095 and rate index 15, far outside both arrays. The out-of-bounds entries are dereferenced and updated as struct minstrel_rate_stats (retry counts, etc.), so the invalid index corrupts adjacent kernel memory. With CONFIG_UBSAN_BOUNDS the access is reported as an array-index-out-of-bounds in minstrel_ht_set_rate(). Only a valid group/rate pair, or U32_MAX to disable fixed rate processing, can be used safely, so reject any other value when the debugfs file is written. The file is only reachable through a root-only debugfs mount, so this is not a privilege boundary. The out-of-bounds access is still a bug that must not be reachable through a writable debugfs attribute. Fixes: 24f7580e852b ("minstrel_ht: fixed rate mode through debugfs") Cc: stable@vger.kernel.org Reported-by: Vega Assisted-by: LLM Signed-off-by: Yuqi Xu Reviewed-by: Ren Wei Link: https://patch.msgid.link/288e6f6d30dc82ff40da502755e6666982f6b735.1789798000.git.xuyuqiabc@gmail.com Signed-off-by: Johannes Berg --- net/mac80211/rc80211_minstrel_ht.c | 30 ++++++++++++++++++++++++++++-- 1 file changed, 28 insertions(+), 2 deletions(-) diff --git a/net/mac80211/rc80211_minstrel_ht.c b/net/mac80211/rc80211_minstrel_ht.c index b73ef3adfcc5..c26221287515 100644 --- a/net/mac80211/rc80211_minstrel_ht.c +++ b/net/mac80211/rc80211_minstrel_ht.c @@ -7,6 +7,7 @@ #include #include #include +#include #include #include #include @@ -1947,14 +1948,39 @@ minstrel_ht_alloc(struct ieee80211_hw *hw) } #ifdef CONFIG_MAC80211_DEBUGFS +static int minstrel_ht_fixed_rate_idx_get(void *data, u64 *val) +{ + *val = *(u32 *)data; + return 0; +} + +static int minstrel_ht_fixed_rate_idx_set(void *data, u64 val) +{ + u32 idx = val; + + /* U32_MAX is the default and keeps fixed rate processing disabled */ + if (val != U32_MAX && + (val > U16_MAX || + MI_RATE_GROUP(idx) >= ARRAY_SIZE(minstrel_mcs_groups) || + MI_RATE_IDX(idx) >= MCS_GROUP_RATES)) + return -EINVAL; + + *(u32 *)data = idx; + return 0; +} + +DEFINE_DEBUGFS_ATTRIBUTE(minstrel_ht_fixed_rate_idx_fops, + minstrel_ht_fixed_rate_idx_get, + minstrel_ht_fixed_rate_idx_set, "%llu\n"); + static void minstrel_ht_add_debugfs(struct ieee80211_hw *hw, void *priv, struct dentry *debugfsdir) { struct minstrel_priv *mp = priv; mp->fixed_rate_idx = (u32) -1; - debugfs_create_u32("fixed_rate_idx", S_IRUGO | S_IWUGO, debugfsdir, - &mp->fixed_rate_idx); + debugfs_create_file("fixed_rate_idx", S_IRUGO | S_IWUGO, debugfsdir, + &mp->fixed_rate_idx, &minstrel_ht_fixed_rate_idx_fops); } #endif From 5bfd4b0b40b79781d900cb2f7da0685070f53c67 Mon Sep 17 00:00:00 2001 From: Yuqi Xu Date: Sat, 19 Sep 2026 16:48:12 +0800 Subject: [PATCH 152/365] wifi: mac80211: handle empty FILS association request payload fils_encrypt_assoc_req() accepts a (Re)Association Request frame whose last element is the FILS Session element. crypt_len is then zero and aes_siv_encrypt() is called with an empty plaintext. kmemdup() returns ZERO_SIZE_PTR for a zero-length allocation, so the unconditional sg_init_one() puts an invalid page into the scatterlist. With CONFIG_DEBUG_SG this triggers a kernel BUG in sg_set_buf(); without the debug check the bogus page is handed to the crypto backend. AES-SIV is defined for empty plaintext and the S2V result is then the complete output, so handle this case before the CTR step, which has nothing to encrypt. Fixes: 39404feee691 ("mac80211: FILS AEAD protection for station mode association frames") Cc: stable@vger.kernel.org Reported-by: Vega Assisted-by: LLM Signed-off-by: Yuqi Xu Reviewed-by: Ren Wei Link: https://patch.msgid.link/325c048982d637a3af80d74ac5b36e9369b93f1e.1789751126.git.xuyuqiabc@gmail.com Signed-off-by: Johannes Berg --- net/mac80211/fils_aead.c | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/net/mac80211/fils_aead.c b/net/mac80211/fils_aead.c index 293590976489..9c4f00d61981 100644 --- a/net/mac80211/fils_aead.c +++ b/net/mac80211/fils_aead.c @@ -89,6 +89,14 @@ static int aes_siv_encrypt(const u8 *key, size_t key_len, if (res) return res; + /* With an empty plaintext there is no ciphertext to produce and + * the S2V result is the complete output. + */ + if (!plain_len) { + memcpy(out, v, AES_BLOCK_SIZE); + return 0; + } + /* Use a temporary buffer of the plaintext to handle need for * overwriting this during AES-CTR. */ From 9eda41e32263806a66835e275efdcc0c5f46bbc4 Mon Sep 17 00:00:00 2001 From: Stanislaw Gruszka Date: Sat, 19 Sep 2026 11:24:54 +0200 Subject: [PATCH 153/365] wifi: cfg80211: fix RTS threshold setting for single-radio PHY Due to an incorrect old_radio_rts_threshold check, we do not set rdev->wiphy.rts_threshold for a single-radio PHY, and the threshold value is not propagated to lower layers. Cc: stable@vger.kernel.org Fixes: 264637941cf4 ("wifi: cfg80211: Add Support to Set RTS Threshold for each Radio") Signed-off-by: Stanislaw Gruszka Link: https://patch.msgid.link/20260919092454.50266-1-stf_xl@wp.pl Signed-off-by: Johannes Berg --- net/wireless/nl80211.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-) diff --git a/net/wireless/nl80211.c b/net/wireless/nl80211.c index 9fafd7673d85..ddce61ec016a 100644 --- a/net/wireless/nl80211.c +++ b/net/wireless/nl80211.c @@ -4563,12 +4563,13 @@ static int nl80211_set_wiphy(struct sk_buff *skb, struct genl_info *info) rdev->wiphy.retry_long = retry_long; if (changed & WIPHY_PARAM_FRAG_THRESHOLD) rdev->wiphy.frag_threshold = frag_threshold; - if ((changed & WIPHY_PARAM_RTS_THRESHOLD) && - old_radio_rts_threshold) { + if (changed & WIPHY_PARAM_RTS_THRESHOLD) { rdev->wiphy.rts_threshold = rts_threshold; - for (i = 0 ; i < rdev->wiphy.n_radio; i++) - rdev->wiphy.radio_cfg[i].rts_threshold = - rdev->wiphy.rts_threshold; + if (old_radio_rts_threshold) { + for (i = 0; i < rdev->wiphy.n_radio; i++) + rdev->wiphy.radio_cfg[i].rts_threshold = + rdev->wiphy.rts_threshold; + } } if (changed & WIPHY_PARAM_COVERAGE_CLASS) rdev->wiphy.coverage_class = coverage_class; From 7e0ad05ebc72fffba2bdf2400400902d94ed7272 Mon Sep 17 00:00:00 2001 From: Nicolas Escande Date: Wed, 12 Aug 2026 16:58:05 +0200 Subject: [PATCH 154/365] wifi: ath11k: reset ar->num_stations on hardware start When (multiple) hw restart occurs, we end up in a situation where we cannot accept new stations / mesh peers. This seems to be the 'num_stations' in 'struct ath11k' that did not get reset properly in this case. In ath11k_mac_op_start(), it was indeed the only accounting var that did not get reset so let's clear it too. This avoids problems like those: [1248828.088023] WARNING: CPU: 0 PID: 12281 at net/mac80211/util.c:1682 ieee80211_reconfig_stations+0x98/0xc0 [1248828.088115] Call trace: [1248828.088116] ieee80211_reconfig_stations+0x98/0xc0 (P) [1248828.088119] ieee80211_reconfig+0x774/0x1438 [1248828.088123] ieee80211_restart_work+0x10c/0x178 [1248828.088127] process_one_work+0x178/0x3c0 [1248828.088130] worker_thread+0x280/0x440 [1248828.088133] kthread+0xe0/0xf8 [1248828.088137] ret_from_fork+0x10/0x20 [1248828.088141] ---[ end trace 0000000000000000 ]--- [1248836.686604] ath11k_pci 0000:04:00.0: refusing to associate station: too many connected already (128) [1248836.686609] ath11k_pci 0000:04:00.0: Failed to add station: xx:xx:xx:xx:xx:xx for VDEV: 1 Tested-on: QCN9074 hw1.0 PCI WLAN.HK.2.9.0.1-01977-QCAHKSWPL_SILICONZ-1 Fixes: d5c65159f289 ("ath11k: driver for Qualcomm IEEE 802.11ax devices") Signed-off-by: Nicolas Escande Reviewed-by: Rameshkumar Sundaram Reviewed-by: Baochen Qiang Link: https://patch.msgid.link/20260812-num-station-v3-1-871d20e043f6@gmail.com Signed-off-by: Jeff Johnson --- drivers/net/wireless/ath/ath11k/mac.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/net/wireless/ath/ath11k/mac.c b/drivers/net/wireless/ath/ath11k/mac.c index ae91b57c8422..c1fa42edd363 100644 --- a/drivers/net/wireless/ath/ath11k/mac.c +++ b/drivers/net/wireless/ath/ath11k/mac.c @@ -6716,6 +6716,7 @@ static int ath11k_mac_op_start(struct ieee80211_hw *hw) ar->num_started_vdevs = 0; ar->num_created_vdevs = 0; ar->num_peers = 0; + ar->num_stations = 0; ar->allocated_vdev_map = 0; /* Configure monitor status ring with default rx_filter to get rx status From 1af8dc9ca726cf53767f9c0661171396f95c146a Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Toke=20H=C3=B8iland-J=C3=B8rgensen?= Date: Tue, 8 Sep 2026 19:05:18 +0200 Subject: [PATCH 155/365] wifi: ath9k: Clean up device initialisation guards MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The ath9k driver has a couple of guards against the RX path running until device initialisation has completed. However, they are problematic for several reasons: - They use smp_wmb() paired with a data_race() annotation, instead of the idiomatic smp_store_release()/smp_load_acquire() pair. - The guard in ath9k_wmi_event_tasklet() is weirdly late in the function. - There are two of them when there's really no reason the rx init gets to run before the whole initialisation is complete. Consolidate the guards on the single ath9k_htc_priv->initialized variable and use smp_store_release()/smp_load_acquire() to access it. Move the store into ath9k_init_device() to make sure it's set in all paths in the driver. Fixes: 24355fcb0d4c ("wifi: ath9k: delay all of ath9k_wmi_event_tasklet() until init is complete") Fixes: b0ec7e55fce6 ("ath9k_htc: fix NULL pointer dereference at ath9k_htc_rxep()") Signed-off-by: Toke Høiland-Jørgensen Link: https://patch.msgid.link/20260908170520.112946-1-toke@redhat.com Signed-off-by: Jeff Johnson --- drivers/net/wireless/ath/ath9k/htc.h | 1 - drivers/net/wireless/ath/ath9k/htc_drv_init.c | 7 +++---- drivers/net/wireless/ath/ath9k/htc_drv_txrx.c | 8 ++------ drivers/net/wireless/ath/ath9k/wmi.c | 12 ++++++------ 4 files changed, 11 insertions(+), 17 deletions(-) diff --git a/drivers/net/wireless/ath/ath9k/htc.h b/drivers/net/wireless/ath/ath9k/htc.h index 6c33e898b300..4d2b75d0212e 100644 --- a/drivers/net/wireless/ath/ath9k/htc.h +++ b/drivers/net/wireless/ath/ath9k/htc.h @@ -281,7 +281,6 @@ struct ath9k_htc_rxbuf { struct ath9k_htc_rx { struct list_head rxbuf; spinlock_t rxbuflock; - bool initialized; }; #define ATH9K_HTC_TX_CLEANUP_INTERVAL 50 /* ms */ diff --git a/drivers/net/wireless/ath/ath9k/htc_drv_init.c b/drivers/net/wireless/ath/ath9k/htc_drv_init.c index 6de78ae85726..911eaa4f776a 100644 --- a/drivers/net/wireless/ath/ath9k/htc_drv_init.c +++ b/drivers/net/wireless/ath/ath9k/htc_drv_init.c @@ -910,6 +910,9 @@ static int ath9k_init_device(struct ath9k_htc_priv *priv, ath9k_init_leds(priv); ath9k_start_rfkill_poll(priv); + /* signal completion to ath9k_htc_rxep() and ath9k_wmi_event_tasklet() */ + smp_store_release(&priv->initialized, true); + return 0; err_world: @@ -966,10 +969,6 @@ int ath9k_htc_probe_device(struct htc_target *htc_handle, struct device *dev, htc_handle->drv_priv = priv; - /* Allow ath9k_wmi_event_tasklet() to operate. */ - smp_wmb(); - priv->initialized = true; - return 0; err_init: diff --git a/drivers/net/wireless/ath/ath9k/htc_drv_txrx.c b/drivers/net/wireless/ath/ath9k/htc_drv_txrx.c index bed7ea2425a0..95f48c7dca15 100644 --- a/drivers/net/wireless/ath/ath9k/htc_drv_txrx.c +++ b/drivers/net/wireless/ath/ath9k/htc_drv_txrx.c @@ -1141,8 +1141,8 @@ void ath9k_htc_rxep(void *drv_priv, struct sk_buff *skb, struct ath9k_htc_rxbuf *rxbuf = NULL, *tmp_buf = NULL; unsigned long flags; - /* Check if ath9k_rx_init() completed. */ - if (!data_race(priv->rx.initialized)) + /* Check if ath9k_init_device() completed. */ + if (!smp_load_acquire(&priv->initialized)) goto err; spin_lock_irqsave(&priv->rx.rxbuflock, flags); @@ -1200,10 +1200,6 @@ int ath9k_rx_init(struct ath9k_htc_priv *priv) list_add_tail(&rxbuf->list, &priv->rx.rxbuf); } - /* Allow ath9k_htc_rxep() to operate. */ - smp_wmb(); - priv->rx.initialized = true; - return 0; err: diff --git a/drivers/net/wireless/ath/ath9k/wmi.c b/drivers/net/wireless/ath/ath9k/wmi.c index 284e8c13b043..44e91815a159 100644 --- a/drivers/net/wireless/ath/ath9k/wmi.c +++ b/drivers/net/wireless/ath/ath9k/wmi.c @@ -146,6 +146,12 @@ void ath9k_wmi_event_tasklet(struct tasklet_struct *t) unsigned long flags; u16 cmd_id; + /* Check if ath9k_init_device() completed. */ + if (!smp_load_acquire(&priv->initialized)) { + tasklet_schedule(t); + return; + } + do { spin_lock_irqsave(&wmi->wmi_lock, flags); skb = __skb_dequeue(&wmi->wmi_event_queue); @@ -155,12 +161,6 @@ void ath9k_wmi_event_tasklet(struct tasklet_struct *t) } spin_unlock_irqrestore(&wmi->wmi_lock, flags); - /* Check if ath9k_htc_probe_device() completed. */ - if (!data_race(priv->initialized)) { - kfree_skb(skb); - continue; - } - hdr = (struct wmi_cmd_hdr *) skb->data; cmd_id = be16_to_cpu(hdr->command_id); wmi_event = skb_pull(skb, sizeof(struct wmi_cmd_hdr)); From 93ce8ea70451e800601bc9444aa33043de15d67b Mon Sep 17 00:00:00 2001 From: Pengpeng Hou Date: Fri, 14 Aug 2026 15:58:37 +0800 Subject: [PATCH 156/365] wifi: ath9k: reject short WMI command responses ath9k_wmi_rsp_callback() removes the validated WMI header and copies the number of bytes expected by the waiting command into its response buffer. A device response shorter than that expectation can therefore be read beyond the skb payload. Record -EMSGSIZE for a short response, complete the waiter, and return the stored status from the command path. This also prevents callers from consuming stale response bytes as a successful reply. Fixes: fb9987d0f748 ("ath9k_htc: Support for AR9271 chipset.") Assisted-by: LLM Signed-off-by: Pengpeng Hou Link: https://patch.msgid.link/20260814075837.20005-1-pengpeng@iscas.ac.cn Signed-off-by: Jeff Johnson --- drivers/net/wireless/ath/ath9k/wmi.c | 16 ++++++++++++++-- drivers/net/wireless/ath/ath9k/wmi.h | 1 + 2 files changed, 15 insertions(+), 2 deletions(-) diff --git a/drivers/net/wireless/ath/ath9k/wmi.c b/drivers/net/wireless/ath/ath9k/wmi.c index 44e91815a159..4a08babb6d9d 100644 --- a/drivers/net/wireless/ath/ath9k/wmi.c +++ b/drivers/net/wireless/ath/ath9k/wmi.c @@ -206,9 +206,16 @@ static void ath9k_wmi_rsp_callback(struct wmi *wmi, struct sk_buff *skb) { skb_pull(skb, sizeof(struct wmi_cmd_hdr)); - if (wmi->cmd_rsp_buf != NULL && wmi->cmd_rsp_len != 0) + if (wmi->cmd_rsp_buf && wmi->cmd_rsp_len) { + if (skb->len < wmi->cmd_rsp_len) { + wmi->cmd_rsp_status = -EMSGSIZE; + goto complete; + } memcpy(wmi->cmd_rsp_buf, skb->data, wmi->cmd_rsp_len); + } + wmi->cmd_rsp_status = 0; +complete: complete(&wmi->cmd_wait); } @@ -300,6 +307,7 @@ static int ath9k_wmi_cmd_issue(struct wmi *wmi, /* record the rsp buffer and length */ wmi->cmd_rsp_buf = rsp_buf; wmi->cmd_rsp_len = rsp_len; + wmi->cmd_rsp_status = 0; wmi->last_seq_id = wmi->tx_seq_id; spin_unlock_irqrestore(&wmi->wmi_lock, flags); @@ -356,9 +364,13 @@ int ath9k_wmi_cmd(struct wmi *wmi, enum wmi_cmd_id cmd_id, return -ETIMEDOUT; } + spin_lock_irqsave(&wmi->wmi_lock, flags); + ret = wmi->cmd_rsp_status; + spin_unlock_irqrestore(&wmi->wmi_lock, flags); + mutex_unlock(&wmi->op_mutex); - return 0; + return ret; out: ath_dbg(common, WMI, "WMI failure for: %s\n", wmi_cmd_to_name(cmd_id)); diff --git a/drivers/net/wireless/ath/ath9k/wmi.h b/drivers/net/wireless/ath/ath9k/wmi.h index 5c3b710b8f31..0bb9d31f1a1c 100644 --- a/drivers/net/wireless/ath/ath9k/wmi.h +++ b/drivers/net/wireless/ath/ath9k/wmi.h @@ -157,6 +157,7 @@ struct wmi { u16 tx_seq_id; u8 *cmd_rsp_buf; u32 cmd_rsp_len; + int cmd_rsp_status; bool stopped; struct list_head pending_tx_events; From aeaca27c1db3106967ffbb2b517d98c835f6d500 Mon Sep 17 00:00:00 2001 From: Georgios Karantzas Date: Fri, 11 Sep 2026 03:11:43 +0300 Subject: [PATCH 157/365] wifi: ath9k_htc: bound TX aggregation to MAX_TX_BUF_SIZE MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit __hif_usb_tx() dequeues up to MAX_TX_AGGR_NUM (20) frames into a single tx_buf of MAX_TX_BUF_SIZE (32768) bytes, limiting the batch by record count but never by cumulative byte length. With large frames (MTU 2304), 20 aggregated frames of 2292 bytes each exceed the allocation (20 * 2296 = 45920 bytes), so the memcpy() in the loop writes up to 13152 bytes past tx_buf->buf before usb_submit_urb(). Peek the queue head and stop before copying any record that would cross MAX_TX_BUF_SIZE, then dispatch the current batch. Leftover skbs remain queued and are drained on the next URB completion. The byte bound changes the loop's exit semantics: it can now exit before i == tx_skb_cnt - 1. Stock only finalized tx_buf->len on that last index (len += offset), so an early break would submit a URB holding only the last record's length while every dequeued skb is freed on completion, silently dropping frames. Make tx_buf->len a running total and advance tx_buf->offset per record instead; the stride round_up(nskb->len + 4, 4) is identical to the stock stride when the loop runs to completion. Tested on hardware with an MTU 2304 flood: the loop stops at record 15 (len = 32144, offset = 32144, within 32768), no oversized URB is submitted, and MTU 1500 pings pass 20/20. Fixes: fb9987d0f748 ("ath9k_htc: Support for AR9271 chipset.") Signed-off-by: Georgios Karantzas Acked-by: Toke Høiland-Jørgensen Link: https://patch.msgid.link/20260911001143.4507-1-gck.kara@gmail.com Signed-off-by: Jeff Johnson --- drivers/net/wireless/ath/ath9k/hif_usb.c | 21 ++++++++++----------- 1 file changed, 10 insertions(+), 11 deletions(-) diff --git a/drivers/net/wireless/ath/ath9k/hif_usb.c b/drivers/net/wireless/ath/ath9k/hif_usb.c index d3491ff08e6e..7d2f292e20c5 100644 --- a/drivers/net/wireless/ath/ath9k/hif_usb.c +++ b/drivers/net/wireless/ath/ath9k/hif_usb.c @@ -14,6 +14,7 @@ * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. */ +#include #include #include "htc.h" @@ -328,11 +329,14 @@ static int __hif_usb_tx(struct hif_device_usb *hif_dev) tx_skb_cnt = min_t(u16, hif_dev->tx.tx_skb_cnt, MAX_TX_AGGR_NUM); for (i = 0; i < tx_skb_cnt; i++) { + nskb = skb_peek(&hif_dev->tx.tx_skb_queue); + if (!nskb) + break; + + if (tx_buf->offset + nskb->len + 4 > MAX_TX_BUF_SIZE) + break; + nskb = __skb_dequeue(&hif_dev->tx.tx_skb_queue); - - /* Should never be NULL */ - BUG_ON(!nskb); - hif_dev->tx.tx_skb_cnt--; buf = tx_buf->buf; @@ -342,13 +346,8 @@ static int __hif_usb_tx(struct hif_device_usb *hif_dev) *hdr++ = cpu_to_le16(ATH_USB_TX_STREAM_MODE_TAG); buf += 4; memcpy(buf, nskb->data, nskb->len); - tx_buf->len = nskb->len + 4; - - if (i < (tx_skb_cnt - 1)) - tx_buf->offset += (((tx_buf->len - 1) / 4) + 1) * 4; - - if (i == (tx_skb_cnt - 1)) - tx_buf->len += tx_buf->offset; + tx_buf->len = tx_buf->offset + nskb->len + 4; + tx_buf->offset += round_up(nskb->len + 4, 4); __skb_queue_tail(&tx_buf->skb_queue, nskb); TX_STAT_INC(hif_dev, skb_queued); From 0ee5c5d804d592a8328a425b9274487d393d3a05 Mon Sep 17 00:00:00 2001 From: Liu Dalin Date: Thu, 27 Aug 2026 09:54:44 +0800 Subject: [PATCH 158/365] rtc: dev: zero-initialize struct rtc_wkalrm to prevent information leak The struct rtc_wkalrm has padding holes after the 'pending' member. When returned to userspace via the RTC_WKALRM_RD ioctl, uninitialized kernel stack data residing in these holes may be leaked. Zero-initialize the struct at declaration to eliminate this information leak risk. Fixes smatch warnings: - drivers/rtc/dev.c:392 rtc_dev_ioctl() warn: check that 'alarm' doesn't leak information (struct has a hole after 'pending') Fixes: 36e14f5fdfdf ("rtc: rename core files") Assisted-by: smatch:2.0 [static analysis] Signed-off-by: Liu Dalin Link: https://patch.msgid.link/B28B7DE18F3A9447+20260827015444.3314227-1-liudalin@kylinsec.com.cn Signed-off-by: Alexandre Belloni --- drivers/rtc/dev.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/rtc/dev.c b/drivers/rtc/dev.c index 8ba7c25d2565..be173c9511fa 100644 --- a/drivers/rtc/dev.c +++ b/drivers/rtc/dev.c @@ -216,7 +216,7 @@ static long rtc_dev_ioctl(struct file *file, struct rtc_device *rtc = file->private_data; const struct rtc_class_ops *ops = rtc->ops; struct rtc_time tm; - struct rtc_wkalrm alarm; + struct rtc_wkalrm alarm = { 0 }; struct rtc_param param; void __user *uarg = (void __user *)arg; From b430d1f6d80c005b486f028910e1ac3c91812710 Mon Sep 17 00:00:00 2001 From: "Johnny.Fan" Date: Mon, 31 Aug 2026 22:00:14 +0800 Subject: [PATCH 159/365] rtc: efi: restore alarm support with runtime capability probe Restore the EFI RTC alarm functionality (read_alarm/set_alarm and procfs output) that was removed by commit 18a3510bc87d ("rtc: efi: Remove wakeup functionality"). The removal was motivated by the observation that many EFI platforms lack RTC wakeup support while also omitting the RT_PROP table, causing the wakeup capability to be incorrectly advertised. While that is true, removing the feature entirely also breaks platforms where the wakeup runtime service actually works. On CIX SoCs, the RTC is an I2C device owned by firmware, and in the ACPI boot configuration the I2C bus it sits on is not exposed to the OS, so the EFI runtime services are the only way Linux can access the RTC at all. Both Linux and Windows use this same interface by design, so the RTC state is consistent across the operating systems installed on the machine, and Windows relies on the EFI wakeup services for its own RTC alarm support. With the alarm support removed from rtc-efi, Linux loses the equivalent capability entirely on these platforms: rtcwake fails for both suspend-to-RAM and power-off wakeup, as there is no other path to program the RTC alarm. This is a real product requirement that CIX customers have explicitly requested, not a hypothetical use case. To avoid exposing a broken alarm on firmware that does not implement the wakeup runtime services, probe the capability at ->probe() time by actually calling GetWakeupTime() -- if it fails, clear RTC_FEATURE_ALARM and the device behaves as before the removal. Also set RTC_FEATURE_ALARM_WAKEUP_ONLY since EFI wakeup time is a wakeup-only alarm, not a periodic/update interrupt source. Acked-by: Ard Biesheuvel Reviewed-by: Feng Tang Reviewed-by: Fugang Duan Signed-off-by: Johnny.Fan Link: https://patch.msgid.link/20260831140014.839222-1-johnny.fan@cixtech.com Signed-off-by: Alexandre Belloni --- drivers/rtc/rtc-efi.c | 80 +++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 78 insertions(+), 2 deletions(-) diff --git a/drivers/rtc/rtc-efi.c b/drivers/rtc/rtc-efi.c index b4f44999ef0f..a8c18ebc5599 100644 --- a/drivers/rtc/rtc-efi.c +++ b/drivers/rtc/rtc-efi.c @@ -52,6 +52,7 @@ compute_wday(efi_time_t *eft, int yday) static void convert_to_efi_time(struct rtc_time *wtime, efi_time_t *eft) { + memset(eft, 0, sizeof(*eft)); eft->year = wtime->tm_year + 1900; eft->month = wtime->tm_mon + 1; eft->day = wtime->tm_mday; @@ -112,6 +113,35 @@ convert_from_efi_time(efi_time_t *eft, struct rtc_time *wtime) return true; } +static int efi_read_alarm(struct device *dev, struct rtc_wkalrm *wkalrm) +{ + efi_time_t eft; + efi_status_t status; + + memset(&eft, 0, sizeof(eft)); + status = efi.get_wakeup_time((efi_bool_t *)&wkalrm->enabled, + (efi_bool_t *)&wkalrm->pending, &eft); + if (status != EFI_SUCCESS) + return -EINVAL; + + if (!convert_from_efi_time(&eft, &wkalrm->time)) + return -EIO; + + return rtc_valid_tm(&wkalrm->time); +} + +static int efi_set_alarm(struct device *dev, struct rtc_wkalrm *wkalrm) +{ + efi_time_t eft; + efi_status_t status; + + convert_to_efi_time(&wkalrm->time, &eft); + + status = efi.set_wakeup_time((efi_bool_t)!!wkalrm->enabled, &eft); + + return status == EFI_SUCCESS ? 0 : -EINVAL; +} + static int efi_read_time(struct device *dev, struct rtc_time *tm) { efi_status_t status; @@ -146,13 +176,21 @@ static int efi_set_time(struct device *dev, struct rtc_time *tm) static int efi_procfs(struct device *dev, struct seq_file *seq) { - efi_time_t eft; + efi_time_t eft, alm; efi_time_cap_t cap; + efi_bool_t enabled, pending; + struct rtc_device *rtc = dev_get_drvdata(dev); memset(&eft, 0, sizeof(eft)); + memset(&alm, 0, sizeof(alm)); memset(&cap, 0, sizeof(cap)); efi.get_time(&eft, &cap); + if (test_bit(RTC_FEATURE_ALARM, rtc->features) && + efi.get_wakeup_time(&enabled, &pending, &alm) != EFI_SUCCESS) { + enabled = 0; + pending = 0; + } seq_printf(seq, "Time\t\t: %u:%u:%u.%09u\n" @@ -168,6 +206,25 @@ static int efi_procfs(struct device *dev, struct seq_file *seq) /* XXX fixme: convert to string? */ seq_printf(seq, "Timezone\t: %u\n", eft.timezone); + if (test_bit(RTC_FEATURE_ALARM, rtc->features)) { + seq_printf(seq, + "Alarm Time\t: %u:%u:%u.%09u\n" + "Alarm Date\t: %u-%u-%u\n" + "Alarm Daylight\t: %u\n" + "Enabled\t\t: %s\n" + "Pending\t\t: %s\n", + alm.hour, alm.minute, alm.second, alm.nanosecond, + alm.year, alm.month, alm.day, + alm.daylight, + enabled == 1 ? "yes" : "no", + pending == 1 ? "yes" : "no"); + + if (alm.timezone == EFI_UNSPECIFIED_TIMEZONE) + seq_puts(seq, "Alarm Timezone\t: unspecified\n"); + else + seq_printf(seq, "Alarm Timezone\t: %d\n", alm.timezone); + } + /* * now prints the capabilities */ @@ -183,6 +240,8 @@ static int efi_procfs(struct device *dev, struct seq_file *seq) static const struct rtc_class_ops efi_rtc_ops = { .read_time = efi_read_time, .set_time = efi_set_time, + .read_alarm = efi_read_alarm, + .set_alarm = efi_set_alarm, .proc = efi_procfs, }; @@ -191,6 +250,7 @@ static int __init efi_rtc_probe(struct platform_device *dev) struct rtc_device *rtc; efi_time_t eft; efi_time_cap_t cap; + efi_bool_t enabled, pending; /* First check if the RTC is usable */ if (efi.get_time(&eft, &cap) != EFI_SUCCESS) @@ -203,7 +263,23 @@ static int __init efi_rtc_probe(struct platform_device *dev) platform_set_drvdata(dev, rtc); rtc->ops = &efi_rtc_ops; - clear_bit(RTC_FEATURE_ALARM, rtc->features); + clear_bit(RTC_FEATURE_UPDATE_INTERRUPT, rtc->features); + + /* + * The EFI_RT_SUPPORTED_WAKEUP_SERVICES bit defaults to enabled + * and only gets cleared when the RT_PROP table explicitly says + * wakeup is unsupported. Many platforms lack an RT_PROP table + * even though they don't implement the wakeup runtime service, + * so probe by actually calling GetWakeupTime() to avoid exposing + * a broken alarm to userspace. + */ + if (efi_rt_services_supported(EFI_RT_SUPPORTED_WAKEUP_SERVICES) && + efi.get_wakeup_time(&enabled, &pending, &eft) == EFI_SUCCESS) { + set_bit(RTC_FEATURE_ALARM, rtc->features); + set_bit(RTC_FEATURE_ALARM_WAKEUP_ONLY, rtc->features); + } else { + clear_bit(RTC_FEATURE_ALARM, rtc->features); + } device_init_wakeup(&dev->dev, true); From 6a4117eee82dd85f11bf898bf66026764011eeb6 Mon Sep 17 00:00:00 2001 From: Aamir Ahmed Date: Sat, 5 Sep 2026 19:38:08 +0100 Subject: [PATCH 160/365] rtc: ac100: Assign .num before accessing .hws Commit f316cdff8d67 ("clk: Annotate struct clk_hw_onecell_data with __counted_by") annotated the hws member of 'struct clk_hw_onecell_data' with __counted_by, which informs the bounds sanitizer (UBSAN_BOUNDS) about the number of elements in .hws[], so that it can warn when .hws[] is accessed out of bounds. As noted in that change, the __counted_by member must be initialized with the number of elements before the first array access happens, otherwise there will be a warning from each access prior to the initialization because the number of elements is zero. This occurs in ac100_rtc_register_clks() due to .num being assigned only after every clkout clock has been stored in .hws[]. With CONFIG_UBSAN_BOUNDS and a compiler that implements __counted_by (GCC 15.1+ or Clang 20.1+), this triggers an array-index-out-of-bounds report during probe, and with CONFIG_UBSAN_TRAP the first store traps. Initialize .num with AC100_CLKOUT_NUM, the number of elements .hws[] was allocated with, right after the allocation. That is the value the loop counter ends up at on the success path anyway, so the provider's behaviour is unchanged. Cc: stable@vger.kernel.org Fixes: f316cdff8d67 ("clk: Annotate struct clk_hw_onecell_data with __counted_by") Assisted-by: LLM Signed-off-by: Aamir Ahmed Reviewed-by: Chen-Yu Tsai Reviewed-by: Gustavo A. R. Silva Link: https://patch.msgid.link/AS8P251MB00013E724A77A355668B6CCEC8B42@AS8P251MB0001.EURP251.PROD.OUTLOOK.COM Signed-off-by: Alexandre Belloni --- drivers/rtc/rtc-ac100.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/rtc/rtc-ac100.c b/drivers/rtc/rtc-ac100.c index bba7115ff3ad..a2f465438fd0 100644 --- a/drivers/rtc/rtc-ac100.c +++ b/drivers/rtc/rtc-ac100.c @@ -317,6 +317,8 @@ static int ac100_rtc_register_clks(struct ac100_rtc_dev *chip) if (!chip->clk_data) return -ENOMEM; + chip->clk_data->num = AC100_CLKOUT_NUM; + chip->rtc_32k_clk = clk_hw_register_fixed_rate(chip->dev, AC100_RTC_32K_NAME, NULL, 0, @@ -360,7 +362,6 @@ static int ac100_rtc_register_clks(struct ac100_rtc_dev *chip) chip->clk_data->hws[i] = &clk->hw; } - chip->clk_data->num = i; ret = of_clk_add_hw_provider(np, of_clk_hw_onecell_get, chip->clk_data); if (ret) goto err_unregister_rtc_32k; From fcf0b58e976e83adf246b014518e49c662b96f5e Mon Sep 17 00:00:00 2001 From: Aamir Ahmed Date: Sat, 5 Sep 2026 21:41:52 +0100 Subject: [PATCH 161/365] rtc: ac100: Fix clock provider use-after-free on probe failure ac100_rtc_register_clks() registers the RTC-32k clock with clk_hw_register_fixed_rate() and the clock provider with of_clk_add_hw_provider(). Neither is device managed, and both are released only from ac100_rtc_remove(). The driver core does not call remove() when probe() fails: really_probe() reaches probe_failed below the device_remove() call and goes straight to releasing the device's managed resources. ac100_rtc_probe() ends with return devm_rtc_register_device(chip->rtc); so if that fails after the clocks have been registered, the provider is left on the global of_clk_providers list holding chip->clk_data, which was allocated with devm_kzalloc() and is freed while probe() unwinds. A later lookup on this device tree node then reads freed memory in of_clk_hw_onecell_get(). Consumers that do exactly that exist in tree: the wifi power sequence nodes on sun8i-a83t-bananapi-m3 and sun8i-a83t-cubietruck-plus take <&ac100_rtc 1>, and the sun9i-a80 boards route osc32k through <&ac100_rtc 0>. The window is narrow, as devm_rtc_register_device() can only fail with -ENOMEM here, but the provider is left dangling whenever it does. The RTC-32k clock is leaked on the same path. Since clk_core_lookup() searches a global list that is not scoped per device, __clk_register() rejects the duplicate name with -EEXIST, so the leak also makes a later probe of the same device fail. The error path that returns -EINVAL when the ADDA 4M parent clock cannot be found leaks the RTC-32k clock in the same way. No provider has been registered at that point, so that one is a leak rather than a use-after-free. Register both with devm_clk_hw_register_fixed_rate() and devm_of_clk_add_hw_provider() so that they are released whenever the device goes away, on a failed probe as well as on unbind. Devres releases in reverse order of acquisition, so the provider is removed before chip->clk_data is freed, and the clkout children are now unregistered before their parent rather than after it. ac100_rtc_unregister_clks() and the remove callback then have nothing left to do and are removed. Fixes: d00a18a42c14 ("rtc: ac100: Add RTC driver for X-Powers AC100") Reported-by: Sashiko AI Closes: https://lore.kernel.org/linux-rtc/20260905184936.155E11F00A3A@smtp.kernel.org/ Assisted-by: LLM Signed-off-by: Aamir Ahmed Link: https://patch.msgid.link/AS8P251MB0001A489CC45A591ABBB9BDCC8B42@AS8P251MB0001.EURP251.PROD.OUTLOOK.COM Signed-off-by: Alexandre Belloni --- drivers/rtc/rtc-ac100.c | 36 +++++++----------------------------- 1 file changed, 7 insertions(+), 29 deletions(-) diff --git a/drivers/rtc/rtc-ac100.c b/drivers/rtc/rtc-ac100.c index a2f465438fd0..c579c4bf5578 100644 --- a/drivers/rtc/rtc-ac100.c +++ b/drivers/rtc/rtc-ac100.c @@ -319,10 +319,10 @@ static int ac100_rtc_register_clks(struct ac100_rtc_dev *chip) chip->clk_data->num = AC100_CLKOUT_NUM; - chip->rtc_32k_clk = clk_hw_register_fixed_rate(chip->dev, - AC100_RTC_32K_NAME, - NULL, 0, - AC100_RTC_32K_RATE); + chip->rtc_32k_clk = devm_clk_hw_register_fixed_rate(chip->dev, + AC100_RTC_32K_NAME, + NULL, 0, + AC100_RTC_32K_RATE); if (IS_ERR(chip->rtc_32k_clk)) { ret = PTR_ERR(chip->rtc_32k_clk); dev_err(chip->dev, "Failed to register RTC-32k clock: %d\n", @@ -356,28 +356,14 @@ static int ac100_rtc_register_clks(struct ac100_rtc_dev *chip) if (ret) { dev_err(chip->dev, "Failed to register clk '%s': %d\n", init.name, ret); - goto err_unregister_rtc_32k; + return ret; } chip->clk_data->hws[i] = &clk->hw; } - ret = of_clk_add_hw_provider(np, of_clk_hw_onecell_get, chip->clk_data); - if (ret) - goto err_unregister_rtc_32k; - - return 0; - -err_unregister_rtc_32k: - clk_unregister_fixed_rate(chip->rtc_32k_clk->clk); - - return ret; -} - -static void ac100_rtc_unregister_clks(struct ac100_rtc_dev *chip) -{ - of_clk_del_provider(chip->dev->of_node); - clk_unregister_fixed_rate(chip->rtc_32k_clk->clk); + return devm_of_clk_add_hw_provider(chip->dev, of_clk_hw_onecell_get, + chip->clk_data); } /* @@ -615,13 +601,6 @@ static int ac100_rtc_probe(struct platform_device *pdev) return devm_rtc_register_device(chip->rtc); } -static void ac100_rtc_remove(struct platform_device *pdev) -{ - struct ac100_rtc_dev *chip = platform_get_drvdata(pdev); - - ac100_rtc_unregister_clks(chip); -} - static const struct of_device_id ac100_rtc_match[] = { { .compatible = "x-powers,ac100-rtc" }, { }, @@ -630,7 +609,6 @@ MODULE_DEVICE_TABLE(of, ac100_rtc_match); static struct platform_driver ac100_rtc_driver = { .probe = ac100_rtc_probe, - .remove = ac100_rtc_remove, .driver = { .name = "ac100-rtc", .of_match_table = of_match_ptr(ac100_rtc_match), From ac41b05d15db3134e839148290d67415ee0bdb58 Mon Sep 17 00:00:00 2001 From: Manush Prajwal Date: Sun, 6 Sep 2026 16:44:09 +0530 Subject: [PATCH 162/365] rtc: mpfs: fix unchecked devm_clk_get() error pointer in probe() devm_clk_get(&pdev->dev, "rtcref")'s return value was passed straight into clk_get_rate() without checking it for an error first, unlike the "rtc" clock a few lines above which is correctly checked with IS_ERR(). clk_get_rate() only guards against a NULL clk, not an error pointer: if (!clk) return 0; ... rate = clk_core_get_rate_recalc(clk->core); so if devm_clk_get() ever returns an error pointer here (for example ERR_PTR(-EPROBE_DEFER), which is the normal, expected outcome if the clkcfg clock-provider this RTC depends on has not registered its clocks yet by the time this driver probes), clk_get_rate() dereferences that error pointer instead of returning 0, crashing instead of letting probe defer. Capture the clock in the existing 'clk' local and check it with IS_ERR() before calling clk_get_rate(), matching the handling already used for the "rtc" clock in this same function. Signed-off-by: Manush Prajwal Fixes: 0b31d703598d ("rtc: Add driver for Microchip PolarFire SoC") Reviewed-by: Conor Dooley Link: https://patch.msgid.link/6a9d4b02.7d74b517.206c4c.75f8@mx.google.com Signed-off-by: Alexandre Belloni --- drivers/rtc/rtc-mpfs.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/drivers/rtc/rtc-mpfs.c b/drivers/rtc/rtc-mpfs.c index ece6de4a6adb..60596b0ea9d8 100644 --- a/drivers/rtc/rtc-mpfs.c +++ b/drivers/rtc/rtc-mpfs.c @@ -256,8 +256,12 @@ static int mpfs_rtc_probe(struct platform_device *pdev) return ret; } + clk = devm_clk_get(&pdev->dev, "rtcref"); + if (IS_ERR(clk)) + return PTR_ERR(clk); + /* prescaler hardware adds 1 to reg value */ - prescaler = clk_get_rate(devm_clk_get(&pdev->dev, "rtcref")) - 1; + prescaler = clk_get_rate(clk) - 1; if (prescaler > MAX_PRESCALER_COUNT) { dev_dbg(&pdev->dev, "invalid prescaler %lu\n", prescaler); return -EINVAL; From 055ef5ce9f67a6a3a1363fa663007f8196cc0fb8 Mon Sep 17 00:00:00 2001 From: Runyu Xiao Date: Wed, 2 Sep 2026 16:03:18 +0800 Subject: [PATCH 163/365] rtc: spear: initialize IRQ state before requesting alarm IRQ devm_request_irq() enables the interrupt before it returns, so the handler may run while probe is still initializing the device. Initialize the MMIO address and spinlock before requesting the alarm IRQ. Fixes: 0942a71e435f ("rtc: add support for spear rtc") Cc: stable@vger.kernel.org Assisted-by: Codex:GPT-5 Signed-off-by: Runyu Xiao Link: https://patch.msgid.link/20260902080318.3498434-1-runyu.xiao@seu.edu.cn Signed-off-by: Alexandre Belloni --- drivers/rtc/rtc-spear.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-) diff --git a/drivers/rtc/rtc-spear.c b/drivers/rtc/rtc-spear.c index 959acff8faff..d12bccb3cf5e 100644 --- a/drivers/rtc/rtc-spear.c +++ b/drivers/rtc/rtc-spear.c @@ -363,6 +363,12 @@ static int spear_rtc_probe(struct platform_device *pdev) if (irq < 0) return irq; + config->ioaddr = devm_platform_ioremap_resource(pdev, 0); + if (IS_ERR(config->ioaddr)) + return PTR_ERR(config->ioaddr); + + spin_lock_init(&config->lock); + status = devm_request_irq(&pdev->dev, irq, spear_rtc_irq, 0, pdev->name, config); if (status) { @@ -371,10 +377,6 @@ static int spear_rtc_probe(struct platform_device *pdev) return status; } - config->ioaddr = devm_platform_ioremap_resource(pdev, 0); - if (IS_ERR(config->ioaddr)) - return PTR_ERR(config->ioaddr); - config->clk = devm_clk_get(&pdev->dev, NULL); if (IS_ERR(config->clk)) return PTR_ERR(config->clk); @@ -383,7 +385,6 @@ static int spear_rtc_probe(struct platform_device *pdev) if (status < 0) return status; - spin_lock_init(&config->lock); platform_set_drvdata(pdev, config); config->rtc->ops = &spear_rtc_ops; From 67f4c1c6a1b51e203d986779299824d1c2c590a6 Mon Sep 17 00:00:00 2001 From: Zihan Xi Date: Wed, 16 Sep 2026 15:29:24 +0000 Subject: [PATCH 164/365] smb: client: fix create context out-of-bounds reads smb2_parse_contexts() validates the complete create-context area but does not limit each record to its Next field before dispatching it. A malformed chain can therefore expose bytes beyond the current context to a handler. The QFid handler also used a full response-structure cast although it only reads DiskFileId. The SMB2/SMB3 lease parsers made the same layout assumption: they read LeaseState and LeaseFlags at canonical offsets rather than at DataOffset. A valid non-canonical DataOffset could therefore yield unrelated in-bounds data, while a short DataLength was still accepted. Limit each context to its Next value, reject offsets before the context header, and reject malformed chains. Bound the name range by the current context and do not dispatch a known handler when DataLength is zero. Read the QFid DiskFileId only when the context data covers that field. Parse the lease context from DataOffset and require DataLength to match the v1 or v2 lease_context size used by ksmbd. A size mismatch skips lease parsing without failing the open. Fixes: b8c32dbb0deb ("CIFS: Request SMB2.1 leases") Fixes: f047390a097e ("CIFS: Add create lease v2 context for SMB3") Fixes: 89a5bfa350fa ("smb3: optimize open to not send query file internal info") Cc: stable@vger.kernel.org Reported-by: Vega Assisted-by: LLM Co-developed-by: Luxing Yin Signed-off-by: Luxing Yin Signed-off-by: Zihan Xi Tested-by: Frank Sorenson Signed-off-by: Paulo Alcantara --- fs/smb/client/smb2ops.c | 28 ++++++++++++++++++-------- fs/smb/client/smb2pdu.c | 44 +++++++++++++++++++++++++++++++---------- 2 files changed, 54 insertions(+), 18 deletions(-) diff --git a/fs/smb/client/smb2ops.c b/fs/smb/client/smb2ops.c index 3464470d3297..aa142420dae2 100644 --- a/fs/smb/client/smb2ops.c +++ b/fs/smb/client/smb2ops.c @@ -4572,25 +4572,37 @@ smb3_create_lease_buf(u8 *lease_key, u8 oplock, u8 *parent_lease_key, __le32 fla static __u8 smb2_parse_lease_buf(void *buf, __u16 *epoch, char *lease_key) { - struct create_lease *lc = (struct create_lease *)buf; + struct create_context *cc = buf; + struct lease_context lc; *epoch = 0; /* not used */ - if (lc->lcontext.LeaseFlags & SMB2_LEASE_FLAG_BREAK_IN_PROGRESS_LE) + if (le32_to_cpu(cc->DataLength) != sizeof(lc)) + return 0; + + memcpy(&lc, (u8 *)cc + le16_to_cpu(cc->DataOffset), sizeof(lc)); + if (lc.LeaseFlags & SMB2_LEASE_FLAG_BREAK_IN_PROGRESS_LE) return SMB2_OPLOCK_LEVEL_NOCHANGE; - return le32_to_cpu(lc->lcontext.LeaseState); + return le32_to_cpu(lc.LeaseState); } static __u8 smb3_parse_lease_buf(void *buf, __u16 *epoch, char *lease_key) { - struct create_lease_v2 *lc = (struct create_lease_v2 *)buf; + struct create_context *cc = buf; + struct lease_context_v2 lc; - *epoch = le16_to_cpu(lc->lcontext.Epoch); - if (lc->lcontext.LeaseFlags & SMB2_LEASE_FLAG_BREAK_IN_PROGRESS_LE) + if (le32_to_cpu(cc->DataLength) != sizeof(lc)) { + *epoch = 0; + return 0; + } + + memcpy(&lc, (u8 *)cc + le16_to_cpu(cc->DataOffset), sizeof(lc)); + *epoch = le16_to_cpu(lc.Epoch); + if (lc.LeaseFlags & SMB2_LEASE_FLAG_BREAK_IN_PROGRESS_LE) return SMB2_OPLOCK_LEVEL_NOCHANGE; if (lease_key) - memcpy(lease_key, &lc->lcontext.LeaseKey, SMB2_LEASE_KEY_SIZE); - return le32_to_cpu(lc->lcontext.LeaseState); + memcpy(lease_key, lc.LeaseKey, SMB2_LEASE_KEY_SIZE); + return le32_to_cpu(lc.LeaseState); } static unsigned int diff --git a/fs/smb/client/smb2pdu.c b/fs/smb/client/smb2pdu.c index 880ce12f50c4..4046500dbe93 100644 --- a/fs/smb/client/smb2pdu.c +++ b/fs/smb/client/smb2pdu.c @@ -2379,11 +2379,17 @@ create_reconnect_durable_buf(struct cifs_fid *fid) static void parse_query_id_ctxt(struct create_context *cc, struct smb2_file_all_info *buf) { - struct create_disk_id_rsp *pdisk_id = (struct create_disk_id_rsp *)cc; + u16 doff = le16_to_cpu(cc->DataOffset); + u32 dlen = le32_to_cpu(cc->DataLength); + u8 *beg; - cifs_dbg(FYI, "parse query id context 0x%llx 0x%llx\n", - pdisk_id->DiskFileId, pdisk_id->VolumeId); - buf->IndexNumber = pdisk_id->DiskFileId; + if (dlen < sizeof(__le64)) + return; + + beg = (u8 *)cc + doff; + memcpy(&buf->IndexNumber, beg, sizeof(__le64)); + cifs_dbg(FYI, "parse query id context 0x%llx\n", + le64_to_cpu(buf->IndexNumber)); } static void @@ -2431,6 +2437,7 @@ int smb2_parse_contexts(struct TCP_Server_Info *server, struct smb2_create_rsp *rsp = rsp_iov->iov_base; struct create_context *cc; size_t rem, off, len; + size_t cc_len; size_t doff, dlen; size_t noff, nlen; char *name; @@ -2453,29 +2460,41 @@ int smb2_parse_contexts(struct TCP_Server_Info *server, buf->IndexNumber = 0; while (rem >= sizeof(*cc)) { + off = le32_to_cpu(cc->Next); + if (off) { + if ((off & 0x7) || off >= rem || off < sizeof(*cc)) + return -EINVAL; + cc_len = off; + } else { + cc_len = rem; + } + doff = le16_to_cpu(cc->DataOffset); dlen = le32_to_cpu(cc->DataLength); - if (check_add_overflow(doff, dlen, &len) || len > rem) + if (doff < sizeof(*cc) || + check_add_overflow(doff, dlen, &len) || len > cc_len) return -EINVAL; noff = le16_to_cpu(cc->NameOffset); nlen = le16_to_cpu(cc->NameLength); - if (noff + nlen > doff) + if (noff < sizeof(*cc) || + check_add_overflow(noff, nlen, &len) || len > cc_len || + (dlen && len > doff)) return -EINVAL; name = (char *)cc + noff; switch (nlen) { case 4: - if (!strncmp(name, SMB2_CREATE_REQUEST_LEASE, 4)) { + if (dlen && !strncmp(name, SMB2_CREATE_REQUEST_LEASE, 4)) { *oplock = server->ops->parse_lease_buf(cc, epoch, lease_key); - } else if (buf && + } else if (dlen && buf && !strncmp(name, SMB2_CREATE_QUERY_ON_DISK_ID, 4)) { parse_query_id_ctxt(cc, buf); } break; case 16: - if (posix && !memcmp(name, smb3_create_tag_posix, 16)) + if (dlen && posix && !memcmp(name, smb3_create_tag_posix, 16)) parse_posix_ctxt(cc, buf, posix); break; default: @@ -2487,13 +2506,18 @@ int smb2_parse_contexts(struct TCP_Server_Info *server, } off = le32_to_cpu(cc->Next); - if (!off) + if (!off) { + rem = 0; break; + } if (check_sub_overflow(rem, off, &rem)) return -EINVAL; cc = (struct create_context *)((u8 *)cc + off); } + if (rem) + return -EINVAL; + if (rsp->OplockLevel != SMB2_OPLOCK_LEVEL_LEASE) *oplock = rsp->OplockLevel; From fa2e9900dd2a3f5a1e7ef5a8c5e8d435feedbfcc Mon Sep 17 00:00:00 2001 From: Zihan Xi Date: Wed, 16 Sep 2026 15:29:25 +0000 Subject: [PATCH 165/365] smb: client: validate POSIX create context length parse_posix_ctxt() reads the fixed nlink, reparse_tag, and mode fields before checking that the POSIX create context contains them. A short context can pass the generic checks and still make these fixed-width reads run past its declared data. The current in-tree smb2_open_file() path passes a NULL posix pointer, so this handler is not reached on the ordinary open path. Still require the POSIX data to cover all three fields before reading them because the helper performs those unguarded reads. Keep the existing soft-failure behavior so malformed optional metadata does not fail the open. Fixes: 69dda3059e7a ("cifs: add SMB2_open() arg to return POSIX data") Cc: stable@vger.kernel.org Reported-by: Vega Assisted-by: LLM Co-developed-by: Luxing Yin Signed-off-by: Luxing Yin Signed-off-by: Zihan Xi Tested-by: Frank Sorenson Signed-off-by: Paulo Alcantara --- fs/smb/client/smb2pdu.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/fs/smb/client/smb2pdu.c b/fs/smb/client/smb2pdu.c index 4046500dbe93..538d708b0404 100644 --- a/fs/smb/client/smb2pdu.c +++ b/fs/smb/client/smb2pdu.c @@ -2396,12 +2396,15 @@ static void parse_posix_ctxt(struct create_context *cc, struct smb2_file_all_info *info, struct create_posix_rsp *posix) { - int sid_len; u8 *beg = (u8 *)cc + le16_to_cpu(cc->DataOffset); - u8 *end = beg + le32_to_cpu(cc->DataLength); + u32 dlen = le32_to_cpu(cc->DataLength); + u8 *end = beg + dlen; + int sid_len; u8 *sid; memset(posix, 0, sizeof(*posix)); + if (dlen < 3 * sizeof(__le32)) + return; posix->nlink = get_unaligned_le32(beg); posix->reparse_tag = get_unaligned_le32(beg + 4); From 566820af017e81497fb5e9d3ad6e7ffe2828bc8b Mon Sep 17 00:00:00 2001 From: Zihan Xi Date: Wed, 16 Sep 2026 15:29:26 +0000 Subject: [PATCH 166/365] smb: client: close handle after create-context parsing failure SMB2_open() accounts a successful CREATE response as a remote open before parsing its create contexts. If smb2_parse_contexts() rejects malformed context data, SMB2_open() returns without closing the handle, leaving the server-side handle open and num_remote_opens elevated. Close the handle after a post-CREATE context parsing failure so the error path releases the remote resource and balances the open count. Fixes: af1689a9b770 ("smb: client: fix potential OOBs in smb2_parse_contexts()") Cc: stable@vger.kernel.org Reported-by: Vega Assisted-by: LLM Co-developed-by: Luxing Yin Signed-off-by: Luxing Yin Signed-off-by: Zihan Xi Tested-by: Frank Sorenson Signed-off-by: Paulo Alcantara --- fs/smb/client/smb2pdu.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/fs/smb/client/smb2pdu.c b/fs/smb/client/smb2pdu.c index 538d708b0404..3d7ead36d1a0 100644 --- a/fs/smb/client/smb2pdu.c +++ b/fs/smb/client/smb2pdu.c @@ -3416,6 +3416,9 @@ SMB2_open(const unsigned int xid, struct cifs_open_parms *oparms, __le16 *path, rc = smb2_parse_contexts(server, &rsp_iov, &oparms->fid->epoch, oparms->fid->lease_key, oplock, file_info, posix); + if (rc) + SMB2_close(xid, tcon, oparms->fid->persistent_fid, + oparms->fid->volatile_fid); trace_smb3_open_done(xid, rsp->PersistentFileId, tcon->tid, ses->Suid, oparms->create_options, oparms->desired_access, From d2ff5fb93ea83034025850266b5eed391f96b825 Mon Sep 17 00:00:00 2001 From: Zihan Xi Date: Wed, 16 Sep 2026 15:29:27 +0000 Subject: [PATCH 167/365] smb: client: clean up failed cached directory opens open_cached_dir() sends CREATE and QUERY_INFO as a compound request. If the CREATE succeeds but a later command returns an error, the function must retain the CREATE FID so common cleanup can issue SMB2_close(). It also must not treat a response error as a valid CREATE. Validate the CREATE response before using its fields, record the FIDs, and mark the handle open before handling errors from later compound commands. Move the -EREMCHG reconnect handling before response validation so a missing response does not hide the reconnect request. Count the handle when it is marked open; confirmed close responses decrement the counter, while existing close retry behavior remains best effort on transport failures. Fixes: b0f6df737a1c ("cifs: cache FILE_ALL_INFO for the shared root handle") Cc: stable@vger.kernel.org Reported-by: Vega Assisted-by: LLM Co-developed-by: Luxing Yin Signed-off-by: Luxing Yin Signed-off-by: Zihan Xi Tested-by: Frank Sorenson Signed-off-by: Paulo Alcantara --- fs/smb/client/cached_dir.c | 32 ++++++++++++++++++++++---------- 1 file changed, 22 insertions(+), 10 deletions(-) diff --git a/fs/smb/client/cached_dir.c b/fs/smb/client/cached_dir.c index 88d5e9a32f28..647fa26da4d2 100644 --- a/fs/smb/client/cached_dir.c +++ b/fs/smb/client/cached_dir.c @@ -8,6 +8,7 @@ #include #include "cifsglob.h" #include "cifsproto.h" +#include "../common/smb2status.h" #include "cifs_debug.h" #include "smb2proto.h" #include "cached_dir.h" @@ -323,25 +324,37 @@ int open_cached_dir(unsigned int xid, struct cifs_tcon *tcon, rc = compound_send_recv(xid, ses, server, flags, 2, rqst, resp_buftype, rsp_iov); - if (rc) { - if (rc == -EREMCHG) { - tcon->need_reconnect = true; - pr_warn_once("server share %s deleted\n", - tcon->tree_name); - } + if (rc == -EREMCHG) { + tcon->need_reconnect = true; + pr_warn_once("server share %s deleted\n", + tcon->tree_name); + } + + if (!rsp_iov[0].iov_base || rsp_iov[0].iov_len < sizeof(*o_rsp)) { + if (!rc) + rc = -EIO; goto oshr_free; } - cfid->is_open = true; - - spin_lock(&cfids->cfid_list_lock); o_rsp = (struct smb2_create_rsp *)rsp_iov[0].iov_base; + if (o_rsp->hdr.Status != STATUS_SUCCESS) { + if (!rc) + rc = -EIO; + goto oshr_free; + } + oparms.fid->persistent_fid = o_rsp->PersistentFileId; oparms.fid->volatile_fid = o_rsp->VolatileFileId; #ifdef CONFIG_CIFS_DEBUG2 oparms.fid->mid = le64_to_cpu(o_rsp->hdr.MessageId); #endif /* CIFS_DEBUG2 */ + cfid->is_open = true; + atomic_inc(&tcon->num_remote_opens); + if (rc) + goto oshr_free; + + spin_lock(&cfids->cfid_list_lock); if (o_rsp->OplockLevel != SMB2_OPLOCK_LEVEL_LEASE) { spin_unlock(&cfids->cfid_list_lock); @@ -408,7 +421,6 @@ int open_cached_dir(unsigned int xid, struct cifs_tcon *tcon, close_cached_dir(cfid); } else { *ret_cfid = cfid; - atomic_inc(&tcon->num_remote_opens); } kfree(utf16_path); From 6c5c547f037bc18f0b8d0b5db5a648f8f630ce85 Mon Sep 17 00:00:00 2001 From: Zihan Xi Date: Wed, 16 Sep 2026 15:29:28 +0000 Subject: [PATCH 168/365] smb: client: close completed creates on compound wait errors compound_send_recv() waits for responses in order. If a later wait is interrupted, or if a later MID fails during response synchronization, an earlier CREATE may already have opened a remote handle. The earlier mid is then released without invoking handle_cancelled_mid(), leaving the remote handle open because no FID was copied to the caller. Mark completed earlier mids as cancelled when a compound wait or MID synchronization aborts. Keep their response buffers attached while the MIDs are synchronized, and transfer them only after synchronization of the processed responses, so the release path can inspect successful CREATE responses and queue SMB2_close() after a later failure. Account for a remote open only after the close work is allocated and before it is queued, since the caller has not yet updated num_remote_opens. Mark the create+close compound used by smb2_unlink() so it is not closed again. Non-CREATE responses and compounds that already include a close keep their existing behavior. Fixes: e0bba0b85481 ("cifs: add compound_send_recv()") Cc: stable@vger.kernel.org Reported-by: Vega Assisted-by: LLM Co-developed-by: Luxing Yin Signed-off-by: Luxing Yin Signed-off-by: Zihan Xi Tested-by: Frank Sorenson Signed-off-by: Paulo Alcantara --- fs/smb/client/smb2inode.c | 2 +- fs/smb/client/smb2misc.c | 9 ++++-- fs/smb/client/transport.c | 67 +++++++++++++++++++++++++++++++-------- 3 files changed, 61 insertions(+), 17 deletions(-) diff --git a/fs/smb/client/smb2inode.c b/fs/smb/client/smb2inode.c index 13fe8e3b48f3..ecd7a65cab08 100644 --- a/fs/smb/client/smb2inode.c +++ b/fs/smb/client/smb2inode.c @@ -1121,7 +1121,7 @@ smb2_unlink(const unsigned int xid, struct cifs_tcon *tcon, const char *name, struct kvec close_iov; int resp_buftype[2]; struct cifs_fid fid; - int flags = 0; + int flags = CIFS_CP_CREATE_CLOSE_OP; __u8 oplock; int rc; diff --git a/fs/smb/client/smb2misc.c b/fs/smb/client/smb2misc.c index 0cfe60ae42c3..5e5cf92d1eb3 100644 --- a/fs/smb/client/smb2misc.c +++ b/fs/smb/client/smb2misc.c @@ -834,7 +834,8 @@ smb2_cancelled_close_fid(struct work_struct *work) */ static int __smb2_handle_cancelled_cmd(struct cifs_tcon *tcon, __u16 cmd, __u64 mid, - __u64 persistent_fid, __u64 volatile_fid) + __u64 persistent_fid, __u64 volatile_fid, + bool account_remote_open) { struct close_cancelled_open *cancelled; @@ -848,6 +849,8 @@ __smb2_handle_cancelled_cmd(struct cifs_tcon *tcon, __u16 cmd, __u64 mid, cancelled->cmd = cmd; cancelled->mid = mid; INIT_WORK(&cancelled->work, smb2_cancelled_close_fid); + if (account_remote_open) + atomic_inc(&tcon->num_remote_opens); WARN_ON(queue_work(cifsiod_wq, &cancelled->work) == false); return 0; @@ -884,7 +887,7 @@ smb2_handle_cancelled_close(struct cifs_tcon *tcon, __u64 persistent_fid, spin_unlock(&tcon->tc_lock); rc = __smb2_handle_cancelled_cmd(tcon, SMB2_CLOSE_HE, 0, - persistent_fid, volatile_fid); + persistent_fid, volatile_fid, false); if (rc) cifs_put_tcon(tcon, netfs_trace_tcon_ref_put_cancelled_close); @@ -912,7 +915,7 @@ smb2_handle_cancelled_mid(struct mid_q_entry *mid, struct TCP_Server_Info *serve le16_to_cpu(hdr->Command), le64_to_cpu(hdr->MessageId), rsp->PersistentFileId, - rsp->VolatileFileId); + rsp->VolatileFileId, true); if (rc) cifs_put_tcon(tcon, netfs_trace_tcon_ref_put_cancelled_mid); diff --git a/fs/smb/client/transport.c b/fs/smb/client/transport.c index 7df5b3447aea..6e21b5f8754a 100644 --- a/fs/smb/client/transport.c +++ b/fs/smb/client/transport.c @@ -805,6 +805,18 @@ cifs_cancelled_callback(struct TCP_Server_Info *server, struct mid_q_entry *mid) release_mid(server, mid); } +static void +cifs_mark_compound_mids_cancelled(struct mid_q_entry **mid, int count) +{ + int i; + + for (i = 0; i < count; i++) { + spin_lock(&mid[i]->mid_lock); + mid[i]->wait_cancelled = true; + spin_unlock(&mid[i]->mid_lock); + } +} + /* * cifs_pick_channel - pick an eligible channel for network operations * @@ -865,6 +877,7 @@ compound_send_recv(const unsigned int xid, struct cifs_ses *ses, int *resp_buf_type, struct kvec *resp_iov) { int i, j, optype, rc = 0; + int num_processed = 0; struct mid_q_entry *mid[MAX_COMPOUND]; bool cancelled_mid[MAX_COMPOUND] = {false}; struct cifs_credits credits[MAX_COMPOUND] = { @@ -1015,6 +1028,14 @@ compound_send_recv(const unsigned int xid, struct cifs_ses *ses, break; } if (rc != 0) { + /* + * A completed CREATE earlier in the compound chain may have + * opened a remote handle even though a later wait was + * interrupted. Mark it cancelled so __release_mid() invokes + * the existing unmatched-open cleanup. + */ + cifs_mark_compound_mids_cancelled(mid, i); + for (; i < num_rqst; i++) { cifs_server_dbg(FYI, "Cancelling wait for mid %llu cmd: %d\n", mid[i]->mid, le16_to_cpu(mid[i]->command)); @@ -1037,6 +1058,14 @@ compound_send_recv(const unsigned int xid, struct cifs_ses *ses, rc = cifs_sync_mid_result(mid[i], server); if (rc != 0) { + /* + * A previous CREATE may have completed before this + * response failed. Mark it cancelled so its remote + * handle is closed when the mid is released. + */ + cifs_mark_compound_mids_cancelled(mid, i); + /* Keep their response buffers for cancelled-mid cleanup. */ + num_processed = 0; /* mark this mid as cancelled to not free it below */ cancelled_mid[i] = true; goto out; @@ -1046,13 +1075,24 @@ compound_send_recv(const unsigned int xid, struct cifs_ses *ses, mid[i]->mid_state != MID_RESPONSE_READY) { rc = smb_EIO1(smb_eio_trace_rx_mid_unready, mid[i]->mid_state); cifs_dbg(FYI, "Bad MID state?\n"); + cifs_mark_compound_mids_cancelled(mid, i); + num_processed = 0; goto out; } rc = server->ops->check_receive(mid[i], server, flags & CIFS_LOG_ERROR); + num_processed = i + 1; + } - if (resp_iov) { +out: + /* + * Delay moving response buffers out of their mids until response + * synchronization completes. This lets cancelled-mid cleanup inspect + * an earlier CREATE response if a later MID fails. + */ + if (resp_iov) { + for (i = 0; i < num_processed; i++) { buf = (char *)mid[i]->resp_buf; resp_iov[i].iov_base = buf; resp_iov[i].iov_len = mid[i]->resp_buf_size; @@ -1071,21 +1111,22 @@ compound_send_recv(const unsigned int xid, struct cifs_ses *ses, /* * Compounding is never used during session establish. */ - spin_lock(&ses->ses_lock); - if ((ses->ses_status == SES_NEW) || (optype & CIFS_NEG_OP) || (optype & CIFS_SESS_OP)) { - struct kvec iov = { - .iov_base = resp_iov[0].iov_base, - .iov_len = resp_iov[0].iov_len - }; - spin_unlock(&ses->ses_lock); - cifs_server_lock(server); - smb311_update_preauth_hash(ses, server, &iov, 1); - cifs_server_unlock(server); + if (num_processed == num_rqst) { spin_lock(&ses->ses_lock); + if ((ses->ses_status == SES_NEW) || (optype & CIFS_NEG_OP) || (optype & CIFS_SESS_OP)) { + struct kvec iov = { + .iov_base = resp_iov[0].iov_base, + .iov_len = resp_iov[0].iov_len + }; + spin_unlock(&ses->ses_lock); + cifs_server_lock(server); + smb311_update_preauth_hash(ses, server, &iov, 1); + cifs_server_unlock(server); + spin_lock(&ses->ses_lock); + } + spin_unlock(&ses->ses_lock); } - spin_unlock(&ses->ses_lock); -out: /* * This will dequeue all mids. After this it is important that the * demultiplex_thread will not process any of these mids any further. From 2e828035d5d736d904c238ae7ec3f77c0d6270bf Mon Sep 17 00:00:00 2001 From: Zihan Xi Date: Wed, 16 Sep 2026 15:29:29 +0000 Subject: [PATCH 169/365] smb: client: preserve create-context parsing errors smb2_compound_op() saves the result from compound_send_recv() in tmp_rc. For SMB2_OP_OPEN_QUERY it then parses the CREATE contexts, but the final assignment of rc from tmp_rc discards a parsing error. A malformed create-context response can therefore be reported as successful to smb2_query_path_info(). Keep a create-context parsing error in tmp_rc so it survives per-command response processing and is returned to the caller. Fixes: b07687edee99 ("cifs: Improve SMB2+ stat() to work also without FILE_READ_ATTRIBUTES") Cc: stable@vger.kernel.org Reported-by: Vega Assisted-by: LLM Co-developed-by: Luxing Yin Signed-off-by: Luxing Yin Signed-off-by: Zihan Xi Tested-by: Frank Sorenson Signed-off-by: Paulo Alcantara --- fs/smb/client/smb2inode.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/fs/smb/client/smb2inode.c b/fs/smb/client/smb2inode.c index ecd7a65cab08..f46a62eae659 100644 --- a/fs/smb/client/smb2inode.c +++ b/fs/smb/client/smb2inode.c @@ -598,8 +598,10 @@ static int smb2_compound_op(const unsigned int xid, struct cifs_tcon *tcon, /* smb2_parse_contexts() fills idata->fi.IndexNumber */ rc = smb2_parse_contexts(server, &rsp_iov[0], &oparms->fid->epoch, oparms->fid->lease_key, &oplock, &idata->fi, NULL); - if (rc) + if (rc) { cifs_dbg(VFS, "rc: %d parsing context of compound op\n", rc); + tmp_rc = rc; + } } for (i = 0; i < num_cmds; i++) { From 6c43c72748fffd29dec15cd1f31e9a32949bc437 Mon Sep 17 00:00:00 2001 From: Melody Wang Date: Mon, 14 Sep 2026 00:57:17 +0000 Subject: [PATCH 170/365] x86/sev: Make vTPM SVSM calls preemption-safe Two functions in the SVSM vTPM guest implementation do not disable preemption when fetching the SVSM Calling Area Address (CAA). The SVSM CAA is a per-CPU structure. When a thread is preempted and migrated to a different CPU after fetching the per-CPU CAA, the SVSM call will execute on the new CPU with the original CPU's CAA. Which is wrong. Move the CAA fetching operation inside svsm_perform_call_protocol() which disables interrupts around the SVSM call and thus runs preemption-safe. Fixes: 770de678bc28 ("x86/sev: Add SVSM vTPM probe/send_command functions") Signed-off-by: Melody Wang Signed-off-by: Borislav Petkov (AMD) Reviewed-by: Stefano Garzarella Cc: stable@vger.kernel.org Link: https://patch.msgid.link/a5bc0d4a2c462a0089109e145c21626b244b2ff0.1789345277.git.huibo.wang@amd.com --- arch/x86/coco/sev/svsm.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) diff --git a/arch/x86/coco/sev/svsm.c b/arch/x86/coco/sev/svsm.c index 916d62cd17dc..2d493d55ff2e 100644 --- a/arch/x86/coco/sev/svsm.c +++ b/arch/x86/coco/sev/svsm.c @@ -74,6 +74,14 @@ int svsm_perform_call_protocol(struct svsm_call *call) flags = native_local_irq_save(); + /* + * 'caa' is a per-CPU variable. To avoid using a stale or incorrect + * 'caa' if the task is preempted or migrated to another CPU after it + * is fetched, always fetch 'caa' and then issue the SVSM call with + * interrupts disabled. This ensures the correct 'caa' is used. + */ + call->caa = svsm_get_caa(); + ghcb = __sev_get_ghcb(&state); do { @@ -321,7 +329,6 @@ int snp_svsm_vtpm_send_command(u8 *buffer) { struct svsm_call call = {}; - call.caa = svsm_get_caa(); call.rax = SVSM_VTPM_CALL(SVSM_VTPM_CMD); call.rcx = __pa(buffer); @@ -345,7 +352,6 @@ bool snp_svsm_vtpm_probe(void) if (!snp_vmpl) return false; - call.caa = svsm_get_caa(); call.rax = SVSM_VTPM_CALL(SVSM_VTPM_QUERY); if (svsm_perform_call_protocol(&call)) From 72b782097e534ab48152dd25aaba40dda5f25f9e Mon Sep 17 00:00:00 2001 From: Jakub Pawlak Date: Mon, 14 Sep 2026 10:43:01 +0200 Subject: [PATCH 171/365] accel/ivpu: Use separate flag for job timeout Use separate flag to mark a job timeout as a reason of starting context_abort_work. This allows to distinguish engine reset reason and clearly adjust reset procedure flow. The flag is cleared in ivpu_prepare_for_reset(), which every recovery and suspend path already funnels through, so that the state is clean after recovery. Cc: stable@vger.kernel.org # v7.1+ Fixes: ade00a6c903f ("accel/ivpu: Perform engine reset instead of device recovery on TDR") Signed-off-by: Jakub Pawlak Reviewed-by: Dawid Osuchowski Signed-off-by: Karol Wachowski Link: https://patch.msgid.link/20260914084301.894028-1-karol.wachowski@linux.intel.com --- drivers/accel/ivpu/ivpu_drv.c | 3 ++- drivers/accel/ivpu/ivpu_drv.h | 2 +- drivers/accel/ivpu/ivpu_job.c | 7 +++---- drivers/accel/ivpu/ivpu_mmu.c | 1 - drivers/accel/ivpu/ivpu_pm.c | 1 + 5 files changed, 7 insertions(+), 7 deletions(-) diff --git a/drivers/accel/ivpu/ivpu_drv.c b/drivers/accel/ivpu/ivpu_drv.c index 95120957f42a..8c1c87e69f91 100644 --- a/drivers/accel/ivpu/ivpu_drv.c +++ b/drivers/accel/ivpu/ivpu_drv.c @@ -515,6 +515,7 @@ void ivpu_prepare_for_reset(struct ivpu_device *vdev) { ivpu_hw_irq_disable(vdev); disable_irq(vdev->irq); + atomic_set(&vdev->job_timeout_detected, 0); flush_work(&vdev->irq_dct_work); flush_work(&vdev->context_abort_work); flush_work(&vdev->job_destroy_work); @@ -710,7 +711,7 @@ static int ivpu_dev_init(struct ivpu_device *vdev) vdev->context_xa_limit.max = IVPU_USER_CONTEXT_MAX_SSID; atomic64_set(&vdev->unique_id_counter, 0); atomic_set(&vdev->job_timeout_counter, 0); - atomic_set(&vdev->faults_detected, 0); + atomic_set(&vdev->job_timeout_detected, 0); xa_init_flags(&vdev->context_xa, XA_FLAGS_ALLOC | XA_FLAGS_LOCK_IRQ); xa_init_flags(&vdev->submitted_jobs_xa, XA_FLAGS_ALLOC1); xa_init_flags(&vdev->db_xa, XA_FLAGS_ALLOC1); diff --git a/drivers/accel/ivpu/ivpu_drv.h b/drivers/accel/ivpu/ivpu_drv.h index 86d7c9966cac..6f4012926478 100644 --- a/drivers/accel/ivpu/ivpu_drv.h +++ b/drivers/accel/ivpu/ivpu_drv.h @@ -171,7 +171,7 @@ struct ivpu_device { struct xarray submitted_jobs_xa; struct ivpu_ipc_consumer job_done_consumer; atomic_t job_timeout_counter; - atomic_t faults_detected; + atomic_t job_timeout_detected; atomic64_t unique_id_counter; diff --git a/drivers/accel/ivpu/ivpu_job.c b/drivers/accel/ivpu/ivpu_job.c index ebb2c865b09a..4689b8ab519d 100644 --- a/drivers/accel/ivpu/ivpu_job.c +++ b/drivers/accel/ivpu/ivpu_job.c @@ -621,7 +621,6 @@ bool ivpu_job_handle_engine_error(struct ivpu_device *vdev, u32 job_id, u32 job_ * status and ensure both are handled in the same way */ job->file_priv->has_mmu_faults = true; - atomic_set(&vdev->faults_detected, 1); queue_work(system_percpu_wq, &vdev->context_abort_work); return true; } @@ -1175,10 +1174,10 @@ static int reset_engine_and_mark_faulty_contexts(struct ivpu_device *vdev) return ret; /* - * If faults are detected, ignore guilty contexts from engine reset as NPU may not be stuck - * and could return currently running good context and faulty contexts are already marked + * If job timeout is detected, read guilty context from engine reset, for other reasons + * faulty context is already known */ - if (atomic_cmpxchg(&vdev->faults_detected, 1, 0) == 1) + if (atomic_cmpxchg(&vdev->job_timeout_detected, 1, 0) == 0) return 0; num_impacted_contexts = resp.payload.engine_reset_done.num_impacted_contexts; diff --git a/drivers/accel/ivpu/ivpu_mmu.c b/drivers/accel/ivpu/ivpu_mmu.c index 41efd8985fa6..b2025274f91d 100644 --- a/drivers/accel/ivpu/ivpu_mmu.c +++ b/drivers/accel/ivpu/ivpu_mmu.c @@ -964,7 +964,6 @@ void ivpu_mmu_irq_evtq_handler(struct ivpu_device *vdev) file_priv = xa_load(&vdev->context_xa, ssid); if (file_priv) { if (!READ_ONCE(file_priv->has_mmu_faults)) { - atomic_set(&vdev->faults_detected, 1); ivpu_mmu_dump_event(vdev, event); WRITE_ONCE(file_priv->has_mmu_faults, true); } diff --git a/drivers/accel/ivpu/ivpu_pm.c b/drivers/accel/ivpu/ivpu_pm.c index c1ce8329790e..de0becbfdffb 100644 --- a/drivers/accel/ivpu/ivpu_pm.c +++ b/drivers/accel/ivpu/ivpu_pm.c @@ -229,6 +229,7 @@ static void ivpu_job_timeout_work(struct work_struct *work) ivpu_jsm_state_dump(vdev); ivpu_dev_coredump(vdev); + atomic_set(&vdev->job_timeout_detected, 1); queue_work(system_percpu_wq, &vdev->context_abort_work); } From 6bdcdb31318c10e6369cff8e5b32c37397e98091 Mon Sep 17 00:00:00 2001 From: Yuqi Xu Date: Mon, 21 Sep 2026 10:52:04 +0800 Subject: [PATCH 172/365] wifi: mac80211: validate TX status rate metadata minstrel_ht_tx_status() converts the HT/VHT rate reported in a TX status into a group and rate index and uses those to index the minstrel_ht rate tables. The existing validation only checked that an entry was present and had tries recorded; it did not check that the reported MCS, spatial stream count and bandwidth are representable by the tables, so malformed metadata passed to minstrel_ht_get_stats() or minstrel_ht_ri_get_stats() could be used as an out-of-bounds index. Validate the rate metadata in the common TX status path instead of only in minstrel_ht, since the values are used by more than rate control. The legacy ieee80211_tx_rate array and the rate_info based entries are sanitized in ieee80211_tx_status_ext() and ieee80211_tx_rate_update() before any consumer runs: an entry that does not describe a valid rate for its encoding is dropped. minstrel_ht keeps the checks that depend on its own tables, i.e. the supported number of spatial streams, the MCS group size and the supported bandwidths. This does not take away the driver's responsibility to report sane values, it only prevents a single bad report from corrupting mac80211 state. No WARN_ON() is added: with panic_on_warn a driver bug would otherwise turn into a denial of service. Fixes: a5f69d94d8e2 ("mac80211: Get rid of search loop for rate group index") Fixes: 9208247d74bc ("mac80211: minstrel_ht: add basic support for VHT rates <= 3SS@80MHz") Reported-by: Vega Assisted-by: LLM Signed-off-by: Yuqi Xu Reviewed-by: Ren Wei Link: https://patch.msgid.link/607e985d71fbbe63ebfc182a37f882f04f980f7f.1789958876.git.xuyuqiabc@gmail.com Signed-off-by: Johannes Berg --- net/mac80211/rc80211_minstrel_ht.c | 59 ++++++++++++++++++++++--- net/mac80211/status.c | 70 ++++++++++++++++++++++++++++++ 2 files changed, 123 insertions(+), 6 deletions(-) diff --git a/net/mac80211/rc80211_minstrel_ht.c b/net/mac80211/rc80211_minstrel_ht.c index c26221287515..45e7a7ea7690 100644 --- a/net/mac80211/rc80211_minstrel_ht.c +++ b/net/mac80211/rc80211_minstrel_ht.c @@ -1194,6 +1194,54 @@ minstrel_ht_update_stats(struct minstrel_priv *mp, struct minstrel_ht_sta *mi) mi->sample_time = jiffies; } +/* + * Check whether an HT/VHT rate from a TX status entry maps to an entry + * in the minstrel_ht rate tables. Values that cannot be represented + * there must not be used for indexing mi->groups[] and the MCS groups. + */ +static bool +minstrel_ht_txstat_rate_valid(struct ieee80211_tx_rate *rate) +{ + unsigned int bw; + + if (!(rate->flags & (IEEE80211_TX_RC_MCS | IEEE80211_TX_RC_VHT_MCS))) + return true; + + if (rate->flags & IEEE80211_TX_RC_MCS) { + /* minstrel_ht supports up to MINSTREL_MAX_STREAMS streams */ + return rate->idx < MINSTREL_MAX_STREAMS * 8; + } + + /* minstrel_ht has no VHT groups for 160 MHz and wider */ + bw = !!(rate->flags & IEEE80211_TX_RC_40_MHZ_WIDTH) + + 2 * !!(rate->flags & IEEE80211_TX_RC_80_MHZ_WIDTH); + if ((rate->flags & IEEE80211_TX_RC_160_MHZ_WIDTH) || bw > BW_80) + return false; + + return ieee80211_rate_get_vht_nss(rate) <= MINSTREL_MAX_STREAMS && + ieee80211_rate_get_vht_mcs(rate) < MCS_GROUP_RATES; +} + +static bool +minstrel_ht_ri_txstat_rate_valid(struct rate_info *rate) +{ + if (!(rate->flags & (RATE_INFO_FLAGS_MCS | RATE_INFO_FLAGS_VHT_MCS))) + return true; + + if (rate->flags & RATE_INFO_FLAGS_MCS) { + /* minstrel_ht supports up to MINSTREL_MAX_STREAMS streams */ + return rate->mcs < MINSTREL_MAX_STREAMS * 8; + } + + /* minstrel_ht has VHT groups only for 20/40/80 MHz */ + if (rate->bw != RATE_INFO_BW_20 && rate->bw != RATE_INFO_BW_40 && + rate->bw != RATE_INFO_BW_80) + return false; + + return rate->nss <= MINSTREL_MAX_STREAMS && + rate->mcs < MCS_GROUP_RATES; +} + static bool minstrel_ht_txstat_valid(struct minstrel_priv *mp, struct minstrel_ht_sta *mi, struct ieee80211_tx_rate *rate) @@ -1206,9 +1254,8 @@ minstrel_ht_txstat_valid(struct minstrel_priv *mp, struct minstrel_ht_sta *mi, if (!rate->count) return false; - if (rate->flags & IEEE80211_TX_RC_MCS || - rate->flags & IEEE80211_TX_RC_VHT_MCS) - return true; + if (rate->flags & (IEEE80211_TX_RC_MCS | IEEE80211_TX_RC_VHT_MCS)) + return minstrel_ht_txstat_rate_valid(rate); for (i = 0; i < ARRAY_SIZE(mp->cck_rates); i++) if (rate->idx == mp->cck_rates[i]) @@ -1236,9 +1283,9 @@ minstrel_ht_ri_txstat_valid(struct minstrel_priv *mp, if (!rate_status->try_count) return false; - if (rate_status->rate_idx.flags & RATE_INFO_FLAGS_MCS || - rate_status->rate_idx.flags & RATE_INFO_FLAGS_VHT_MCS) - return true; + if (rate_status->rate_idx.flags & + (RATE_INFO_FLAGS_MCS | RATE_INFO_FLAGS_VHT_MCS)) + return minstrel_ht_ri_txstat_rate_valid(&rate_status->rate_idx); for (i = 0; i < ARRAY_SIZE(mp->cck_rates); i++) { if (rate_status->rate_idx.legacy == diff --git a/net/mac80211/status.c b/net/mac80211/status.c index 3d811f652603..414979e2a1d9 100644 --- a/net/mac80211/status.c +++ b/net/mac80211/status.c @@ -1164,6 +1164,73 @@ void ieee80211_tx_status_skb(struct ieee80211_hw *hw, struct sk_buff *skb) } EXPORT_SYMBOL(ieee80211_tx_status_skb); +/* + * Check whether the HT/VHT rate information in a TX status entry is + * valid for its encoding. This is not specific to any rate control + * algorithm; all status consumers rely on the values being sane. + */ +static bool ieee80211_tx_status_rate_info_valid(const struct rate_info *rate) +{ + if (rate->flags & RATE_INFO_FLAGS_MCS) + return rate->mcs < 32; + + if (rate->flags & RATE_INFO_FLAGS_VHT_MCS) + return rate->nss >= 1 && rate->nss <= 8 && rate->mcs <= 11; + + return true; +} + +static bool +ieee80211_tx_status_tx_rate_valid(const struct ieee80211_tx_rate *rate) +{ + if (rate->flags & IEEE80211_TX_RC_MCS) + return rate->idx < 32; + + if (rate->flags & IEEE80211_TX_RC_VHT_MCS) + return ieee80211_rate_get_vht_mcs(rate) <= 11; + + return true; +} + +/* + * Drop TX status rate entries that don't describe a valid rate. The + * status information is used by rate control and by other mac80211 + * code, and a malformed entry must not be able to corrupt state beyond + * the driver that reported it. + */ +static void +ieee80211_tx_status_drop_invalid_rates(struct ieee80211_tx_status *status) +{ + int i; + + for (i = 0; i < status->n_rates; i++) { + struct ieee80211_rate_status *rs = &status->rates[i]; + + if (ieee80211_tx_status_rate_info_valid(&rs->rate_idx)) + continue; + + rs->try_count = 0; + memset(&rs->rate_idx, 0, sizeof(rs->rate_idx)); + } + + if (!status->info) + return; + + for (i = 0; i < IEEE80211_TX_MAX_RATES; i++) { + struct ieee80211_tx_rate *rate; + + rate = &status->info->status.rates[i]; + if (rate->idx < 0) + break; + + if (ieee80211_tx_status_tx_rate_valid(rate)) + continue; + + rate->idx = -1; + rate->count = 0; + } +} + void ieee80211_tx_status_ext(struct ieee80211_hw *hw, struct ieee80211_tx_status *status) { @@ -1176,6 +1243,8 @@ void ieee80211_tx_status_ext(struct ieee80211_hw *hw, bool acked, noack_success, ack_signal_valid; u16 tx_time_est; + ieee80211_tx_status_drop_invalid_rates(status); + if (pubsta) { sta = container_of(pubsta, struct sta_info, sta); @@ -1300,6 +1369,7 @@ void ieee80211_tx_rate_update(struct ieee80211_hw *hw, .sta = pubsta, }; + ieee80211_tx_status_drop_invalid_rates(&status); rate_control_tx_status(local, &status); if (ieee80211_hw_check(&local->hw, HAS_RATE_CONTROL)) From e588e83e5c7a0f6d7a77dbbb58927c8c8cfd598e Mon Sep 17 00:00:00 2001 From: Zhao Li Date: Thu, 10 Sep 2026 14:54:02 +0800 Subject: [PATCH 173/365] wifi: mac80211: shut down RX BA session timer on teardown Tearing down an RX BA session can leave the session timer queued on a freed tid_rx. Later timer-wheel processing and the timer callback itself then run against freed memory. An RX BAR handler can dereference tid_rx under RCU before teardown unpublishes it and stay in its read-side critical section while teardown deletes the session timer. The stale handler can then rearm the timer after timer_delete_sync() returns and before the RCU callback frees tid_rx. Use timer_shutdown_sync() so rearm attempts become ineffective once teardown starts. Keep reorder_timer unchanged: teardown marks the session removed under reorder_lock and its release path checks that state before rearming. session_timer has no equivalent guard because ieee80211_rx_h_ctrl() rearms it before taking reorder_lock. Fixes: a87f736d942c ("mac80211: use RCU for RX aggregation") Cc: stable@kernel.org # 5.15+ Assisted-by: LLM sparse kasan Signed-off-by: Zhao Li Signed-off-by: Johannes Berg --- net/mac80211/agg-rx.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/net/mac80211/agg-rx.c b/net/mac80211/agg-rx.c index 9629e00069a1..01d6703767c6 100644 --- a/net/mac80211/agg-rx.c +++ b/net/mac80211/agg-rx.c @@ -102,7 +102,7 @@ void __ieee80211_stop_rx_ba_session(struct sta_info *sta, u16 tid, if (!tid_rx) return; - timer_delete_sync(&tid_rx->session_timer); + timer_shutdown_sync(&tid_rx->session_timer); /* make sure ieee80211_sta_reorder_release() doesn't re-arm the timer */ spin_lock_bh(&tid_rx->reorder_lock); From a10a2f80476d166fd81c361ec0319493671b694e Mon Sep 17 00:00:00 2001 From: Zhao Li Date: Thu, 10 Sep 2026 14:52:51 +0800 Subject: [PATCH 174/365] wifi: mac80211: keep fallback association elements alive Associating to a non-transmitted MBSSID profile reads freed memory while determining the operating channel and applying HT capabilities. When an association response omits WMM, HT, or VHT information, ieee80211_assoc_config_link() parses the cached BSS IEs and copies the missing element pointers out of that fallback parse. For a non-transmitted MBSSID profile those pointers refer to merged-profile storage owned by the parse object, and the fallback block frees that object at its end. The copied pointers are used afterwards to determine the operating channel, configure bandwidth, and apply HT capabilities. Keep the fallback parse result alive until association setup finishes. Fixes: 5023b14cf4df ("mac80211: support profile split between elements") Cc: stable@kernel.org Assisted-by: LLM sparse kasan [reorder variable declaration] Signed-off-by: Zhao Li Signed-off-by: Johannes Berg --- net/mac80211/mlme.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/net/mac80211/mlme.c b/net/mac80211/mlme.c index f51167f0fc46..3c31c16860ab 100644 --- a/net/mac80211/mlme.c +++ b/net/mac80211/mlme.c @@ -5877,6 +5877,7 @@ static bool ieee80211_assoc_config_link(struct ieee80211_link_data *link, bool is_6ghz = cbss->channel->band == NL80211_BAND_6GHZ; bool is_s1g = cbss->channel->band == NL80211_BAND_S1GHZ; const struct cfg80211_bss_ies *bss_ies = NULL; + struct ieee802_11_elems *bss_elems = NULL; struct ieee80211_supported_band *sband; struct ieee802_11_elems *elems; u16 capab_info; @@ -6007,7 +6008,6 @@ static bool ieee80211_assoc_config_link(struct ieee80211_link_data *link, (is_5ghz && link->u.mgd.conn.mode >= IEEE80211_CONN_MODE_VHT && (!elems->vht_cap_elem || !elems->vht_operation)))) { const struct cfg80211_bss_ies *ies; - struct ieee802_11_elems *bss_elems; rcu_read_lock(); ies = rcu_dereference(cbss->ies); @@ -6069,7 +6069,6 @@ static bool ieee80211_assoc_config_link(struct ieee80211_link_data *link, "AP bug: VHT operation missing from AssocResp\n"); } } - kfree(bss_elems); } /* @@ -6342,6 +6341,7 @@ static bool ieee80211_assoc_config_link(struct ieee80211_link_data *link, ret = true; out: kfree(elems); + kfree(bss_elems); kfree(bss_ies); return ret; } From cc45f313d85533d647ce619326302aad6f797c14 Mon Sep 17 00:00:00 2001 From: Zhao Li Date: Mon, 14 Sep 2026 23:28:25 +0800 Subject: [PATCH 175/365] wifi: cfg80211: preserve hidden-group beacon IE ownership A hidden-group probe-response entry borrows its beacon IEs from the group's beacon entry. cfg80211_update_assoc_bss_entry() can pass such an entry to cfg80211_update_known_bss() when merging scan entries after a channel switch. Copying the borrowed beacon_ies into the associated entry gives the same allocation two owners. Releasing the group's beacon entry can then leave the associated entry pointing to freed IEs, which are freed again when that entry is released. Skip beacon IE replacement when the source entry belongs to a hidden group. Preserve probe-response processing, the subsequent metadata updates, and the existing rejection path for conflicting hidden beacon updates. Fixes: 0afd425b1b64 ("cfg80211: fix duplicated scan entries after channel switch") Reported-by: Yilin Zhang Closes: https://lore.kernel.org/linux-wireless/20260901080427.1983015-1-yilinzhang@moonshot.ai/ Assisted-by: Codex:gpt-6-astra Signed-off-by: Zhao Li Signed-off-by: Johannes Berg --- net/wireless/scan.c | 22 ++++++++++++---------- 1 file changed, 12 insertions(+), 10 deletions(-) diff --git a/net/wireless/scan.c b/net/wireless/scan.c index caa9c6495f20..d35c826c4b1a 100644 --- a/net/wireless/scan.c +++ b/net/wireless/scan.c @@ -1992,20 +1992,22 @@ cfg80211_update_known_bss(struct cfg80211_registered_device *rdev, return false; } - old = rcu_access_pointer(known->pub.beacon_ies); + if (!new->pub.hidden_beacon_bss) { + old = rcu_access_pointer(known->pub.beacon_ies); - rcu_assign_pointer(known->pub.beacon_ies, new->pub.beacon_ies); + rcu_assign_pointer(known->pub.beacon_ies, new->pub.beacon_ies); - /* Override IEs if they were from a beacon before */ - if (old == rcu_access_pointer(known->pub.ies)) - rcu_assign_pointer(known->pub.ies, new->pub.beacon_ies); + /* Override IEs if they were from a beacon before */ + if (old == rcu_access_pointer(known->pub.ies)) + rcu_assign_pointer(known->pub.ies, new->pub.beacon_ies); - cfg80211_update_hidden_bsses(known, - rcu_access_pointer(new->pub.beacon_ies), - old); + cfg80211_update_hidden_bsses(known, + rcu_access_pointer(new->pub.beacon_ies), + old); - if (old) - kfree_rcu((struct cfg80211_bss_ies *)old, rcu_head); + if (old) + kfree_rcu((struct cfg80211_bss_ies *)old, rcu_head); + } } known->pub.beacon_interval = new->pub.beacon_interval; From 7f93ca31f7fd9b42e7d692f59514436d448ac908 Mon Sep 17 00:00:00 2001 From: Felix Fietkau Date: Wed, 16 Sep 2026 11:48:32 +0200 Subject: [PATCH 176/365] wifi: mac80211: prevent AP VLAN tx from other interfaces Since the station lookup was factored out of ieee80211_build_hdr(), ieee80211_lookup_ra_sta() uses sta_info_get_bss() for all AP and AP VLAN frames, not only for control port frames. A unicast frame sent on one interface thus matches a station of a different VLAN of the same BSS and is transmitted to it. Only match stations of the interface the frame is sent on, except for control port frames, since those may be sent for VLAN stations directly on the BSS interface. Fixes: 97ffe75791b3 ("mac80211: factor out station lookup from ieee80211_build_hdr()") Signed-off-by: Felix Fietkau Link: https://patch.msgid.link/20260916094833.164219-1-nbd@nbd.name Signed-off-by: Johannes Berg --- net/mac80211/ieee80211_i.h | 2 +- net/mac80211/tx.c | 24 +++++++++++++++--------- 2 files changed, 16 insertions(+), 10 deletions(-) diff --git a/net/mac80211/ieee80211_i.h b/net/mac80211/ieee80211_i.h index d05f59467399..9374c2942489 100644 --- a/net/mac80211/ieee80211_i.h +++ b/net/mac80211/ieee80211_i.h @@ -2475,7 +2475,7 @@ static inline bool ieee80211_require_encrypted_assoc(__le16 fc, /* sta_out needs to be checked for ERR_PTR() before using */ int ieee80211_lookup_ra_sta(struct ieee80211_sub_if_data *sdata, struct sk_buff *skb, - struct sta_info **sta_out); + struct sta_info **sta_out, bool bss); static inline void ieee80211_tx_skb_tid_band(struct ieee80211_sub_if_data *sdata, diff --git a/net/mac80211/tx.c b/net/mac80211/tx.c index 814399989b5e..c8217f6d146e 100644 --- a/net/mac80211/tx.c +++ b/net/mac80211/tx.c @@ -2532,7 +2532,7 @@ static inline bool ieee80211_is_tdls_setup(struct sk_buff *skb) int ieee80211_lookup_ra_sta(struct ieee80211_sub_if_data *sdata, struct sk_buff *skb, - struct sta_info **sta_out) + struct sta_info **sta_out, bool bss) { struct sta_info *sta; @@ -2553,7 +2553,10 @@ int ieee80211_lookup_ra_sta(struct ieee80211_sub_if_data *sdata, *sta_out = ERR_PTR(-ENOENT); return 0; } - sta = sta_info_get_bss(sdata, skb->data); + if (bss) + sta = sta_info_get_bss(sdata, skb->data); + else + sta = sta_info_get(sdata, skb->data); break; #ifdef CONFIG_MAC80211_MESH case NL80211_IFTYPE_MESH_POINT: @@ -4419,7 +4422,8 @@ void __ieee80211_subif_start_xmit(struct sk_buff *skb, ieee80211_mesh_xmit_fast(sdata, skb, ctrl_flags)) goto out; - if (ieee80211_lookup_ra_sta(sdata, skb, &sta)) + if (ieee80211_lookup_ra_sta(sdata, skb, &sta, + skb->protocol == sdata->control_port_protocol)) goto out_free; if (IS_ERR(sta)) @@ -4848,7 +4852,10 @@ static void __ieee80211_subif_start_xmit_8023(struct sk_buff *skb, rcu_read_lock(); - if (ieee80211_lookup_ra_sta(sdata, skb, &sta)) { + if (unlikely(sdata->control_port_protocol == ehdr->h_proto)) + goto skip_offload; + + if (ieee80211_lookup_ra_sta(sdata, skb, &sta, false)) { kfree_skb(skb); goto out; } @@ -4869,8 +4876,7 @@ static void __ieee80211_subif_start_xmit_8023(struct sk_buff *skb, link = &sdata->deflink; key = rcu_dereference(link->default_multicast_key); } else if (unlikely(IS_ERR_OR_NULL(sta) || !sta->uploaded || - !test_sta_flag(sta, WLAN_STA_AUTHORIZED) || - sdata->control_port_protocol == ehdr->h_proto)) { + !test_sta_flag(sta, WLAN_STA_AUTHORIZED))) { goto skip_offload; } else { key = rcu_dereference(sta->ptk[sta->ptk_idx]); @@ -4931,7 +4937,7 @@ ieee80211_build_data_template(struct ieee80211_sub_if_data *sdata, rcu_read_lock(); - if (ieee80211_lookup_ra_sta(sdata, skb, &sta)) { + if (ieee80211_lookup_ra_sta(sdata, skb, &sta, false)) { kfree_skb(skb); skb = ERR_PTR(-EINVAL); goto out; @@ -5002,7 +5008,7 @@ static bool ieee80211_tx_pending_skb(struct ieee80211_local *local, } result = ieee80211_tx(sdata, NULL, skb, true); } else if (info->flags & IEEE80211_TX_CTL_HW_80211_ENCAP) { - if (ieee80211_lookup_ra_sta(sdata, skb, &sta)) { + if (ieee80211_lookup_ra_sta(sdata, skb, &sta, true)) { dev_kfree_skb(skb); return true; } @@ -6639,7 +6645,7 @@ int ieee80211_tx_control_port(struct wiphy *wiphy, struct net_device *dev, * AF_PACKET */ rcu_read_lock(); - err = ieee80211_lookup_ra_sta(sdata, skb, &sta); + err = ieee80211_lookup_ra_sta(sdata, skb, &sta, true); if (err) { dev_kfree_skb(skb); rcu_read_unlock(); From 0d6526f82c3cdefcca47f73f5fc08dc6f335eac6 Mon Sep 17 00:00:00 2001 From: Tim Chen Date: Mon, 21 Sep 2026 17:37:22 -0700 Subject: [PATCH 177/365] sched/cache: Keep nr_pref_llc_running in the runnable domain, to fix LLC mis-scheduling bug alb_break_llc() decides whether to break LLC preference during active load balance. It does so by testing that every runnable fair task on the source rq prefers its LLC: env->src_rq->nr_pref_llc_running == env->src_rq->cfs.h_nr_runnable But the two counters cover different sets. nr_pref_llc_running is updated in account_llc_enqueue()/account_llc_dequeue(), next to cfs_rq->nr_queued, so it follows queued tasks. h_nr_runnable is updated in set_delayed()/ clear_delayed() and drops delay-dequeued tasks. So under DELAY_DEQUEUE, a preferring task that goes to sleep stays counted in nr_pref_llc_running while h_nr_runnable falls. The equality then breaks, alb_break_llc() returns false, and active balance is free to pull a task off its preferred LLC. Active balance only moves runnable tasks, and this is the only LLC check it consults: once the stopper runs, LBF_ACTIVE_LB skips the per-task test in can_migrate_task(). The runnable set is the one we want. Fix it on the counter side. A task should be counted in nr_pref_llc_running exactly while it is both queued on its preferred LLC (pref_llc_queued) and runnable (!sched_delayed). Define that membership once in task_pref_llc_runnable(), and adjust the counter only through pref_llc_running_inc()/pref_llc_running_dec() from the four sites that change either input: account_llc_enqueue(), account_llc_dequeue(), set_delayed() and clear_delayed(). Gating every update on the same predicate keeps the delay, wake and dequeue paths from double-counting or underflowing; see the comments at those sites for the ordering. nr_llc_running and sd->llc_counts are not touched and stay on queued semantics. Fixes: 714059f79ff0 ("sched/cache: Handle moving single tasks to/from their preferred LLC") Closes: https://lore.kernel.org/lkml/20260827135000.735138-1-zhanxusheng@xiaomi.com/ Reported-by: Zhan Xusheng Suggested-by: Chen Yu Signed-off-by: Tim Chen Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Reviewed-by: Kayra Cizmeci Cc: # v7.2.x Link: https://patch.msgid.link/06af61afedac32e6477f57feb4d658f6c411c3af.1790035273.git.tim.c.chen@linux.intel.com --- kernel/sched/fair.c | 63 +++++++++++++++++++++++++++++++++++++++++---- 1 file changed, 58 insertions(+), 5 deletions(-) diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index 7455a83a6a99..de3d589fa8eb 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -1546,6 +1546,28 @@ static bool invalid_llc_nr(struct mm_struct *mm, struct task_struct *p, (scale * per_cpu(sd_llc_size, cpu))); } +/* + * A task counts in nr_pref_llc_running while it is queued on its preferred + * LLC (pref_llc_queued) and runnable (!sched_delayed), keeping the counter in + * the runnable domain so alb_break_llc() can compare it with h_nr_runnable. + */ +static bool task_pref_llc_runnable(struct task_struct *p) +{ + return p->pref_llc_queued && !p->se.sched_delayed; +} + +static void pref_llc_running_inc(struct rq *rq, struct task_struct *p) +{ + if (task_pref_llc_runnable(p)) + rq->nr_pref_llc_running++; +} + +static void pref_llc_running_dec(struct rq *rq, struct task_struct *p) +{ + if (task_pref_llc_runnable(p)) + rq->nr_pref_llc_running--; +} + static void account_llc_enqueue(struct rq *rq, struct task_struct *p) { int pref_llc, pref_llc_queued; @@ -1557,7 +1579,6 @@ static void account_llc_enqueue(struct rq *rq, struct task_struct *p) pref_llc_queued = (pref_llc == task_llc(p)); rq->nr_llc_running++; - rq->nr_pref_llc_running += pref_llc_queued; /* * Record whether p is enqueued on its preferred @@ -1575,6 +1596,9 @@ static void account_llc_enqueue(struct rq *rq, struct task_struct *p) */ p->pref_llc_queued = pref_llc_queued; + /* Skipped while delayed; clear_delayed() adds it back on wake. */ + pref_llc_running_inc(rq, p); + sd = rcu_dereference_all(rq->sd); if (sd && (unsigned int)pref_llc < sd->llc_max) sd->llc_counts[pref_llc]++; @@ -1591,7 +1615,12 @@ static void account_llc_dequeue(struct rq *rq, struct task_struct *p) rq->nr_llc_running--; if (p->pref_llc_queued) { - rq->nr_pref_llc_running--; + /* + * Skipped if still delayed (set_delayed() already removed it); + * clearing pref_llc_queued below also stops clear_delayed() + * from re-adding it. + */ + pref_llc_running_dec(rq, p); /* * Update the status in case * other logic might query @@ -1995,6 +2024,7 @@ void init_sched_mm(struct task_struct *p) * polluting account_llc_enqueue(). */ p->preferred_llc = -1; + p->pref_llc_queued = 0; } #else /* CONFIG_SCHED_CACHE */ @@ -2016,6 +2046,10 @@ static void account_llc_enqueue(struct rq *rq, struct task_struct *p) {} static void account_llc_dequeue(struct rq *rq, struct task_struct *p) {} +static void pref_llc_running_inc(struct rq *rq, struct task_struct *p) {} + +static void pref_llc_running_dec(struct rq *rq, struct task_struct *p) {} + #endif /* CONFIG_SCHED_CACHE */ /* @@ -6390,15 +6424,27 @@ static __always_inline void return_cfs_rq_runtime(struct cfs_rq *cfs_rq); static void set_delayed(struct sched_entity *se) { - se->sched_delayed = 1; - /* * Delayed se of cfs_rq have no tasks queued on them. * Do not adjust h_nr_runnable since __dequeue_task() * will account it for blocked tasks. + * + * This check can be removed because when flat pick + * patches get merged as only task can get delayed, + * same for clear_delayed(). */ - if (!entity_is_task(se)) + if (!entity_is_task(se)) { + se->sched_delayed = 1; return; + } + + /* + * Drop a task leaving the runnable set. + * Needs to be called before sched_delayed is set. + * clear_delayed() mirrors this after clearing the flag. + */ + pref_llc_running_dec(rq_of(cfs_rq_of(se)), task_of(se)); + se->sched_delayed = 1; for_each_sched_entity(se) { struct cfs_rq *cfs_rq = cfs_rq_of(se); @@ -6420,6 +6466,13 @@ static void clear_delayed(struct sched_entity *se) if (!entity_is_task(se)) return; + /* + * Re-add on wake, after sched_delayed is cleared. On a final delayed + * dequeue account_llc_dequeue() already cleared pref_llc_queued, so + * this does nothing. + */ + pref_llc_running_inc(rq_of(cfs_rq_of(se)), task_of(se)); + for_each_sched_entity(se) { struct cfs_rq *cfs_rq = cfs_rq_of(se); From d6013e2465d98d524b030a81c1223882a1bb7e4c Mon Sep 17 00:00:00 2001 From: Lu Wang Date: Mon, 21 Sep 2026 17:37:23 -0700 Subject: [PATCH 178/365] sched/cache: Honor migrate_llc_task semantics in active load balance, to fix LLC mis-scheduling bug Cache aware scheduling introduced the migrate_llc_task migration type to direct tasks toward their preferred LLC, but its semantics can be lost when passive load balance falls back to active load balance (ALB). This may allow ALB to select a candidate whose preferred LLC does not match the destination, moving it away from its preferred LLC. Example scenario: src_rq has two runnable tasks, p1 and p2. p1 prefers dst_rq (dst_llc), while p2 prefers src_rq (src_llc). In this case, migrate_llc_task is set because src_rq has at least one task, p1, that wants to migrate to dst_rq. In ALB, can_migrate_task() finds p2 and returns true for it, thus moving p2 out of its preferred LLC. Solution: The CPU stopper in ALB constructs a fresh lb_env that does not inherit migration_type from the passive load-balance pass. Two approaches are possible: (a) Add a new member to struct rq so ALB can inherit migrate_llc_task from the passive LB that triggered it. (b) Define a new flag LBF_ACTIVE_LB_LLC and select the stopper callback at kick time to preserve the migration semantics across the asynchronous boundary. We choose (b) because it avoids passing migration_type through the stopper, which would affect the meaning of migration_type for delayed-dequeue tasks. Fixes: e4c9a4cb244a ("sched/cache: Add migrate_llc_task migration type for cache-aware balancing") Suggested-by: Chen Yu Signed-off-by: Lu Wang Signed-off-by: Tim Chen Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Reviewed-by: Tim Chen Reviewed-by: Chen Yu Cc: # v7.2.x Link: https://patch.msgid.link/cb39f64a17fc2b76097264aaec74a2d6dfff4315.1790035273.git.tim.c.chen@linux.intel.com --- kernel/sched/fair.c | 57 ++++++++++++++++++++++++++++++++++++++++----- 1 file changed, 51 insertions(+), 6 deletions(-) diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index de3d589fa8eb..514bd54ccd56 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -10448,6 +10448,7 @@ enum migration_type { #define LBF_SOME_PINNED 0x08 #define LBF_ACTIVE_LB 0x10 #define LBF_LLC_PINNED 0x20 +#define LBF_ACTIVE_LB_LLC 0x40 struct lb_env { struct sched_domain *sd; @@ -10866,6 +10867,21 @@ alb_break_llc(struct lb_env *env) return false; } +/* + * Returns true if p's preferred LLC does not match the destination CPU + * under migrate_llc_task semantics. Passive LB passes migrate_llc_task + * in env->migration_type, while active LB carries LBF_ACTIVE_LB_LLC in + * env->flags to avoid overwriting env->migration_type. + */ +static inline bool +migrate_llc_task_wrong_dst(struct task_struct *p, struct lb_env *env) +{ + return sched_cache_enabled() && + (env->migration_type == migrate_llc_task || + env->flags & LBF_ACTIVE_LB_LLC) && + READ_ONCE(p->preferred_llc) != llc_id(env->dst_cpu); +} + /* * Check if migrating task p from env->src_cpu to * env->dst_cpu breaks LLC localiy. @@ -10894,8 +10910,7 @@ static bool migrate_degrades_llc(struct task_struct *p, struct lb_env *env) * run on env->dst_cpu, skip the tasks do not prefer * env->dst_cpu, and find the one that prefers. */ - if (env->migration_type == migrate_llc_task && - READ_ONCE(p->preferred_llc) != llc_id(env->dst_cpu)) + if (migrate_llc_task_wrong_dst(p, env)) return true; if (can_migrate_llc_task(env, p) != mig_forbid) @@ -10917,6 +10932,12 @@ alb_break_llc(struct lb_env *env) return false; } +static inline bool +migrate_llc_task_wrong_dst(struct task_struct *p, struct lb_env *env) +{ + return false; +} + static inline bool migrate_degrades_llc(struct task_struct *p, struct lb_env *env) { @@ -11016,7 +11037,7 @@ int can_migrate_task(struct task_struct *p, struct lb_env *env) * 4) too many balance attempts have failed. */ if (env->flags & LBF_ACTIVE_LB) - return 1; + return !migrate_llc_task_wrong_dst(p, env); degrades = migrate_degrades_locality(p, env); if (!degrades) { @@ -13415,6 +13436,20 @@ static int need_active_balance(struct lb_env *env) } static int active_load_balance_cpu_stop(void *data); +static int active_load_balance_llc_cpu_stop(void *data); + +/* + * migration_type is checked elsewhere to decide migration policy, so + * it shouldn't be repurposed just to flag an LLC-directed active + * balance across the stopper. Pick the callback here instead. + */ +static inline cpu_stop_fn_t alb_stop_fn(struct lb_env *env) +{ + if (env->migration_type == migrate_llc_task) + return active_load_balance_llc_cpu_stop; + + return active_load_balance_cpu_stop; +} static int should_we_balance(struct lb_env *env) { @@ -13760,7 +13795,7 @@ static int sched_balance_rq(int this_cpu, struct rq *this_rq, } if (active_balance) { stop_one_cpu_nowait(cpu_of(busiest), - active_load_balance_cpu_stop, busiest, + alb_stop_fn(&env), busiest, &busiest->active_balance_work); } preempt_enable(); @@ -13865,7 +13900,7 @@ update_next_balance(struct sched_domain *sd, unsigned long *next_balance) * least 1 task to be running on each physical CPU where possible, and * avoids physical / logical imbalances. */ -static int active_load_balance_cpu_stop(void *data) +static int __active_load_balance_cpu_stop(void *data, unsigned int lb_flags) { struct rq *busiest_rq = data; int busiest_cpu = cpu_of(busiest_rq); @@ -13915,7 +13950,7 @@ static int active_load_balance_cpu_stop(void *data) .src_cpu = busiest_rq->cpu, .src_rq = busiest_rq, .idle = CPU_IDLE, - .flags = LBF_ACTIVE_LB, + .flags = LBF_ACTIVE_LB | lb_flags, }; schedstat_inc(sd->alb_count); @@ -13943,6 +13978,16 @@ static int active_load_balance_cpu_stop(void *data) return 0; } +static int active_load_balance_cpu_stop(void *data) +{ + return __active_load_balance_cpu_stop(data, 0); +} + +static int active_load_balance_llc_cpu_stop(void *data) +{ + return __active_load_balance_cpu_stop(data, LBF_ACTIVE_LB_LLC); +} + /* * Scale the max sched_balance_rq interval with the number of CPUs in the system. * This trades load-balance latency on larger machines for less cross talk. From 28f9c0e0a0b94c5d3e1b634db545f6e1f94858c5 Mon Sep 17 00:00:00 2001 From: Tim Chen Date: Mon, 21 Sep 2026 17:37:24 -0700 Subject: [PATCH 179/365] sched/cache: Decouple sched_cache_group from mm to fix UAF Currently the sched cache grouping is by mm and the scheduling statistics sched_cache_stat lives in the mm structure. This ties the life cycle of scheduling stats with mm. In account_mm_sched(), the scheduling stats are accessed by task->mm->sc_stat. However, a task may be switching mm on one CPU when another CPU is running account_mm_sched(), and possibly accessing the old mm that was freed. This problem was found when running tests with KASAN by Hyunwoo: https://lore.kernel.org/lkml/apPb-Dr4nPYuHQOK@v4bel/ Instead of serializing the mm access by introducing extra acquisition of rq lock in the mm free path, extract sched_cache_stat from mm_struct, rename it as sched_cache_group and manage its life cycle apart from mm_struct with its own ref counting. This allows us in the next patch access sched_cache_group directly from task, and add a refcount on sched_cache_group when a task links to it. This prevents the use after free issue when accessing stale and released old mm and its sched cache stat a task switches to a new mm while account_mm_sched() is done elsewhere. The other benefit of this restructure is in the future, the grouping of tasks to a LLC would have the flexibility to be associated with a user defined grouping, or cgroup, cookie group, numa_group or others instead of just with a single mm address space. Rename sched_cache_stat to sched_cache_group and turn it into a refcounted object allocated from mm_struct. The mm_struct now holds a pointer (sched_cache_grp) to this object instead of embedding it. Fixes: df0d98475954 ("sched/cache: Introduce infrastructure for cache-aware load balancing") Closes: https://lore.kernel.org/lkml/apPb-Dr4nPYuHQOK@v4bel/ Closes: https://lore.kernel.org/all/343a7e07-7fad-4979-9c9b-82ec038c293c@linux.dev/ Reported-by: Hyunwoo Kim Reported-by: Zenghui Yu (Huawei) Co-developed-by: Chen Yu Signed-off-by: Chen Yu Signed-off-by: Tim Chen Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Cc: #7.2.x Link: https://patch.msgid.link/91fd1e3266707c865bc9abecfb3e17bc676712df.1790035273.git.tim.c.chen@linux.intel.com --- include/linux/mm_types.h | 15 ++-- include/linux/sched.h | 6 +- kernel/exit.c | 11 ++- kernel/sched/fair.c | 173 ++++++++++++++++++++++++++++----------- 4 files changed, 144 insertions(+), 61 deletions(-) diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h index 6d815f6440c9..f3e5a2fadbe5 100644 --- a/include/linux/mm_types.h +++ b/include/linux/mm_types.h @@ -1226,7 +1226,7 @@ struct mm_struct { struct mm_mm_cid mm_cid; /* sched_cache related statistics */ - struct sched_cache_stat sc_stat; + struct sched_cache_group *sched_cache_grp; #ifdef CONFIG_MMU atomic_long_t pgtables_bytes; /* size of all page tables */ #endif @@ -1624,8 +1624,9 @@ static inline unsigned int mm_cid_size(void) #endif /* CONFIG_SCHED_MM_CID */ #ifdef CONFIG_SCHED_CACHE -void mm_init_sched(struct mm_struct *mm, - struct sched_cache_time __percpu *pcpu_sched); +int mm_init_sched(struct mm_struct *mm, + struct sched_cache_time __percpu *pcpu_sched); +void mm_destroy_sched(struct mm_struct *mm); static inline int mm_alloc_sched_noprof(struct mm_struct *mm) { @@ -1635,17 +1636,11 @@ static inline int mm_alloc_sched_noprof(struct mm_struct *mm) if (!pcpu_sched) return -ENOMEM; - mm_init_sched(mm, pcpu_sched); - return 0; + return mm_init_sched(mm, pcpu_sched); } #define mm_alloc_sched(...) alloc_hooks(mm_alloc_sched_noprof(__VA_ARGS__)) -static inline void mm_destroy_sched(struct mm_struct *mm) -{ - free_percpu(mm->sc_stat.pcpu_sched); - mm->sc_stat.pcpu_sched = NULL; -} #else /* !CONFIG_SCHED_CACHE */ static inline int mm_alloc_sched(struct mm_struct *mm) { return 0; } diff --git a/include/linux/sched.h b/include/linux/sched.h index 705970d07614..e14ad43522c8 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -2405,7 +2405,7 @@ struct sched_cache_time { unsigned long epoch; }; -struct sched_cache_stat { +struct sched_cache_group { struct sched_cache_time __percpu *pcpu_sched; raw_spinlock_t lock; unsigned long epoch; @@ -2413,11 +2413,13 @@ struct sched_cache_stat { unsigned long next_scan; unsigned long footprint; int cpu; + refcount_t refcnt; + struct rcu_head rcu; } ____cacheline_aligned_in_smp; #else -struct sched_cache_stat { }; +struct sched_cache_group { }; #endif diff --git a/kernel/exit.c b/kernel/exit.c index 424c44a42a4d..024350e9b48c 100644 --- a/kernel/exit.c +++ b/kernel/exit.c @@ -558,18 +558,23 @@ void mm_update_next_owner(struct mm_struct *mm) */ static void exit_mm_sched_cache(struct mm_struct *mm) { + struct sched_cache_group *grp; unsigned long fp, sub; if (!current->total_numa_faults) return; /* * No lock protection due to performance considerations. - * Make sure mm->sc_stat.footprint does not become + * Make sure the group footprint does not become * negative. */ - fp = READ_ONCE(mm->sc_stat.footprint); + grp = READ_ONCE(mm->sched_cache_grp); + if (!grp) + return; + + fp = READ_ONCE(grp->footprint); sub = min(fp, current->total_numa_faults); - WRITE_ONCE(mm->sc_stat.footprint, fp - sub); + WRITE_ONCE(grp->footprint, fp - sub); } #else static inline void exit_mm_sched_cache(struct mm_struct *mm) diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index 514bd54ccd56..f0a9586332fd 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -1492,12 +1492,17 @@ static bool exceed_llc_capacity(struct mm_struct *mm, int cpu) return true; if (static_branch_likely(&sched_numa_balancing)) { + struct sched_cache_group *grp = READ_ONCE(mm->sched_cache_grp); + + if (!grp) + return true; + /* * TBD: RDT exclusive LLC ways reserved should be * excluded. */ llc = sd->llc_bytes; - footprint = READ_ONCE(mm->sc_stat.footprint); + footprint = READ_ONCE(grp->footprint); /* * Scale the LLC size by 256*llc_aggr_tolerance @@ -1529,6 +1534,7 @@ static bool exceed_llc_capacity(struct mm_struct *mm, int cpu) static bool invalid_llc_nr(struct mm_struct *mm, struct task_struct *p, int cpu) { + struct sched_cache_group *grp; int scale; if (get_nr_threads(p) <= 1) @@ -1542,7 +1548,11 @@ static bool invalid_llc_nr(struct mm_struct *mm, struct task_struct *p, if (scale == INT_MAX) return false; - return !fits_capacity((mm->sc_stat.nr_running_avg * cpu_smt_num_threads), + grp = READ_ONCE(mm->sched_cache_grp); + if (!grp) + return true; + + return !fits_capacity((READ_ONCE(grp->nr_running_avg) * cpu_smt_num_threads), (scale * per_cpu(sd_llc_size, cpu))); } @@ -1648,12 +1658,20 @@ static void account_llc_dequeue(struct rq *rq, struct task_struct *p) } } -void mm_init_sched(struct mm_struct *mm, - struct sched_cache_time __percpu *_pcpu_sched) +int mm_init_sched(struct mm_struct *mm, + struct sched_cache_time __percpu *_pcpu_sched) { + struct sched_cache_group *grp; unsigned long epoch = 0; int i; + grp = kzalloc_obj(*grp); + if (!grp) { + free_percpu(_pcpu_sched); + mm->sched_cache_grp = NULL; + return -ENOMEM; + } + for_each_possible_cpu(i) { struct sched_cache_time *pcpu_sched = per_cpu_ptr(_pcpu_sched, i); struct rq *rq = cpu_rq(i); @@ -1664,18 +1682,51 @@ void mm_init_sched(struct mm_struct *mm, epoch = rq->cpu_epoch; } - raw_spin_lock_init(&mm->sc_stat.lock); - mm->sc_stat.epoch = epoch; - mm->sc_stat.cpu = -1; - mm->sc_stat.next_scan = jiffies; - mm->sc_stat.nr_running_avg = 0; - mm->sc_stat.footprint = 0; + raw_spin_lock_init(&grp->lock); + grp->epoch = epoch; + grp->cpu = -1; + grp->next_scan = jiffies; + grp->nr_running_avg = 0; + grp->footprint = 0; + refcount_set(&grp->refcnt, 1); /* - * The update to mm->sc_stat should not be reordered - * before initialization to mm's other fields, in case + * The update to grp->pcpu_sched should not be reordered + * before initialization to grp's other fields, in case * the readers may get invalid mm_sched_epoch, etc. */ - smp_store_release(&mm->sc_stat.pcpu_sched, _pcpu_sched); + smp_store_release(&grp->pcpu_sched, _pcpu_sched); + /* + * Publish the group last. Not every reader qualifies it by + * grp->pcpu_sched - can_migrate_llc_task() only checks that the + * pointer is non-NULL before reading grp->footprint and + * grp->nr_running_avg - so a reachable group must already be + * fully initialized. + */ + smp_store_release(&mm->sched_cache_grp, grp); + return 0; +} + +static void sched_cache_group_free_rcu(struct rcu_head *rcu) +{ + struct sched_cache_group *grp = + container_of(rcu, struct sched_cache_group, rcu); + + free_percpu(grp->pcpu_sched); + kfree(grp); +} + +static void sched_cache_group_put(struct sched_cache_group *grp) +{ + if (!grp || !refcount_dec_and_test(&grp->refcnt)) + return; + + call_rcu(&grp->rcu, sched_cache_group_free_rcu); +} + +void mm_destroy_sched(struct mm_struct *mm) +{ + sched_cache_group_put(mm->sched_cache_grp); + mm->sched_cache_grp = NULL; } /* because why would C be fully specified */ @@ -1729,11 +1780,16 @@ static unsigned long fraction_mm_sched(struct rq *rq, static int get_pref_llc(struct task_struct *p, struct mm_struct *mm) { int mm_sched_llc = -1, mm_sched_cpu; + struct sched_cache_group *grp; if (!mm) return -1; - mm_sched_cpu = READ_ONCE(mm->sc_stat.cpu); + grp = READ_ONCE(mm->sched_cache_grp); + if (!grp) + return -1; + + mm_sched_cpu = READ_ONCE(grp->cpu); if (mm_sched_cpu != -1) { mm_sched_llc = llc_id(mm_sched_cpu); @@ -1764,6 +1820,7 @@ static inline void account_mm_sched(struct rq *rq, struct task_struct *p, s64 delta_exec) { struct sched_cache_time *pcpu_sched; + struct sched_cache_group *grp; struct mm_struct *mm = p->mm; int mm_sched_llc = -1; unsigned long epoch; @@ -1777,10 +1834,14 @@ void account_mm_sched(struct rq *rq, struct task_struct *p, s64 delta_exec) * init_task, kthreads and user thread created * by user_mode_thread() don't have mm. */ - if (!mm || !mm->sc_stat.pcpu_sched) + if (!mm) return; - pcpu_sched = per_cpu_ptr(mm->sc_stat.pcpu_sched, cpu_of(rq)); + grp = READ_ONCE(mm->sched_cache_grp); + if (!grp || !grp->pcpu_sched) + return; + + pcpu_sched = per_cpu_ptr(grp->pcpu_sched, cpu_of(rq)); scoped_guard (raw_spinlock, &rq->cpu_epoch_lock) { __update_mm_sched(rq, pcpu_sched); @@ -1793,11 +1854,11 @@ void account_mm_sched(struct rq *rq, struct task_struct *p, s64 delta_exec) * If this process hasn't hit task_cache_work() for a while invalidate * its preferred state. */ - if ((long)(epoch - READ_ONCE(mm->sc_stat.epoch)) > llc_epoch_affinity_timeout || + if ((long)(epoch - READ_ONCE(grp->epoch)) > llc_epoch_affinity_timeout || invalid_llc_nr(mm, p, cpu_of(rq)) || exceed_llc_capacity(mm, cpu_of(rq))) { - if (READ_ONCE(mm->sc_stat.cpu) != -1) - WRITE_ONCE(mm->sc_stat.cpu, -1); + if (READ_ONCE(grp->cpu) != -1) + WRITE_ONCE(grp->cpu, -1); } mm_sched_llc = get_pref_llc(p, mm); @@ -1814,30 +1875,35 @@ void account_mm_sched(struct rq *rq, struct task_struct *p, s64 delta_exec) static void task_tick_cache(struct rq *rq, struct task_struct *p) { struct callback_head *work = &p->cache_work; + struct sched_cache_group *grp; struct mm_struct *mm = p->mm; unsigned long epoch; if (!sched_cache_enabled()) return; - if (!mm || p->flags & PF_KTHREAD || - !mm->sc_stat.pcpu_sched) + if (!mm || p->flags & PF_KTHREAD) + return; + + grp = READ_ONCE(mm->sched_cache_grp); + if (!grp || !grp->pcpu_sched) return; epoch = rq->cpu_epoch; /* avoid moving backwards */ - if (time_after_eq(mm->sc_stat.epoch, epoch)) + if (time_after_eq(grp->epoch, epoch)) return; - guard(raw_spinlock)(&mm->sc_stat.lock); + guard(raw_spinlock)(&grp->lock); if (work->next == work) { task_work_add(p, work, TWA_RESUME); - WRITE_ONCE(mm->sc_stat.epoch, epoch); + WRITE_ONCE(grp->epoch, epoch); } } -static void get_scan_cpumasks(cpumask_var_t cpus, struct task_struct *p) +static void get_scan_cpumasks(cpumask_var_t cpus, struct task_struct *p, + struct sched_cache_group *grp) { #ifdef CONFIG_NUMA_BALANCING int cpu, curr_cpu, nid, pref_nid; @@ -1845,7 +1911,7 @@ static void get_scan_cpumasks(cpumask_var_t cpus, struct task_struct *p) if (!static_branch_likely(&sched_numa_balancing)) goto out; - cpu = READ_ONCE(p->mm->sc_stat.cpu); + cpu = READ_ONCE(grp->cpu); if (cpu != -1) nid = cpu_to_node(cpu); curr_cpu = task_cpu(p); @@ -1906,6 +1972,7 @@ static void task_cache_work(struct callback_head *work) unsigned long next_scan, now = jiffies; struct task_struct *p = current, *cur; unsigned long curr_m_a_occ = 0; + struct sched_cache_group *grp; struct mm_struct *mm = p->mm; unsigned long m_a_occ = 0; cpumask_var_t cpus; @@ -1917,12 +1984,16 @@ static void task_cache_work(struct callback_head *work) if (p->flags & PF_EXITING) return; - next_scan = READ_ONCE(mm->sc_stat.next_scan); + grp = READ_ONCE(mm->sched_cache_grp); + if (!grp) + return; + + next_scan = READ_ONCE(grp->next_scan); if (time_before(now, next_scan)) return; /* only 1 thread is allowed to scan */ - if (!try_cmpxchg(&mm->sc_stat.next_scan, &next_scan, + if (!try_cmpxchg(&grp->next_scan, &next_scan, now + max_t(unsigned long, READ_ONCE(llc_epoch_period), 1))) return; @@ -1930,8 +2001,8 @@ static void task_cache_work(struct callback_head *work) curr_cpu = task_cpu(p); if (invalid_llc_nr(mm, p, curr_cpu) || exceed_llc_capacity(mm, curr_cpu)) { - if (READ_ONCE(mm->sc_stat.cpu) != -1) - WRITE_ONCE(mm->sc_stat.cpu, -1); + if (READ_ONCE(grp->cpu) != -1) + WRITE_ONCE(grp->cpu, -1); return; } @@ -1942,7 +2013,7 @@ static void task_cache_work(struct callback_head *work) scoped_guard (cpus_read_lock) { guard(rcu)(); - get_scan_cpumasks(cpus, p); + get_scan_cpumasks(cpus, p, grp); for_each_cpu(cpu, cpus) { /* XXX sched_cluster_active */ @@ -1955,7 +2026,7 @@ static void task_cache_work(struct callback_head *work) for_each_cpu(i, sched_domain_span(sd)) { occ = fraction_mm_sched(cpu_rq(i), - per_cpu_ptr(mm->sc_stat.pcpu_sched, i)); + per_cpu_ptr(grp->pcpu_sched, i)); a_occ += occ; if (occ > m_occ) { m_occ = occ; @@ -1988,7 +2059,7 @@ static void task_cache_work(struct callback_head *work) m_a_cpu = m_cpu; } - if (llc_id(cpu) == llc_id(READ_ONCE(mm->sc_stat.cpu))) + if (llc_id(cpu) == llc_id(READ_ONCE(grp->cpu))) curr_m_a_occ = a_occ; cpumask_andnot(cpus, cpus, sched_domain_span(sd)); @@ -1997,7 +2068,7 @@ static void task_cache_work(struct callback_head *work) if (m_a_occ > (2 * curr_m_a_occ)) { /* - * Avoid switching sc_stat.cpu too fast. + * Avoid switching sched_cache_grp->cpu too fast. * The reason to choose 2X is because: * 1. It is better to keep the preferred LLC stable, * rather than changing it frequently and cause migrations @@ -2006,10 +2077,10 @@ static void task_cache_work(struct callback_head *work) * 3. 2X is chosen based on test results, as it delivers * the optimal performance gain so far. */ - WRITE_ONCE(mm->sc_stat.cpu, m_a_cpu); + WRITE_ONCE(grp->cpu, m_a_cpu); } - update_avg_scale(&mm->sc_stat.nr_running_avg, nr_running); + update_avg_scale(&grp->nr_running_avg, nr_running); free_cpumask_var(cpus); } @@ -3726,6 +3797,7 @@ static int preferred_group_nid(struct task_struct *p, int nid) static void task_numa_placement(struct task_struct *p) __context_unsafe(/* conditional locking */) { + struct sched_cache_group __maybe_unused *grp; int seq, nid, max_nid = NUMA_NO_NODE; unsigned long max_faults = 0; unsigned long fault_types[2] = { 0, 0 }; @@ -3818,19 +3890,23 @@ static void task_numa_placement(struct task_struct *p) * heuristic and occasional lost updates are tolerable. * * If a task exits, its corresponding footprint must - * be subtracted from the mm->sc_stat.footprint, otherwise - * the mm->sc_stat.footprint will not converge: - * the exiting thread's footprint remains unchanged/undecayed - * in mm->sc_stat.footprint. See exit_mm(). + * be subtracted from the mm->sched_cache_grp->footprint, + * otherwise the mm->sched_cache_grp->footprint will not + * converge: the exiting thread's footprint remains + * unchanged/undecayed in mm->sched_cache_grp->footprint. + * See exit_mm(). * * Lost updates and unsynchronized subtraction * in exit_mm() can cause footprint + diff to * go negative. Clamp to zero to prevent the * unsigned footprint from wrapping. */ - new_fp = (long)READ_ONCE(p->mm->sc_stat.footprint) + diff; - WRITE_ONCE(p->mm->sc_stat.footprint, - max(new_fp, 0L)); + grp = READ_ONCE(p->mm->sched_cache_grp); + if (!grp) + continue; + + new_fp = (long)READ_ONCE(grp->footprint) + diff; + WRITE_ONCE(grp->footprint, max(new_fp, 0L)); #endif } @@ -10778,6 +10854,7 @@ static inline bool task_misfits_asym_cpu(struct lb_env *env, struct task_struct static enum llc_mig can_migrate_llc_task(struct lb_env *env, struct task_struct *p) { + struct sched_cache_group *grp; struct mm_struct *mm; bool to_pref; int cpu, src_cpu, dst_cpu; @@ -10791,15 +10868,19 @@ static enum llc_mig can_migrate_llc_task(struct lb_env *env, if (!mm) return mig_unrestricted; - cpu = READ_ONCE(mm->sc_stat.cpu); + grp = READ_ONCE(mm->sched_cache_grp); + if (!grp) + return mig_unrestricted; + + cpu = READ_ONCE(grp->cpu); if (cpu < 0 || cpus_share_cache(src_cpu, dst_cpu)) return mig_unrestricted; /* skip cache aware load balance for too many threads */ if (invalid_llc_nr(mm, p, dst_cpu) || exceed_llc_capacity(mm, dst_cpu)) { - if (READ_ONCE(mm->sc_stat.cpu) != -1) - WRITE_ONCE(mm->sc_stat.cpu, -1); + if (READ_ONCE(grp->cpu) != -1) + WRITE_ONCE(grp->cpu, -1); return mig_unrestricted; } From b636fef85bda7d1bab9c0a45067ab1508d79d946 Mon Sep 17 00:00:00 2001 From: Tim Chen Date: Mon, 21 Sep 2026 17:37:25 -0700 Subject: [PATCH 180/365] sched/cache: Introduce task_struct->sched_cache_grp to fix UAF Add a sched_cache_grp pointer to task_struct so that scheduler code can access the cache group directly via the task, without going through mm->sched_cache_grp. This decouples the scheduler's hot-path accesses from the mm_struct. Each task holds its own refcount on the sched_cache_group, separate from the reference held by its mm_struct. The reference is acquired in copy_mm() (fork) and exec_mmap() (exec), and released in exit_mm(). This fixes the use-after-free when account_mm_sched() reaches the group through a task whose mm is being switched, as reported by Hyunwoo: https://lore.kernel.org/lkml/apPb-Dr4nPYuHQOK@v4bel/ Convert all scheduler code in fair.c and exit.c to use p->sched_cache_grp instead of p->mm->sched_cache_grp. Keep the fork/exec/exit reference management out of the generic mm paths: add sched_cache_fork(), sched_cache_fork_cleanup(), sched_cache_exec_mmap() and sched_cache_exit_mm() in kernel/sched/cache_sched.c (with empty stubs for !CONFIG_SCHED_CACHE), so fs/exec.c, kernel/fork.c and kernel/exit.c each call one helper instead of open-coding the refcounting under #ifdef. Also add sched_cache_group_get() and task_cache_group_get(). Fixes: df0d98475954 ("sched/cache: Introduce infrastructure for cache-aware load balancing") Closes: https://lore.kernel.org/lkml/apPb-Dr4nPYuHQOK@v4bel/ Closes: https://lore.kernel.org/all/343a7e07-7fad-4979-9c9b-82ec038c293c@linux.dev/ Reported-by: Hyunwoo Kim Reported-by: Zenghui Yu (Huawei) Co-developed-by: Chen Yu Signed-off-by: Chen Yu Signed-off-by: Tim Chen Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Cc: #7.2.x Link: https://patch.msgid.link/ae7081dc54736bf115215f9867abb2711a7403fb.1790035273.git.tim.c.chen@linux.intel.com --- fs/exec.c | 1 + include/linux/sched.h | 14 +++ kernel/exit.c | 33 +------ kernel/fork.c | 2 + kernel/sched/fair.c | 196 +++++++++++++++++++++++++++++------------- 5 files changed, 154 insertions(+), 92 deletions(-) diff --git a/fs/exec.c b/fs/exec.c index 819643408e6d..a5269b5e00df 100644 --- a/fs/exec.c +++ b/fs/exec.c @@ -882,6 +882,7 @@ static int exec_mmap(struct linux_binprm *bprm) active_mm = tsk->active_mm; tsk->active_mm = mm; tsk->mm = mm; + sched_cache_exec_mmap(tsk, mm); mm_init_cid(mm, tsk); exec_state = task_exec_state_replace(tsk, exec_state); /* diff --git a/include/linux/sched.h b/include/linux/sched.h index e14ad43522c8..d35ae49a991f 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1433,6 +1433,7 @@ struct task_struct { #ifdef CONFIG_SCHED_CACHE struct callback_head cache_work; + struct sched_cache_group __rcu *sched_cache_grp; int preferred_llc; /* 1: task was enqueued to its preferred LLC, 0 otherwise */ int pref_llc_queued; @@ -2417,10 +2418,23 @@ struct sched_cache_group { struct rcu_head rcu; } ____cacheline_aligned_in_smp; +struct sched_cache_group *sched_cache_group_get(struct sched_cache_group *grp); +struct sched_cache_group *task_cache_group_get(struct task_struct *p); + +void sched_cache_fork(struct task_struct *p); +void sched_cache_fork_cleanup(struct task_struct *p); +void sched_cache_exec_mmap(struct task_struct *p, struct mm_struct *mm); +void sched_cache_exit_mm(struct task_struct *p); + #else struct sched_cache_group { }; +static inline void sched_cache_fork(struct task_struct *p) { } +static inline void sched_cache_fork_cleanup(struct task_struct *p) { } +static inline void sched_cache_exec_mmap(struct task_struct *p, struct mm_struct *mm) { } +static inline void sched_cache_exit_mm(struct task_struct *p) { } + #endif #ifndef MODULE diff --git a/kernel/exit.c b/kernel/exit.c index 024350e9b48c..282328d2b4cf 100644 --- a/kernel/exit.c +++ b/kernel/exit.c @@ -551,37 +551,6 @@ void mm_update_next_owner(struct mm_struct *mm) } #endif /* CONFIG_MEMCG */ -#if defined(CONFIG_SCHED_CACHE) && defined(CONFIG_NUMA_BALANCING) -/* - * Subtract the memory footprint of the current task from - * mm. - */ -static void exit_mm_sched_cache(struct mm_struct *mm) -{ - struct sched_cache_group *grp; - unsigned long fp, sub; - - if (!current->total_numa_faults) - return; - /* - * No lock protection due to performance considerations. - * Make sure the group footprint does not become - * negative. - */ - grp = READ_ONCE(mm->sched_cache_grp); - if (!grp) - return; - - fp = READ_ONCE(grp->footprint); - sub = min(fp, current->total_numa_faults); - WRITE_ONCE(grp->footprint, fp - sub); -} -#else -static inline void exit_mm_sched_cache(struct mm_struct *mm) -{ -} -#endif /* CONFIG_SCHED_CACHE CONFIG_NUMA_BALANCING */ - /* * Turn us into a lazy TLB process if we * aren't already.. @@ -594,7 +563,7 @@ static void exit_mm(void) if (!mm) return; - exit_mm_sched_cache(mm); + sched_cache_exit_mm(current); mmap_read_lock(mm); mmgrab_lazy_tlb(mm); diff --git a/kernel/fork.c b/kernel/fork.c index 5ef413368912..10f2d05d816a 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -1599,6 +1599,7 @@ static int copy_mm(u64 clone_flags, struct task_struct *tsk) tsk->mm = mm; tsk->active_mm = mm; + sched_cache_fork(tsk); return 0; } @@ -2602,6 +2603,7 @@ __latent_entropy struct task_struct *copy_process( bad_fork_cleanup_namespaces: exit_nsproxy_namespaces(p); bad_fork_cleanup_mm: + sched_cache_fork_cleanup(p); if (p->mm) { mm_clear_owner(p->mm, p); mmput(p->mm); diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index f0a9586332fd..974a7dfe3215 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -1478,7 +1478,7 @@ static inline int get_sched_cache_scale(int mul) return (1 + (tol - 1) * mul); } -static bool exceed_llc_capacity(struct mm_struct *mm, int cpu) +static bool exceed_llc_capacity(struct sched_cache_group *grp, int cpu) { #ifdef CONFIG_NUMA_BALANCING unsigned long llc, footprint; @@ -1492,11 +1492,6 @@ static bool exceed_llc_capacity(struct mm_struct *mm, int cpu) return true; if (static_branch_likely(&sched_numa_balancing)) { - struct sched_cache_group *grp = READ_ONCE(mm->sched_cache_grp); - - if (!grp) - return true; - /* * TBD: RDT exclusive LLC ways reserved should be * excluded. @@ -1531,10 +1526,9 @@ static bool exceed_llc_capacity(struct mm_struct *mm, int cpu) return false; } -static bool invalid_llc_nr(struct mm_struct *mm, struct task_struct *p, +static bool invalid_llc_nr(struct sched_cache_group *grp, struct task_struct *p, int cpu) { - struct sched_cache_group *grp; int scale; if (get_nr_threads(p) <= 1) @@ -1548,10 +1542,6 @@ static bool invalid_llc_nr(struct mm_struct *mm, struct task_struct *p, if (scale == INT_MAX) return false; - grp = READ_ONCE(mm->sched_cache_grp); - if (!grp) - return true; - return !fits_capacity((READ_ONCE(grp->nr_running_avg) * cpu_smt_num_threads), (scale * per_cpu(sd_llc_size, cpu))); } @@ -1723,6 +1713,96 @@ static void sched_cache_group_put(struct sched_cache_group *grp) call_rcu(&grp->rcu, sched_cache_group_free_rcu); } +DEFINE_FREE(sched_cache_group_put, struct sched_cache_group *, + sched_cache_group_put(_T)); + +#define rcu_deref_sched_cache_grp(tsk) \ + rcu_dereference_check((tsk)->sched_cache_grp, (tsk) == current) + +static struct sched_cache_group *sched_cache_replace_grp(struct task_struct *p, + struct sched_cache_group *new) +{ + struct sched_cache_group *old; + + old = rcu_deref_sched_cache_grp(p); + rcu_assign_pointer(p->sched_cache_grp, new); + + return old; +} + +struct sched_cache_group *sched_cache_group_get(struct sched_cache_group *grp) +{ + /* + * refcount_inc_not_zero() is the acquire primitive for lockless + * (RCU) lookups; plain refcount_inc() would scribble the count if + * it already reached zero. Return NULL in that case. + */ + if (grp && !refcount_inc_not_zero(&grp->refcnt)) + grp = NULL; + + return grp; +} + +struct sched_cache_group *task_cache_group_get(struct task_struct *p) +{ + guard(rcu)(); + return sched_cache_group_get(rcu_dereference(p->sched_cache_grp)); +} + +void sched_cache_fork(struct task_struct *p) +{ + /* + * The child takes its own reference on the mm's cache group, separate + * from the reference held by the mm. @p is not yet visible to readers, + * so a plain initializing store is enough. + */ + RCU_INIT_POINTER(p->sched_cache_grp, + sched_cache_group_get(p->mm->sched_cache_grp)); +} + +void sched_cache_fork_cleanup(struct task_struct *p) +{ + /* + * A fork that fails after sched_cache_fork() never reaches exit_mm(), + * so drop the reference here. @p never became visible, so there are no + * concurrent readers and the reference we hold keeps the group alive. + */ + sched_cache_group_put(rcu_access_pointer(p->sched_cache_grp)); + RCU_INIT_POINTER(p->sched_cache_grp, NULL); +} + +void sched_cache_exec_mmap(struct task_struct *p, struct mm_struct *mm) +{ + struct sched_cache_group *old; + + /* + * Acquire the new reference before publishing the pointer, then drop + * the old one. @p is current and the only writer of its own pointer. + */ + old = sched_cache_replace_grp(p, sched_cache_group_get(mm->sched_cache_grp)); + sched_cache_group_put(old); +} + +void sched_cache_exit_mm(struct task_struct *p) +{ + struct sched_cache_group *grp = sched_cache_replace_grp(p, NULL); + +#ifdef CONFIG_NUMA_BALANCING + /* + * Subtract this task's footprint from the group before dropping the + * reference, so the group footprint converges as its threads exit. + * Unlocked for performance; clamp to avoid underflow. + */ + if (grp && p->total_numa_faults) { + unsigned long fp = READ_ONCE(grp->footprint); + unsigned long sub = min(fp, p->total_numa_faults); + + WRITE_ONCE(grp->footprint, fp - sub); + } +#endif + sched_cache_group_put(grp); +} + void mm_destroy_sched(struct mm_struct *mm) { sched_cache_group_put(mm->sched_cache_grp); @@ -1777,15 +1857,10 @@ static unsigned long fraction_mm_sched(struct rq *rq, return div64_u64(NICE_0_LOAD * pcpu_sched->runtime, rq->cpu_runtime + 1); } -static int get_pref_llc(struct task_struct *p, struct mm_struct *mm) +static int get_pref_llc(struct task_struct *p, struct sched_cache_group *grp) { int mm_sched_llc = -1, mm_sched_cpu; - struct sched_cache_group *grp; - if (!mm) - return -1; - - grp = READ_ONCE(mm->sched_cache_grp); if (!grp) return -1; @@ -1819,9 +1894,8 @@ static unsigned int task_running_on_cpu(int cpu, struct task_struct *p); static inline void account_mm_sched(struct rq *rq, struct task_struct *p, s64 delta_exec) { + struct sched_cache_group *grp = rcu_dereference_all(p->sched_cache_grp); struct sched_cache_time *pcpu_sched; - struct sched_cache_group *grp; - struct mm_struct *mm = p->mm; int mm_sched_llc = -1; unsigned long epoch; @@ -1832,12 +1906,8 @@ void account_mm_sched(struct rq *rq, struct task_struct *p, s64 delta_exec) return; /* * init_task, kthreads and user thread created - * by user_mode_thread() don't have mm. + * by user_mode_thread() don't have a cache group. */ - if (!mm) - return; - - grp = READ_ONCE(mm->sched_cache_grp); if (!grp || !grp->pcpu_sched) return; @@ -1855,13 +1925,13 @@ void account_mm_sched(struct rq *rq, struct task_struct *p, s64 delta_exec) * its preferred state. */ if ((long)(epoch - READ_ONCE(grp->epoch)) > llc_epoch_affinity_timeout || - invalid_llc_nr(mm, p, cpu_of(rq)) || - exceed_llc_capacity(mm, cpu_of(rq))) { + invalid_llc_nr(grp, p, cpu_of(rq)) || + exceed_llc_capacity(grp, cpu_of(rq))) { if (READ_ONCE(grp->cpu) != -1) WRITE_ONCE(grp->cpu, -1); } - mm_sched_llc = get_pref_llc(p, mm); + mm_sched_llc = get_pref_llc(p, grp); /* task not on rq accounted later in account_entity_enqueue() */ if (task_running_on_cpu(rq->cpu, p) && @@ -1874,19 +1944,15 @@ void account_mm_sched(struct rq *rq, struct task_struct *p, s64 delta_exec) static void task_tick_cache(struct rq *rq, struct task_struct *p) { + struct sched_cache_group *grp = rcu_dereference_all(p->sched_cache_grp); struct callback_head *work = &p->cache_work; - struct sched_cache_group *grp; - struct mm_struct *mm = p->mm; unsigned long epoch; if (!sched_cache_enabled()) return; - if (!mm || p->flags & PF_KTHREAD) - return; - - grp = READ_ONCE(mm->sched_cache_grp); - if (!grp || !grp->pcpu_sched) + if (!grp || p->flags & PF_KTHREAD || + !grp->pcpu_sched) return; epoch = rq->cpu_epoch; @@ -1968,14 +2034,13 @@ static inline void update_avg_scale(u64 *avg, u64 sample) static void task_cache_work(struct callback_head *work) { + struct sched_cache_group *grp __free(sched_cache_group_put) = NULL; + cpumask_var_t cpus __free(free_cpumask_var) = CPUMASK_VAR_NULL; int cpu, m_a_cpu = -1, nr_running = 0, curr_cpu; unsigned long next_scan, now = jiffies; struct task_struct *p = current, *cur; unsigned long curr_m_a_occ = 0; - struct sched_cache_group *grp; - struct mm_struct *mm = p->mm; unsigned long m_a_occ = 0; - cpumask_var_t cpus; WARN_ON_ONCE(work != &p->cache_work); @@ -1984,7 +2049,12 @@ static void task_cache_work(struct callback_head *work) if (p->flags & PF_EXITING) return; - grp = READ_ONCE(mm->sched_cache_grp); + /* + * A reference makes sure grp is not released by others. The rcu + * lock can not be held till after zalloc_cpumask_var() below, + * because the latter might sleep. + */ + grp = task_cache_group_get(p); if (!grp) return; @@ -1999,8 +2069,8 @@ static void task_cache_work(struct callback_head *work) return; curr_cpu = task_cpu(p); - if (invalid_llc_nr(mm, p, curr_cpu) || - exceed_llc_capacity(mm, curr_cpu)) { + if (invalid_llc_nr(grp, p, curr_cpu) || + exceed_llc_capacity(grp, curr_cpu)) { if (READ_ONCE(grp->cpu) != -1) WRITE_ONCE(grp->cpu, -1); @@ -2033,9 +2103,13 @@ static void task_cache_work(struct callback_head *work) m_cpu = i; } + /* + * rcu_access_pointer() is used because the + * pointer is only compared, never dereferenced. + */ cur = rcu_dereference_all(cpu_rq(i)->curr); if (cur && !(cur->flags & (PF_EXITING | PF_KTHREAD)) && - cur->mm == mm) + rcu_access_pointer(cur->sched_cache_grp) == grp) nr_running++; } @@ -2081,7 +2155,6 @@ static void task_cache_work(struct callback_head *work) } update_avg_scale(&grp->nr_running_avg, nr_running); - free_cpumask_var(cpus); } void init_sched_mm(struct task_struct *p) @@ -2090,6 +2163,13 @@ void init_sched_mm(struct task_struct *p) init_task_work(work, task_cache_work); work->next = work; + /* + * dup_task_struct() copies the parent's task_struct, including its + * sched_cache_grp, for which the child holds no reference. Clear it + * here - before copy_mm() runs - so the child never carries a + * borrowed pointer that the fork error path would put. + */ + RCU_INIT_POINTER(p->sched_cache_grp, NULL); /* * Reset new task's preference to avoid * polluting account_llc_enqueue(). @@ -3890,10 +3970,9 @@ static void task_numa_placement(struct task_struct *p) * heuristic and occasional lost updates are tolerable. * * If a task exits, its corresponding footprint must - * be subtracted from the mm->sched_cache_grp->footprint, - * otherwise the mm->sched_cache_grp->footprint will not - * converge: the exiting thread's footprint remains - * unchanged/undecayed in mm->sched_cache_grp->footprint. + * be subtracted from p->sched_cache_grp->footprint, + * otherwise the footprint will not converge: the + * exiting thread's footprint remains unchanged/undecayed. * See exit_mm(). * * Lost updates and unsynchronized subtraction @@ -3901,12 +3980,14 @@ static void task_numa_placement(struct task_struct *p) * go negative. Clamp to zero to prevent the * unsigned footprint from wrapping. */ - grp = READ_ONCE(p->mm->sched_cache_grp); - if (!grp) - continue; + scoped_guard(rcu) { + grp = rcu_dereference(p->sched_cache_grp); - new_fp = (long)READ_ONCE(grp->footprint) + diff; - WRITE_ONCE(grp->footprint, max(new_fp, 0L)); + if (grp) { + new_fp = (long)READ_ONCE(grp->footprint) + diff; + WRITE_ONCE(grp->footprint, max(new_fp, 0L)); + } + } #endif } @@ -10855,7 +10936,6 @@ static enum llc_mig can_migrate_llc_task(struct lb_env *env, struct task_struct *p) { struct sched_cache_group *grp; - struct mm_struct *mm; bool to_pref; int cpu, src_cpu, dst_cpu; @@ -10864,11 +10944,7 @@ static enum llc_mig can_migrate_llc_task(struct lb_env *env, src_cpu = env->src_cpu; dst_cpu = env->dst_cpu; - mm = p->mm; - if (!mm) - return mig_unrestricted; - - grp = READ_ONCE(mm->sched_cache_grp); + grp = rcu_dereference_all(p->sched_cache_grp); if (!grp) return mig_unrestricted; @@ -10877,8 +10953,8 @@ static enum llc_mig can_migrate_llc_task(struct lb_env *env, return mig_unrestricted; /* skip cache aware load balance for too many threads */ - if (invalid_llc_nr(mm, p, dst_cpu) || - exceed_llc_capacity(mm, dst_cpu)) { + if (invalid_llc_nr(grp, p, dst_cpu) || + exceed_llc_capacity(grp, dst_cpu)) { if (READ_ONCE(grp->cpu) != -1) WRITE_ONCE(grp->cpu, -1); return mig_unrestricted; From 65efcccddc83d6a19e8a2e2a6117811e39e589bf Mon Sep 17 00:00:00 2001 From: Chen Yu Date: Mon, 21 Sep 2026 17:37:26 -0700 Subject: [PATCH 181/365] sched/cache: Skip kernel threads for cache aware scheduling to rubustify the code Kernel thread should not be covered by cache aware scheduling as it borrows the statistics from the user space thread. Filter the kernel thread in account_mm_sched(). In theory a kernel thread does not have any valid cache group, so !grp should gate the kernel thread. Add the PF_KTHREAD check explicitly here for safety reasons, to guard against future modifications and to pair with task_tick_cache(). Fixes: df0d98475954 ("sched/cache: Introduce infrastructure for cache-aware load balancing") Signed-off-by: Chen Yu Signed-off-by: Tim Chen Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Cc: # 7.2.x Link: https://patch.msgid.link/058f0c6ea7b991c177a17de347fa3157f25489a7.1790035273.git.tim.c.chen@linux.intel.com --- kernel/sched/fair.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index 974a7dfe3215..57360f5cdde4 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -1907,8 +1907,14 @@ void account_mm_sched(struct rq *rq, struct task_struct *p, s64 delta_exec) /* * init_task, kthreads and user thread created * by user_mode_thread() don't have a cache group. + * In theory a kernel thread does not have any valid + * cache group, because sched_cache_fork() is not + * invoked for a kernel thread - !grp should gate the + * kernel thread. Use the PF_KTHREAD check explicitly + * here for safety reasons, to guard against future + * modifications and to pair with task_tick_cache(). */ - if (!grp || !grp->pcpu_sched) + if (p->flags & PF_KTHREAD || !grp || !grp->pcpu_sched) return; pcpu_sched = per_cpu_ptr(grp->pcpu_sched, cpu_of(rq)); From 3cb0243767fd033bdce95f4f1b5882172a2f8119 Mon Sep 17 00:00:00 2001 From: Davi Chaves Azevedo Date: Mon, 21 Sep 2026 17:37:27 -0700 Subject: [PATCH 182/365] sched/cache: Refresh LLC capacity across CPU hotplug, to fix capacity underestimation bug The scheduler scales LLC capacity by the fraction of cache-sharing CPUs covered by a domain: llc_bytes = cache_size * span_weight / shared_weight During CPU teardown, sched_cpu_deactivate() rebuilds scheduler domains before cacheinfo_cpu_pre_down() removes the CPU from shared_cpu_map. The new domains therefore use the old sharing weight. The later call to sched_update_llc_bytes() looks up the departing CPU's sd_llc, which has already been detached, and returns without correcting the surviving CPUs. On a Ryzen 5 7535U with twelve logical CPUs sharing a 16 MiB LLC, offlining one SMT sibling left the remaining CPUs with: llc_bytes = floor(16777216 * 11 / 12) = 15379114 bytes The correct capacity is still 16777216 bytes. On systems with active cache-aware scheduling, an underestimated capacity can cause exceed_llc_capacity() to reject aggregation for a process whose footprint would fit. Unchanged cpuset partitions sharing the physical cache can also retain stale capacity when a CPU comes online in another partition. Pass the cache-sharing mask already retained by cacheinfo to the scheduler update. Refresh every surviving CPU using its own LLC domain so that each partition receives the correct share. This also preserves the correction needed as cache-sharing maps grow during boot. Keep the existing CPU-hotplug and scheduler-domain synchronization. The update remains on the hotplug path; no steady-state scheduling operation or persistent allocation is added. Fixes: 7030513a0877 ("sched/cache: Calculate the LLC size and store it in sched_domain") Signed-off-by: Davi Chaves Azevedo Signed-off-by: Tim Chen Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Reviewed-by: Chen Yu Reviewed-by: Tim Chen Reviewed-by: K Prateek Nayak Tested-by: Chen Yu Tested-by: K Prateek Nayak Cc: # v7.2.x Link: https://patch.msgid.link/6751d93e15889e624796c74db0bfe66603d60b1b.1790035273.git.tim.c.chen@linux.intel.com --- drivers/base/cacheinfo.c | 11 ++++++----- include/linux/sched/topology.h | 4 ++-- kernel/sched/topology.c | 22 +++++++++++++--------- 3 files changed, 21 insertions(+), 16 deletions(-) diff --git a/drivers/base/cacheinfo.c b/drivers/base/cacheinfo.c index 9f9c72727a05..7a47a392568a 100644 --- a/drivers/base/cacheinfo.c +++ b/drivers/base/cacheinfo.c @@ -1040,9 +1040,10 @@ static int cacheinfo_cpu_online(unsigned int cpu) rc = cache_add_dev(cpu); if (rc) goto err; - if (cpu_map_shared_cache(true, cpu, &cpu_map)) + if (cpu_map_shared_cache(true, cpu, &cpu_map)) { update_per_cpu_data_slice_size(true, cpu, cpu_map); - sched_update_llc_bytes(cpu); + sched_update_llc_bytes(cpu_map); + } return 0; err: free_cache_attributes(cpu); @@ -1059,10 +1060,10 @@ static int cacheinfo_cpu_pre_down(unsigned int cpu) cpu_cache_sysfs_exit(cpu); free_cache_attributes(cpu); - if (nr_shared > 1) + if (nr_shared > 1) { update_per_cpu_data_slice_size(false, cpu, cpu_map); - - sched_update_llc_bytes(cpu); + sched_update_llc_bytes(cpu_map); + } return 0; } diff --git a/include/linux/sched/topology.h b/include/linux/sched/topology.h index b5d9d7c2b8ad..f96812d71c51 100644 --- a/include/linux/sched/topology.h +++ b/include/linux/sched/topology.h @@ -281,9 +281,9 @@ static inline int task_node(const struct task_struct *p) } #ifdef CONFIG_SCHED_CACHE -extern void sched_update_llc_bytes(unsigned int cpu); +extern void sched_update_llc_bytes(const struct cpumask *cpus); #else -static inline void sched_update_llc_bytes(unsigned int cpu) { } +static inline void sched_update_llc_bytes(const struct cpumask *cpus) { } #endif #endif /* _LINUX_SCHED_TOPOLOGY_H */ diff --git a/kernel/sched/topology.c b/kernel/sched/topology.c index 0248227d983a..3dab0253976f 100644 --- a/kernel/sched/topology.c +++ b/kernel/sched/topology.c @@ -985,8 +985,8 @@ void sched_cache_active_set(void) } /* - * Update the bottom sched_domain's llc_bytes for @cpu and all its - * LLC siblings. Called from cacheinfo_cpu_online() or + * Update the bottom sched_domain's llc_bytes for @cpus sharing a physical + * LLC. Called from cacheinfo_cpu_online() or * cacheinfo_cpu_pre_down() with cpu hotplug lock held. * * Note: get_effective_llc_bytes() returns 0 on PowerPC. @@ -996,17 +996,13 @@ void sched_cache_active_set(void) * and does not populates the per-CPU struct cpu_cacheinfo array * that get_cpu_cacheinfo_llc() reads. */ -void sched_update_llc_bytes(unsigned int cpu) +void sched_update_llc_bytes(const struct cpumask *cpus) { struct sched_domain *sd, *sdp; unsigned int i; sched_domains_mutex_lock(); - sdp = rcu_dereference_sched_domain(per_cpu(sd_llc, cpu)); - if (!sdp) - goto unlock; - /* * ci->shared_cpu_map is built incrementally as CPUs come * online, so the first CPU in an LLC initially sees @@ -1014,14 +1010,22 @@ void sched_update_llc_bytes(unsigned int cpu) * get_effective_llc_bytes(). Re-evaluating every LLC * sibling on each online event corrects this once the full * shared_cpu_map is known. + * + * The departing CPU's domains have already been detached when + * cacheinfo removes it. Use the surviving cache siblings instead. + * They may belong to different cpuset partitions, so use each CPU's + * own LLC domain to scale its share of the physical cache. */ - for_each_cpu(i, sched_domain_span(sdp)) { + for_each_cpu(i, cpus) { + sdp = rcu_dereference_sched_domain(per_cpu(sd_llc, i)); + if (!sdp) + continue; + sd = rcu_dereference_sched_domain(cpu_rq(i)->sd); if (sd) sd->llc_bytes = get_effective_llc_bytes(i, sdp); } -unlock: sched_domains_mutex_unlock(); } From cec38d5c098a350dcf084d345025136ade7e6d1e Mon Sep 17 00:00:00 2001 From: Sean Christopherson Date: Mon, 21 Sep 2026 12:14:09 -0700 Subject: [PATCH 183/365] perf/x86/intel: Ensure KVM guest PEBS path doesn't set unwanted PERF_GLOBAL_CTRL bits When reinstating PEBS counters into PERF_GLOBAL_CTRL for a KVM guest, mask the value with perf's desired/original PERF_GLOBAL_CTRL value to ensure KVM doesn't unintentionally set reserved bits in PERF_GLOBAL_CTRL. E.g. if the guest's PEBS_ENABLE value had bit 63, "Enable Precise Store", set, then using the raw guest PEBS value would propagate bit 63 to the guest's PERF_GLOBAL_CTRL value (which thankfully would be a failed VM-Entry, not a VMX Abort). The only reason this bug isn't reachable is because KVM doesn't support "Enable Precise Store" (which is probably a KVM bug?), i.e. bit 63 can't be set in kvm_pmu->pebs_enable and thus not in arr[pebs_enable].guest. In other words, this _should_ be a glorified NOP in the current code base. Fixes: c59a1f106f5c ("KVM: x86/pmu: Add IA32_PEBS_ENABLE MSR emulation for extended PEBS") Signed-off-by: Sean Christopherson Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Reviewed-by: Dapeng Mi Link: https://patch.msgid.link/20260921191418.950933-2-seanjc@google.com --- arch/x86/events/intel/core.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/x86/events/intel/core.c b/arch/x86/events/intel/core.c index 1ac2ca35db53..0400b5111db9 100644 --- a/arch/x86/events/intel/core.c +++ b/arch/x86/events/intel/core.c @@ -5359,7 +5359,7 @@ static struct perf_guest_switch_msr *intel_guest_get_msrs(int *nr, void *data) arr[pebs_enable].guest &= ~kvm_pmu->host_cross_mapped_mask; arr[global_ctrl].guest &= ~kvm_pmu->host_cross_mapped_mask; /* Set hw GLOBAL_CTRL bits for PEBS counter when it runs for guest */ - arr[global_ctrl].guest |= arr[pebs_enable].guest; + arr[global_ctrl].guest |= intel_ctrl & arr[pebs_enable].guest; } return arr; From 4b64dbdc5861477f148e13d1ed127e7fe7182e4f Mon Sep 17 00:00:00 2001 From: Sean Christopherson Date: Mon, 21 Sep 2026 12:14:10 -0700 Subject: [PATCH 184/365] perf/x86/intel: Don't write PEBS_ENABLED on host<=>guest xfers if CPU has PEBS isolation, to fix stuck PEBS_ENABLED When filling the list of MSRs to be loaded by KVM on VM-Enter and VM-Exit, *never* insert an entry for PEBS_ENABLED if the CPU properly isolates PEBS events, in which case disabling counters via PERF_GLOBAL_CTRL is sufficient to prevent unwanted PEBS events in the guest (or host). Because perf loads PEBS_ENABLE with the unfiltered cpu_hw_events.pebs_enabled, i.e. with both host and guest masks, there is no need to load different values for the guest versus host, perf+KVM can and should simply control which counters are enabled/disabled via PERF_GLOBAL_CTRL. Avoiding touching PEBS_ENABLED "fixes" a bug where PEBS_ENABLED can end up with "stuck" bits if a PEBS event is throttled between generating the list and actually entering the guest (Intel CPUs can't arbtitrarily block NMIs). Fixes in quotes because leaving PEBS_ENABLED as-is doesn't fix the underlying problem of perf (via PMIs) being able to modify state after the perf<=>KVM handoff. But not writing PEBS_ENABLED is desirable no matter what, as stating the obvious, leaving PEBS_ENABLED as-is avoids three MSR writes on every VMX transition: one each on entry/exit, and one more explicit WRMSR to zero PEBS_ENABLED before VM-Entry (KVM assumes the only reason PEBS_ENABLED is in the load list is if the CPU lacks PEBS isolation and thus needs a quiescent period). Opportunistically add comments to (better) explain the rules for generating the set of PEBS counters that will be active while the guest is running, along with a FIXME for the suspected hack-a-fix where perf disables guest PEBS if _any_ PEBS event is configured to count in the host (commit 854250329c02 ("KVM: x86/pmu: Disable guest PEBS temporarily in two rare situations") doesn't explain the motivation, at all). Fixes: c59a1f106f5c ("KVM: x86/pmu: Add IA32_PEBS_ENABLE MSR emulation for extended PEBS") Signed-off-by: Sean Christopherson Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Reviewed-by: Dapeng Mi Link: https://patch.msgid.link/20260921191418.950933-3-seanjc@google.com --- arch/x86/events/intel/core.c | 55 ++++++++++++++++++++++++------------ 1 file changed, 37 insertions(+), 18 deletions(-) diff --git a/arch/x86/events/intel/core.c b/arch/x86/events/intel/core.c index 0400b5111db9..5ac8fe31871b 100644 --- a/arch/x86/events/intel/core.c +++ b/arch/x86/events/intel/core.c @@ -5292,12 +5292,15 @@ static struct perf_guest_switch_msr *intel_guest_get_msrs(int *nr, void *data) struct kvm_pmu *kvm_pmu = (struct kvm_pmu *)data; u64 intel_ctrl = hybrid(cpuc->pmu, intel_ctrl); u64 pebs_mask = cpuc->pebs_enabled & x86_pmu.pebs_capable; - int global_ctrl, pebs_enable; + u64 guest_pebs_mask; + int global_ctrl; /* * In addition to obeying exclude_guest/exclude_host, remove bits being * used for PEBS when running a guest, because PEBS writes to virtual - * addresses (not physical addresses). + * addresses (not physical addresses). If the guest wants to utilize + * PEBS, and PEBS can be safely enabled in the guest, bits for the guest's + * PEBS-enabled counters will be OR'd back in as appropriate. */ *nr = 0; global_ctrl = (*nr)++; @@ -5344,24 +5347,40 @@ static struct perf_guest_switch_msr *intel_guest_get_msrs(int *nr, void *data) }; } - pebs_enable = (*nr)++; - arr[pebs_enable] = (struct perf_guest_switch_msr){ - .msr = MSR_IA32_PEBS_ENABLE, - .host = cpuc->pebs_enabled & ~cpuc->intel_ctrl_guest_mask, - .guest = pebs_mask & ~cpuc->intel_ctrl_host_mask & kvm_pmu->pebs_enable, - }; + /* + * Restrict guest PEBS events to counters that (a) perf supports, (b) + * the guest wants to use for PEBS, (c) are not excluded from counting + * in the guest, and (d) _are_ excluded from counting in the host. + */ + guest_pebs_mask = pebs_mask & intel_ctrl & kvm_pmu->pebs_enable & + ~cpuc->intel_ctrl_host_mask & + cpuc->intel_ctrl_guest_mask; - if (arr[pebs_enable].host) { - /* Disable guest PEBS if host PEBS is enabled. */ - arr[pebs_enable].guest = 0; - } else { - /* Disable guest PEBS thoroughly for cross-mapped PEBS counters. */ - arr[pebs_enable].guest &= ~kvm_pmu->host_cross_mapped_mask; - arr[global_ctrl].guest &= ~kvm_pmu->host_cross_mapped_mask; - /* Set hw GLOBAL_CTRL bits for PEBS counter when it runs for guest */ - arr[global_ctrl].guest |= intel_ctrl & arr[pebs_enable].guest; - } + /* + * Disable counters where the guest PMC is different than the host PMC + * being used on behalf of the guest, as the PEBS record includes + * PERF_GLOBAL_STATUS, i.e. the guest will see overflow status for the + * wrong counter(s). + */ + guest_pebs_mask &= ~kvm_pmu->host_cross_mapped_mask; + /* + * FIXME: Allow guest and host usage of PEBS events to co-exist instead + * of disabling guest PEBS entirely if the host is using PEBS. + * What exactly goes wrong if guest and host are using PEBS is + * unknown. + */ + if (pebs_mask & ~cpuc->intel_ctrl_guest_mask) + guest_pebs_mask = 0; + + /* + * Do NOT mess with PEBS_ENABLED. As above, disabling counters via + * PERF_GLOBAL_CTRL is sufficient, and loading a stale PEBS_ENABLED, + * e.g. on VM-Exit, can put the system in a bad state. Simply enable + * counters in PERF_GLOBAL_CTRL, as perf load PEBS_ENABLED with the + * full value, i.e. perf *also* relies on PERF_GLOBAL_CTRL. + */ + arr[global_ctrl].guest |= guest_pebs_mask; return arr; } From d06260e99eb93d2942b7af4ccd789eb8a6c829d3 Mon Sep 17 00:00:00 2001 From: Sean Christopherson Date: Mon, 21 Sep 2026 12:14:11 -0700 Subject: [PATCH 185/365] perf/x86/intel: Don't pointlessly context switch DS_AREA (and PEBS config) if PEBS is unused When filling the list of MSRs to be loaded by KVM on VM-Enter and VM-Exit, load the guest values for DS_AREA and (conditionally) MSR_PEBS_DATA_CFG if and only if PEBS will be active in the guest, i.e. only if a PEBS record may be generated while running the guest. As shown by the !pebs_ept path, it's perfectly safe to run with the host's DS_AREA, so long as PEBS-enabled counters are disabled via PERF_GLOBAL_CTRL. Omitting DS_AREA and MSR_PEBS_DATA_CFG when PEBS is unused saves two MSR writes per MSR on each VMX transition, i.e. eliminates two/four pointless MSR writes on each VMX roundtrip when PEBS isn't being used by the guest. Fixes: c59a1f106f5c ("KVM: x86/pmu: Add IA32_PEBS_ENABLE MSR emulation for extended PEBS") Signed-off-by: Sean Christopherson Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Reviewed-by: Jim Mattson Reviewed-by: Dapeng Mi Link: https://patch.msgid.link/20260921191418.950933-4-seanjc@google.com --- arch/x86/events/intel/core.c | 39 +++++++++++++++++++++++------------- 1 file changed, 25 insertions(+), 14 deletions(-) diff --git a/arch/x86/events/intel/core.c b/arch/x86/events/intel/core.c index 5ac8fe31871b..75c74e8fe2f4 100644 --- a/arch/x86/events/intel/core.c +++ b/arch/x86/events/intel/core.c @@ -5330,23 +5330,14 @@ static struct perf_guest_switch_msr *intel_guest_get_msrs(int *nr, void *data) return arr; } + /* + * If the guest won't use PEBS or the CPU doesn't support PEBS in the + * guest, then there's nothing more to do as disabling PMCs via + * PERF_GLOBAL_CTRL is sufficient on CPUs with guest/host isolation. + */ if (!kvm_pmu || !x86_pmu.pebs_ept) return arr; - arr[(*nr)++] = (struct perf_guest_switch_msr){ - .msr = MSR_IA32_DS_AREA, - .host = (unsigned long)cpuc->ds, - .guest = kvm_pmu->ds_area, - }; - - if (x86_pmu.intel_cap.pebs_baseline) { - arr[(*nr)++] = (struct perf_guest_switch_msr){ - .msr = MSR_PEBS_DATA_CFG, - .host = cpuc->active_pebs_data_cfg, - .guest = kvm_pmu->pebs_data_cfg, - }; - } - /* * Restrict guest PEBS events to counters that (a) perf supports, (b) * the guest wants to use for PEBS, (c) are not excluded from counting @@ -5373,6 +5364,26 @@ static struct perf_guest_switch_msr *intel_guest_get_msrs(int *nr, void *data) if (pebs_mask & ~cpuc->intel_ctrl_guest_mask) guest_pebs_mask = 0; + /* + * Context switch DS_AREA and PEBS_DATA_CFG if and only if PEBS will be + * active in the guest; if no records will be generated while the guest + * is running, then simply keep the host values resident in hardware. + */ + arr[(*nr)++] = (struct perf_guest_switch_msr){ + .msr = MSR_IA32_DS_AREA, + .host = (unsigned long)cpuc->ds, + .guest = guest_pebs_mask ? kvm_pmu->ds_area : (unsigned long)cpuc->ds, + }; + + if (x86_pmu.intel_cap.pebs_baseline) { + arr[(*nr)++] = (struct perf_guest_switch_msr){ + .msr = MSR_PEBS_DATA_CFG, + .host = cpuc->active_pebs_data_cfg, + .guest = guest_pebs_mask ? kvm_pmu->pebs_data_cfg : + cpuc->active_pebs_data_cfg, + }; + } + /* * Do NOT mess with PEBS_ENABLED. As above, disabling counters via * PERF_GLOBAL_CTRL is sufficient, and loading a stale PEBS_ENABLED, From a391618e1d563f099e4c2a704f45d08329ccdf7c Mon Sep 17 00:00:00 2001 From: Sean Christopherson Date: Mon, 21 Sep 2026 12:14:12 -0700 Subject: [PATCH 186/365] perf/x86/intel: Make @data a mandatory param for intel_guest_get_msrs() Drop "support" for passing a NULL @data/@kvm_pmu param when getting guest MSRs. KVM, the only in-tree user, unconditionally passes a non-NULL pointer, and carrying code that suggests @data may be NULL is confusing, e.g. incorrectly implies that there are scenarios where KVM doesn't pass a PMU context. Fixes: 8183a538cd95 ("KVM: x86/pmu: Add IA32_DS_AREA MSR emulation to support guest DS") Signed-off-by: Sean Christopherson Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Reviewed-by: Jim Mattson Reviewed-by: Dapeng Mi Link: https://patch.msgid.link/20260921191418.950933-5-seanjc@google.com --- arch/x86/events/intel/core.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/arch/x86/events/intel/core.c b/arch/x86/events/intel/core.c index 75c74e8fe2f4..757d4082de60 100644 --- a/arch/x86/events/intel/core.c +++ b/arch/x86/events/intel/core.c @@ -5331,11 +5331,11 @@ static struct perf_guest_switch_msr *intel_guest_get_msrs(int *nr, void *data) } /* - * If the guest won't use PEBS or the CPU doesn't support PEBS in the - * guest, then there's nothing more to do as disabling PMCs via - * PERF_GLOBAL_CTRL is sufficient on CPUs with guest/host isolation. + * If the CPU doesn't support PEBS in the guest, then there's nothing + * more to do as disabling PMCs via PERF_GLOBAL_CTRL is sufficient on + * CPUs with guest/host isolation. */ - if (!kvm_pmu || !x86_pmu.pebs_ept) + if (!x86_pmu.pebs_ept) return arr; /* From 89dc568e8c0be60e05e5fcd0b528c79077d7b84b Mon Sep 17 00:00:00 2001 From: Dapeng Mi Date: Thu, 17 Sep 2026 09:52:23 +0800 Subject: [PATCH 187/365] perf/x86/intel: Fix GRT PEBS load/store direction for latency events, to fix sample classification On Gracemont, intel_grt_pebs_event_constraints[] applies LAT_CONSTRAINT constraints to MEM_UOPS_RETIRED.{LOAD,STORE}_LATENCY, but does not set explicit LOAD/STORE flags for those events. The PEBS latency path (pebs_latency_data(), via __grt_latency_data()) uses the event flags to determine memory operation direction. Without an explicit STORE flag, samples from MEM_UOPS_RETIRED.STORE_LATENCY can be misclassified as LOADs. Set explicit LOAD/STORE flags in intel_grt_pebs_event_constraints[] for: - MEM_UOPS_RETIRED.LOAD_LATENCY - MEM_UOPS_RETIRED.STORE_LATENCY Also update __grt_latency_data() to explicitly interpret these flags when assigning the sampled memory operation direction. This fixes incorrect STORE sample classification. Fixes: 39a41278f041 ("perf/x86/intel: Fix PEBS memory access info encoding for ADL") Signed-off-by: Dapeng Mi Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Cc: # v7.2+ Link: https://patch.msgid.link/20260917015234.981153-2-dapeng1.mi@linux.intel.com --- arch/x86/events/intel/ds.c | 16 +++++++++++++--- 1 file changed, 13 insertions(+), 3 deletions(-) diff --git a/arch/x86/events/intel/ds.c b/arch/x86/events/intel/ds.c index b98029b44052..fff98f457576 100644 --- a/arch/x86/events/intel/ds.c +++ b/arch/x86/events/intel/ds.c @@ -455,6 +455,7 @@ static inline void pebs_set_tlb_lock(u64 *val, bool tlb, bool lock) static u64 __grt_latency_data(struct perf_event *event, u64 status, u8 dse, bool tlb, bool lock, bool blk) { + union perf_mem_data_src src; u64 val; WARN_ON_ONCE(is_hybrid() && @@ -470,7 +471,16 @@ static u64 __grt_latency_data(struct perf_event *event, u64 status, else val |= P(BLK, NA); - return val; + src.val = val; + + if (event->hw.flags & + (PERF_X86_EVENT_PEBS_LDLAT | PERF_X86_EVENT_PEBS_LD_HSW)) + src.mem_op = P(OP, LOAD); + if (event->hw.flags & + (PERF_X86_EVENT_PEBS_STLAT | PERF_X86_EVENT_PEBS_ST_HSW)) + src.mem_op = P(OP, STORE); + + return src.val; } u64 grt_latency_data(struct perf_event *event, u64 status) @@ -1291,8 +1301,8 @@ struct event_constraint intel_glm_pebs_event_constraints[] = { struct event_constraint intel_grt_pebs_event_constraints[] = { /* Allow all events as PEBS with no flags */ - INTEL_HYBRID_LAT_CONSTRAINT(0x5d0, 0x3), - INTEL_HYBRID_LAT_CONSTRAINT(0x6d0, 0x3f), + INTEL_HYBRID_LDLAT_CONSTRAINT(0x5d0, 0x3), + INTEL_HYBRID_STLAT_CONSTRAINT(0x6d0, 0x3f), EVENT_CONSTRAINT_END }; From e961d6db42d1f1b66c81438969a648f08573bd9c Mon Sep 17 00:00:00 2001 From: Dapeng Mi Date: Thu, 17 Sep 2026 09:52:24 +0800 Subject: [PATCH 188/365] perf/x86/intel: Fix CMT PEBS load/store direction for latency events, to fix sample classification The same bug exists on Crestmont as on Gracemont: intel_cmt_pebs_event_constraints[] applies LAT_CONSTRAINT constraints to MEM_UOPS_RETIRED.{LOAD,STORE}_LATENCY, but does not set explicit LOAD/STORE flags for those events. The PEBS latency path (pebs_latency_data(), via cmt_latency_data) uses the event flags to determine memory operation direction. Without an explicit STORE flag, samples from MEM_UOPS_RETIRED.STORE_LATENCY can be misclassified as LOADs. Set explicit LOAD/STORE flags in intel_cmt_pebs_event_constraints[] for: - MEM_UOPS_RETIRED.LOAD_LATENCY - MEM_UOPS_RETIRED.STORE_LATENCY This fixes incorrect STORE sample classification. Fixes: e99fb45436ea ("perf/x86/intel: Update event constraints and cache_extra_regsfor MTL") Signed-off-by: Dapeng Mi Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Cc: # v7.2+ Link: https://patch.msgid.link/20260917015234.981153-3-dapeng1.mi@linux.intel.com --- arch/x86/events/intel/ds.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/arch/x86/events/intel/ds.c b/arch/x86/events/intel/ds.c index fff98f457576..fa7e0268b4f2 100644 --- a/arch/x86/events/intel/ds.c +++ b/arch/x86/events/intel/ds.c @@ -1308,8 +1308,8 @@ struct event_constraint intel_grt_pebs_event_constraints[] = { struct event_constraint intel_cmt_pebs_event_constraints[] = { /* Allow all events as PEBS with no flags */ - INTEL_HYBRID_LAT_CONSTRAINT(0x5d0, 0x3), - INTEL_HYBRID_LAT_CONSTRAINT(0x6d0, 0xff), + INTEL_HYBRID_LDLAT_CONSTRAINT(0x5d0, 0x3), + INTEL_HYBRID_STLAT_CONSTRAINT(0x6d0, 0xff), EVENT_CONSTRAINT_END }; From 2777ec9852277a06ae68fee0c4f1a32783e4a999 Mon Sep 17 00:00:00 2001 From: Nemesa Garg Date: Wed, 9 Sep 2026 16:33:32 +0530 Subject: [PATCH 189/365] drm/i915/psr: Clear stale sel fetch enable bits on sel fetch disable MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Selective fetch is dropped while pipe CRC is active, and the planes keep their SEL_FETCH_PLANE_CTL / SEL_FETCH_CUR_CTL enable bit set in hardware over that. A plane disabled while selective fetch is off never gets the bit cleared, as the disable path is guarded by enable_psr2_sel_fetch. Once selective fetch comes back the hardware resumes fetching for a plane that is no longer enabled and keeps its DDB range reserved. Clear the bits as selective fetch is turned off instead. Atomic check has both the old and the new crtc state, so record the transition there and let the plane and cursor arm paths write the registers to 0 for that commit. v2: Drop the old_crtc_state->hw.active check. [Jouni] Fixes: b1f5279b5981 ("drm/i915/psr: Move plane sel fetch configuration into plane source files") Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/work_items/8739 Assisted-by: Copilot:Claude-Opus-5 Signed-off-by: Nemesa Garg Reviewed-by: Jouni Högander Signed-off-by: Suraj Kandpal Link: https://patch.msgid.link/20260909110332.3528029-3-nemesa.garg@intel.com (cherry picked from commit a4c0e7f80429eda6990960971aebd4e4b9533cc6) Signed-off-by: Jani Nikula --- drivers/gpu/drm/i915/display/intel_cursor.c | 7 +++++-- .../gpu/drm/i915/display/intel_display_types.h | 2 ++ drivers/gpu/drm/i915/display/intel_psr.c | 15 +++++++++++++++ .../gpu/drm/i915/display/skl_universal_plane.c | 9 ++++----- 4 files changed, 26 insertions(+), 7 deletions(-) diff --git a/drivers/gpu/drm/i915/display/intel_cursor.c b/drivers/gpu/drm/i915/display/intel_cursor.c index 0673f16f6fd0..824ffeef0103 100644 --- a/drivers/gpu/drm/i915/display/intel_cursor.c +++ b/drivers/gpu/drm/i915/display/intel_cursor.c @@ -536,7 +536,8 @@ static void i9xx_cursor_disable_sel_fetch_arm(struct intel_dsb *dsb, struct intel_display *display = to_intel_display(plane); enum pipe pipe = plane->pipe; - if (!crtc_state->enable_psr2_sel_fetch) + if (!crtc_state->enable_psr2_sel_fetch && + !crtc_state->clear_psr2_sel_fetch) return; intel_de_write_dsb(display, dsb, SEL_FETCH_CUR_CTL(pipe), 0); @@ -569,8 +570,10 @@ static void i9xx_cursor_update_sel_fetch_arm(struct intel_dsb *dsb, struct intel_display *display = to_intel_display(plane); enum pipe pipe = plane->pipe; - if (!crtc_state->enable_psr2_sel_fetch) + if (!crtc_state->enable_psr2_sel_fetch) { + i9xx_cursor_disable_sel_fetch_arm(dsb, plane, crtc_state); return; + } if (drm_rect_height(&plane_state->psr2_sel_fetch_area) > 0) { if (crtc_state->enable_psr2_su_region_et) { diff --git a/drivers/gpu/drm/i915/display/intel_display_types.h b/drivers/gpu/drm/i915/display/intel_display_types.h index 43d53a98dae7..f6a9b0de1ade 100644 --- a/drivers/gpu/drm/i915/display/intel_display_types.h +++ b/drivers/gpu/drm/i915/display/intel_display_types.h @@ -1187,6 +1187,8 @@ struct intel_crtc_state { bool has_sel_update; bool enable_psr2_sel_fetch; bool enable_psr2_su_region_et; + /* Drop the stale selective fetch enable bits as selective fetch is turned off */ + bool clear_psr2_sel_fetch; bool req_psr2_sdp_prior_scanline; bool has_panel_replay; bool link_off_after_as_sdp_when_pr_active; diff --git a/drivers/gpu/drm/i915/display/intel_psr.c b/drivers/gpu/drm/i915/display/intel_psr.c index 40e3d7095996..c28bcb1a0184 100644 --- a/drivers/gpu/drm/i915/display/intel_psr.c +++ b/drivers/gpu/drm/i915/display/intel_psr.c @@ -2883,6 +2883,8 @@ int intel_psr2_sel_fetch_update(struct intel_atomic_state *state, struct intel_crtc *crtc) { struct intel_display *display = to_intel_display(state); + const struct intel_crtc_state *old_crtc_state = + intel_atomic_get_old_crtc_state(state, crtc); struct intel_crtc_state *crtc_state = intel_atomic_get_new_crtc_state(state, crtc); struct intel_plane_state *new_plane_state, *old_plane_state; struct intel_plane *plane; @@ -2895,6 +2897,19 @@ int intel_psr2_sel_fetch_update(struct intel_atomic_state *state, bool full_update = false, su_area_changed; int i, ret; + /* + * Selective fetch is not always usable, for instance it is dropped + * while pipe CRC is active. The planes keep their selective fetch + * enable bit set in hardware over that, and a plane disabled while + * selective fetch is off never gets the bit cleared. Once selective + * fetch comes back the hardware would resume fetching for a plane that + * is no longer enabled and keep its DDB range reserved, so have the + * plane update drop the bit for every plane of the pipe as selective + * fetch is turned off. + */ + crtc_state->clear_psr2_sel_fetch = old_crtc_state->enable_psr2_sel_fetch && + !crtc_state->enable_psr2_sel_fetch; + if (!crtc_state->enable_psr2_sel_fetch) return 0; diff --git a/drivers/gpu/drm/i915/display/skl_universal_plane.c b/drivers/gpu/drm/i915/display/skl_universal_plane.c index 07a683293352..eb5ed981b40f 100644 --- a/drivers/gpu/drm/i915/display/skl_universal_plane.c +++ b/drivers/gpu/drm/i915/display/skl_universal_plane.c @@ -885,7 +885,8 @@ static void icl_plane_disable_sel_fetch_arm(struct intel_dsb *dsb, struct intel_display *display = to_intel_display(plane); enum pipe pipe = plane->pipe; - if (!crtc_state->enable_psr2_sel_fetch) + if (!crtc_state->enable_psr2_sel_fetch && + !crtc_state->clear_psr2_sel_fetch) return; intel_de_write_dsb(display, dsb, SEL_FETCH_PLANE_CTL(pipe, plane->id), 0); @@ -1634,10 +1635,8 @@ static void icl_plane_update_sel_fetch_arm(struct intel_dsb *dsb, struct intel_display *display = to_intel_display(plane); enum pipe pipe = plane->pipe; - if (!crtc_state->enable_psr2_sel_fetch) - return; - - if (drm_rect_height(&plane_state->psr2_sel_fetch_area) > 0) + if (crtc_state->enable_psr2_sel_fetch && + drm_rect_height(&plane_state->psr2_sel_fetch_area) > 0) intel_de_write_dsb(display, dsb, SEL_FETCH_PLANE_CTL(pipe, plane->id), SEL_FETCH_PLANE_CTL_ENABLE); else From 5271d81f99dd01d983d439930eb056952485e15e Mon Sep 17 00:00:00 2001 From: Ankit Nautiyal Date: Mon, 7 Sep 2026 09:15:55 +0530 Subject: [PATCH 190/365] drm/i915/quirks: Limit eDP rate to HBR2 on HP Pavilion Plus 14-ew1 The eDP panel on the HP Pavilion Plus Laptop 14-ew1xxx advertises HBR3 while leaving the TPS4 support bit clear. The output however flickers, once link is trained with HBR3. Until commit 8c9006283e4b ("Revert "drm/i915/dp: Reject HBR3 when sink doesn't support TPS4"") such sinks were capped at HBR2 by the TPS4 check which incidentally kept this panel stable. That check was reverted because other panels legitimately need HBR3 without advertising TPS4, and the per-machine QUIRK_EDP_LIMIT_RATE_HBR2 was introduced to handle the affected machines instead. Add the machine to the list of devices that need the QUIRK_EDP_LIMIT_RATE_HBR2. Fixes: 8c9006283e4b ("Revert "drm/i915/dp: Reject HBR3 when sink doesn't support TPS4"") Reported-by: Annoy Cc Closes: https://gitlab.freedesktop.org/drm/i915/kernel/-/work_items/16743 Cc: # v6.18+ Tested-by: Annoy Cc Signed-off-by: Ankit Nautiyal Reviewed-by: Nemesa Garg Link: https://patch.msgid.link/20260907034555.2753846-1-ankit.k.nautiyal@intel.com (cherry picked from commit 550b703fdbb2a2022faa75b4b11ab135241afbd9) Signed-off-by: Jani Nikula --- drivers/gpu/drm/i915/display/intel_quirks.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/drivers/gpu/drm/i915/display/intel_quirks.c b/drivers/gpu/drm/i915/display/intel_quirks.c index 33245f44c0d5..7d7db774d8c7 100644 --- a/drivers/gpu/drm/i915/display/intel_quirks.c +++ b/drivers/gpu/drm/i915/display/intel_quirks.c @@ -257,6 +257,9 @@ static struct intel_quirk intel_quirks[] = { /* Dell XPS 13 7390 2-in-1 */ { 0x8a52, 0x1028, 0x08b0, quirk_edp_limit_rate_hbr2 }, + /* HP Pavilion Plus Laptop 14-ew1xxx */ + { 0x7d55, 0x103c, 0x8c31, quirk_edp_limit_rate_hbr2 }, + /* Xiaomi Book Pro 14 2026 */ { 0xb081, 0x1d72, 0x2424, quirk_disable_psr2 }, }; From 8302c5f475fa5a4ed36a7d32c7b965023d5d7066 Mon Sep 17 00:00:00 2001 From: Dapeng Mi Date: Thu, 17 Sep 2026 09:52:25 +0800 Subject: [PATCH 191/365] perf/x86/intel: Fix DKT PEBS load/store direction for latency events, to fix sample classification Same bug exists on Darkmont as on Gracemont: intel_dkt_pebs_event_constraints[] applies LAT_CONSTRAINT constraints to MEM_UOPS_RETIRED.{LOAD,STORE}_LATENCY, but does not set explicit LOAD/STORE flags for those events. The PEBS latency path (pebs_latency_data(), via cmt_latency_data) uses the event flags to determine memory operation direction. Without an explicit STORE flag, samples from MEM_UOPS_RETIRED.STORE_LATENCY can be misclassified as LOADs. Set explicit LOAD/STORE flags in intel_dkt_pebs_event_constraints[] for: - MEM_UOPS_RETIRED.LOAD_LATENCY - MEM_UOPS_RETIRED.STORE_LATENCY This fixes incorrect STORE sample classification. Additionally remove INTEL_HYBRID_LAT_CONSTRAINT() since no one uses it anymore. Fixes: 65fd435095bb ("perf/x86/intel: Update event constraints for PTL") Signed-off-by: Dapeng Mi Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Cc: # v7.2+ Link: https://patch.msgid.link/20260917015234.981153-4-dapeng1.mi@linux.intel.com --- arch/x86/events/intel/ds.c | 4 ++-- arch/x86/events/perf_event.h | 4 ---- 2 files changed, 2 insertions(+), 6 deletions(-) diff --git a/arch/x86/events/intel/ds.c b/arch/x86/events/intel/ds.c index fa7e0268b4f2..ea5b331c58d1 100644 --- a/arch/x86/events/intel/ds.c +++ b/arch/x86/events/intel/ds.c @@ -1315,8 +1315,8 @@ struct event_constraint intel_cmt_pebs_event_constraints[] = { struct event_constraint intel_dkt_pebs_event_constraints[] = { /* Allow all events as PEBS with no flags */ - INTEL_HYBRID_LAT_CONSTRAINT(0x5d0, 0xff), - INTEL_HYBRID_LAT_CONSTRAINT(0x6d0, 0xff), + INTEL_HYBRID_LDLAT_CONSTRAINT(0x5d0, 0xff), + INTEL_HYBRID_STLAT_CONSTRAINT(0x6d0, 0xff), EVENT_CONSTRAINT_END }; diff --git a/arch/x86/events/perf_event.h b/arch/x86/events/perf_event.h index 4680cba91340..fab9da78a5c7 100644 --- a/arch/x86/events/perf_event.h +++ b/arch/x86/events/perf_event.h @@ -517,10 +517,6 @@ struct cpu_hw_events { __EVENT_CONSTRAINT(c, n, INTEL_ARCH_EVENT_MASK|X86_ALL_EVENT_FLAGS, \ HWEIGHT(n), 0, PERF_X86_EVENT_PEBS_ST) -#define INTEL_HYBRID_LAT_CONSTRAINT(c, n) \ - __EVENT_CONSTRAINT(c, n, INTEL_ARCH_EVENT_MASK|X86_ALL_EVENT_FLAGS, \ - HWEIGHT(n), 0, PERF_X86_EVENT_PEBS_LAT_HYBRID) - #define INTEL_HYBRID_LDLAT_CONSTRAINT(c, n) \ __EVENT_CONSTRAINT(c, n, INTEL_ARCH_EVENT_MASK|X86_ALL_EVENT_FLAGS, \ HWEIGHT(n), 0, PERF_X86_EVENT_PEBS_LAT_HYBRID|PERF_X86_EVENT_PEBS_LD_HSW) From acbe9a3b60b9a6ace8ef11fe898f586251c84592 Mon Sep 17 00:00:00 2001 From: Imre Deak Date: Mon, 7 Sep 2026 20:44:12 +0300 Subject: [PATCH 192/365] drm/i915/dp_mst: Fix configuring FEC for a disconnected stream During an atomic commit after all the MST stream CRTC state is computed the driver ensures that the FEC is configured the same way (enabled or disabled) for all the streams on a given MST topology's link. drm_dp_mst_port_downstream_of_parent() used to determine if a stream is downstream of an MST port will return false if the whole topology is disconnected, since in that case it can't verify that the port/ parent_port passed to it is in the given MST topology. This is a problem during the above FEC configuration check, since intel_dp_mst_check_dsc_change()->get_pipes_downstream_of_mst_ports() will not return all the stream CRTCs/pipes for the topology as expected. Since passing parent_port==NULL to get_pipes_downstream_of_mst_port() is meant to return all the streams for the given topology (i.e. mst_mgr) skip checking if an MST port is downstream of a parent port in this case. This fixes a problem where the FEC configuration check explained above failed to ensure that all streams' FEC is configured the same way if the topology was disconnected, leading to a FEC state mismatch error. Cc: stable@vger.kernel.org # v6.10+ Closes: https://gitlab.freedesktop.org/drm/i915/kernel/-/work_items/16073 Closes: https://gitlab.freedesktop.org/drm/i915/kernel/-/work_items/16384 Reviewed-by: Luca Coelho Signed-off-by: Imre Deak Link: https://patch.msgid.link/20260907174413.741851-1-imre.deak@intel.com (cherry picked from commit 270681fbffbba2b6ccf5b7e3c34b8b563b36167f) Signed-off-by: Jani Nikula --- drivers/gpu/drm/i915/display/intel_dp_mst.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/i915/display/intel_dp_mst.c b/drivers/gpu/drm/i915/display/intel_dp_mst.c index 57daed0b0b36..9e65b4a1c71d 100644 --- a/drivers/gpu/drm/i915/display/intel_dp_mst.c +++ b/drivers/gpu/drm/i915/display/intel_dp_mst.c @@ -852,7 +852,8 @@ static u8 get_pipes_downstream_of_mst_port(struct intel_atomic_state *state, if (&connector->mst.dp->mst.mgr != mst_mgr) continue; - if (connector->mst.port != parent_port && + if (parent_port && + connector->mst.port != parent_port && !drm_dp_mst_port_downstream_of_parent(mst_mgr, connector->mst.port, parent_port)) From a443e0b8d647c1401b110d9f919d8c6cb8607260 Mon Sep 17 00:00:00 2001 From: Imre Deak Date: Mon, 7 Sep 2026 20:44:13 +0300 Subject: [PATCH 193/365] drm/i915/dp_mst: Fix configuring TUs for a disconnected stream During an atomic commit after all the MST stream CRTC state is computed the driver ensures that the sum of TUs of all the streams on a given MST topology link is within limits (63 for 8b10 and 64 for 128b132b). For a disconnected stream the DRM MST core's BW verification doesn't ensure this, because the topology state it uses for this is destroyed as soon as the stream (i.e. MST connector/port) is disconnected. The driver should keep the link state valid even for such disconnected streams, as userspace may disable them one-by-one only in a deferred way. Ensure the link's sum of TUs stays within limits in this case by simply reusing the maximum link BPP limit from the stream's (i.e. CRTC's) old state. The disconnection can happen either via the whole topology getting disconnected or via only the given stream's port getting disconnected. Check for both of these conditions separately, as a connector gets unregistered after a link disconnect event only in a deferred way. Cc: stable@vger.kernel.org # v6.10+ Link: https://gitlab.freedesktop.org/drm/i915/kernel/-/work_items/16073 Link: https://gitlab.freedesktop.org/drm/i915/kernel/-/work_items/16384 Reviewed-by: Luca Coelho Signed-off-by: Imre Deak Link: https://patch.msgid.link/20260907174413.741851-2-imre.deak@intel.com (cherry picked from commit ee00f8fbb2b202002ab90834e02e9ba372773a36) Signed-off-by: Jani Nikula --- drivers/gpu/drm/i915/display/intel_dp_mst.c | 21 ++++++++++++++++++++ drivers/gpu/drm/i915/display/intel_dp_mst.h | 2 ++ drivers/gpu/drm/i915/display/intel_link_bw.c | 3 ++- 3 files changed, 25 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/i915/display/intel_dp_mst.c b/drivers/gpu/drm/i915/display/intel_dp_mst.c index 9e65b4a1c71d..fb3942f56b1f 100644 --- a/drivers/gpu/drm/i915/display/intel_dp_mst.c +++ b/drivers/gpu/drm/i915/display/intel_dp_mst.c @@ -2168,6 +2168,27 @@ bool intel_dp_mst_crtc_needs_modeset(struct intel_atomic_state *state, return false; } +bool intel_dp_mst_stream_disconnected(struct intel_atomic_state *state, + const struct intel_crtc *crtc) +{ + struct intel_connector *connector; + + connector = get_connector_in_state_for_crtc(state, crtc); + if (!connector) + return false; + + if (!connector->mst.dp) + return false; + + if (!connector->mst.dp->mst.mgr.mst_state) + return true; + + if (drm_connector_is_unregistered(&connector->base)) + return true; + + return false; +} + /** * intel_dp_mst_prepare_probe - Prepare an MST link for topology probing * @intel_dp: DP port object diff --git a/drivers/gpu/drm/i915/display/intel_dp_mst.h b/drivers/gpu/drm/i915/display/intel_dp_mst.h index ab09b487c6bb..8ce89242c05c 100644 --- a/drivers/gpu/drm/i915/display/intel_dp_mst.h +++ b/drivers/gpu/drm/i915/display/intel_dp_mst.h @@ -28,6 +28,8 @@ int intel_dp_mst_atomic_check_link(struct intel_atomic_state *state, struct intel_link_bw_limits *limits); bool intel_dp_mst_crtc_needs_modeset(struct intel_atomic_state *state, struct intel_crtc *crtc); +bool intel_dp_mst_stream_disconnected(struct intel_atomic_state *state, + const struct intel_crtc *crtc); void intel_dp_mst_prepare_probe(struct intel_dp *intel_dp); bool intel_dp_mst_verify_dpcd_state(struct intel_dp *intel_dp); diff --git a/drivers/gpu/drm/i915/display/intel_link_bw.c b/drivers/gpu/drm/i915/display/intel_link_bw.c index b47474a3e9fe..e71e76d6fd3e 100644 --- a/drivers/gpu/drm/i915/display/intel_link_bw.c +++ b/drivers/gpu/drm/i915/display/intel_link_bw.c @@ -64,7 +64,8 @@ void intel_link_bw_init_limits(struct intel_atomic_state *state, intel_atomic_get_new_crtc_state(state, crtc); int forced_bpp_x16 = get_forced_link_bpp_x16(state, crtc); - if (state->base.duplicated && crtc_state) { + if ((state->base.duplicated && crtc_state) || + intel_dp_mst_stream_disconnected(state, crtc)) { limits->max_bpp_x16[pipe] = crtc_state->max_link_bpp_x16; if (intel_dsc_enabled_on_link(crtc_state)) limits->link_dsc_pipes |= BIT(pipe); From bb2635be7646a6a9e40a27becb936fe3cdcccf8d Mon Sep 17 00:00:00 2001 From: Jani Nikula Date: Tue, 15 Sep 2026 19:06:20 +0300 Subject: [PATCH 194/365] drm/i915/dp: use EXPORT_SYMBOL_IF_KUNIT() for kunit helpers Use EXPORT_SYMBOL_IF_KUNIT() instead of the regular EXPORT_SYMBOL() to export the symbols to the kunit namespace. Otherwise, the symbols get exported for all the kernel to see, and the corresponding MODULE_IMPORT_NS("EXPORTED_FOR_KUNIT_TESTING") in the tests is meaningless. Fixes: 2eb9982ff179 ("drm/i915/kunit: Export link training and caps funcs for testing") Cc: Imre Deak Reviewed-by: Imre Deak Link: https://patch.msgid.link/20260915160620.779372-1-jani.nikula@intel.com Signed-off-by: Jani Nikula (cherry picked from commit 4ffdb772716e4279d63dfaaadf965da73e799aeb) --- drivers/gpu/drm/i915/display/intel_dp_link_caps.c | 6 ++++-- drivers/gpu/drm/i915/display/intel_dp_link_training.c | 4 ++-- 2 files changed, 6 insertions(+), 4 deletions(-) diff --git a/drivers/gpu/drm/i915/display/intel_dp_link_caps.c b/drivers/gpu/drm/i915/display/intel_dp_link_caps.c index 98657aa4d3d5..abec3e2519ca 100644 --- a/drivers/gpu/drm/i915/display/intel_dp_link_caps.c +++ b/drivers/gpu/drm/i915/display/intel_dp_link_caps.c @@ -3,6 +3,8 @@ * Copyright © 2026 Intel Corporation */ +#include + #include #include #include @@ -1302,14 +1304,14 @@ void intel_dp_link_caps_cleanup(struct intel_dp_link_caps *link_caps) const struct intel_dp_link_caps_test_ops i915_display_dp_link_caps_test_ops = { INTEL_DP_LINK_CAPS_TEST_OPS_INIT }; -EXPORT_SYMBOL(i915_display_dp_link_caps_test_ops); +EXPORT_SYMBOL_IF_KUNIT(i915_display_dp_link_caps_test_ops); #else const struct intel_dp_link_caps_test_ops intel_display_dp_link_caps_test_ops = { INTEL_DP_LINK_CAPS_TEST_OPS_INIT }; -EXPORT_SYMBOL(intel_display_dp_link_caps_test_ops); +EXPORT_SYMBOL_IF_KUNIT(intel_display_dp_link_caps_test_ops); #endif /* I915 */ diff --git a/drivers/gpu/drm/i915/display/intel_dp_link_training.c b/drivers/gpu/drm/i915/display/intel_dp_link_training.c index cb92cff90614..9a692f4fdfee 100644 --- a/drivers/gpu/drm/i915/display/intel_dp_link_training.c +++ b/drivers/gpu/drm/i915/display/intel_dp_link_training.c @@ -2825,14 +2825,14 @@ void intel_dp_link_training_cleanup(struct intel_dp_link_training *link_training const struct intel_dp_link_training_test_ops i915_display_dp_link_training_test_ops = { INTEL_DP_LINK_TRAINING_TEST_OPS_INIT }; -EXPORT_SYMBOL(i915_display_dp_link_training_test_ops); +EXPORT_SYMBOL_IF_KUNIT(i915_display_dp_link_training_test_ops); #else const struct intel_dp_link_training_test_ops intel_display_dp_link_training_test_ops = { INTEL_DP_LINK_TRAINING_TEST_OPS_INIT }; -EXPORT_SYMBOL(intel_display_dp_link_training_test_ops); +EXPORT_SYMBOL_IF_KUNIT(intel_display_dp_link_training_test_ops); #endif /* I915 */ From d2da6696e0c4e60414706e607029d0bb0330c67e Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Christian=20K=C3=B6nig?= Date: Thu, 3 Sep 2026 13:36:21 +0200 Subject: [PATCH 195/365] drm/i915: fix incorrect RCU teardown order MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit i915_gem_busy_ioctl uses dma_resv_for_each_fence_unlocked() to iterate over the fences in an GEM object without holding a reference but only the RCU read side lock. What can happen here is that the GEM object is destroyed concurrently while i915_gem_busy_ioctl is still running. This won't free the GEM objects memory, but still drops all the dma_fence references. Now when dma_resv_for_each_fence_unlocked() sees a destroyed dma_fence it assumes that a new fence list was installed and re-starts the loop. But in the case of a destroyed GEM object a new fence list is never installed, only the old one freed and therefore the iteration never finishes resulting in an endless loop. The solution is to drop the fence references only after the RCU grace period. The fixes tag is not necessary the patch introducing the problem, but the one making it so worse that we need to address it. This problem was pointed out by Sashiko-bot. Signed-off-by: Christian König Fixes: 912ff2ebd695 ("drm/i915: use the new iterator in i915_gem_busy_ioctl v2") CC: stable@vger.kernel.org Reviewed-by: Tvrtko Ursulin Signed-off-by: Tvrtko Ursulin Link: https://lore.kernel.org/r/20260903113621.54660-1-christian.koenig@amd.com (cherry picked from commit 5113479556025093bf8133bb2dcaa33be2d50921) Signed-off-by: Jani Nikula --- drivers/gpu/drm/i915/gem/i915_gem_object.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_object.c b/drivers/gpu/drm/i915/gem/i915_gem_object.c index 5172d3982654..9e01f8b2079a 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_object.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_object.c @@ -89,6 +89,7 @@ struct drm_i915_gem_object *i915_gem_object_alloc(void) void i915_gem_object_free(struct drm_i915_gem_object *obj) { + dma_resv_fini(&obj->base._resv); return kmem_cache_free(slab_objects, obj); } @@ -144,7 +145,6 @@ void __i915_gem_object_fini(struct drm_i915_gem_object *obj) { mutex_destroy(&obj->mm.get_page.lock); mutex_destroy(&obj->mm.get_dma_page.lock); - dma_resv_fini(&obj->base._resv); } /** From 335b0642812ee0f2b7798b634ce507b8b7c9fe0c Mon Sep 17 00:00:00 2001 From: Dapeng Mi Date: Thu, 17 Sep 2026 09:52:26 +0800 Subject: [PATCH 196/365] perf/x86/intel: Update arw_latency_data() mem-op direction handling Align arw_latency_data() with other *_latency_data() helpers by explicitly decoding LOAD/STORE event flags when setting the sampled memory operation direction. This keeps the latency data path behavior consistent across platforms and avoids relying on implicit direction inference. Signed-off-by: Dapeng Mi Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Link: https://patch.msgid.link/20260917015234.981153-5-dapeng1.mi@linux.intel.com --- arch/x86/events/intel/ds.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/arch/x86/events/intel/ds.c b/arch/x86/events/intel/ds.c index ea5b331c58d1..c0004b08b087 100644 --- a/arch/x86/events/intel/ds.c +++ b/arch/x86/events/intel/ds.c @@ -538,7 +538,11 @@ static u64 arw_latency_data(struct perf_event *event, u64 status) val |= P(BLK, NA); src.val = val; - if (event->hw.flags & PERF_X86_EVENT_PEBS_ST_HSW) + if (event->hw.flags & + (PERF_X86_EVENT_PEBS_LDLAT | PERF_X86_EVENT_PEBS_LD_HSW)) + src.mem_op = P(OP, LOAD); + if (event->hw.flags & + (PERF_X86_EVENT_PEBS_STLAT | PERF_X86_EVENT_PEBS_ST_HSW)) src.mem_op = P(OP, STORE); return src.val; From 7c944595cc43a664019edc22505b6d6f6039be5e Mon Sep 17 00:00:00 2001 From: Dapeng Mi Date: Thu, 17 Sep 2026 09:52:27 +0800 Subject: [PATCH 197/365] perf/x86/intel: Remove incorrect LionCove PEBS data-source constraints On Lion Cove, PEBS data source is valid only for these events: - MEM_TRANS_RETIRED.LOAD_LATENCY (0x1cd) - MEM_TRANS_RETIRED.STORE_SAMPLE (0x2cd) The perfmon database (https://github.com/intel/perfmon) previously tagged additional memory events such as MEM_INST_RETIRED.STLB_MISS_LOADS with L1_Hit_Indication, implying PEBS data-source support, which is incorrect. The database has since been fixed, but intel_lnc_pebs_event_constraints[] still follows the old definition and marks those events as data-source capable. As a result, get_data_src() may decode data-source information for events that do not provide valid PEBS data-source data and mislead users. Remove those non-data-source memory events from the Lion Cove PEBS constraint table so matching falls back to the regular non-PEBS constraints, which already provide the same counter constraints. Also update lnc_latency_data() to decode LOAD/STORE flags explicitly when setting memory operation direction, for consistency with other *_latency_data() helpers. Fixes: a932aa0e868f ("perf/x86: Add Lunar Lake and Arrow Lake support") Signed-off-by: Dapeng Mi Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Cc: Link: https://patch.msgid.link/20260917015234.981153-6-dapeng1.mi@linux.intel.com --- arch/x86/events/intel/ds.c | 22 +++++----------------- 1 file changed, 5 insertions(+), 17 deletions(-) diff --git a/arch/x86/events/intel/ds.c b/arch/x86/events/intel/ds.c index c0004b08b087..cbcc0b36536b 100644 --- a/arch/x86/events/intel/ds.c +++ b/arch/x86/events/intel/ds.c @@ -577,7 +577,11 @@ static u64 lnc_latency_data(struct perf_event *event, u64 status) val |= P(BLK, NA); src.val = val; - if (event->hw.flags & PERF_X86_EVENT_PEBS_ST_HSW) + if (event->hw.flags & + (PERF_X86_EVENT_PEBS_LDLAT | PERF_X86_EVENT_PEBS_LD_HSW)) + src.mem_op = P(OP, LOAD); + if (event->hw.flags & + (PERF_X86_EVENT_PEBS_STLAT | PERF_X86_EVENT_PEBS_ST_HSW)) src.mem_op = P(OP, STORE); return src.val; @@ -1510,24 +1514,8 @@ struct event_constraint intel_lnc_pebs_event_constraints[] = { INTEL_FLAGS_UEVENT_CONSTRAINT(0x012a, 0x1), /* OCR.* events */ INTEL_FLAGS_UEVENT_CONSTRAINT(0x012b, 0x1), /* OCR.* events */ - INTEL_FLAGS_UEVENT_CONSTRAINT(0x04a4, 0x1), /* TOPDOWN.BAD_SPEC_SLOTS */ - INTEL_FLAGS_UEVENT_CONSTRAINT(0x08a4, 0x1), /* TOPDOWN.BR_MISPREDICT_SLOTS */ - INTEL_FLAGS_UEVENT_CONSTRAINT(0x10a4, 0x8), /* TOPDOWN.MEMORY_BOUND_SLOTS */ - INTEL_HYBRID_LDLAT_CONSTRAINT(0x1cd, 0x3fc), INTEL_HYBRID_STLAT_CONSTRAINT(0x2cd, 0x3), - INTEL_FLAGS_UEVENT_CONSTRAINT_DATALA_LD(0x11d0, 0xf), /* MEM_INST_RETIRED.STLB_MISS_LOADS */ - INTEL_FLAGS_UEVENT_CONSTRAINT_DATALA_ST(0x12d0, 0xf), /* MEM_INST_RETIRED.STLB_MISS_STORES */ - INTEL_FLAGS_UEVENT_CONSTRAINT_DATALA_LD(0x21d0, 0xf), /* MEM_INST_RETIRED.LOCK_LOADS */ - INTEL_FLAGS_UEVENT_CONSTRAINT_DATALA_LD(0x41d0, 0xf), /* MEM_INST_RETIRED.SPLIT_LOADS */ - INTEL_FLAGS_UEVENT_CONSTRAINT_DATALA_ST(0x42d0, 0xf), /* MEM_INST_RETIRED.SPLIT_STORES */ - INTEL_FLAGS_UEVENT_CONSTRAINT_DATALA_LD(0x81d0, 0xf), /* MEM_INST_RETIRED.ALL_LOADS */ - INTEL_FLAGS_UEVENT_CONSTRAINT_DATALA_ST(0x82d0, 0xf), /* MEM_INST_RETIRED.ALL_STORES */ - INTEL_FLAGS_UEVENT_CONSTRAINT(0x87d0, 0x3ff), /* MEM_INST_RETIRED.ANY */ - - INTEL_FLAGS_EVENT_CONSTRAINT_DATALA_LD_RANGE(0xd1, 0xd4, 0xf), - - INTEL_FLAGS_EVENT_CONSTRAINT(0xd0, 0xf), /* * Everything else is handled by PMU_FL_PEBS_ALL, because we From 9f93d33ad65af9853974c1ec4ba1f937e8ba1376 Mon Sep 17 00:00:00 2001 From: Dapeng Mi Date: Thu, 17 Sep 2026 09:52:28 +0800 Subject: [PATCH 198/365] perf/x86/intel: Remove incorrect Panther Cove PEBS data-source constraints Same issue exists on Panther Cove, PEBS data source is valid only for these events: - MEM_TRANS_RETIRED.LOAD_LATENCY (0x1cd) - MEM_TRANS_RETIRED.STORE_SAMPLE (0x2cd) The perfmon database (https://github.com/intel/perfmon) previously tagged additional memory events such as MEM_INST_RETIRED.STLB_MISS_LOADS with L1_Hit_Indication, implying PEBS data-source support, which is incorrect. The database has since been fixed, but intel_pnc_pebs_event_constraints[] still follows the old definition and marks those events as data-source capable. As a result, get_data_src() may decode data-source information for events that do not provide valid PEBS data-source data and mislead users. Remove those non-data-source memory events from the Pather Cove PEBS constraint table so matching falls back to the regular non-PEBS constraints, which already provide the same counter constraints. Also update pnc_latency_data() to decode LOAD/STORE flags explicitly when setting memory operation direction, for consistency with other *_latency_data() helpers. Fixes: d345b6bb8860 ("perf/x86/intel: Add core PMU support for DMR") Signed-off-by: Dapeng Mi Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Cc: # v7.0+ Link: https://patch.msgid.link/20260917015234.981153-7-dapeng1.mi@linux.intel.com --- arch/x86/events/intel/ds.c | 18 +++++------------- 1 file changed, 5 insertions(+), 13 deletions(-) diff --git a/arch/x86/events/intel/ds.c b/arch/x86/events/intel/ds.c index cbcc0b36536b..72179bcf6d0d 100644 --- a/arch/x86/events/intel/ds.c +++ b/arch/x86/events/intel/ds.c @@ -639,7 +639,11 @@ u64 pnc_latency_data(struct perf_event *event, u64 status) val |= P(BLK, NA); src.val = val; - if (event->hw.flags & PERF_X86_EVENT_PEBS_ST_HSW) + if (event->hw.flags & + (PERF_X86_EVENT_PEBS_LDLAT | PERF_X86_EVENT_PEBS_LD_HSW)) + src.mem_op = P(OP, LOAD); + if (event->hw.flags & + (PERF_X86_EVENT_PEBS_STLAT | PERF_X86_EVENT_PEBS_ST_HSW)) src.mem_op = P(OP, STORE); return src.val; @@ -1528,18 +1532,6 @@ struct event_constraint intel_lnc_pebs_event_constraints[] = { struct event_constraint intel_pnc_pebs_event_constraints[] = { INTEL_HYBRID_LDLAT_CONSTRAINT(0x1cd, 0xfc), INTEL_HYBRID_STLAT_CONSTRAINT(0x2cd, 0x3), - INTEL_FLAGS_UEVENT_CONSTRAINT_DATALA_LD(0x11d0, 0xf), /* MEM_INST_RETIRED.STLB_MISS_LOADS */ - INTEL_FLAGS_UEVENT_CONSTRAINT_DATALA_ST(0x12d0, 0xf), /* MEM_INST_RETIRED.STLB_MISS_STORES */ - INTEL_FLAGS_UEVENT_CONSTRAINT_DATALA_LD(0x21d0, 0xf), /* MEM_INST_RETIRED.LOCK_LOADS */ - INTEL_FLAGS_UEVENT_CONSTRAINT_DATALA_LD(0x41d0, 0xf), /* MEM_INST_RETIRED.SPLIT_LOADS */ - INTEL_FLAGS_UEVENT_CONSTRAINT_DATALA_ST(0x42d0, 0xf), /* MEM_INST_RETIRED.SPLIT_STORES */ - INTEL_FLAGS_UEVENT_CONSTRAINT_DATALA_LD(0x81d0, 0xf), /* MEM_INST_RETIRED.ALL_LOADS */ - INTEL_FLAGS_UEVENT_CONSTRAINT_DATALA_ST(0x82d0, 0xf), /* MEM_INST_RETIRED.ALL_STORES */ - - INTEL_FLAGS_EVENT_CONSTRAINT_DATALA_LD_RANGE(0xd1, 0xd4, 0xf), - - INTEL_FLAGS_EVENT_CONSTRAINT(0xd0, 0xf), - INTEL_FLAGS_EVENT_CONSTRAINT(0xd6, 0xf), /* * Everything else is handled by PMU_FL_PEBS_ALL, because we From 0ac5d6ca2c3b7d047963b67dccef17902dc6c017 Mon Sep 17 00:00:00 2001 From: Dapeng Mi Date: Thu, 17 Sep 2026 09:52:29 +0800 Subject: [PATCH 199/365] perf/x86/intel: Fix Panther Cove PEBS data-source snoop states For Panther Cove, the snoop states for the data source encodings "Prefetch Promotion" and "Cross Core Prefetch Promotion" should be SNOOP_NONE instead of SNOOP_MISS. Fix the incorrect snooping states for Panther Cove. Fixes: d2bdcde9626c ("perf/x86/intel: Add support for PEBS memory auxiliary info field in DMR") Signed-off-by: Dapeng Mi Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Cc: # v7.0+ Link: https://patch.msgid.link/20260917015234.981153-8-dapeng1.mi@linux.intel.com --- arch/x86/events/intel/ds.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/arch/x86/events/intel/ds.c b/arch/x86/events/intel/ds.c index 72179bcf6d0d..d0329a7eb8a5 100644 --- a/arch/x86/events/intel/ds.c +++ b/arch/x86/events/intel/ds.c @@ -277,8 +277,8 @@ static u64 pnc_pebs_l2_hit_data_source[PNC_PEBS_DATA_SOURCE_MAX] = { 0, /* 0x06: Reserved */ OP_LH | P(LVL, L2) | LEVEL(L2) | P(SNOOP, HIT), /* 0x07: L2 Hit Snoop HIT */ OP_LH | P(LVL, L2) | LEVEL(L2) | P(SNOOP, HITM), /* 0x08: L2 Hit Snoop Hit Modified */ - OP_LH | P(LVL, L2) | LEVEL(L2) | P(SNOOP, MISS), /* 0x09: Prefetch Promotion */ - OP_LH | P(LVL, L2) | LEVEL(L2) | P(SNOOP, MISS), /* 0x0a: Cross Core Prefetch Promotion */ + OP_LH | P(LVL, L2) | LEVEL(L2) | P(SNOOP, NONE), /* 0x09: Prefetch Promotion */ + OP_LH | P(LVL, L2) | LEVEL(L2) | P(SNOOP, NONE), /* 0x0a: Cross Core Prefetch Promotion */ 0, /* 0x0b: Reserved */ 0, /* 0x0c: Reserved */ 0, /* 0x0d: Reserved */ From 858b37ca19d3f695f7fa94cd14be5a856fd8e7d7 Mon Sep 17 00:00:00 2001 From: Dapeng Mi Date: Thu, 17 Sep 2026 09:52:30 +0800 Subject: [PATCH 200/365] perf/x86/intel: Delete dead NVL PEBS data-source initcall Nova Lake now uses the OMR data-source table for PEBS data-source decoding and no longer depends on the legacy static pebs_data_source[] mapping. Remove the dead intel_pmu_pebs_data_source_lnl() initialization call for NVL. Fixes: c847a208f43b ("perf/x86/intel: Add core PMU support for Novalake") Signed-off-by: Dapeng Mi Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Link: https://patch.msgid.link/20260917015234.981153-9-dapeng1.mi@linux.intel.com --- arch/x86/events/intel/core.c | 2 -- 1 file changed, 2 deletions(-) diff --git a/arch/x86/events/intel/core.c b/arch/x86/events/intel/core.c index 757d4082de60..a7dbbed8f94e 100644 --- a/arch/x86/events/intel/core.c +++ b/arch/x86/events/intel/core.c @@ -8824,8 +8824,6 @@ __init int intel_pmu_init(void) /* Initialize Atom core specific PerfMon capabilities.*/ pmu = &x86_pmu.hybrid_pmu[X86_HYBRID_PMU_ATOM_IDX]; intel_pmu_init_arw(&pmu->pmu); - - intel_pmu_pebs_data_source_lnl(); break; default: From 04a7ef3b7aa3af34202285043e3423a2e3983595 Mon Sep 17 00:00:00 2001 From: Dapeng Mi Date: Thu, 17 Sep 2026 09:52:31 +0800 Subject: [PATCH 201/365] perf/x86/intel: Constrain Panther Cove UOPS_DISPATCHED events to PMCs 0-3 Per the latest Panther Cove event definitions, the following events are only supported on PMCs 0-3: - UOPS_DISPATCHED.INT_EU_ALL (0x1b2) - UOPS_DISPATCHED.ALU (0x2b2) Add explicit event constraints for these two events so scheduling does not place them on unsupported counters. Fixes: d345b6bb8860 ("perf/x86/intel: Add core PMU support for DMR") Signed-off-by: Dapeng Mi Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Cc: # v7.0+ Link: https://patch.msgid.link/20260917015234.981153-10-dapeng1.mi@linux.intel.com --- arch/x86/events/intel/core.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/arch/x86/events/intel/core.c b/arch/x86/events/intel/core.c index a7dbbed8f94e..ecf6d6325dfe 100644 --- a/arch/x86/events/intel/core.c +++ b/arch/x86/events/intel/core.c @@ -506,6 +506,8 @@ static struct event_constraint intel_pnc_event_constraints[] = { INTEL_EVENT_CONSTRAINT(0xce, 0x1), INTEL_UEVENT_CONSTRAINT(0x01b1, 0x8), + INTEL_UEVENT_CONSTRAINT(0x01b2, 0xf), + INTEL_UEVENT_CONSTRAINT(0x02b2, 0xf), INTEL_UEVENT_CONSTRAINT(0x0847, 0xf), INTEL_UEVENT_CONSTRAINT(0x0446, 0xf), INTEL_UEVENT_CONSTRAINT(0x0846, 0xf), From ff1621adfdd7cfb73f2ccdd3856cd3462ac98ac5 Mon Sep 17 00:00:00 2001 From: Dapeng Mi Date: Thu, 17 Sep 2026 09:52:32 +0800 Subject: [PATCH 202/365] perf/x86/intel: Fix precise OMR event scheduling for DMR/NVL The latest perfmon event database introduces below precise OMR event support for DMR/NVL: - MEM_LOAD_L2_MISS_RETIRED.* (event 0xd6) - MEM_STORE_L2_MISS_RETIRED.* (event 0x4f) These events use the same OMR MSRs as the existing OMR events, but they are not listed in intel_pnc_extra_regs[]. As a result, perf cannot assign the required OMR extra registers when scheduling them. Add the new precise OMR events to intel_pnc_extra_regs[] so they can be scheduled with the correct OMR MSRs. MEM_LOAD_L2_MISS_RETIRED.* remains limited to GP counters 0-3, while MEM_STORE_L2_MISS_RETIRED.* is available on all GP counters. Signed-off-by: Dapeng Mi Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Link: https://patch.msgid.link/20260917015234.981153-11-dapeng1.mi@linux.intel.com --- arch/x86/events/intel/core.c | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/arch/x86/events/intel/core.c b/arch/x86/events/intel/core.c index ecf6d6325dfe..22715602f7f8 100644 --- a/arch/x86/events/intel/core.c +++ b/arch/x86/events/intel/core.c @@ -522,6 +522,14 @@ static struct extra_reg intel_pnc_extra_regs[] __read_mostly = { INTEL_UEVENT_EXTRA_REG(0x022a, MSR_OMR_1, 0x40ffffff0000ffffull, OMR_1), INTEL_UEVENT_EXTRA_REG(0x042a, MSR_OMR_2, 0x40ffffff0000ffffull, OMR_2), INTEL_UEVENT_EXTRA_REG(0x082a, MSR_OMR_3, 0x40ffffff0000ffffull, OMR_3), + INTEL_UEVENT_EXTRA_REG(0x014f, MSR_OMR_0, 0x40ffffff0000ffffull, OMR_0), + INTEL_UEVENT_EXTRA_REG(0x024f, MSR_OMR_1, 0x40ffffff0000ffffull, OMR_1), + INTEL_UEVENT_EXTRA_REG(0x044f, MSR_OMR_2, 0x40ffffff0000ffffull, OMR_2), + INTEL_UEVENT_EXTRA_REG(0x084f, MSR_OMR_3, 0x40ffffff0000ffffull, OMR_3), + INTEL_UEVENT_EXTRA_REG(0x01d6, MSR_OMR_0, 0x40ffffff0000ffffull, OMR_0), + INTEL_UEVENT_EXTRA_REG(0x02d6, MSR_OMR_1, 0x40ffffff0000ffffull, OMR_1), + INTEL_UEVENT_EXTRA_REG(0x04d6, MSR_OMR_2, 0x40ffffff0000ffffull, OMR_2), + INTEL_UEVENT_EXTRA_REG(0x08d6, MSR_OMR_3, 0x40ffffff0000ffffull, OMR_3), INTEL_UEVENT_PEBS_LDLAT_EXTRA_REG(0x01cd), INTEL_UEVENT_EXTRA_REG(0x02c6, MSR_PEBS_FRONTEND, 0x9, FE), INTEL_UEVENT_EXTRA_REG(0x03c6, MSR_PEBS_FRONTEND, 0x7fff1f, FE), From 0102c8c7fdfc4bd700971daff2b94470f4eafae4 Mon Sep 17 00:00:00 2001 From: Dapeng Mi Date: Thu, 17 Sep 2026 09:52:33 +0800 Subject: [PATCH 203/365] perf/x86/intel: Rename DMR offcore_rsp attribute to offmodule_rsp DMR introduces Offmodule Response events in place of the legacy Offcore Response events, but it still exposes the inherited offcore_rsp PMU attribute for programming the corresponding MSR data. Rename the DMR PMU attribute to offmodule_rsp so the sysfs interface matches the underlying event name and avoids user & tooling confusion. Signed-off-by: Dapeng Mi Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Link: https://patch.msgid.link/20260917015234.981153-12-dapeng1.mi@linux.intel.com --- arch/x86/events/intel/core.c | 23 ++++++++++++++++++++--- 1 file changed, 20 insertions(+), 3 deletions(-) diff --git a/arch/x86/events/intel/core.c b/arch/x86/events/intel/core.c index 22715602f7f8..3843dbe530a5 100644 --- a/arch/x86/events/intel/core.c +++ b/arch/x86/events/intel/core.c @@ -6601,6 +6601,8 @@ static void intel_pmu_filter(struct pmu *pmu, int cpu, bool *ret) PMU_FORMAT_ATTR(offcore_rsp, "config1:0-63"); +PMU_FORMAT_ATTR(offmodule_rsp, "config1:0-63"); + PMU_FORMAT_ATTR(ldlat, "config1:0-15"); PMU_FORMAT_ATTR(frontend, "config1:0-23"); @@ -6649,6 +6651,20 @@ static struct attribute *skl_format_attr[] = { NULL, }; +static struct attribute *pnc_format_attr_rtm[] = { + &format_attr_in_tx.attr, + &format_attr_in_tx_cp.attr, + &format_attr_offmodule_rsp.attr, + &format_attr_ldlat.attr, + NULL +}; + +static struct attribute *pnc_format_attr[] = { + &format_attr_offmodule_rsp.attr, + &format_attr_ldlat.attr, + NULL +}; + static __initconst const struct x86_pmu core_pmu = { .name = "core", .handle_irq = x86_pmu_handle_irq, @@ -8589,6 +8605,8 @@ __init int intel_pmu_init(void) case INTEL_DIAMONDRAPIDS_X: intel_pmu_init_pnc(NULL); x86_pmu.pebs_latency_data = pnc_latency_data; + extra_attr = boot_cpu_has(X86_FEATURE_RTM) ? + pnc_format_attr_rtm : pnc_format_attr; pr_cont("Panthercove events, "); name = "panthercove"; @@ -8597,13 +8615,12 @@ __init int intel_pmu_init(void) glc_common: intel_pmu_init_glc(NULL); intel_pmu_pebs_data_source_skl(true); - + extra_attr = boot_cpu_has(X86_FEATURE_RTM) ? + hsw_format_attr : nhm_format_attr; glc_base: x86_pmu.pebs_ept = 1; x86_pmu.hw_config = hsw_hw_config; x86_pmu.get_event_constraints = glc_get_event_constraints; - extra_attr = boot_cpu_has(X86_FEATURE_RTM) ? - hsw_format_attr : nhm_format_attr; extra_skl_attr = skl_format_attr; mem_attr = glc_events_attrs; td_attr = glc_td_events_attrs; From d4d9ccbad527af115b8e83a7a2b74785c0e8dfea Mon Sep 17 00:00:00 2001 From: Dapeng Mi Date: Thu, 17 Sep 2026 09:52:34 +0800 Subject: [PATCH 204/365] perf/x86/intel: Rename NVL offcore_rsp attribute to offmodule_rsp NVL introduces Offmodule Response events in place of the legacy Offcore Response events, but it still exposes the inherited offcore_rsp PMU attribute for programming the corresponding MSR data. Rename the NVL PMU attribute to offmodule_rsp so the sysfs interface matches the underlying event name and avoids user & tooling confusion. Signed-off-by: Dapeng Mi Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Link: https://patch.msgid.link/20260917015234.981153-13-dapeng1.mi@linux.intel.com --- arch/x86/events/intel/core.c | 20 +++++++++++++++++++- 1 file changed, 19 insertions(+), 1 deletion(-) diff --git a/arch/x86/events/intel/core.c b/arch/x86/events/intel/core.c index 3843dbe530a5..3ef80882843e 100644 --- a/arch/x86/events/intel/core.c +++ b/arch/x86/events/intel/core.c @@ -7530,6 +7530,7 @@ static struct attribute *adl_hybrid_tsx_attrs[] = { FORMAT_ATTR_HYBRID(in_tx, hybrid_big); FORMAT_ATTR_HYBRID(in_tx_cp, hybrid_big); FORMAT_ATTR_HYBRID(offcore_rsp, hybrid_big_small_tiny); +FORMAT_ATTR_HYBRID(offmodule_rsp, hybrid_big_small_tiny); FORMAT_ATTR_HYBRID(ldlat, hybrid_big_small_tiny); FORMAT_ATTR_HYBRID(frontend, hybrid_big); @@ -7568,6 +7569,23 @@ static struct attribute *mtl_hybrid_extra_attr[] = { NULL }; +static struct attribute *nvl_hybrid_extra_attr_rtm[] = { + ADL_HYBRID_RTM_FORMAT_ATTR, + FORMAT_HYBRID_PTR(offmodule_rsp), + FORMAT_HYBRID_PTR(ldlat), + FORMAT_HYBRID_PTR(frontend), + FORMAT_HYBRID_PTR(snoop_rsp), + NULL +}; + +static struct attribute *nvl_hybrid_extra_attr[] = { + FORMAT_HYBRID_PTR(offmodule_rsp), + FORMAT_HYBRID_PTR(ldlat), + FORMAT_HYBRID_PTR(frontend), + FORMAT_HYBRID_PTR(snoop_rsp), + NULL +}; + static bool is_attr_for_this_pmu(struct kobject *kobj, struct attribute *attr) { struct device *dev = kobj_to_dev(kobj); @@ -8842,7 +8860,7 @@ __init int intel_pmu_init(void) mem_attr = mtl_hybrid_mem_attrs; tsx_attr = adl_hybrid_tsx_attrs; extra_attr = boot_cpu_has(X86_FEATURE_RTM) ? - mtl_hybrid_extra_attr_rtm : mtl_hybrid_extra_attr; + nvl_hybrid_extra_attr_rtm : nvl_hybrid_extra_attr; /* Initialize big core specific PerfMon capabilities.*/ pmu = &x86_pmu.hybrid_pmu[X86_HYBRID_PMU_CORE_IDX]; From 09b7040a1b79f4f61cdad6d9af972e045bb7c498 Mon Sep 17 00:00:00 2001 From: Niklas Schnelle Date: Wed, 16 Sep 2026 17:14:10 +0200 Subject: [PATCH 205/365] s390/pci: Fix leak of struct pci_dev reference in zpci_report_status() In zpci_report_status(), a reference to the pdev associated with the zdev being reported about is acquired using pci_get_slot(). This reference needs to be dropped with pci_dev_put(), but this call is missing, thus leaking the reference. On subsequent hot unplug, this will cause the struct pci_dev to not be released, leaking memory and preventing reattach. At the same time, the only existing caller already holds a pdev reference. So instead of reacquiring and then dropping another reference, simply pass the existing pdev pointer to zpci_report_status(). This gets rid of the need for pci_get_slot() as well as the zdev->zbus check. Cc: stable@vger.kernel.org Fixes: 4ec6054e7321 ("s390/pci: Report PCI error recovery results via SCLP") Signed-off-by: Niklas Schnelle Reviewed-by: Benjamin Block Reviewed-by: Farhan Ali Signed-off-by: Heiko Carstens --- arch/s390/pci/pci_event.c | 2 +- arch/s390/pci/pci_report.c | 11 +++++------ arch/s390/pci/pci_report.h | 4 +++- 3 files changed, 9 insertions(+), 8 deletions(-) diff --git a/arch/s390/pci/pci_event.c b/arch/s390/pci/pci_event.c index f317a1465dad..3b4941b65840 100644 --- a/arch/s390/pci/pci_event.c +++ b/arch/s390/pci/pci_event.c @@ -298,7 +298,7 @@ static pci_ers_result_t zpci_event_attempt_error_recovery(struct pci_dev *pdev, pci_uevent_ers(pdev, PCI_ERS_RESULT_RECOVERED); out_unlock: device_unlock(&pdev->dev); - zpci_report_status(zdev, "recovery", status_str); + zpci_report_status(zdev, pdev, "recovery", status_str); return ers_res; } diff --git a/arch/s390/pci/pci_report.c b/arch/s390/pci/pci_report.c index 7030f7052926..867419779219 100644 --- a/arch/s390/pci/pci_report.c +++ b/arch/s390/pci/pci_report.c @@ -89,7 +89,8 @@ static struct debug_view debug_log_view = { /** * zpci_report_status - Report the status of operations on a PCI device - * @zdev: The PCI device for which to report status + * @zdev: The zPCI device for which to report status + * @pdev: The PCI device associated with the zdev if any, NULL otherwise * @operation: A string representing the operation reported * @status: A string representing the status of the operation * @@ -103,15 +104,15 @@ static struct debug_view debug_log_view = { * * Return: 0 on success an error code < 0 otherwise. */ -int zpci_report_status(struct zpci_dev *zdev, const char *operation, const char *status) +int zpci_report_status(struct zpci_dev *zdev, struct pci_dev *pdev, + const char *operation, const char *status) { struct zpci_report_error *report; struct pci_driver *driver = NULL; - struct pci_dev *pdev = NULL; char *buf, *end; int ret; - if (!zdev || !zdev->zbus) + if (!zdev) return -ENODEV; /* Protected virtualization hosts get nothing from us */ @@ -121,8 +122,6 @@ int zpci_report_status(struct zpci_dev *zdev, const char *operation, const char report = (void *)get_zeroed_page(GFP_KERNEL); if (!report) return -ENOMEM; - if (zdev->zbus->bus) - pdev = pci_get_slot(zdev->zbus->bus, zdev->devfn); if (pdev) driver = to_pci_driver(pdev->dev.driver); diff --git a/arch/s390/pci/pci_report.h b/arch/s390/pci/pci_report.h index e08003d51a97..dd7b0b05001c 100644 --- a/arch/s390/pci/pci_report.h +++ b/arch/s390/pci/pci_report.h @@ -8,9 +8,11 @@ */ #ifndef __S390_PCI_REPORT_H #define __S390_PCI_REPORT_H +#include struct zpci_dev; -int zpci_report_status(struct zpci_dev *zdev, const char *operation, const char *status); +int zpci_report_status(struct zpci_dev *zdev, struct pci_dev *pdev, + const char *operation, const char *status); #endif /* __S390_PCI_REPORT_H */ From 0261aef4b15efcee2860ab857e5cb05e9bfa47b0 Mon Sep 17 00:00:00 2001 From: Niklas Schnelle Date: Wed, 16 Sep 2026 17:14:11 +0200 Subject: [PATCH 206/365] s390/pci: Fix missing device lock in zpci_report_status() When pdev is non-NULL, zpci_report_status() accesses the device's driver. To get a consistent state matching the recovery, the device lock needs to be held. Do so by expanding the existing device lock critical section. The lock only needs to be held when the pdev is non-NULL, so extract the pdev-specific reporting into a helper function which also adds a lockdep assertion to detect calls without the device lock held. Cc: stable@vger.kernel.org Fixes: 4ec6054e7321 ("s390/pci: Report PCI error recovery results via SCLP") Signed-off-by: Niklas Schnelle Reviewed-by: Benjamin Block Reviewed-by: Farhan Ali Signed-off-by: Heiko Carstens --- arch/s390/pci/pci_event.c | 2 +- arch/s390/pci/pci_report.c | 21 +++++++++++++++------ 2 files changed, 16 insertions(+), 7 deletions(-) diff --git a/arch/s390/pci/pci_event.c b/arch/s390/pci/pci_event.c index 3b4941b65840..ec93f34b6e19 100644 --- a/arch/s390/pci/pci_event.c +++ b/arch/s390/pci/pci_event.c @@ -297,8 +297,8 @@ static pci_ers_result_t zpci_event_attempt_error_recovery(struct pci_dev *pdev, driver->err_handler->resume(pdev); pci_uevent_ers(pdev, PCI_ERS_RESULT_RECOVERED); out_unlock: - device_unlock(&pdev->dev); zpci_report_status(zdev, pdev, "recovery", status_str); + device_unlock(&pdev->dev); return ers_res; } diff --git a/arch/s390/pci/pci_report.c b/arch/s390/pci/pci_report.c index 867419779219..72ecabf7003c 100644 --- a/arch/s390/pci/pci_report.c +++ b/arch/s390/pci/pci_report.c @@ -87,6 +87,19 @@ static struct debug_view debug_log_view = { NULL }; +static ssize_t zpci_report_pdev(struct pci_dev *pdev, char *buf, size_t size) +{ + struct pci_driver *driver; + const char *start = buf; + char *end = buf + size; + + device_lock_assert(&pdev->dev); + buf += scnprintf(buf, end - buf, "state: %s\n", zpci_state_str(pdev->error_state)); + driver = to_pci_driver(pdev->dev.driver); + buf += scnprintf(buf, end - buf, "driver: %s\n", (driver) ? driver->name : "n/a"); + return buf - start; +} + /** * zpci_report_status - Report the status of operations on a PCI device * @zdev: The zPCI device for which to report status @@ -108,7 +121,6 @@ int zpci_report_status(struct zpci_dev *zdev, struct pci_dev *pdev, const char *operation, const char *status) { struct zpci_report_error *report; - struct pci_driver *driver = NULL; char *buf, *end; int ret; @@ -122,16 +134,13 @@ int zpci_report_status(struct zpci_dev *zdev, struct pci_dev *pdev, report = (void *)get_zeroed_page(GFP_KERNEL); if (!report) return -ENOMEM; - if (pdev) - driver = to_pci_driver(pdev->dev.driver); buf = report->data.log_data; end = report->data.log_data + ZPCI_REPORT_DATA_SIZE; buf += scnprintf(buf, end - buf, "report: %s\n", operation); buf += scnprintf(buf, end - buf, "status: %s\n", status); - buf += scnprintf(buf, end - buf, "state: %s\n", - (pdev) ? zpci_state_str(pdev->error_state) : "n/a"); - buf += scnprintf(buf, end - buf, "driver: %s\n", (driver) ? driver->name : "n/a"); + if (pdev) + buf += zpci_report_pdev(pdev, buf, end - buf); ret = debug_dump(pci_debug_msg_id, &debug_log_view, buf, end - buf, true); if (ret < 0) pr_err("Reading PCI debug messages failed with code %d\n", ret); From a1120bea9bc8ea9d9ab2f9904a63b9a228bf2ffc Mon Sep 17 00:00:00 2001 From: Niklas Schnelle Date: Wed, 16 Sep 2026 17:14:12 +0200 Subject: [PATCH 207/365] s390/pci: Report SCLP status on error events when no pdev is associated With commit 4ec6054e7321 ("s390/pci: Report PCI error recovery results via SCLP") SCLP reports are generated when recovery is performed in response to an error event. If such an error event arrives but no pdev is currently associated with the zdev, e.g. because it was removed or not yet probed, no report is generated. Fix this by generating a report specific to an error event for a zdev without an associated pdev. Cc: stable@vger.kernel.org Fixes: 4ec6054e7321 ("s390/pci: Report PCI error recovery results via SCLP") Signed-off-by: Niklas Schnelle Reviewed-by: Benjamin Block Reviewed-by: Farhan Ali Signed-off-by: Heiko Carstens --- arch/s390/pci/pci_event.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/arch/s390/pci/pci_event.c b/arch/s390/pci/pci_event.c index ec93f34b6e19..7b538c6ae11b 100644 --- a/arch/s390/pci/pci_event.c +++ b/arch/s390/pci/pci_event.c @@ -361,8 +361,10 @@ static void __zpci_event_error(struct zpci_ccdf_err *ccdf) __zpci_event_print_error(pdev, ccdf); - if (!pdev) + if (!pdev) { + zpci_report_status(zdev, NULL, "error event", "no pdev bound"); goto no_pdev; + } switch (ccdf->pec) { case 0x002a: /* Error event concerns FMB */ From 3a43be7a1fd06a35cf9e621b88283b3b6e7d281c Mon Sep 17 00:00:00 2001 From: Niklas Schnelle Date: Wed, 16 Sep 2026 17:14:13 +0200 Subject: [PATCH 208/365] s390/pci: Don't report recovery success on skipped recovery When a PCI device is already in the permanent failure state, recovery is skipped, but the SCLP recovery report still shows success. Fix this by changing the status string to explicitly state that recovery was skipped due to permanent failure. Cc: stable@vger.kernel.org Fixes: 4ec6054e7321 ("s390/pci: Report PCI error recovery results via SCLP") Signed-off-by: Niklas Schnelle Reviewed-by: Benjamin Block Reviewed-by: Farhan Ali Signed-off-by: Heiko Carstens --- arch/s390/pci/pci_event.c | 1 + 1 file changed, 1 insertion(+) diff --git a/arch/s390/pci/pci_event.c b/arch/s390/pci/pci_event.c index 7b538c6ae11b..d6af4015223e 100644 --- a/arch/s390/pci/pci_event.c +++ b/arch/s390/pci/pci_event.c @@ -226,6 +226,7 @@ static pci_ers_result_t zpci_event_attempt_error_recovery(struct pci_dev *pdev, device_lock(&pdev->dev); if (pdev->error_state == pci_channel_io_perm_failure) { ers_res = PCI_ERS_RESULT_DISCONNECT; + status_str = "skipped (permanent failure)"; goto out_unlock; } pdev->error_state = pci_channel_io_frozen; From f4d04425e66af2ecee9d1a49ae0484436f3c2fd1 Mon Sep 17 00:00:00 2001 From: Peter Oberparleiter Date: Mon, 21 Sep 2026 15:16:48 +0200 Subject: [PATCH 209/365] s390/cmf: Fix virtual vs physical address confusion The measurement block address is an absolute address. Define the associated schib_config and schib fields as dma64_t to enable automatic detection of incorrect assignments. Also add the missing virt_to_dma64() translation. Without this fix, a wrong address will be used by firmware when storing extended format channel measurement data on kernels built with CONFIG_RANDOMIZE_IDENTITY_BASE=y. Fixes: 14edd0d73bfe ("s390/cmf: fix virtual vs physical address confusion") Cc: stable@vger.kernel.org Signed-off-by: Peter Oberparleiter Reviewed-by: Heiko Carstens Signed-off-by: Heiko Carstens --- drivers/s390/cio/cio.h | 5 +++-- drivers/s390/cio/cmf.c | 2 +- 2 files changed, 4 insertions(+), 3 deletions(-) diff --git a/drivers/s390/cio/cio.h b/drivers/s390/cio/cio.h index bad142c536e1..6d28a62bc67a 100644 --- a/drivers/s390/cio/cio.h +++ b/drivers/s390/cio/cio.h @@ -7,6 +7,7 @@ #include #include #include +#include #include #include #include @@ -49,7 +50,7 @@ struct pmcw { /* Target SCHIB configuration. */ struct schib_config { - u64 mba; + dma64_t mba; u32 intparm; u16 mbi; u32 isc:3; @@ -66,7 +67,7 @@ struct schib_config { struct schib { struct pmcw pmcw; /* path management control word */ union scsw scsw; /* subchannel status word */ - __u64 mba; /* measurement block address */ + dma64_t mba; /* measurement block address */ __u8 mda[4]; /* model dependent area */ } __attribute__ ((packed,aligned(4))); diff --git a/drivers/s390/cio/cmf.c b/drivers/s390/cio/cmf.c index 92ab3d546fe4..66b14fedbd18 100644 --- a/drivers/s390/cio/cmf.c +++ b/drivers/s390/cio/cmf.c @@ -183,7 +183,7 @@ static int set_schib(struct ccw_device *cdev, u32 mme, int mbfc, sch->config.mbfc = mbfc; /* address can be either a block address or a block index */ if (mbfc) - sch->config.mba = address; + sch->config.mba = address ? virt_to_dma64((void *)address) : 0; else sch->config.mbi = address; From 4467df89dbca6a3e9dbc343a315324bb192603d6 Mon Sep 17 00:00:00 2001 From: Mikhail Zaslonko Date: Wed, 16 Sep 2026 18:53:30 +0200 Subject: [PATCH 210/365] s390/debug: Reject NULL debug info in debug_dump() debug_dump() passes id on to debug_info_copy(), which dereferences in->name unchecked. debug_unregister(), debug_set_size() and debug_register_view() guard against a NULL id but debug_dump() does not. Nothing reaches this today, but add the check for consistency. Reported-by: Sashiko Closes: https://lore.kernel.org/all/20260910110147.96E851F000FF@smtp.kernel.org/ Signed-off-by: Mikhail Zaslonko Reviewed-by: Heiko Carstens Signed-off-by: Heiko Carstens --- arch/s390/kernel/debug.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/arch/s390/kernel/debug.c b/arch/s390/kernel/debug.c index b5bf8284dbfc..354ce78fc0cf 100644 --- a/arch/s390/kernel/debug.c +++ b/arch/s390/kernel/debug.c @@ -823,6 +823,9 @@ ssize_t debug_dump(debug_info_t *id, struct debug_view *view, file_private_info_t *p_info; size_t size, offset = 0; + if (!id) + return -EINVAL; + /* Need space for '\0' byte */ if (buf_size < 1) return 0; From 28e29992b034acffc9342df216c06097825ce610 Mon Sep 17 00:00:00 2001 From: Mikhail Zaslonko Date: Fri, 18 Sep 2026 17:13:51 +0200 Subject: [PATCH 211/365] s390/debug: Do not register views for failed static debug areas __REGISTER_STATIC_DEBUG_INFO() calls debug_register_view() unconditionally, even when debug_register_static() has failed. In that case _debug_register() was never reached and id->debugfs_root_entry is still NULL, so debugfs_create_file() places the view file in the debugfs root directory. For sclp_err this leaves a /sys/kernel/debug/hex_ascii file with nothing to indicate which debug log it belongs to. debug_register_static() is not exported and the macro is its only caller, so let it return an error code and skip the view registration when it fails. No debugfs files are created for such an area then. Reproduce by booting with s390dbf=sclp_err::100000000. The sclp_err registration fails, no s390dbf/sclp_err/ directory is created, and a hex_ascii file appears in the debugfs root instead. Fixes: d72541f94512 ("s390/debug: add early tracing support") Signed-off-by: Mikhail Zaslonko Reviewed-by: Heiko Carstens Signed-off-by: Heiko Carstens --- arch/s390/include/asm/debug.h | 8 ++++++-- arch/s390/kernel/debug.c | 12 +++++++++--- 2 files changed, 15 insertions(+), 5 deletions(-) diff --git a/arch/s390/include/asm/debug.h b/arch/s390/include/asm/debug.h index 39d484c59774..ad438d6352c8 100644 --- a/arch/s390/include/asm/debug.h +++ b/arch/s390/include/asm/debug.h @@ -460,7 +460,11 @@ static int VNAME(var, active_entries)[EARLY_AREAS] __initdata #define __REGISTER_STATIC_DEBUG_INFO(var, name, pages, areas, view) \ static int __init VNAME(var, reg)(void) \ { \ - debug_register_static(&var, (pages), (areas)); \ + int rc; \ + \ + rc = debug_register_static(&var, (pages), (areas)); \ + if (rc) \ + return rc; \ debug_register_view(&var, (view)); \ return 0; \ } \ @@ -493,7 +497,7 @@ static debug_info_t __refdata var = \ static debug_info_t __used __section(".s390dbf_info") *VNAME(var, info) = &var; \ __REGISTER_STATIC_DEBUG_INFO(var, name, pages, nr_areas, view) -void debug_register_static(debug_info_t *id, int pages_per_area, int nr_areas); +int debug_register_static(debug_info_t *id, int pages_per_area, int nr_areas); #endif /* MODULE */ diff --git a/arch/s390/kernel/debug.c b/arch/s390/kernel/debug.c index 354ce78fc0cf..142dc1487187 100644 --- a/arch/s390/kernel/debug.c +++ b/arch/s390/kernel/debug.c @@ -953,8 +953,12 @@ EXPORT_SYMBOL(debug_register); * * Note: This function is called automatically via an initcall generated by * DEFINE_STATIC_DEBUG_INFO. + * + * Return: + * - 0 on success + * - negative error code on failure */ -void debug_register_static(debug_info_t *id, int pages_per_area, int nr_areas) +int debug_register_static(debug_info_t *id, int pages_per_area, int nr_areas) { unsigned long flags; debug_info_t *copy; @@ -962,7 +966,7 @@ void debug_register_static(debug_info_t *id, int pages_per_area, int nr_areas) if (!initialized) { pr_err("Tried to register debug feature %s too early\n", id->name); - return; + return -EINVAL; } debug_get_param(id->name, &id->level, &pages_per_area, false); @@ -978,7 +982,7 @@ void debug_register_static(debug_info_t *id, int pages_per_area, int nr_areas) id->active_entries = NULL; raw_spin_unlock_irqrestore(&id->lock, flags); - return; + return -ENOMEM; } /* Replace static trace area with dynamic copy. */ @@ -996,6 +1000,8 @@ void debug_register_static(debug_info_t *id, int pages_per_area, int nr_areas) mutex_lock(&debug_mutex); _debug_register(id); mutex_unlock(&debug_mutex); + + return 0; } /* Remove debugfs entries. */ From 012bfcd5a51082d5a65f096dfb9ca652b5267965 Mon Sep 17 00:00:00 2001 From: Mikhail Zaslonko Date: Wed, 16 Sep 2026 18:06:26 +0200 Subject: [PATCH 212/365] s390/debug: Fix NULL pointer dereference in debug_info_copy() When debug_register_static() fails, it clears areas, active_pages and active_entries but leaves the area bounds unchanged. Copying such an area, either by opening its view file or via debug_dump(), makes debug_info_copy() dereference the NULL pointers. Skip the copy loop when the source has no areas. Closes: https://lore.kernel.org/r/20260903132123.12F271F00A3F@smtp.kernel.org Fixes: d72541f94512 ("s390/debug: add early tracing support") Signed-off-by: Mikhail Zaslonko Reviewed-by: Peter Oberparleiter Acked-by: Heiko Carstens Signed-off-by: Heiko Carstens --- arch/s390/kernel/debug.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/arch/s390/kernel/debug.c b/arch/s390/kernel/debug.c index 142dc1487187..3b68f5c58333 100644 --- a/arch/s390/kernel/debug.c +++ b/arch/s390/kernel/debug.c @@ -434,7 +434,8 @@ static debug_info_t *debug_info_copy(debug_info_t *in, int mode) debug_info_free(rc); } while (1); - if (mode == NO_AREAS) + /* debug_register_static() failure leaves areas NULL, bounds intact */ + if (mode == NO_AREAS || !in->areas) goto out; for (i = 0; i < in->nr_areas; i++) { From 10180a277549339020b08000206092c07e0bff5a Mon Sep 17 00:00:00 2001 From: Sean Christopherson Date: Mon, 21 Sep 2026 14:16:07 -0700 Subject: [PATCH 213/365] KVM: x86: Re-pend GET_NESTED_STATE_PAGES if getting said pages fails Re-pend GET_NESTED_STATE_PAGES before exiting to userspace if getting the nested pages fails in the KVM_RUN path. If userspace re-runs the vCPU, and vmcs02 holds valid PFNs from the *previous* run of L2, then KVM could re-enter L2 with stale, unpinned PFNs mapped into e.g. the vAPIC page. Note, both SVM and VMX (as of commit 11722439fb20 ("KVM: nVMX: Ensure KVM_REQ_GET_NESTED_STATE_PAGES is cleared on VM-Exit") ensure the request is cleared on VM-Exit (including the "forced" case), i.e. there is no risk of double-mapping due to emulated VMLAUNCH/VMRESUME/VMRUN *and* the request trying to map the nested pages. Fixes: 671ddc700fd0 ("KVM: nVMX: Don't leak L1 MMIO regions to L2") Cc: stable@vger.kernel.org Reported-by: Jinwoo Lee Closes: https://lore.kernel.org/all/20260813043932.3214460-1-rkskek9254@gmail.com Reported-by: Stefan Teodorescu Link: https://patch.msgid.link/20260921211608.1030158-2-seanjc@google.com Signed-off-by: Sean Christopherson --- arch/x86/kvm/x86.c | 1 + 1 file changed, 1 insertion(+) diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c index a137dc6dd8c6..4ec17aaff413 100644 --- a/arch/x86/kvm/x86.c +++ b/arch/x86/kvm/x86.c @@ -8072,6 +8072,7 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu) if (kvm_check_request(KVM_REQ_GET_NESTED_STATE_PAGES, vcpu)) { if (unlikely(!kvm_nested_call(get_nested_state_pages)(vcpu))) { + kvm_make_request(KVM_REQ_GET_NESTED_STATE_PAGES, vcpu); r = 0; goto out; } From c1214f293d77c6e425a379f90d09bdb4815993a8 Mon Sep 17 00:00:00 2001 From: Sean Christopherson Date: Mon, 21 Sep 2026 14:16:08 -0700 Subject: [PATCH 214/365] KVM: x86: Fill kvm_run exit fields in common get_nested_state_pages() error paths Fill kvm_run with "internal error, emulation" in the common error handling paths for getting nested state pages, as requiring each check to manually fill kvm_run is error prone and requires a non-trivial amount of copy+paste. Specifically, both SVM and VMX fail to fill kvm_run if load_pdptrs() fails, and SVM fails to fill kvm_run if kvm_hv_verify_vp_assist() fails. If those flows fail, the *best* case scenario is that KVM will exit to userspace with KVM_EXIT_UNKNOWN. The worst case scenario is that KVM exits with a stale exit_reason and confuses userspace. Note, SVM never exits to userspace if something goes sideways when dealing with vmcb12 assets while emulating VMRUN, i.e. lack of SVM-specific code is not a bug. Fixes: 0f85722341b0 ("KVM: nVMX: delay loading of PDPTRs to KVM_REQ_GET_NESTED_STATE_PAGES") Fixes: 232f75d3b4b5 ("KVM: nSVM: call nested_svm_load_cr3 on nested state load") Fixes: 3f4a812edf5c ("KVM: nSVM: hyper-v: Enable L2 TLB flush") Cc: stable@vger.kernel.org Reported-by: Jinwoo Lee Closes: https://lore.kernel.org/all/20260813043932.3214460-1-rkskek9254@gmail.com Reported-by: Stefan Teodorescu Link: https://patch.msgid.link/20260921211608.1030158-3-seanjc@google.com Signed-off-by: Sean Christopherson --- arch/x86/kvm/svm/nested.c | 7 +------ arch/x86/kvm/vmx/nested.c | 15 +++++---------- arch/x86/kvm/x86.c | 3 +++ 3 files changed, 9 insertions(+), 16 deletions(-) diff --git a/arch/x86/kvm/svm/nested.c b/arch/x86/kvm/svm/nested.c index 73f37b050d0a..f9090b601efa 100644 --- a/arch/x86/kvm/svm/nested.c +++ b/arch/x86/kvm/svm/nested.c @@ -2125,13 +2125,8 @@ static bool svm_get_nested_state_pages(struct kvm_vcpu *vcpu) return false; } - if (!nested_svm_merge_msrpm(vcpu)) { - vcpu->run->exit_reason = KVM_EXIT_INTERNAL_ERROR; - vcpu->run->internal.suberror = - KVM_INTERNAL_ERROR_EMULATION; - vcpu->run->internal.ndata = 0; + if (!nested_svm_merge_msrpm(vcpu)) return false; - } if (kvm_hv_verify_vp_assist(vcpu)) return false; diff --git a/arch/x86/kvm/vmx/nested.c b/arch/x86/kvm/vmx/nested.c index 151873407abd..40c1a5f6fa8a 100644 --- a/arch/x86/kvm/vmx/nested.c +++ b/arch/x86/kvm/vmx/nested.c @@ -3465,10 +3465,6 @@ static bool nested_get_vmcs12_pages(struct kvm_vcpu *vcpu) } else { pr_debug_ratelimited("%s: no backing for APIC-access address in vmcs12\n", __func__); - vcpu->run->exit_reason = KVM_EXIT_INTERNAL_ERROR; - vcpu->run->internal.suberror = - KVM_INTERNAL_ERROR_EMULATION; - vcpu->run->internal.ndata = 0; return false; } } @@ -3539,11 +3535,6 @@ static bool vmx_get_nested_state_pages(struct kvm_vcpu *vcpu) if (!nested_get_evmcs_page(vcpu)) { pr_debug_ratelimited("%s: enlightened vmptrld failed\n", __func__); - vcpu->run->exit_reason = KVM_EXIT_INTERNAL_ERROR; - vcpu->run->internal.suberror = - KVM_INTERNAL_ERROR_EMULATION; - vcpu->run->internal.ndata = 0; - return false; } #endif @@ -3915,8 +3906,12 @@ static int nested_vmx_run(struct kvm_vcpu *vcpu, bool launch) vmentry_failed: vcpu->arch.nested_run_pending = 0; - if (status == NVMX_VMENTRY_KVM_INTERNAL_ERROR) + if (status == NVMX_VMENTRY_KVM_INTERNAL_ERROR) { + vcpu->run->exit_reason = KVM_EXIT_INTERNAL_ERROR; + vcpu->run->internal.suberror = KVM_INTERNAL_ERROR_EMULATION; + vcpu->run->internal.ndata = 0; return 0; + } if (status == NVMX_VMENTRY_VMEXIT) return 1; WARN_ON_ONCE(status != NVMX_VMENTRY_VMFAIL); diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c index 4ec17aaff413..aad065d035fb 100644 --- a/arch/x86/kvm/x86.c +++ b/arch/x86/kvm/x86.c @@ -8072,6 +8072,9 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu) if (kvm_check_request(KVM_REQ_GET_NESTED_STATE_PAGES, vcpu)) { if (unlikely(!kvm_nested_call(get_nested_state_pages)(vcpu))) { + vcpu->run->exit_reason = KVM_EXIT_INTERNAL_ERROR; + vcpu->run->internal.suberror = KVM_INTERNAL_ERROR_EMULATION; + vcpu->run->internal.ndata = 0; kvm_make_request(KVM_REQ_GET_NESTED_STATE_PAGES, vcpu); r = 0; goto out; From 277d3623d99a4fc2623bfb7d191b649ca380605d Mon Sep 17 00:00:00 2001 From: Zeng Chi Date: Mon, 21 Sep 2026 18:24:42 +0800 Subject: [PATCH 215/365] KVM: Don't treat reserved xarray entries as having memory attributes kvm_vm_set_mem_attributes() reserves an xarray entry for every gfn in the range before storing the new attributes, so that the store loop can't fail partway through. If one of the reservations fails, e.g. with -ENOMEM, the entries that were already reserved are left in the array. That is harmless as far as xa_reserve() is concerned, as the reserved entries read back as NULL via xa_load(), but it confuses the "does this range have no attributes at all" check: if (!attrs) return !xas_find(&xas, end - 1); A reserved entry is XA_ZERO_ENTRY, not NULL, and xas_find() returns it as present. So a leftover reservation makes KVM report that a fully shared range has attributes even though kvm_get_memory_attributes() returns none for every gfn in the range. On x86, the next time mixed-attribute tracking is recomputed for the range (memslot creation, or a later attribute change that straddles the 2MiB page), hugepage_has_attrs() treats a fully shared 2MiB range as mixed and refuses to map it with a hugepage, until userspace happens to set attributes on the range again. Drop the shortcut and handle the !attrs case in the per-index loop, using xas_next_entry() to find the next non-NULL entry. xas_next_entry() is essentially an optimized xas_find(), so the effective change is that the !attrs lookup now goes through xas_retry() like the attrs != 0 case, i.e. reserved entries are skipped and retry entries restart the walk. Don't check the index when no entry is found, as the xarray leaves the xas index in a bogus state in that case; no entry simply means the rest of the range has no attributes. KVM never stores a non-NULL entry with a value of zero (clearing stores NULL), but such an entry would be returned by xas_next_entry() and trip the index check, so WARN if one is ever seen. Fixes: 5a475554db1e ("KVM: Introduce per-page memory attributes") Cc: stable@vger.kernel.org Suggested-by: Sean Christopherson Cc: David Ballesteros Signed-off-by: Zeng Chi Link: https://patch.msgid.link/20260921102442.1232375-1-zeng_chi911@163.com [sean: expand comment to elaborate on xarray APIs, split optimization out] Signed-off-by: Sean Christopherson --- virt/kvm/kvm_main.c | 28 +++++++++++++++++++++++++--- 1 file changed, 25 insertions(+), 3 deletions(-) diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c index 65eb26a0520d..2a046e95f95e 100644 --- a/virt/kvm/kvm_main.c +++ b/virt/kvm/kvm_main.c @@ -2447,14 +2447,36 @@ bool kvm_range_has_memory_attributes(struct kvm *kvm, gfn_t start, gfn_t end, return (kvm_get_memory_attributes(kvm, start) & mask) == attrs; guard(rcu)(); - if (!attrs) - return !xas_find(&xas, end - 1); + /* + * Lookup the entry for each index instead of iterating over the xarray + * as KVM deletes/nullifies entries to represent "no attributes", and + * the xas index is effectively invalid when no entry is found. I.e. + * matching non-zero attributes for *every* entry effectively requires + * a manually lookup for each index. + * + * Skip pre-allocated, reserved entries, or restart the lookup if the + * xarray was concurrently modified, via xas_retry() ("retry" means the + * entry holds an internal xarray value, i.e. is either invalid or NULL + * from the caller's perspective). + * + * Use xas_next() when looking for non-zero attributes to optimize for + * the case where the start of the range (or the entire range) doesn't + * have any attributes, as xas_next() returns literally the next entry, + * whereas xas_next_entry() returns the next non-NULL entry (bounded by + * a maximum index). + */ for (index = start; index < end; index++) { do { - entry = xas_next(&xas); + entry = attrs ? xas_next(&xas) : + xas_next_entry(&xas, end - 1); } while (xas_retry(&xas, entry)); + if (!entry) + return !attrs; + + WARN_ON_ONCE(!xa_to_value(entry)); + if (xas.xa_index != index || (xa_to_value(entry) & mask) != attrs) return false; From 382e5d514b6f35bdda2ab9044b4eed23d2ec4254 Mon Sep 17 00:00:00 2001 From: David Ballesteros Date: Tue, 15 Sep 2026 17:53:57 +0000 Subject: [PATCH 216/365] KVM: Ensure memory attributes xarray nodes are accounted to the caller's memcg Explicitly instantiate the memory attributes xarray with XA_FLAGS_ACCOUNT to ensure that all allocations are accounted to the memcg. Frustratingly, memory allocations done in the "fastpath" do not honor the passed in gfp, even for an explicit xa_reserve(). Only the rare, slow path __xas_nomem() honors the original gfp. E.g. xa_reserve(..., GFP_KERNEL_ACCOUNT) | -> ... | -> __xa_cmpxchg_raw() | -> xas_store() <== does not take @gfp | -> xas_create() | -> xas_alloc() The bug was confirmed by observing that a process in a cgroup limited to 256 MiB grew radix_tree_node slab by ~512 MiB while its memory.current stayed near 0. Fixes: 5a475554db1e ("KVM: Introduce per-page memory attributes") Cc: stable@vger.kernel.org Assisted-by: Claude-Code:claude-opus-5 Signed-off-by: David Ballesteros Link: https://patch.msgid.link/20260915175335.138547-4-davimaba.v@proton.me [sean: rewrite changelog, tag for stable] Signed-off-by: Sean Christopherson --- virt/kvm/kvm_main.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c index 2a046e95f95e..df643ec3ea75 100644 --- a/virt/kvm/kvm_main.c +++ b/virt/kvm/kvm_main.c @@ -1117,7 +1117,7 @@ static struct kvm *kvm_create_vm(unsigned long type, const char *fdname) rcuwait_init(&kvm->mn_memslots_update_rcuwait); xa_init(&kvm->vcpu_array); #ifdef CONFIG_KVM_GENERIC_MEMORY_ATTRIBUTES - xa_init(&kvm->mem_attr_array); + xa_init_flags(&kvm->mem_attr_array, XA_FLAGS_ACCOUNT); #endif INIT_LIST_HEAD(&kvm->gpc_list); From 119e9db233ad0f4cbd4cfb9f9385a7bca378b37d Mon Sep 17 00:00:00 2001 From: Zeng Chi Date: Mon, 21 Sep 2026 18:24:42 +0800 Subject: [PATCH 217/365] KVM: Don't pre-reserve xarray entries when storing empty/NULL attributes Skip the xarray reservation loop when clearing all memory attributes, as storing NULL only erases the entry and never needs to allocate, so no reservation (and no cleanup of a failed one) is required in that case. Suggested-by: Sean Christopherson Cc: David Ballesteros Signed-off-by: Zeng Chi Link: https://patch.msgid.link/20260921102442.1232375-1-zeng_chi911@163.com [sean: split to separate patch] Signed-off-by: Sean Christopherson --- virt/kvm/kvm_main.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c index df643ec3ea75..85f42289748d 100644 --- a/virt/kvm/kvm_main.c +++ b/virt/kvm/kvm_main.c @@ -2593,9 +2593,10 @@ static int kvm_vm_set_mem_attributes(struct kvm *kvm, gfn_t start, gfn_t end, /* * Reserve memory ahead of time to avoid having to deal with failures - * partway through setting the new attributes. + * partway through setting the new attributes. Storing NULL never + * allocates, so no reservations are needed when clearing. */ - for (i = start; i < end; i++) { + for (i = start; entry && i < end; i++) { r = xa_reserve(&kvm->mem_attr_array, i, GFP_KERNEL_ACCOUNT); if (r) goto out_unlock; From 141008dec73521ccf64878517460cec8b3297251 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Szymon=20Aceda=C5=84ski?= Date: Wed, 16 Sep 2026 19:30:30 +0200 Subject: [PATCH 218/365] drm/xe: Limit sg segment size to PAGE_SIZE on Xen PV MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Fix display corruption on Xen PV dom0, where DMA buffers are not guaranteed machine-contiguous, in which case bounce buffering kicks in, breaking xe's memory coherency assumptions. Apply the same workaround i915 carries in i915_sg_segment_size() since commit 78a07fe777c4 ("drm/i915: stop abusing swiotlb_max_segment"). Fixes: dd08ebf6c352 ("drm/xe: Introduce a new DRM driver for Intel GPUs") Reported-by: Marek Marczykowski-Górecki Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/work_items/8382 Link: https://lore.kernel.org/xen-devel/aYtznP_tT6xNPwf-@mail-itl/ Link: https://lore.kernel.org/all/20221020110308.1582518-1-hch@lst.de/ # i915 counterpart Cc: Christoph Hellwig Cc: Robert Beckett Cc: stable@vger.kernel.org # v6.8+ Signed-off-by: Szymon Acedański Reviewed-by: Thomas Hellström Signed-off-by: Thomas Hellström Link: https://patch.msgid.link/20260916173030.3223833-1-accek@invisiblethingslab.com (cherry picked from commit 77f704158f099b952681f207478a22d5b8218edb) Signed-off-by: Rodrigo Vivi --- drivers/gpu/drm/xe/xe_bo.h | 19 +++++++++++++++++++ 1 file changed, 19 insertions(+) diff --git a/drivers/gpu/drm/xe/xe_bo.h b/drivers/gpu/drm/xe/xe_bo.h index e8081af5bfc1..152bfcffe399 100644 --- a/drivers/gpu/drm/xe/xe_bo.h +++ b/drivers/gpu/drm/xe/xe_bo.h @@ -9,6 +9,8 @@ #include #include +#include + #include "xe_bo_types.h" #include "xe_ggtt.h" #include "xe_macros.h" @@ -575,6 +577,23 @@ static inline unsigned int xe_sg_segment_size(struct device *dev) struct scatterlist __maybe_unused sg; size_t max = BIT_ULL(sizeof(sg.length) * 8) - 1; + /* + * For Xen PV guests pages aren't contiguous in DMA (machine) address + * space. The DMA API takes care of that both in dma_alloc_* (by + * calling into the hypervisor to make the pages contiguous) and in + * dma_map_* (by bounce buffering). But xe (like i915, see commit + * 78a07fe777c4) ignores the coherency aspects of the DMA API and thus + * can't cope with bounce buffering actually happening, so add a hack + * here to force small allocations and mappings when running in PV + * mode on Xen. + * + * Note this will still break if bounce buffering is required for other + * reasons, like confidential computing hypervisors or PCIe root ports + * with addressing limitations. + */ + if (xen_pv_domain()) + return PAGE_SIZE; + max = min_t(size_t, max, dma_max_mapping_size(dev)); /* From 90f467577ebc28c5bc4ba2f0f1b83a4419fb0740 Mon Sep 17 00:00:00 2001 From: Tangudu Tilak Tirumalesh Date: Wed, 16 Sep 2026 15:35:44 +0530 Subject: [PATCH 219/365] drm/xe: harden adjust_idledly() against divide-by-zero and overflow adjust_idledly() has several corner-case issues flagged during review: 1. If xe_gt_clock_init() failed to recognise the crystal clock, gt->info.timestamp_base is 0, which makes idledly_units_ps also 0. The subsequent DIV_ROUND_CLOSEST(..., idledly_units_ps) is then a divide-by-zero and panics the kernel. 2. The tick-to-ns conversions are done in u32: idledly * idledly_units_ps, (maxcnt - 1) * 1000 Both overflow u32 before DIV_ROUND_CLOSEST() sees them. 3. If IDLE_WAIT_TIME reads back as 0, maxcnt evaluates to 0 and the maxcnt - 1 clamp wraps to 0xFFFFFFFF in u32. 4. The register only stores whole ticks, so the clamped ns value has to be converted to ticks and back. DIV_ROUND_CLOSEST() can round that conversion up past maxcnt: maxcnt = 640 ns, one tick = 666664 ps clamp: maxcnt - 1 = 639 ns ns -> ticks: 639000 / 666664 = 0.958 -> rounds to 1 tick tick -> ns: 1 * 666664 / 1000 = 667 ns 667 ns is programmed into RING_IDLEDLY, but 667 >= maxcnt (640), so xe_gt_WARN_ON() fires again on every subsequent init. Return early if timestamp_base is 0 (the unknown-crystal path). Do the conversions in u64 via the *_ULL() helpers so they cannot wrap. Clamp with a floor (DIV_ROUND_DOWN_ULL) so the programmed delay stays strictly below maxcnt, and guard the maxcnt == 0 case with a zero delay while still writing RING_IDLEDLY so INHIBIT_SWITCH_UNTIL_PREEMPTED is cleared. v2: Drop the redundant warn on the timestamp_base == 0 path; xe_gt_clock_init() already warns on an unrecognised crystal clock. Keep the early return to avoid the divide-by-zero. - Vinay v3: Field-mask the RING_IDLEDLY write with REG_FIELD_PREP(IDLE_DELAY, ...) instead of writing the raw tick count, which could clobber INHIBIT_SWITCH_UNTIL_PREEMPTED and reserved bits. Split the inhibit-switch clear from the maxcnt clamp so a set inhibit bit no longer forces a needless delay overwrite when the delay itself is already valid. Use gt_to_xe(gt) instead of gt_to_xe(hwe->gt). Fixes: d2de4410a88f ("drm/xe: Apply Wa_16023105232") Cc: stable@vger.kernel.org Assisted-by: GitHub_Copilot:claude-opus-4.8 Signed-off-by: Tangudu Tilak Tirumalesh Reviewed-by: Vinay Belgaumkar Link: https://patch.msgid.link/20260916100545.779894-2-tilak.tirumalesh.tangudu@intel.com Signed-off-by: Matt Roper (cherry picked from commit d864065ea25e9d12897c175de9176ce46677e176) Signed-off-by: Rodrigo Vivi --- drivers/gpu/drm/xe/xe_hw_engine.c | 27 +++++++++++++++++++++------ 1 file changed, 21 insertions(+), 6 deletions(-) diff --git a/drivers/gpu/drm/xe/xe_hw_engine.c b/drivers/gpu/drm/xe/xe_hw_engine.c index 010499766fce..d868af8e3301 100644 --- a/drivers/gpu/drm/xe/xe_hw_engine.c +++ b/drivers/gpu/drm/xe/xe_hw_engine.c @@ -592,22 +592,37 @@ static void adjust_idledly(struct xe_hw_engine *hwe) u32 idledly_units_ps = 8 * gt->info.timestamp_base; u32 maxcnt_units_ns = 640; bool inhibit_switch = 0; + bool wa_applied = false; + + if (!IS_SRIOV_VF(gt_to_xe(gt)) && XE_GT_WA(gt, 16023105232)) { + /* xe_gt_clock_init() warns and zeroes timestamp_base on unknown crystal clock. */ + if (!idledly_units_ps) + return; - if (!IS_SRIOV_VF(gt_to_xe(hwe->gt)) && XE_GT_WA(gt, 16023105232)) { idledly = xe_mmio_read32(>->mmio, RING_IDLEDLY(hwe->mmio_base)); maxcnt = xe_mmio_read32(>->mmio, RING_PWRCTX_MAXCNT(hwe->mmio_base)); inhibit_switch = idledly & INHIBIT_SWITCH_UNTIL_PREEMPTED; idledly = REG_FIELD_GET(IDLE_DELAY, idledly); - idledly = DIV_ROUND_CLOSEST(idledly * idledly_units_ps, 1000); + idledly = DIV_ROUND_CLOSEST_ULL((u64)idledly * idledly_units_ps, 1000); maxcnt = REG_FIELD_GET(IDLE_WAIT_TIME, maxcnt); maxcnt *= maxcnt_units_ns; - if (xe_gt_WARN_ON(gt, idledly >= maxcnt || inhibit_switch)) { - idledly = DIV_ROUND_CLOSEST(((maxcnt - 1) * 1000), - idledly_units_ps); - xe_mmio_write32(>->mmio, RING_IDLEDLY(hwe->mmio_base), idledly); + /* Clear the inhibit switch without disturbing a valid delay. */ + if (inhibit_switch) + wa_applied = true; + + if (xe_gt_WARN_ON(gt, idledly >= maxcnt)) { + /* Floor below maxcnt; write 0 to still clear the inhibit bit. */ + idledly = maxcnt ? + DIV_ROUND_DOWN_ULL((u64)(maxcnt - 1) * 1000, + idledly_units_ps) : 0; + wa_applied = true; } + + if (wa_applied) + xe_mmio_write32(>->mmio, RING_IDLEDLY(hwe->mmio_base), + REG_FIELD_PREP(IDLE_DELAY, idledly)); } } From 41112a787f9182c7f2d27122817861e3f22ef928 Mon Sep 17 00:00:00 2001 From: Florian Schmaus Date: Sun, 20 Sep 2026 16:35:25 +0200 Subject: [PATCH 220/365] PM: hibernate: Freeze kernel threads after image preallocation Commit 783c81098445 ("PM: hibernate: call preallocate_image() after freeze prepare") moved hibernate_preallocate_memory() after dpm_prepare() so that device drivers have the opportunity to release pinned/unswappable memory during their ->prepare() callback before memory is preallocated for the snapshot image. However, that commit also placed hibernate_preallocate_memory() after freeze_kernel_threads(). While it was assumed during review that swap I/O submitted via submit_bio() is synchronous and would not depend on frozen kernel threads, this does not hold in practice. Calling hibernate_preallocate_memory() with kernel threads frozen leads to intermittent deadlocks during hibernation. Inside hibernate_preallocate_memory(), shrink_all_memory() is invoked with .may_writepage = 1 and .may_swap = 1 to aggressively reclaim and swap out pages. Any writeback or swap I/O that relies on freezable kernel threads, block device helpers, or WQ_FREEZABLE workqueues (such as those in storage drivers, device mapper, or filesystems) deadlocks waiting on tasks that are stuck in the refrigerator. Fix this by reordering hibernation_snapshot(): 1. Call dpm_prepare(PMSG_FREEZE) first, allowing device drivers to release pinned resources while kernel threads are still active. 2. Call hibernate_preallocate_memory() second, performing page reclaim and swapout while storage layers, workqueues, and kernel threads are alive. 3. Call freeze_kernel_threads() third, only after all memory preallocation and swap I/O have completed. Additionally, restore the call to swsusp_free() in the cleanup path so that preallocated image memory is properly freed if freeze_kernel_threads() fails or if TEST_FREEZER is enabled. Fixes: 783c81098445 ("PM: hibernate: call preallocate_image() after freeze prepare") Signed-off-by: Florian Schmaus Reviewed-by: Mario Limonciello (AMD) Tested-by: Matthew Leach Reviewed-by: Matthew Leach Link: https://patch.msgid.link/20260920-fix-hibernation-v1-1-f9940c2d7d7f@geekplace.eu Signed-off-by: Rafael J. Wysocki --- kernel/power/hibernate.c | 26 ++++++++++++++------------ 1 file changed, 14 insertions(+), 12 deletions(-) diff --git a/kernel/power/hibernate.c b/kernel/power/hibernate.c index d2479c69d71a..c13f68ab7f6e 100644 --- a/kernel/power/hibernate.c +++ b/kernel/power/hibernate.c @@ -408,9 +408,18 @@ int hibernation_snapshot(int platform_mode) if (error) goto Close; + error = dpm_prepare(PMSG_FREEZE); + if (error) + goto Complete; + + /* Preallocate image memory before freezing kernel threads and shutting down devices. */ + error = hibernate_preallocate_memory(); + if (error) + goto Complete; + error = freeze_kernel_threads(); if (error) - goto Close; + goto Cleanup; if (hibernation_test(TEST_FREEZER)) { @@ -422,15 +431,6 @@ int hibernation_snapshot(int platform_mode) goto Thaw; } - error = dpm_prepare(PMSG_FREEZE); - if (error) - goto Complete; - - /* Preallocate image memory before shutting down devices. */ - error = hibernate_preallocate_memory(); - if (error) - goto Complete; - console_suspend_all(); pm_restrict_gfp_mask(); @@ -464,10 +464,12 @@ int hibernation_snapshot(int platform_mode) platform_end(platform_mode); return error; - Complete: - dpm_complete(PMSG_RECOVER); Thaw: thaw_kernel_threads(); + Cleanup: + swsusp_free(); + Complete: + dpm_complete(PMSG_RECOVER); goto Close; } From ec0d89150a9381d591344a9f6f5428655c227a7f Mon Sep 17 00:00:00 2001 From: Manaf Meethalavalappu Pallikunhi Date: Tue, 22 Sep 2026 17:49:02 +0530 Subject: [PATCH 221/365] thermal: gov_step_wise: Fix stale mitigation vote with non-zero lower bounds MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit When two or more thermal zones bind to a common cooling device and one zone uses a non-zero instance->lower value, there is a bug where the instance holds a stale mitigation vote even after its trip is cleared. Problem scenario: - thermal-zone1: Trip at 50°C, cooling-map with lower=0 - thermal-zone2: Trip at 55°C, cooling-map with lower=2 - Both zones share the same cooling device (e.g., CPU) Issue flow: 1. Both trips trigger, zone1 requests state 5, zone2 also mitigates 2. Zone2 trip clears (temp < 53°C due to hysteresis) 3. When throttle=false and trend=THERMAL_TREND_DROPPING: - Current code checks: if (cur_state <= instance->lower) return THERMAL_NO_TARGET - Since cur_state (5) > instance->lower (2), it returns instance->lower (2) - This is the BUG where it returns instance->lower even though trip is cleared 4. Zone2's passive polling stops (tz->passive reaches 0) - no more updates for zone2 5. Zone2's stale vote of 2 persists indefinitely 6. Even when zone1 wants to reduce cooling to state, the cooling device cannot go below state 2 due to zone2's stale vote When a trip is cleared (throttle == false), always return THERMAL_NO_TARGET instead of instance->lower. Remove the unnecessary check comparing cur_state with instance->lower. Since passive polling is already deactivated when the trip is cleared, the instance should always be deactivated regardless of its current cooling state. This ensures that instances with non-zero lower bounds do not retain stale mitigation votes after their trips are cleared. Fixes: 042a3d80f118 ("thermal: core: Move passive polling management to the core") Signed-off-by: Manaf Meethalavalappu Pallikunhi Link: https://patch.msgid.link/20260922-step_wise_multi_zone_stale_vote_fix-v1-1-789f68dab229@oss.qualcomm.com Signed-off-by: Rafael J. Wysocki --- drivers/thermal/gov_step_wise.c | 10 ++++------ 1 file changed, 4 insertions(+), 6 deletions(-) diff --git a/drivers/thermal/gov_step_wise.c b/drivers/thermal/gov_step_wise.c index ea277c466d8d..4fa4377f0d42 100644 --- a/drivers/thermal/gov_step_wise.c +++ b/drivers/thermal/gov_step_wise.c @@ -65,14 +65,12 @@ static unsigned long get_target_state(struct thermal_instance *instance, min(instance->lower + 1, instance->upper), instance->upper); } else if (trend == THERMAL_TREND_DROPPING) { - if (cur_state <= instance->lower) - return THERMAL_NO_TARGET; - /* - * If 'throttle' is false, no mitigation is necessary, so - * request the lower state for this instance. + * If 'throttle' is false, no mitigation is necessary and + * passive polling is already deactivated, so clear this + * instance state by returning THERMAL_NO_TARGET. */ - return instance->lower; + return THERMAL_NO_TARGET; } return instance->target; From cc319238e3f6668f867beb381ce93727c69b7317 Mon Sep 17 00:00:00 2001 From: Tangudu Tilak Tirumalesh Date: Wed, 16 Sep 2026 15:35:45 +0530 Subject: [PATCH 222/365] drm/xe: Add wa_14025941587 to xe2, xe3 and xe3p platforms Avoid programming the IDLEDLY timer to less than 5 microseconds. Apply wa_14025941587 to Graphics Versions 20.01 to 35.11 and Media Versions 13.01 to 35.03 v2: Use xe_rtp_match_not_sriov_vf, move to local variable Remove warn and other knits - Matt R v3: Add verbose comment - Tejas v4: Restore IDLE_DLY register on engine reset. Add it to GUC save-restore list. -Vivek v5: Extend WA to Media Versions 13.01 to 35.03 - Vinay v6: Avoid clearing inhibit switch - Bala Refactor code accordingly by adding idle_reg_val. v7: Rebased with the divide-by-zero/overflow guards living in a separate hardening patch. v8: Preserve the Wa_16023105232 floor (DIV_ROUND_DOWN_ULL) and the maxcnt == 0 guard from the hardening patch. Round up (DIV_ROUND_UP_ULL) the Wa_14025941587 minimum conversion instead, so the tick-quantized delay cannot round back below 5 us. v9: Evaluate the Wa_16023105232 xe_gt_WARN_ON() against the value read from hardware instead of the Wa_14025941587-bumped value, so it no longer fires on the driver's own floor. Re-check the rounded-up tick value against maxcnt and floor it if tick quantization pushed it back to/above maxcnt, logging via xe_gt_dbg since this is the driver's own value, not a hardware anomaly. Assisted-by: GitHub_Copilot:claude-opus-4.8 Signed-off-by: Tangudu Tilak Tirumalesh Reviewed-by: Vinay Belgaumkar Link: https://patch.msgid.link/20260916100545.779894-3-tilak.tirumalesh.tangudu@intel.com Signed-off-by: Matt Roper (cherry picked from commit 9453c528fc909076468ff10df1c2e334ca5a9b00) Signed-off-by: Rodrigo Vivi --- drivers/gpu/drm/xe/xe_guc_ads.c | 2 +- drivers/gpu/drm/xe/xe_hw_engine.c | 95 ++++++++++++++++++++++++------ drivers/gpu/drm/xe/xe_wa_oob.rules | 2 + 3 files changed, 80 insertions(+), 19 deletions(-) diff --git a/drivers/gpu/drm/xe/xe_guc_ads.c b/drivers/gpu/drm/xe/xe_guc_ads.c index ff8eee3831aa..58a3f6293ce4 100644 --- a/drivers/gpu/drm/xe/xe_guc_ads.c +++ b/drivers/gpu/drm/xe/xe_guc_ads.c @@ -864,7 +864,7 @@ static unsigned int guc_mmio_regset_write(struct xe_guc_ads *ads, } } - if (XE_GT_WA(hwe->gt, 16023105232)) + if (XE_GT_WA(hwe->gt, 16023105232) || XE_GT_WA(hwe->gt, 14025941587)) guc_mmio_regset_write_one(ads, regset_map, RING_IDLEDLY(hwe->mmio_base), count++); diff --git a/drivers/gpu/drm/xe/xe_hw_engine.c b/drivers/gpu/drm/xe/xe_hw_engine.c index d868af8e3301..9680d2a5adec 100644 --- a/drivers/gpu/drm/xe/xe_hw_engine.c +++ b/drivers/gpu/drm/xe/xe_hw_engine.c @@ -585,44 +585,103 @@ static void hw_engine_init_early(struct xe_gt *gt, struct xe_hw_engine *hwe, xe_reg_whitelist_process_engine(hwe); } +static u32 idledly_floor_ticks(u32 idledly_ns, u32 idledly_units_ps) +{ + return DIV_ROUND_DOWN_ULL((u64)idledly_ns * 1000, idledly_units_ps); +} + static void adjust_idledly(struct xe_hw_engine *hwe) { struct xe_gt *gt = hwe->gt; - u32 idledly, maxcnt; + u32 idledly, idledly_hw, idledly_reg_val, maxcnt; u32 idledly_units_ps = 8 * gt->info.timestamp_base; u32 maxcnt_units_ns = 640; - bool inhibit_switch = 0; + bool inhibit_switch = false; bool wa_applied = false; + bool clamped_below_maxcnt = false; + + if ((!IS_SRIOV_VF(gt_to_xe(gt)) && XE_GT_WA(gt, 16023105232)) || + XE_GT_WA(gt, 14025941587)) { + u32 mincnt_idledly_ns = 5000; - if (!IS_SRIOV_VF(gt_to_xe(gt)) && XE_GT_WA(gt, 16023105232)) { /* xe_gt_clock_init() warns and zeroes timestamp_base on unknown crystal clock. */ if (!idledly_units_ps) return; - idledly = xe_mmio_read32(>->mmio, RING_IDLEDLY(hwe->mmio_base)); + idledly_reg_val = xe_mmio_read32(>->mmio, RING_IDLEDLY(hwe->mmio_base)); maxcnt = xe_mmio_read32(>->mmio, RING_PWRCTX_MAXCNT(hwe->mmio_base)); - inhibit_switch = idledly & INHIBIT_SWITCH_UNTIL_PREEMPTED; - idledly = REG_FIELD_GET(IDLE_DELAY, idledly); + inhibit_switch = idledly_reg_val & INHIBIT_SWITCH_UNTIL_PREEMPTED; + idledly = REG_FIELD_GET(IDLE_DELAY, idledly_reg_val); idledly = DIV_ROUND_CLOSEST_ULL((u64)idledly * idledly_units_ps, 1000); + idledly_hw = idledly; maxcnt = REG_FIELD_GET(IDLE_WAIT_TIME, maxcnt); maxcnt *= maxcnt_units_ns; - /* Clear the inhibit switch without disturbing a valid delay. */ - if (inhibit_switch) - wa_applied = true; - - if (xe_gt_WARN_ON(gt, idledly >= maxcnt)) { - /* Floor below maxcnt; write 0 to still clear the inhibit bit. */ - idledly = maxcnt ? - DIV_ROUND_DOWN_ULL((u64)(maxcnt - 1) * 1000, - idledly_units_ps) : 0; + /* + * Wa_14025941587 is applied before Wa_16023105232, which takes + * priority if the two ever conflict (not expected in practice). + */ + if (XE_GT_WA(gt, 14025941587) && + idledly < mincnt_idledly_ns) { + idledly = mincnt_idledly_ns; wa_applied = true; } - if (wa_applied) - xe_mmio_write32(>->mmio, RING_IDLEDLY(hwe->mmio_base), - REG_FIELD_PREP(IDLE_DELAY, idledly)); + if (XE_GT_WA(gt, 16023105232)) { + /* Clear the inhibit switch without disturbing a valid delay. */ + if (inhibit_switch) { + idledly_reg_val &= ~INHIBIT_SWITCH_UNTIL_PREEMPTED; + wa_applied = true; + } + + /* Warn only on the value read from hardware. */ + xe_gt_WARN_ON(gt, idledly_hw >= maxcnt); + + if (idledly >= maxcnt) { + /* maxcnt may be 0 if IDLE_WAIT_TIME is unprogrammed. */ + idledly = maxcnt ? maxcnt - 1 : 0; + clamped_below_maxcnt = true; + wa_applied = true; + } + } + + if (wa_applied) { + u32 idledly_ticks; + + /* + * Wa_16023105232 requires idledly < maxcnt, so floor + * that clamp; otherwise round up to guarantee the + * Wa_14025941587 minimum survives tick quantization. + */ + if (clamped_below_maxcnt) + idledly_ticks = idledly_floor_ticks(idledly, idledly_units_ps); + else + idledly_ticks = DIV_ROUND_UP_ULL((u64)idledly * 1000, + idledly_units_ps); + + /* + * Tick quantization can still push the rounded-up value + * to/above maxcnt; re-floor here so Wa_16023105232 keeps + * priority even in that case. + */ + if (!clamped_below_maxcnt && XE_GT_WA(gt, 16023105232) && + (u64)idledly_ticks * idledly_units_ps >= (u64)maxcnt * 1000) { + xe_gt_dbg(gt, "idledly %s: %u ticks would exceed maxcnt=%u, so flooring\n", + hwe->name, idledly_ticks, maxcnt); + idledly = maxcnt ? maxcnt - 1 : 0; + idledly_ticks = idledly_floor_ticks(idledly, idledly_units_ps); + } + + idledly_reg_val &= ~IDLE_DELAY; + idledly_reg_val |= REG_FIELD_PREP(IDLE_DELAY, idledly_ticks); + xe_gt_dbg(gt, "idledly %s: set %u max=%u inh=%u ts=%u\n", + hwe->name, idledly, maxcnt, + !!inhibit_switch, gt->info.timestamp_base); + xe_mmio_write32(>->mmio, + RING_IDLEDLY(hwe->mmio_base), + idledly_reg_val); + } } } diff --git a/drivers/gpu/drm/xe/xe_wa_oob.rules b/drivers/gpu/drm/xe/xe_wa_oob.rules index dd69ad07f7a9..3001155f8d09 100644 --- a/drivers/gpu/drm/xe/xe_wa_oob.rules +++ b/drivers/gpu/drm/xe/xe_wa_oob.rules @@ -72,3 +72,5 @@ 16029897822 MEDIA_VERSION(3500) GRAPHICS_VERSION(3510) 14027054324 GRAPHICS_VERSION(3511) +14025941587 GRAPHICS_VERSION_RANGE(2001, 3511), FUNC(xe_rtp_match_not_sriov_vf) + MEDIA_VERSION_RANGE(1301, 3503), FUNC(xe_rtp_match_not_sriov_vf) From c6b51091cafff9ce6c03c1416aa13864d17ab97c Mon Sep 17 00:00:00 2001 From: Takashi Sakamoto Date: Tue, 22 Sep 2026 22:26:39 +0900 Subject: [PATCH 223/365] firewire: cdev: fix back-transition for iso_resource_auto client resource The todo member of iso_resource_auto structure represents the state of the client resource and normally transitions in the following order: ISO_RES_AUTO_ALLOC -> ISO_RES_AUTO_REALLOC -> ISO_RES_AUTO_DEALLOC However, concurrent access from the work item and the file descriptor release function can cause the state to transition backwards from ISO_RES_AUTO_DEALLOC to ISO_RES_AUTO_REALLOC. Prevent the back-transition by checking the current state before updating it in the work item. Fixes: fcabbf40fae5 ("firewire: core: move allocation/reallocation paths into specific branch after isoc resource management in cdev") Link: https://lore.kernel.org/r/20260922132639.191593-1-o-takashi@sakamocchi.jp Signed-off-by: Takashi Sakamoto --- drivers/firewire/core-cdev.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/drivers/firewire/core-cdev.c b/drivers/firewire/core-cdev.c index e49d8a58be09..664952a67a11 100644 --- a/drivers/firewire/core-cdev.c +++ b/drivers/firewire/core-cdev.c @@ -1397,8 +1397,10 @@ static void iso_resource_auto_work(struct work_struct *work) } else { // Transit from allocation to reallocation, except if the client requested // deallocation in the meantime. - scoped_guard(spinlock_irq, &client->lock) - r->todo = ISO_RES_AUTO_REALLOC; + scoped_guard(spinlock_irq, &client->lock) { + if (r->todo == ISO_RES_AUTO_ALLOC) + r->todo = ISO_RES_AUTO_REALLOC; + } if (channel >= 0) r->params.channels_mask = BIT_ULL(channel); From a1259e92e1e892f009bbebf23e1207b02e2d5708 Mon Sep 17 00:00:00 2001 From: ZhangGuoDong Date: Mon, 21 Sep 2026 09:41:28 +0800 Subject: [PATCH 224/365] smb: client: update POSIX extension specification references The POSIX extension specifications have now been published: https://smb3posix.org/ Suggested-by: Paulo Alcantara Reviewed-by: ChenXiaoSong Signed-off-by: ZhangGuoDong Reviewed-by: Namjae Jeon Signed-off-by: Paulo Alcantara --- fs/smb/client/smb2pdu.h | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) diff --git a/fs/smb/client/smb2pdu.h b/fs/smb/client/smb2pdu.h index ab6c667bebc0..b37a1e3941a1 100644 --- a/fs/smb/client/smb2pdu.h +++ b/fs/smb/client/smb2pdu.h @@ -224,8 +224,7 @@ struct smb2_file_id_extd_directory_info { extern char smb2_padding[7]; /* - * See POSIX-SMB2 2.2.14.2.16 - * Link: https://gitlab.com/samba-team/smb3-posix-spec/-/blob/master/smb3_posix_extensions.md + * See POSIX-SMB2 2.1.3.2.1 */ struct create_posix_rsp { u32 nlink; @@ -238,6 +237,7 @@ struct create_posix_rsp { #define SMB2_QUERY_DIRECTORY_IOV_SIZE 2 /* + * See POSIX-FSCC 2.2.1 * SMB2-only POSIX info level for query dir * * See posix_info_sid_size(), posix_info_extra_size() and @@ -256,13 +256,17 @@ struct smb2_posix_info { __le64 Inode; __le32 DeviceId; __le32 Zero; - /* beginning of POSIX Create Context Response */ + /* + * Beginning of POSIX Create Context Response + * See POSIX-SMB2 2.1.3.2.1 + */ __le32 HardLinks; __le32 ReparseTag; __le32 Mode; /* * var sized owner SID * var sized group SID + * End of POSIX Create Context Response * le32 filenamelength * u8 filename[] */ From 031fe051abb3c1f36c2ba6346931fcdb3cddaeae Mon Sep 17 00:00:00 2001 From: Fredric Cover Date: Tue, 22 Sep 2026 10:37:59 -0700 Subject: [PATCH 225/365] smb: client: use GFP_KERNEL in get_targets() Currently, get_targets() uses GFP_ATOMIC to allocate the cache target iterator and to duplicate t->name. The callers of get_targets() are in a sleepable context; therefore, switch to GFP_KERNEL to reduce risk of failure and reduce pressure on emergency pools in low-memory scenarios. Signed-off-by: Fredric Cover Signed-off-by: Paulo Alcantara --- fs/smb/client/dfs_cache.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/fs/smb/client/dfs_cache.c b/fs/smb/client/dfs_cache.c index 29dfd7595941..26956e33037e 100644 --- a/fs/smb/client/dfs_cache.c +++ b/fs/smb/client/dfs_cache.c @@ -798,13 +798,13 @@ static int get_targets(struct cache_entry *ce, struct dfs_cache_tgt_list *tl) INIT_LIST_HEAD(head); list_for_each_entry(t, &ce->tlist, list) { - it = kzalloc_obj(*it, GFP_ATOMIC); + it = kzalloc_obj(*it, GFP_KERNEL); if (!it) { rc = -ENOMEM; goto err_free_it; } - it->it_name = kstrdup(t->name, GFP_ATOMIC); + it->it_name = kstrdup(t->name, GFP_KERNEL); if (!it->it_name) { kfree(it); rc = -ENOMEM; From 76ebb69da677d2a4c5e59bf758b6424d511f5684 Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Thu, 17 Sep 2026 13:49:03 +0200 Subject: [PATCH 226/365] Revert "selftests/filesystems: add mntns cleanup test" This reverts commit 3452eecbcc954ef859b7d19309c121a78d6d0e31. The test checks that the mounts of a destroyed mount namespace stay connected. The commit that made them stay connected is reverted next because it leaks superblocks and loop devices, so drop the test with it. Link: https://patch.msgid.link/20260917-work-put_mnt_ns-revert-v1-1-34d9b8679e68@kernel.org Signed-off-by: Christian Brauner (Amutable) --- tools/testing/selftests/Makefile | 1 - .../filesystems/mntns_cleanup/.gitignore | 2 - .../filesystems/mntns_cleanup/Makefile | 6 -- .../mntns_cleanup/mntns_cleanup_test.c | 58 ------------------- 4 files changed, 67 deletions(-) delete mode 100644 tools/testing/selftests/filesystems/mntns_cleanup/.gitignore delete mode 100644 tools/testing/selftests/filesystems/mntns_cleanup/Makefile delete mode 100644 tools/testing/selftests/filesystems/mntns_cleanup/mntns_cleanup_test.c diff --git a/tools/testing/selftests/Makefile b/tools/testing/selftests/Makefile index 330d061f6366..273853937c25 100644 --- a/tools/testing/selftests/Makefile +++ b/tools/testing/selftests/Makefile @@ -49,7 +49,6 @@ TARGETS += filesystems/move_mount TARGETS += filesystems/empty_mntns TARGETS += filesystems/fsmount_ns TARGETS += filesystems/fscontext_ns -TARGETS += filesystems/mntns_cleanup TARGETS += filesystems/xattr TARGETS += firmware TARGETS += fpu diff --git a/tools/testing/selftests/filesystems/mntns_cleanup/.gitignore b/tools/testing/selftests/filesystems/mntns_cleanup/.gitignore deleted file mode 100644 index 493fbcf8d9ec..000000000000 --- a/tools/testing/selftests/filesystems/mntns_cleanup/.gitignore +++ /dev/null @@ -1,2 +0,0 @@ -# SPDX-License-Identifier: GPL-2.0-only -mntns_cleanup_test diff --git a/tools/testing/selftests/filesystems/mntns_cleanup/Makefile b/tools/testing/selftests/filesystems/mntns_cleanup/Makefile deleted file mode 100644 index 0e09e7030a5c..000000000000 --- a/tools/testing/selftests/filesystems/mntns_cleanup/Makefile +++ /dev/null @@ -1,6 +0,0 @@ -# SPDX-License-Identifier: GPL-2.0 -TEST_GEN_PROGS := mntns_cleanup_test - -CFLAGS += -Wall -O2 -g $(KHDR_INCLUDES) - -include ../../lib.mk diff --git a/tools/testing/selftests/filesystems/mntns_cleanup/mntns_cleanup_test.c b/tools/testing/selftests/filesystems/mntns_cleanup/mntns_cleanup_test.c deleted file mode 100644 index 5209712568b1..000000000000 --- a/tools/testing/selftests/filesystems/mntns_cleanup/mntns_cleanup_test.c +++ /dev/null @@ -1,58 +0,0 @@ -// SPDX-License-Identifier: GPL-2.0 - -#define _GNU_SOURCE -#include -#include -#include -#include -#include -#include - -#include "../../kselftest_harness.h" - -FIXTURE(mntns_cleanup) { -}; - -FIXTURE_SETUP(mntns_cleanup) -{ - if (geteuid() != 0) - SKIP(return, "test requires CAP_SYS_ADMIN"); - - ASSERT_EQ(unshare(CLONE_NEWNS), 0); - ASSERT_EQ(mount("", "/", NULL, MS_REC | MS_PRIVATE, NULL), 0); - - rmdir("/mnt_dir"); - ASSERT_EQ(mkdir("/mnt_dir", 0755), 0); - ASSERT_EQ(mount("tmpfs", "/mnt_dir", "tmpfs", 0, NULL), 0); - ASSERT_EQ(mkdir("/mnt_dir/hidden", 0755), 0); - ASSERT_EQ(mkdir("/mnt_dir/hidden/secret", 0755), 0); - ASSERT_EQ(mount("tmpfs", "/mnt_dir/hidden", "tmpfs", 0, NULL), 0); -} - -FIXTURE_TEARDOWN(mntns_cleanup) -{ -} - -/* Mounts must stay connected when a mount namespace is cleaned up. */ -TEST_F(mntns_cleanup, keeps_mounts_connected) -{ - int fd, sfd, err; - - fd = open("/mnt_dir", O_PATH | O_DIRECTORY | O_CLOEXEC); - ASSERT_GE(fd, 0); - - /* Destroy the namespace; the fd keeps /mnt_dir alive. */ - ASSERT_EQ(unshare(CLONE_NEWNS), 0); - - sfd = openat(fd, "hidden/secret", O_RDONLY); - err = errno; - if (sfd >= 0) - close(sfd); - close(fd); - - ASSERT_LT(sfd, 0) - TH_LOG("mount namespace teardown revealed what the overmount covered"); - ASSERT_EQ(err, ENOENT); -} - -TEST_HARNESS_MAIN From 2e2142a35d809c3cc726d3b39e7aeee98d9ab71c Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Thu, 17 Sep 2026 13:49:04 +0200 Subject: [PATCH 227/365] Revert "put_mnt_ns(): leave mounts connected" This reverts commit 0342482a4d15358fe6931606caf58968de5d1d38. Keeping every mount of a dying namespace attached to its parent changes who owns it. An attached but unmounted mount is owned by its parent: the parent's final mntput() is what drops the child's reference and takes its superblock down. Before that commit only the namespace root was in that position and every other mount dropped its own reference in namespace_unlock(). That ownership rule turns any reference from a child's superblock back to one of its ancestors into a cycle. The obvious case is a loop device: unshare -m sh -c 'mount -o loop /var/tmp/img /var/tmp/mp' mount(8) opens the image from inside the new namespace, so the loop device's backing file pins that namespace's copy of the root mount. When the namespace dies the loop mount stays attached to that copy and is owned by it. The copy can't reach zero because the backing file holds it. The backing file is only put once the ext4 superblock releases the block device and autoclear runs, and that needs the loop mount to go first. Nothing breaks the cycle. The superblock, the loop device and, since the root copy keeps all of its children. systemd-sysext sets up its loop devices in a private mount namespace and relies on autoclear when that namespace goes away, which is how this was found. Anything that holds a file on an ancestor from a mounted filesystem has the same problem. That includes ecryptfs lower paths, erofs file-backed mounts, fuse passthrough backing files. That's a bigger fix and not an -rc change. Revert. Fixes: 0342482a4d15 ("put_mnt_ns(): leave mounts connected") Reported-by: Michael Vogt Link: https://gist.github.com/mvo5/63ef46482349f3b1c3957d463a0c9c6f Link: https://patch.msgid.link/20260917-work-put_mnt_ns-revert-v1-2-34d9b8679e68@kernel.org Signed-off-by: Christian Brauner (Amutable) --- fs/namespace.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/namespace.c b/fs/namespace.c index 5e41021eaa63..580877e46b1a 100644 --- a/fs/namespace.c +++ b/fs/namespace.c @@ -6301,7 +6301,7 @@ void put_mnt_ns(struct mnt_namespace *ns) guard(namespace_excl)(); emptied_ns = ns; guard(mount_writer)(); - umount_tree(ns->root, UMOUNT_CONNECTED); + umount_tree(ns->root, 0); } struct vfsmount *kern_mount(struct file_system_type *type) From 1feb5d39b05afd902ed9fc902ec5b15be03a4bdb Mon Sep 17 00:00:00 2001 From: Bartosz Golaszewski Date: Tue, 22 Sep 2026 10:52:28 +0200 Subject: [PATCH 228/365] gpio: cdev: fix kernel stack leak to user-space in error path If we fail to acquire the GPIO chip guard in gpio_desc_to_lineinfo(), we return immediately before zeroing the info struct we'll end up passing to the user-space later in lineinfo_get_v1(). This may leak the kernel stack contents. Make gpio_desc_to_lineinfo() return int so that the -ENODEV returned on failure to acquire the guard can be propagated to the callers. While not strictly necessary: move the memset() before trying to acquire the SRCU read lock too for good measure. Fixes: d83cee3d2bb1 ("gpio: protect the pointer to gpio_chip in gpio_device with SRCU") Cc: stable@vger.kernel.org Reported-by: Sashiko Closes: https://sashiko.dev/#/patchset/20260912123529.7951-1-tzungbi%40kernel.org?part=3 Reviewed-by: Kent Gibson Link: https://patch.msgid.link/20260922-gpio-cdev-stack-leak-fixes-v3-1-7a0c7a4299d5@oss.qualcomm.com Signed-off-by: Bartosz Golaszewski --- drivers/gpio/gpiolib-cdev.c | 32 +++++++++++++++++++++++++------- 1 file changed, 25 insertions(+), 7 deletions(-) diff --git a/drivers/gpio/gpiolib-cdev.c b/drivers/gpio/gpiolib-cdev.c index d1105b7ae437..5d53bfcdf726 100644 --- a/drivers/gpio/gpiolib-cdev.c +++ b/drivers/gpio/gpiolib-cdev.c @@ -2172,18 +2172,19 @@ static void gpio_v2_line_info_changed_to_v1( #endif /* CONFIG_GPIO_CDEV_V1 */ -static void gpio_desc_to_lineinfo(struct gpio_desc *desc, - struct gpio_v2_line_info *info, bool atomic) +static int gpio_desc_to_lineinfo(struct gpio_desc *desc, + struct gpio_v2_line_info *info, bool atomic) { u32 debounce_period_us; unsigned long dflags; const char *label; + memset(info, 0, sizeof(*info)); + CLASS(gpio_chip_guard, guard)(desc); if (!guard.gc) - return; + return -ENODEV; - memset(info, 0, sizeof(*info)); info->offset = gpiod_hwgpio(desc); if (desc->name) @@ -2258,6 +2259,8 @@ static void gpio_desc_to_lineinfo(struct gpio_desc *desc, debounce_period_us; info->num_attrs++; } + + return 0; } struct gpio_chardev_data { @@ -2309,6 +2312,7 @@ static int lineinfo_get_v1(struct gpio_chardev_data *cdev, void __user *ip, struct gpio_desc *desc; struct gpioline_info lineinfo; struct gpio_v2_line_info lineinfo_v2; + int ret; if (copy_from_user(&lineinfo, ip, sizeof(lineinfo))) return -EFAULT; @@ -2326,7 +2330,10 @@ static int lineinfo_get_v1(struct gpio_chardev_data *cdev, void __user *ip, return -EBUSY; } - gpio_desc_to_lineinfo(desc, &lineinfo_v2, false); + ret = gpio_desc_to_lineinfo(desc, &lineinfo_v2, false); + if (ret) + return ret; + gpio_v2_line_info_to_v1(&lineinfo_v2, &lineinfo); if (copy_to_user(ip, &lineinfo, sizeof(lineinfo))) { @@ -2344,6 +2351,7 @@ static int lineinfo_get(struct gpio_chardev_data *cdev, void __user *ip, { struct gpio_desc *desc; struct gpio_v2_line_info lineinfo; + int ret; if (copy_from_user(&lineinfo, ip, sizeof(lineinfo))) return -EFAULT; @@ -2363,7 +2371,10 @@ static int lineinfo_get(struct gpio_chardev_data *cdev, void __user *ip, if (test_and_set_bit(lineinfo.offset, cdev->watched_lines)) return -EBUSY; } - gpio_desc_to_lineinfo(desc, &lineinfo, false); + + ret = gpio_desc_to_lineinfo(desc, &lineinfo, false); + if (ret) + return ret; if (copy_to_user(ip, &lineinfo, sizeof(lineinfo))) { if (watch) @@ -2489,6 +2500,7 @@ static int lineinfo_changed_notify(struct notifier_block *nb, struct lineinfo_changed_ctx *ctx; struct gpio_desc *desc = data; struct file *fp; + int ret; if (!test_bit(gpiod_hwgpio(desc), cdev->watched_lines)) return NOTIFY_DONE; @@ -2519,7 +2531,13 @@ static int lineinfo_changed_notify(struct notifier_block *nb, ctx->chg.event_type = action; ctx->chg.timestamp_ns = ktime_get_ns(); - gpio_desc_to_lineinfo(desc, &ctx->chg.info, true); + + ret = gpio_desc_to_lineinfo(desc, &ctx->chg.info, true); + if (ret) { + fput(fp); + return NOTIFY_DONE; + } + /* Keep the GPIO device alive until we emit the event. */ ctx->gdev = gpio_device_get(desc->gdev); ctx->cdev = cdev; From 2d2a2d7aa98741b58f54cacc99b52024e4d865f9 Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Wed, 9 Sep 2026 21:30:34 +0200 Subject: [PATCH 229/365] super: make iterate_supers_type() deletion-safe iterate_supers_type() drops sb_lock while invoking the callback and keeps only a passive reference to the current superblock. That reference keeps the object allocated, but does not keep its s_instances node linked. After the iterator releases s_umount, final teardown can unlink the current s_instances node. The iterator then advances through a reinitialized node. With the current hlist it stops without visiting the remaining superblocks. The unlink moved from generic_shutdown_super() to kill_super_notify(), but the cursor lifetime has been unsafe since the helper was introduced. The CIFS DFS lookup can consequently miss a matching superblock and return -EINVAL. Move removal from fs_supers to put_super(), alongside removal from super_blocks, so a passive reference keeps both list nodes linked. Keep the filesystem module reference until then, since unlinking s_instances may touch type->fs_supers. Make sget_fc() skip SB_DEAD superblocks before invoking test(), and set SB_DEAD under sb_lock to serialize with those callbacks. This allows kernfs to free its private information after kill_anon_super() returns. Keep matching SB_DYING superblocks until SB_DEAD is set so concurrent mounts still wait for teardown before retrying. Fixes: 43e15cdbefea ("new helper: iterate_supers_type()") Reported-by: Karl Mehltretter Closes: https://lore.kernel.org/r/20260903013336.92081-1-kmehltretter@gmail.com Suggested-by: Jan Kara Cc: stable@vger.kernel.org Tested-by: Karl Mehltretter [kmehltretter: supplied the commit message] Signed-off-by: Karl Mehltretter Link: https://patch.msgid.link/20260909193034.7467-1-kmehltretter@gmail.com Reviewed-by: Jan Kara Signed-off-by: Christian Brauner (Amutable) --- fs/kernfs/mount.c | 4 ++-- fs/super.c | 39 +++++++++++++++++++-------------------- 2 files changed, 21 insertions(+), 22 deletions(-) diff --git a/fs/kernfs/mount.c b/fs/kernfs/mount.c index f183a96778b9..a57399021c8b 100644 --- a/fs/kernfs/mount.c +++ b/fs/kernfs/mount.c @@ -434,8 +434,8 @@ void kernfs_kill_sb(struct super_block *sb) up_write(&root->kernfs_supers_rwsem); /* - * Remove the superblock from fs_supers/s_instances - * so we can't find it, before freeing kernfs_super_info. + * Mark the superblock dead so sget_fc() can't find it, + * before freeing kernfs_super_info. */ kill_anon_super(sb); kfree(info); diff --git a/fs/super.c b/fs/super.c index 01db6124e409..e41cae9e5c62 100644 --- a/fs/super.c +++ b/fs/super.c @@ -433,15 +433,19 @@ static struct super_block *alloc_super(struct file_system_type *type, int flags, void put_super(struct super_block *s) { if (refcount_dec_and_test(&s->s_passive)) { + struct file_system_type *type = s->s_type; spin_lock(&sb_lock); list_del_init(&s->s_list); + hlist_del_init(&s->s_instances); spin_unlock(&sb_lock); WARN_ON(s->s_dentry_lru.node); WARN_ON(s->s_inode_lru.node); WARN_ON(s->s_mounts); call_rcu(&s->rcu, destroy_super_rcu); + /* The unlink above may touch type->fs_supers, so drop it last. */ + put_filesystem(type); } } @@ -558,17 +562,6 @@ static void kill_super_notify(struct super_block *sb) if (sb->s_flags & SB_DEAD) return; - /* - * Remove it from @fs_supers so it isn't found by new - * sget_fc() walkers anymore. Any concurrent mounter still - * managing to grab a temporary reference is guaranteed to - * already see SB_DYING and will wait until we notify them about - * SB_DEAD. - */ - spin_lock(&sb_lock); - hlist_del_init(&sb->s_instances); - spin_unlock(&sb_lock); - /* Drop sget_fc()'s claim; a never-registered entry stays with the sb. */ if (sb->s_super_dev->sd_dev) { super_dev_put(sb->s_super_dev); @@ -577,11 +570,15 @@ static void kill_super_notify(struct super_block *sb) /* * Let concurrent mounts know that this thing is really dead. - * We don't need @sb->s_umount here as every concurrent caller - * will see SB_DYING and either discard the superblock or wait - * for SB_DEAD. + * sget_fc() skips SB_DEAD superblocks and calls test() under + * sb_lock, so set it under sb_lock: once we return no test() + * runs on this superblock anymore and none will start. Everyone + * else already saw SB_DYING and either discarded the superblock + * or waits for SB_DEAD. */ + spin_lock(&sb_lock); super_wake(sb, SB_DEAD); + spin_unlock(&sb_lock); } /** @@ -608,7 +605,6 @@ void deactivate_locked_super(struct super_block *s) list_lru_destroy(&s->s_dentry_lru); list_lru_destroy(&s->s_inode_lru); - put_filesystem(fs); put_super(s); } else { super_unlock_excl(s); @@ -795,12 +791,12 @@ void generic_shutdown_super(struct super_block *sb) } /* * Broadcast to everyone that grabbed a temporary reference to this - * superblock before we removed it from @fs_supers that the superblock - * is dying. Every walker of @fs_supers outside of sget_fc() will now - * discard this superblock and treat it as dead. + * superblock that it is dying. Every walker of @fs_supers outside + * of sget_fc() will now discard this superblock and treat it as + * dead. * - * We leave the superblock on @fs_supers so it can be found by - * sget_fc() until we passed sb->kill_sb(). + * sget_fc() keeps finding the superblock until SB_DEAD is set, so + * a concurrent mounter waits until we passed sb->kill_sb(). */ super_wake(sb, SB_DYING); super_unlock_excl(sb); @@ -879,6 +875,9 @@ struct super_block *sget_fc(struct fs_context *fc, spin_lock(&sb_lock); if (test) { hlist_for_each_entry(old, &fc->fs_type->fs_supers, s_instances) { + /* Only unlinked at the last passive reference. */ + if (super_flags(old, SB_DEAD)) + continue; if (test(old, fc)) goto share_extant_sb; } From e9438ab5328a177c9c0e5df87eb92a7162841e98 Mon Sep 17 00:00:00 2001 From: Ridham Khurana Date: Tue, 22 Sep 2026 09:20:58 +0000 Subject: [PATCH 230/365] gpio: zynq: fix runtime PM leak on request error path pm_runtime_get_sync() leaves the usage counter incremented even when it fails, and zynq_gpio_request() returns the error without dropping it. gpiolib does not call ->free() when ->request() fails, so zynq_gpio_free(), which holds the only matching pm_runtime_put(), never runs. The reference is leaked and the controller can no longer runtime-suspend, so its clock stays enabled. Switch to pm_runtime_resume_and_get(), which only increments the usage counter on success. Fixes: 3242ba117e9b ("gpio: Add driver for Zynq GPIO controller") Cc: stable@vger.kernel.org Signed-off-by: Ridham Khurana Link: https://patch.msgid.link/20260922092102.1053513-1-khurana.ridham222@gmail.com Signed-off-by: Bartosz Golaszewski --- drivers/gpio/gpio-zynq.c | 10 +--------- 1 file changed, 1 insertion(+), 9 deletions(-) diff --git a/drivers/gpio/gpio-zynq.c b/drivers/gpio/gpio-zynq.c index 15a79d9a2e9e..13e4c5f0e297 100644 --- a/drivers/gpio/gpio-zynq.c +++ b/drivers/gpio/gpio-zynq.c @@ -798,15 +798,7 @@ static int zynq_gpio_runtime_resume(struct device *dev) static int zynq_gpio_request(struct gpio_chip *chip, unsigned int offset) { - int ret; - - ret = pm_runtime_get_sync(chip->parent); - - /* - * If the device is already active pm_runtime_get() will return 1 on - * success, but gpio_request still needs to return 0. - */ - return ret < 0 ? ret : 0; + return pm_runtime_resume_and_get(chip->parent); } static void zynq_gpio_free(struct gpio_chip *chip, unsigned int offset) From cca4980630b3c7a85f53cb43c6018184ce5d4e37 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Sun, 20 Sep 2026 16:16:39 -0700 Subject: [PATCH 231/365] perf/core: Fix a refcount leak in attach_perf_ctx_data() The attach_perf_ctx_data() can race on global and !global cases. The global case is protected by global_ctx_data_rwsem and shares a single reference count using perf_ctx_data.global field. But when it races with !global case, it may miss to set the global field and result in a reference count leak. CPU1 CPU2 ---------------------------------------------------------------- attach_task_ctx_data(.global=1) attach_task_ctx_data(.global=0) cd1 = alloc_perf_ctx_data(); cd2 = alloc_perf_ctx_data(); // { .global = 0, .refcount = 1 }; try_cmpxchg(); // success, // task->perf_ctx_data = cd2 try_cmpxhg(); // fail; old = cd2 refcount_inc_not_zero(&old->refcount); // success // old.refcount = 2 free_perf_ctx_data(cd1); Then later detach_global_ctx_data() will see the data but it's not marked as global, so it won't call detach_task_ctx_data(). Fixes: 506e64e710ff ("perf: attach/detach PMU specific data") Assisted-by: Sashiko.dev:Gemini-3.1-pro Signed-off-by: Namhyung Kim Signed-off-by: Peter Zijlstra (Intel) Link: https://patch.msgid.link/20260920231639.11910-1-namhyung@kernel.org --- kernel/events/core.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/kernel/events/core.c b/kernel/events/core.c index db7b76d6b68a..e180134bad5e 100644 --- a/kernel/events/core.c +++ b/kernel/events/core.c @@ -5454,6 +5454,8 @@ attach_task_ctx_data(struct task_struct *task, struct kmem_cache *ctx_cache, } if (refcount_inc_not_zero(&old->refcount)) { + if (global) + old->global = true; free_perf_ctx_data(cd); /* unused */ return 0; } From 36bb85cf36cab15fb611cb44b78a5df06e4e69a2 Mon Sep 17 00:00:00 2001 From: Puranjay Mohan Date: Mon, 10 Aug 2026 06:35:34 -0700 Subject: [PATCH 232/365] perf/core: Fix NULL pmu_ctx passed to pmu->sched_task() perf_pmu_sched_task() returns early when cpuctx->task_ctx is set, and cpc->task_epc is only non-NULL while a task context is scheduled in on this CPU. __perf_pmu_sched_task() therefore always passes NULL: Unable to handle kernel NULL pointer dereference at virtual address 00 pc : armv8pmu_sched_task+0x14/0x50 Call trace: armv8pmu_sched_task+0x14/0x50 (P) perf_pmu_sched_task+0xac/0x108 __perf_event_task_sched_out+0x6c/0xe0 Pass &cpc->epc instead, the CPU-wide context for this PMU, which the function already dereferences a few lines up to find pmu. armv8pmu_sched_task() is the only in-tree implementation that dereferences the argument, and it only reads ->pmu, so the oops needs BRBE, added in v6.17. Fixes: bd2756811766 ("perf: Rewrite core context handling") Signed-off-by: Puranjay Mohan Signed-off-by: Peter Zijlstra (Intel) Tested-by: Yifan Wu Cc: stable@vger.kernel.org Link: https://patch.msgid.link/20260810133540.1947118-2-puranjay@kernel.org --- kernel/events/core.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/kernel/events/core.c b/kernel/events/core.c index e180134bad5e..7ce72f366f6c 100644 --- a/kernel/events/core.c +++ b/kernel/events/core.c @@ -3914,7 +3914,7 @@ static void __perf_pmu_sched_task(struct perf_cpu_pmu_context *cpc, perf_ctx_lock(cpuctx, cpuctx->task_ctx); perf_pmu_disable(pmu); - pmu->sched_task(cpc->task_epc, task, sched_in); + pmu->sched_task(&cpc->epc, task, sched_in); perf_pmu_enable(pmu); perf_ctx_unlock(cpuctx, cpuctx->task_ctx); From 3d8d74100954a3b17e5c5e37adfe14e16b1db103 Mon Sep 17 00:00:00 2001 From: Puranjay Mohan Date: Mon, 10 Aug 2026 06:35:35 -0700 Subject: [PATCH 233/365] perf/core: Run sched_task() for PMUs with only CPU-wide events perf_pmu_sched_task() returns early when cpuctx->task_ctx is set and leaves the work to perf_ctx_sched_task_cb(), which only walks ctx->pmu_ctx_list. A PMU whose events are all CPU-wide is not on that list, so nothing calls its sched_task(). With perf record -b -e cycles -a -- ls armv8pmu_sched_task() is skipped on every switch to a task that has a perf context but no event on that PMU, and BRBE records leak across the task boundary. intel_pmu_lbr_add() calls perf_sched_cb_inc() unconditionally too, so LBR records leak the same way on x86. Drop the early return and skip only the CPCs that perf_ctx_sched_task_cb() handles. That one needs a gate of its own to make the split exact: it tests cpc->sched_cb_usage, which perf_sched_cb_inc() sets per CPU for every branch stack user, so a task with an event for that PMU pinned to another CPU would be handled twice. On x86 the second __intel_pmu_lbr_restore() finds lbr_stack_state == LBR_NONE and calls intel_pmu_lbr_reset(), throwing away the callstack the first one restored. cpc->task_epc is set only while a task context is scheduled in, and there is one epc per PMU on ctx->pmu_ctx_list, so the two gates are inverses. For the CPCs perf_pmu_sched_task() picks up, the callback now runs outside the perf_ctx_disable() and perf_ctx_enable() pair in perf_event_context_sched_in(). __perf_pmu_sched_task() disables the PMU around the call itself. Fixes: bd2756811766 ("perf: Rewrite core context handling") Signed-off-by: Puranjay Mohan Signed-off-by: Peter Zijlstra (Intel) Tested-by: Yifan Wu Link: https://patch.msgid.link/20260810133540.1947118-3-puranjay@kernel.org Cc: stable@vger.kernel.org --- kernel/events/core.c | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-) diff --git a/kernel/events/core.c b/kernel/events/core.c index 7ce72f366f6c..634d2ccbab82 100644 --- a/kernel/events/core.c +++ b/kernel/events/core.c @@ -3764,6 +3764,9 @@ static void perf_ctx_sched_task_cb(struct perf_event_context *ctx, list_for_each_entry(pmu_ctx, &ctx->pmu_ctx_list, pmu_ctx_entry) { cpc = this_cpc(pmu_ctx->pmu); + if (cpc->task_epc != pmu_ctx) + continue; + if (cpc->sched_cb_usage && pmu_ctx->pmu->sched_task) pmu_ctx->pmu->sched_task(pmu_ctx, task, sched_in); } @@ -3924,15 +3927,17 @@ static void perf_pmu_sched_task(struct task_struct *prev, struct task_struct *next, bool sched_in) { - struct perf_cpu_context *cpuctx = this_cpu_ptr(&perf_cpu_context); struct perf_cpu_pmu_context *cpc, *cpc2; - /* cpuctx->task_ctx will be handled in perf_event_context_sched_in/out */ - if (prev == next || cpuctx->task_ctx) + if (prev == next) return; - list_for_each_entry_safe(cpc, cpc2, this_cpu_ptr(&sched_cb_list), sched_cb_entry) + list_for_each_entry_safe(cpc, cpc2, this_cpu_ptr(&sched_cb_list), sched_cb_entry) { + if (cpc->task_epc) + continue; + __perf_pmu_sched_task(cpc, sched_in ? next : prev, sched_in); + } } static void perf_event_switch(struct task_struct *task, From 24b620729e53d978b3e425f55bc66efd3bab1f59 Mon Sep 17 00:00:00 2001 From: Puranjay Mohan Date: Mon, 10 Aug 2026 06:35:36 -0700 Subject: [PATCH 234/365] perf/core: Fill branch entries with a single assignment perf_clear_branch_entry_bitfields() clears the bitfields of struct perf_branch_entry one by one and leaves from/to alone, since callers overwrite those straight away. The list has to be kept in sync with the struct by hand and has already fallen behind: new_type and priv were added to perf_branch_entry and never added here. Only BRBE writes those two, and neither for every record. brbe_set_perf_entry_type() leaves new_type alone for a branch type it does not recognise, and priv is not set for source-only records. arm_pmuv3.c allocates the per-CPU branch stack with kmalloc(), so such a record reaches userspace with whatever the slot held: uninitialised kmalloc() data on the first pass over the buffer, the previous record's values after that. Nothing under arch/x86/events/ writes either field, so only arm64 is affected. Assign the whole entry at each site instead. Everything not named is then zero, and there is no list to keep in sync. The bitfields add up to exactly 64 bits, so the struct has no padding to leave undefined. perf_clear_branch_entry_bitfields() has no callers left, so remove it. perf_entry_from_brbe_regset() assigns an empty literal instead, since it fills from/to conditionally. PERF_BR_SPEC_NA is 0, so dropping the explicit spec assignment changes nothing. Fixes: b190bc4ac9e6 ("perf: Extend branch type classification") Fixes: 5402d25aa571 ("perf: Capture branch privilege information") Suggested-by: Peter Zijlstra Signed-off-by: Puranjay Mohan Signed-off-by: Peter Zijlstra (Intel) Tested-by: Yifan Wu Link: https://patch.msgid.link/20260810133540.1947118-4-puranjay@kernel.org --- arch/x86/events/amd/brs.c | 9 +++--- arch/x86/events/amd/lbr.c | 16 +++++----- arch/x86/events/intel/lbr.c | 63 ++++++++++++++++++++----------------- drivers/perf/arm_brbe.c | 2 +- include/linux/perf_event.h | 17 ---------- 5 files changed, 47 insertions(+), 60 deletions(-) diff --git a/arch/x86/events/amd/brs.c b/arch/x86/events/amd/brs.c index dc564688f3d7..54b13faba116 100644 --- a/arch/x86/events/amd/brs.c +++ b/arch/x86/events/amd/brs.c @@ -343,11 +343,10 @@ void amd_brs_drain(void) if (!amd_brs_match_plm(event, from, to)) continue; - perf_clear_branch_entry_bitfields(br+nr); - - br[nr].from = from; - br[nr].to = to; - + br[nr] = (struct perf_branch_entry){ + .from = from, + .to = to, + }; nr++; } empty: diff --git a/arch/x86/events/amd/lbr.c b/arch/x86/events/amd/lbr.c index 9d9c961989d5..a55646fcb846 100644 --- a/arch/x86/events/amd/lbr.c +++ b/arch/x86/events/amd/lbr.c @@ -184,13 +184,6 @@ void amd_pmu_lbr_read(void) entry.to.split.reserved) continue; - perf_clear_branch_entry_bitfields(br + out); - - br[out].from = sign_ext_branch_ip(entry.from.split.ip); - br[out].to = sign_ext_branch_ip(entry.to.split.ip); - br[out].mispred = entry.from.split.mispredict; - br[out].predicted = !br[out].mispred; - /* * Set branch speculation information using the status of * the valid and spec bits. @@ -208,7 +201,14 @@ void amd_pmu_lbr_read(void) * speculative and took the correct path */ idx = (entry.to.split.valid << 1) | entry.to.split.spec; - br[out].spec = lbr_spec_map[idx]; + + br[out] = (struct perf_branch_entry){ + .from = sign_ext_branch_ip(entry.from.split.ip), + .to = sign_ext_branch_ip(entry.to.split.ip), + .mispred = entry.from.split.mispredict, + .predicted = !entry.from.split.mispredict, + .spec = lbr_spec_map[idx], + }; out++; } diff --git a/arch/x86/events/intel/lbr.c b/arch/x86/events/intel/lbr.c index cbe5c762008d..22e2a06d5786 100644 --- a/arch/x86/events/intel/lbr.c +++ b/arch/x86/events/intel/lbr.c @@ -756,10 +756,10 @@ void intel_pmu_lbr_read_32(struct cpu_hw_events *cpuc) rdmsrq(x86_pmu.lbr_from + lbr_idx, msr_lastbranch.lbr); - perf_clear_branch_entry_bitfields(br); - - br->from = msr_lastbranch.from; - br->to = msr_lastbranch.to; + *br = (struct perf_branch_entry){ + .from = msr_lastbranch.from, + .to = msr_lastbranch.to, + }; br++; } cpuc->lbr_stack.nr = i; @@ -847,14 +847,15 @@ void intel_pmu_lbr_read_64(struct cpu_hw_events *cpuc) if (abort && x86_pmu.lbr_double_abort && out > 0) out--; - perf_clear_branch_entry_bitfields(br+out); - br[out].from = from; - br[out].to = to; - br[out].mispred = mis; - br[out].predicted = pred; - br[out].in_tx = in_tx; - br[out].abort = abort; - br[out].cycles = cycles; + br[out] = (struct perf_branch_entry){ + .from = from, + .to = to, + .mispred = mis, + .predicted = pred, + .in_tx = in_tx, + .abort = abort, + .cycles = cycles, + }; out++; } cpuc->lbr_stack.nr = out; @@ -905,6 +906,7 @@ static void intel_pmu_store_lbr(struct cpu_hw_events *cpuc, struct perf_branch_entry *e; struct lbr_entry *lbr; u64 from, to, info; + bool mispred; int i; for (i = 0; i < x86_pmu.lbr_nr; i++) { @@ -921,24 +923,27 @@ static void intel_pmu_store_lbr(struct cpu_hw_events *cpuc, to = rdlbr_to(i, lbr); info = rdlbr_info(i, lbr); - perf_clear_branch_entry_bitfields(e); + mispred = get_lbr_mispred(info); - e->from = from; - e->to = to; - e->mispred = get_lbr_mispred(info); - e->predicted = !e->mispred; - e->in_tx = !!(info & LBR_INFO_IN_TX); - e->abort = !!(info & LBR_INFO_ABORT); - e->cycles = get_lbr_cycles(info); - e->type = get_lbr_br_type(info); - - /* - * Leverage the reserved field of cpuc->lbr_entries[i] to - * temporarily store the branch counters information. - * The later code will decide what content can be disclosed - * to the perf tool. Pleae see intel_pmu_lbr_counters_reorder(). - */ - e->reserved = (info >> LBR_INFO_BR_CNTR_OFFSET) & LBR_INFO_BR_CNTR_FULL_MASK; + *e = (struct perf_branch_entry){ + .from = from, + .to = to, + .mispred = mispred, + .predicted = !mispred, + .in_tx = !!(info & LBR_INFO_IN_TX), + .abort = !!(info & LBR_INFO_ABORT), + .cycles = get_lbr_cycles(info), + .type = get_lbr_br_type(info), + /* + * Leverage the reserved field of + * cpuc->lbr_entries[i] to temporarily store the + * branch counters information. The later code will + * decide what content can be disclosed to the perf + * tool. Pleae see intel_pmu_lbr_counters_reorder(). + */ + .reserved = (info >> LBR_INFO_BR_CNTR_OFFSET) & + LBR_INFO_BR_CNTR_FULL_MASK, + }; } cpuc->lbr_stack.nr = i; diff --git a/drivers/perf/arm_brbe.c b/drivers/perf/arm_brbe.c index ba554e0c846c..254be4da8ae2 100644 --- a/drivers/perf/arm_brbe.c +++ b/drivers/perf/arm_brbe.c @@ -604,7 +604,7 @@ static bool perf_entry_from_brbe_regset(int index, struct perf_branch_entry *ent return false; brbinf = bregs.brbinf; - perf_clear_branch_entry_bitfields(entry); + *entry = (struct perf_branch_entry){ }; if (brbe_record_is_complete(brbinf)) { entry->from = bregs.brbsrc; entry->to = bregs.brbtgt; diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index 5842552294c1..915c6fd3f084 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -1467,23 +1467,6 @@ static inline u32 perf_sample_data_size(struct perf_sample_data *data, return size; } -/* - * Clear all bitfields in the perf_branch_entry. - * The to and from fields are not cleared because they are - * systematically modified by caller. - */ -static inline void perf_clear_branch_entry_bitfields(struct perf_branch_entry *br) -{ - br->mispred = 0; - br->predicted = 0; - br->in_tx = 0; - br->abort = 0; - br->cycles = 0; - br->type = 0; - br->spec = PERF_BR_SPEC_NA; - br->reserved = 0; -} - extern void perf_output_sample(struct perf_output_handle *handle, struct perf_event_header *header, struct perf_sample_data *data, From 40eb4b18ad167b63644c27fa67a7400ba050379a Mon Sep 17 00:00:00 2001 From: Chandrashekar Devegowda Date: Tue, 22 Sep 2026 12:10:20 +0530 Subject: [PATCH 235/365] Bluetooth: btintel: fix buffer over-read in btintel_hw_error() The "Exception info %s" print uses skb->data + 1, but the 12-byte payload is not guaranteed to be NUL-terminated and can be read past its end. Bound the print with "%.*s" and skb->len - 1. Fixes: 973bb97e5aee ("Bluetooth: btintel: Add generic function for handling hardware errors") Cc: stable@vger.kernel.org Signed-off-by: Chandrashekar Devegowda Signed-off-by: Luiz Augusto von Dentz --- drivers/bluetooth/btintel.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/bluetooth/btintel.c b/drivers/bluetooth/btintel.c index 909a265fd906..e35fff90b31d 100644 --- a/drivers/bluetooth/btintel.c +++ b/drivers/bluetooth/btintel.c @@ -289,7 +289,8 @@ void btintel_hw_error(struct hci_dev *hdev, u8 code) goto unlock; } - bt_dev_err(hdev, "Exception info %s", (char *)(skb->data + 1)); + bt_dev_err(hdev, "Exception info %.*s", (int)(skb->len - 1), + (char *)(skb->data + 1)); kfree_skb(skb); From 7dbeb6935ac331b0cba6c41cabd8e9d02d5f64ea Mon Sep 17 00:00:00 2001 From: Chandrashekar Devegowda Date: Tue, 22 Sep 2026 12:10:21 +0530 Subject: [PATCH 236/365] Bluetooth: btintel_pcie: fix plen overflow in btintel_pcie_recv_frame() plen is u16 and computed as HDR_SIZE + dlen. For ACL/ISO frames dlen is a 16-bit field, so a value >= 0xFFFC wraps the sum, bypassing the "skb->len < plen" check and letting skb_trim() truncate the packet. Widen plen to u32 so the addition cannot wrap. Fixes: c2b636b3f788 ("Bluetooth: btintel_pcie: Add support for PCIe transport") Cc: stable@vger.kernel.org Signed-off-by: Chandrashekar Devegowda Signed-off-by: Luiz Augusto von Dentz --- drivers/bluetooth/btintel_pcie.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/bluetooth/btintel_pcie.c b/drivers/bluetooth/btintel_pcie.c index 2819e001797b..d05af27790f9 100644 --- a/drivers/bluetooth/btintel_pcie.c +++ b/drivers/bluetooth/btintel_pcie.c @@ -1197,7 +1197,7 @@ static int btintel_pcie_recv_frame(struct btintel_pcie_data *data, { int ret; u8 pkt_type; - u16 plen; + u32 plen; u32 pcie_pkt_type; void *pdata; struct hci_dev *hdev = data->hdev; From f771f96557a4d953a7e4d045870547271e36e819 Mon Sep 17 00:00:00 2001 From: Chandrashekar Devegowda Date: Tue, 22 Sep 2026 12:10:22 +0530 Subject: [PATCH 237/365] Bluetooth: btintel_pcie: reject oversized TX packets in send_frame() btintel_pcie_send_frame() eventually copies skb->data into a fixed BTINTEL_PCIE_BUFFER_SIZE (4096) DMA buffer via memcpy() in btintel_pcie_prepare_tx(), with a 4-byte PCIe type header prepended. A user-space process with HCI_CHANNEL_USER access can inject an oversized packet and overflow the buffer. Reject packets larger than BTINTEL_PCIE_BUFFER_SIZE - BTINTEL_PCIE_HCI_TYPE_LEN on entry. Fixes: 6e65a09f9275 ("Bluetooth: btintel_pcie: Add *setup* function to download firmware") Cc: stable@vger.kernel.org Signed-off-by: Chandrashekar Devegowda Signed-off-by: Luiz Augusto von Dentz --- drivers/bluetooth/btintel_pcie.c | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/drivers/bluetooth/btintel_pcie.c b/drivers/bluetooth/btintel_pcie.c index d05af27790f9..f333b01f7465 100644 --- a/drivers/bluetooth/btintel_pcie.c +++ b/drivers/bluetooth/btintel_pcie.c @@ -2219,6 +2219,16 @@ static int btintel_pcie_send_frame(struct hci_dev *hdev, if (test_bit(BTINTEL_PCIE_RECOVERY_IN_PROGRESS, &data->flags)) return -ENODEV; + /* Account for the 4-byte PCIe type header prepended before the + * DMA copy. Written as a subtraction to avoid wrap-around on + * attacker-controlled skb->len. + */ + if (skb->len > BTINTEL_PCIE_BUFFER_SIZE - BTINTEL_PCIE_HCI_TYPE_LEN) { + bt_dev_err(hdev, "Packet too large: %u > %u", skb->len, + BTINTEL_PCIE_BUFFER_SIZE - BTINTEL_PCIE_HCI_TYPE_LEN); + return -EMSGSIZE; + } + /* Due to the fw limitation, the type header of the packet should be * 4 bytes unlike 1 byte for UART. In UART, the firmware can read * the first byte to get the packet type and redirect the rest of data From 74847177eca5638827ab74cdf77f0b10fc40ca6c Mon Sep 17 00:00:00 2001 From: Linmao Li Date: Tue, 22 Sep 2026 10:01:41 +0800 Subject: [PATCH 238/365] Bluetooth: hci_core: Fix inquiry cache timestamps on 64-bit systems On 64-bit systems, outgoing BR/EDR connections always fall back to page scan repetition mode R2 with no clock offset once the system has been up for more than five minutes, even when inquiry found the peer only seconds earlier. This lengthens paging and can increase the risk of a Page Timeout. The inquiry cache stores jiffies in __u32 timestamps, but its age helpers subtract them from unsigned long jiffies. INITIAL_JIFFIES casts -300 * HZ through unsigned int, so jiffies crosses 2^32 five minutes after boot on 64-bit systems. Assigning it to __u32 then drops the upper 32 bits. In one trace, a 7.6-second-old entry (HZ=1000) was reported as 2^32 + 7620 ticks old and rejected by hci_acl_create_conn_sync(). hci_inquiry() is affected by the same truncation when checking the whole cache. 32-bit systems are unaffected because unsigned long is 32 bits wide there. Use unsigned long for both timestamps so they have the same width as jiffies on 32-bit and 64-bit systems, and update the debugfs format specifier accordingly. Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Cc: stable@vger.kernel.org Signed-off-by: Linmao Li Signed-off-by: Luiz Augusto von Dentz --- include/net/bluetooth/hci_core.h | 4 ++-- net/bluetooth/hci_debugfs.c | 2 +- 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/include/net/bluetooth/hci_core.h b/include/net/bluetooth/hci_core.h index 4105c446ca98..ca77ad0d9393 100644 --- a/include/net/bluetooth/hci_core.h +++ b/include/net/bluetooth/hci_core.h @@ -62,7 +62,7 @@ struct inquiry_entry { NAME_PENDING, NAME_KNOWN, } name_state; - __u32 timestamp; + unsigned long timestamp; struct inquiry_data data; }; @@ -78,7 +78,7 @@ struct discovery_state { struct list_head all; /* All devices found during inquiry */ struct list_head unknown; /* Name state not known */ struct list_head resolve; /* Name needs to be resolved */ - __u32 timestamp; + unsigned long timestamp; bdaddr_t last_adv_addr; u8 last_adv_addr_type; s8 last_adv_rssi; diff --git a/net/bluetooth/hci_debugfs.c b/net/bluetooth/hci_debugfs.c index aadffaaff20e..2559fb6324d6 100644 --- a/net/bluetooth/hci_debugfs.c +++ b/net/bluetooth/hci_debugfs.c @@ -364,7 +364,7 @@ static int inquiry_cache_show(struct seq_file *f, void *p) list_for_each_entry(e, &cache->all, all) { struct inquiry_data *data = &e->data; - seq_printf(f, "%pMR %d %d %d 0x%.2x%.2x%.2x 0x%.4x %d %d %u\n", + seq_printf(f, "%pMR %d %d %d 0x%.2x%.2x%.2x 0x%.4x %d %d %lu\n", &data->bdaddr, data->pscan_rep_mode, data->pscan_period_mode, data->pscan_mode, data->dev_class[2], From d1ac915fb4184faad1b62148be50f317de7c24b5 Mon Sep 17 00:00:00 2001 From: Sang-Hoon Choi Date: Tue, 22 Sep 2026 03:49:28 +0900 Subject: [PATCH 239/365] Bluetooth: hci_core: free the HCI ID if naming fails hci_register_dev() allocates an ID before calling dev_set_name(). If naming fails, it returns without releasing that ID. The device has not been registered and hdev->id has not been assigned, so the normal unregister path cannot release it. Free the allocated ID before returning the naming error. Fixes: dcda165706b9 ("Bluetooth: hci_core: Fix build warnings") Reported-by: Changyul Lee Assisted-by: LLM Signed-off-by: Sang-Hoon Choi Reported-by: Changyul Lee Signed-off-by: Luiz Augusto von Dentz --- net/bluetooth/hci_core.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/net/bluetooth/hci_core.c b/net/bluetooth/hci_core.c index d183efaf9063..bbe391525530 100644 --- a/net/bluetooth/hci_core.c +++ b/net/bluetooth/hci_core.c @@ -2557,8 +2557,10 @@ int hci_register_dev(struct hci_dev *hdev) return id; error = dev_set_name(&hdev->dev, "hci%u", id); - if (error) + if (error) { + ida_free(&hci_index_ida, id); return error; + } hdev->name = dev_name(&hdev->dev); hdev->id = id; From c9c15d4d8956df8c564f7c210e4dc5b9b820d97a Mon Sep 17 00:00:00 2001 From: Sai Teja Aluvala Date: Tue, 22 Sep 2026 18:33:11 +0530 Subject: [PATCH 240/365] Bluetooth: btintel: validate DDC record lengths Parse DDC record lengths in unsigned storage so 0xFF does not wrap. Reject records smaller than the mandatory three-byte header, records exceeding U8_MAX, and records exceeding remaining firmware bytes before issuing Intel_Write_DDC. This issue was reported by Claude Mythos. Fixes: 145f2368c5fd ("Bluetooth: btintel: Add Device Configuration support") Signed-off-by: Sai Teja Aluvala Signed-off-by: Luiz Augusto von Dentz --- drivers/bluetooth/btintel.c | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-) diff --git a/drivers/bluetooth/btintel.c b/drivers/bluetooth/btintel.c index e35fff90b31d..c69a5d226e88 100644 --- a/drivers/bluetooth/btintel.c +++ b/drivers/bluetooth/btintel.c @@ -409,8 +409,16 @@ int btintel_load_ddc_config(struct hci_dev *hdev, const char *ddc_name) /* DDC file contains one or more DDC structure which has * Length (1 byte), DDC ID (2 bytes), and DDC value (Length - 2). */ - while (fw->size > fw_ptr - fw->data) { - u8 cmd_plen = fw_ptr[0] + sizeof(u8); + while (fw->size > (size_t)(fw_ptr - fw->data)) { + size_t remaining = fw->size - (fw_ptr - fw->data); + unsigned int cmd_plen = fw_ptr[0] + 1U; + + if (cmd_plen < 3 || cmd_plen > U8_MAX || cmd_plen > remaining) { + bt_dev_err(hdev, "Malformed DDC record (plen=%u, remaining=%zu)", + cmd_plen, remaining); + release_firmware(fw); + return -EINVAL; + } skb = __hci_cmd_sync(hdev, 0xfc8b, cmd_plen, fw_ptr, HCI_INIT_TIMEOUT); From 4cbe530c0233c7413aaaeb029a4f32dd6aadacbb Mon Sep 17 00:00:00 2001 From: Karl Mehltretter Date: Sat, 19 Sep 2026 19:10:58 +0200 Subject: [PATCH 241/365] gpio: tps65219: Fix GPIO input value reads TPS65219_MFP_GPIO_STATUS_MASK is already BIT(4). Passing it to BIT() again tests bit 16, which cannot be set in the 8-bit MFP_CTRL register, so GPIO0 is always reported low when configured as an input. Test the register value with the mask directly. Fixes: 57e30e00bd5b ("gpio: tps65219: add GPIO support for TPS65219 PMIC") Cc: stable@vger.kernel.org Assisted-by: LLM Signed-off-by: Karl Mehltretter Reviewed-by: Jonathan Cormier Link: https://patch.msgid.link/20260919171100.90430-2-kmehltretter@gmail.com Signed-off-by: Bartosz Golaszewski --- drivers/gpio/gpio-tps65219.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpio/gpio-tps65219.c b/drivers/gpio/gpio-tps65219.c index 457fd8a589e8..b25c6f727768 100644 --- a/drivers/gpio/gpio-tps65219.c +++ b/drivers/gpio/gpio-tps65219.c @@ -79,7 +79,7 @@ static int tps65219_gpio_get(struct gpio_chip *gc, unsigned int offset) if (ret) return ret; - ret = !!(val & BIT(TPS65219_MFP_GPIO_STATUS_MASK)); + ret = !!(val & TPS65219_MFP_GPIO_STATUS_MASK); dev_warn(dev, "GPIO%d = %d, MULTI_DEVICE_ENABLE, not a standard GPIO\n", offset, ret); /* From 93cf8cedeaaa05714f709b539cfb976e0b80c830 Mon Sep 17 00:00:00 2001 From: Karl Mehltretter Date: Sat, 19 Sep 2026 19:10:59 +0200 Subject: [PATCH 242/365] gpio: tps65219: Use the variant-specific direction callback The TPS65214 template installs its own get_direction callback because its direction bit is in GENERAL_CONFIG. The shared get and direction callbacks nevertheless call tps65219_gpio_get_direction() directly and interpret the unrelated TPS65219 MFP bit. On TPS65214 this can reject reads from an input and skip the change from input to output. Call the callback selected by the gpio_chip template instead. Fixes: 1b6ab07c0c80 ("gpio: tps65219: Add support for TI TPS65214 PMIC") Cc: stable@vger.kernel.org Assisted-by: LLM Signed-off-by: Karl Mehltretter Link: https://patch.msgid.link/20260919171100.90430-3-kmehltretter@gmail.com Signed-off-by: Bartosz Golaszewski --- drivers/gpio/gpio-tps65219.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/drivers/gpio/gpio-tps65219.c b/drivers/gpio/gpio-tps65219.c index b25c6f727768..479a80ef7654 100644 --- a/drivers/gpio/gpio-tps65219.c +++ b/drivers/gpio/gpio-tps65219.c @@ -87,7 +87,7 @@ static int tps65219_gpio_get(struct gpio_chip *gc, unsigned int offset) * status bit. */ - if (tps65219_gpio_get_direction(gc, offset) == GPIO_LINE_DIRECTION_OUT) + if (gc->get_direction(gc, offset) == GPIO_LINE_DIRECTION_OUT) return -ENOTSUPP; return ret; @@ -176,7 +176,7 @@ static int tps65219_gpio_direction_input(struct gpio_chip *gc, unsigned int offs return -ENOTSUPP; } - if (tps65219_gpio_get_direction(gc, offset) == GPIO_LINE_DIRECTION_IN) + if (gc->get_direction(gc, offset) == GPIO_LINE_DIRECTION_IN) return 0; return gpio->change_dir(gc, offset, GPIO_LINE_DIRECTION_IN); @@ -190,7 +190,7 @@ static int tps65219_gpio_direction_output(struct gpio_chip *gc, unsigned int off if (offset != TPS6521X_GPIO0_IDX) return 0; - if (tps65219_gpio_get_direction(gc, offset) == GPIO_LINE_DIRECTION_OUT) + if (gc->get_direction(gc, offset) == GPIO_LINE_DIRECTION_OUT) return 0; return gpio->change_dir(gc, offset, GPIO_LINE_DIRECTION_OUT); From 270437f3fe62516f16482742a7762a075e7a9457 Mon Sep 17 00:00:00 2001 From: Karl Mehltretter Date: Sat, 19 Sep 2026 19:11:00 +0200 Subject: [PATCH 243/365] gpio: tps65219: Fix TPS65214 GPIO direction programming GPIO_LINE_DIRECTION_OUT and GPIO_LINE_DIRECTION_IN have the values 0 and 1, respectively, while the TPS65214 GPIO_CONFIG field is BIT(1). regmap_update_bits() masks the supplied value, so passing either direction value clears the field and selects input mode. Translate the GPIO direction to the register encoding used by tps65214_gpio_get_direction(), setting GPIO_CONFIG for output and clearing it for input. Fixes: 1b6ab07c0c80 ("gpio: tps65219: Add support for TI TPS65214 PMIC") Cc: stable@vger.kernel.org Assisted-by: LLM Signed-off-by: Karl Mehltretter Link: https://patch.msgid.link/20260919171100.90430-4-kmehltretter@gmail.com Signed-off-by: Bartosz Golaszewski --- drivers/gpio/gpio-tps65219.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/drivers/gpio/gpio-tps65219.c b/drivers/gpio/gpio-tps65219.c index 479a80ef7654..6958466455d0 100644 --- a/drivers/gpio/gpio-tps65219.c +++ b/drivers/gpio/gpio-tps65219.c @@ -158,8 +158,10 @@ static int tps65214_gpio_change_direction(struct gpio_chip *gc, unsigned int off if (ret) dev_err(dev, "GPIO%d configured as VSEL, not GPIO\n", offset); + val = direction == GPIO_LINE_DIRECTION_OUT ? + TPS65214_GPIO0_DIR_MASK : 0; ret = regmap_update_bits(gpio->tps->regmap, TPS65219_REG_GENERAL_CONFIG, - TPS65214_GPIO0_DIR_MASK, direction); + TPS65214_GPIO0_DIR_MASK, val); if (ret) dev_err(dev, "Fail to change direction to %u for GPIO%d.\n", direction, offset); From 5b76268dac968612f7283d59b539036de955b7d9 Mon Sep 17 00:00:00 2001 From: Vineeth Vijayan Date: Tue, 22 Sep 2026 22:48:39 +0200 Subject: [PATCH 244/365] s390/cio: Fix NULL pointer dereference in ccw_device_get_util_str() The channel path registry entry associated with a CHPID may be removed while the subchannel's PMCW still references that CHPID. In this case, chpid_to_chp() can return NULL, leading to a NULL pointer dereference. Add the missing NULL check before dereferencing the returned pointer. Fixes: 199652309a4d ("s390/cio: add helper to query utility strings per given ccw device") Cc: stable@vger.kernel.org Signed-off-by: Vineeth Vijayan Reviewed-by: Peter Oberparleiter Signed-off-by: Heiko Carstens --- drivers/s390/cio/device_ops.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/s390/cio/device_ops.c b/drivers/s390/cio/device_ops.c index f2f7f8cba410..1f7e83fd5087 100644 --- a/drivers/s390/cio/device_ops.c +++ b/drivers/s390/cio/device_ops.c @@ -517,6 +517,8 @@ u8 *ccw_device_get_util_str(struct ccw_device *cdev, int chp_idx) chp_id_init(&chpid); chpid.id = sch->schib.pmcw.chpid[chp_idx]; chp = chpid_to_chp(chpid); + if (!chp) + return NULL; util_str = kmalloc(sizeof(chp->desc_fmt3.util_str), GFP_KERNEL); if (!util_str) From 883b776abe48fed8ef615f4ff7e91113a6c4dffb Mon Sep 17 00:00:00 2001 From: Jan Kara Date: Tue, 22 Sep 2026 12:41:02 +0200 Subject: [PATCH 245/365] isofs: Fix handling of directories with tight blocks Thomas has reported that after commit b2eb2e288604 ("isofs: Drop support of directory entries straddling blocks") some isofs images of Fedora miss some entries in some directories. The problem is triggered when directory entries are packed in a block in such a way that they exactly fit the block (which BTW mkisofs doesn't do so I didn't catch this bug when testing with my images). Fix readdir and lookup code to properly transition to the next block when directory block is tightly packed. Link: https://bugzilla.redhat.com/show_bug.cgi?id=2535353 Reported-by: Thomas Schmitt Reported-by: Matthias Goergens Fixes: b2eb2e288604 ("isofs: Drop support of directory entries straddling blocks") Reviewed-by: Thomas Schmitt Signed-off-by: Jan Kara --- fs/isofs/dir.c | 20 ++++++++------------ fs/isofs/namei.c | 16 +++++++++------- 2 files changed, 17 insertions(+), 19 deletions(-) diff --git a/fs/isofs/dir.c b/fs/isofs/dir.c index c7ca7603e97a..28741251d56e 100644 --- a/fs/isofs/dir.c +++ b/fs/isofs/dir.c @@ -110,25 +110,21 @@ static int do_isofs_readdir(struct inode *inode, struct file *file, return 0; } - de = (struct iso_directory_record *) (bh->b_data + offset); - - de_len = *(unsigned char *)de; - + de = (struct iso_directory_record *)(bh->b_data + offset); /* - * If the length byte is zero, we should move on to the next - * CDROM sector. If we are at the end of the directory, we - * kick out of the while loop. + * If we are at the end of a block (or at its zero-padded + * tail), move on to the next CDROM sector. If we are at the + * end of the directory, we'll abort the while loop. */ - - if (de_len == 0) { + if (offset >= bufsize || de->length[0] == 0) { brelse(bh); bh = NULL; - ctx->pos = (ctx->pos + ISOFS_BLOCK_SIZE) & ~(ISOFS_BLOCK_SIZE - 1); - block = ctx->pos >> bufbits; + block++; + ctx->pos = (loff_t)block << bufbits; offset = 0; continue; } - + de_len = de->length[0]; block_saved = block; offset_saved = offset; offset += de_len; diff --git a/fs/isofs/namei.c b/fs/isofs/namei.c index 010682f5901a..4fba1bf7f016 100644 --- a/fs/isofs/namei.c +++ b/fs/isofs/namei.c @@ -74,18 +74,20 @@ isofs_find_entry(struct inode *dir, struct dentry *dentry, return 0; } - de = (struct iso_directory_record *) (bh->b_data + offset); - - de_len = *(unsigned char *) de; - if (!de_len) { + de = (struct iso_directory_record *)(bh->b_data + offset); + /* + * If we are at the end of the block or at its zero-padded + * tail, move to the next block. + */ + if (offset >= bufsize || de->length[0] == 0) { brelse(bh); bh = NULL; - f_pos = (f_pos + ISOFS_BLOCK_SIZE) & ~(ISOFS_BLOCK_SIZE - 1); - block = f_pos >> bufbits; + block++; + f_pos = block << bufbits; offset = 0; continue; } - + de_len = de->length[0]; block_saved = bh->b_blocknr; offset_saved = offset; offset += de_len; From 4409a85735cdca5c4c400c0dad1feee091872eee Mon Sep 17 00:00:00 2001 From: Liang Luo Date: Wed, 23 Sep 2026 19:07:37 +0800 Subject: [PATCH 246/365] sched_ext: Count SCX_EV_SUB_BYPASS_DISPATCH in the dispatch fallback When a descendant scheduler enters bypass mode, its tasks are parked in the bypass DSQs of the nearest non-bypassing ancestor, which is then responsible for running them. On behalf of such a non-bypassing host, scx_dispatch_sched() consumes those bypass DSQs from two places: the attempt made every SCX_BYPASS_HOST_NTH dispatches, and the end-of-dispatch fallback that keeps the CPU from going idle while bypassed descendants still have tasks queued. The former increments SCX_EV_SUB_BYPASS_DISPATCH but the latter does not, even though both perform the same scx_consume_dispatch_q() on the same bypass DSQ. The descendant bypass dispatches done by the fallback are therefore missing from the counter exposed via sysfs, scx_dump_state() and the scx_bpf_events() kfunc, which under-reports the actual number of such dispatches. Add the missing __scx_add_event() so the fallback counts them too. When @sch itself is bypassing, scx_dispatch_sched() takes the earlier self-bypass branch and returns before reaching these host paths; that mode is accounted for by SCX_EV_BYPASS_DISPATCH at enqueue time and is intentionally left unchanged. Fixes: 025b1bd41965 ("sched_ext: Implement hierarchical bypass mode") Signed-off-by: Liang Luo Signed-off-by: Tejun Heo --- kernel/sched/ext/inlines.h | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/kernel/sched/ext/inlines.h b/kernel/sched/ext/inlines.h index ed423bcc26b8..2ff5479334cb 100644 --- a/kernel/sched/ext/inlines.h +++ b/kernel/sched/ext/inlines.h @@ -129,8 +129,10 @@ scx_dispatch_sched(struct scx_sched *sch, struct rq *rq, * scheduler's ops.dispatch() doesn't yield any tasks. */ if (scx_bypass_dsp_enabled(sch) && - scx_consume_dispatch_q(sch, rq, scx_bypass_dsq(sch, cpu), 0)) + scx_consume_dispatch_q(sch, rq, scx_bypass_dsq(sch, cpu), 0)) { + __scx_add_event(sch, SCX_EV_SUB_BYPASS_DISPATCH, 1); return SCX_DSP_LOCAL; + } return SCX_DSP_NONE; } From b8d1d5b63a8ef532038eebd9d97d406860385668 Mon Sep 17 00:00:00 2001 From: "Masami Hiramatsu (Google)" Date: Tue, 22 Sep 2026 13:24:55 +0900 Subject: [PATCH 247/365] x86/mce: Fix hardware debug register corruption on task migration In exc_machine_check_user(), local_db_save() and local_db_restore() are invoked in the outer entry stubs (DEFINE_IDTENTRY_MCE_USER, DEFINE_FREDENTRY_MCE, and DEFINE_IDTENTRY_RAW), surrounding exc_machine_check_user(). However, exc_machine_check_user() calls irqentry_exit_to_user_mode(), which handles pending thread work and may schedule() if TIF_NEED_RESCHED is set. If the task migrates to another CPU during schedule(), local_db_restore() runs on the new CPU with the dr7 state saved from the old CPU. This corrupts the new CPU's DR7 hardware debug register and leaves the old CPU's DR7 disabled. In short, local_db_save() and local_db_restore() pair must be run on the same CPU. To fix this, move local_db_save() and local_db_restore() inside exc_machine_check_user() and exc_machine_check_kernel(). In exc_machine_check_user(), DR7 is saved and restored strictly around do_machine_check() to avoid schedule() during migration. In exc_machine_check_kernel(), local_db_save() is called at the entry point to prevent early memory accesses from triggering nested #DB exceptions, and restored on all exits. Fixes: cd840e424f27 ("x86/entry, mce: Disallow #DB during #MC") Assisted-by: LLM Signed-off-by: Masami Hiramatsu (Google) Signed-off-by: Borislav Petkov (AMD) Acked-by: Peter Zijlstra (Intel) Cc: Link: https://patch.msgid.link/179005109564.388919.3937970081044095776.stgit@devnote2 --- arch/x86/kernel/cpu/mce/core.c | 27 ++++++++++----------------- 1 file changed, 10 insertions(+), 17 deletions(-) diff --git a/arch/x86/kernel/cpu/mce/core.c b/arch/x86/kernel/cpu/mce/core.c index ab469605fc89..39f238952e14 100644 --- a/arch/x86/kernel/cpu/mce/core.c +++ b/arch/x86/kernel/cpu/mce/core.c @@ -2108,6 +2108,9 @@ bool filter_mce(struct mce *m) static __always_inline void exc_machine_check_kernel(struct pt_regs *regs) { irqentry_state_t irq_state; + unsigned long dr7; + + dr7 = local_db_save(); WARN_ON_ONCE(user_mode(regs)); @@ -2116,20 +2119,26 @@ static __always_inline void exc_machine_check_kernel(struct pt_regs *regs) * mce_check_crashing_cpu() for details. */ if (mca_cfg.initialized && mce_check_crashing_cpu()) - return; + goto out; irq_state = irqentry_nmi_enter(regs); do_machine_check(regs); irqentry_nmi_exit(regs, irq_state); +out: + local_db_restore(dr7); } static __always_inline void exc_machine_check_user(struct pt_regs *regs) { + unsigned long dr7; + irqentry_enter_from_user_mode(regs); + dr7 = local_db_save(); do_machine_check(regs); + local_db_restore(dr7); irqentry_exit_to_user_mode(regs); } @@ -2138,21 +2147,13 @@ static __always_inline void exc_machine_check_user(struct pt_regs *regs) /* MCE hit kernel mode */ DEFINE_IDTENTRY_MCE(exc_machine_check) { - unsigned long dr7; - - dr7 = local_db_save(); exc_machine_check_kernel(regs); - local_db_restore(dr7); } /* The user mode variant. */ DEFINE_IDTENTRY_MCE_USER(exc_machine_check) { - unsigned long dr7; - - dr7 = local_db_save(); exc_machine_check_user(regs); - local_db_restore(dr7); } #ifdef CONFIG_X86_FRED @@ -2169,28 +2170,20 @@ DEFINE_IDTENTRY_MCE_USER(exc_machine_check) */ DEFINE_FREDENTRY_MCE(exc_machine_check) { - unsigned long dr7; - - dr7 = local_db_save(); if (user_mode(regs)) exc_machine_check_user(regs); else exc_machine_check_kernel(regs); - local_db_restore(dr7); } #endif #else /* 32bit unified entry point */ DEFINE_IDTENTRY_RAW(exc_machine_check) { - unsigned long dr7; - - dr7 = local_db_save(); if (user_mode(regs)) exc_machine_check_user(regs); else exc_machine_check_kernel(regs); - local_db_restore(dr7); } #endif From 4fde448225123442c5796f54b7a4400e2d3cbaf6 Mon Sep 17 00:00:00 2001 From: Mario Limonciello Date: Tue, 8 Sep 2026 14:05:59 -0500 Subject: [PATCH 248/365] x86/PCI: Disable enhanced atomics on AMD NBIO 7.7 and 7.11 Multiple users report data corruption during 64-bit DMA transfers on systems with AMD NBIO 7.7 and 7.11 controllers. This occurs when BIOS enables AMD "enhanced atomic operations" on PCIe Root Ports. When enhanced atomics are enabled, any 64-bit DMA access may be corrupted. Disable enhanced atomics using SMN for NBIO 7.7 and 7.11 based models. Reported-by: Mikael Etienne Closes: https://lore.kernel.org/178789300872.392066.15963676631650361573@gmail.com/ Reported-by: Arthur Husband Closes: https://lore.kernel.org/20260406222335.379935-1-artmoty@gmail.com/ Reported-by: Alvin Lim Closes: https://lore.kernel.org/20260621100844.1224301-1-alvinwylim@gmail.com/ Signed-off-by: Mario Limonciello [bhelgaas: commit log, s/IOVA/DMA/ in comment] Signed-off-by: Bjorn Helgaas Cc: stable@vger.kernel.org Cc: David Laight Cc: John Smith Cc: Lennert Buytenhek Cc: Niklas Cassel Cc: Roland Waltersson Link: https://patch.msgid.link/20260908190600.226485-2-mario.limonciello@amd.com --- arch/x86/pci/fixup.c | 99 ++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 99 insertions(+) diff --git a/arch/x86/pci/fixup.c b/arch/x86/pci/fixup.c index b301c6c8df75..795d81c7a4de 100644 --- a/arch/x86/pci/fixup.c +++ b/arch/x86/pci/fixup.c @@ -886,6 +886,105 @@ static void quirk_clear_strap_no_soft_reset_dev2_f0(struct pci_dev *dev) } } DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_AMD, 0x15b8, quirk_clear_strap_no_soft_reset_dev2_f0); + +/* + * Enhanced atomic operations can cause corruption with 64-bit DMA + * on these devices. + */ +#define RX_ENH_ATOMIC_EN BIT(8) + +static const u32 nbio_7_7_pcie_smn_addrs[] = { + 0x111401d0, + 0x111411d0, + 0x111421d0, + 0x111431d0, + 0x111441d0, + 0x112401d0, + 0x112411d0, + 0x112421d0, + 0x112431d0, + 0x112441d0, + 0x112451d0, + 0x113401d0, + 0x114401d0, +}; + +static const u32 nbio_7_11_pcie_smn_addrs[] = { + 0x112401d0, + 0x112411d0, + 0x112421d0, + 0x112431d0, + 0x112441d0, + 0x112451d0, + 0x113401d0, + 0x113411d0, + 0x113421d0, + 0x113431d0, + 0x113441d0, + 0x113451d0, +}; + +static void quirk_amd_nbio_enhanced_atomic(struct pci_dev *host_bridge, + const u32 *smn_addrs, + size_t nr_smn_addrs) +{ + bool changed = false; + size_t i; + u32 data; + int ret; + + for (i = 0; i < nr_smn_addrs; i++) { + ret = amd_smn_read(0, smn_addrs[i], &data); + if (ret) + continue; + if (!(data & RX_ENH_ATOMIC_EN)) + continue; + data = data & ~RX_ENH_ATOMIC_EN; + ret = amd_smn_write(0, smn_addrs[i], data); + if (ret) + continue; + if (changed) + continue; + ret = amd_smn_read(0, smn_addrs[i], &data); + if (ret) + continue; + if (data & RX_ENH_ATOMIC_EN) + continue; + changed = true; + } + + if (changed) + pci_info(host_bridge, "enhanced atomics disabled\n"); +} + +static void quirk_amd_nbio_7_7_disable_enhanced_atomic(struct pci_dev *dev) +{ + quirk_amd_nbio_enhanced_atomic(dev, nbio_7_7_pcie_smn_addrs, + ARRAY_SIZE(nbio_7_7_pcie_smn_addrs)); +} + +static void quirk_amd_nbio_7_11_disable_enhanced_atomic(struct pci_dev *dev) +{ + quirk_amd_nbio_enhanced_atomic(dev, nbio_7_11_pcie_smn_addrs, + ARRAY_SIZE(nbio_7_11_pcie_smn_addrs)); +} + +/* Phoenix, Hawk Point (NBIO 7.7) */ +DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_AMD, 0x14E8, + quirk_amd_nbio_7_7_disable_enhanced_atomic); +DECLARE_PCI_FIXUP_RESUME(PCI_VENDOR_ID_AMD, 0x14E8, + quirk_amd_nbio_7_7_disable_enhanced_atomic); + +/* Strix, Krackan, Strix Halo (NBIO 7.11) */ +DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_AMD, 0x1507, + quirk_amd_nbio_7_11_disable_enhanced_atomic); +DECLARE_PCI_FIXUP_RESUME(PCI_VENDOR_ID_AMD, 0x1507, + quirk_amd_nbio_7_11_disable_enhanced_atomic); +DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_AMD, 0x1122, + quirk_amd_nbio_7_11_disable_enhanced_atomic); +DECLARE_PCI_FIXUP_RESUME(PCI_VENDOR_ID_AMD, 0x1122, + quirk_amd_nbio_7_11_disable_enhanced_atomic); + #endif /* From 80320b278fea07ffcda3f57b67b61658e0a4e1ca Mon Sep 17 00:00:00 2001 From: Matthias Goergens Date: Thu, 24 Sep 2026 01:52:03 +0800 Subject: [PATCH 249/365] ata: libata-scsi: bound the ATA passthru sense descriptor writes When an ATA PASS-THROUGH command to an ATAPI device fails, the sense buffer holds the device's REQUEST SENSE reply, and ata_scsi_set_passthru_sense_fields() trusts its additional length byte, sb[7], when adding the ATA Status Return descriptor. A faulty or malicious device can use that to make the kernel read and write past the 96-byte buffer in three ways: - scsi_sense_desc_find() is passed sb[7] + 8 as the buffer length, so its clamp against sb[7] does nothing and the walk runs off the end. - A type-9 descriptor found near the end is filled in unchecked. - A new descriptor at sb[8 + len] needs len + 22 bytes, not len + 14, so len 75..82 writes up to 8 bytes past the end. Reproduced with KASAN under qemu, with the emulated ATAPI REQUEST SENSE reply patched: BUG: KASAN: slab-out-of-bounds in scsi_sense_desc_find+0x1a5/0x210 BUG: KASAN: slab-out-of-bounds in ata_scsi_qc_complete+0x1a15/0x1a50 Both are gone with this patch, and a valid descriptor is still filled in. Fixes: 97981926224a ("ata: libata-scsi: Do not overwrite valid sense data when CK_COND=1") Cc: stable@vger.kernel.org Reviewed-by: Damien Le Moal Signed-off-by: Matthias Goergens Link: https://lore.kernel.org/r/20260923175203.1576825-1-matthias.goergens@gmail.com Signed-off-by: Niklas Cassel --- drivers/ata/libata-scsi.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) diff --git a/drivers/ata/libata-scsi.c b/drivers/ata/libata-scsi.c index 7e22bbc38238..8f9aa97a519d 100644 --- a/drivers/ata/libata-scsi.c +++ b/drivers/ata/libata-scsi.c @@ -261,12 +261,18 @@ static void ata_scsi_set_passthru_sense_fields(struct ata_queued_cmd *qc) /* descriptor format */ len = sb[7]; - desc = (char *)scsi_sense_desc_find(sb, len + 8, 9); + desc = (char *)scsi_sense_desc_find(sb, SCSI_SENSE_BUFFERSIZE, 9); if (!desc) { - if (SCSI_SENSE_BUFFERSIZE < len + 14) + /* + * The descriptor is written at sb[8 + len] and is 14 + * bytes long, so it needs len + 22 bytes of buffer. + */ + if (len + 22 > SCSI_SENSE_BUFFERSIZE) return; sb[7] = len + 14; desc = sb + 8 + len; + } else if (desc - sb > SCSI_SENSE_BUFFERSIZE - 14) { + return; } desc[0] = 9; desc[1] = 12; From c5fd4eaad50d620c7e09ac2082b2fb55ee54170e Mon Sep 17 00:00:00 2001 From: Wentao Liang Date: Wed, 16 Sep 2026 09:49:56 +0000 Subject: [PATCH 250/365] drm/amd/display: Fix dc stream excess put in dm_update_crtc_state() In dm_update_crtc_state(), when a modeset is required the newly created stream is stored in dm_new_crtc_state->stream and an extra reference is taken with dc_stream_retain(). The reference returned by create_validate_stream_for_sink() is released as an extra reference at the skip_modeset label, leaving the stream owned by the new CRTC state. If amdgpu_dm_check_crtc_color_mgmt() fails afterwards, the code jumps to the fail label which releases new_stream again. Since the extra reference was already released at skip_modeset, this drops the reference owned by dm_new_crtc_state->stream and the stream is released while the atomic state still points to it, leading to a premature free of the dc stream. Set new_stream to NULL after releasing the extra reference at the skip_modeset label so that a later goto fail cannot release the reference owned by the new CRTC state. Fixes: 7cd4b70091a5 ("drm/amd/display: Rework CRTC color management") Signed-off-by: Wentao Liang Signed-off-by: Alex Deucher (cherry picked from commit 102a47065a62dc8f6bbbb47cf082a2934282eb08) Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c index 36d2f86f000a..91fdf3de7202 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -5583,8 +5583,10 @@ static int dm_update_crtc_state(struct amdgpu_display_manager *dm, skip_modeset: /* Release extra reference */ - if (new_stream) + if (new_stream) { dc_stream_release(new_stream); + new_stream = NULL; + } new_stream = NULL; /* From 3022bdfe3e6d776e9273d6892f7c193138ca0666 Mon Sep 17 00:00:00 2001 From: Prike Liang Date: Fri, 31 Jul 2026 11:44:37 +0800 Subject: [PATCH 251/365] drm/amdgpu: move userq fence wait out of signalling section The eviction fence suspend worker waits for every pending userq fence from inside a dma_fence_begin_signalling() critical section. Waiting on another DMA fence while responsible for signalling one violates the cross-driver fence contract and is reported by lockdep as a dma_fence_map dependency. Move the wait before dma_fence_begin_signalling(). Keep userq_mutex held so queue lifetime remains stable while inspecting last_fence. Fixes: fc61df151617 ("drm/amdgpu: annotate eviction fence signaling path") Signed-off-by: Prike Liang Reviewed-by: Vitaly Prosyak Signed-off-by: Alex Deucher (cherry picked from commit 3bd4fbc5ed89621340b5cd249869092691a9c81f) Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.c | 3 +++ drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c | 4 +--- drivers/gpu/drm/amd/amdgpu/amdgpu_userq.h | 1 + 3 files changed, 5 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.c index f6b7522c3c82..f8652fd0525d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.c @@ -68,6 +68,9 @@ amdgpu_eviction_fence_suspend_worker(struct work_struct *work) mutex_lock(&uq_mgr->userq_mutex); + /* Fence waits are not allowed in a fence signalling critical section. */ + amdgpu_userq_wait_for_signal(uq_mgr); + /* * This is intentionally after taking the userq_mutex since we do * allocate memory while holding this lock, but only after ensuring that diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c index e43bda0cab3f..280bdeb38970 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c @@ -1272,7 +1272,7 @@ amdgpu_userq_evict_all(struct amdgpu_userq_mgr *uq_mgr) return ret; } -static void +void amdgpu_userq_wait_for_signal(struct amdgpu_userq_mgr *uq_mgr) { struct amdgpu_usermode_queue *queue; @@ -1291,8 +1291,6 @@ amdgpu_userq_wait_for_signal(struct amdgpu_userq_mgr *uq_mgr) void amdgpu_userq_evict(struct amdgpu_userq_mgr *uq_mgr) { - /* Wait for any pending userqueue fence work to finish */ - amdgpu_userq_wait_for_signal(uq_mgr); amdgpu_userq_evict_all(uq_mgr); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.h index 6412a7f7b6ef..488dc21d7c81 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.h @@ -162,6 +162,7 @@ void amdgpu_userq_mgr_cancel_reset_work(struct amdgpu_device *adev); void amdgpu_userq_mgr_cancel_resume(struct amdgpu_userq_mgr *userq_mgr); void amdgpu_userq_mgr_fini(struct amdgpu_userq_mgr *userq_mgr); +void amdgpu_userq_wait_for_signal(struct amdgpu_userq_mgr *uq_mgr); void amdgpu_userq_evict(struct amdgpu_userq_mgr *uq_mgr); void amdgpu_userq_ensure_ev_fence(struct amdgpu_userq_mgr *userq_mgr, From cd195f1616b2bb5fb7765465326c4d5d64a620a0 Mon Sep 17 00:00:00 2001 From: Sunil Khatri Date: Thu, 17 Sep 2026 18:56:45 +0530 Subject: [PATCH 252/365] drm/amdgpu/userq: fix double jiffies conversion in hang detect timeout MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Function amdgpu_userq_start_hang_detect_work() calls msecs_to_jiffies() on adev->gfx_timeout/compute_timeout/sdma_timeout before arming hang_detect_work. These timeout values already hold jiffies values from amdgpu_device_get_job_timeout_settings() at device init. This silently shrinks the real hang-detect deadline to (2 * HZ) ms instead of the intended timeout. e.g. 500ms instead of the 2000ms default on a CONFIG_HZ=250 kernel, only coincidentally correct at HZ=1000. The shortened window is easily exceeded by ordinary fence-completion latency, causing hang_detect_work to fire and trigger a per-queue or full GPU reset for queues that are not actually hung. Pass the jiffies value directly to queue_delayed_work() instead of converting it a second time. Fixes: fc3336be9c62 ("drm/amd/amdgpu: Add independent hang detect work for user queue fence") Signed-off-by: Sunil Khatri Reviewed-by: Christian König Signed-off-by: Alex Deucher (cherry picked from commit 13d44ca033cb74756c2aef0ade54a75cdf2f6271) Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c index 280bdeb38970..cc8e7af18834 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c @@ -184,27 +184,27 @@ static void amdgpu_userq_hang_detect_work(struct work_struct *work) void amdgpu_userq_start_hang_detect_work(struct amdgpu_usermode_queue *queue) { struct amdgpu_device *adev; - unsigned long timeout_ms; + unsigned long timeout_jiffies; adev = queue->userq_mgr->adev; /* Determine timeout based on queue type */ switch (queue->queue_type) { case AMDGPU_RING_TYPE_GFX: - timeout_ms = adev->gfx_timeout; + timeout_jiffies = adev->gfx_timeout; break; case AMDGPU_RING_TYPE_COMPUTE: - timeout_ms = adev->compute_timeout; + timeout_jiffies = adev->compute_timeout; break; case AMDGPU_RING_TYPE_SDMA: - timeout_ms = adev->sdma_timeout; + timeout_jiffies = adev->sdma_timeout; break; default: - timeout_ms = adev->gfx_timeout; + timeout_jiffies = adev->gfx_timeout; break; } queue_delayed_work(adev->reset_domain->wq, &queue->hang_detect_work, - msecs_to_jiffies(timeout_ms)); + timeout_jiffies); } void amdgpu_userq_process_fence_irq(struct amdgpu_device *adev, u32 doorbell) From f952ed353a27b46c86c9525a39b7a642850b8139 Mon Sep 17 00:00:00 2001 From: Sunil Khatri Date: Thu, 17 Sep 2026 18:56:45 +0530 Subject: [PATCH 253/365] drm/amdgpu/vcn5.0.1: fix video_timeout unit mismatch in jpeg reset wait MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit vcn_v5_0_1_reset_jpeg_pre_helper() passes adev->video_timeout directly to amdgpu_fence_wait_polling(), whose timeout parameter is documented and implemented in usecs (busy-wait loop decrementing by udelay(2)). adev->video_timeout is set in jiffies by amdgpu_device_get_job_timeout_settings(), via msecs_to_jiffies(). Passing it unconverted means the intended ~2s wait for outstanding JPEG fences to complete before the JPEG queue is torn down actually lasts only a couple of microseconds (HZ jiffies interpreted as usecs), so pending jobs are almost never given a real chance to finish before the reset path forces completion in the following helper. Convert the jiffies value to usecs with jiffies_to_usecs() before passing it to amdgpu_fence_wait_polling(). Fixes: fab47d2db5ca ("drm/amdgpu/vcn5.0.1: rework reset handling") Cc: Jesse.Zhang Assisted-by: Claude:claude-sonnet-5 Signed-off-by: Sunil Khatri Reviewed-by: Christian König Signed-off-by: Alex Deucher (cherry picked from commit b8334fec8b90ebffcaa01001a23edca9f29a05e9) Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.c b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.c index 1a07c3bf4425..011afc0fdc87 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.c @@ -1335,7 +1335,8 @@ static int vcn_v5_0_1_reset_jpeg_pre_helper(struct amdgpu_device *adev, int inst /* if Jobs are still pending after timeout, * We'll handle them in the bottom helper */ - amdgpu_fence_wait_polling(ring, wait_seq, adev->video_timeout); + amdgpu_fence_wait_polling(ring, wait_seq, + jiffies_to_usecs(adev->video_timeout)); } return 0; From 6b13ddbf5bb8deec337f9b4887f579097a953f6a Mon Sep 17 00:00:00 2001 From: Sunil Khatri Date: Thu, 17 Sep 2026 18:56:45 +0530 Subject: [PATCH 254/365] drm/amdgpu/vcn4.0.3: fix video_timeout unit mismatch in jpeg reset wait MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit vcn_v4_0_3_reset_jpeg_pre_helper() passes adev->video_timeout directly to amdgpu_fence_wait_polling(), whose timeout parameter is documented and implemented in usecs (busy-wait loop decrementing by udelay(2)). adev->video_timeout is set in jiffies by amdgpu_device_get_job_timeout_settings(), via msecs_to_jiffies(). Passing it unconverted means the intended ~2s wait for outstanding JPEG fences to complete before the JPEG queue is torn down actually lasts only a couple of microseconds (HZ jiffies interpreted as usecs), so pending jobs are almost never given a real chance to finish before the reset path forces completion in the following helper. Convert the jiffies value to usecs with jiffies_to_usecs() before passing it to amdgpu_fence_wait_polling(). Fixes: d25c67fd9d6f ("drm/amdgpu/vcn4.0.3: rework reset handling") Cc: Jesse.Zhang Assisted-by: Claude:claude-sonnet-5 Signed-off-by: Sunil Khatri Reviewed-by: Christian König Signed-off-by: Alex Deucher (cherry picked from commit 5feabbd673c10ebee22b880e4d812f08974d2ef7) Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c index 179b892fb410..62e2e04314dc 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c @@ -1689,7 +1689,8 @@ static int vcn_v4_0_3_reset_jpeg_pre_helper(struct amdgpu_device *adev, int inst /* if Jobs are still pending after timeout, * We'll handle them in the bottom helper */ - amdgpu_fence_wait_polling(ring, wait_seq, adev->video_timeout); + amdgpu_fence_wait_polling(ring, wait_seq, + jiffies_to_usecs(adev->video_timeout)); } return 0; From c3a31087b1c8df679b653c1a09d9abe5ff7ec8ef Mon Sep 17 00:00:00 2001 From: Asad Kamal Date: Fri, 28 Aug 2026 15:59:21 +0800 Subject: [PATCH 255/365] drm/amdkfd: fix use-after-free and multi-container gap in kfd_dev_mapping kfd_dev_mapping caches the address_space of the first /dev/kfd opener so that the GPU reset path can call unmap_mapping_range() to zap all userspace mappings of doorbell and MMIO ranges. This design has two bugs that both manifest under SRIOV with multiple containers: 1. Use-after-free / rwsem deadlock. The cached pointer refers to an inode owned by the first opener's container. When that container exits and its inode is released, kfd_dev_mapping becomes a dangling pointer. A subsequent GPU reset dereferences it inside unmap_mapping_range(), which takes i_mmap_rwsem on the freed inode, causing a hard hang observable as an uninterruptible rwsem wait. 2. Multi-container gap. Only the first opener's address_space is cached; VMAs created by later openers live in a different address_space and are never reached by unmap_mapping_range(). After a GPU reset those stale mappings keep doorbell and MMIO pages accessible to guest userspace with no GPU behind them, risking PCIe transaction timeouts and NMI panics. Fix both bugs with the same approach used by DRM core (drm_drv.c): create a private pseudo-filesystem at module init time and allocate one anonymous inode from it. In kfd_open() redirect every opener's file->f_mapping to that inode's address_space. The inode is module-owned, lives exactly as long as the amdgpu module, and collects VMAs from all openers in one address_space. A single unmap_mapping_range() call in the reset path then correctly reaches every container's mappings with no dangling pointer risk. The hang manifests as an NMI backtrace on the GPU reset workqueue stuck spinning in rwsem_down_read_slowpath() with a corrupted i_mmap_rwsem: Workqueue: amdgpu-reset-dev xgpu_ai_mailbox_flr_work [amdgpu] Call Trace: kvm_wait+0x1f/0x40 __pv_queued_spin_lock_slowpath+0x31d/0x3a0 _raw_spin_lock_irq+0x51/0x80 rwsem_down_read_slowpath+0xb3/0x550 down_read+0x48/0xd0 unmap_mapping_range+0x71/0x140 kfd_dev_unmap_mapping_range+0x5b/0x140 [amdgpu] amdgpu_amdkfd_clear_kfd_mapping+0xd8/0x190 [amdgpu] amdgpu_device_gpu_recover+0x232/0x450 [amdgpu] xgpu_ai_mailbox_flr_work+0xb5/0xc0 [amdgpu] process_one_work+0x18e/0x3e0 worker_thread+0x2e3/0x420 kthread+0x10a/0x230 Fixes: 70cadefcc616 ("drm/amdgpu: unmap all user mappings of framebuffer and doorbell before mode1 reset") Signed-off-by: Asad Kamal Reviewed-by: Lijo Lazar Signed-off-by: Alex Deucher (cherry picked from commit 1128b4a52de1572e87431de837fd9850cb99542c) Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdkfd/kfd_chardev.c | 71 +++++++++++++++++++----- 1 file changed, 57 insertions(+), 14 deletions(-) diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c index 504a286368eb..344da6c0e96a 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c @@ -35,6 +35,7 @@ #include #include #include +#include #include #include #include @@ -70,18 +71,54 @@ static const struct class kfd_class = { }; /* - * Cache the address space of the chardev on first open so that the reset - * path can drop all userspace mappings of doorbell and MMIO ranges via - * unmap_mapping_range(). + * Private pseudo-filesystem for KFD, Provides a stable, module-owned + * inode whose address_space is the unmap target for all /dev/kfd + * openers during GPU reset. */ -static struct address_space *kfd_dev_mapping; +static struct vfsmount *kfd_fs_mnt; +static int kfd_fs_cnt; + +static int kfd_fs_init_fs_context(struct fs_context *fc) +{ + return init_pseudo(fc, 0x4b464400 /* "KFD" */) ? 0 : -ENOMEM; +} + +static struct file_system_type kfd_fs_type = { + .name = "kfd", + .init_fs_context = kfd_fs_init_fs_context, + .kill_sb = kill_anon_super, +}; + +static struct inode *kfd_fs_inode_new(void) +{ + struct inode *inode; + int r; + + r = simple_pin_fs(&kfd_fs_type, &kfd_fs_mnt, &kfd_fs_cnt); + if (r < 0) + return ERR_PTR(r); + + inode = alloc_anon_inode(kfd_fs_mnt->mnt_sb); + if (IS_ERR(inode)) + simple_release_fs(&kfd_fs_mnt, &kfd_fs_cnt); + + return inode; +} + +static void kfd_fs_inode_free(struct inode *inode) +{ + if (inode) { + iput(inode); + simple_release_fs(&kfd_fs_mnt, &kfd_fs_cnt); + } +} + +static struct inode *kfd_anon_inode; void kfd_dev_unmap_mapping_range(loff_t const holebegin, loff_t const holelen) { - struct address_space *mapping = READ_ONCE(kfd_dev_mapping); - - if (mapping) - unmap_mapping_range(mapping, holebegin, holelen, 1); + if (kfd_anon_inode) + unmap_mapping_range(kfd_anon_inode->i_mapping, holebegin, holelen, 1); } static inline struct kfd_process_device *kfd_lock_pdd_by_id(struct kfd_process *p, __u32 gpu_id) @@ -107,6 +144,13 @@ int kfd_chardev_init(void) { int err = 0; + kfd_anon_inode = kfd_fs_inode_new(); + if (IS_ERR(kfd_anon_inode)) { + err = PTR_ERR(kfd_anon_inode); + kfd_anon_inode = NULL; + return err; + } + kfd_char_dev_major = register_chrdev(0, kfd_dev_name, &kfd_fops); err = kfd_char_dev_major; if (err < 0) @@ -130,6 +174,8 @@ int kfd_chardev_init(void) err_class_create: unregister_chrdev(kfd_char_dev_major, kfd_dev_name); err_register_chrdev: + kfd_fs_inode_free(kfd_anon_inode); + kfd_anon_inode = NULL; return err; } @@ -138,6 +184,8 @@ void kfd_chardev_exit(void) device_destroy(&kfd_class, MKDEV(kfd_char_dev_major, 0)); class_unregister(&kfd_class); unregister_chrdev(kfd_char_dev_major, kfd_dev_name); + kfd_fs_inode_free(kfd_anon_inode); + kfd_anon_inode = NULL; kfd_device = NULL; } @@ -150,12 +198,7 @@ static int kfd_open(struct inode *inode, struct file *filep) if (iminor(inode) != 0) return -ENODEV; - /* - * /dev/kfd is a single chardev so all opens share one inode. Cache - * its address_space on the first open for use by the reset path. - */ - if (!READ_ONCE(kfd_dev_mapping)) - cmpxchg(&kfd_dev_mapping, NULL, inode->i_mapping); + filep->f_mapping = kfd_anon_inode->i_mapping; is_32bit_user_mode = in_compat_syscall(); From aea841bc62a76242396610d22d8ff40c13065f64 Mon Sep 17 00:00:00 2001 From: Wentao Liang Date: Wed, 16 Sep 2026 10:01:39 +0000 Subject: [PATCH 256/365] drm/amdgpu: Fix vmid_wait fence leak in amdgpu_ring_init() amdgpu_ring_init() initializes ring->vmid_wait with a reference to the stub fence taken via dma_fence_get_stub(). When a later step of the initialization fails, e.g. amdgpu_fence_driver_init_ring(), a writeback slot allocation or the ring buffer allocation, the function returns an error without releasing the stub fence reference and the reference is leaked if the ring is torn down without amdgpu_ring_fini(). Move the stub fence assignment to the end of the initialization, right before the ring is registered with the GPU scheduler, where no further failure is possible. The stub fence is only consumed by command submission handling in amdgpu_ids.c once the ring is up and running, so nothing reads it during the error-prone part of the initialization. Fixes: 48e9fbd1a284 ("drm/amdgpu: initialize the vmid_wait with the stub fence") Signed-off-by: Wentao Liang Signed-off-by: Alex Deucher (cherry picked from commit f2b96986851203e9c50ca0d13aaa3581ca3e8ebd) Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c index 686c92e96025..5922406d0a03 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c @@ -254,7 +254,6 @@ int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring, ring->adev = adev; ring->num_hw_submission = sched_hw_submission; ring->sched_score = sched_score; - ring->vmid_wait = dma_fence_get_stub(); ring->idx = adev->num_rings++; adev->rings[ring->idx] = ring; @@ -374,6 +373,7 @@ int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring, ring->max_dw = max_dw; ring->hw_prio = hw_prio; + ring->vmid_wait = dma_fence_get_stub(); if (!ring->no_scheduler && ring->funcs->type < AMDGPU_HW_IP_NUM) { hw_ip = ring->funcs->type; From a997baa61179b450bd55c4810c7ccfed3b753a54 Mon Sep 17 00:00:00 2001 From: Wentao Liang Date: Wed, 16 Sep 2026 09:54:32 +0000 Subject: [PATCH 257/365] drm/amdgpu: Fix acpi device leak in amdgpu_acpi_enumerate_xcc() amdgpu_acpi_enumerate_xcc() looks up each XCC ACPI device with acpi_dev_get_first_match_dev(), which takes a reference to the device. The reference is dropped with acpi_dev_put() after the XCC info is initialized, but if the kzalloc_obj() allocation of the XCC info fails the function returns -ENOMEM without releasing the reference, leaking the last reference to the ACPI device. Drop the ACPI device reference on the allocation failure path before returning. Fixes: 4d5275ab0b18 ("drm/amdgpu: Add parsing of acpi xcc objects") Reviewed-by: Lijo Lazar Signed-off-by: Wentao Liang Signed-off-by: Alex Deucher (cherry picked from commit 9211ef48b31ec66999cf55e04d0cbc60cd855fd5) Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c index 7f5abb03be1b..8b8acf98fdfe 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c @@ -1167,8 +1167,10 @@ int amdgpu_acpi_enumerate_xcc(void) } xcc_info = kzalloc_obj(struct amdgpu_acpi_xcc_info); - if (!xcc_info) + if (!xcc_info) { + acpi_dev_put(acpi_dev); return -ENOMEM; + } INIT_LIST_HEAD(&xcc_info->list); xcc_info->handle = acpi_device_handle(acpi_dev); From b4f7b4459b1b155e4c4977a6482b5df2cf08758c Mon Sep 17 00:00:00 2001 From: Wentao Liang Date: Wed, 16 Sep 2026 09:55:36 +0000 Subject: [PATCH 258/365] drm/amdgpu: Fix last_update fence leak in amdgpu_vm_init() amdgpu_vm_init() initializes vm->last_update, vm->last_unlocked and vm->last_tlb_flush with references to the stub fence taken via dma_fence_get_stub(). The error label at the end of the function releases the last_unlocked and last_tlb_flush references with dma_fence_put(), but the reference stored in vm->last_update is never dropped, so whenever the page table root creation, the reservation of the root BO or the PASID registration fails, the stub fence reference leaks. Drop the vm->last_update reference together with the other stub fence references on the error path. Fixes: 187916e6ed9d ("drm/amdgpu: install stub fence into potential unused fence pointers") Signed-off-by: Wentao Liang Signed-off-by: Alex Deucher (cherry picked from commit e7979c84fc05a176bdf855ee664871b1648404c9) Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index bb04101b0fb5..4a63b472f68e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -2678,6 +2678,7 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm, amdgpu_bo_unref(&root_bo); error_free_delayed: + dma_fence_put(vm->last_update); dma_fence_put(vm->last_tlb_flush); dma_fence_put(vm->last_unlocked); ttm_lru_bulk_move_fini(&adev->mman.bdev, &vm->lru_bulk_move); From 2b86ab1bd6673c525adda88819d7658ba9e784ec Mon Sep 17 00:00:00 2001 From: Wentao Liang Date: Wed, 16 Sep 2026 09:58:05 +0000 Subject: [PATCH 259/365] drm/amdgpu: Fix runtime PM leak in amdgpu_debugfs_test_ib_show() amdgpu_debugfs_test_ib_show() resumes the device with pm_runtime_get_sync() before taking the reset domain semaphore with down_write_killable(). If the write lock acquisition is interrupted, the function returns without calling pm_runtime_put_autosuspend(), leaking the runtime PM reference acquired for the device and keeping the GPU awake. Drop the runtime PM reference on the interrupted down_write_killable() error path before returning. Fixes: 6049db43d6dd ("drm/amdgpu: change reset lock from mutex to rw_semaphore") Signed-off-by: Wentao Liang Signed-off-by: Alex Deucher (cherry picked from commit ec30a576c2d4c0364549e6c04218f50704ef56c8) Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c index 132d054900b5..aca1a8045afa 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c @@ -1780,8 +1780,10 @@ static int amdgpu_debugfs_test_ib_show(struct seq_file *m, void *unused) /* Avoid accidently unparking the sched thread during GPU reset */ r = down_write_killable(&adev->reset_domain->sem); - if (r) + if (r) { + pm_runtime_put_autosuspend(dev->dev); return r; + } /* hold on the scheduler */ for (i = 0; i < AMDGPU_MAX_RINGS; i++) { From 0fd5e9ddf362b1253b3c94b858371f90400e5c4a Mon Sep 17 00:00:00 2001 From: Alex Hung Date: Wed, 23 Sep 2026 11:31:18 -0600 Subject: [PATCH 260/365] drm/amd/display: Relax DML frame limit with UBSAN [WHY] UBSAN instrumentation adds checks and handler calls and increases stack usage in the large DML calculation functions, similar to KASAN and KCSAN. With UBSAN enabled these files exceed the default -Wframe-larger-than limit and fail to build when -Werror is in effect. Reproduced with LLVM (make LLVM=1, clang 19.1.1), CONFIG_UBSAN=y, CONFIG_GCOV_PROFILE_ALL=y and CONFIG_DRM_AMDGPU_WERROR=y on x86_64. [HOW] Include CONFIG_UBSAN in the sanitizer check that selects the higher per-file frame warning limit in the dml and dml2_0 Makefiles. Suggested-by: Leo Li Assisted-by: Copilot:Claude-Opus-5.5 Signed-off-by: Alex Hung Signed-off-by: Alex Deucher (cherry picked from commit ebf8b0fd8508b744f85a8eee82b745b1d3502dd0) Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/display/dc/dml/Makefile | 2 +- drivers/gpu/drm/amd/display/dc/dml2_0/Makefile | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/dml/Makefile b/drivers/gpu/drm/amd/display/dc/dml/Makefile index 10d4ace04d4f..096520bd27c6 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/Makefile +++ b/drivers/gpu/drm/amd/display/dc/dml/Makefile @@ -29,7 +29,7 @@ dml_ccflags := $(CC_FLAGS_FPU) dml_rcflags := $(CC_FLAGS_NO_FPU) ifneq ($(CONFIG_FRAME_WARN),0) - ifeq ($(filter y,$(CONFIG_KASAN)$(CONFIG_KCSAN)),y) + ifeq ($(filter y,$(CONFIG_KASAN)$(CONFIG_KCSAN)$(CONFIG_UBSAN)),y) ifeq ($(CONFIG_CC_IS_CLANG)$(CONFIG_COMPILE_TEST),yy) frame_warn_limit := 4096 else diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/Makefile b/drivers/gpu/drm/amd/display/dc/dml2_0/Makefile index 5388bf094fbc..4d682e92df1d 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2_0/Makefile +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/Makefile @@ -28,7 +28,7 @@ dml2_ccflags := $(CC_FLAGS_FPU) dml2_rcflags := $(CC_FLAGS_NO_FPU) ifneq ($(CONFIG_FRAME_WARN),0) - ifeq ($(filter y,$(CONFIG_KASAN)$(CONFIG_KCSAN)),y) + ifeq ($(filter y,$(CONFIG_KASAN)$(CONFIG_KCSAN)$(CONFIG_UBSAN)),y) ifeq ($(CONFIG_CC_IS_CLANG)$(CONFIG_COMPILE_TEST),yy) frame_warn_limit := 4096 else From 18779dd84515db093fedb4ebaf0998c9b165a5fb Mon Sep 17 00:00:00 2001 From: Ivan Lipski Date: Fri, 21 Aug 2026 00:04:53 -0400 Subject: [PATCH 261/365] drm/amd/display: Bump frame warning limit for clang builds of dml [Why&How] When building the DML files with clang without any sanitizer or LTO, the following -Wframe-larger-than errors break the build under CONFIG_WERROR: display_mode_vba_30.c: error: stack frame size (2512) exceeds limit (2048) in 'dml30_ModeSupportAndSystemConfigurationFull' display_mode_vba_31.c: error: stack frame size (2416) exceeds limit (2048) in 'dml31_ModeSupportAndSystemConfigurationFull' display_mode_vba_314.c: error: stack frame size (2392) exceeds limit (2048) in 'dml314_ModeSupportAndSystemConfigurationFull' Clang consistently spills more than gcc, pushing the frame past the 2048 byte limit. Apply an existing approach of increasing the warn stack size to the non-sanitizer path so plain clang builds use a 3072 byte limit. Closes: https://gitlab.freedesktop.org/drm/amd/-/work_items/5642 Signed-off-by: Ivan Lipski Signed-off-by: Alex Deucher (cherry picked from commit 21711b6e66bb7b41b1aec67b2d99aafe768c8fcb) Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/display/dc/dml/Makefile | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/dc/dml/Makefile b/drivers/gpu/drm/amd/display/dc/dml/Makefile index 096520bd27c6..79eeb3721b99 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/Makefile +++ b/drivers/gpu/drm/amd/display/dc/dml/Makefile @@ -36,7 +36,11 @@ ifneq ($(CONFIG_FRAME_WARN),0) frame_warn_limit := 3072 endif else - frame_warn_limit := 2048 + ifeq ($(CONFIG_CC_IS_CLANG),y) + frame_warn_limit := 3072 + else + frame_warn_limit := 2048 + endif endif ifeq ($(call test-lt, $(CONFIG_FRAME_WARN), $(frame_warn_limit)),y) From a0bb6fac53fa7cf1cadb487b43d4c9276a6b82e3 Mon Sep 17 00:00:00 2001 From: Zhan Xusheng Date: Fri, 18 Sep 2026 21:29:15 +0800 Subject: [PATCH 262/365] sched/core: Account PSI IRQ time to the execution context, not the scheduling context psi_account_irqtime() has two callers which share rq->psi_irq_time, and they disagree about the context: __schedule() passes the outgoing rq->curr, sched_tick() passes rq->donor. Under proxy execution the donor is blocked on a mutex while rq->curr burns the CPU. The tick charges PSI_IRQ_FULL to the donor's cgroup and advances the timestamp, so the call from __schedule() then finds delta <= 0 and charges nothing. The delta is not counted twice, it lands on the wrong cgroup. Pass rq->curr, which is what the call read before commit af0c8b2bf67b ("sched: Split scheduler and execution contexts") renamed 'curr' to 'donor' across sched_tick(). Without CONFIG_SCHED_PROXY_EXEC the two rq members are a union, so this only changes anything where that option is set, and it depends on EXPERT. Fixes: af0c8b2bf67b ("sched: Split scheduler and execution contexts") Signed-off-by: Zhan Xusheng Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Link: https://patch.msgid.link/20260918132915.1236312-1-zhanxusheng@xiaomi.com --- kernel/sched/core.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 0b846a13c628..1fe40de6ebe3 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -5798,7 +5798,7 @@ void sched_tick(void) curr = rq->curr; donor = rq->donor; - psi_account_irqtime(rq, donor, NULL); + psi_account_irqtime(rq, curr, NULL); update_rq_clock(rq); hw_pressure = arch_scale_hw_pressure(cpu_of(rq)); From 7b824c293a6b56de8285a97984c507cba56bc4c4 Mon Sep 17 00:00:00 2001 From: Brajesh Gupta Date: Tue, 22 Sep 2026 09:56:55 +0530 Subject: [PATCH 263/365] drm/imagination: Propagate map failures correctly from pvr_mmu_map_sgl() Map failure from pvr_mmu_map_sgl() interface was not returned correctly to pvr_mmu_map() interface. This resulted in pvr_mmu_map() interface to continue instead of returning an error to caller. Fix it by returning a proper error code from pvr_mmu_map_sgl() interface. Call stack for crash: [ 1179.286237] Unable to handle kernel NULL pointer dereference at virtual address 0000000000000008 [ 1179.295067] Mem abort info: [ 1179.297877] ESR = 0x0000000096000004 [ 1179.301656] EC = 0x25: DABT (current EL), IL = 32 bits [ 1179.306987] SET = 0, FnV = 0 [ 1179.310048] EA = 0, S1PTW = 0 [ 1179.313198] FSC = 0x04: level 0 translation fault [ 1179.318096] Data abort info: [ 1179.320993] ISV = 0, ISS = 0x00000004, ISS2 = 0x00000000 [ 1179.326483] CM = 0, WnR = 0, TnD = 0, TagAccess = 0 [ 1179.331546] GCS = 0, Overlay = 0, DirtyBit = 0, Xs = 0 [ 1179.336895] user pgtable: 4k pages, 48-bit VAs, pgdp=000000009822a000 [ 1179.343402] [0000000000000008] pgd=0000000000000000, p4d=0000000000000000 [ 1179.350243] Internal error: Oops: 0000000096000004 [#2] SMP [ 1179.355908] Modules linked in: powervr gpu_sched drm_shmem_helper drm_gpuvm drm_exec xhci_plat_hcd xhci_hcd dwc3 usbcore usb_common snd_soc_simple_card snd_soc_simple_card_utils dwc3_am62 at24 sa2ul sha512 libsha512 sha256 authenc sch_fq_codel fuse dm_mod ipv6 [ 1179.378992] CPU: 1 UID: 1000 PID: 680 Comm: deqp-vk Tainted: G D 6.17.0 #1 PREEMPT [ 1179.388120] Tainted: [D]=DIE [ 1179.390994] Hardware name: Texas Instruments AM625 SK (DT) [ 1179.396467] pstate: 00000005 (nzcv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--) [ 1179.403415] pc : pvr_mmu_op_context_unmap_curr_page+0x6c/0x134 [powervr] [ 1179.410140] lr : pvr_mmu_op_context_unmap_curr_page+0x58/0x134 [powervr] [ 1179.416848] sp : ffff8000839ab8c0 [ 1179.420153] x29: ffff8000839ab8c0 x28: 0000000000000001 x27: 000000008f386000 [ 1179.427283] x26: ffff000016d1df98 x25: 0000000000247000 x24: 00000000000001e6 [ 1179.434413] x23: 0000000000000002 x22: 000000000000ffff x21: 0000000000000247 [ 1179.441540] x20: 0000000000000245 x19: ffff000016d1df60 x18: 0000000000000002 [ 1179.448668] x17: 0000000000000000 x16: 0000000000000000 x15: 0000000000000001 [ 1179.455793] x14: 0000000000060810 x13: ffff80007fffffff x12: ffff000004190480 [ 1179.462921] x11: ffff8000853f7000 x10: ffff8000811ae000 x9 : ffff0000041900b8 [ 1179.470051] x8 : 0000000000000000 x7 : 00000000990c4001 x6 : 0000000000000007 [ 1179.477177] x5 : ffff000016d1df60 x4 : 0000000000000000 x3 : ffff00000a7d8000 [ 1179.484306] x2 : 00000000000001ff x1 : 0000000000000000 x0 : 0000000000000000 [ 1179.491433] Call trace: [ 1179.493872] pvr_mmu_op_context_unmap_curr_page+0x6c/0x134 [powervr] (P) [ 1179.500582] pvr_mmu_map+0x31c/0x388 [powervr] [ 1179.505027] pvr_vm_gpuva_map+0x40/0x88 [powervr] [ 1179.509732] __drm_gpuvm_sm_map+0x250/0x44c [drm_gpuvm] [ 1179.514952] drm_gpuvm_sm_map+0x48/0x5c [drm_gpuvm] [ 1179.519822] pvr_vm_bind_op_exec+0x64/0x70 [powervr] [ 1179.524785] pvr_vm_map+0x1f8/0x2a8 [powervr] [ 1179.529142] pvr_ioctl_vm_map+0x12c/0x188 [powervr] [ 1179.534018] drm_ioctl_kernel+0xb8/0x128 [ 1179.537941] drm_ioctl+0x21c/0x4ec [ 1179.541337] __arm64_sys_ioctl+0xac/0x108 [ 1179.545344] invoke_syscall+0x44/0x100 [ 1179.549091] el0_svc_common.constprop.0+0x40/0xe0 [ 1179.553790] do_el0_svc+0x1c/0x28 [ 1179.557106] el0_svc+0x34/0xf0 [ 1179.560159] el0t_64_sync_handler+0xd0/0xe4 [ 1179.564334] el0t_64_sync+0x198/0x19c [ 1179.567996] Code: 54000300 35000360 f9402261 79409a62 (f9400421) [ 1179.574081] ---[ end trace 0000000000000000 ]--- Fixes: ff5f643de0bf ("drm/imagination: Add GEM and VM related code") Reviewed-by: Alexandru Dadu Reviewed-by: Alessio Belle Cc: stable@vger.kernel.org Link: https://patch.msgid.link/20260922-mmu_fix-v4-1-12f1a871456a@imgtec.com Signed-off-by: Brajesh Gupta --- drivers/gpu/drm/imagination/pvr_mmu.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/imagination/pvr_mmu.c b/drivers/gpu/drm/imagination/pvr_mmu.c index 3cac482e1034..23261d9ad3fd 100644 --- a/drivers/gpu/drm/imagination/pvr_mmu.c +++ b/drivers/gpu/drm/imagination/pvr_mmu.c @@ -12,6 +12,7 @@ #include "pvr_rogue_mmu_defs.h" #include +#include #include #include #include @@ -2553,7 +2554,9 @@ pvr_mmu_map_sgl(struct pvr_mmu_op_context *op_ctx, struct scatterlist *sgl, err_destroy_pages: memcpy(&op_ctx->curr_page, &ptr_copy, sizeof(op_ctx->curr_page)); - err = pvr_mmu_op_context_unmap_curr_page(op_ctx, page); + if (pvr_mmu_op_context_unmap_curr_page(op_ctx, page)) + drm_err(from_pvr_device(op_ctx->mmu_ctx->pvr_dev), + "%s : Failure in unmapping pages\n", __func__); return err; } From 0a8224058a5835297dcf4a46bbcd16f77a9fe424 Mon Sep 17 00:00:00 2001 From: Brajesh Gupta Date: Tue, 22 Sep 2026 09:56:56 +0530 Subject: [PATCH 264/365] drm/imagination: Fix page count for page table for map() interface The GPU virtual start address wasn't included in the calculation for the amount of page tables required for mapping a BO object in map() interface. It resulted in map failure later due to not enough pages at L0/L1 level. Update pvr_mmu_op_context_create() interface to pass device address as well to allow correct calculation for page table memory. If L0 tables cover 2MB (0x200000), the range defined by device address 0x80001ff000 (general heap at 2MB - 4KB) and size 0x2000 (two 4KB pages) requires two L0 pages to be mapped, but without the base address a range of 0x2000 computes to a single L0 page which is not enough. Fixes: ff5f643de0bf ("drm/imagination: Add GEM and VM related code") Reviewed-by: Alexandru Dadu Reviewed-by: Alessio Belle Cc: stable@vger.kernel.org Link: https://patch.msgid.link/20260922-mmu_fix-v4-2-12f1a871456a@imgtec.com Signed-off-by: Brajesh Gupta --- drivers/gpu/drm/imagination/pvr_mmu.c | 14 ++++++++------ drivers/gpu/drm/imagination/pvr_mmu.h | 2 +- drivers/gpu/drm/imagination/pvr_vm.c | 4 ++-- 3 files changed, 11 insertions(+), 9 deletions(-) diff --git a/drivers/gpu/drm/imagination/pvr_mmu.c b/drivers/gpu/drm/imagination/pvr_mmu.c index 23261d9ad3fd..62eae7fcd5a2 100644 --- a/drivers/gpu/drm/imagination/pvr_mmu.c +++ b/drivers/gpu/drm/imagination/pvr_mmu.c @@ -2336,6 +2336,7 @@ void pvr_mmu_op_context_destroy(struct pvr_mmu_op_context *op_ctx) * pvr_mmu_op_context_create() - Create an MMU op context. * @ctx: MMU context associated with owning VM context. * @sgt: Scatter gather table containing pages pinned for use by this context. + * @device_addr: Virtual device address at the start of the requested mapping. * @sgt_offset: Start offset of the requested device-virtual memory mapping. * @size: Size in bytes of the requested device-virtual memory mapping. For an * unmapping, this should be zero so that no page tables are allocated. @@ -2347,8 +2348,9 @@ void pvr_mmu_op_context_destroy(struct pvr_mmu_op_context *op_ctx) */ struct pvr_mmu_op_context * pvr_mmu_op_context_create(struct pvr_mmu_context *ctx, struct sg_table *sgt, - u64 sgt_offset, u64 size) + u64 device_addr, u64 sgt_offset, u64 size) { + u64 start_addr = device_addr + sgt_offset; int err; struct pvr_mmu_op_context *op_ctx = kzalloc_obj(*op_ctx); @@ -2364,16 +2366,16 @@ pvr_mmu_op_context_create(struct pvr_mmu_context *ctx, struct sg_table *sgt, if (size) { /* * The number of page table objects we need to prealloc is - * indicated by the mapping size, start offset and the sizes + * indicated by the mapping size, start address and the sizes * of the areas mapped per PT or PD. The range calculation is * identical to that for the index into a table for a device * address, so we reuse those functions here. */ - const u32 l1_start_idx = pvr_page_table_l2_idx(sgt_offset); - const u32 l1_end_idx = pvr_page_table_l2_idx(sgt_offset + size); + const u32 l1_start_idx = pvr_page_table_l2_idx(start_addr); + const u32 l1_end_idx = pvr_page_table_l2_idx(start_addr + size); const u32 l1_count = l1_end_idx - l1_start_idx + 1; - const u32 l0_start_idx = pvr_page_table_l1_idx(sgt_offset); - const u32 l0_end_idx = pvr_page_table_l1_idx(sgt_offset + size); + const u32 l0_start_idx = pvr_page_table_l1_idx(start_addr); + const u32 l0_end_idx = pvr_page_table_l1_idx(start_addr + size); const u32 l0_count = l0_end_idx - l0_start_idx + 1; /* diff --git a/drivers/gpu/drm/imagination/pvr_mmu.h b/drivers/gpu/drm/imagination/pvr_mmu.h index a8ecd460168d..2c02d61ba0a2 100644 --- a/drivers/gpu/drm/imagination/pvr_mmu.h +++ b/drivers/gpu/drm/imagination/pvr_mmu.h @@ -99,7 +99,7 @@ dma_addr_t pvr_mmu_get_root_table_dma_addr(struct pvr_mmu_context *ctx); void pvr_mmu_op_context_destroy(struct pvr_mmu_op_context *op_ctx); struct pvr_mmu_op_context * pvr_mmu_op_context_create(struct pvr_mmu_context *ctx, - struct sg_table *sgt, u64 sgt_offset, u64 size); + struct sg_table *sgt, u64 device_addr, u64 sgt_offset, u64 size); int pvr_mmu_map(struct pvr_mmu_op_context *op_ctx, u64 size, u64 flags, u64 device_addr); diff --git a/drivers/gpu/drm/imagination/pvr_vm.c b/drivers/gpu/drm/imagination/pvr_vm.c index ceb78694cd98..55cc999f3708 100644 --- a/drivers/gpu/drm/imagination/pvr_vm.c +++ b/drivers/gpu/drm/imagination/pvr_vm.c @@ -276,7 +276,7 @@ pvr_vm_bind_op_map_init(struct pvr_vm_bind_op *bind_op, goto err_bind_op_fini; bind_op->mmu_op_ctx = - pvr_mmu_op_context_create(vm_ctx->mmu_ctx, sgt, offset, size); + pvr_mmu_op_context_create(vm_ctx->mmu_ctx, sgt, device_addr, offset, size); err = PTR_ERR_OR_ZERO(bind_op->mmu_op_ctx); if (err) { bind_op->mmu_op_ctx = NULL; @@ -318,7 +318,7 @@ pvr_vm_bind_op_unmap_init(struct pvr_vm_bind_op *bind_op, } bind_op->mmu_op_ctx = - pvr_mmu_op_context_create(vm_ctx->mmu_ctx, NULL, 0, 0); + pvr_mmu_op_context_create(vm_ctx->mmu_ctx, NULL, device_addr, 0, 0); err = PTR_ERR_OR_ZERO(bind_op->mmu_op_ctx); if (err) { bind_op->mmu_op_ctx = NULL; From 45585c3aa285854face65293acc95eff73063d6d Mon Sep 17 00:00:00 2001 From: Pengpeng Hou Date: Sun, 20 Sep 2026 11:43:29 +0800 Subject: [PATCH 265/365] drm/imagination: clamp freelist reconstruction requests The firmware reconstruction count controls accesses to the request's fixed freelist ID array and the copy into the fixed response array. Neither access currently bounds the count to those protocol arrays. Clamp the count to the request capacity, which is shared by the response layout, and use that count consistently for reconstruction and response publication. Keep the firmware recovery exchange instead of dropping an oversized request without a response, as discussed with the firmware maintainer. The issue was found by our static-analysis tool. Fixes: 6eedddab733b ("drm/imagination: Implement free list and HWRT create and destroy ioctls") Assisted-by: gpt 5 Signed-off-by: Pengpeng Hou Reviewed-by: Alessio Belle Link: https://patch.msgid.link/20260920034329.16614-1-hppiscas@163.com Signed-off-by: Brajesh Gupta --- drivers/gpu/drm/imagination/pvr_free_list.c | 15 ++++++++++++--- 1 file changed, 12 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/imagination/pvr_free_list.c b/drivers/gpu/drm/imagination/pvr_free_list.c index e85cac83834c..faf5e586d8dc 100644 --- a/drivers/gpu/drm/imagination/pvr_free_list.c +++ b/drivers/gpu/drm/imagination/pvr_free_list.c @@ -8,6 +8,7 @@ #include "pvr_vm.h" #include +#include #include #include #include @@ -612,13 +613,21 @@ pvr_free_list_process_reconstruct_req(struct pvr_device *pvr_dev, }; struct rogue_fwif_freelists_reconstruction_data *resp = &resp_cmd.cmd_data.free_lists_reconstruction_data; + u32 count = min_t(u32, req->freelist_count, + ARRAY_SIZE(req->freelist_ids)); - for (u32 i = 0; i < req->freelist_count; i++) + if (count != req->freelist_count) { + drm_warn_once(from_pvr_device(pvr_dev), + "Requested reconstruction of %u freelists, limiting to %u\n", + req->freelist_count, count); + } + + for (u32 i = 0; i < count; i++) pvr_free_list_reconstruct(pvr_dev, req->freelist_ids[i]); - resp->freelist_count = req->freelist_count; + resp->freelist_count = count; memcpy(resp->freelist_ids, req->freelist_ids, - req->freelist_count * sizeof(resp->freelist_ids[0])); + count * sizeof(resp->freelist_ids[0])); WARN_ON(pvr_kccb_send_cmd(pvr_dev, &resp_cmd, NULL)); } From 15814c01ac57643edf1662a076d2961332f277a5 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Wed, 23 Sep 2026 21:44:30 -0400 Subject: [PATCH 266/365] drm/amd/display: Bump frame warning limit for all builds of dml Some configs with gcc are also now affected. Signed-off-by: Alex Deucher (cherry picked from commit 76b9706e7fa1a6e374703170128b1be2f590cda7) --- drivers/gpu/drm/amd/display/dc/dml/Makefile | 6 +----- drivers/gpu/drm/amd/display/dc/dml2_0/Makefile | 2 +- 2 files changed, 2 insertions(+), 6 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/dml/Makefile b/drivers/gpu/drm/amd/display/dc/dml/Makefile index 79eeb3721b99..91465ac05c97 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/Makefile +++ b/drivers/gpu/drm/amd/display/dc/dml/Makefile @@ -36,11 +36,7 @@ ifneq ($(CONFIG_FRAME_WARN),0) frame_warn_limit := 3072 endif else - ifeq ($(CONFIG_CC_IS_CLANG),y) - frame_warn_limit := 3072 - else - frame_warn_limit := 2048 - endif + frame_warn_limit := 3072 endif ifeq ($(call test-lt, $(CONFIG_FRAME_WARN), $(frame_warn_limit)),y) diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/Makefile b/drivers/gpu/drm/amd/display/dc/dml2_0/Makefile index 4d682e92df1d..39ee2d1999dd 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2_0/Makefile +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/Makefile @@ -35,7 +35,7 @@ ifneq ($(CONFIG_FRAME_WARN),0) frame_warn_limit := 3072 endif else - frame_warn_limit := 2056 + frame_warn_limit := 3072 endif ifeq ($(call test-lt, $(CONFIG_FRAME_WARN), $(frame_warn_limit)),y) From e66cf1625ec4a3fe68346119f371def713fd0a4d Mon Sep 17 00:00:00 2001 From: Namjae Jeon Date: Wed, 23 Sep 2026 20:25:49 +0900 Subject: [PATCH 267/365] smb: client: use finish_no_open() for non-regular inodes An O_CREAT open can find an existing symlink or another non-regular inode. cifs_atomic_open() calls finish_open() on it and attaches a cifsFileInfo. Symlink inodes have no CIFS release operation, so the dentry reference held by cifsFileInfo is leaked. FMODE_OPENED also prevents the VFS from following the symlink. Track whether cifs_do_create() returned an open server handle. For non-regular inodes, close the handle if present, remove the pending open, and call finish_no_open() so the VFS can continue the lookup. Do not set FMODE_CREATED unless a regular file was opened. For O_NOFOLLOW with __O_REGULAR, return -ELOOP before the VFS's -EFTYPE check. Defer closing a legacy POSIX handle on a non-regular inode until after inode lookup. This avoids closing it again if lookup fails. Fixes: d2c127197dfc ("cifs: implement i_op->atomic_open()") Signed-off-by: Namjae Jeon Signed-off-by: Paulo Alcantara Cc: stable@vger.kernel.org --- fs/smb/client/dir.c | 52 +++++++++++++++++++++++++++++++++------------ 1 file changed, 39 insertions(+), 13 deletions(-) diff --git a/fs/smb/client/dir.c b/fs/smb/client/dir.c index 6fa6d48fdfd3..1a56fa4d0e89 100644 --- a/fs/smb/client/dir.c +++ b/fs/smb/client/dir.c @@ -199,7 +199,7 @@ static int __cifs_do_create(struct inode *dir, struct dentry *direntry, struct tcon_link *tlink, unsigned int oflags, umode_t mode, __u32 *oplock, struct cifs_fid *fid, struct cifs_open_info_data *buf, - struct inode **inode) + struct inode **inode, bool *opened) { int rc = -ENOENT; int create_options = CREATE_NOT_DIR; @@ -216,6 +216,7 @@ static int __cifs_do_create(struct inode *dir, struct dentry *direntry, __le32 lease_flags = 0; *inode = NULL; + *opened = false; *oplock = 0; if (tcon->ses->server->oplocks) *oplock = REQ_OPLOCK; @@ -232,6 +233,7 @@ static int __cifs_do_create(struct inode *dir, struct dentry *direntry, oflags, oplock, &fid->netfid, xid); switch (rc) { case 0: + *opened = true; if (newinode == NULL) { /* query inode info */ goto cifs_create_get_file_info; @@ -253,11 +255,9 @@ static int __cifs_do_create(struct inode *dir, struct dentry *direntry, /* * The server may allow us to open things like * FIFOs, but the client isn't set up to deal - * with that. If it's not a regular file, just - * close it and proceed as if it were a normal - * lookup. + * with that. Keep the handle until the caller + * can finish the lookup. */ - CIFSSMBClose(xid, tcon, fid->netfid); goto cifs_create_get_file_info; } /* success, no need to query */ @@ -384,6 +384,7 @@ static int __cifs_do_create(struct inode *dir, struct dentry *direntry, } return rc; } + *opened = true; if (rdwr_for_fscache == 2) cifs_invalidate_cache(dir, FSCACHE_INVAL_DIO_WRITE); @@ -475,7 +476,7 @@ static int __cifs_do_create(struct inode *dir, struct dentry *direntry, return rc; out_err: - if (server->ops->close) + if (*opened && server->ops->close) server->ops->close(xid, tcon, fid); if (newinode) iput(newinode); @@ -487,7 +488,7 @@ static int cifs_do_create(struct inode *dir, struct dentry *direntry, unsigned int oflags, umode_t mode, __u32 *oplock, struct cifs_fid *fid, struct cifs_open_info_data *buf, - struct inode **inode) + struct inode **inode, bool *opened) { void *page = alloc_dentry_path(); const char *full_path; @@ -496,10 +497,11 @@ static int cifs_do_create(struct inode *dir, struct dentry *direntry, full_path = build_path_from_dentry(direntry, page); if (IS_ERR(full_path)) { rc = PTR_ERR(full_path); + *opened = false; } else { rc = __cifs_do_create(dir, direntry, full_path, xid, tlink, oflags, mode, oplock, - fid, buf, inode); + fid, buf, inode, opened); } free_dentry_path(page); return rc; @@ -529,6 +531,8 @@ int cifs_atomic_open(struct inode *dir, struct dentry *direntry, struct inode *inode; unsigned int xid; __u32 oplock; + bool is_regular; + bool opened; int rc; if (unlikely(cifs_forced_shutdown(cifs_sb))) @@ -581,12 +585,26 @@ int cifs_atomic_open(struct inode *dir, struct dentry *direntry, cifs_add_pending_open(&fid, tlink, &open); rc = cifs_do_create(dir, direntry, xid, tlink, oflags, mode, - &oplock, &fid, &buf, &inode); + &oplock, &fid, &buf, &inode, &opened); if (rc) { cifs_del_pending_open(&open); goto out; } + is_regular = S_ISREG(inode->i_mode); + if (!is_regular || !opened) { + if (opened && server->ops->close) + server->ops->close(xid, tcon, &fid); + cifs_del_pending_open(&open); + if (S_ISLNK(inode->i_mode) && + (oflags & (O_NOFOLLOW | __O_REGULAR)) == + (O_NOFOLLOW | __O_REGULAR) && !(oflags & O_EXCL)) { + iput(inode); + rc = -ELOOP; + goto out; + } + } + if (d_in_lookup(direntry)) { alias = d_splice_alias(inode, direntry); if (!IS_ERR_OR_NULL(alias)) @@ -595,9 +613,15 @@ int cifs_atomic_open(struct inode *dir, struct dentry *direntry, d_instantiate(direntry, inode); } - if ((oflags & (O_CREAT | O_EXCL)) == (O_CREAT | O_EXCL)) + if (is_regular && opened && + (oflags & (O_CREAT | O_EXCL)) == (O_CREAT | O_EXCL)) file->f_mode |= FMODE_CREATED; + if (!is_regular || !opened) { + rc = finish_no_open(file, NULL); + goto out; + } + rc = finish_open(file, direntry, generic_file_open); if (rc) { if (server->ops->close) @@ -660,6 +684,7 @@ int cifs_create(struct mnt_idmap *idmap, struct inode *dir, struct inode *inode; struct cifs_fid fid; __u32 oplock; + bool opened; struct cifs_open_info_data buf = {}; cifs_dbg(FYI, "cifs_create parent inode = 0x%p name is: %pd and dentry = 0x%p\n", @@ -682,10 +707,10 @@ int cifs_create(struct mnt_idmap *idmap, struct inode *dir, server->ops->new_lease_key(&fid); rc = cifs_do_create(dir, direntry, xid, tlink, oflags, - mode, &oplock, &fid, &buf, &inode); + mode, &oplock, &fid, &buf, &inode, &opened); if (!rc) { d_instantiate(direntry, inode); - if (server->ops->close) + if (opened && server->ops->close) server->ops->close(xid, tcon, &fid); } @@ -1078,6 +1103,7 @@ int cifs_tmpfile(struct mnt_idmap *idmap, struct inode *dir, struct inode *inode; unsigned int xid; __u32 oplock; + bool opened; int namelen; int rc; @@ -1116,7 +1142,7 @@ int cifs_tmpfile(struct mnt_idmap *idmap, struct inode *dir, namelen = scnprintf(name, namesize, CIFS_TMPNAME_PREFIX "%x", atomic_inc_return(&cifs_tmpcounter)); rc = __cifs_do_create(dir, dentry, path, xid, tlink, oflags, - mode, &oplock, &fid, NULL, &inode); + mode, &oplock, &fid, NULL, &inode, &opened); if (!rc) { rc = d_mark_tmpfile_name(file, &QSTR_LEN(name, namelen)); if (rc) { From 1765a153d985c231357145e26798f9408db10e42 Mon Sep 17 00:00:00 2001 From: Guopeng Zhang Date: Thu, 24 Sep 2026 17:15:36 +0800 Subject: [PATCH 268/365] cgroup/pids: Restore pids.events notifications in local mode A fork rejected by the pids controller increments the counter reported by pids.events. When local event accounting is selected, however, pids_event() returns after notifying only events_local_file, leaving pids.events pollers asleep. On legacy hierarchies, pids.events.local does not exist. With pids_localevents, pids.events reports the same local counter. In both cases, pids.events changes without generating a notification. This can be reproduced with a pids_localevents mount: mkdir /tmp/test mount -t cgroup2 -o pids_localevents none /tmp/test mkdir /tmp/test/t echo 1 > /tmp/test/t/pids.max cat /tmp/test/t/pids.events # max 0 timeout 3 inotifywait -e modify /tmp/test/t/pids.events & sh -c 'echo $$ > /tmp/test/t/cgroup.procs; (true &)' 2>/dev/null wait cat /tmp/test/t/pids.events # max 1 Without this patch, inotifywait times out without reporting an event. Notify pids.events before returning from the local event path. Fixes: 3f26a885a068 ("cgroup/pids: Add pids.events.local") Cc: stable@vger.kernel.org # v6.11+ Signed-off-by: Guopeng Zhang Signed-off-by: Tejun Heo --- kernel/cgroup/pids.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/kernel/cgroup/pids.c b/kernel/cgroup/pids.c index ecbb839d2acb..78cdc0558d0c 100644 --- a/kernel/cgroup/pids.c +++ b/kernel/cgroup/pids.c @@ -253,6 +253,11 @@ static void pids_event(struct pids_cgroup *pids_forking, } if (!cgroup_subsys_on_dfl(pids_cgrp_subsys) || cgrp_dfl_root.flags & CGRP_ROOT_PIDS_LOCAL_EVENTS) { + /* + * pids.events reports the local counter on legacy hierarchies + * and when pids_localevents is enabled. + */ + cgroup_file_notify(&p->events_file); cgroup_file_notify(&p->events_local_file); return; } From 11536ee3d3e0b1bd35b6f3f8df55a6053eb0c71d Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Wed, 23 Sep 2026 08:32:03 +0000 Subject: [PATCH 269/365] MAINTAINERS: update Eric Dumazet's email address My @google.com address is going away soon, let's switch to @kernel.org. Signed-off-by: Eric Dumazet Link: https://patch.msgid.link/20260923083203.383923-1-edumazet@google.com Signed-off-by: Jakub Kicinski --- .mailmap | 2 ++ MAINTAINERS | 8 ++++---- 2 files changed, 6 insertions(+), 4 deletions(-) diff --git a/.mailmap b/.mailmap index 389b94a0124e..b8d0dc185714 100644 --- a/.mailmap +++ b/.mailmap @@ -286,6 +286,8 @@ Ed L. Cashin Elliot Berman Enric Balletbo i Serra Enric Balletbo i Serra +Eric Dumazet +Eric Dumazet Erik Kaneda Ethan Carter Edwards Ethan Edwards Eugen Hristev diff --git a/MAINTAINERS b/MAINTAINERS index c52ec2d7d3c1..c915b114c99c 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -18885,7 +18885,7 @@ F: net/core/drop_monitor.c NETWORKING DRIVERS M: Andrew Lunn M: "David S. Miller" -M: Eric Dumazet +M: Eric Dumazet M: Jakub Kicinski M: Paolo Abeni L: netdev@vger.kernel.org @@ -19002,7 +19002,7 @@ K: struct\s+phy_port|phy_port_ NETWORKING [GENERAL] M: "David S. Miller" -M: Eric Dumazet +M: Eric Dumazet M: Jakub Kicinski M: Paolo Abeni R: Simon Horman @@ -19228,7 +19228,7 @@ F: net/ipv6/seg6* F: tools/testing/selftests/net/srv6* NETWORKING [TCP] -M: Eric Dumazet +M: Eric Dumazet M: Neal Cardwell R: Kuniyuki Iwashima L: netdev@vger.kernel.org @@ -19261,7 +19261,7 @@ F: net/tls/ F: tools/testing/selftests/net/tls.c NETWORKING [SOCKETS] -M: Eric Dumazet +M: Eric Dumazet M: Kuniyuki Iwashima M: Paolo Abeni M: Willem de Bruijn From cb97bf3d4f91453b881acaf8e9f0cc47bb40b604 Mon Sep 17 00:00:00 2001 From: Viken Dadhaniya Date: Mon, 21 Sep 2026 17:17:01 +0530 Subject: [PATCH 270/365] i2c: qcom-geni: Fix hardcoded clock index in SE_GENI_CLK_SEL qcom_geni_i2c_conf() writes a hardcoded 0 to SE_GENI_CLK_SEL, which selects an index from the hardware clock performance table. This always picks the first table entry regardless of the actual source clock configuration. On platforms where the matching entry is not at index 0, the wrong source clock divider is active and the I2C bus runs at an incorrect frequency. Use geni_se_clk_freq_match() in geni_i2c_clk_map_idx() to find the performance table index for the source clock (32 MHz or 19.2 MHz). Store the resolved index in a new clk_idx field in geni_i2c_dev and write it to SE_GENI_CLK_SEL instead of the hardcoded 0. Fixes: 37692de5d523 ("i2c: i2c-qcom-geni: Add bus driver for the Qualcomm GENI I2C controller") Signed-off-by: Viken Dadhaniya Cc: # v4.19+ Reviewed-by: Mukesh Kumar Savaliya Signed-off-by: Andi Shyti Link: https://patch.msgid.link/20260921-i2c-fix-se-clk-conf-v2-1-8b5537ceff2d@oss.qualcomm.com --- drivers/i2c/busses/i2c-qcom-geni.c | 38 ++++++++++++++++++++++++++---- 1 file changed, 33 insertions(+), 5 deletions(-) diff --git a/drivers/i2c/busses/i2c-qcom-geni.c b/drivers/i2c/busses/i2c-qcom-geni.c index e9e41a174c20..6a39e21d5ed4 100644 --- a/drivers/i2c/busses/i2c-qcom-geni.c +++ b/drivers/i2c/busses/i2c-qcom-geni.c @@ -82,6 +82,9 @@ enum geni_i2c_err_code { #define XFER_TIMEOUT HZ #define RST_TIMEOUT HZ +#define GENI_SE_CLK_32MHZ (32 * HZ_PER_MHZ) +#define GENI_SE_CLK_19P2MHZ 19200000UL + struct geni_i2c_desc { bool no_dma_support; unsigned int tx_fifo_depth; @@ -127,6 +130,7 @@ struct geni_i2c_dev { spinlock_t lock; u32 clk_freq_out; const struct geni_i2c_clk_fld *clk_fld; + u32 clk_idx; void *dma_buf; size_t xfer_len; dma_addr_t dma_addr; @@ -197,19 +201,44 @@ static const struct geni_i2c_clk_fld geni_i2c_clk_map_32mhz[] = { static int geni_i2c_clk_map_idx(struct geni_i2c_dev *gi2c) { const struct geni_i2c_clk_fld *itr; + unsigned long res_freq; - if (clk_get_rate(gi2c->se.clk) == 32 * HZ_PER_MHZ) + /* + * Frequency counter tables are calibrated for a specific source + * clock frequency and are not valid for any multiple of it + * (e.g. 64 MHz, 128 MHz). + * Use exact=true and verify res_freq matches req_freq literally + * to reject harmonics: a 64 MHz clock that divides evenly to + * 32 MHz would pass exact matching but produce double the intended + * I2C frequency with these counter values. + */ + if (!geni_se_clk_freq_match(&gi2c->se, GENI_SE_CLK_32MHZ, + &gi2c->clk_idx, &res_freq, true) && + res_freq == GENI_SE_CLK_32MHZ) { itr = geni_i2c_clk_map_32mhz; - else + } else if (!geni_se_clk_freq_match(&gi2c->se, GENI_SE_CLK_19P2MHZ, + &gi2c->clk_idx, &res_freq, true) && + res_freq == GENI_SE_CLK_19P2MHZ) { itr = geni_i2c_clk_map_19p2mhz; + } else { + dev_err(gi2c->se.dev, + "Unsupported SE source clock: must be exactly 32 MHz or 19.2 MHz\n"); + return -EINVAL; + } while (itr->clk_freq_out != 0) { if (itr->clk_freq_out == gi2c->clk_freq_out) { gi2c->clk_fld = itr; + dev_dbg(gi2c->se.dev, + "I2C clk selected: freq: %u Hz, clk_idx: %u\n", + gi2c->clk_freq_out, gi2c->clk_idx); return 0; } itr++; } + + dev_err(gi2c->se.dev, "Unsupported I2C output frequency %u Hz\n", gi2c->clk_freq_out); + return -EINVAL; } @@ -219,7 +248,7 @@ static int qcom_geni_i2c_conf(struct geni_se *se, unsigned long freq) const struct geni_i2c_clk_fld *itr = gi2c->clk_fld; u32 val; - writel_relaxed(0, gi2c->se.base + SE_GENI_CLK_SEL); + writel_relaxed(gi2c->clk_idx, gi2c->se.base + SE_GENI_CLK_SEL); val = (itr->clk_div << CLK_DIV_SHFT) | SER_CLK_EN; writel_relaxed(val, gi2c->se.base + GENI_SER_M_CLK_CFG); @@ -1111,8 +1140,7 @@ static int geni_i2c_resources_init(struct geni_se *se) ret = geni_i2c_clk_map_idx(gi2c); if (ret) - return dev_err_probe(gi2c->se.dev, ret, "Invalid clk frequency %d Hz\n", - gi2c->clk_freq_out); + return ret; return geni_icc_set_bw_ab(&gi2c->se, GENI_DEFAULT_BW, GENI_DEFAULT_BW, Bps_to_icc(gi2c->clk_freq_out)); From f050c3e61d2a1aaece3170d459e4ce5fa2486c12 Mon Sep 17 00:00:00 2001 From: Vineet Gupta Date: Wed, 23 Sep 2026 15:01:38 -0700 Subject: [PATCH 271/365] ARC: arch_cmpxchg_relaxed to use size of pointed type not pointer Bradley spotted this snafu when working on two-byte cmpxchg emulation. He originally folded this fix into his series but the fix itself uncovered a pre-existing issue, missing arch_atomic64_cmpxchg_relaxed which just needed to be wired up. Fixes: e188f3330a13 ("ARC: cmpxchg/xchg: rewrite as macros to make type safe") Reported-by: kernel test robot Closes: https://lore.kernel.org/oe-kbuild-all/202609240539.AkulvjVe-lkp@intel.com/ Suggested-by: Bradley Morgan Signed-off-by: Vineet Gupta --- arch/arc/include/asm/atomic64-arcv2.h | 6 ++++++ arch/arc/include/asm/cmpxchg.h | 4 ++-- 2 files changed, 8 insertions(+), 2 deletions(-) diff --git a/arch/arc/include/asm/atomic64-arcv2.h b/arch/arc/include/asm/atomic64-arcv2.h index 73080a664369..1ca4a1047fb3 100644 --- a/arch/arc/include/asm/atomic64-arcv2.h +++ b/arch/arc/include/asm/atomic64-arcv2.h @@ -156,6 +156,12 @@ static inline u64 __arch_cmpxchg64_relaxed(volatile void *ptr, u64 old, u64 new) } #define arch_cmpxchg64_relaxed __arch_cmpxchg64_relaxed +static inline s64 arch_atomic64_cmpxchg_relaxed(atomic64_t *v, s64 old, s64 new) +{ + return arch_cmpxchg64_relaxed(&v->counter, old, new); +} +#define arch_atomic64_cmpxchg_relaxed arch_atomic64_cmpxchg_relaxed + static inline s64 arch_atomic64_xchg(atomic64_t *ptr, s64 new) { s64 prev; diff --git a/arch/arc/include/asm/cmpxchg.h b/arch/arc/include/asm/cmpxchg.h index 76f43db0890f..66b3743c9400 100644 --- a/arch/arc/include/asm/cmpxchg.h +++ b/arch/arc/include/asm/cmpxchg.h @@ -46,7 +46,7 @@ __typeof__(*(ptr)) _n_ = (new); \ __typeof__(*(ptr)) _prev_; \ \ - switch(sizeof((_p_))) { \ + switch (sizeof(*(_p_))) { \ case 1: \ _prev_ = (__typeof__(*(ptr)))cmpxchg_emu_u8((volatile u8 *__force)_p_, (uintptr_t)_o_, (uintptr_t)_n_); \ break; \ @@ -102,7 +102,7 @@ __typeof__(ptr) _p_ = (ptr); \ __typeof__(*(ptr)) _val_ = (val); \ \ - switch(sizeof(*(_p_))) { \ + switch (sizeof(*(_p_))) { \ case 4: \ _val_ = __arch_xchg(_p_, _val_); \ break; \ From 113dcdfadf30ea11fbbdfcd4f6ea87687655cc5b Mon Sep 17 00:00:00 2001 From: Matthias Goergens Date: Fri, 25 Sep 2026 13:23:29 +0800 Subject: [PATCH 272/365] MAINTAINERS: name the libata/linux for-next branch The T: entry for LIBATA SUBSYSTEM (Serial and Parallel ATA drivers) names libata/linux without a branch. The repository's HEAD pointer points to branch master, which has no active development. Active development is on the for-next branch. Name the branch so the entry identifies where development happens. Documentation/process/submitting-patches.rst sends contributors to the T: entry to find the tree to prepare patches against, so a branch-less entry whose HEAD is already in mainline points them to the wrong branch. Reviewed-by: Damien Le Moal Signed-off-by: Matthias Goergens Link: https://lore.kernel.org/r/20260925052329.2683619-1-matthias.goergens@gmail.com Signed-off-by: Niklas Cassel --- MAINTAINERS | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/MAINTAINERS b/MAINTAINERS index 3a19da74d00c..3e7ede95c7cf 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -14911,7 +14911,7 @@ M: Damien Le Moal M: Niklas Cassel L: linux-ide@vger.kernel.org S: Maintained -T: git git://git.kernel.org/pub/scm/linux/kernel/git/libata/linux.git +T: git git://git.kernel.org/pub/scm/linux/kernel/git/libata/linux.git for-next F: Documentation/ABI/testing/sysfs-ata F: Documentation/devicetree/bindings/ata/ F: drivers/ata/ From ae146bc1abdeb4607abf2975b858c053024e8ac1 Mon Sep 17 00:00:00 2001 From: Amir Goldstein Date: Mon, 21 Sep 2026 12:40:13 +0200 Subject: [PATCH 273/365] ovl: fix UAF in ovl_do_mkdir() debug print ovl_do_mkdir() prints the input dentry with %pd after vfs_mkdir(). Since commit fe497f0759e0 ("VFS: change vfs_mkdir() to unlock on failure."), vfs_mkdir() calls end_creating() on the input dentry on failure and may replace it on success, so the post-call %pd can use-after-free the dentry when CONFIG_OVERLAY_FS_DEBUG is enabled. Print the dentry before the call and only the result afterward. Reported-by: syzbot+ced26b784bf977d223dd@syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=ced26b784bf977d223dd Fixes: fe497f0759e0 ("VFS: change vfs_mkdir() to unlock on failure.") Signed-off-by: Amir Goldstein Link: https://patch.msgid.link/20260921104013.40475-1-amir73il@gmail.com Signed-off-by: Christian Brauner (Amutable) --- fs/overlayfs/overlayfs.h | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/fs/overlayfs/overlayfs.h b/fs/overlayfs/overlayfs.h index e0d8c6152e9f..7f3558372c59 100644 --- a/fs/overlayfs/overlayfs.h +++ b/fs/overlayfs/overlayfs.h @@ -254,8 +254,10 @@ static inline struct dentry *ovl_do_mkdir(struct ovl_fs *ofs, { struct dentry *ret; + /* vfs_mkdir() drops @dentry on failure and may replace it on success */ + pr_debug("mkdir(%pd2, 0%o)\n", dentry, mode); ret = vfs_mkdir(ovl_upper_mnt_idmap(ofs), dir, dentry, mode, NULL); - pr_debug("mkdir(%pd2, 0%o) = %i\n", dentry, mode, PTR_ERR_OR_ZERO(ret)); + pr_debug("...mkdir = %i\n", PTR_ERR_OR_ZERO(ret)); return ret; } From 76d8e697242e7e4e30dd08eff7c278728f9dd8e2 Mon Sep 17 00:00:00 2001 From: Drif Abdelmalek Mohamed Said Date: Fri, 18 Sep 2026 23:42:04 +0100 Subject: [PATCH 274/365] dcache: unpoison the inline name buffer in __d_alloc() syzbot reported: BUG: KMSAN: uninit-value in dentry_string_cmp fs/dcache.c:291 [inline] BUG: KMSAN: uninit-value in dentry_cmp fs/dcache.c:322 [inline] BUG: KMSAN: uninit-value in __d_lookup_rcu+0x37d/0x5e0 fs/dcache.c:2522 dentry_string_cmp fs/dcache.c:291 [inline] dentry_cmp fs/dcache.c:322 [inline] __d_lookup_rcu+0x37d/0x5e0 fs/dcache.c:2522 lookup_fast+0x194/0xa40 fs/namei.c:1854 lookup_fast_for_open fs/namei.c:4545 [inline] open_last_lookups fs/namei.c:4579 [inline] path_openat+0x9ef/0x6540 fs/namei.c:4856 do_file_open+0x2aa/0x680 fs/namei.c:4888 do_sys_openat2+0x17c/0x390 fs/open.c:1395 do_sys_open fs/open.c:1401 [inline] __do_sys_openat fs/open.c:1417 [inline] __se_sys_openat fs/open.c:1412 [inline] __x64_sys_openat+0x240/0x300 fs/open.c:1412 x64_sys_call+0x2445/0x3ea0 arch/x86/include/generated/asm/syscalls_64.h:258 do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline] do_syscall_64+0x15d/0x3c0 arch/x86/entry/syscall_64.c:94 entry_SYSCALL_64_after_hwframe+0x77/0x7f Uninit was stored to memory at: copy_name fs/dcache.c:3031 [inline] __d_move+0xd29/0x21f0 fs/dcache.c:3099 d_move+0x71/0xf0 fs/dcache.c:3147 vfs_rename+0x2619/0x2770 fs/namei.c:6085 filename_renameat2+0xa59/0x1230 fs/namei.c:6188 __do_sys_rename fs/namei.c:6232 [inline] __se_sys_rename+0xc5/0x5c0 fs/namei.c:6228 __x64_sys_rename+0x78/0xb0 fs/namei.c:6228 x64_sys_call+0x329/0x3ea0 arch/x86/include/generated/asm/syscalls_64.h:83 do_syscall_x64 arch/x86/entry/syscall_64.c:63 do_syscall_64+0x15d/0x3c0 arch/x86/entry/syscall_64.c:94 entry_SYSCALL_64_after_hwframe+0x77/0x7f Uninit was created at: slab_post_alloc_hook mm/slub.c:4617 [inline] slab_alloc_node mm/slub.c:4939 [inline] kmem_cache_alloc_lru_noprof+0x376/0x1230 mm/s __d_alloc+0x52/0x9f0 fs/dcache.c:1902 d_alloc+0x57/0x300 fs/dcache.c:1981 lookup_one_qstr_excl+0x19d/0x7a0 fs/namei.c:1806 __start_renaming+0x341/0x850 fs/namei.c:3888 filename_renameat2+0x625/0x1230 fs/namei.c:6163 __do_sys_rename fs/namei.c:6232 [inline] __se_sys_rename+0xc5/0x5c0 fs/namei.c:6228 __x64_sys_rename+0x78/0xb0 fs/namei.c:6228 x64_sys_call+0x329/0x3ea0 arch/x86/include/generated/asm/syscalls_64.h:83 do_syscall_x64 arch/x86/entry/syscall_64.c:63 do_syscall_64+0x15d/0x3c0 arch/x86/entry/syscall_64.c:94 entry_SYSCALL_64_after_hwframe+0x77/0x7f The race is between a concurrent open() and rename() of the same path. __d_alloc() only stores the name itself and its terminating NUL, so the rest of the inline buffer (d_shortname, DNAME_INLINE_LEN bytes) is left uninitialized. copy_name(), called from rename(), copies that buffer as a whole, so the uninitialized tail is propagated into the dentry that is being moved. Meanwhile __d_lookup_rcu(), called from open(), is an optimistic lockless lookup: it checks d_name.hash_len first and leaves the seqcount retry to its caller, so it can end up comparing against a dentry whose name a rename is rewriting in place, using a stale (longer) length. The comparison then runs past the terminating NUL and reads bytes of the uninitialized tail, which KMSAN reports. The read is harmless by design: it stays inside the buffer, the name is still NUL-terminated, and the result is thrown away by the seqcount retry. It is not specific to KMSAN either - with CONFIG_DCACHE_WORD_ACCESS enabled the very same bytes are read by read_word_at_a_time(), which is __no_sanitize_or_inline and therefore invisible to KMSAN. KMSAN builds only see the instrumented byte-at-a-time dentry_string_cmp() because CONFIG_DCACHE_WORD_ACCESS is disabled when KMSAN is enabled on x86: commit 7cf8f44a5a1c ("x86: fs: kmsan: disable CONFIG_DCACHE_WORD_ACCESS") Zeroing the inline buffer would hide the report, but it would add a memset() to a hot allocation path just to initialize bytes that are never used as part of a name. Instead, tell KMSAN the inline buffer is initialized: kmsan_unpoison_memory() compiles to nothing unless CONFIG_KMSAN is set, and doing it at allocation time is enough for every dentry, because copy_name() and swap_names() copy the whole buffer and thus propagate its shadow. Reported-by: syzbot+7ff3adde89dd795ad4c4@syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=7ff3adde89dd795ad4c4 Signed-off-by: Drif Abdelmalek Mohamed Said Changes in v3: - Annotate for KMSAN instead of zeroing, as suggested in review: the read is harmless, so unpoison the inline buffer in __d_alloc() with kmsan_unpoison_memory() (a no-op unless CONFIG_KMSAN) rather than adding a memset() to the dentry allocation path. - Document why only KMSAN builds report this at all: with CONFIG_DCACHE_WORD_ACCESS the same read goes through read_word_at_a_time(), which KMSAN does not instrument. - Rewrite the commit message; the previous one had several truncated lines. Link: https://patch.msgid.link/20260918224204.3056-1-drifabdelmalekmohamedsaid@gmail.com Reviewed-by: Jan Kara Signed-off-by: Christian Brauner (Amutable) --- fs/dcache.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/fs/dcache.c b/fs/dcache.c index 1b1a81f10da6..a66be85f9d01 100644 --- a/fs/dcache.c +++ b/fs/dcache.c @@ -1916,6 +1916,10 @@ static struct dentry *__d_alloc(struct super_block *sb, const struct qstr *name) * be overwriting an internal NUL character */ dentry->d_shortname.string[DNAME_INLINE_LEN-1] = 0; + + /* Racy __d_lookup_rcu() walk may read past the NUL; harmless */ + kmsan_unpoison_memory(dentry->d_shortname.string, DNAME_INLINE_LEN); + if (unlikely(!name)) { name = &slash_name; dname = dentry->d_shortname.string; From 5bfa9f1a9dcb6ecb607adbc1c0226605c972935b Mon Sep 17 00:00:00 2001 From: Andrea Parri Date: Thu, 24 Sep 2026 11:21:39 +0200 Subject: [PATCH 275/365] kprobes: Fix permanent hang when flushing the kprobe optimizer Writing 0 to /proc/sys/debug/kprobes-optimization while a kprobe is jump-optimized never returns. The writer sleeps in D state forever with kprobe_sysctl_mutex held, so any later read or write of that sysctl hangs as well. For example, with vfs_read+9 as an optimizable address in this build: # cd /sys/kernel/tracing # echo 'p:myprobe vfs_read+9' >> kprobe_events # echo 1 > events/kprobes/myprobe/enable # # wait until /sys/kernel/debug/kprobes/list shows [OPTIMIZED] # echo 0 > /proc/sys/debug/kprobes-optimization INFO: task sh:246 blocked for more than 10 seconds. Call Trace: __schedule+0x1176/0x4f70 schedule+0xdc/0x2c0 schedule_timeout+0x17b/0x260 wait_for_completion+0x173/0x3c0 wait_for_kprobe_optimizer_locked+0xbc/0x130 proc_kprobes_optimization_handler+0x156/0x1b0 proc_sys_call_handler+0x324/0x490 vfs_write+0x52d/0xfe0 ksys_write+0xff/0x200 do_syscall_64+0x106/0x630 entry_SYSCALL_64_after_hwframe+0x77/0x7f ... INFO: task cat:265 is blocked on a mutex likely owned by task sh:246. wait_for_kprobe_optimizer_locked() reinitializes optimizer_completion, asks the optimizer thread to flush and sleeps in wait_for_completion(). The thread drains the (un)optimizing lists, but calls complete() only if completion_done() is true, i.e. if the completion is already done, which never happens while someone waits. disarm_all_kprobes() and kprobe_trace_self_tests_init() wait the same way. Calling complete() unconditionally would not be enough: the waiter drops kprobe_mutex while it sleeps, and nothing else serializes the sysctl handler against the debugfs "enabled" file. A second flusher that still finds the lists non-empty, e.g. because a disabled probe is queued for unoptimizing, reinitializes the completion under the first: sysctl write debugfs "enabled" write unoptimize_all_kprobes() wait_for_kprobe_optimizer_locked() init_completion(c) mutex_unlock(&kprobe_mutex) wait_for_completion(c) disarm_all_kprobes() wait_for_kprobe_optimizer_locked() init_completion(c) // c->wait is reset, the first // waiter is off the queue mutex_unlock(&kprobe_mutex) wait_for_completion(c) kprobe_optimizer() complete(c) // wakes the debugfs writer only where c is &optimizer_completion. Lining up the two writes during an optimizer pass loses the sysctl writer this way. Replace the completion with a counter of optimizer passes, bumped at the end of each pass and signalled with wake_up_var_locked(), both under kprobe_mutex. A flusher samples the count and waits with wait_var_event_mutex(), which drops kprobe_mutex only while sleeping, so a new count means a whole pass ran in the meantime. Nothing is reinitialized, so several flushers can sleep in the wait at once. Link: https://lore.kernel.org/all/20260924092142.199198-1-parri.andrea@gmail.com/ Fixes: 73c12f209462 ("kprobes: Use dedicated kthread for kprobe optimizer") Cc: stable@vger.kernel.org Assisted-by: LLM Signed-off-by: Andrea Parri Signed-off-by: Masami Hiramatsu (Google) --- kernel/kprobes.c | 22 ++++++++++++++-------- 1 file changed, 14 insertions(+), 8 deletions(-) diff --git a/kernel/kprobes.c b/kernel/kprobes.c index 6337da5cab9e..4edd8ca5c657 100644 --- a/kernel/kprobes.c +++ b/kernel/kprobes.c @@ -42,6 +42,7 @@ #include #include #include +#include #include #include @@ -526,7 +527,8 @@ enum { OPTIMIZER_ST_FLUSHING = 2, }; -static DECLARE_COMPLETION(optimizer_completion); +/* Bumped at the end of each kprobe_optimizer() pass, under 'kprobe_mutex' */ +static unsigned long optimizer_passes; #define OPTIMIZE_DELAY 5 @@ -654,9 +656,9 @@ static void kprobe_optimizer(void) do_free_cleaned_kprobes(); } - /* Step 5: Kick optimizer again if needed. But if there is a flush requested, */ - if (completion_done(&optimizer_completion)) - complete(&optimizer_completion); + /* Step 5: Wake up flushers, and kick optimizer again if needed. */ + optimizer_passes++; + wake_up_var_locked(&optimizer_passes, &kprobe_mutex); if (!list_empty(&optimizing_list) || !list_empty(&unoptimizing_list)) kick_kprobe_optimizer(); /*normal kick*/ @@ -708,7 +710,8 @@ static void wait_for_kprobe_optimizer_locked(void) lockdep_assert_held(&kprobe_mutex); while (!list_empty(&optimizing_list) || !list_empty(&unoptimizing_list)) { - init_completion(&optimizer_completion); + unsigned long passes = optimizer_passes; + /* * Set state to OPTIMIZER_ST_FLUSHING and wake up the thread if it's * idle. If it's already kicked, it will see the state change. @@ -717,9 +720,12 @@ static void wait_for_kprobe_optimizer_locked(void) OPTIMIZER_ST_FLUSHING) != OPTIMIZER_ST_FLUSHING) wake_up(&kprobe_optimizer_wait); - mutex_unlock(&kprobe_mutex); - wait_for_completion(&optimizer_completion); - mutex_lock(&kprobe_mutex); + /* + * kprobe_optimizer() holds 'kprobe_mutex' for a whole pass, which + * this drops while sleeping, so a new count means a full pass ran. + */ + wait_var_event_mutex(&optimizer_passes, + optimizer_passes != passes, &kprobe_mutex); } } From aa5e44b29ffe4eaa08cc2237fd65bc2596bc023e Mon Sep 17 00:00:00 2001 From: Hui Peng Date: Sat, 19 Sep 2026 20:48:08 +0000 Subject: [PATCH 276/365] autofs: fix sbi->pipe file reference leak in autofs_kill_sb() When autofs_fill_super() fails before clearing AUTOFS_SBI_CATATONIC (for example, when find_get_pid() fails on an invalid pgrp mount option, or when an fs_context is closed before mounting), deactivate_locked_super() invokes autofs_kill_sb() -> autofs_catatonic_mode(sbi). Because AUTOFS_SBI_CATATONIC is still set in sbi->flags, autofs_catatonic_mode() returns early without calling fput(sbi->pipe), permanently leaking the pipe struct file reference. Explicitly release sbi->pipe in autofs_kill_sb() if it is still non-NULL after autofs_catatonic_mode(). Fixes: ebc921ca9b92 ("autofs: copy autofs4 to autofs") Signed-off-by: Hui Peng Link: https://patch.msgid.link/20260919204808.2812930-1-benquike@gmail.com Signed-off-by: Christian Brauner (Amutable) --- fs/autofs/inode.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/fs/autofs/inode.c b/fs/autofs/inode.c index 6b15a3717ba7..066c16f2ea56 100644 --- a/fs/autofs/inode.c +++ b/fs/autofs/inode.c @@ -51,6 +51,10 @@ void autofs_kill_sb(struct super_block *sb) if (sbi) { /* Free wait queues, close pipe */ autofs_catatonic_mode(sbi); + if (sbi->pipe) { + fput(sbi->pipe); + sbi->pipe = NULL; + } put_pid(sbi->oz_pgrp); } From 1c1a342aceec79528b3ba51f376eca2be5928fe7 Mon Sep 17 00:00:00 2001 From: Sagnik Sasmal Date: Thu, 10 Sep 2026 22:44:05 +0000 Subject: [PATCH 277/365] mtd: spinand: Do not update the QE bit on devices without one Commit be0b86c648bf ("mtd: spinand: Gather all the bus interface steps in one single function") moved quad-enable setup into spinand_configure_chip(). The new code only determines whether quad mode is needed when SPINAND_HAS_QE_BIT is set, but calls spinand_init_quad_enable() unconditionally. This clears configuration register bit 0 on devices without a QE bit. That bit is not universally a QE bit. On the Winbond W25N02KV it is H-DIS, which disables the active-low HOLD function. Clearing H-DIS enables HOLD during single and dual I/O operations. If IO3 is not kept high, the flash can pause a command and ignore clock and data. H-DIS is not restored by the FFh reset command, allowing the incorrect state to survive an SoC warm reboot while the flash remains powered. Before the refactoring, spinand_init_quad_enable() returned without touching the configuration register on devices without SPINAND_HAS_QE_BIT. Restore that behavior by only calling the helper when the flag is set. Return zero explicitly once SSDR configuration completes, as all errors are returned immediately. This avoids returning an uninitialized value when neither optional configuration step runs. The regression was reproduced on a JioRouter JIDU6401 with an MT7986 SoC and a W25N02KV. With Linux 6.18.44, sysupgrade failed and the following warm reboot hung in BL2. With this change applied, both sysupgrade and warm reboot completed successfully. Fixes: be0b86c648bf ("mtd: spinand: Gather all the bus interface steps in one single function") Cc: stable@vger.kernel.org Suggested-by: Miquel Raynal Assisted-by: LLM Signed-off-by: Sagnik Sasmal Signed-off-by: Miquel Raynal --- drivers/mtd/nand/spi/core.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/drivers/mtd/nand/spi/core.c b/drivers/mtd/nand/spi/core.c index 43bbe2629b85..95353777d7f7 100644 --- a/drivers/mtd/nand/spi/core.c +++ b/drivers/mtd/nand/spi/core.c @@ -1813,11 +1813,11 @@ static int spinand_configure_chip(struct spinand_device *spinand) spinand->ssdr_op_templates.write_cache->data.buswidth == 4 || spinand->ssdr_op_templates.update_cache->data.buswidth == 4) quad_enable = true; - } - ret = spinand_init_quad_enable(spinand, quad_enable); - if (ret) - return ret; + ret = spinand_init_quad_enable(spinand, quad_enable); + if (ret) + return ret; + } if (spinand->configure_chip) { ret = spinand->configure_chip(spinand, SSDR); @@ -1825,7 +1825,7 @@ static int spinand_configure_chip(struct spinand_device *spinand) return ret; } - return ret; + return 0; } static int spinand_init_flash(struct spinand_device *spinand) From 35d442ed1f86465e49df3119fb898f985186db13 Mon Sep 17 00:00:00 2001 From: Andrea Parri Date: Tue, 22 Sep 2026 16:55:30 +0200 Subject: [PATCH 278/365] bpf: fs/xattr: don't assume the inode is locked in path_unlink/path_rmdir bpf_lsm_has_d_inode_locked() makes the verifier rewrite bpf_[set|remove]_dentry_xattr() to the _locked variants, which assume that the caller already holds the inode's i_rwsem. The path_unlink and path_rmdir hooks are listed, but security_path_unlink() and security_path_rmdir() run before vfs_unlink()/vfs_rmdir() take the victim inode's i_rwsem, so a sleepable BPF LSM program attached to either hook mutates the victim's xattrs without the lock held. Drop the two path hooks from d_inode_locked_hooks so that the verifier keeps the locking bpf_[set|remove]_dentry_xattr() variants, which take the lock themselves. Fixes: 56467292794b8 ("bpf: fs/xattr: Add BPF kfuncs to set and remove xattrs") Cc: stable@vger.kernel.org Signed-off-by: Andrea Parri Link: https://patch.msgid.link/20260922145530.369775-1-parri.andrea@gmail.com Signed-off-by: Christian Brauner (Amutable) --- fs/bpf_fs_kfuncs.c | 4 ---- 1 file changed, 4 deletions(-) diff --git a/fs/bpf_fs_kfuncs.c b/fs/bpf_fs_kfuncs.c index 6cb877267978..357a379ef92a 100644 --- a/fs/bpf_fs_kfuncs.c +++ b/fs/bpf_fs_kfuncs.c @@ -472,10 +472,6 @@ BTF_ID(func, bpf_lsm_inode_rmdir) BTF_ID(func, bpf_lsm_inode_setattr) BTF_ID(func, bpf_lsm_inode_setxattr) BTF_ID(func, bpf_lsm_inode_unlink) -#ifdef CONFIG_SECURITY_PATH -BTF_ID(func, bpf_lsm_path_unlink) -BTF_ID(func, bpf_lsm_path_rmdir) -#endif /* CONFIG_SECURITY_PATH */ BTF_SET_END(d_inode_locked_hooks) bool bpf_lsm_has_d_inode_locked(const struct bpf_prog *prog) From b78b728e21c32ec4c330b299f657fb1eb02dffc2 Mon Sep 17 00:00:00 2001 From: Hao Ge Date: Wed, 23 Sep 2026 14:37:59 +0800 Subject: [PATCH 279/365] netfs: Fix missing alloc tagging of direct mempool allocations Commit 1d78d56c43ef ("netfs: Fix folio_queue ENOMEM in writeback by adding a mempool") added a mempool for the folio_queues and made the request, subrequest and folio_queue allocations distinguish between writeback and everything else. Writeback is part of memory reclaim and must not fail due to ENOMEM, so it allocates under GFP_NOFS through mempool_alloc(), which may dip into the pool's reserve and, if that runs empty, wait for elements to be returned. The GFP_KERNEL paths, which can return -ENOMEM to their callers, invoke the pool's ->alloc() callback directly instead. The direct call, however, skips the alloc_hooks() wrapper that the mempool_alloc() macro provides. The pool callbacks, mempool_alloc_slab() and mempool_kmalloc(), call kmem_cache_alloc_noprof() and kmalloc_noprof() and rely on current->alloc_tag having been set by the caller. With CONFIG_MEM_ALLOC_PROFILING_DEBUG=y this leads to current->alloc_tag not set WARNING: ./include/linux/alloc_tag.h:161 at __alloc_tagging_slab_alloc_hook alloc_tag was not set WARNING: ./include/linux/alloc_tag.h:166 at __alloc_tagging_slab_free_hook at allocation and free time respectively, as reported when reading files on a CIFS mount. The allocations are also missing from /proc/allocinfo. Wrap the direct ->alloc() invocations in alloc_hooks() with a new mempool_alloc_noreserve() helper in include/linux/mempool.h, next to the other alloc_hooks()-wrapped macros such as mempool_alloc(). The GFP_KERNEL paths keep their failable allocation semantics, they just get tagged now. Fixes: 1d78d56c43ef ("netfs: Fix folio_queue ENOMEM in writeback by adding a mempool") Reported-by: Erhard Furtner Closes: https://lore.kernel.org/all/0b004319-9ef7-437c-a4dd-174d6a9a83db@mailbox.org/ Tested-by: Erhard Furtner Suggested-by: Suren Baghdasaryan Cc: stable@vger.kernel.org Signed-off-by: Hao Ge Link: https://patch.msgid.link/20260923063759.34667-1-hao.ge@linux.dev Acked-by: Vlastimil Babka (SUSE) Signed-off-by: Christian Brauner (Amutable) --- fs/netfs/objects.c | 4 ++-- fs/netfs/rolling_buffer.c | 2 +- include/linux/mempool.h | 7 +++++++ 3 files changed, 10 insertions(+), 3 deletions(-) diff --git a/fs/netfs/objects.c b/fs/netfs/objects.c index 7f6a3e912602..ad549daa9c79 100644 --- a/fs/netfs/objects.c +++ b/fs/netfs/objects.c @@ -34,7 +34,7 @@ struct netfs_io_request *netfs_alloc_request(struct address_space *mapping, rreq = mempool_alloc(mempool, gfp); } else { - rreq = mempool->alloc(gfp, mempool->pool_data); + rreq = mempool_alloc_noreserve(mempool, gfp); if (!rreq) return ERR_PTR(-ENOMEM); } @@ -214,7 +214,7 @@ struct netfs_io_subrequest *netfs_alloc_subrequest(struct netfs_io_request *rreq struct kmem_cache *cache = mempool->pool_data; if (rreq->gfp == GFP_KERNEL) - subreq = mempool->alloc(rreq->gfp, mempool->pool_data); + subreq = mempool_alloc_noreserve(mempool, rreq->gfp); else subreq = mempool_alloc(mempool, rreq->gfp); if (!subreq) diff --git a/fs/netfs/rolling_buffer.c b/fs/netfs/rolling_buffer.c index 424e77a9a109..d30d5ef6d86e 100644 --- a/fs/netfs/rolling_buffer.c +++ b/fs/netfs/rolling_buffer.c @@ -29,7 +29,7 @@ struct folio_queue *netfs_folioq_alloc(unsigned int rreq_id, gfp_t gfp, struct folio_queue *fq; if (gfp == GFP_KERNEL) - fq = netfs_folioq_pool.alloc(gfp, netfs_folioq_pool.pool_data); + fq = mempool_alloc_noreserve(&netfs_folioq_pool, gfp); else fq = mempool_alloc(&netfs_folioq_pool, gfp); if (fq) { diff --git a/include/linux/mempool.h b/include/linux/mempool.h index a0fa6d43e0dc..6da502aef2f7 100644 --- a/include/linux/mempool.h +++ b/include/linux/mempool.h @@ -70,6 +70,13 @@ int mempool_alloc_bulk_noprof(struct mempool *pool, void **elem, #define mempool_alloc_bulk(...) \ alloc_hooks(mempool_alloc_bulk_noprof(__VA_ARGS__)) +/* + * Allocate a new element without dipping into the pool's reserves or + * waiting. Returns NULL on failure. + */ +#define mempool_alloc_noreserve(_pool, _gfp) \ + alloc_hooks((_pool)->alloc(_gfp, (_pool)->pool_data)) + void *mempool_alloc_preallocated(struct mempool *pool) __malloc; void mempool_free(void *element, struct mempool *pool); unsigned int mempool_free_bulk(struct mempool *pool, void **elem, From aa650d87a498844060eecb1d336f4d46f4d1b0b3 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?J=C3=A9r=C3=A9my=20Jean?= Date: Tue, 22 Sep 2026 20:27:35 +0000 Subject: [PATCH 280/365] audit: fix exe mark UAF in kill_rules() MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit kill_rules() removes mixed AUDIT_DIR and AUDIT_EXE rules when an audit tree is pruned. It drops entry->rule.exe before removing the rule from the RCU-visible filter lists. After a rule has been installed with AUDIT_ADD_RULE, which requires CAP_AUDIT_CONTROL, removing the watched directory can race with another task that is still evaluating the rule. In that case, fsnotify can free the executable mark before the reader reaches audit_mark_compare(), causing a use-after-free. KASAN reports: BUG: KASAN: slab-use-after-free in audit_mark_compare+0x8d/0xa0 Unlink the published rules from the RCU-visible lists and retain them on tree->rules for cleanup. If any rule has an executable mark, wait for a single RCU grace period before removing the marks and scheduling the entries for freeing. Otherwise, call_rcu() already provides the required deferred freeing without a synchronous wait. Cc: stable@vger.kernel.org Fixes: 34d99af52ad4 ("audit: implement audit by executable") Assisted-by: Codex:gpt-5 Signed-off-by: Jérémy Jean Reviewed-by: Ricardo Robaina Tested-by: Ricardo Robaina Reviewed-by: Bradley Morgan Signed-off-by: Paul Moore --- kernel/audit_tree.c | 36 ++++++++++++++++++++++++++---------- 1 file changed, 26 insertions(+), 10 deletions(-) diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c index 1ed19b775912..f2e81be8265e 100644 --- a/kernel/audit_tree.c +++ b/kernel/audit_tree.c @@ -545,21 +545,37 @@ static void kill_rules(struct audit_context *context, struct audit_tree *tree) { struct audit_krule *rule, *next; struct audit_entry *entry; + bool need_sync = false; + + list_for_each_entry_safe(rule, next, &tree->rules, rlist) { + entry = container_of(rule, struct audit_entry, rule); + + if (rule->tree) { + /* not a half-baked one */ + audit_tree_log_remove_rule(context, rule); + rule->tree = NULL; + list_del_rcu(&entry->list); + list_del(&entry->rule.list); + if (entry->rule.exe) + need_sync = true; + } else { + list_del_init(&rule->rlist); + } + } + + if (list_empty(&tree->rules)) + return; + + if (need_sync) + synchronize_rcu(); list_for_each_entry_safe(rule, next, &tree->rules, rlist) { entry = container_of(rule, struct audit_entry, rule); list_del_init(&rule->rlist); - if (rule->tree) { - /* not a half-baked one */ - audit_tree_log_remove_rule(context, rule); - if (entry->rule.exe) - audit_remove_mark(entry->rule.exe); - rule->tree = NULL; - list_del_rcu(&entry->list); - list_del(&entry->rule.list); - call_rcu(&entry->rcu, audit_free_rule_rcu); - } + if (entry->rule.exe) + audit_remove_mark(entry->rule.exe); + call_rcu(&entry->rcu, audit_free_rule_rcu); } } From 9814077275eca36ebf8d510d2f076d235ff9f51a Mon Sep 17 00:00:00 2001 From: Fan Wu Date: Tue, 22 Sep 2026 20:13:48 -0700 Subject: [PATCH 281/365] ipe: fix use-after-free when auditing a newly loaded policy new_policy() audits the policy after ipe_new_policyfs_node() publishes it and drops the new directory's inode lock. A concurrent delete can free the policy while ipe_audit_policy_load() is still using it. Audit the successful load under that lock. Fixes: f44554b5067b ("audit,ipe: add IPE auditing support") Cc: stable@vger.kernel.org Assisted-by: LLM [FW: remove model name according to latest guideline] Signed-off-by: Fan Wu --- security/ipe/fs.c | 8 +++----- security/ipe/policy_fs.c | 3 +++ 2 files changed, 6 insertions(+), 5 deletions(-) diff --git a/security/ipe/fs.c b/security/ipe/fs.c index 076c111c85c8..847a76afb93d 100644 --- a/security/ipe/fs.c +++ b/security/ipe/fs.c @@ -159,18 +159,16 @@ static ssize_t new_policy(struct file *f, const char __user *data, } rc = ipe_new_policyfs_node(p); - if (rc) - goto out; out: kfree(copy); if (rc < 0) { ipe_free_policy(p); ipe_audit_policy_load(ERR_PTR(rc)); - } else { - ipe_audit_policy_load(p); + return rc; } - return (rc < 0) ? rc : len; + + return len; } static const struct file_operations np_fops = { diff --git a/security/ipe/policy_fs.c b/security/ipe/policy_fs.c index 9d92d8a14b13..a7aeb57483c6 100644 --- a/security/ipe/policy_fs.c +++ b/security/ipe/policy_fs.c @@ -481,6 +481,9 @@ int ipe_new_policyfs_node(struct ipe_policy *p) inode_lock(root); p->policyfs = policyfs; root->i_private = p; + /* Only audit signed policies from userspace */ + if (p->pkcs7) + ipe_audit_policy_load(p); inode_unlock(root); return 0; From 2776e9c28513a1c855a94792b292cbcc533418c8 Mon Sep 17 00:00:00 2001 From: Fan Wu Date: Tue, 22 Sep 2026 20:13:49 -0700 Subject: [PATCH 282/365] ipe: protect the dm-verity root hash with RCU ipe_bdev_setintegrity() frees the old root hash when dm-verity publishes a new one on ->preresume, while policy evaluation can still be dereferencing it. Protect the root hash with RCU. The evaluation path already runs under rcu_read_lock(). Fixes: e155858dd995 ("ipe: add support for dm-verity as a trust provider") Cc: stable@vger.kernel.org Assisted-by: LLM [FW: remove model name according to latest guideline] Signed-off-by: Fan Wu --- security/ipe/eval.c | 12 ++++++++---- security/ipe/eval.h | 2 +- security/ipe/hooks.c | 22 +++++++++++++++++----- 3 files changed, 26 insertions(+), 10 deletions(-) diff --git a/security/ipe/eval.c b/security/ipe/eval.c index 21439c5be336..4a7be96c84f0 100644 --- a/security/ipe/eval.c +++ b/security/ipe/eval.c @@ -134,10 +134,14 @@ static bool evaluate_boot_verified(const struct ipe_eval_ctx *const ctx) static bool evaluate_dmv_roothash(const struct ipe_eval_ctx *const ctx, struct ipe_prop *p) { - return !!ctx->ipe_bdev && - !!ctx->ipe_bdev->root_hash && - ipe_digest_eval(p->value, - ctx->ipe_bdev->root_hash); + const struct digest_info *root_hash; + + if (!ctx->ipe_bdev) + return false; + + root_hash = rcu_dereference(ctx->ipe_bdev->root_hash); + + return root_hash && ipe_digest_eval(p->value, root_hash); } #else static bool evaluate_dmv_roothash(const struct ipe_eval_ctx *const ctx, diff --git a/security/ipe/eval.h b/security/ipe/eval.h index fef65a36468c..b26d1147d070 100644 --- a/security/ipe/eval.h +++ b/security/ipe/eval.h @@ -27,7 +27,7 @@ struct ipe_bdev { #ifdef CONFIG_IPE_PROP_DM_VERITY_SIGNATURE bool dm_verity_signed; #endif /* CONFIG_IPE_PROP_DM_VERITY_SIGNATURE */ - struct digest_info *root_hash; + struct digest_info __rcu *root_hash; }; #endif /* CONFIG_IPE_PROP_DM_VERITY */ diff --git a/security/ipe/hooks.c b/security/ipe/hooks.c index 0ae54a880405..d878b52ceec8 100644 --- a/security/ipe/hooks.c +++ b/security/ipe/hooks.c @@ -9,6 +9,7 @@ #include #include #include +#include #include "ipe.h" #include "hooks.h" @@ -232,7 +233,20 @@ void ipe_bdev_free_security(struct block_device *bdev) { struct ipe_bdev *blob = ipe_bdev(bdev); - ipe_digest_free(blob->root_hash); + ipe_digest_free(rcu_access_pointer(blob->root_hash)); +} + +static void ipe_set_dmverity_roothash(struct ipe_bdev *blob, + struct digest_info *info) +{ + struct digest_info *old; + + /* Protected by device-mapper's md->suspend_lock */ + old = rcu_replace_pointer(blob->root_hash, info, true); + if (old) { + synchronize_rcu(); + ipe_digest_free(old); + } } #ifdef CONFIG_IPE_PROP_DM_VERITY_SIGNATURE @@ -280,8 +294,7 @@ int ipe_bdev_setintegrity(struct block_device *bdev, enum lsm_integrity_type typ return -EINVAL; if (!value) { - ipe_digest_free(blob->root_hash); - blob->root_hash = NULL; + ipe_set_dmverity_roothash(blob, NULL); return 0; } @@ -301,8 +314,7 @@ int ipe_bdev_setintegrity(struct block_device *bdev, enum lsm_integrity_type typ info->digest_len = digest->digest_len; - ipe_digest_free(blob->root_hash); - blob->root_hash = info; + ipe_set_dmverity_roothash(blob, info); return 0; err: From aadf655e8ea9a8ffa24bb8d69f504572e8d18181 Mon Sep 17 00:00:00 2001 From: Binbin Deng <18983559317@163.com> Date: Tue, 22 Sep 2026 21:19:19 +0800 Subject: [PATCH 283/365] net/mctp: publish the mctp_dev only after it is initialised mctp_add_dev() links the new mctp_dev into dev->mctp_ptr before mdev->dev is assigned, so a reader that picks the device up in that window dereferences a net_device pointer that is still NULL: KASAN: null-ptr-deref in range [0x00000000000000b0-0x00000000000000b7] Call Trace: ? __pfx_mctp_sendmsg (./include/linux/sockptr.h:49) ? __pfx_mctp_dst_output (net/mctp/route.c:41) ? selinux_socket_sendmsg (security/selinux/hooks.c:5278) ____sys_sendmsg (net/socket.c:775 (discriminator 1) net/socket.c:790 (discriminator 1) net/socket.c:2684 (discriminator 1)) ? __pfx_____sys_sendmsg (net/socket.c:1131) ? __pfx_copy_msghdr_from_user (net/socket.c:2590) ? update_cfs_rq_load_avg (kernel/sched/fair.c:5478) ___sys_sendmsg (net/socket.c:2738) ? __pfx____sys_sendmsg (net/socket.c:2625) ? perf_event_task_tick (./include/linux/rcupdate.h:838) ? sched_tick (kernel/sched/core.c:5818) ? clockevents_program_event (kernel/time/clockevents.c:372) ? fdget (./include/linux/rcupdate.h:873 fs/file.c:1100) __sys_sendmsg (net/socket.c:2770) ? __pfx___sys_sendmsg (net/socket.c:2751) do_syscall_64 (arch/x86/entry/syscall_64.c:63) entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:121) Fix by publishing only once the object is fully initialised: __mctp_dev_get() runs under rcu_read_lock() only and hands this object to readers as soon as mctp_ptr is assigned, so they must never observe mdev->dev == NULL. Fixes: 583be982d934 ("mctp: Add device handling and netlink interface") Signed-off-by: Binbin Deng <18983559317@163.com> Reviewed-by: Eric Dumazet Link: https://patch.msgid.link/729830b8.130.1a0c945579d.Coremail.18983559317@163.com Signed-off-by: Jakub Kicinski --- net/mctp/device.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/net/mctp/device.c b/net/mctp/device.c index 822120e860c8..9f76ff092559 100644 --- a/net/mctp/device.c +++ b/net/mctp/device.c @@ -344,13 +344,13 @@ static struct mctp_dev *mctp_add_dev(struct net_device *dev) mdev->net = mctp_default_net(dev_net(dev)); - /* associate to net_device */ refcount_set(&mdev->refs, 1); - rcu_assign_pointer(dev->mctp_ptr, mdev); - dev_hold(dev); mdev->dev = dev; + /* Associate to net_device, publish the mdev. */ + rcu_assign_pointer(dev->mctp_ptr, mdev); + return mdev; } From 0bb8eab29b55045281a4963d4557f2776cdf8cfa Mon Sep 17 00:00:00 2001 From: Jamal Hadi Salim Date: Thu, 24 Sep 2026 04:32:49 -0400 Subject: [PATCH 284/365] net/sched: cls_api: reclaim an empty proto on the error path Two racing tc filter add requests on the same chain/prio of an unlocked classifier both run change() on the shared proto and both can fail: the winner's tcf_chain_tp_delete_empty() attempt gives up because the loser's handle is still in the idr, and the loser's error path drops only its own reference without a second reclamation attempt. The empty proto stays linked in the chain, holding the chain reference, a block reference and the classifier module reference until the chain or block is torn down. Reclaim the proto on the error path of any failed request that holds a proto reference. The reclamation is emptiness-gated: tcf_chain_tp_delete_empty() unlinks the proto only when delete_empty() admits it is empty, so a live shared proto is never unlinked. A proto the request created is reclaimed unconditionally - it is the only owner, so marking it for deletion is safe even without a delete_empty callback. Classifiers without one (the check marks the proto unconditionally) are rtnl-serialized, so the raced window this guard closes cannot arise for them. This is a follow-up to commit d4e359b3608a ("net/sched: cls_api: fix teardown of an adopted proto on insert-race loss"), which stopped the loser of the insert race from unlinking the winner's live proto but left the empty-proto residual in place. Conditions to recreate: - CONFIG_NET_CLS_FLOWER=y; veth pair - tc qdisc add dev veth0 ingress - two concurrent `tc filter add dev veth0 ingress protocol ip pref 1 flower skip_sw ... action drop` (both fail in fl_hw_replace_filter after publishing their handle in the idr); repeat in a loop - an empty flower tp stays linked after both requests fail; visible as a bare `filter protocol ip pref 1 flower chain 0` header in `tc filter show` with no filter entries - CAP_NET_ADMIN (namespace-local via unshare -Urn suffices) Fixes: 8b64678e0af8 ("net: sched: refactor tp insert/delete for concurrent execution") Reported-by: Sashiko (nipa) Closes: https://netdev-ai.bots.linux.dev/sashiko/#/patchset/20260805134049.927864-1-victor@mojatatu.com Signed-off-by: Jamal Hadi Salim Reviewed-by: Simon Horman Link: https://patch.msgid.link/QDISC-JCOT.v1.20260910090924@mojatatu.com Signed-off-by: Jakub Kicinski --- net/sched/cls_api.c | 15 ++++++++++++++- 1 file changed, 14 insertions(+), 1 deletion(-) diff --git a/net/sched/cls_api.c b/net/sched/cls_api.c index c47d2ee13641..a9f54988561f 100644 --- a/net/sched/cls_api.c +++ b/net/sched/cls_api.c @@ -2463,7 +2463,20 @@ static int tc_new_tfilter(struct sk_buff *skb, struct nlmsghdr *n, } errout: - if (err && tp_state == TP_CREATED) + if (err && !IS_ERR_OR_NULL(tp) && + (tp_state == TP_CREATED || tp->ops->delete_empty)) + /* + * The request is dropping its reference to tp. If it was + * the last user (the idr is empty now), reclaim the proto. + * A tp this request created is reclaimed unconditionally: + * it is the only owner, so marking it for deletion is + * safe. Otherwise only classifiers with a delete_empty + * callback are reclaimed -- the callback admits an empty + * proto only, so a live shared proto is never unlinked. + * Classifiers without one (deleting is set + * unconditionally) are rtnl-serialized, so the raced + * window this guard closes cannot arise for them. + */ tcf_chain_tp_delete_empty(chain, tp, rtnl_held, NULL); errout_tp: if (chain) { From 0e7fe2b0ca45343b53345edc174898b448ccdaa4 Mon Sep 17 00:00:00 2001 From: Nicolai Buchwitz Date: Thu, 24 Sep 2026 12:19:21 +0200 Subject: [PATCH 285/365] net: bcmgenet: allocate RX buffers as page fragments Since the page_pool conversion every RX buffer is a whole page and the skb truesize is the page, although the hardware writes at most 2 KiB of it. On 64 KiB pages a packet therefore counts 65792 bytes against the socket buffer where it used to count 2752. As a result a UDP socket with the default buffer starts to drop after three packets, and the RX rings pin 16 MiB for 512 KiB of buffers. Fix this and allocate the buffers as page fragments, so the truesize is what a packet occupies. 4 KiB pages stay one buffer per page. Sync each buffer in the refill path, as page_pool can only sync a whole page on recycle. On 4 KiB pages that is twice what the hardware wrote. Fixes: 7bc054c2d4ed ("net: bcmgenet: convert RX path to page_pool") Reported-by: Karl Mehltretter Closes: https://lore.kernel.org/all/20260924065839.56793-1-kmehltretter@gmail.com/ Signed-off-by: Nicolai Buchwitz Reviewed-by: Simon Horman Link: https://patch.msgid.link/20260924101922.2675127-1-nb@tipi-net.de Signed-off-by: Jakub Kicinski --- .../net/ethernet/broadcom/genet/bcmgenet.c | 29 ++++++++++++++----- .../net/ethernet/broadcom/genet/bcmgenet.h | 2 ++ 2 files changed, 23 insertions(+), 8 deletions(-) diff --git a/drivers/net/ethernet/broadcom/genet/bcmgenet.c b/drivers/net/ethernet/broadcom/genet/bcmgenet.c index 21668e41b696..f725d26e6020 100644 --- a/drivers/net/ethernet/broadcom/genet/bcmgenet.c +++ b/drivers/net/ethernet/broadcom/genet/bcmgenet.c @@ -57,6 +57,9 @@ */ #define GENET_RSB_PAD (sizeof(struct status_64) + 2) +/* RX buffer plus the skb_shared_info napi_build_skb() places behind it */ +#define GENET_RX_BUF_SIZE SKB_HEAD_ALIGN(RX_BUF_LENGTH) + /* Tx/Rx DMA register offset, skip 256 descriptors */ #define WORDS_PER_BD(p) (p->hw_params->words_per_bd) #define DMA_DESC_SIZE (WORDS_PER_BD(priv) * sizeof(u32)) @@ -2254,11 +2257,12 @@ static int bcmgenet_rx_refill(struct bcmgenet_rx_ring *ring, struct enet_cb *cb) { struct bcmgenet_priv *priv = ring->priv; + unsigned int size = GENET_RX_BUF_SIZE; + unsigned int offset; dma_addr_t mapping; struct page *page; - page = page_pool_alloc_pages(ring->page_pool, - GFP_ATOMIC); + page = page_pool_dev_alloc(ring->page_pool, &offset, &size); if (!page) { priv->mib.alloc_rx_buff_failed++; netif_err(priv, rx_err, priv->dev, @@ -2267,9 +2271,13 @@ static int bcmgenet_rx_refill(struct bcmgenet_rx_ring *ring, } /* page_pool handles DMA mapping via PP_FLAG_DMA_MAP */ - mapping = page_pool_get_dma_addr(page); + mapping = page_pool_get_dma_addr(page) + offset; + dma_sync_single_for_device(&priv->pdev->dev, mapping, RX_BUF_LENGTH, + DMA_FROM_DEVICE); cb->rx_page = page; + cb->rx_offset = offset; + cb->rx_size = size; dmadesc_set_addr(priv, cb->bd_addr, mapping); return 0; @@ -2323,6 +2331,7 @@ static unsigned int bcmgenet_desc_rx(struct bcmgenet_rx_ring *ring, while ((rxpktprocessed < rxpkttoprocess) && (rxpktprocessed < budget)) { + unsigned int rx_offset, rx_size; struct status_64 *status; struct page *rx_page; void *hard_start; @@ -2332,6 +2341,8 @@ static unsigned int bcmgenet_desc_rx(struct bcmgenet_rx_ring *ring, /* Save the received page before refilling */ rx_page = cb->rx_page; + rx_offset = cb->rx_offset; + rx_size = cb->rx_size; if (bcmgenet_rx_refill(ring, cb)) { BCMGENET_STATS64_INC(stats, dropped); @@ -2341,10 +2352,10 @@ static unsigned int bcmgenet_desc_rx(struct bcmgenet_rx_ring *ring, /* Sync the full buffer; the HW may have written anywhere * up to RX_BUF_LENGTH. */ - page_pool_dma_sync_for_cpu(ring->page_pool, rx_page, 0, + page_pool_dma_sync_for_cpu(ring->page_pool, rx_page, rx_offset, RX_BUF_LENGTH); - hard_start = page_address(rx_page); + hard_start = page_address(rx_page) + rx_offset; status = (struct status_64 *)hard_start; dma_length_status = status->length_status; @@ -2410,7 +2421,7 @@ static unsigned int bcmgenet_desc_rx(struct bcmgenet_rx_ring *ring, /* Build SKB from the page - data starts at hard_start, * frame begins after RSB(64) + pad(2) = 66 bytes. */ - skb = napi_build_skb(hard_start, PAGE_SIZE); + skb = napi_build_skb(hard_start, rx_size); if (unlikely(!skb)) { BCMGENET_STATS64_INC(stats, dropped); page_pool_put_full_page(ring->page_pool, rx_page, @@ -2762,14 +2773,16 @@ static void bcmgenet_init_tx_ring(struct bcmgenet_priv *priv, static int bcmgenet_rx_ring_create_pool(struct bcmgenet_priv *priv, struct bcmgenet_rx_ring *ring) { + /* Buffers share a page. bcmgenet_rx_refill() syncs each one for the + * device, PP_FLAG_DMA_SYNC_DEV would sync the whole page. + */ struct page_pool_params pp_params = { .order = 0, - .flags = PP_FLAG_DMA_MAP | PP_FLAG_DMA_SYNC_DEV, + .flags = PP_FLAG_DMA_MAP, .pool_size = ring->size, .nid = NUMA_NO_NODE, .dev = &priv->pdev->dev, .dma_dir = DMA_FROM_DEVICE, - .max_len = RX_BUF_LENGTH, }; int err; diff --git a/drivers/net/ethernet/broadcom/genet/bcmgenet.h b/drivers/net/ethernet/broadcom/genet/bcmgenet.h index 22a958ba9902..86f2aed20dbe 100644 --- a/drivers/net/ethernet/broadcom/genet/bcmgenet.h +++ b/drivers/net/ethernet/broadcom/genet/bcmgenet.h @@ -470,6 +470,8 @@ struct bcmgenet_rx_stats64 { struct enet_cb { struct sk_buff *skb; struct page *rx_page; + unsigned int rx_offset; + unsigned int rx_size; void __iomem *bd_addr; DEFINE_DMA_UNMAP_ADDR(dma_addr); DEFINE_DMA_UNMAP_LEN(dma_len); From a7bfaba4823e3c165bb2004c74eff7c096672bc7 Mon Sep 17 00:00:00 2001 From: Qishuai Liu Date: Wed, 23 Sep 2026 13:20:48 +0000 Subject: [PATCH 286/365] ipv6: fix prefix route expiry in modify_prefix_route() modify_prefix_route() is given the lifetime in clock_t relative to now, but fib6_set_expires() wants an absolute jiffies value. So when a permanent address is changed to a finite valid_lft, the prefix route ends up already expired and GC removes it. Steps to reproduce: ip link add dummy9 type dummy ip link set dummy9 up ip -6 addr add 2001:db8:9::1/64 dev dummy9 ip -6 addr change 2001:db8:9::1/64 dev dummy9 valid_lft 3600 preferred_lft 3600 ip -6 route show dev dummy9 # expires is negative Fixes: 8308f3ff1753 ("net/ipv6: Add support for specifying metric of connected routes") Signed-off-by: Qishuai Liu Reviewed-by: Hangbin Liu Reviewed-by: Ido Schimmel Link: https://patch.msgid.link/20260923132048.3272878-1-lqs@lqs.me Signed-off-by: Jakub Kicinski --- net/ipv6/addrconf.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c index 9d89be7e0544..c90ee6dd7446 100644 --- a/net/ipv6/addrconf.c +++ b/net/ipv6/addrconf.c @@ -4874,7 +4874,7 @@ static int modify_prefix_route(struct net *net, struct inet6_ifaddr *ifp, fib6_clean_expires(f6i); fib6_may_remove_gc_list(net, f6i); } else { - fib6_set_expires(f6i, expires); + fib6_set_expires(f6i, jiffies + clock_t_to_jiffies(expires)); fib6_add_gc_list(f6i); } From 12c1f6e03f944e399bd2c88441dca5dc702b95a5 Mon Sep 17 00:00:00 2001 From: Sean Christopherson Date: Wed, 23 Sep 2026 09:37:20 -0700 Subject: [PATCH 287/365] KVM: SEV: Free have_run_cpus during VM destruction even if VM is no longer SEV Unconditionally free SEV's "have run CPUs" cpumask in the VM destroy path, i.e. even for what appear to be non-SEV VMs, as an SEV VM becomes a non-SEV VM if its state is intra-host migrated. Alternatively, the mask could be freed in sev_migrate_from() when "converting" the source VM, but that gets annoying because ideally KVM would nullify the mask to guard against UAF, and nullifying the mask would need be conditioned on CPUMASK_OFFSTACK=y. Freeing the mask during sev_migrate_from() is also not robust against other KVM bugs, though that's kind of a moot point since any such bugs would show up even if sev->active is never set. I.e. KVM must get that side of things correct. But, that's not a great reason to add more code just to make things marginally less robust. Fixes: 6f38f8c57464 ("KVM: SVM: Flush cache only on CPUs running SEV guest") Cc: stable@vger.kernel.org Reported-by: Stefan Teodorescu Signed-off-by: Sean Christopherson Message-ID: <20260923163721.1584779-2-seanjc@google.com> Signed-off-by: Paolo Bonzini --- arch/x86/kvm/svm/sev.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/arch/x86/kvm/svm/sev.c b/arch/x86/kvm/svm/sev.c index 5705723f1f41..cdc1c04f60da 100644 --- a/arch/x86/kvm/svm/sev.c +++ b/arch/x86/kvm/svm/sev.c @@ -2980,13 +2980,17 @@ void sev_vm_destroy(struct kvm *kvm) struct list_head *head = &sev->regions_list; struct list_head *pos, *q; + /* + * Free the mask even if the VM is not *currently* an SEV VM, as it may + * have been an SEV VM prior to intra-host migration. + */ + free_cpumask_var(sev->have_run_cpus); + if (!sev_guest(kvm)) return; WARN_ON(!list_empty(&sev->mirror_vms)); - free_cpumask_var(sev->have_run_cpus); - /* * If this is a mirror VM, remove it from the owner's list of a mirrors * and skip ASID cleanup (the ASID is tied to the lifetime of the owner). From 93de2a6a4b91b72607136dd656edf03fb399d27f Mon Sep 17 00:00:00 2001 From: Sean Christopherson Date: Wed, 23 Sep 2026 09:37:21 -0700 Subject: [PATCH 288/365] KVM: SEV: Do cache maintenance on the source VM during intra-host migration Manually perform cache maintenance on the source VM during intra-host migration to ensure no stale data is left in CPU caches after the VM is destroyed. Because the source VM is "converted" to a non-SEV VM, KVM's memory reclaim flows won't trigger cache maintenance, e.g. when all guest memory is reclaimed in response to detaching from the mmu_notifier. Note, relying on the destination VM to do cache maintenance isn't an option as KVM doesn't require identical guest memory configurations, i.e. the source VM may have access to memory that the destination VM does not. Enforcing equivalent memory configurations is infeasible, as it would require a *deep* comparison of memslots, e.g. to verify that not only are the memslot identical, but what the memslots point at is also identical. Fixes: b56639318bb2 ("KVM: SEV: Add support for SEV intra host migration") Cc: stable@vger.kernel.org Reported-by: Stefan Teodorescu Signed-off-by: Sean Christopherson Message-ID: <20260923163721.1584779-3-seanjc@google.com> Signed-off-by: Paolo Bonzini --- arch/x86/kvm/svm/sev.c | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/arch/x86/kvm/svm/sev.c b/arch/x86/kvm/svm/sev.c index cdc1c04f60da..63eb2155a774 100644 --- a/arch/x86/kvm/svm/sev.c +++ b/arch/x86/kvm/svm/sev.c @@ -2048,6 +2048,12 @@ static void sev_migrate_from(struct kvm *dst_kvm, struct kvm *src_kvm) src->pages_locked = 0; src->es_active = false; + /* + * Do cache maintenance on the source VM as it is no longer an SEV VM, + * i.e. memory reclaim flows won't trigger cache maintenance on the VM. + */ + sev_writeback_caches(src_kvm); + list_cut_before(&dst->regions_list, &src->regions_list, &src->regions_list); mutex_lock(&sev_mirror_lock); @@ -2187,6 +2193,10 @@ int sev_vm_move_enc_context_from(struct kvm *kvm, unsigned int source_fd) * the set of CPUs from the source. If a CPU was used to run a vCPU in * the source VM but is never used for the destination VM, then the CPU * can only have cached memory that was accessible to the source VM. + * Furthermore, KVM *must* perform cache maintenance on the source VM, + * as the source VM may have access to memory that the destination VM + * does not, i.e. KVM could skip flushes if memory is reclaimed from + * the old VM but not the new VM. */ if (!zalloc_cpumask_var(&dst_sev->have_run_cpus, GFP_KERNEL_ACCOUNT)) { ret = -ENOMEM; From 31c88350b7dd1522792f726f79607f31bb55c50f Mon Sep 17 00:00:00 2001 From: Hui Peng Date: Thu, 24 Sep 2026 04:27:27 +0000 Subject: [PATCH 289/365] cgroup/cpuset: Return PERR_NOCPUS in remote_partition_enable() on subpartitions_cpus conflict When a remote partition is created underneath an existing local partition via a non-partition (PRS_MEMBER) intermediate cgroup, update_prstate() sees parent->partition_root_state == PRS_MEMBER and calls remote_partition_enable(). Commit 86888c7bd117 ("cgroup/cpuset: Add warnings to catch inconsistency in exclusive CPUs") replaced the cpumask_intersects(tmp->new_cpus, subpartitions_cpus) error check in remote_partition_enable() with WARN_ON_ONCE(). As a result, remote_partition_enable() emits a warning and proceeds to enable the remote partition on CPUs that are already owned by the ancestor local partition in subpartitions_cpus. This can be reproduced on Linux 7.3.0-rc3 with: mkdir -p /tmp/cg1 mount -t cgroup2 none /tmp/cg1 echo "+cpuset" > /tmp/cg1/cgroup.subtree_control mkdir /tmp/cg1/A echo 1 > /tmp/cg1/A/cpuset.cpus echo 1 > /tmp/cg1/A/cpuset.cpus.exclusive echo root > /tmp/cg1/A/cpuset.cpus.partition echo "+cpuset" > /tmp/cg1/A/cgroup.subtree_control mkdir /tmp/cg1/A/B echo 1 > /tmp/cg1/A/B/cpuset.cpus echo 1 > /tmp/cg1/A/B/cpuset.cpus.exclusive echo "+cpuset" > /tmp/cg1/A/B/cgroup.subtree_control mkdir /tmp/cg1/A/B/D echo 1 > /tmp/cg1/A/B/D/cpuset.cpus echo 1 > /tmp/cg1/A/B/D/cpuset.cpus.exclusive echo root > /tmp/cg1/A/B/D/cpuset.cpus.partition which triggers: WARNING: kernel/cgroup/cpuset.c:1594 at remote_partition_enable+0x1c1/0x300 and leaves both /tmp/cg1/A and /tmp/cg1/A/B/D as active root partitions claiming exclusive CPU 1. Fix this by returning PERR_NOCPUS when tmp->new_cpus intersects subpartitions_cpus in remote_partition_enable(), matching the error code used by remote_cpus_update() for the same subpartitions_cpus conflict, and add a regression test case to tools/testing/selftests/cgroup/test_cpuset_prs.sh. Tested in QEMU on Linux 7.3.0-rc3 using the reproducer above and tools/testing/selftests/cgroup/test_cpuset_prs.sh. Fixes: 86888c7bd117 ("cgroup/cpuset: Add warnings to catch inconsistency in exclusive CPUs") Suggested-by: Guopeng Zhang Cc: stable@vger.kernel.org Assisted-by: LLM Signed-off-by: Hui Peng Reviewed-by: Waiman Long Signed-off-by: Tejun Heo --- kernel/cgroup/cpuset.c | 3 ++- tools/testing/selftests/cgroup/test_cpuset_prs.sh | 2 ++ 2 files changed, 4 insertions(+), 1 deletion(-) diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c index 8f24171b6055..1fcec89a28b9 100644 --- a/kernel/cgroup/cpuset.c +++ b/kernel/cgroup/cpuset.c @@ -1591,10 +1591,11 @@ static int remote_partition_enable(struct cpuset *cs, int new_prs, * above it or remote partition root underneath it is not allowed. */ compute_excpus(cs, tmp->new_cpus); - WARN_ON_ONCE(cpumask_intersects(tmp->new_cpus, subpartitions_cpus)); if (!cpumask_intersects(tmp->new_cpus, cpu_active_mask) || cpumask_subset(top_cpuset.effective_cpus, tmp->new_cpus)) return PERR_INVCPUS; + if (cpumask_intersects(tmp->new_cpus, subpartitions_cpus)) + return PERR_NOCPUS; if (((new_prs == PRS_ISOLATED) && !isolated_cpus_can_update(tmp->new_cpus, NULL)) || prstate_housekeeping_conflict(new_prs, tmp->new_cpus)) diff --git a/tools/testing/selftests/cgroup/test_cpuset_prs.sh b/tools/testing/selftests/cgroup/test_cpuset_prs.sh index 131d8b4551ef..7efd5e645767 100755 --- a/tools/testing/selftests/cgroup/test_cpuset_prs.sh +++ b/tools/testing/selftests/cgroup/test_cpuset_prs.sh @@ -298,6 +298,8 @@ TEST_MATRIX=( " C0-4:X2-4 C1-4:X2-4:P2 C2-4:X4:P1 \ . . . X1 . 0 A1:0-1|A2:2-4|A3:2-4 \ A1:P0|A2:P2|A3:P-1 2-4" + " CX1-3:P1 CX1-3 CX1-3 . . . P1 . 0 A1:1-3|A2:1-3|A3:1-3 \ + A1:P1|A2:P0|A3:P-1" # Remote partition offline tests " C0-3 C1-3 C2-3 . X2-3 X2-3 X2-3:P2:O2=0 . 0 A1:0-1|A2:1|A3:3 A1:P0|A3:P2 2-3" From 94480606a677deb68d5622cf0ded88626514b3f2 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Sat, 26 Sep 2026 09:23:02 -1000 Subject: [PATCH 290/365] sched_ext: Add a size argument to scx_bpf_cid_topo() so struct scx_cid_topo can grow scx_bpf_cid_topo() copies struct scx_cid_topo into a buffer the BPF program sized from its own vmlinux.h while the verifier sizes the write from the running kernel's BTF. The struct may grow and each growth then breaks every scheduler built against the older layout, rejected at load or written past its buffer. This is the usual hole for a struct handed to BPF, closed elsewhere with a size argument, and it was missed here. Take the buffer size, copy the smaller of it and the kernel's struct and set the rest to -1. Accesses to the copy are CO-RE relocated, so the struct can grow by appending fields, which its comment now states. The kfunc changes in place: the cid interface is still being finalized and no released scheduler uses the current form. Fixes: e9b55af47edf ("sched_ext: Add topological CPU IDs (cids)") Cc: stable@vger.kernel.org # v7.2+ Signed-off-by: Tejun Heo Reviewed-by: Andrea Righi --- kernel/sched/ext/cid.c | 26 +++++++++++++++--------- kernel/sched/ext/types.h | 4 ++++ tools/sched_ext/include/scx/common.bpf.h | 2 +- 3 files changed, 21 insertions(+), 11 deletions(-) diff --git a/kernel/sched/ext/cid.c b/kernel/sched/ext/cid.c index 39f88deb94bc..d4d5f537832a 100644 --- a/kernel/sched/ext/cid.c +++ b/kernel/sched/ext/cid.c @@ -912,30 +912,36 @@ bool scx_cmask_empty(const struct scx_cmask *m) /** * scx_bpf_cid_topo - Copy out per-cid topology info * @cid: cid to look up - * @out__uninit: where to copy the topology info; fully written by this call + * @out: where to copy the topology info + * @out__sz: size of @out, the program's sizeof(struct scx_cid_topo) * @aux: implicit BPF argument to access bpf_prog_aux hidden from BPF progs * - * Fill @out__uninit with the topology info for @cid. Trigger scx_error() if - * @cid is out of range. If @cid is valid but in the no-topo section, all fields - * are set to -1. All fields are also set to -1 when no cid tables have been - * published yet, which a program may observe while racing the root enable. + * Fill @out with the topology info for @cid. Trigger scx_error() if @cid is out + * of range. If @cid is valid but in the no-topo section, all fields are set to + * -1. All fields are also set to -1 when no cid tables have been published yet, + * which a program may observe while racing the root enable. + * + * The program's struct may be older or newer than the kernel's. The smaller of + * @out__sz and the kernel's size is copied and the rest of @out is set to -1. */ -__bpf_kfunc void scx_bpf_cid_topo(s32 cid, struct scx_cid_topo *out__uninit, +__bpf_kfunc void scx_bpf_cid_topo(s32 cid, struct scx_cid_topo *out, size_t out__sz, const struct bpf_prog_aux *aux) { + size_t len = min(out__sz, sizeof(*out)); struct scx_cid_topo *topo; struct scx_sched *sch; + /* the error cases and fields the kernel lacks read as -1 */ + memset(out, 0xff, out__sz); + guard(rcu)(); sch = scx_prog_sched(aux); topo = rcu_dereference(scx_cid_topo); - if (unlikely(!sch) || !cid_valid(sch, cid) || unlikely(!topo)) { - *out__uninit = SCX_CID_TOPO_NEG; + if (unlikely(!sch) || !cid_valid(sch, cid) || unlikely(!topo)) return; - } - *out__uninit = topo[cid]; + memcpy(out, &topo[cid], len); } __bpf_kfunc_end_defs(); diff --git a/kernel/sched/ext/types.h b/kernel/sched/ext/types.h index 943d8d429a2c..139176cf9fc6 100644 --- a/kernel/sched/ext/types.h +++ b/kernel/sched/ext/types.h @@ -70,6 +70,10 @@ enum scx_consts { * smaller shards if the LLC exceeds the target size. No-topo cids are packed * into their own max-sized shards. * + * New fields are appended, never inserted: scx_bpf_cid_topo() copies this + * struct out sized by the program's own layout, and an older program's copy + * must stay a prefix of the kernel's. + * * @core_cid: first cid of this cid's core (smt-sibling group) * @core_idx: global index of that core, in [0, nr_cores_at_init) * @llc_cid: first cid of this cid's LLC diff --git a/tools/sched_ext/include/scx/common.bpf.h b/tools/sched_ext/include/scx/common.bpf.h index 2ddb01a059fd..22f24ebef8a9 100644 --- a/tools/sched_ext/include/scx/common.bpf.h +++ b/tools/sched_ext/include/scx/common.bpf.h @@ -106,7 +106,7 @@ u64 scx_bpf_now(void) __ksym __weak; void scx_bpf_events(struct scx_event_stats *events, size_t events__sz) __ksym __weak; s32 scx_bpf_cpu_to_cid(s32 cpu) __ksym __weak; s32 scx_bpf_cid_to_cpu(s32 cid) __ksym __weak; -void scx_bpf_cid_topo(s32 cid, struct scx_cid_topo *out) __ksym __weak; +void scx_bpf_cid_topo(s32 cid, struct scx_cid_topo *out, size_t out__sz) __ksym __weak; void scx_bpf_kick_cid(s32 cid, u64 flags) __ksym __weak; s32 scx_bpf_task_cid(const struct task_struct *p) __ksym __weak; s32 scx_bpf_this_cid(void) __ksym __weak; From c79cf15eb35a4c703bf18f8c33c47e80dd660f8e Mon Sep 17 00:00:00 2001 From: Hao Ge Date: Thu, 27 Aug 2026 11:05:03 +0800 Subject: [PATCH 291/365] module: fix lost error code from codetag_load_module() If codetag_load_module() fails, err is not set to reflect the failure and load_module() returns 0 after the module has been torn down. Also, if the module is a livepatch, mod->klp_info allocated by copy_module_elf() leaks on this error path. Free it via a new livepatch_cleanup label. Link: https://lore.kernel.org/20260827030503.49171-1-hao.ge@linux.dev Fixes: 044d2aee6c57 ("alloc_tag: handle module codetag load errors as module load failures") Signed-off-by: Hao Ge Signed-off-by: Andrew Morton Reported-by: Sashiko Suggested-by: Petr Pavlu Reviewed-by: Bradley Morgan Cc: Aaron Tomlin Cc: Luis Chamberalin Cc: Sami Tolvanen Cc: Suren Baghdasaryan Cc: --- kernel/module/main.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/kernel/module/main.c b/kernel/module/main.c index d0e1e0bd2ad0..c1b34dc1e89a 100644 --- a/kernel/module/main.c +++ b/kernel/module/main.c @@ -3581,8 +3581,9 @@ static int load_module(struct load_info *info, const char __user *uargs, goto sysfs_cleanup; } - if (codetag_load_module(mod)) - goto sysfs_cleanup; + err = codetag_load_module(mod); + if (err) + goto livepatch_cleanup; /* Get rid of temporary copy. */ free_copy(info, flags); @@ -3592,6 +3593,9 @@ static int load_module(struct load_info *info, const char __user *uargs, return do_init_module(mod); + livepatch_cleanup: + if (is_livepatch_module(mod)) + free_module_elf(mod); sysfs_cleanup: mod_sysfs_teardown(mod); coming_cleanup: From 72c3d682c165b836b54a3d07a01f72e39c47022c Mon Sep 17 00:00:00 2001 From: Baolin Wang Date: Mon, 14 Sep 2026 10:12:41 +0800 Subject: [PATCH 292/365] mm: shmem: ignore sysfs configs for shmem forced collapse According to Documentation/mm/transhuge.rst, MADV_COLLAPSE is expected to ignore any THP or mTHP interface settings. However, after commit 26c7d8413aaf ("mm: thp: support "THPeligible" semantics for mTHP with anonymous shmem"), performing MADV_COLLAPSE on shmem will depend on /sys/.../hugepages-2048kB/shmem_enabled being set to "inherit" (although that is the default), which can cause MADV_COLLAPSE to fail unexpectedly. This could cause a userspace-visible performance regression. Fix this by returning the result of shmem_huge_global_enabled() directly when MADV_COLLAPSE is requested, allowing PMD-order collapse while ignoring shmem THP/mTHP settings. Link: https://lore.kernel.org/063f655b4d6c4234f3aa27ed6ecab10283ecb880.1789351825.git.baolin.wang@linux.alibaba.com Fixes: 26c7d8413aaf ("mm: thp: support "THPeligible" semantics for mTHP with anonymous shmem") Signed-off-by: Baolin Wang Signed-off-by: Andrew Morton Reported-by: Kiryl Shutsemau (Meta) Closes: https://lore.kernel.org/all/20260908125105.1510704-7-kirill@shutemov.name/ Reviewed-by: Kiryl Shutsemau (Meta) Acked-by: Zi Yan Reviewed-by: Lance Yang Reviewed-by: Barry Song Acked-by: Qi Zheng Acked-by: David Hildenbrand (Arm) Reviewed-by: Lorenzo Stoakes (ARM) Cc: Dev Jain Cc: Hugh Dickins Cc: Liam R. Howlett Cc: Ryan Roberts Cc: --- mm/shmem.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/mm/shmem.c b/mm/shmem.c index 897fa2b61346..848316eaa7f4 100644 --- a/mm/shmem.c +++ b/mm/shmem.c @@ -1847,8 +1847,11 @@ unsigned long shmem_allowable_huge_orders(struct inode *inode, global_orders = shmem_huge_global_enabled(inode, index, write_end, shmem_huge_force, vma, vm_flags); - /* Tmpfs huge pages allocation */ - if (!vma || !vma_is_anon_shmem(vma)) + /* + * Tmpfs huge pages allocation or forced collapse ignores + * sysfs configs. + */ + if (!vma || !vma_is_anon_shmem(vma) || shmem_huge_force) return global_orders; /* From 347c6ed8cf7768883252e106994de62c7947f0a0 Mon Sep 17 00:00:00 2001 From: Arnd Bergmann Date: Tue, 15 Sep 2026 22:23:47 +0200 Subject: [PATCH 293/365] alloc_tag: avoid implicit padding in uapi The implied padding causes a harmless warning when testing the uapi headers with -Wpadded that could in theory indicate incompatibilities or data leaks: ./usr/include/linux/alloc_tag.h:41:1: error: padding struct size to alignment boundary with 7 bytes [-Werror=padded] The code here is fine, but it's better to make the padding explicit and avoid the warning here. Link: https://lore.kernel.org/20260916065830.1619425-1-arnd@kernel.org Link: https://lore.kernel.org/20260915202404.3568029-1-arnd@kernel.org Fixes: 1d581ab2348c ("alloc_tag: add ioctl to /proc/allocinfo") Signed-off-by: Arnd Bergmann Signed-off-by: Andrew Morton Acked-by: Suren Baghdasaryan Acked-by: SJ Park Acked-by: Hao Ge --- include/uapi/linux/alloc_tag.h | 1 + 1 file changed, 1 insertion(+) diff --git a/include/uapi/linux/alloc_tag.h b/include/uapi/linux/alloc_tag.h index e3ad94444864..7d4618bea043 100644 --- a/include/uapi/linux/alloc_tag.h +++ b/include/uapi/linux/alloc_tag.h @@ -38,6 +38,7 @@ struct allocinfo_counter { __u64 bytes; __u64 calls; __u8 accurate; + __u8 pad[7]; } __attribute__((aligned(8))); struct allocinfo_tag_data { From 52ae167ce16609c7e6fffef588b3c2c26de2db19 Mon Sep 17 00:00:00 2001 From: SJ Park Date: Wed, 16 Sep 2026 06:50:19 -0700 Subject: [PATCH 294/365] mm/damon/core: don't skip damos_adjust_quota() while esz is not zero DAMOS could unexpectedly stop working when a user disables quota using the online parameters commit feature. Fix it by correcting a wrong quota unset check in damos_adjust_quota(). DAMON users could disable all quotas by unsetting time and size quotas, and removing all quota goals. The intention of disabling quotas would be making DAMOS run at full speed. When such quota disabled setup is detected, damos_adjust_quota() skips all its work. The skipped works include effective size quota (damos_quota->esz) updates and charged quota amount (damos_quota->charged_sz) resets. The intention is to avoid doing unnecessary work when quotas are disabled. However, users could do the setup while effective size quota is non-zero, by doing the disabling with the online DAMON parameters commit feature. In this case, because the effective size quota exists, DAMOS will keep working with the quota until it is fully charged. After the effective quota is fully charged, the charged quota amount (damos_quota->charged_sz) cannot be reset because damos_adjust_quota() skips it. Then, DAMOS stops working until the quota is newly set or DAMON is entirely restarted. The problem happens because damos_adjust_quota() assumes the user setup for disabling quota immediately disabled it. In reality, the quota is still working until the effective size quota is also updated to zero. Other logic for catching that uses damos_quota_is_set(), which understands the fact and therefore checks the effective size quota in addition to the user setup. Fix the issue by using damos_quota_is_set() in damos_adjust_quota() to determine if its works should be skipped. The user impact is a non-deterministic and unexpected DAMOS stop behavior. That is, users would disable quotas using the online parameters commit feature, expecting DAMOS will run at full speed. However, depending on the timing, the setup can be updated while the effective size quota is non-zero. Due to the above mentioned internal mechanism, DAMOS stops working instead of running at full speed. It is unexpected behavior. It is also non-deterministic because sometimes the setup is done when the effective size quota is zero, depending on the timing. It doesn't cause critical issues like crashes or leaks. Users can simply set a reasonable quota again, or restart DAMON. But definitely it is an unexpected and non-deterministic behavior that makes it difficult to reliably use. Also investigating the root cause of the behavior would be quite difficult. Link: https://lore.kernel.org/20260916135020.86483-1-sj@kernel.org Fixes: da87878010e5 ("mm/damon/sysfs: support online inputs update") Signed-off-by: SJ Park Signed-off-by: Andrew Morton Reviewed-by: Kunwu Chan Cc: # 5.19.x --- mm/damon/core.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/mm/damon/core.c b/mm/damon/core.c index 1764620903f2..ab3c4d754964 100644 --- a/mm/damon/core.c +++ b/mm/damon/core.c @@ -3147,7 +3147,7 @@ static void damos_adjust_quota(struct damon_ctx *c, struct damos *s) unsigned long cumulated_sz, cached_esz; unsigned int score, max_score = 0; - if (!quota->ms && !quota->sz && list_empty("a->goals)) + if (!damos_quota_is_set(quota)) return; /* First charge window */ From 22ab0647764c38924669673634ae333578d77768 Mon Sep 17 00:00:00 2001 From: Andrey Ryabinin Date: Wed, 16 Sep 2026 19:51:13 +0200 Subject: [PATCH 295/365] kasan: unpoison task stack below watermark only in generic mode CPU resume and BPF exception handling can discard stack frames without running their compiler-generated epilogues. Generic KASAN needs kasan_unpoison_task_stack_below() to clear the redzones left behind by those frames before the stack is reused. CONFIG_KASAN_STACK also enables this helper for SW_TAGS. The helper derives the stack base from an untagged stack pointer, so kasan_unpoison() writes KASAN_TAG_KERNEL (0xff) into the shadow for [base, watermark). For a vmapped task stack, vm_area->addr still carries the original random allocation tag. This creates a tag mismatch at the stack base even if that memory has never held an instrumented stack object. When the task exits and its stack is not cached, thread_stack_free_rcu() passes vm_area->addr to vfree(). In RCU callback context this reaches vfree_atomic(), whose llist_add() writes to the allocation base through the tagged pointer and triggers a false KASAN invalid-access report: BUG: KASAN: invalid-access in vfree_atomic+0x90/0x150 Write of size 8 at addr c2ffffc0a8f70000 by task rcuop/7/75 Pointer tag: [c2], memory tag: [ff] Call trace: __hwasan_store8_noabort+0xe8/0xf8 vfree_atomic+0x90/0x150 vfree+0x220/0x298 thread_stack_free_rcu+0x3c/0x4c rcu_do_batch+0x308/0xaf0 rcu_nocb_cb_kthread+0x33c/0x708 The deferred-free path started using the tagged vm_area->addr in commit 449e0b4ed5a1 ("fork: clean-up naming of vm_stack/vm_struct variables in vmap stacks code"). Previously it freed the untagged pointer derived from tsk->stack, so the write was never tag-checked. SW_TAGS does not need the blanket shadow reset to make subsequent stack accesses valid: untagged kernel pointers have the match-all 0xff tag, and the compiler initializes the shadow tags of instrumented stack objects before use. Return early unless CONFIG_KASAN_GENERIC is enabled. Keep the mode check in the common helper so it covers both resume and bpf_throw(). Link: https://lore.kernel.org/20260916175113.1327454-1-ryabinin.a.a@gmail.com Fixes: 449e0b4ed5a1 ("fork: clean-up naming of vm_stack/vm_struct variables in vmap stacks code") Signed-off-by: Andrey Ryabinin Signed-off-by: Andrew Morton Reported-by: Shaobo Huang Closes: https://lore.kernel.org/all/20260806123020.90869-1-huangshaobo3@xiaomi.com/ Assisted-by: LLM Cc: Alexander Potapenko Cc: Andrey Konovalov Cc: David Hildenbrand Cc: Dmitry Vyukov Cc: Lorenzo Stoakes Cc: Vincenzo Frascino Cc: --- mm/kasan/common.c | 13 ++++++++++++- 1 file changed, 12 insertions(+), 1 deletion(-) diff --git a/mm/kasan/common.c b/mm/kasan/common.c index b7d05c2a6d93..7d21c4db68d2 100644 --- a/mm/kasan/common.c +++ b/mm/kasan/common.c @@ -114,12 +114,23 @@ void kasan_unpoison_task_stack(struct task_struct *task) /* Unpoison the stack for the current task beyond a watermark sp value. */ asmlinkage void kasan_unpoison_task_stack_below(const void *watermark) { + void *base; + + /* + * Only the generic mode needs this. In the tag-based mode the + * compiler fully initializes the shadow of stack variables on + * function entry, so stale tags left below the watermark are + * harmless. + */ + if (!IS_ENABLED(CONFIG_KASAN_GENERIC)) + return; + /* * Calculate the task stack base address. Avoid using 'current' * because this function is called by early resume code which hasn't * yet set up the percpu register (%gs). */ - void *base = (void *)((unsigned long)watermark & ~(THREAD_SIZE - 1)); + base = (void *)((unsigned long)watermark & ~(THREAD_SIZE - 1)); kasan_unpoison(base, watermark - base, false); } From 711d886fe76e28854814759f51ee175e279ef8c7 Mon Sep 17 00:00:00 2001 From: "David Hildenbrand (Arm)" Date: Fri, 18 Sep 2026 18:34:24 +0200 Subject: [PATCH 296/365] MAINTAINERS: split up MEMORY MANAGEMENT - MEMORY POLICY AND MIGRATION Patch series "MAINTAINERS: rework MEMORY MANAGEMENT - MEMORY POLICY". Let's rework MEMORY MANAGEMENT - MEMORY POLICY AND MIGRATION. As I can use some maintenance help in that area, add Gregory as a new co-maintainer for the split out MEMORY MANAGEMENT - NUMA PLACEMENT section. This patch (of 3): Let's split it up into MIGRATION and NUMA PLACEMENT. The latter is a better fitting description for mempolicy.c. Make a best guess about which pieces existing reviewers are interested in. Move the split sections to keep alphabetical order. Link: https://lore.kernel.org/20260918-maintainers-mempolicy-v1-1-9a94cac6d135@kernel.org Signed-off-by: David Hildenbrand (Arm) Signed-off-by: Andrew Morton Acked-by: Lorenzo Stoakes (ARM) Acked-by: Zi Yan Reviewed-by: Joshua Hahn Reviewed-by: SJ Park Acked-by: Vlastimil Babka (SUSE) Cc: Alistair Popple Cc: Byungchul Park Cc: Gregory Price Cc: "Huang, Ying" Cc: Liam R. Howlett Cc: Matthew Brost Cc: Michal Hocko Cc: Mike Rapoport Cc: Rakie Kim Cc: Suren Baghdasaryan --- MAINTAINERS | 62 ++++++++++++++++++++++++++++++++--------------------- 1 file changed, 37 insertions(+), 25 deletions(-) diff --git a/MAINTAINERS b/MAINTAINERS index 140eafcbbd78..3f70a28df4b1 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -17208,31 +17208,6 @@ F: include/trace/events/ksm.h F: mm/ksm.c F: mm/mm_slot.h -MEMORY MANAGEMENT - MEMORY POLICY AND MIGRATION -M: Andrew Morton -M: David Hildenbrand -R: Zi Yan -R: Matthew Brost -R: Joshua Hahn -R: Rakie Kim -R: Byungchul Park -R: Gregory Price -R: Ying Huang -R: Alistair Popple -L: linux-mm@kvack.org -S: Maintained -W: http://www.linux-mm.org -T: git git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm -F: Documentation/ABI/testing/sysfs-kernel-mm-mempolicy -F: Documentation/ABI/testing/sysfs-kernel-mm-mempolicy-weighted-interleave -F: include/linux/mempolicy.h -F: include/uapi/linux/mempolicy.h -F: include/linux/migrate.h -F: include/linux/migrate_mode.h -F: mm/mempolicy.c -F: mm/migrate.c -F: mm/migrate_device.c - MEMORY MANAGEMENT - MGLRU (MULTI-GEN LRU) M: Andrew Morton R: Kairui Song @@ -17255,6 +17230,25 @@ F: include/linux/mmzone.h F: mm/vmscan.c F: mm/workingset.c +MEMORY MANAGEMENT - MIGRATION +M: Andrew Morton +M: David Hildenbrand +R: Zi Yan +R: Matthew Brost +R: Joshua Hahn +R: Byungchul Park +R: Gregory Price +R: Ying Huang +R: Alistair Popple +L: linux-mm@kvack.org +S: Maintained +W: http://www.linux-mm.org +T: git git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm +F: include/linux/migrate.h +F: include/linux/migrate_mode.h +F: mm/migrate.c +F: mm/migrate_device.c + MEMORY MANAGEMENT - MISC M: Andrew Morton M: David Hildenbrand @@ -17307,6 +17301,24 @@ F: mm/numa.c F: mm/numa_emulation.c F: mm/numa_memblks.c +MEMORY MANAGEMENT - NUMA PLACEMENT +M: Andrew Morton +M: David Hildenbrand +R: Zi Yan +R: Joshua Hahn +R: Rakie Kim +R: Gregory Price +R: Ying Huang +L: linux-mm@kvack.org +S: Maintained +W: http://www.linux-mm.org +T: git git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm +F: Documentation/ABI/testing/sysfs-kernel-mm-mempolicy +F: Documentation/ABI/testing/sysfs-kernel-mm-mempolicy-weighted-interleave +F: include/linux/mempolicy.h +F: include/uapi/linux/mempolicy.h +F: mm/mempolicy.c + MEMORY MANAGEMENT - OOM KILLER M: Michal Hocko R: David Rientjes From 4050a73a90f456887dc35d5bbfa4f4790eb5c2cf Mon Sep 17 00:00:00 2001 From: "David Hildenbrand (Arm)" Date: Fri, 18 Sep 2026 18:34:25 +0200 Subject: [PATCH 297/365] MAINTAINERS: move memory tiering under MEMORY MANAGEMENT - NUMA PLACEMENT NUMA PLACEMENT is a better place for memory tiering. My best guess is that existing MISC reviewers are not that interested in memory tiering, so don't carry any over. Link: https://lore.kernel.org/20260918-maintainers-mempolicy-v1-2-9a94cac6d135@kernel.org Signed-off-by: David Hildenbrand (Arm) Signed-off-by: Andrew Morton Acked-by: Lorenzo Stoakes (ARM) Acked-by: Zi Yan Reviewed-by: Joshua Hahn Reviewed-byt: SJ Park Acked-by: Vlastimil Babka (SUSE) Cc: Alistair Popple Cc: Byungchul Park Cc: Gregory Price Cc: "Huang, Ying" Cc: Liam R. Howlett Cc: Matthew Brost Cc: Michal Hocko Cc: Mike Rapoport Cc: Rakie Kim Cc: Suren Baghdasaryan --- MAINTAINERS | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/MAINTAINERS b/MAINTAINERS index 3f70a28df4b1..d69f112e5abb 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -17264,7 +17264,6 @@ W: http://www.linux-mm.org T: git git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm F: Documentation/ABI/testing/sysfs-kernel-mm F: Documentation/ABI/testing/sysfs-kernel-mm-cma -F: Documentation/ABI/testing/sysfs-kernel-mm-memory-tiers F: Documentation/ABI/testing/sysfs-kernel-mm-numa F: Documentation/admin-guide/mm/ F: Documentation/mm/ @@ -17272,7 +17271,6 @@ F: drivers/char/mem.c F: include/linux/cma.h F: include/linux/dmapool.h F: include/linux/ioremap.h -F: include/linux/memory-tiers.h F: include/linux/page_idle.h F: mm/backing-dev.c F: mm/cma.c @@ -17284,7 +17282,6 @@ F: mm/early_ioremap.c F: mm/fadvise.c F: mm/ioremap.c F: mm/mapping_dirty_helpers.c -F: mm/memory-tiers.c F: mm/page_idle.c F: mm/pgalloc-track.h F: mm/process_vm_access.c @@ -17313,10 +17310,13 @@ L: linux-mm@kvack.org S: Maintained W: http://www.linux-mm.org T: git git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm +F: Documentation/ABI/testing/sysfs-kernel-mm-memory-tiers F: Documentation/ABI/testing/sysfs-kernel-mm-mempolicy F: Documentation/ABI/testing/sysfs-kernel-mm-mempolicy-weighted-interleave +F: include/linux/memory-tiers.h F: include/linux/mempolicy.h F: include/uapi/linux/mempolicy.h +F: mm/memory-tiers.c F: mm/mempolicy.c MEMORY MANAGEMENT - OOM KILLER From b3f0c1a4e41f356c3126a676ad55c0f6881e7514 Mon Sep 17 00:00:00 2001 From: "David Hildenbrand (Arm)" Date: Fri, 18 Sep 2026 18:34:26 +0200 Subject: [PATCH 298/365] MAINTAINERS: make Gregory a co-maintainer of MEMORY MANAGEMENT - NUMA PLACEMENT Gregory is extremely familiar with NUMA placement handling and volunteer to help co-maintain the numa placement bits. So add him as a co-maintainer. Link: https://lore.kernel.org/20260918-maintainers-mempolicy-v1-3-9a94cac6d135@kernel.org Signed-off-by: David Hildenbrand (Arm) Signed-off-by: Andrew Morton Acked-by: Lorenzo Stoakes (ARM) Acked-by: Gregory Price (Meta) Acked-by: Zi Yan Reviewed-by: Joshua Hahn Acked-by: SJ Park Acked-by: Vlastimil Babka (SUSE) Cc: Alistair Popple Cc: Byungchul Park Cc: "Huang, Ying" Cc: Liam R. Howlett Cc: Matthew Brost Cc: Michal Hocko Cc: Mike Rapoport Cc: Rakie Kim Cc: Suren Baghdasaryan --- MAINTAINERS | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/MAINTAINERS b/MAINTAINERS index d69f112e5abb..9fcda07a32a0 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -17301,10 +17301,10 @@ F: mm/numa_memblks.c MEMORY MANAGEMENT - NUMA PLACEMENT M: Andrew Morton M: David Hildenbrand +M: Gregory Price R: Zi Yan R: Joshua Hahn R: Rakie Kim -R: Gregory Price R: Ying Huang L: linux-mm@kvack.org S: Maintained From 6cc8f549f6888e36036ab14d84a8ec46b4a393c0 Mon Sep 17 00:00:00 2001 From: Joseph Qi Date: Wed, 23 Sep 2026 08:39:20 +0800 Subject: [PATCH 299/365] MAINTAINERS: add Heming Zhao as ocfs2 reviewer Heming Zhao has contributed ocfs2 a lot recent years, both as a author and reviewer. So add him as ocfs2 reviewer. Link: https://lore.kernel.org/20260923003920.2382398-1-joseph.qi@linux.alibaba.com Signed-off-by: Joseph Qi Signed-off-by: Andrew Morton Acked-by: Mark Fasheh Cc: Heming Zhao Cc: Joel Becker --- MAINTAINERS | 1 + 1 file changed, 1 insertion(+) diff --git a/MAINTAINERS b/MAINTAINERS index 9fcda07a32a0..23de9fc4c2a7 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -20568,6 +20568,7 @@ ORACLE CLUSTER FILESYSTEM 2 (OCFS2) M: Mark Fasheh M: Joel Becker M: Joseph Qi +R: Heming Zhao L: ocfs2-devel@lists.linux.dev S: Supported W: http://ocfs2.wiki.kernel.org From 976d5e0ddac885b0219252f004f6c561c344df9d Mon Sep 17 00:00:00 2001 From: "Lorenzo Stoakes (ARM)" Date: Wed, 23 Sep 2026 18:45:41 +0100 Subject: [PATCH 300/365] mm/vma: predicate setting mmap_prepare VMA fields on new vma alloc It only makes sense to manipulate VMA fields if a new VMA was allocated, rather than merged. VMA merging does not compare vm_ops or vm_private_data, so a merged VMA keeps its own, which is also what the legacy f_op->mmap path does since it never touches an existing VMA. Currently, these fields will get overwritten by whatever state is established in the mmap_prepare hook, and if the VMA was merged, vm_ops->mapped will not have been called, so this could destructively clear existing state without replacing it with anything valid. There is an implicit requirement that vm_private_data and vm_ops are fungible across VMAs which means that losing the 'new' state is fine. However in this case the 'old' state is being overwritten by potentially invalid 'new' state, so this must be rectified. Additionally constify have_mmap_prepare while here. All existing in-tree users either derive state from the file or are unmergeable due to VMA flags, so this has no direct impact. Link: https://lore.kernel.org/20260923-fix-mmap-prepare-overwrite-v1-1-3b3f1bfcdf5e@kernel.org Fixes: c84bf6dd2b83 ("mm: introduce new .mmap_prepare() file callback") Signed-off-by: Lorenzo Stoakes (ARM) Signed-off-by: Andrew Morton Reviewed-by: Suren Baghdasaryan Reviewed-by: Gregory Price (Meta) Acked-by: Zi Yan Acked-by: Vlastimil Babka (SUSE) Cc: Liam R. Howlett Cc: Jann Horn Cc: Pedro Falcato Cc: --- mm/vma.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/mm/vma.c b/mm/vma.c index 9f0a0acf694a..6cde67883fb0 100644 --- a/mm/vma.c +++ b/mm/vma.c @@ -2849,7 +2849,7 @@ static unsigned long __mmap_region(struct file *file, unsigned long addr, { struct mm_struct *mm = current->mm; struct vm_area_struct *vma = NULL; - bool have_mmap_prepare = file && file->f_op->mmap_prepare; + const bool have_mmap_prepare = file && file->f_op->mmap_prepare; VMA_ITERATOR(vmi, mm, addr); const pgoff_t anon_pgoff = addr >> PAGE_SHIFT; MMAP_STATE(map, mm, &vmi, addr, len, pgoff, anon_pgoff, vma_flags, file); @@ -2892,7 +2892,7 @@ static unsigned long __mmap_region(struct file *file, unsigned long addr, allocated_new = true; } - if (have_mmap_prepare) + if (have_mmap_prepare && allocated_new) set_vma_user_defined_fields(vma, &map); __mmap_complete(&map, vma); From db6365ced4d5855e321f772b240c0e473bcfcdd5 Mon Sep 17 00:00:00 2001 From: Pavankumar Kondeti Date: Fri, 25 Sep 2026 15:25:12 +0530 Subject: [PATCH 301/365] workqueue: Fix NULL current_pwq deref in flush dependency check check_flush_dependency() uses current_wq_worker() to determine whether the caller is a workqueue worker and then dereferences worker->current_pwq to test whether the current workqueue is WQ_MEM_RECLAIM. current_wq_worker() only means that %current has PF_WQ_WORKER set. A kworker can reach check_flush_dependency() while it is not executing a work item. One such path is worker_thread() acting as the pool manager, where create_worker() does GFP_KERNEL allocation and the allocation path invokes the OOM notifier. In that state worker->current_pwq is NULL because current_pwq is set only by process_one_work() and cleared again after the work function returns. [ 416.760634][ T375] Call trace: [ 416.760638][ T375] check_flush_dependency+0x80/0x120 (P) [ 416.760648][ T375] __flush_work+0x98/0x224 [ 416.760657][ T375] flush_work+0x30/0x44 [ 416.760665][ T375] ... [ 416.760710][ T375] blocking_notifier_call_chain+0x58/0xa0 [ 416.760719][ T375] out_of_memory+0xb4/0x458 [ 416.760730][ T375] __alloc_pages_may_oom+0x11c/0x1a8 [ 416.760739][ T375] __alloc_pages_slowpath+0x314/0x46c [ 416.760746][ T375] __alloc_frozen_pages_noprof+0x110/0x1a4 [ 416.760753][ T375] new_slab+0x12c/0x484 [ 416.760759][ T375] ___slab_alloc+0x7a8/0xc7c [ 416.760765][ T375] __slab_alloc+0x74/0xd8 [ 416.760772][ T375] __kmalloc_cache_node_noprof+0x2ac/0x304 [ 416.760779][ T375] alloc_worker+0x28/0x60 [ 416.760785][ T375] create_worker+0x4c/0x20c [ 416.760790][ T375] worker_thread+0xe8/0x2b8 [ 416.760796][ T375] kthread+0x1a8/0x200 [ 416.760805][ T375] ret_from_fork+0x10/0x20 Guard the WQ_MEM_RECLAIM-worker warning with worker->current_pwq. If the kworker is not currently executing a work item, there is no current workqueue to diagnose with that warning. The PF_MEMALLOC warning is left unchanged so explicit reclaim context flushing a !WQ_MEM_RECLAIM target is still reported. Fixes: fca839c00a12 ("workqueue: warn if memory reclaim tries to flush !WQ_MEM_RECLAIM workqueue") Cc: stable@vger.kernel.org Assisted-by: LLM Signed-off-by: Pavankumar Kondeti Signed-off-by: Tejun Heo --- kernel/workqueue.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/kernel/workqueue.c b/kernel/workqueue.c index b8bec1689b7a..4a73d305d88a 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -3918,7 +3918,7 @@ static void check_flush_dependency(struct workqueue_struct *target_wq, WARN_ONCE(current->flags & PF_MEMALLOC, "workqueue: PF_MEMALLOC task %d(%s) is flushing !WQ_MEM_RECLAIM %s:%ps", current->pid, current->comm, target_wq->name, target_func); - WARN_ONCE(worker && ((worker->current_pwq->wq->flags & + WARN_ONCE(worker && worker->current_pwq && ((worker->current_pwq->wq->flags & (WQ_MEM_RECLAIM | __WQ_LEGACY)) == WQ_MEM_RECLAIM), "workqueue: WQ_MEM_RECLAIM %s:%ps is flushing !WQ_MEM_RECLAIM %s:%ps", worker->current_pwq->wq->name, worker->current_func, From a442c8a89fb9c531f9d6ab86120c31decd778d51 Mon Sep 17 00:00:00 2001 From: Florian Westphal Date: Fri, 4 Sep 2026 20:03:07 +0200 Subject: [PATCH 302/365] netfilter: ipset: do not update comments from kernel-side adds 'Fixes' commit stopped calling ip_set_init_comment() for hash types from kernel-side-adds (xtables .. -j SET). ip_set_init_comment() says: "The kadt functions don't use the comment extensions in any way." But bitmap set type calls the function from kadt cb too. While this appears to be safe (serialized via the set spinlock), it seems better to not call the init function either, least of all to keep behaviour consistent. ip_set_list calls ip_set_init_comment() only from uadt cb, it can be kept as-is. This was triggered by yet another LLM review, hinting that the existing rcu_dereference_protected() cannot be downgraded to only check if the nfnl mutex is held. Fixes: f30415929be8 ("netfilter: ipset: do not update comments from kernel-side hash adds") Signed-off-by: Florian Westphal Signed-off-by: Pablo Neira Ayuso --- net/netfilter/ipset/ip_set_bitmap_gen.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/net/netfilter/ipset/ip_set_bitmap_gen.h b/net/netfilter/ipset/ip_set_bitmap_gen.h index d6a7e6604542..ae376fa3e7a3 100644 --- a/net/netfilter/ipset/ip_set_bitmap_gen.h +++ b/net/netfilter/ipset/ip_set_bitmap_gen.h @@ -159,7 +159,7 @@ mtype_add(struct ip_set *set, void *value, const struct ip_set_ext *ext, if (SET_WITH_COUNTER(set)) ip_set_init_counter(ext_counter(x, set), ext); - if (SET_WITH_COMMENT(set)) + if (SET_WITH_COMMENT(set) && !ext->target) ip_set_init_comment(set, ext_comment(x, set), ext); if (SET_WITH_SKBINFO(set)) ip_set_init_skbinfo(ext_skbinfo(x, set), ext); From 72d3fcf802c45d00b300f25b848a93c3a2bd7c7e Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Sun, 27 Sep 2026 13:55:01 -0700 Subject: [PATCH 303/365] Linux 7.3-rc5 --- Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/Makefile b/Makefile index 751a08643bf8..6b8b812f23b9 100644 --- a/Makefile +++ b/Makefile @@ -2,7 +2,7 @@ VERSION = 7 PATCHLEVEL = 3 SUBLEVEL = 0 -EXTRAVERSION = -rc4 +EXTRAVERSION = -rc5 NAME = Baby Opossum Posse # *DOCUMENTATION* From 4febc02cd02f021f2b6d0768332be643ef14a12d Mon Sep 17 00:00:00 2001 From: Felix Fietkau Date: Sat, 26 Sep 2026 14:02:16 +0200 Subject: [PATCH 304/365] wifi: mac80211: set info->band for 802.3 encap offload frames Unlike the 802.11 TX paths, ieee80211_8023_xmit() does not set info->band from the channel context on non-MLD interfaces. It stays at 0, so ieee80211_get_tx_rates() uses the wrong band for these frames. Set it the same way as ieee80211_build_hdr(): drop the frame if there is no channel context, and leave the band at 0 on MLD interfaces. Reported-by: Andrea Pesaresi Fixes: 50ff477a8639 ("mac80211: add 802.11 encapsulation offloading support") Cc: stable@vger.kernel.org Signed-off-by: Felix Fietkau Link: https://patch.msgid.link/20260926120216.4054712-1-nbd@nbd.name Signed-off-by: Johannes Berg --- net/mac80211/tx.c | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/net/mac80211/tx.c b/net/mac80211/tx.c index c8217f6d146e..b88acb18e04d 100644 --- a/net/mac80211/tx.c +++ b/net/mac80211/tx.c @@ -4735,6 +4735,7 @@ static void ieee80211_8023_xmit(struct ieee80211_sub_if_data *sdata, { struct ieee80211_tx_info *info; struct ieee80211_local *local = sdata->local; + struct ieee80211_chanctx_conf *chanctx_conf; struct tid_ampdu_tx *tid_tx = NULL; struct sk_buff *seg, *next; unsigned int skbs = 0, len = 0; @@ -4784,6 +4785,16 @@ static void ieee80211_8023_xmit(struct ieee80211_sub_if_data *sdata, sdata = container_of(sdata->bss, struct ieee80211_sub_if_data, u.ap); + /* MLD transmissions must not rely on the band */ + if (!ieee80211_vif_is_mld(&sdata->vif)) { + chanctx_conf = rcu_dereference(sdata->vif.bss_conf.chanctx_conf); + if (unlikely(!chanctx_conf)) { + kfree_skb_list(skb); + return; + } + info->band = chanctx_conf->def.chan->band; + } + info->flags |= IEEE80211_TX_CTL_HW_80211_ENCAP; info->control.vif = &sdata->vif; From 56e236759f0a8dc910c2ae7cd3c161f298ee5794 Mon Sep 17 00:00:00 2001 From: Ruide Cao Date: Wed, 23 Sep 2026 02:02:23 +0800 Subject: [PATCH 305/365] wifi: mac80211: reject invalid 320 MHz CSA bandwidth The HT/VHT channel definition validator can receive a 320 MHz bandwidth indication from a received CSA frame on a non-6 GHz link. It warns for this unsupported width but continues with an uninitialized vht_operation.chan_width, which is then read by ieee80211_chandef_vht_oper(). With panic_on_warn enabled, this lets a received frame panic the kernel. Reject the channel definition before entering the VHT operation conversion. This preserves the existing CSA fallback while avoiding both the warning and the uninitialized read. Fixes: 21c3f8f95554 ("wifi: mac80211: refactor STA CSA parsing flows") Cc: stable@vger.kernel.org Reported-by: Vega Assisted-by: LLM Signed-off-by: Ruide Cao Signed-off-by: Ren Wei Link: https://patch.msgid.link/bd68e360e06135c750397cb6be003a01a834a179.1787222001.git.vega.cover-letter@nebusec.ai Signed-off-by: Johannes Berg --- net/mac80211/spectmgmt.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/net/mac80211/spectmgmt.c b/net/mac80211/spectmgmt.c index 880f4625775d..f5e4970869b6 100644 --- a/net/mac80211/spectmgmt.c +++ b/net/mac80211/spectmgmt.c @@ -111,8 +111,8 @@ validate_chandef_by_ht_vht_oper(struct ieee80211_sub_if_data *sdata, switch (chan_width) { case NL80211_CHAN_WIDTH_320: - WARN_ON(1); - break; + chandef->chan = NULL; + return; case NL80211_CHAN_WIDTH_160: vht_oper.chan_width = IEEE80211_VHT_CHANWIDTH_80MHZ; vht_oper.center_freq_seg1_idx = vht_oper.center_freq_seg0_idx; From 6f63e919fe1e335b8abcb3a28bfd4804a98d875a Mon Sep 17 00:00:00 2001 From: "Cen Zhang (Microsoft)" Date: Thu, 24 Sep 2026 22:44:08 -0400 Subject: [PATCH 306/365] wifi: mac80211: fix slab-out-of-bounds read in ieee80211_monitor_select_queue() ieee80211_monitor_select_queue() validates the skb length using skb->len before dereferencing pointers into skb->data to access the 802.11 header. However, skb->len includes data in both the linear head buffer and non-linear fragments/pages. When AF_PACKET sends a packet large enough to become non-linear, the 802.11 header at skb->data + len_rthdr may extend past the linear head buffer even though skb->len appears sufficient. This leads to a slab-out-of-bounds read when accessing hdr->frame_control, as the kernel reads beyond the allocated skb head buffer: BUG: KASAN: slab-out-of-bounds in ieee80211_monitor_select_queue+0x1ed/0x220 syzbot has hit the same issues. Fix this by checking skb_headlen(skb) (which gives the length of the linear data region) instead of skb->len before dereferencing skb->data pointers. This ensures the required bytes are actually present in the linear portion of the skb. Apply the same fix to ieee80211_validate_radiotap_len() which has the same class of bug: it uses skb->len to validate accesses to skb->data, and to the corresponding checks in ieee80211_monitor_start_xmit(): for drivers advertising NETIF_F_SG the skb is not linearized before ndo_start_xmit, so the 802.11 header can be read past the linear buffer there as well. Fixes: cf0277e714a0 ("mac80211: fix skb buffering issue") Fixes: 9b8a74e3482f ("[MAC80211]: Improve sanity checks on injected packets") Reported-by: AutonomousCodeSecurity@microsoft.com Reported-by: syzbot+610e40369bc02181bad0@syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=610e40369bc02181bad0 Reported-by: syzbot+878643e0580bc580f883@syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=878643e0580bc580f883 Assisted-by: GitHub-Copilot:claude-opus-4.6 Signed-off-by: Cen Zhang (Microsoft) Reviewed-by: Francis Perron Link: https://patch.msgid.link/20260925024408.32143-1-cenzhang@linux.microsoft.com Signed-off-by: Johannes Berg --- net/mac80211/iface.c | 4 ++-- net/mac80211/tx.c | 10 +++++----- 2 files changed, 7 insertions(+), 7 deletions(-) diff --git a/net/mac80211/iface.c b/net/mac80211/iface.c index 889c32fd8de1..88942317fec8 100644 --- a/net/mac80211/iface.c +++ b/net/mac80211/iface.c @@ -998,8 +998,8 @@ static u16 ieee80211_monitor_select_queue(struct net_device *dev, len_rthdr = ieee80211_get_radiotap_len(skb->data); hdr = (struct ieee80211_hdr *)(skb->data + len_rthdr); - if (skb->len < len_rthdr + 2 || - skb->len < len_rthdr + ieee80211_hdrlen(hdr->frame_control)) + if (skb_headlen(skb) < len_rthdr + 2 || + skb_headlen(skb) < len_rthdr + ieee80211_hdrlen(hdr->frame_control)) return 0; /* doesn't matter, frame will be dropped */ return ieee80211_select_queue_80211(sdata, skb, hdr); diff --git a/net/mac80211/tx.c b/net/mac80211/tx.c index b88acb18e04d..d51f810dd2c5 100644 --- a/net/mac80211/tx.c +++ b/net/mac80211/tx.c @@ -2091,7 +2091,7 @@ static bool ieee80211_validate_radiotap_len(struct sk_buff *skb) (struct ieee80211_radiotap_header *)skb->data; /* check for not even having the fixed radiotap header part */ - if (unlikely(skb->len < sizeof(struct ieee80211_radiotap_header))) + if (unlikely(skb_headlen(skb) < sizeof(struct ieee80211_radiotap_header))) return false; /* too short to be possibly valid */ /* is it a header version we can trust to find length from? */ @@ -2099,7 +2099,7 @@ static bool ieee80211_validate_radiotap_len(struct sk_buff *skb) return false; /* only version 0 is supported */ /* does the skb contain enough to deliver on the alleged length? */ - if (unlikely(skb->len < ieee80211_get_radiotap_len(skb->data))) + if (unlikely(skb_headlen(skb) < ieee80211_get_radiotap_len(skb->data))) return false; /* skb too short for claimed rt header extent */ return true; @@ -2388,13 +2388,13 @@ netdev_tx_t ieee80211_monitor_start_xmit(struct sk_buff *skb, skb_set_network_header(skb, len_rthdr); skb_set_transport_header(skb, len_rthdr); - if (skb->len < len_rthdr + 2) + if (skb_headlen(skb) < len_rthdr + 2) goto fail; hdr = (struct ieee80211_hdr *)(skb->data + len_rthdr); hdrlen = ieee80211_hdrlen(hdr->frame_control); - if (skb->len < len_rthdr + hdrlen) + if (skb_headlen(skb) < len_rthdr + hdrlen) goto fail; /* @@ -2402,7 +2402,7 @@ netdev_tx_t ieee80211_monitor_start_xmit(struct sk_buff *skb, * carrying a rfc1042 header */ if (ieee80211_is_data(hdr->frame_control) && - skb->len >= len_rthdr + hdrlen + sizeof(rfc1042_header) + 2) { + skb_headlen(skb) >= len_rthdr + hdrlen + sizeof(rfc1042_header) + 2) { u8 *payload = (u8 *)hdr + hdrlen; if (ether_addr_equal(payload, rfc1042_header)) From 024e05f73a4c1263d031614bc36700ec135eecb4 Mon Sep 17 00:00:00 2001 From: Chengfeng Ye Date: Sun, 27 Sep 2026 19:04:50 +0800 Subject: [PATCH 307/365] Bluetooth: hci_conn: Lock parent access during enhanced SCO setup Bluetooth: hci_conn: Lock parent access during enhanced SCO setup hci_enhanced_setup_sync() runs on the request workqueue without the hci_dev_lock held by its caller when setup was queued. Its CVSD capability check and find_next_esco_param() dereference conn->parent while the receive workqueue can unlink and release that parent. The following interleaving can cause a use-after-free: hci_enhanced_setup_sync() hci_disconn_complete_evt() load conn->parent hci_dev_lock() hci_conn_del(ACL parent) unlink SCO child drop link's parent reference clear child->parent release ACL parent bt_link_release() kfree(parent) hci_dev_unlock() read parent->features[0][3] The reference held for the queued SCO child does not keep its ACL parent alive after unlinking. Commit 42de40abe25d ("Bluetooth: hci_conn: fix the SCO setup context lifetime") protects the child stored in the queued context, but leaves these parent accesses unprotected. With a 40 ms diagnostic delay after loading conn->parent, an instrumented kernel based on fd179f8a05be, which already contains 42de40abe25d, reported: BUG: KASAN: slab-use-after-free in hci_enhanced_setup_sync+0xda5/0xdf0 Read of size 1 at addr ffff888102204047 by task kworker/u17:0/93 Call Trace: hci_enhanced_setup_sync+0xda5/0xdf0 hci_cmd_sync_work+0x13c/0x290 process_one_work+0x6b4/0x10e0 Allocated by task 92: __hci_conn_add+0x304/0x1df0 hci_connect_acl+0x349/0x3e0 hci_connect_sco+0x3b/0x9a0 sco_sock_connect+0x475/0xca0 Freed by task 94: kfree+0x121/0x3c0 bt_link_release+0x79/0xa0 device_release+0xc8/0x240 kobject_put+0x14d/0x280 hci_conn_del+0x524/0xe30 hci_disconn_complete_evt+0x403/0x8c0 hci_event_packet+0x71b/0xb20 hci_rx_work+0x293/0x730 The accessed address is 71 bytes into the freed ACL parent, at its features[0][3] byte; the queued SCO child is a different object. The diagnostic preserves the loaded parent across the delay, matching the unmodified compiled capability check, and does not change the parent references or teardown path. Hold hci_dev_lock() across the codec switch, including every call to find_next_esco_param(), and release it on all selection errors. Keep configure_datapath_sync() outside the critical section because it waits for HCI events. Preserve parameter selection and existing return values. Fixes: e07a06b4eb41 ("Bluetooth: Convert SCO configure_datapath to hci_sync") Cc: stable@vger.kernel.org Assisted-by: GPT-6-Astra Signed-off-by: Chengfeng Ye Signed-off-by: Luiz Augusto von Dentz --- net/bluetooth/hci_conn.c | 18 +++++++++++++----- 1 file changed, 13 insertions(+), 5 deletions(-) diff --git a/net/bluetooth/hci_conn.c b/net/bluetooth/hci_conn.c index 96195d2fd10f..cf44452e0766 100644 --- a/net/bluetooth/hci_conn.c +++ b/net/bluetooth/hci_conn.c @@ -302,11 +302,13 @@ static int hci_enhanced_setup_sync(struct hci_dev *hdev, void *data) cp.tx_bandwidth = cpu_to_le32(0x00001f40); cp.rx_bandwidth = cpu_to_le32(0x00001f40); + hci_dev_lock(hdev); + switch (conn->codec.id) { case BT_CODEC_MSBC: if (!find_next_esco_param(conn, esco_param_msbc, ARRAY_SIZE(esco_param_msbc))) - return -EINVAL; + goto unlock; param = &esco_param_msbc[conn->attempt - 1]; cp.tx_coding_format.id = 0x05; @@ -332,7 +334,7 @@ static int hci_enhanced_setup_sync(struct hci_dev *hdev, void *data) case BT_CODEC_TRANSPARENT: if (!find_next_esco_param(conn, esco_param_msbc, ARRAY_SIZE(esco_param_msbc))) - return -EINVAL; + goto unlock; param = &esco_param_msbc[conn->attempt - 1]; cp.tx_coding_format.id = 0x03; @@ -359,11 +361,11 @@ static int hci_enhanced_setup_sync(struct hci_dev *hdev, void *data) if (conn->parent && lmp_esco_capable(conn->parent)) { if (!find_next_esco_param(conn, esco_param_cvsd, ARRAY_SIZE(esco_param_cvsd))) - return -EINVAL; + goto unlock; param = &esco_param_cvsd[conn->attempt - 1]; } else { if (conn->attempt > ARRAY_SIZE(sco_param_cvsd)) - return -EINVAL; + goto unlock; param = &sco_param_cvsd[conn->attempt - 1]; } cp.tx_coding_format.id = 2; @@ -386,9 +388,11 @@ static int hci_enhanced_setup_sync(struct hci_dev *hdev, void *data) cp.out_transport_unit_size = 16; break; default: - return -EINVAL; + goto unlock; } + hci_dev_unlock(hdev); + cp.retrans_effort = param->retrans_effort; cp.pkt_type = __cpu_to_le16(param->pkt_type); cp.max_latency = __cpu_to_le16(param->max_latency); @@ -397,6 +401,10 @@ static int hci_enhanced_setup_sync(struct hci_dev *hdev, void *data) return -EIO; return 0; + +unlock: + hci_dev_unlock(hdev); + return -EINVAL; } static bool hci_setup_sync_conn(struct hci_conn *conn, __u16 handle) From 1a572db43c057566771b485598ad2e66cb390821 Mon Sep 17 00:00:00 2001 From: Chengfeng Ye Date: Sun, 27 Sep 2026 01:29:35 +0800 Subject: [PATCH 308/365] Bluetooth: hci_core: Serialize fragmented ISO packet queueing The fragmented path in hci_queue_iso() calls __skb_queue_tail() without holding conn->data_q.lock. The socket lock serializes senders, but the transmit worker can concurrently remove packets from this queue using skb_dequeue(). After an insertion saves the old tail pointer, hci_sched_iso() can dequeue that packet and pass it to the driver. The driver can free the packet before the insertion resumes and writes to the old tail's next pointer, causing a use-after-free. Concurrent updates can also corrupt the queue. KASAN reported: BUG: KASAN: slab-use-after-free in hci_send_iso+0xc7d/0xe10 Call Trace: hci_send_iso+0xc7d/0xe10 iso_sock_sendmsg+0x710/0x900 __sys_sendto+0x34a/0x3a0 Allocated by task 86: __alloc_skb+0xdd/0x820 alloc_skb_with_frags+0x7e/0x770 sock_alloc_send_pskb+0x67a/0x820 bt_skb_sendmsg.constprop.0+0xc0/0x6c0 iso_sock_sendmsg+0x44e/0x900 Freed by task 90: kmem_cache_free+0xcb/0x3d0 vhci_read+0x33f/0x4d0 vfs_read+0x177/0xa20 Hold the queue lock across the entire fragment batch, as hci_queue_acl() does, to serialize insertion against dequeue while preserving fragment ordering. Fixes: 26afbd826ee3 ("Bluetooth: Add initial implementation of CIS connections") Cc: stable@vger.kernel.org Signed-off-by: Chengfeng Ye Signed-off-by: Luiz Augusto von Dentz --- net/bluetooth/hci_core.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/net/bluetooth/hci_core.c b/net/bluetooth/hci_core.c index bbe391525530..5d1cdb1e7d8b 100644 --- a/net/bluetooth/hci_core.c +++ b/net/bluetooth/hci_core.c @@ -3327,6 +3327,8 @@ static void hci_queue_iso(struct hci_conn *conn, struct sk_buff_head *queue, skb_shinfo(skb)->frag_list = NULL; + spin_lock_bh(&queue->lock); + __skb_queue_tail(queue, skb); do { @@ -3341,6 +3343,8 @@ static void hci_queue_iso(struct hci_conn *conn, struct sk_buff_head *queue, __skb_queue_tail(queue, skb); } while (list); + + spin_unlock_bh(&queue->lock); } bt_dev_dbg(hdev, "hcon %p queued %d", conn, skb_queue_len(queue)); From b1f24c1ab52345e810c42a1a81286cb5a7c92252 Mon Sep 17 00:00:00 2001 From: Nguyen Ngoc Thang Date: Sun, 27 Sep 2026 22:18:02 +0700 Subject: [PATCH 309/365] Bluetooth: hci_sync: don't drain cmd_sync backlog on unregister hci_unregister_dev() disables cmd_work and cmd_timer, then calls hci_cmd_sync_clear(), whose cancel_work_sync() waits for hci_cmd_sync_work() to return. That worker dequeues and runs every entry on cmd_sync_work_list. With cmd_work disabled nothing reaches the controller, so each queued HCI command waits the full HCI_CMD_TIMEOUT (2s). The backlog has no bound. Userspace can keep queuing MGMT commands such as MGMT_OP_GET_CLOCK_INFO against a controller that doesn't answer. Closing /dev/vhci then blocks in vhci_release() for backlog * 2s: INFO: task syz-executor:5749 blocked for more than 143 seconds. cancel_work_sync hci_cmd_sync_clear hci_unregister_dev vhci_release In a local reproduction the backlog held more than 5000 entries, which comes to hours of hang. Stop the worker from taking new entries once HCI_UNREGISTER is set, and wake any request still waiting with -ENODEV before cancelling the work. The entries left over are destroyed with -ECANCELED by the existing sweep in hci_cmd_sync_clear(). They stay on the list until the work has stopped, so hci_cmd_sync_dequeue() and friends still see them. A callback already running may still issue another command, which delays unregister by at most one timeout per remaining command, not by the whole backlog. Fixes: 008ee9eb8a11 ("Bluetooth: hci_sync: Fix not processing all entries on cmd_sync_work") Reported-by: syzbot+217e3f1283cafe80586e@syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=217e3f1283cafe80586e Signed-off-by: Nguyen Ngoc Thang Signed-off-by: Luiz Augusto von Dentz --- net/bluetooth/hci_sync.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/net/bluetooth/hci_sync.c b/net/bluetooth/hci_sync.c index 74e2b04c84b2..e41fc1427d34 100644 --- a/net/bluetooth/hci_sync.c +++ b/net/bluetooth/hci_sync.c @@ -312,6 +312,10 @@ static void hci_cmd_sync_work(struct work_struct *work) while (1) { struct hci_cmd_sync_work_entry *entry; + /* Leave the backlog to hci_cmd_sync_clear() */ + if (hci_dev_test_flag(hdev, HCI_UNREGISTER)) + break; + mutex_lock(&hdev->cmd_sync_work_lock); entry = list_first_entry_or_null(&hdev->cmd_sync_work_list, struct hci_cmd_sync_work_entry, @@ -658,6 +662,8 @@ void hci_cmd_sync_clear(struct hci_dev *hdev) { struct hci_cmd_sync_work_entry *entry, *tmp; + /* cmd_work is disabled, the pending request can only time out */ + hci_cmd_sync_cancel_sync(hdev, ENODEV); cancel_work_sync(&hdev->cmd_sync_work); cancel_work_sync(&hdev->reenable_adv_work); From 8ed67535fdff40ca652bcd4ac7da68d11d01bd34 Mon Sep 17 00:00:00 2001 From: Chengfeng Ye Date: Sun, 27 Sep 2026 01:28:31 +0800 Subject: [PATCH 310/365] Bluetooth: hci_sync: Fix inquiry cache use-after-free The sync command worker holds hdev->req_lock, but inquiry-cache updates and flushes use hdev->lock. Both hci_acl_create_conn_sync() and hci_stop_discovery_sync() look up entries and read their fields without taking hdev->lock. After either lookup returns, a concurrent HCIINQUIRY ioctl can acquire hdev->lock and flush the cache, freeing the entry. The worker then reads the freed entry while preparing a create-connection or remote-name-cancel command. The list traversal also races with cache updates and removal. KASAN reported these accesses: BUG: KASAN: slab-use-after-free in hci_acl_create_conn_sync+0x5f1/0x650 Workqueue: hci0 hci_cmd_sync_work Call Trace: hci_acl_create_conn_sync+0x5f1/0x650 hci_cmd_sync_work+0x13c/0x290 Allocated by task 90: hci_inquiry_cache_update+0x3e6/0x7d0 hci_inquiry_result_evt+0x3cb/0x560 Freed by task 93: hci_inquiry_cache_flush+0x111/0x2b0 hci_inquiry+0x2f2/0x780 hci_sock_ioctl+0x269/0x5f0 BUG: KASAN: slab-use-after-free in hci_stop_discovery_sync+0x3b1/0x3c0 Workqueue: hci0 hci_cmd_sync_work Call Trace: hci_stop_discovery_sync+0x3b1/0x3c0 hci_cmd_sync_work+0x173/0x300 Allocated by task 86: hci_inquiry_cache_update+0x483/0x940 hci_inquiry_result_evt+0x3cb/0x560 Freed by task 91: hci_inquiry_cache_flush+0x13e/0x2f0 hci_inquiry+0x2f2/0x780 hci_sock_ioctl+0x269/0x5f0 Hold hdev->lock across each lookup and all reads from its result. Copy the remote address before unlocking so discovery cancellation does not retain a cache entry pointer. Release the lock before sending synchronous HCI commands, since their completion handlers may need the same lock. Fixes: cf75ad8b41d2 ("Bluetooth: hci_sync: Convert MGMT_SET_POWERED") Fixes: 45340097ce6e ("Bluetooth: hci_conn: Only do ACL connections sequentially") Cc: stable@vger.kernel.org Signed-off-by: Chengfeng Ye Signed-off-by: Luiz Augusto von Dentz --- net/bluetooth/hci_sync.c | 13 +++++++++++-- 1 file changed, 11 insertions(+), 2 deletions(-) diff --git a/net/bluetooth/hci_sync.c b/net/bluetooth/hci_sync.c index e41fc1427d34..c01c8b58d9e8 100644 --- a/net/bluetooth/hci_sync.c +++ b/net/bluetooth/hci_sync.c @@ -5770,6 +5770,7 @@ int hci_stop_discovery_sync(struct hci_dev *hdev) { struct discovery_state *d = &hdev->discovery; struct inquiry_entry *e; + bdaddr_t addr; int err; bt_dev_dbg(hdev, "state %u", hdev->discovery.state); @@ -5805,15 +5806,21 @@ int hci_stop_discovery_sync(struct hci_dev *hdev) return 0; if (d->state == DISCOVERY_RESOLVING || d->state == DISCOVERY_STOPPING) { + hci_dev_lock(hdev); e = hci_inquiry_cache_lookup_resolve(hdev, BDADDR_ANY, NAME_PENDING); - if (!e) + if (!e) { + hci_dev_unlock(hdev); return 0; + } + + bacpy(&addr, &e->data.bdaddr); + hci_dev_unlock(hdev); /* Ignore cancel errors since it should interfere with stopping * of the discovery. */ - hci_remote_name_cancel_sync(hdev, &e->data.bdaddr); + hci_remote_name_cancel_sync(hdev, &addr); } return 0; @@ -7242,6 +7249,7 @@ static int hci_acl_create_conn_sync(struct hci_dev *hdev, void *data) bacpy(&cp.bdaddr, &conn->dst); cp.pscan_rep_mode = 0x02; + hci_dev_lock(hdev); ie = hci_inquiry_cache_lookup(hdev, &conn->dst); if (ie) { if (inquiry_entry_age(ie) <= INQUIRY_ENTRY_AGE_MAX) { @@ -7253,6 +7261,7 @@ static int hci_acl_create_conn_sync(struct hci_dev *hdev, void *data) memcpy(conn->dev_class, ie->data.dev_class, 3); } + hci_dev_unlock(hdev); cp.pkt_type = cpu_to_le16(conn->pkt_type); if (lmp_rswitch_capable(hdev) && !(hdev->link_mode & HCI_LM_MASTER)) From 2d696c1ed3468dcb62578a3911e687110d463520 Mon Sep 17 00:00:00 2001 From: Chengfeng Ye Date: Sun, 27 Sep 2026 01:27:23 +0800 Subject: [PATCH 311/365] Bluetooth: RFCOMM: Fix initial port reference race For RFCOMM_RELEASE_ONHUP devices, rfcomm_tty_install() and __rfcomm_release_dev() both use RFCOMM_TTY_OWNED to decide who drops the initial tty_port reference. The release path tests the bit separately from the install path setting it, so both can drop that reference. The synchronous hangup does not prevent this race: port->tty is not assigned until tty_port_open(), after installation. The following interleaving is possible: 1. Install and release each obtain a reference with rfcomm_dev_get(). 2. Release observes RFCOMM_TTY_OWNED clear. 3. Install sets RFCOMM_TTY_OWNED and drops the initial reference. 4. Release drops the same initial reference again. 5. TTY cleanup drops its reference and frees the device. 6. Release performs its final tty_port_put() on the freed port. KASAN reported: BUG: KASAN: slab-use-after-free in tty_port_put+0x22/0x190 Write of size 4 at addr ffff8881001d3d5c by task poc/92 Call Trace: tty_port_put+0x22/0x190 rfcomm_dev_ioctl+0x1d4/0x1930 sock_do_ioctl+0x110/0x260 sock_ioctl+0x380/0x590 __x64_sys_ioctl+0x134/0x1c0 Allocated by task 88: rfcomm_dev_ioctl+0x8f6/0x1930 sock_do_ioctl+0x110/0x260 sock_ioctl+0x380/0x590 __x64_sys_ioctl+0x134/0x1c0 Freed by task 65: kfree+0x131/0x3c0 rfcomm_dev_destruct+0x23b/0x2f0 release_one_tty+0xc1/0x370 process_one_work+0x661/0x1090 Use test_and_set_bit() in both paths so that only the caller that changes RFCOMM_TTY_OWNED from clear to set drops the initial reference. Each path keeps its own lookup reference until release or TTY cleanup, preserving the existing callback ordering and error handling. Fixes: 80ea73378af4 ("Bluetooth: Fix unreleased rfcomm_dev reference") Cc: stable@vger.kernel.org Assisted-by: LLM Signed-off-by: Chengfeng Ye Signed-off-by: Luiz Augusto von Dentz --- net/bluetooth/rfcomm/tty.c | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) diff --git a/net/bluetooth/rfcomm/tty.c b/net/bluetooth/rfcomm/tty.c index b2c1060394e6..dc3cdf614def 100644 --- a/net/bluetooth/rfcomm/tty.c +++ b/net/bluetooth/rfcomm/tty.c @@ -462,7 +462,7 @@ static int __rfcomm_release_dev(void __user *arg) /* Shut down TTY synchronously before freeing rfcomm_dev */ tty_port_tty_vhangup(&dev->port); - if (!test_bit(RFCOMM_TTY_OWNED, &dev->status)) + if (!test_and_set_bit(RFCOMM_TTY_OWNED, &dev->status)) tty_port_put(&dev->port); tty_port_put(&dev->port); @@ -724,10 +724,9 @@ static int rfcomm_tty_install(struct tty_driver *driver, struct tty_struct *tty) * when the last process closes the tty. The behaviour is expected by * userspace. */ - if (test_bit(RFCOMM_RELEASE_ONHUP, &dev->flags)) { - set_bit(RFCOMM_TTY_OWNED, &dev->status); + if (test_bit(RFCOMM_RELEASE_ONHUP, &dev->flags) && + !test_and_set_bit(RFCOMM_TTY_OWNED, &dev->status)) tty_port_put(&dev->port); - } return 0; } From 81a2345f1984f0b36d3226571bc196316a01b648 Mon Sep 17 00:00:00 2001 From: Chengfeng Ye Date: Sun, 27 Sep 2026 14:42:50 +0800 Subject: [PATCH 312/365] Bluetooth: Serialize SMP remote OOB data access build_pairing_cmd() looks up remote OOB data and copies its contents without holding hdev->lock, which serializes the list's writers. After SMP finds an entry, a concurrent management Remove Remote OOB Data command can unlink and free it before SMP reads its present flag or copies its random and confirmation values. Removal can also invalidate an entry while the lookup is still traversing the list. KASAN reported: BUG: KASAN: slab-use-after-free in build_pairing_cmd+0x948/0x9b0 Call Trace: build_pairing_cmd+0x948/0x9b0 smp_recv_cb+0x459f/0x8110 l2cap_recv_frame+0xf14/0x9190 l2cap_recv_acldata+0xa64/0xd40 hci_rx_work+0x4ca/0x730 Allocated by task 87: hci_add_remote_oob_data+0x11d/0x530 add_remote_oob_data+0x282/0x400 hci_sock_sendmsg+0x1033/0x1ea0 Freed by task 93: hci_remote_oob_data_clear+0x108/0x1c0 remove_remote_oob_data+0x198/0x220 hci_sock_sendmsg+0x1033/0x1ea0 Taking hdev->lock in build_pairing_cmd() would recurse for callers that already hold it and invert the device-to-L2CAP lock order on the receive path. Add a per-device remote_oob_lock instead, held across the SMP lookup and copies and by the add, remove and clear helpers. Cover initialization and in-place updates as well, so SMP cannot read partially initialized or updated OOB values. Release the mutex on allocation failure, preserving the existing error return. The new critical sections acquire no device, connection or channel locks. Writers retain their existing hdev->lock protection, which continues to serialize the other readers without changing their locking or behavior. Link: https://lore.kernel.org/r/00660cd3-7d71-13a4-f617-229e6defb701@gmail.com Fixes: 02b05bd8b0a6 ("Bluetooth: Set SMP OOB flag if OOB data is available") Cc: stable@vger.kernel.org Signed-off-by: Chengfeng Ye Signed-off-by: Luiz Augusto von Dentz --- include/net/bluetooth/hci_core.h | 1 + net/bluetooth/hci_core.c | 12 +++++++++++- net/bluetooth/smp.c | 2 ++ 3 files changed, 14 insertions(+), 1 deletion(-) diff --git a/include/net/bluetooth/hci_core.h b/include/net/bluetooth/hci_core.h index ca77ad0d9393..fa2a367731b5 100644 --- a/include/net/bluetooth/hci_core.h +++ b/include/net/bluetooth/hci_core.h @@ -562,6 +562,7 @@ struct hci_dev { struct list_head link_keys; struct list_head long_term_keys; struct list_head identity_resolving_keys; + struct mutex remote_oob_lock; struct list_head remote_oob_data; struct list_head le_accept_list; struct list_head le_resolv_list; diff --git a/net/bluetooth/hci_core.c b/net/bluetooth/hci_core.c index 5d1cdb1e7d8b..bc662e24a8e4 100644 --- a/net/bluetooth/hci_core.c +++ b/net/bluetooth/hci_core.c @@ -1489,8 +1489,10 @@ int hci_remove_remote_oob_data(struct hci_dev *hdev, bdaddr_t *bdaddr, BT_DBG("%s removing %pMR (%u)", hdev->name, bdaddr, bdaddr_type); + mutex_lock(&hdev->remote_oob_lock); list_del(&data->list); kfree(data); + mutex_unlock(&hdev->remote_oob_lock); return 0; } @@ -1499,10 +1501,12 @@ void hci_remote_oob_data_clear(struct hci_dev *hdev) { struct oob_data *data, *n; + mutex_lock(&hdev->remote_oob_lock); list_for_each_entry_safe(data, n, &hdev->remote_oob_data, list) { list_del(&data->list); kfree(data); } + mutex_unlock(&hdev->remote_oob_lock); } int hci_add_remote_oob_data(struct hci_dev *hdev, bdaddr_t *bdaddr, @@ -1511,11 +1515,14 @@ int hci_add_remote_oob_data(struct hci_dev *hdev, bdaddr_t *bdaddr, { struct oob_data *data; + mutex_lock(&hdev->remote_oob_lock); data = hci_find_remote_oob_data(hdev, bdaddr, bdaddr_type); if (!data) { data = kmalloc_obj(*data); - if (!data) + if (!data) { + mutex_unlock(&hdev->remote_oob_lock); return -ENOMEM; + } bacpy(&data->bdaddr, bdaddr); data->bdaddr_type = bdaddr_type; @@ -1548,6 +1555,8 @@ int hci_add_remote_oob_data(struct hci_dev *hdev, bdaddr_t *bdaddr, BT_DBG("%s for %pMR", hdev->name, bdaddr); + mutex_unlock(&hdev->remote_oob_lock); + return 0; } @@ -2485,6 +2494,7 @@ struct hci_dev *hci_alloc_dev_priv(int sizeof_priv) mutex_init(&hdev->lock); mutex_init(&hdev->req_lock); mutex_init(&hdev->mgmt_pending_lock); + mutex_init(&hdev->remote_oob_lock); ida_init(&hdev->unset_handle_ida); diff --git a/net/bluetooth/smp.c b/net/bluetooth/smp.c index d23f9d0729c4..2df303f6a38f 100644 --- a/net/bluetooth/smp.c +++ b/net/bluetooth/smp.c @@ -662,6 +662,7 @@ static void build_pairing_cmd(struct l2cap_conn *conn, else bdaddr_type = BDADDR_LE_RANDOM; + mutex_lock(&hdev->remote_oob_lock); oob_data = hci_find_remote_oob_data(hdev, &hcon->dst, bdaddr_type); if (oob_data && oob_data->present) { @@ -672,6 +673,7 @@ static void build_pairing_cmd(struct l2cap_conn *conn, SMP_DBG("OOB Remote Confirmation: %16phN", smp->pcnf); SMP_DBG("OOB Remote Random: %16phN", smp->rr); } + mutex_unlock(&hdev->remote_oob_lock); } else { authreq &= ~SMP_AUTH_SC; From f4845ab191a4f686a5da481fbba90ed73bb693b0 Mon Sep 17 00:00:00 2001 From: Chengfeng Ye Date: Sun, 27 Sep 2026 01:04:02 +0800 Subject: [PATCH 313/365] Bluetooth: hci_core: Serialize ACL scheduling with channel deletion hci_chan_sent() selects a channel under RCU but drops the read lock before accessing chan->conn and returning the channel. The ACL and LE schedulers then use its packet queue and update its transmit counters without any protection against channel deletion. After TX selects a channel and releases RCU, a disconnect command timeout on the separate request workqueue can run hci_conn_failed() and l2cap_conn_del(). hci_chan_del() can then unlink the channel, complete synchronize_rcu(), purge its queue and free it before TX resumes. This causes use-after-free both in hci_chan_sent() and in its callers. KASAN reported: BUG: KASAN: slab-use-after-free in hci_chan_sent+0x892/0x9b0 Workqueue: hci0 hci_tx_work Call Trace: hci_chan_sent+0x892/0x9b0 hci_tx_work+0x5e6/0xb70 Allocated by task 91: hci_chan_create+0xe3/0x350 l2cap_conn_add.part.0+0x12/0xa30 l2cap_chan_connect+0x110d/0x1b60 l2cap_sock_connect+0x310/0x530 Freed by task 99: hci_chan_del+0x11f/0x170 l2cap_conn_del+0x4f1/0x800 l2cap_connect_cfm+0x88c/0xd30 hci_conn_failed+0x150/0x250 hci_abort_conn_sync+0x3e3/0x800 hci_cmd_sync_run+0x7e/0xc0 hci_abort_conn+0x105/0x1f0 disconnect_sync+0x157/0x290 hci_cmd_sync_work+0x13c/0x290 Hold the existing device mutex across channel selection and transmission in both schedulers to serialize them with channel teardown. Keep timeout handling outside the critical sections because hci_link_tx_to() acquires the same mutex. This also permits the transmit path to sleep, unlike extending the RCU read-side critical section across packet submission. Fixes: 3eff45eaf817 ("Bluetooth: convert tx_task to workqueue") Cc: stable@vger.kernel.org Assisted-by: LLM Signed-off-by: Chengfeng Ye Signed-off-by: Luiz Augusto von Dentz --- net/bluetooth/hci_core.c | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/net/bluetooth/hci_core.c b/net/bluetooth/hci_core.c index bc662e24a8e4..a92ab4db4676 100644 --- a/net/bluetooth/hci_core.c +++ b/net/bluetooth/hci_core.c @@ -3675,6 +3675,8 @@ static void hci_sched_acl_pkt(struct hci_dev *hdev) __check_timeout(hdev, cnt, ACL_LINK); + hci_dev_lock(hdev); + while (hdev->acl_cnt && (chan = hci_chan_sent(hdev, ACL_LINK, "e))) { u32 priority = (skb_peek(&chan->data_q))->priority; @@ -3706,6 +3708,8 @@ static void hci_sched_acl_pkt(struct hci_dev *hdev) if (cnt != hdev->acl_cnt) hci_prio_recalculate(hdev, ACL_LINK); + + hci_dev_unlock(hdev); } static void hci_sched_acl(struct hci_dev *hdev) @@ -3734,6 +3738,8 @@ static void hci_sched_le(struct hci_dev *hdev) __check_timeout(hdev, *cnt, LE_LINK); + hci_dev_lock(hdev); + tmp = *cnt; while (*cnt && (chan = hci_chan_sent(hdev, LE_LINK, "e))) { u32 priority = (skb_peek(&chan->data_q))->priority; @@ -3762,6 +3768,8 @@ static void hci_sched_le(struct hci_dev *hdev) if (*cnt != tmp) hci_prio_recalculate(hdev, LE_LINK); + + hci_dev_unlock(hdev); } /* Schedule iso */ From 848a7a91c7f03675d3bd8db6f7721cbf9cb50e21 Mon Sep 17 00:00:00 2001 From: Chengfeng Ye Date: Sun, 27 Sep 2026 01:04:03 +0800 Subject: [PATCH 314/365] Bluetooth: hci_core: Serialize SCO and ISO scheduling with teardown hci_low_sent() selects a connection under RCU but drops the read lock before calculating its quota and returning it to the scheduler. The connection is then used without lifetime protection by hci_sched_iso() and hci_sched_sco(). After the TX worker drops the RCU read lock, hci_abort_conn_sync() on hdev->req_workqueue can remove the connection from the hash, complete synchronize_rcu(), purge its queues and release it. The TX worker on hdev->workqueue can then access the freed connection in hci_quote_sent() or while dequeuing packets and updating conn->sent. KASAN reported: BUG: KASAN: slab-use-after-free in hci_low_sent+0x730/0x840 Workqueue: hci0 hci_tx_work Call Trace: hci_low_sent+0x730/0x840 hci_sched_iso+0x25e/0x4d0 hci_tx_work+0x239/0xcb0 Allocated by task 93: __hci_conn_add+0x16f/0x1b40 hci_bind_bis+0x782/0x17b0 hci_connect_bis+0xa0/0x510 iso_sock_connect+0x589/0x1050 Freed by task 88: kfree+0x131/0x3c0 device_release+0xc8/0x240 kobject_put+0x14d/0x280 hci_conn_del+0x55a/0xe80 hci_disconnect_sync+0x156/0x180 hci_abort_conn_sync+0x3e7/0x940 hci_cmd_sync_work+0x13c/0x290 Hold hci_dev_lock() across connection selection and transmission in both SCO and ISO scheduling, serializing them with connection teardown. This also prevents queuing completion timestamps after the connection queues have been purged. Extending RCU across transmission would be unsafe because the transmit path can sleep. Keep the ISO timeout check outside the mutex since hci_link_tx_to() takes it itself. The preceding channel fix already holds this mutex in the ACL and LE schedulers, which call the SCO scheduler between packets. Move the SCO body to __hci_sched_sco(), assert that its caller holds the mutex, and use it directly from these locked paths. Keep a locking hci_sched_sco() wrapper for the direct calls from hci_tx_work(). This avoids recursively acquiring the device mutex while preserving the scheduling order. Remove the obsolete claim that connection removal disables TX. Fixes: bf4c63252490 ("Bluetooth: convert conn hash to RCU") Cc: stable@vger.kernel.org Assisted-by: LLM Signed-off-by: Chengfeng Ye Signed-off-by: Luiz Augusto von Dentz --- net/bluetooth/hci_core.c | 26 ++++++++++++++++++-------- 1 file changed, 18 insertions(+), 8 deletions(-) diff --git a/net/bluetooth/hci_core.c b/net/bluetooth/hci_core.c index a92ab4db4676..2076689eb302 100644 --- a/net/bluetooth/hci_core.c +++ b/net/bluetooth/hci_core.c @@ -3418,9 +3418,6 @@ static struct hci_conn *hci_low_sent(struct hci_dev *hdev, __u8 type, struct hci_conn *conn = NULL, *c; unsigned int num = 0, min = ~0; - /* We don't have to lock device here. Connections are always - * added and removed with TX task disabled. */ - rcu_read_lock(); list_for_each_entry_rcu(c, &h->list, list) { @@ -3625,13 +3622,15 @@ static void __check_timeout(struct hci_dev *hdev, unsigned int cnt, u8 type) } /* Schedule SCO */ -static void hci_sched_sco(struct hci_dev *hdev, __u8 type) +static void __hci_sched_sco(struct hci_dev *hdev, __u8 type) { struct hci_conn *conn; struct sk_buff *skb; int quote, *cnt; unsigned int pkts = hdev->sco_pkts; + lockdep_assert_held(&hdev->lock); + bt_dev_dbg(hdev, "type %u", type); if (!hci_conn_num(hdev, type) || !pkts) @@ -3666,6 +3665,13 @@ static void hci_sched_sco(struct hci_dev *hdev, __u8 type) queue_work(hdev->workqueue, &hdev->tx_work); } +static void hci_sched_sco(struct hci_dev *hdev, __u8 type) +{ + hci_dev_lock(hdev); + __hci_sched_sco(hdev, type); + hci_dev_unlock(hdev); +} + static void hci_sched_acl_pkt(struct hci_dev *hdev) { unsigned int cnt = hdev->acl_cnt; @@ -3701,8 +3707,8 @@ static void hci_sched_acl_pkt(struct hci_dev *hdev) chan->conn->sent++; /* Send pending SCO packets right away */ - hci_sched_sco(hdev, SCO_LINK); - hci_sched_sco(hdev, ESCO_LINK); + __hci_sched_sco(hdev, SCO_LINK); + __hci_sched_sco(hdev, ESCO_LINK); } } @@ -3761,8 +3767,8 @@ static void hci_sched_le(struct hci_dev *hdev) chan->conn->sent++; /* Send pending SCO packets right away */ - hci_sched_sco(hdev, SCO_LINK); - hci_sched_sco(hdev, ESCO_LINK); + __hci_sched_sco(hdev, SCO_LINK); + __hci_sched_sco(hdev, ESCO_LINK); } } @@ -3788,6 +3794,8 @@ static void hci_sched_iso(struct hci_dev *hdev, __u8 type) __check_timeout(hdev, *cnt, type); + hci_dev_lock(hdev); + while (*cnt && (conn = hci_low_sent(hdev, type, "e))) { while (quote-- && (skb = skb_dequeue(&conn->data_q))) { BT_DBG("skb %p len %d", skb, skb->len); @@ -3801,6 +3809,8 @@ static void hci_sched_iso(struct hci_dev *hdev, __u8 type) (*cnt)--; } } + + hci_dev_unlock(hdev); } static void hci_tx_work(struct work_struct *work) From 6b2462f07cb74cd0a5cb868b36c09e9f51f56d3f Mon Sep 17 00:00:00 2001 From: Christian Melendez Date: Tue, 22 Sep 2026 23:14:39 +0200 Subject: [PATCH 315/365] MAINTAINERS: add missing entry for b53.rst Documentation/networking/dsa/b53.rst is not covered by the BROADCOM B53/SF2 ETHERNET SWITCH DRIVER entry. Signed-off-by: Christian Melendez Link: https://patch.msgid.link/20260922211439.369917-2-chrismelnu@gmail.com Signed-off-by: Jakub Kicinski --- MAINTAINERS | 1 + 1 file changed, 1 insertion(+) diff --git a/MAINTAINERS b/MAINTAINERS index c915b114c99c..7d334b504bc3 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -5143,6 +5143,7 @@ L: netdev@vger.kernel.org L: openwrt-devel@lists.openwrt.org (subscribers-only) S: Supported F: Documentation/devicetree/bindings/net/dsa/brcm,b53.yaml +F: Documentation/networking/dsa/b53.rst F: drivers/net/dsa/b53/* F: drivers/net/dsa/bcm_sf2* F: include/linux/dsa/brcm.h From eb7b84af0ec5c71b4920e999611b466b3784496c Mon Sep 17 00:00:00 2001 From: Florian Fainelli Date: Tue, 22 Sep 2026 16:24:31 -0700 Subject: [PATCH 316/365] net: systemport: Fix buffer overflow in bcm_sysport_get_stats() When running on SYSTEMPORT Lite, certain statistics are unsupported and skipped during bcm_sysport_get_stats(). The variable 'j' tracks the compacted index into the destination data buffer, whereas 'i' iterates over all elements in bcm_sysport_gstrings_stats. Because the buffer allocated by ethtool is sized only according to bcm_sysport_get_sset_count(), storing values at data[i] instead of data[j] writes past the allocated array bounds, leading to memory corruption. Fix this by writing to data[j] instead of data[i]. Fixes: 10377ba7673d ("net: systemport: Support 64bit statistics") Reviewed-by: Nicolai Buchwitz Signed-off-by: Florian Fainelli Link: https://patch.msgid.link/20260922232440.598918-2-florian.fainelli@broadcom.com Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/broadcom/bcmsysport.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/net/ethernet/broadcom/bcmsysport.c b/drivers/net/ethernet/broadcom/bcmsysport.c index 4d06c6ba6641..db627cd15fb7 100644 --- a/drivers/net/ethernet/broadcom/bcmsysport.c +++ b/drivers/net/ethernet/broadcom/bcmsysport.c @@ -482,10 +482,10 @@ static void bcm_sysport_get_stats(struct net_device *dev, s->type == BCM_SYSPORT_STAT_NETDEV64) { do { start = u64_stats_fetch_begin(syncp); - data[i] = *(u64 *)p; + data[j] = *(u64 *)p; } while (u64_stats_fetch_retry(syncp, start)); } else - data[i] = *(u32 *)p; + data[j] = *(u32 *)p; j++; } From 242935adfeb096f09b29fb2cfcf12f9eb2911fc6 Mon Sep 17 00:00:00 2001 From: Florian Fainelli Date: Tue, 22 Sep 2026 16:24:35 -0700 Subject: [PATCH 317/365] net: systemport: Fix RUNT MIB counter register offset calculation In UniMAC hardware, there is a 0xC byte gap between the RX MIB counters and the TX MIB counters, and a second 0xC byte gap between the TX MIB counters and the RX RUNT MIB counters. In bcm_sysport_update_mib_counters(), 'offset' was only set to UMAC_MIB_STAT_OFFSET (0xC) for all non-RX counters, omitting the second 0xC gap for BCM_SYSPORT_STAT_RUNT counters. As a result, all 4 RUNT MIB counters were read from unmapped gap register space. Fix this by setting offset to 2 * UMAC_MIB_STAT_OFFSET (0x18) when reading BCM_SYSPORT_STAT_RUNT counters. Fixes: 80105befdb4b ("net: systemport: add Broadcom SYSTEMPORT Ethernet MAC driver") Reviewed-by: Nicolai Buchwitz Signed-off-by: Florian Fainelli Link: https://patch.msgid.link/20260922232440.598918-6-florian.fainelli@broadcom.com Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/broadcom/bcmsysport.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/drivers/net/ethernet/broadcom/bcmsysport.c b/drivers/net/ethernet/broadcom/bcmsysport.c index db627cd15fb7..78909f516823 100644 --- a/drivers/net/ethernet/broadcom/bcmsysport.c +++ b/drivers/net/ethernet/broadcom/bcmsysport.c @@ -391,8 +391,10 @@ static void bcm_sysport_update_mib_counters(struct bcm_sysport_priv *priv) if (priv->is_lite) continue; - if (s->type != BCM_SYSPORT_STAT_MIB_RX) + if (s->type == BCM_SYSPORT_STAT_MIB_TX) offset = UMAC_MIB_STAT_OFFSET; + else if (s->type == BCM_SYSPORT_STAT_RUNT) + offset = 2 * UMAC_MIB_STAT_OFFSET; val = umac_readl(priv, UMAC_MIB_START + j + offset); break; case BCM_SYSPORT_STAT_RXCHK: From 04c824940d52871c49866a4ac67b504146f7cf35 Mon Sep 17 00:00:00 2001 From: Anirudh Virdi Date: Thu, 24 Sep 2026 12:12:44 +0530 Subject: [PATCH 318/365] net/mlx5: Fix slab-out-of-bounds when handling team device events The mlx5_handle_changeupper_event() and mlx5_handle_changeinfodata_event() functions check for LAG masters (which includes both bonding and team devices) but only call bonding-specific APIs. When processing team device events, calling bond_slave_get_rcu() and bond_is_slave_inactive() on team_port structures causes KASAN to detect an out-of-bounds memory access since team_port is smaller than bond slave. Fix this by wrapping the bond-specific API calls with a check for bonding devices. This allows the function to still process LAG events for both bonding and teams but only calls bond-specific functions when dealing with actual bonding devices. For mlx5_handle_changeinfodata_event(), keep the explicit bond check as mlx5 doesn't handle state information for team ports anyway. Tested with Mellanox ConnectX-5 on Linux 7.2.0-rc6: - Bonding: PASS (no regressions) - Team device: PASS (no KASAN errors) Fixes: 54493a08e21f ("net/mlx5: Lag, record inactive state of bond device") Suggested-by: Mark Bloch Signed-off-by: Anirudh Virdi Reviewed-by: Mark Bloch Link: https://patch.msgid.link/20260924064244.94045-1-avirdi@redhat.com Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c index dd14cdc378de..3b34bec559e0 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c @@ -1943,9 +1943,11 @@ static int mlx5_handle_changeupper_event(struct mlx5_lag *ldev, } } if (i < MLX5_MAX_PORTS) { - slave = bond_slave_get_rcu(ndev_tmp); - if (slave) - has_inactive |= bond_is_slave_inactive(slave); + if (netif_is_bond_master(upper)) { + slave = bond_slave_get_rcu(ndev_tmp); + if (slave) + has_inactive |= bond_is_slave_inactive(slave); + } bond_status |= (1 << idx); } @@ -2029,7 +2031,7 @@ static int mlx5_handle_changeinfodata_event(struct mlx5_lag *ldev, bool has_inactive = 0; int idx; - if (!netif_is_lag_master(ndev)) + if (!netif_is_bond_master(ndev)) return 0; rcu_read_lock(); From 056994563935e049093f100efaad7e63ad113814 Mon Sep 17 00:00:00 2001 From: Geetha sowjanya Date: Tue, 22 Sep 2026 15:51:48 +0530 Subject: [PATCH 319/365] octeontx2-pf: Fix aura BPID assignment when CONFIG_DCB is enabled Previously, BPID assignment under CONFIG_DCB assumed `queue_to_pfc_map` was always initialized. For SDP VFs this leads to invalid memory access as it was not initialized. This patch adds a NULL check for `queue_to_pfc_map` before dereferencing it. Also, simplifies the logic by always assigning a default BPID first, then conditionally overriding it if CONFIG_DCB is enabled and the map exists. Fixes: 184fb40f731b ("octeontx2-pf: Avoid adding dcbnl_ops for LBK and SDP vf") Signed-off-by: Geetha sowjanya Signed-off-by: Subrat Pandey Reviewed-by: Simon Horman Link: https://patch.msgid.link/20260922102149.2078168-2-subratp@marvell.com Signed-off-by: Jakub Kicinski --- .../net/ethernet/marvell/octeontx2/nic/cn20k.c | 11 +---------- .../ethernet/marvell/octeontx2/nic/otx2_common.c | 16 ++++++---------- .../ethernet/marvell/octeontx2/nic/otx2_common.h | 9 +++++++++ 3 files changed, 16 insertions(+), 20 deletions(-) diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/cn20k.c b/drivers/net/ethernet/marvell/octeontx2/nic/cn20k.c index 8e41431c7f9c..e92b77789f08 100644 --- a/drivers/net/ethernet/marvell/octeontx2/nic/cn20k.c +++ b/drivers/net/ethernet/marvell/octeontx2/nic/cn20k.c @@ -242,15 +242,6 @@ int cn20k_register_pfvf_mbox_intr(struct otx2_nic *pf, int numvfs) #define RQ_BP_LVL_AURA (255 - ((85 * 256) / 100)) /* BP when 85% is full */ -static u8 cn20k_aura_bpid_idx(struct otx2_nic *pfvf, int aura_id) -{ -#ifdef CONFIG_DCB - return pfvf->queue_to_pfc_map[aura_id]; -#else - return 0; -#endif -} - static int cn20k_tc_get_entry_index(struct otx2_flow_config *flow_cfg, struct otx2_tc_flow *node) { @@ -580,7 +571,7 @@ static int cn20k_aura_aq_init(struct otx2_nic *pfvf, int aura_id, if (pfvf->nix_blkaddr == BLKADDR_NIX1) aq->aura.bp_ena = 1; - bpid_idx = cn20k_aura_bpid_idx(pfvf, aura_id); + bpid_idx = otx2_get_bpid_idx(pfvf, aura_id); aq->aura.bpid = pfvf->bpid[bpid_idx]; /* Set backpressure level for RQ's Aura */ diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c index 175992188c18..0f3f405d9f1b 100644 --- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c @@ -1055,6 +1055,7 @@ int otx2_cq_init(struct otx2_nic *pfvf, u16 qidx) struct nix_aq_enq_req *aq; struct otx2_cq_queue *cq; struct otx2_pool *pool; + u8 bpid_idx; cq = &qset->cq[qidx]; cq->cq_idx = qidx; @@ -1132,11 +1133,8 @@ int otx2_cq_init(struct otx2_nic *pfvf, u16 qidx) if (!is_otx2_lbkvf(pfvf->pdev)) { /* Enable receive CQ backpressure */ aq->cq.bp_ena = 1; -#ifdef CONFIG_DCB - aq->cq.bpid = pfvf->bpid[pfvf->queue_to_pfc_map[qidx]]; -#else - aq->cq.bpid = pfvf->bpid[0]; -#endif + bpid_idx = otx2_get_bpid_idx(pfvf, qidx); + aq->cq.bpid = pfvf->bpid[bpid_idx]; /* Set backpressure level is same as cq pass level */ aq->cq.bp = RQ_PASS_LVL_CQ(pfvf->hw.rq_skid, qset->rqe_cnt); @@ -1378,6 +1376,7 @@ int otx2_aura_aq_init(struct otx2_nic *pfvf, int aura_id, { struct npa_aq_enq_req *aq; struct otx2_pool *pool; + u8 bpid_idx; int err; pool = &pfvf->qset.pool[pool_id]; @@ -1433,11 +1432,8 @@ int otx2_aura_aq_init(struct otx2_nic *pfvf, int aura_id, */ if (pfvf->nix_blkaddr == BLKADDR_NIX1) aq->aura.bp_ena = 1; -#ifdef CONFIG_DCB - aq->aura.nix0_bpid = pfvf->bpid[pfvf->queue_to_pfc_map[aura_id]]; -#else - aq->aura.nix0_bpid = pfvf->bpid[0]; -#endif + bpid_idx = otx2_get_bpid_idx(pfvf, aura_id); + aq->aura.nix0_bpid = pfvf->bpid[bpid_idx]; /* Set backpressure level for RQ's Aura */ aq->aura.bp = RQ_BP_LVL_AURA; diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h index eecee612b7b2..44cd56a0c2c4 100644 --- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h @@ -1029,6 +1029,15 @@ static inline int otx2_tc_flower_rule_cnt(struct otx2_nic *pfvf) return pfvf->flow_cfg->nr_flows; } +static inline u8 otx2_get_bpid_idx(struct otx2_nic *pfvf, int qidx) +{ +#ifdef CONFIG_DCB + if (pfvf->queue_to_pfc_map) + return pfvf->queue_to_pfc_map[qidx]; +#endif + return 0; +} + /* MSI-X APIs */ void otx2_free_cints(struct otx2_nic *pfvf, int n); void otx2_set_cints_affinity(struct otx2_nic *pfvf); From cdef07f3002f708a345392ee28d0d9bfac8383a1 Mon Sep 17 00:00:00 2001 From: Baha Mesleh Date: Tue, 22 Sep 2026 15:51:49 +0530 Subject: [PATCH 320/365] octeontx2-af: mcs: Fix SC resource cleanup loop The SC resource cleanup and statistics loops were incorrectly iterating over secy.max instead of sc.max, so the highest SC id could be left allocated across teardown or FLR. Use sc.max as the loop bound. Fixes: cfc14181d497 ("octeontx2-af: cn10k: mcs: Manage the MCS block hardware resources") Signed-off-by: Baha Mesleh Signed-off-by: Subrat Pandey Reviewed-by: Simon Horman Link: https://patch.msgid.link/20260922102149.2078168-3-subratp@marvell.com Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/marvell/octeontx2/af/mcs.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/net/ethernet/marvell/octeontx2/af/mcs.c b/drivers/net/ethernet/marvell/octeontx2/af/mcs.c index 211c10aa5880..30c672d526a0 100644 --- a/drivers/net/ethernet/marvell/octeontx2/af/mcs.c +++ b/drivers/net/ethernet/marvell/octeontx2/af/mcs.c @@ -314,7 +314,7 @@ int mcs_clear_all_stats(struct mcs *mcs, u16 pcifunc, int dir) } /* Clear SC stats */ - for (id = 0; id < map->secy.max; id++) { + for (id = 0; id < map->sc.max; id++) { if (map->sc2pf_map[id] != pcifunc) continue; mcs_clear_stats(mcs, MCS_SC_STATS, id, dir); @@ -738,7 +738,7 @@ int mcs_free_all_rsrc(struct mcs *mcs, int dir, u16 pcifunc) } /* free sc entries */ - for (id = 0; id < map->secy.max; id++) { + for (id = 0; id < map->sc.max; id++) { if (map->sc2pf_map[id] != pcifunc) continue; mcs_free_rsrc(&map->sc, map->sc2pf_map, id, pcifunc); From 512ccd3d0e91e791fb37442aa0a2599aca19783d Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?J=C3=A9r=C3=A9my=20Jean?= Date: Thu, 24 Sep 2026 20:21:05 +0000 Subject: [PATCH 321/365] tipc: prevent GCM nonce reuse on peer key changes MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit TIPC can encrypt traffic between nodes using a different transmit key for each node. In this mode, the AES-GCM nonce for a packet sent to a known peer consists of a 32-bit prefix (a per-key salt XOR the peer's address) followed by a 64-bit counter. That counter is stored in the peer's RX crypto object. When the peer reports a change in which key it uses to receive packets, TIPC resets this counter. The sender can still be using the same TX key and salt, so subsequent packets reuse earlier nonces. This nonce reuse breaks confidentiality and exposes GCM's authentication key. This makes forgeries trivial: an attacker can exploit CTR malleability to alter captured ciphertexts and use the recovered authentication key to compute a valid tag for the modified ciphertext, under the same key and nonce. Use the TX key's existing aead->seqno counter instead. All encryptions using that key object share the same atomic counter, so concurrent encryptions get distinct nonce counter values. The counter survives key activation and peer reconnection, and peer key-status reports cannot reset it. This prevents those transitions from causing nonce reuse while the same TX key remains installed. The nonce format is unchanged, and receivers do not require consecutive counter values, so sharing the counter across peers remains compatible with existing receivers. A pre-existing check still invokes key revocation in the unlikely event that the counter wraps to zero. Fixes: fc1b6d6de220 ("tipc: introduce TIPC encryption & authentication") Signed-off-by: Jérémy Jean Reviewed-by: Tung Nguyen Link: https://patch.msgid.link/20260924202105.3722778-1-Jeremy.Jean@oss.cyber.gouv.fr Signed-off-by: Jakub Kicinski --- net/tipc/crypto.c | 20 +++++--------------- 1 file changed, 5 insertions(+), 15 deletions(-) diff --git a/net/tipc/crypto.c b/net/tipc/crypto.c index 16f1ed1f6b1b..4409bdb70fe9 100644 --- a/net/tipc/crypto.c +++ b/net/tipc/crypto.c @@ -144,7 +144,7 @@ struct tipc_tfm { * @rcu: struct rcu_head * @key: the aead key * @gen: the key's generation - * @seqno: the key seqno (cluster scope) + * @seqno: the per-key TX nonce counter * @refcnt: the key reference counter */ struct tipc_aead { @@ -190,7 +190,6 @@ struct tipc_crypto_stats { * @rekeying_intv: rekeying interval (in minutes) * @stats: the crypto statistics * @name: the crypto name - * @sndnxt: the per-peer sndnxt (TX) * @timer1: general timer 1 (jiffies) * @timer2: general timer 2 (jiffies) * @working: the crypto is working or not @@ -219,7 +218,6 @@ struct tipc_crypto { struct tipc_crypto_stats __percpu *stats; char name[48]; - atomic64_t sndnxt ____cacheline_aligned; unsigned long timer1; unsigned long timer2; union { @@ -1051,14 +1049,11 @@ static int tipc_ehdr_build(struct net *net, struct tipc_aead *aead, WARN_ON(skb_headroom(skb) < ehsz); ehdr = (struct tipc_ehdr *)skb_push(skb, ehsz); - /* Obtain a seqno first: - * Use the key seqno (= cluster wise) if dest is unknown or we're in - * cluster key mode, otherwise it's better for a per-peer seqno! + /* + * Keep the nonce unique for the lifetime of the TX key, + * including key state changes and peer reconnection. */ - if (!__rx || aead->mode == CLUSTER_KEY) - seqno = atomic64_inc_return(&aead->seqno); - else - seqno = atomic64_inc_return(&__rx->sndnxt); + seqno = atomic64_inc_return(&aead->seqno); /* Revoke the key if seqno is wrapped around */ if (unlikely(!seqno)) @@ -1237,7 +1232,6 @@ void tipc_crypto_key_flush(struct tipc_crypto *c) tipc_crypto_key_set_state(c, 0, 0, 0); for (k = KEY_MIN; k <= KEY_MAX; k++) tipc_crypto_key_detach(c->aead[k], &c->lock); - atomic64_set(&c->sndnxt, 0); spin_unlock_bh(&c->lock); } @@ -1384,8 +1378,6 @@ static struct tipc_aead *tipc_crypto_key_pick_tx(struct tipc_crypto *tx, * It also considers if peer has no key, then we need to make own master key * (if any) taking over i.e. starting grace period and also trigger key * distributing process. - * - * The "per-peer" sndnxt is also reset when the peer key has switched. */ static void tipc_crypto_key_synch(struct tipc_crypto *rx, struct sk_buff *skb) { @@ -1436,7 +1428,6 @@ static void tipc_crypto_key_synch(struct tipc_crypto *rx, struct sk_buff *skb) if (cur) tipc_aead_users_dec(tx->aead[cur], 0); - atomic64_set(&rx->sndnxt, 0); /* Mark the point TX key users changed */ tx->timer1 = jiffies; @@ -1501,7 +1492,6 @@ int tipc_crypto_start(struct tipc_crypto **crypto, struct net *net, tipc_crypto_key_set_state(c, 0, 0, 0); atomic_set(&c->key_distr, 0); atomic_set(&c->peer_rx_active, 0); - atomic64_set(&c->sndnxt, 0); c->timer1 = jiffies; c->timer2 = jiffies; c->rekeying_intv = TIPC_REKEYING_INTV_DEF; From bdea4980182e3badac74f04df6c07492b91c569e Mon Sep 17 00:00:00 2001 From: Mina Almasry Date: Fri, 25 Sep 2026 14:41:11 +0000 Subject: [PATCH 322/365] net: page_pool: fix use-after-free in page_pool_recycle_ring_bulk() MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit With CONFIG_PAGE_POOL_STATS=y, page_pool_recycle_ring_bulk() updates the recycle stats after dropping the producer lock. If it just recycled the last inflight netmems of a pool being destroyed, page_pool_release() can pass its producer-lock barrier and free the pool before recycle_stat_add() runs. Commit fcc680a647ba7 ("page_pool: allow mixing PPs within one bulk") moved this stat update after the unlock. Commit 271683bb2cf32 ("page_pool: Fix use-after-free in page_pool_recycle_in_ring") later added the barrier, but only fixed page_pool_recycle_in_ring(). Move the update back under the lock. Fixes: fcc680a647ba7 ("page_pool: allow mixing PPs within one bulk") Link: https://lore.kernel.org/r/179028939171.2160803.706522228583664641@kernel.org Cc: Kaifeng Wang Cc: Dong Chenchen Signed-off-by: Mina Almasry Reviewed-by: Simon Horman Reviewed-by: Toke Høiland-Jørgensen Link: https://patch.msgid.link/20260925144127.1445667-1-almasrymina@google.com Signed-off-by: Jakub Kicinski --- net/core/page_pool.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/net/core/page_pool.c b/net/core/page_pool.c index 08d7f35cf608..d7c88c0b67e5 100644 --- a/net/core/page_pool.c +++ b/net/core/page_pool.c @@ -951,8 +951,8 @@ static void page_pool_recycle_ring_bulk(struct page_pool *pool, } } - page_pool_producer_unlock(pool, in_softirq); recycle_stat_add(pool, ring, i); + page_pool_producer_unlock(pool, in_softirq); /* Hopefully all pages were returned into ptr_ring */ if (likely(i == bulk_len)) From 88095728511e0df9c89528944f5a1762ad298983 Mon Sep 17 00:00:00 2001 From: Lorenzo Bianconi Date: Wed, 23 Sep 2026 11:14:28 +0200 Subject: [PATCH 323/365] net: stmmac: fix rx Scatter-Gather support When a received frame is larger than dma_buf_sz, the DMA scatters it across multiple RX descriptors (rx Scatter-Gather). The secondary RX buffer (sec_page) was only allocated and programmed when split-header (SPH) was active, so for regular frames buffer2 was neither allocated nor backed by a valid mapping. As soon as an incoming frame overflowed buffer1, the DMA wrote the overflow into the unmapped secondary-buffer address, triggering an SMMU translation fault on IOMMU-based platforms: arm-smmu 15000000.iommu: Unhandled context fault: fsr=0x402, iova=0x00000000, fsynr=0x7f0011, cbfrsynra=0x1c90, cb=11 arm-smmu 15000000.iommu: FSR = 00000402 [Format=2 TF], SID=0x1c90 arm-smmu 15000000.iommu: FSYNR0 = 007f0011 [S1CBNDX=127 WNR PLVL=1] Enable scatter-gather for non-SPH frames on cores that can program an independent secondary RX buffer (GMAC4/XGMAC): allocate and mark buffer2 as valid in stmmac_init_rx_buffers() and stmmac_rx_refill(), and account for it in the buffer length computation. Legacy cores have no set_sec_addr op, so they keep buffer2 disabled. Since buffer2 is handed to the DMA at page offset 0, the page pool sync window is widened to cover both buffers in every mode: offset is set to 0 and max_len to dma_buf_sz + stmmac_rx_offset(). The FCS can straddle the buffer1/buffer2 boundary and a descriptor boundary, so it is now stripped from the tail of the assembled frame with pskb_trim() instead of from a single buffer, avoiding an unsigned underflow for frames that overflow a buffer by 1..3 bytes. For single-buffer frames the XDP program must not see the FCS, so it is removed from the XDP buffer before the program runs. Native XDP currently only supports single-buffer (linear) frames. An oversized frame accepted by the MAC (jumbo enabled) is received via buffer2; the XDP program only sees buffer1, so on a TX/REDIRECT verdict the frame is forwarded truncated. XDP multi-buffer support to handle this case is planned as a follow-up. AF_XDP zero-copy RX is not covered by this change: a ZC queue still programs buffer2 at DMA address 0 (and XGMAC has no buffer2-valid bit), so an oversized frame overflowing buffer1 can still trigger the same SMMU translation fault. Handling is planned as a follow-up. Fixes: 88ebe2cf7f3f ("net: stmmac: Rework stmmac_rx()") Signed-off-by: Lorenzo Bianconi Link: https://patch.msgid.link/20260923-stmmac-rx-sg-fix-v3-1-ed26fea7180d@oss.qualcomm.com Signed-off-by: Jakub Kicinski --- .../net/ethernet/stmicro/stmmac/stmmac_main.c | 83 +++++++++++-------- 1 file changed, 49 insertions(+), 34 deletions(-) diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c index 3f34d491c959..ec62fa7418f4 100644 --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c @@ -1629,6 +1629,16 @@ static void stmmac_clear_descriptors(struct stmmac_priv *priv, stmmac_clear_tx_descriptors(priv, dma_conf, queue); } +static bool stmmac_rx_check_buf2_cap(struct stmmac_priv *priv) +{ + /* Only cores that can program an independent secondary RX buffer + * (used for scatter-gather overflow or split-header payload) back + * buffer2. Legacy cores have no set_sec_addr op, so buffer2 is + * never handed to the hardware there. + */ + return priv->hw->desc && priv->hw->desc->set_sec_addr; +} + /** * stmmac_init_rx_buffers - init the RX descriptor buffer. * @priv: driver private structure @@ -1659,16 +1669,13 @@ static int stmmac_init_rx_buffers(struct stmmac_priv *priv, buf->page_offset = stmmac_rx_offset(priv); } - if (priv->sph_active && !buf->sec_page) { + if (stmmac_rx_check_buf2_cap(priv) && !buf->sec_page) { buf->sec_page = page_pool_alloc_pages(rx_q->page_pool, gfp); if (!buf->sec_page) return -ENOMEM; buf->sec_addr = page_pool_get_dma_addr(buf->sec_page); stmmac_set_desc_sec_addr(priv, p, buf->sec_addr, true); - } else { - buf->sec_page = NULL; - stmmac_set_desc_sec_addr(priv, p, buf->sec_addr, false); } buf->addr = page_pool_get_dma_addr(buf->page) + buf->page_offset; @@ -2256,13 +2263,8 @@ static int __alloc_dma_rx_desc_resources(struct stmmac_priv *priv, pp_params.nid = dev_to_node(priv->device); pp_params.dev = priv->device; pp_params.dma_dir = xdp_prog ? DMA_BIDIRECTIONAL : DMA_FROM_DEVICE; - pp_params.offset = stmmac_rx_offset(priv); - pp_params.max_len = dma_conf->dma_buf_sz; - - if (priv->sph_active) { - pp_params.offset = 0; - pp_params.max_len += stmmac_rx_offset(priv); - } + pp_params.offset = 0; + pp_params.max_len = dma_conf->dma_buf_sz + stmmac_rx_offset(priv); rx_q->page_pool = page_pool_create(&pp_params); if (IS_ERR(rx_q->page_pool)) { @@ -5099,7 +5101,7 @@ static inline void stmmac_rx_refill(struct stmmac_priv *priv, u32 queue) break; } - if (priv->sph_active && !buf->sec_page) { + if (stmmac_rx_check_buf2_cap(priv) && !buf->sec_page) { buf->sec_page = page_pool_alloc_pages(rx_q->page_pool, gfp); if (!buf->sec_page) break; @@ -5110,10 +5112,8 @@ static inline void stmmac_rx_refill(struct stmmac_priv *priv, u32 queue) buf->addr = page_pool_get_dma_addr(buf->page) + buf->page_offset; stmmac_set_desc_addr(priv, p, buf->addr); - if (priv->sph_active) - stmmac_set_desc_sec_addr(priv, p, buf->sec_addr, true); - else - stmmac_set_desc_sec_addr(priv, p, buf->sec_addr, false); + stmmac_set_desc_sec_addr(priv, p, buf->sec_addr, + stmmac_rx_check_buf2_cap(priv)); stmmac_refill_desc3(priv, rx_q, p); rx_q->rx_count_frames++; @@ -5144,7 +5144,9 @@ static unsigned int stmmac_rx_buf1_len(struct stmmac_priv *priv, unsigned int plen = 0, hlen = 0; int coe = priv->hw->rx_csum; - /* Not first descriptor, buffer is always zero */ + /* Not first descriptor, SPH enabled: buffer1 only carries the + * split header of the first descriptor, so it is zero here. + */ if (priv->sph_active && len) return 0; @@ -5155,14 +5157,16 @@ static unsigned int stmmac_rx_buf1_len(struct stmmac_priv *priv, return hlen; } - /* First descriptor, not last descriptor and not split header */ + /* Not last descriptor and not split header: buffer1 is fully filled */ if (status & rx_not_ls) return priv->dma_conf.dma_buf_sz; plen = stmmac_get_rx_frame_len(priv, p, coe); - /* First descriptor and last descriptor and not split header */ - return min_t(unsigned int, priv->dma_conf.dma_buf_sz, plen); + /* Last descriptor and not split header: buffer1 holds the remaining + * bytes of the frame, up to dma_buf_sz + */ + return min_t(unsigned int, priv->dma_conf.dma_buf_sz, plen - len); } static unsigned int stmmac_rx_buf2_len(struct stmmac_priv *priv, @@ -5172,8 +5176,7 @@ static unsigned int stmmac_rx_buf2_len(struct stmmac_priv *priv, int coe = priv->hw->rx_csum; unsigned int plen = 0; - /* Not split header, buffer is not available */ - if (!priv->sph_active) + if (!stmmac_rx_check_buf2_cap(priv)) return 0; /* For GMAC4, when split header is enabled, in some rare cases, the @@ -5190,14 +5193,15 @@ static unsigned int stmmac_rx_buf2_len(struct stmmac_priv *priv, * Thus 'plen - len' always gives the correct length of buf2. */ - /* Not GMAC4 and not last descriptor */ - if (priv->plat->core_type != DWMAC_CORE_GMAC4 && (status & rx_not_ls)) + /* Not GMAC4, or non-SPH and not last descriptor */ + if ((priv->plat->core_type != DWMAC_CORE_GMAC4 || !priv->sph_active) && + (status & rx_not_ls)) return priv->dma_conf.dma_buf_sz; /* GMAC4 or last descriptor */ plen = stmmac_get_rx_frame_len(priv, p, coe); - return plen - len; + return plen > len ? plen - len : 0; } static int stmmac_xdp_xmit_xdpf(struct stmmac_priv *priv, int queue, @@ -5720,6 +5724,7 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue) unsigned int desc_size; struct sk_buff *skb = NULL; struct stmmac_xdp_buff ctx; + bool fcs_stripped = false; int xdp_status = 0; int bufsz; @@ -5809,24 +5814,25 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue) len += buf2_len; /* ACS is disabled; strip manually. */ - if (likely(!(status & rx_not_ls))) { - if (buf2_len) { - buf2_len -= ETH_FCS_LEN; - len -= ETH_FCS_LEN; - } else if (buf1_len) { - buf1_len -= ETH_FCS_LEN; - len -= ETH_FCS_LEN; - } - } + if (likely(!(status & rx_not_ls))) + len -= ETH_FCS_LEN; if (!skb) { unsigned int pre_len, sync_len; + /* Each frame starts here: reset the FCS handling */ + fcs_stripped = false; + dma_sync_single_for_cpu(priv->device, buf->addr, buf1_len, dma_dir); net_prefetch(page_address(buf->page) + buf->page_offset); + if (stmmac_xdp_is_enabled(priv) && !buf2_len) { + buf1_len -= ETH_FCS_LEN; + fcs_stripped = true; + } + xdp_init_buff(&ctx.xdp, bufsz, &rx_q->xdp_rxq); xdp_prepare_buff(&ctx.xdp, page_address(buf->page), buf->page_offset, buf1_len, true); @@ -5927,6 +5933,15 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue) /* Got entire packet into SKB. Finish it. */ + /* Remove FCS if needed */ + if (!fcs_stripped && pskb_trim(skb, len)) { + dev_kfree_skb_any(skb); + skb = NULL; + rx_dropped++; + count++; + continue; + } + stmmac_get_rx_hwtstamp(priv, p, np, skb); if (priv->hw->hw_vlan_en) From 8ae3f4067ba4ae3a15581392f6dac422e3d3def0 Mon Sep 17 00:00:00 2001 From: Andrea Mayer Date: Fri, 25 Sep 2026 15:38:07 +0200 Subject: [PATCH 324/365] seg6: ensure packet data is writable before modifying SRH and IPv6 DA advance_nextseg() modifies the SRH Segments Left field and the IPv6 destination address without ensuring the packet data is writable. seg6_next_csid_advance_arg() has the same problem when it advances the NEXT-C-SID argument in the destination address. The skb may be cloned, for example by an AF_PACKET socket receiving on the ingress device. advance_nextseg() and seg6_next_csid_advance_arg() then write into the packet data shared with the clone. A read from that socket can return the modified packet instead of the received one. The simplified path below shows this for advance_nextseg(): __netif_receive_skb_one_core __netif_receive_skb_core deliver_skb [orig: users=2, cloned=0] packet_rcv skb_clone clone queued to the AF_PACKET socket consume_skb(orig) [orig: users=1, cloned=1] ipv6_rcv ip6_rcv_core skb_share_check: no-op [orig: users=1] [...] input_action_end_core advance_nextseg writes into the data shared with the clone Call skb_ensure_writable() in advance_nextseg() and in seg6_next_csid_advance_arg() before they modify the packet data. skb_ensure_writable() may reallocate skb->head, which invalidates the pointers into the packet data taken before the call. advance_nextseg() now returns a valid SRH pointer, or NULL if skb_ensure_writable() fails. seg6_next_csid_advance_arg() takes the pointer to the destination address from the skb after skb_ensure_writable(). On failure, the callers drop the packet with SKB_DROP_REASON_NOMEM. Fixes: 140f04c33bbc ("ipv6: sr: implement several seg6local actions") Fixes: 848f3c0d4769 ("seg6: add NEXT-C-SID support for SRv6 End behavior") Signed-off-by: Andrea Mayer Link: https://patch.msgid.link/20260925133807.32-1-andrea.mayer@uniroma2.it Signed-off-by: Jakub Kicinski --- net/ipv6/seg6_local.c | 140 +++++++++++++++++++++++++++++++++++------- 1 file changed, 117 insertions(+), 23 deletions(-) diff --git a/net/ipv6/seg6_local.c b/net/ipv6/seg6_local.c index d1070aec7b72..53f7b2452575 100644 --- a/net/ipv6/seg6_local.c +++ b/net/ipv6/seg6_local.c @@ -278,13 +278,43 @@ static bool decap_and_validate(struct sk_buff *skb, int proto) return true; } -static void advance_nextseg(struct ipv6_sr_hdr *srh, struct in6_addr *daddr) +/* advance the SRH to the next segment and set the IPv6 DA accordingly. + * skb pointers may change: after this call, the caller must evaluate again + * any pointer into the packet data. + * + * This function returns: + * - the SRH on success; + * - NULL when the skb cannot be made writable. In this case, the function + * sets the drop reason to SKB_DROP_REASON_NOMEM. + */ +static struct ipv6_sr_hdr *advance_nextseg(struct sk_buff *skb, + struct ipv6_sr_hdr *srh, + enum skb_drop_reason *reason) { struct in6_addr *addr; + int srhoff; + int wlen; + + srhoff = (unsigned char *)srh - skb->data; + /* we write only the Segment Left field and the IPv6 DA. The segment + * list is only read, and seg6_get_srh() already pulled it into the + * linear area. + */ + wlen = srhoff + sizeof(*srh); + + if (unlikely(skb_ensure_writable(skb, wlen))) { + *reason = SKB_DROP_REASON_NOMEM; + return NULL; + } + + /* skb_ensure_writable() may change skb pointers; evaluate srh again */ + srh = (struct ipv6_sr_hdr *)(skb->data + srhoff); srh->segments_left--; addr = srh->segments + srh->segments_left; - *daddr = *addr; + ipv6_hdr(skb)->daddr = *addr; + + return srh; } static int @@ -382,12 +412,26 @@ static bool seg6_next_csid_is_arg_zero(const struct in6_addr *addr, return true; } -/* assume that DA.Argument length > 0 */ -static void seg6_next_csid_advance_arg(struct in6_addr *addr, - const struct seg6_flavors_info *finfo) +/* assume that DA.Argument length > 0. + * skb pointers may change: after this call, the caller must evaluate again + * any pointer into the packet data. + * + * This function returns: + * - SKB_NOT_DROPPED_YET on success; + * - SKB_DROP_REASON_NOMEM when the skb cannot be made writable. + */ +static enum skb_drop_reason +seg6_next_csid_advance_arg(struct sk_buff *skb, + const struct seg6_flavors_info *finfo) { __u8 fnc_octects = seg6_flv_lcnode_func_octects(finfo); __u8 blk_octects = seg6_flv_lcblock_octects(finfo); + struct in6_addr *addr; + + if (unlikely(skb_ensure_writable(skb, sizeof(struct ipv6hdr)))) + return SKB_DROP_REASON_NOMEM; + + addr = &ipv6_hdr(skb)->daddr; /* advance DA.Argument */ memmove(&addr->s6_addr[blk_octects], @@ -395,6 +439,8 @@ static void seg6_next_csid_advance_arg(struct in6_addr *addr, 16 - blk_octects - fnc_octects); memset(&addr->s6_addr[16 - fnc_octects], 0x00, fnc_octects); + + return SKB_NOT_DROPPED_YET; } static int input_action_end_finish(struct sk_buff *skb, @@ -408,31 +454,42 @@ static int input_action_end_finish(struct sk_buff *skb, static int input_action_end_core(struct sk_buff *skb, struct seg6_local_lwt *slwt) { + enum skb_drop_reason reason = SKB_DROP_REASON_NOT_SPECIFIED; struct ipv6_sr_hdr *srh; + int err = -EINVAL; srh = get_and_validate_srh(skb); if (!srh) goto drop; - advance_nextseg(srh, &ipv6_hdr(skb)->daddr); + srh = advance_nextseg(skb, srh, &reason); + if (!srh) { + err = -ENOMEM; + goto drop; + } return input_action_end_finish(skb, slwt); drop: - kfree_skb(skb); - return -EINVAL; + kfree_skb_reason(skb, reason); + return err; } static int end_next_csid_core(struct sk_buff *skb, struct seg6_local_lwt *slwt) { const struct seg6_flavors_info *finfo = &slwt->flv_info; struct in6_addr *daddr = &ipv6_hdr(skb)->daddr; + enum skb_drop_reason reason; if (seg6_next_csid_is_arg_zero(daddr, finfo)) return input_action_end_core(skb, slwt); /* update DA */ - seg6_next_csid_advance_arg(daddr, finfo); + reason = seg6_next_csid_advance_arg(skb, finfo); + if (reason) { + kfree_skb_reason(skb, reason); + return -ENOMEM; + } return input_action_end_finish(skb, slwt); } @@ -448,19 +505,25 @@ static int input_action_end_x_finish(struct sk_buff *skb, static int input_action_end_x_core(struct sk_buff *skb, struct seg6_local_lwt *slwt) { + enum skb_drop_reason reason = SKB_DROP_REASON_NOT_SPECIFIED; struct ipv6_sr_hdr *srh; + int err = -EINVAL; srh = get_and_validate_srh(skb); if (!srh) goto drop; - advance_nextseg(srh, &ipv6_hdr(skb)->daddr); + srh = advance_nextseg(skb, srh, &reason); + if (!srh) { + err = -ENOMEM; + goto drop; + } return input_action_end_x_finish(skb, slwt); drop: - kfree_skb(skb); - return -EINVAL; + kfree_skb_reason(skb, reason); + return err; } static int end_x_next_csid_core(struct sk_buff *skb, @@ -468,12 +531,17 @@ static int end_x_next_csid_core(struct sk_buff *skb, { const struct seg6_flavors_info *finfo = &slwt->flv_info; struct in6_addr *daddr = &ipv6_hdr(skb)->daddr; + enum skb_drop_reason reason; if (seg6_next_csid_is_arg_zero(daddr, finfo)) return input_action_end_x_core(skb, slwt); /* update DA */ - seg6_next_csid_advance_arg(daddr, finfo); + reason = seg6_next_csid_advance_arg(skb, finfo); + if (reason) { + kfree_skb_reason(skb, reason); + return -ENOMEM; + } return input_action_end_x_finish(skb, slwt); } @@ -760,10 +828,12 @@ static bool seg6_pop_srh(struct sk_buff *skb, int srhoff) */ static int end_flv8986_core(struct sk_buff *skb, struct seg6_local_lwt *slwt) { + enum skb_drop_reason reason = SKB_DROP_REASON_NOT_SPECIFIED; const struct seg6_flavors_info *finfo = &slwt->flv_info; enum seg6_local_flv_action action; enum seg6_local_pktinfo pinfo; struct ipv6_sr_hdr *srh; + int err = -EINVAL; __u32 flvmask; int srhoff; @@ -787,10 +857,18 @@ static int end_flv8986_core(struct sk_buff *skb, struct seg6_local_lwt *slwt) switch (action) { case SEG6_LOCAL_FLV_ACT_END: /* process the packet as the "standard" End behavior */ - advance_nextseg(srh, &ipv6_hdr(skb)->daddr); + srh = advance_nextseg(skb, srh, &reason); + if (!srh) { + err = -ENOMEM; + goto drop; + } break; case SEG6_LOCAL_FLV_ACT_PSP: - advance_nextseg(srh, &ipv6_hdr(skb)->daddr); + srh = advance_nextseg(skb, srh, &reason); + if (!srh) { + err = -ENOMEM; + goto drop; + } if (unlikely(!seg6_pop_srh(skb, srhoff))) goto drop; @@ -807,8 +885,8 @@ static int end_flv8986_core(struct sk_buff *skb, struct seg6_local_lwt *slwt) return input_action_end_finish(skb, slwt); drop: - kfree_skb(skb); - return -EINVAL; + kfree_skb_reason(skb, reason); + return err; } /* regular endpoint function */ @@ -847,21 +925,27 @@ static int input_action_end_x(struct sk_buff *skb, struct seg6_local_lwt *slwt) static int input_action_end_t(struct sk_buff *skb, struct seg6_local_lwt *slwt) { + enum skb_drop_reason reason = SKB_DROP_REASON_NOT_SPECIFIED; struct ipv6_sr_hdr *srh; + int err = -EINVAL; srh = get_and_validate_srh(skb); if (!srh) goto drop; - advance_nextseg(srh, &ipv6_hdr(skb)->daddr); + srh = advance_nextseg(skb, srh, &reason); + if (!srh) { + err = -ENOMEM; + goto drop; + } seg6_lookup_nexthop(skb, NULL, slwt->table); return dst_input(skb); drop: - kfree_skb(skb); - return -EINVAL; + kfree_skb_reason(skb, reason); + return err; } /* decapsulate and forward inner L2 frame on specified interface */ @@ -1375,6 +1459,7 @@ static int input_action_end_b6(struct sk_buff *skb, struct seg6_local_lwt *slwt) static int input_action_end_b6_encap(struct sk_buff *skb, struct seg6_local_lwt *slwt) { + enum skb_drop_reason reason = SKB_DROP_REASON_NOT_SPECIFIED; struct ipv6_sr_hdr *srh; int err = -EINVAL; @@ -1382,7 +1467,11 @@ static int input_action_end_b6_encap(struct sk_buff *skb, if (!srh) goto drop; - advance_nextseg(srh, &ipv6_hdr(skb)->daddr); + srh = advance_nextseg(skb, srh, &reason); + if (!srh) { + err = -ENOMEM; + goto drop; + } skb_reset_inner_headers(skb); skb->encapsulation = 1; @@ -1398,7 +1487,7 @@ static int input_action_end_b6_encap(struct sk_buff *skb, return dst_input(skb); drop: - kfree_skb(skb); + kfree_skb_reason(skb, reason); return err; } @@ -1434,6 +1523,7 @@ static int input_action_end_bpf(struct sk_buff *skb, struct seg6_local_lwt *slwt) { struct seg6_bpf_srh_state *srh_state; + enum skb_drop_reason reason; struct ipv6_sr_hdr *srh; int ret; @@ -1442,7 +1532,11 @@ static int input_action_end_bpf(struct sk_buff *skb, kfree_skb(skb); return -EINVAL; } - advance_nextseg(srh, &ipv6_hdr(skb)->daddr); + srh = advance_nextseg(skb, srh, &reason); + if (!srh) { + kfree_skb_reason(skb, reason); + return -ENOMEM; + } /* The access to the per-CPU buffer srh_state is protected by running * always in softirq context (with disabled BH). On PREEMPT_RT the From 036778a4fad19b54b73ae229a1901665aa8750a2 Mon Sep 17 00:00:00 2001 From: Nicolai Buchwitz Date: Fri, 25 Sep 2026 15:52:36 +0200 Subject: [PATCH 325/365] net: ethtool: let tsconfig reach a PHY-only timestamp provider TSCONFIG_GET and TSCONFIG_SET reject a device that implements neither hwtstamp NDO, even when its PHY can serve the request. The ioctls they meant to replace handle it, so the two interfaces disagree on the same hardware and user space has to pick one. Accept the default timestamping PHY and an already installed provider on both sides. On the set side the test moves into ethnl_set_tsconfig() as the validate callback runs without rtnl. On the get side it stays ahead of ethnl_ops_begin(), so a device that can serve nothing keeps failing with EOPNOTSUPP and a dump still skips it rather than stopping there. A netdev provider needs ndo_hwtstamp_set to be programmed at all, so don't pick that source without it, and test for ndo_hwtstamp_get before calling it. Fixes: 6e9e2eed4f39 ("net: ethtool: Add support for tsconfig command to get/set hwtstamp config") Signed-off-by: Nicolai Buchwitz Reviewed-by: Maxime Chevallier Reviewed-by: Kory Maincent Link: https://patch.msgid.link/20260925135237.3432266-3-nb@tipi-net.de Signed-off-by: Jakub Kicinski --- net/core/dev_ioctl.c | 3 ++- net/ethtool/tsconfig.c | 24 ++++++++++-------------- 2 files changed, 12 insertions(+), 15 deletions(-) diff --git a/net/core/dev_ioctl.c b/net/core/dev_ioctl.c index 164643140a52..f6029f60c1dc 100644 --- a/net/core/dev_ioctl.c +++ b/net/core/dev_ioctl.c @@ -267,7 +267,8 @@ int dev_get_hwtstamp_phylib(struct net_device *dev, hwprov->phydev) return phy_hwtstamp_get(hwprov->phydev, cfg); - if (hwprov->source == HWTSTAMP_SOURCE_NETDEV) + if (hwprov->source == HWTSTAMP_SOURCE_NETDEV && + dev->netdev_ops->ndo_hwtstamp_get) return dev->netdev_ops->ndo_hwtstamp_get(dev, cfg); return -EOPNOTSUPP; diff --git a/net/ethtool/tsconfig.c b/net/ethtool/tsconfig.c index 6be3aa5d4bc1..2db0e7ba8b9f 100644 --- a/net/ethtool/tsconfig.c +++ b/net/ethtool/tsconfig.c @@ -1,6 +1,7 @@ // SPDX-License-Identifier: GPL-2.0-only #include +#include #include #include @@ -42,7 +43,9 @@ static int tsconfig_prepare_data(const struct ethnl_req_info *req_base, struct kernel_hwtstamp_config cfg = {}; int ret; - if (!dev->netdev_ops->ndo_hwtstamp_get) + if (!dev->netdev_ops->ndo_hwtstamp_get && + !phy_is_default_hwtstamp(dev->phydev) && + !netdev_ops_lock_dereference(dev->hwprov, dev)) return -EOPNOTSUPP; ret = ethnl_ops_begin(dev); @@ -248,17 +251,6 @@ static int tsconfig_send_reply(struct net_device *dev, struct genl_info *info) return ret; } -static int ethnl_set_tsconfig_validate(struct ethnl_req_info *req_base, - struct genl_info *info) -{ - const struct net_device_ops *ops = req_base->dev->netdev_ops; - - if (!ops->ndo_hwtstamp_set || !ops->ndo_hwtstamp_get) - return -EOPNOTSUPP; - - return 1; -} - static struct hwtstamp_provider * tsconfig_set_hwprov_from_desc(struct net_device *dev, struct genl_info *info, @@ -272,7 +264,7 @@ tsconfig_set_hwprov_from_desc(struct net_device *dev, int ret; ret = ethtool_net_get_ts_info_by_phc(dev, &ts_info, hwprov_desc); - if (!ret) { + if (!ret && dev->netdev_ops->ndo_hwtstamp_set) { /* Found */ source = HWTSTAMP_SOURCE_NETDEV; } else { @@ -313,6 +305,11 @@ static int ethnl_set_tsconfig(struct ethnl_req_info *req_base, if (!netif_device_present(dev)) return -ENODEV; + if (!dev->netdev_ops->ndo_hwtstamp_set && + !phy_is_default_hwtstamp(dev->phydev) && + !netdev_ops_lock_dereference(dev->hwprov, dev)) + return -EOPNOTSUPP; + if (tb[ETHTOOL_A_TSCONFIG_HWTSTAMP_PROVIDER]) { struct hwtstamp_provider_desc __hwprov_desc = {.index = -1}; struct hwtstamp_provider *__hwprov; @@ -459,6 +456,5 @@ const struct ethnl_request_ops ethnl_tsconfig_request_ops = { .reply_size = tsconfig_reply_size, .fill_reply = tsconfig_fill_reply, - .set_validate = ethnl_set_tsconfig_validate, .set = ethnl_set_tsconfig, }; From 37e02c42a00be692c06343e11779aadc45f45971 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Fri, 25 Sep 2026 13:52:43 +0000 Subject: [PATCH 326/365] net: restrict SO_RESERVE_MEM to TCP sockets and cap max value Commit 2bb2f5fb21b0 ("net: add new socket option SO_RESERVE_MEM") and commit d00c8ee31729 ("net: fix possible NULL deref in sock_reserve_memory") only checked sk_has_account(sk), which is true for both TCP and UDP sockets. However, SO_RESERVE_MEM and sk_unused_reserved_mem() are currently only supported by TCP: - On UDP sockets, sk->sk_forward_alloc is protected by sk->sk_receive_queue.lock, whereas sock_reserve_memory() and sock_release_reserved_memory() only acquire lock_sock(sk). Concurrent UDP packet reception/release and setsockopt(SO_RESERVE_MEM) corrupt sk_forward_alloc and memcg accounting. - udp_rmem_release() reclaims excess sk_forward_alloc without accounting for sk_unused_reserved_mem(sk). Restrict sock_reserve_memory() to TCP sockets (sk_is_tcp(sk)) for now. Supporting SO_RESERVE_MEM for UDP (acquiring sk_receive_queue.lock and honoring sk_unused_reserved_mem() in udp_rmem_release()) can be done in a future net-next series if needed. In addition, reject val > SZ_1G with -EINVAL in sk_setsockopt(SO_RESERVE_MEM). Without an upper bound, values near INT_MAX cause sk_mem_pages(delta) and (pages << PAGE_SHIFT) to overflow 32-bit signed int, corrupting sk->sk_forward_alloc and sk->sk_reserved_mem. Using a page-aligned cap (SZ_1G) ensures that the page-rounded sk->sk_reserved_mem reported by getsockopt(SO_RESERVE_MEM) can always be passed back to setsockopt(SO_RESERVE_MEM). Fixes: 2bb2f5fb21b0 ("net: add new socket option SO_RESERVE_MEM") Reported-by: Cai Xinchen Closes: https://lore.kernel.org/netdev/5a88421d-10ef-4fca-9acb-85a27a3c1173@huawei.com/ Signed-off-by: Eric Dumazet Reviewed-by: Wei Wang Link: https://patch.msgid.link/20260925135244.3715196-2-edumazet@google.com Signed-off-by: Jakub Kicinski --- net/core/sock.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/net/core/sock.c b/net/core/sock.c index d23333bb4f3f..e8551df8330f 100644 --- a/net/core/sock.c +++ b/net/core/sock.c @@ -1034,7 +1034,7 @@ static int sock_reserve_memory(struct sock *sk, int bytes) bool charged; int pages; - if (!mem_cgroup_sk_enabled(sk) || !sk_has_account(sk)) + if (!mem_cgroup_sk_enabled(sk) || !sk_is_tcp(sk)) return -EOPNOTSUPP; if (!bytes) @@ -1661,7 +1661,7 @@ int sk_setsockopt(struct sock *sk, int level, int optname, { int delta; - if (val < 0) { + if (val < 0 || val > SZ_1G) { ret = -EINVAL; break; } From e955f14fa8467d70aaeea5830b4d1773d5b25500 Mon Sep 17 00:00:00 2001 From: Zhan Xusheng Date: Mon, 28 Sep 2026 10:02:04 +0800 Subject: [PATCH 327/365] sysctl: Negate before converting in the int read path proc_int_k2u_conv_kop() reports the sign through *negp and the magnitude through *u_ptr, but for a negative value it hands the sign-extended int to the converter and negates the result: *u_ptr = k_ptr_op ? -k_ptr_op((ulong)val) : -(ulong)val; With div_hz() and CONFIG_HZ=1000 a stored -1000 becomes (ulong)-1000 / 1000 == 18446744073709550, and negating that wraps: # echo -1 > /proc/sys/net/ipv4/tcp_fin_timeout # cat /proc/sys/net/ipv4/tcp_fin_timeout -18428297329635842066 The magnitude is HZ dependent but the wrap is not: the negation happens after the division at every CONFIG_HZ. Take the magnitude first and convert that, which is what the open-coded version did before commit 2dc164a48e6f ("sysctl: Create converter functions with two new macros") folded it into a macro. The k_ptr_op == NULL branch was already correct. All three int converters that pass a k_ptr_op are affected: proc_dointvec_jiffies(), proc_dointvec_userhz_jiffies() and proc_dointvec_ms_jiffies(). Fixes: 2dc164a48e6f ("sysctl: Create converter functions with two new macros") Cc: stable@vger.kernel.org Reviewed-by: Bradley Morgan Signed-off-by: Zhan Xusheng Signed-off-by: Joel Granados --- kernel/sysctl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/kernel/sysctl.c b/kernel/sysctl.c index f7b75985d542..38597f26b34c 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -483,7 +483,7 @@ int proc_int_k2u_conv_kop(ulong *u_ptr, const int *k_ptr, bool *negp, if (val < 0) { *negp = true; - *u_ptr = k_ptr_op ? -k_ptr_op((ulong)val) : -(ulong)val; + *u_ptr = k_ptr_op ? k_ptr_op(-(ulong)val) : -(ulong)val; } else { *negp = false; *u_ptr = k_ptr_op ? k_ptr_op((ulong)val) : (ulong) val; From fdd6553e1e53b0421d89c7469c8a36d2eadb1115 Mon Sep 17 00:00:00 2001 From: Zhan Xusheng Date: Mon, 28 Sep 2026 10:02:05 +0800 Subject: [PATCH 328/365] time/jiffies: Saturate in mult_hz() instead of wrapping Return ULONG_MAX for values that don't fit an unsigned long. proc_int_u2k_conv_uop() now correctly rejects a result above INT_MAX. This is the erroneous behaviour that is being fixed. The input has to exceed ULONG_MAX / HZ for the product to wrap, so the value below is specific to CONFIG_HZ=1000: # echo 18446744073709552 > /proc/sys/net/ipv4/tcp_keepalive_time # cat /proc/sys/net/ipv4/tcp_keepalive_time 0 That value is now rejected with an error. The original bound ("*u_ptr > INT_MAX / HZ") was removed in commit 2dc164a48e6f ("sysctl: Create converter functions with two new macros"). Fixes: 2dc164a48e6f ("sysctl: Create converter functions with two new macros") Cc: stable@vger.kernel.org Signed-off-by: Zhan Xusheng Signed-off-by: Joel Granados --- kernel/time/jiffies.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/kernel/time/jiffies.c b/kernel/time/jiffies.c index 80c354811538..9b3487d40cd6 100644 --- a/kernel/time/jiffies.c +++ b/kernel/time/jiffies.c @@ -101,6 +101,8 @@ void __init register_refined_jiffies(long cycles_per_second) #ifdef CONFIG_SYSCTL static ulong mult_hz(const ulong val) { + if (val >= ULONG_MAX / HZ) + return ULONG_MAX; return val * HZ; } From 0f17d2e3dd9cda3ee6d94033d5e5489af168920d Mon Sep 17 00:00:00 2001 From: Xuanqiang Luo Date: Thu, 24 Sep 2026 13:52:52 +0800 Subject: [PATCH 329/365] pfcp: fix socket lifetime on netdevice registration failure pfcp_newlink() creates the UDP socket before register_netdevice(). If registration fails after pfcp_dev_init() succeeds, the core calls pfcp_dev_uninit(), which releases the socket and clears pfcp->sk. The newlink error path then releases it again, causing a NULL pointer dereference. This was observed with failslab fault injection: FAULT_INJECTION: forcing a failure. kobject: kobject_add_internal failed for pfcp0 (error: -12 parent: net) BUG: KASAN: null-ptr-deref in udp_tunnel_sock_release+0x1c/0x50 Read of size 8 at addr 0000000000000120 by task ip/1037 Call trace: show_stack+0x18/0x24 (C) dump_stack_lvl+0x78/0x90 print_report+0x468/0x5cc kasan_report+0xa4/0xf0 __asan_load8+0x7c/0xd0 udp_tunnel_sock_release+0x1c/0x50 pfcp_newlink+0x128/0x184 rtnl_newlink+0x848/0xe44 rtnetlink_rcv_msg+0x468/0x514 netlink_rcv_skb+0xc0/0x1f0 rtnetlink_rcv+0x18/0x24 netlink_unicast+0x4b8/0x558 netlink_sendmsg+0x2b8/0x584 ... Return from pfcp_del_sock() if pfcp->sk is NULL. Fixes: 76c8764ef36a5 ("pfcp: add PFCP module") Cc: stable@vger.kernel.org Signed-off-by: Xuanqiang Luo Reviewed-by: Simon Horman Link: https://patch.msgid.link/20260924055252.33488-1-xuanqiang.luo@linux.dev Signed-off-by: Paolo Abeni --- drivers/net/pfcp.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/drivers/net/pfcp.c b/drivers/net/pfcp.c index e1cca779d2ec..3ecd43cafdf0 100644 --- a/drivers/net/pfcp.c +++ b/drivers/net/pfcp.c @@ -105,6 +105,9 @@ static int pfcp_encap_recv(struct sock *sk, struct sk_buff *skb) static void pfcp_del_sock(struct pfcp_dev *pfcp) { + if (!pfcp->sk) + return; + udp_tunnel_sock_release(pfcp->sk); pfcp->sk = NULL; } From 006c8b834aed8b02bd8fb9fb05e9e7ead37d1763 Mon Sep 17 00:00:00 2001 From: Zihan Xi Date: Thu, 24 Sep 2026 05:15:21 +0000 Subject: [PATCH 330/365] net: gso: limit recursive IP-in-IP segmentation IP-in-IP GSO can re-enter inet_gso_segment() or ipv6_gso_segment() for each nested IP header. encap_level tracks header bytes, not callback depth, so a deep chain can exhaust the kernel stack. Making inet_gso_segment() stackable introduced unbounded IPv4 nesting; IPIP GSO/TSO later made the path reachable. The IPv6 stackable path was introduced separately and uses the same guard. Count IPv4 and IPv6 GSO handler entries in skb_gso_cb, initialized once per top-level GSO operation and preserved across GRE/UDP context changes. Use the existing IP_TUNNEL_RECURSION_LIMIT for both handlers. The first five entries pass, and the sixth returns -EINVAL before dispatching another GSO callback. Fixes: 3347c9602955 ("ipv4: gso: make inet_gso_segment() stackable") Cc: stable@vger.kernel.org Reported-by: Vega Closes: https://lore.kernel.org/all/cover.1790157745.git.zihanx@nebusec.ai/ Assisted-by: LLM Co-developed-by: Luxing Yin Signed-off-by: Luxing Yin Signed-off-by: Zihan Xi Reviewed-by: Willem de Bruijn Link: https://patch.msgid.link/20260924051521.32568-2-zihanx@nebusec.ai Signed-off-by: Paolo Abeni --- include/net/gso.h | 8 ++++++++ net/core/gso.c | 1 + net/ipv4/af_inet.c | 3 +++ net/ipv6/ip6_offload.c | 4 ++++ 4 files changed, 16 insertions(+) diff --git a/include/net/gso.h b/include/net/gso.h index 29975440cad5..0749230d414e 100644 --- a/include/net/gso.h +++ b/include/net/gso.h @@ -19,10 +19,18 @@ struct skb_gso_cb { int encap_level; __wsum csum; __u16 csum_start; + /* Number of IPv4/IPv6 GSO handler entries for this packet. */ + u8 recursion_counter; }; #define SKB_GSO_CB_OFFSET 32 #define SKB_GSO_CB(skb) ((struct skb_gso_cb *)((skb)->cb + SKB_GSO_CB_OFFSET)) +static inline bool gso_recursion_inc_test(struct sk_buff *skb, + unsigned int limit) +{ + return ++SKB_GSO_CB(skb)->recursion_counter > limit; +} + static inline int skb_tnl_header_len(const struct sk_buff *inner_skb) { return (skb_mac_header(inner_skb) - inner_skb->head) - diff --git a/net/core/gso.c b/net/core/gso.c index bcd156372f4d..e96ef6350064 100644 --- a/net/core/gso.c +++ b/net/core/gso.c @@ -117,6 +117,7 @@ struct sk_buff *__skb_gso_segment(struct sk_buff *skb, SKB_GSO_CB(skb)->mac_offset = skb_headroom(skb); SKB_GSO_CB(skb)->encap_level = 0; + SKB_GSO_CB(skb)->recursion_counter = 0; skb_reset_mac_header(skb); skb_reset_mac_len(skb); diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c index 32d006c1a8ee..4ce38c99fef9 100644 --- a/net/ipv4/af_inet.c +++ b/net/ipv4/af_inet.c @@ -1375,6 +1375,9 @@ struct sk_buff *inet_gso_segment(struct sk_buff *skb, int id; skb_reset_network_header(skb); + if (unlikely(gso_recursion_inc_test(skb, + IP_TUNNEL_RECURSION_LIMIT))) + goto out; nhoff = skb_network_header(skb) - skb_mac_header(skb); if (unlikely(!pskb_may_pull(skb, sizeof(*iph)))) goto out; diff --git a/net/ipv6/ip6_offload.c b/net/ipv6/ip6_offload.c index 78f50c93c536..700f1861d8eb 100644 --- a/net/ipv6/ip6_offload.c +++ b/net/ipv6/ip6_offload.c @@ -17,6 +17,7 @@ #include #include #include +#include #include "ip6_offload.h" #include "tcpv6_offload.c" @@ -104,6 +105,9 @@ static struct sk_buff *ipv6_gso_segment(struct sk_buff *skb, bool gso_partial; skb_reset_network_header(skb); + if (unlikely(gso_recursion_inc_test(skb, + IP_TUNNEL_RECURSION_LIMIT))) + goto out; nhoff = skb_network_header(skb) - skb_mac_header(skb); if (unlikely(!pskb_may_pull(skb, sizeof(*ipv6h)))) goto out; From f8932b2c0c9ae6fb577bcc05a2080c8498f91c0d Mon Sep 17 00:00:00 2001 From: Hannu Varjoranta Date: Fri, 25 Sep 2026 14:52:37 +0300 Subject: [PATCH 331/365] gve: DQO: accept TSO packets with non-protocol gso_type bits Since commit 014c607f86ab ("gve: add support for UDP GSO for DQO format"), gve_prep_tso() matches shinfo->gso_type exactly. gso_type is a bitmask, though: SKB_GSO_DODGY is set on every GSO packet that comes from tun/tap, packet sockets or other untrusted sources, and SKB_GSO_TCP_ECN and SKB_GSO_TCP_FIXEDID can be set as well. Such packets hit the default case, gve_tx_add_skb_dqo() fails and gve_try_tx_skb() drops them, counting them in tx_dropped. These packets do reach the driver: tcp_gso_segment() passes DODGY skbs through unsegmented to devices that support TSO, after recomputing gso_segs, so drivers must tolerate the bit. This breaks virtual machines behind a tap on GCE instances using the DQO queue formats. Every TSO packet forwarded from a guest (gso_type SKB_GSO_TCPV4 | SKB_GSO_DODGY) is dropped, and guest uploads slow to a crawl of retransmissions or stall entirely, while the host's own TSO traffic (gso_type SKB_GSO_TCPV4) is unaffected. On an n4 instance (DQO-QPL) with a Cloud Hypervisor guest, a 64 MB upload from the guest went from a 60 s timeout at ~0.9 MB/s to 0.19 s with this change, with tx_dropped no longer increasing. Host TCP with ECN is hit as well: gve advertises NETIF_F_TSO_ECN, so a TSO packet carrying CWR has SKB_GSO_TCP_ECN set and is dropped too. Restore the bitmask test that commit 1b9f75634441 ("gve: ignore nonrelevant GSO type bits when processing TSO headers") introduced for the same problem, keeping the UDP GSO support. While here, reload shinfo after skb_cow_head(). If the head was cloned, pskb_expand_head() moves skb_shared_info to the new head, and the pointer cached at function entry can then refer to memory that another clone frees. Fixes: 014c607f86ab ("gve: add support for UDP GSO for DQO format") Signed-off-by: Hannu Varjoranta Reviewed-by: Eric Dumazet Reviewed-by: Ankit Garg Reviewed-by: Harshitha Ramamurthy Link: https://patch.msgid.link/20260925115237.56465-1-hannu@varjosoft.com Signed-off-by: Paolo Abeni --- drivers/net/ethernet/google/gve/gve_tx_dqo.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) diff --git a/drivers/net/ethernet/google/gve/gve_tx_dqo.c b/drivers/net/ethernet/google/gve/gve_tx_dqo.c index 616c1921aebe..ad99cbb77e29 100644 --- a/drivers/net/ethernet/google/gve/gve_tx_dqo.c +++ b/drivers/net/ethernet/google/gve/gve_tx_dqo.c @@ -595,25 +595,25 @@ static int gve_prep_tso(struct sk_buff *skb) err = skb_cow_head(skb, 0); if (err < 0) return err; + shinfo = skb_shinfo(skb); l4_start = skb_transport_offset(skb); paylen = skb->len - l4_start; - switch (shinfo->gso_type) { - case SKB_GSO_TCPV4: - case SKB_GSO_TCPV6: + /* gso_type is a bitmask: SKB_GSO_DODGY, SKB_GSO_TCP_ECN and + * SKB_GSO_TCP_FIXEDID may be set alongside the protocol bit. + */ + if (shinfo->gso_type & (SKB_GSO_TCPV4 | SKB_GSO_TCPV6)) { tcp = tcp_hdr(skb); csum_replace_by_diff(&tcp->check, (__force __wsum)htonl(paylen)); header_len = skb_tcp_all_headers(skb); - break; - case SKB_GSO_UDP_L4: + } else if (shinfo->gso_type & SKB_GSO_UDP_L4) { udp = udp_hdr(skb); csum_replace_by_diff(&udp->check, (__force __wsum)htonl(paylen)); header_len = sizeof(struct udphdr) + l4_start; - break; - default: + } else { return -EINVAL; } From 0923198be4ae714b46350f742cb19ccd75bfeafa Mon Sep 17 00:00:00 2001 From: Sagar Taunk Date: Fri, 25 Sep 2026 08:55:04 +0000 Subject: [PATCH 332/365] rust: net: netlink: validate attribute length before casting to `c_int` `put()` trusted an unchecked `as` cast from `usize` to `c_int`. When the length exceeds `i32::MAX` that cast wraps around to a negative value. This ultimately resulted in a kernel panic when the reinterpreted value via `__nla_reserve()` and `skb_put()` became enormous. Validate payload and header both fit together in a `u16`, rejecting any payload that wouldn't leave room for `NLA_HDRLEN`. Fixes: 5eaa5fbb6e6c ("rust: netlink: add raw netlink abstraction") Reviewed-by: Alice Ryhl Reviewed-by: Alexandre Courbot Reviewed-by: Carlos Llamas Reviewed-by: Gary Guo Signed-off-by: Sagar Taunk Link: https://patch.msgid.link/20260925085449.22553-1-sagartaunk@proton.me Signed-off-by: Paolo Abeni --- rust/kernel/net/netlink.rs | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/rust/kernel/net/netlink.rs b/rust/kernel/net/netlink.rs index 22ef3dde36fa..2f8733a694a9 100644 --- a/rust/kernel/net/netlink.rs +++ b/rust/kernel/net/netlink.rs @@ -11,6 +11,7 @@ use kernel::{ alloc::{self, AllocError}, error::to_result, + num::casts::u16_as_usize, prelude::*, transmute::AsBytes, types::Opaque, @@ -86,9 +87,17 @@ fn put(&mut self, attrtype: c_int, value: &T) -> Result where T: ?Sized + AsBytes, { + // `nla_len` is a 16-bit field that encodes the total attribute length + // (header + payload). Subtracting the header size from `u16::MAX` gives + // the largest payload that still fits within that field. + const MAX_PAYLOAD_LEN: usize = u16_as_usize(u16::MAX) - size_of::(); + let skb = self.skb.skb.as_ptr(); let len = size_of_val(value); let ptr = core::ptr::from_ref(value).cast::(); + if len > MAX_PAYLOAD_LEN { + return Err(EMSGSIZE); + } // SAFETY: `skb` is valid by `NetlinkSkBuff` type invariants, and the provided value is // readable and initialized for its `size_of` bytes. to_result(unsafe { bindings::nla_put(skb, attrtype, len as c_int, ptr) }) From d031465acc362866f27e4f8f79cfecf1037fe2e5 Mon Sep 17 00:00:00 2001 From: Jamal Hadi Salim Date: Sat, 26 Sep 2026 14:03:27 -0400 Subject: [PATCH 333/365] net/sched: fq_codel: match the no-drop threshold to the packet size commit d9ebd8f9aa8b ("net/sched: fq_codel: clamp default quantum and mtu") clamped both q->quantum and q->cparams.mtu to [256, FQ_CODEL_QUANTUM_MAX]. The two fields mean different things: quantum is a DRR credit that wants the 256 floor, but cparams.mtu is the CoDel no-drop threshold (codel_impl.h "*backlog <= params->mtu"). On a link whose maximum transmitted packet size is below 256, the floor extends CoDel's minimum-backlog exemption beyond one packet and delays drop or mark eligibility by several small packets. Split the clamp. quantum keeps [256, FQ_CODEL_QUANTUM_MAX]; cparams.mtu tracks psched_mtu() (the device MTU plus its hard-header length) with only the upper bound that guards the original overflow (psched_mtu() wrapping to ~2 GiB on a huge-MTU device). Conditions to recreate the bug: attach an fq_codel qdisc on a link whose MTU plus hard_header_len is below 256 (e.g. a CAN interface). At that MTU the no-drop threshold must equal the device MTU plus its hard-header length; before this patch it was forced to 256. Basic Testing done: with dev->mtu=100 and hard_header_len=14, a return probe on fq_codel_init() observed cparams.mtu change from 256 to 114 Fixes: d9ebd8f9aa8b ("net/sched: fq_codel: clamp default quantum and mtu") Reported-by: Sashiko (nipa) Link: https://netdev-ai.bots.linux.dev/sashiko/#/patchset/20260819143136.57350-1-jhs@mojatatu.com Signed-off-by: Jamal Hadi Salim Link: https://patch.msgid.link/QDISC-34MS.v1.20260925165535@mojatatu.com Signed-off-by: Paolo Abeni --- net/sched/sch_fq_codel.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/net/sched/sch_fq_codel.c b/net/sched/sch_fq_codel.c index 969b2510b0b8..e6c87a32950f 100644 --- a/net/sched/sch_fq_codel.c +++ b/net/sched/sch_fq_codel.c @@ -517,8 +517,8 @@ static int fq_codel_init(struct Qdisc *sch, struct nlattr *opt, q->flows_cnt = 1024; q->memory_limit = 32 << 20; /* 32 MBytes */ q->drop_batch_size = 64; - mtu = clamp_t(u32, psched_mtu(qdisc_dev(sch)), 256, FQ_CODEL_QUANTUM_MAX); - q->quantum = mtu; + mtu = min_t(u32, psched_mtu(qdisc_dev(sch)), FQ_CODEL_QUANTUM_MAX); + q->quantum = clamp_t(u32, mtu, 256, FQ_CODEL_QUANTUM_MAX); INIT_LIST_HEAD(&q->new_flows); INIT_LIST_HEAD(&q->old_flows); codel_params_init(&q->cparams); From 54518e0e827f4ca9229ae657022c60bf60f5c1bf Mon Sep 17 00:00:00 2001 From: Jamal Hadi Salim Date: Sat, 26 Sep 2026 14:03:28 -0400 Subject: [PATCH 334/365] net/sched: sch_codel: match the no-drop threshold to the packet size commit 6439461f1618 ("net/sched: sch_codel: clamp default mtu to avoid disabling CoDel") clamped q->params.mtu to [256, 1 << 20]. The value is the CoDel no-drop threshold (codel_impl.h "*backlog <= params->mtu"), so on a link whose maximum transmitted packet size is below 256 the floor extends CoDel's minimum-backlog exemption beyond one packet and delays drop or mark eligibility by several small packets. Keep the upper bound that guards the original overflow (psched_mtu() wrapping to ~2 GiB on a huge-MTU device) but drop the 256 floor, so the threshold tracks the real device packet size. Conditions to recreate the bug: attach a codel qdisc on a link whose MTU plus hard_header_len is below 256 (e.g. a CAN interface). At that MTU the no-drop threshold must equal the device MTU plus its hard-header length; before this patch it was forced to 256. Requires CAP_NET_ADMIN in a user namespace. Fixes: 6439461f1618 ("net/sched: sch_codel: clamp default mtu to avoid disabling CoDel") Reported-by: Sashiko (nipa) Link: https://netdev-ai.bots.linux.dev/sashiko/#/patchset/20260819143136.57350-1-jhs@mojatatu.com Signed-off-by: Jamal Hadi Salim Link: https://patch.msgid.link/QDISC-34MS.v1.20260925165535@mojatatu.com.2 Signed-off-by: Paolo Abeni --- net/sched/sch_codel.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/net/sched/sch_codel.c b/net/sched/sch_codel.c index 6aa5829d6961..a1ba2e355f2d 100644 --- a/net/sched/sch_codel.c +++ b/net/sched/sch_codel.c @@ -205,7 +205,7 @@ static int codel_init(struct Qdisc *sch, struct nlattr *opt, codel_params_init(&q->params); codel_vars_init(&q->vars); codel_stats_init(&q->stats); - q->params.mtu = clamp_t(u32, psched_mtu(qdisc_dev(sch)), 256, 1 << 20); + q->params.mtu = min_t(u32, psched_mtu(qdisc_dev(sch)), 1 << 20); if (opt) { int err = codel_change(sch, opt, extack); From 2cf81fc9505b148b1b70068303ea6ea4b1212403 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ilkka=20Lappetel=C3=A4inen?= Date: Fri, 25 Sep 2026 15:48:29 +0300 Subject: [PATCH 335/365] net/packet: prevent TX_RING byte count overflow MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit tpacket_snd() drains every SEND_REQUEST frame in a TX ring and adds each valid packet length to the signed int len_sum. Userspace can recycle completed frames concurrently, so a single blocking send call is not bounded by the ring size. After enough successful transmissions len_sum wraps into the negative range. In particular, 65536 65535-byte frames followed by a 65007-byte frame produce 0xfffffdef, or -EIOCBQUEUED. sock_sendmsg_nosec() treats that as an impossible return and triggers a BUG. On systems configured to panic on oops, a CAP_NET_RAW holder can panic the host kernel. Stop before the next frame would make the byte count exceed INT_MAX. The positive short count leaves that frame pending for a later send call. A two-thread TPACKET_V2 reproducer triggered the BUG on an arm64 build of Linux 7.3-rc4. With this change, the same reproducer returned the expected positive short count without an oops. Fixes: 69e3c75f4d54 ("net: TX_RING and packet mmap") Cc: stable@vger.kernel.org Signed-off-by: Ilkka Lappeteläinen Reviewed-by: Willem de Bruijn Link: https://patch.msgid.link/20260925124829.6595-1-ilkka.lappetelainen@gmail.com Signed-off-by: Jakub Kicinski --- net/packet/af_packet.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c index 7c83e01526ed..8865c1a42667 100644 --- a/net/packet/af_packet.c +++ b/net/packet/af_packet.c @@ -2874,6 +2874,9 @@ static int tpacket_snd(struct packet_sock *po, struct msghdr *msg) vnet_hdr.hdr_len); has_vnet_hdr = true; } + /* Prevent the byte count from wrapping into a negative errno */ + if (unlikely(tp_len > INT_MAX - len_sum)) + break; copylen = max_t(int, copylen, hard_header_len); skb = sock_alloc_send_skb(&po->sk, hlen + tlen + sizeof(struct sockaddr_ll) + From febec6a7076499d16b37f556724a942633a270a6 Mon Sep 17 00:00:00 2001 From: Alireza Asgari Date: Thu, 24 Sep 2026 13:00:24 +0000 Subject: [PATCH 336/365] net: fix checksum offsets in skb_splice_from_iter() skb_splice_from_iter() appends multiple page fragments before updating skb->len. However, skb_splice_csum_page() uses that unchanged length as the offset for every fragment checksum. If bytes already appended during the call have an odd total length, a later fragment's checksum is combined with the wrong parity. For example, prime a UDP socket with sendto(MSG_MORE) and splice two distinct pipe buffers containing "abc" and "DEFGH". Uncorking reports success, but the receiver discards the packet for a bad checksum. This also affects IPv6 and fragments beginning near a page boundary. Pass the initial skb length plus the bytes already spliced to the checksum helper. Keep the existing final length update and partial-progress error handling unchanged. CHECKSUM_PARTIAL does not use this helper and remains unaffected. Fixes: 2e910b95329c ("net: Add a function to splice pages into an skbuff for MSG_SPLICE_PAGES") Cc: stable@vger.kernel.org Signed-off-by: Alireza Asgari Link: https://patch.msgid.link/20260924-fix-udp-splice-checksum-v1-1-fe61d65447a8@asgari.net Signed-off-by: Jakub Kicinski --- net/core/skbuff.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) diff --git a/net/core/skbuff.c b/net/core/skbuff.c index c3042d822afa..4aea06d5167d 100644 --- a/net/core/skbuff.c +++ b/net/core/skbuff.c @@ -7409,7 +7409,8 @@ nodefer: kfree_skb_napi_cache(skb); } static void skb_splice_csum_page(struct sk_buff *skb, struct page *page, - size_t offset, size_t len) + size_t offset, size_t len, + unsigned int csum_offset) { const char *kaddr; __wsum csum; @@ -7417,7 +7418,7 @@ static void skb_splice_csum_page(struct sk_buff *skb, struct page *page, kaddr = kmap_local_page(page); csum = csum_partial(kaddr + offset, len, 0); kunmap_local(kaddr); - skb->csum = csum_block_add(skb->csum, csum, skb->len); + skb->csum = csum_block_add(skb->csum, csum, csum_offset); } /** @@ -7477,7 +7478,8 @@ ssize_t skb_splice_from_iter(struct sk_buff *skb, struct iov_iter *iter, } if (skb->ip_summed == CHECKSUM_NONE) - skb_splice_csum_page(skb, page, off, part); + skb_splice_csum_page(skb, page, off, part, + skb->len + spliced); off = 0; spliced += part; From 1221246d3bd08d8b5228777e1a9b3361ca946e64 Mon Sep 17 00:00:00 2001 From: Alireza Asgari Date: Thu, 24 Sep 2026 13:00:25 +0000 Subject: [PATCH 337/365] selftests: net: cover UDP splice checksum fragment alignment Exercise software UDP checksums when splice() transfers multiple pipe fragments into a corked datagram. Use separate source pipes so short writes cannot merge, and use odd fragment lengths to expose checksum-position errors between fragments. Cover IPv4, IPv6 and IPv4-mapped destinations, connected and unconnected sockets, resident prefixes, even and odd fragment lengths, and a fragment starting near a page boundary. Ordinary sends provide a control. Check the complete payload and reject extra datagrams after a successful send. Keep receive waits bounded and skip unavailable address families or pipe capacities. The test uses only loopback sockets with ephemeral ports and has no dependency on the application that exposed the regression. Signed-off-by: Alireza Asgari Link: https://patch.msgid.link/20260924-fix-udp-splice-checksum-v1-2-fe61d65447a8@asgari.net Signed-off-by: Jakub Kicinski --- tools/testing/selftests/net/.gitignore | 1 + tools/testing/selftests/net/Makefile | 1 + .../selftests/net/udp_splice_checksum.c | 378 ++++++++++++++++++ 3 files changed, 380 insertions(+) create mode 100644 tools/testing/selftests/net/udp_splice_checksum.c diff --git a/tools/testing/selftests/net/.gitignore b/tools/testing/selftests/net/.gitignore index c9f46031ac73..dacd36ed8455 100644 --- a/tools/testing/selftests/net/.gitignore +++ b/tools/testing/selftests/net/.gitignore @@ -55,6 +55,7 @@ tools tun txring_overwrite txtimestamp +udp_splice_checksum udpgso udpgso_bench_rx udpgso_bench_tx diff --git a/tools/testing/selftests/net/Makefile b/tools/testing/selftests/net/Makefile index 3ee3378f8b26..590b33e16d9f 100644 --- a/tools/testing/selftests/net/Makefile +++ b/tools/testing/selftests/net/Makefile @@ -199,6 +199,7 @@ TEST_GEN_PROGS := \ tap \ tcp_port_share \ tls \ + udp_splice_checksum \ # end of TEST_GEN_PROGS TEST_FILES := \ diff --git a/tools/testing/selftests/net/udp_splice_checksum.c b/tools/testing/selftests/net/udp_splice_checksum.c new file mode 100644 index 000000000000..dc61444e6481 --- /dev/null +++ b/tools/testing/selftests/net/udp_splice_checksum.c @@ -0,0 +1,378 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* + * Exercise UDP software checksums over multiple spliced pipe fragments. + * All traffic stays on loopback; no privileges or external services are needed. + * Exit status: 0 = pass, 1 = failure, 4 = all cases skipped. + */ +#define _GNU_SOURCE + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "kselftest.h" + +#define TIMEOUT_MS 200 + +struct test_case { + const char *name; + const char *body; + const char *prefix; + size_t fragments[3]; + bool ordinary; + bool page_tail; +}; + +static const struct test_case cases[] = { + { "ordinary-control", "abcDEFGH", "", { 3, 5 }, true, false }, + { "uniform-odd-fragments", "BBBBBBBB", "", { 3, 5 }, false, false }, + { "even-fragments", "abcdEFGH", "", { 4, 4 }, false, false }, + { "odd-fragments", "abcDEFGH", "", { 3, 5 }, false, false }, + { "odd-prefix-odd-fragments", "abcDEFGH", "P", { 3, 5 }, false, false }, + { "even-prefix-odd-fragments", "abcDEFGH", "PQ", { 3, 5 }, false, false }, + { "page-tail-odd-fragments", "abcDEFGH", "", { 3, 5 }, false, true }, + { "three-fragments", "abcDEFGhijkl", "", { 3, 4, 5 }, false, false }, + { "odd-prefix-three-fragments", + "abcDEFGhijkl", + "P", + { 3, 4, 5 }, + false, + false }, +}; + +static const char *const families[] = { "IPv4", "IPv6", "IPv4-mapped" }; + +static int error(const char *what) +{ + ksft_perror(what); + return 1; +} + +static bool unavailable(void) +{ + return errno == EAFNOSUPPORT || errno == EPROTONOSUPPORT || + errno == EADDRNOTAVAIL; +} + +static long long now_ms(void) +{ + struct timespec ts; + + if (clock_gettime(CLOCK_MONOTONIC, &ts)) + ksft_exit_fail_perror("clock_gettime"); + return (long long)ts.tv_sec * 1000 + ts.tv_nsec / 1000000; +} + +static int receive_one(int fd, const char *expected, size_t length) +{ + struct pollfd pfd = { .fd = fd, .events = POLLIN }; + long long deadline = now_ms() + TIMEOUT_MS; + int remaining, ret; + char received[64]; + ssize_t count; + + for (;;) { + remaining = deadline - now_ms(); + if (remaining < 0) + remaining = 0; + ret = poll(&pfd, 1, remaining); + if (ret < 0) { + if (errno == EINTR) + continue; + return error("poll"); + } + if (!ret) { + ksft_print_msg("no datagram within %d ms after send\n", + TIMEOUT_MS); + return 1; + } + count = recv(fd, received, sizeof(received), + MSG_DONTWAIT | MSG_TRUNC); + if (count < 0) { + if (errno == EINTR || errno == EAGAIN) + continue; + return error("recv"); + } + break; + } + if (count != (ssize_t)length || memcmp(received, expected, length)) { + ksft_print_msg("received %zd bytes; expected %zu matching\n", + count, length); + return 1; + } + count = recv(fd, received, sizeof(received), MSG_DONTWAIT | MSG_TRUNC); + if (count >= 0) { + ksft_print_msg("unexpected extra datagram (%zd bytes)\n", + count); + return 1; + } + if (errno != EAGAIN) + return error("extra-datagram check"); + return 0; +} + +/* Separate source pipes prevent adjacent small writes merging into one buffer. */ +static int append_fragment(int dest, const char *data, size_t length) +{ + int source[2], ret = 1; + + if (pipe2(source, O_NONBLOCK | O_CLOEXEC)) + return error("source pipe"); + if (write(source[1], data, length) != (ssize_t)length) { + error("source write"); + goto out; + } + if (splice(source[0], NULL, dest, NULL, length, SPLICE_F_NONBLOCK) != + (ssize_t)length) { + error("pipe-to-pipe splice"); + goto out; + } + ret = 0; +out: + close(source[0]); + close(source[1]); + return ret; +} + +static int pipe_capacity(int fd, unsigned int slots, const char **skip_reason) +{ + long required = slots * sysconf(_SC_PAGESIZE); + int capacity = fcntl(fd, F_GETPIPE_SZ); + + if (capacity < 0) + return error("query pipe capacity"); + if (capacity >= required || + fcntl(fd, F_SETPIPE_SZ, required) >= required) + return 0; + ksft_print_msg("cannot reserve %u pipe slots: %s\n", slots, + strerror(errno)); + *skip_reason = "required pipe capacity unavailable"; + return KSFT_SKIP; +} + +/* Leave three bytes at a page's end, followed by five bytes on the next page. */ +static int append_page_tail(int dest, const char *body, + const char **skip_reason) +{ + long page = sysconf(_SC_PAGESIZE); + int source[2], ret = 1; + char *data; + + if (page < 8) { + ksft_print_msg("invalid page size %ld\n", page); + return 1; + } + data = malloc(page + 5); + if (!data) + return error("malloc"); + memset(data, 'x', page - 3); + memcpy(data + page - 3, body, 8); + if (pipe2(source, O_NONBLOCK | O_CLOEXEC)) { + error("page-tail source pipe"); + goto free_data; + } + ret = pipe_capacity(source[1], 2, skip_reason); + if (ret) + goto out; + ret = 1; + if (write(source[1], data, page + 5) != page + 5) { + error("page-tail write"); + goto out; + } + if (read(source[0], data, page - 3) != page - 3) { + error("page-tail discard"); + goto out; + } + if (splice(source[0], NULL, dest, NULL, 8, SPLICE_F_NONBLOCK) != 8) { + error("page-tail pipe-to-pipe splice"); + goto out; + } + ret = 0; +out: + close(source[0]); + close(source[1]); +free_data: + free(data); + return ret; +} + +static int run_case(unsigned int family, bool connected, + const struct test_case *test, const char **skip_reason) +{ + size_t body_length = strlen(test->body), offset = 0; + int tx_family = family == 0 ? AF_INET : AF_INET6; + int rx_family = family == 1 ? AF_INET6 : AF_INET; + int rx = -1, tx = -1, pipefd[2] = { -1, -1 }; + size_t prefix_length = strlen(test->prefix); + struct sockaddr_storage address = { 0 }; + socklen_t address_length; + struct sockaddr_in6 *v6; + struct sockaddr_in *v4; + int ret = 1, zero = 0; + char expected[64]; + unsigned int i; + ssize_t count; + + v4 = (struct sockaddr_in *)&address; + v6 = (struct sockaddr_in6 *)&address; + *skip_reason = "address family unavailable"; + memcpy(expected, test->prefix, prefix_length); + memcpy(expected + prefix_length, test->body, body_length); + rx = socket(rx_family, SOCK_DGRAM | SOCK_CLOEXEC, 0); + if (rx < 0) { + ret = unavailable() ? KSFT_SKIP : error("receive socket"); + goto out; + } + if (rx_family == AF_INET) { + v4->sin_family = AF_INET; + v4->sin_addr.s_addr = htonl(INADDR_LOOPBACK); + address_length = sizeof(*v4); + } else { + v6->sin6_family = AF_INET6; + v6->sin6_addr = in6addr_loopback; + address_length = sizeof(*v6); + } + if (bind(rx, (struct sockaddr *)&address, address_length)) { + ret = unavailable() ? KSFT_SKIP : error("bind loopback"); + goto out; + } + if (getsockname(rx, (struct sockaddr *)&address, &address_length)) { + error("getsockname"); + goto out; + } + if (family == 2) { + unsigned short port = v4->sin_port; + + memset(&address, 0, sizeof(address)); + v6->sin6_family = AF_INET6; + v6->sin6_port = port; + inet_pton(AF_INET6, "::ffff:127.0.0.1", &v6->sin6_addr); + address_length = sizeof(*v6); + } + tx = socket(tx_family, SOCK_DGRAM | SOCK_CLOEXEC, 0); + if (tx < 0) { + ret = unavailable() ? KSFT_SKIP : error("send socket"); + goto out; + } + if (family == 2 && + setsockopt(tx, IPPROTO_IPV6, IPV6_V6ONLY, &zero, sizeof(zero))) { + error("disable IPV6_V6ONLY"); + goto out; + } + if (connected && + connect(tx, (struct sockaddr *)&address, address_length)) { + error("connect"); + goto out; + } + if (test->ordinary) { + count = sendto(tx, expected, prefix_length + body_length, 0, + (struct sockaddr *)&address, address_length); + if (count != (ssize_t)(prefix_length + body_length)) { + error("ordinary sendto"); + goto out; + } + } else { + if (pipe2(pipefd, O_NONBLOCK | O_CLOEXEC)) { + error("assembled pipe"); + goto out; + } + ret = pipe_capacity(pipefd[1], test->fragments[2] ? 3 : 2, + skip_reason); + if (ret) + goto out; + ret = 1; + if (test->page_tail) { + ret = append_page_tail(pipefd[1], test->body, + skip_reason); + if (ret) + goto out; + ret = 1; + } else { + for (i = 0; i < ARRAY_SIZE(test->fragments); i++) { + if (!test->fragments[i]) + break; + if (append_fragment(pipefd[1], + test->body + offset, + test->fragments[i])) + goto out; + offset += test->fragments[i]; + } + } + /* Even an empty prefix establishes the destination and UDP cork. */ + count = sendto(tx, test->prefix, prefix_length, MSG_MORE, + (struct sockaddr *)&address, address_length); + if (count != (ssize_t)prefix_length) { + error("prime sendto(MSG_MORE)"); + goto out; + } + count = splice(pipefd[0], NULL, tx, NULL, body_length, + SPLICE_F_NONBLOCK | SPLICE_F_MORE); + if (count != (ssize_t)body_length) { + ksft_print_msg("socket splice: %zd, expected %zu: %s\n", + count, body_length, + count < 0 ? strerror(errno) : + "short transfer"); + goto out; + } + if (sendto(tx, "", 0, 0, (struct sockaddr *)&address, + address_length)) { + error("commit sendto"); + goto out; + } + } + ret = receive_one(rx, expected, prefix_length + body_length); +out: + if (pipefd[0] >= 0) { + close(pipefd[0]); + close(pipefd[1]); + } + if (tx >= 0) + close(tx); + if (rx >= 0) + close(rx); + return ret; +} + +int main(void) +{ + unsigned int family, connected, i; + const char *connection, *skip; + int ret; + + ksft_print_header(); + ksft_set_plan(ARRAY_SIZE(families) * 2 * ARRAY_SIZE(cases)); + for (family = 0; family < ARRAY_SIZE(families); family++) { + for (connected = 0; connected < 2; connected++) { + connection = connected ? "connected" : "unconnected"; + for (i = 0; i < ARRAY_SIZE(cases); i++) { + ret = run_case(family, connected, &cases[i], + &skip); + if (ret == KSFT_SKIP) + ksft_test_result_skip("%s %s %s: %s\n", + families[family], + connection, + cases[i].name, + skip); + else + ksft_test_result(!ret, "%s %s %s\n", + families[family], + connection, + cases[i].name); + } + } + } + if (!ksft_get_pass_cnt() && !ksft_get_fail_cnt()) { + /* ksft_exit_skip() would add a result beyond the plan. */ + ksft_print_cnts(); + return KSFT_SKIP; + } + ksft_finished(); +} From 8b9993437429c66d11430ae61ffa5937a6abb4ae Mon Sep 17 00:00:00 2001 From: Yoshihisa Yamamoto Date: Thu, 24 Sep 2026 10:36:57 +0000 Subject: [PATCH 338/365] net: pcs: rzn1-miic: Clear MIIC protection state before unprotect sequence On RZ/T2H and RZ/N1D MIIC, writing 0x0000 to the MIIC protection register clears the protection state machine. The initial protection state cannot be assumed to be constant, as it may be influenced by previous activity before the driver takes ownership of the hardware. Clear the protection state before issuing the unprotect sequence so that register access always starts from a known state, regardless of any previous activity. Fixes: 7dc54d3b8d91 ("net: pcs: add Renesas MII converter driver") Signed-off-by: Yoshihisa Yamamoto Reviewed-by: Lad Prabhakar Link: https://patch.msgid.link/TYCPR01MB74815955A10817B64BD6FB93A4812@TYCPR01MB7481.jpnprd01.prod.outlook.com Signed-off-by: Jakub Kicinski --- drivers/net/pcs/pcs-rzn1-miic.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/drivers/net/pcs/pcs-rzn1-miic.c b/drivers/net/pcs/pcs-rzn1-miic.c index cb74861e823c..3aa513467060 100644 --- a/drivers/net/pcs/pcs-rzn1-miic.c +++ b/drivers/net/pcs/pcs-rzn1-miic.c @@ -270,6 +270,9 @@ static struct miic_port *phylink_pcs_to_miic_port(struct phylink_pcs *pcs) static void miic_unlock_regs(struct miic *miic) { + /* Clear protection state */ + writel(0x0000, miic->base + MIIC_PRCMD); + /* Unprotect register writes */ writel(0x00A5, miic->base + MIIC_PRCMD); writel(0x0001, miic->base + MIIC_PRCMD); From 19419faf58dfe5bcc8a9e60e622976c5f8e3ddc1 Mon Sep 17 00:00:00 2001 From: Nicolai Buchwitz Date: Mon, 28 Sep 2026 12:56:41 +0200 Subject: [PATCH 339/365] net: bcmgenet: fix NULL dereference in set_coalesce before first open The Rx ring priv and index are only set in bcmgenet_init_rx_ring(), which runs at open. Setting the Rx coalescing parameters on an interface that was never opened, e.g. "ethtool -C eth0 rx-usecs 50", dereferences a NULL ring->priv. The oops happens with RTNL held, so networking stays stuck until a reboot: Unable to handle kernel paging request at virtual address 0000000000002788 pc : bcmgenet_set_rx_coalesce.isra.0+0x10/0x88 lr : bcmgenet_set_coalesce+0xe8/0x180 Call trace: bcmgenet_set_rx_coalesce.isra.0+0x10/0x88 (P) __ethnl_set_coalesce.isra.0+0x490/0x570 ethnl_set_coalesce+0x48/0xc0 ethnl_default_set_doit+0xf4/0x230 genl_family_rcv_msg_doit+0xe8/0x160 genl_rcv_msg+0x220/0x2a0 netlink_rcv_skb+0x68/0x140 genl_rcv+0x40/0x60 netlink_unicast+0x338/0x3c0 netlink_sendmsg+0x19c/0x3f8 __sock_sendmsg+0x64/0xc0 __sys_sendto+0x124/0x198 __arm64_sys_sendto+0x30/0x48 Set priv and index at probe. Writing the registers while the interface is down is harmless because open resets the MAC and bcmgenet_init_rx_coalesce() reapplies the stored values. It was observed on a Raspberry Pi CM4, but should apply to all other genet devices. Fixes: 9f4ca05827a2 ("net: bcmgenet: Add support for adaptive RX coalescing") Signed-off-by: Nicolai Buchwitz Reviewed-by: Florian Fainelli Link: https://patch.msgid.link/20260928105642.856210-1-nb@tipi-net.de Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/broadcom/genet/bcmgenet.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/drivers/net/ethernet/broadcom/genet/bcmgenet.c b/drivers/net/ethernet/broadcom/genet/bcmgenet.c index f725d26e6020..e8908916558b 100644 --- a/drivers/net/ethernet/broadcom/genet/bcmgenet.c +++ b/drivers/net/ethernet/broadcom/genet/bcmgenet.c @@ -4151,8 +4151,11 @@ static int bcmgenet_probe(struct platform_device *pdev) netif_set_real_num_rx_queues(priv->dev, priv->hw_params->rx_queues + 1); /* Set default coalescing parameters */ - for (i = 0; i <= priv->hw_params->rx_queues; i++) + for (i = 0; i <= priv->hw_params->rx_queues; i++) { + priv->rx_rings[i].priv = priv; + priv->rx_rings[i].index = i; priv->rx_rings[i].rx_max_coalesced_frames = 1; + } /* Initialize u64 stats seq counter for 32bit machines */ for (i = 0; i <= GENET_MAX_MQ_CNT; i++) { From ea4d4b5dddb5121e64f56fd8e0720bd8b447b63d Mon Sep 17 00:00:00 2001 From: Jamal Hadi Salim Date: Mon, 28 Sep 2026 08:46:16 -0400 Subject: [PATCH 340/365] net: cap skb->queue_mapping when the tx queue is picked skbedit can set skb->queue_mapping and raise the per-CPU skip_txqueue flag so __dev_queue_xmit() honours the mapping. __dev_queue_xmit() cleared the flag before sch_handle_egress() and only read it afterwards, so the flag was not confined to the xmit that set it: a nested xmit (mirred redirect or mirror, or a drop after skbedit) could set the flag and the outer xmit would consume it for an skb that never went through skbedit. A forwarded packet still carries the ingress NIC's rx_queue + 1 in skb->queue_mapping, so the outer device then indexes its tx queue state with that stale value. Taprio's child array q->qdiscs[] is sized to the device's queue count, so taprio_enqueue() indexes past its allocation and dereferences the result as a struct Qdisc *. We (ab)use the skb->nf_skip_egress which means "skip netfilter egress for this packet" to tag to "am I in tc egress?". Despite the overload I dont see it as a conflict since the marker is set only around the single sch_handle_egress() call and ingress path is guarded by tc_at_ingress. I will send a followup(net-next) patch once this hits net-next to rename the skb->nf_skip_egress bit/flag to skb->skip_egress Arm the flag only from the egress classifier that can use it: raise skip_txqueue from tcf_skbedit_act() only when it runs inside sch_handle_egress(), thanks to skb->nf_skip_egress. An egress qdisc classifier runs in q->enqueue(), after the tx queue has been picked, so a mapping it sets cannot affect the current packet; arming the flag there only pollutes it for a later xmit. Then own the flag for the xmit frame the egress hook runs in: save the incoming value and clear it just before sch_handle_egress(), and restore it after the hook - on the consumed (drop) path, or, in the same call that reads it, on the surviving path. The save and the restores stay inside the egress_needed_key static branch, so a packet pays for them only when egress hooks are active (2f1e85b1aee4). Store the value netdev_cap_txqueue() selected back into skb->queue_mapping in netdev_tx_queue_mapping(), as netdev_core_pick_tx() already does, so the skip_txqueue path never hands a later reader on the xmit path a mapping the device cannot serve. A store made still later in the same frame, by a tc BPF program attached to a transmit qdisc, is outside this path and is not re-capped; a separate followup will resolve that path. netdev_xmit_skip_txqueue() returns the previous flag value so the save-and-clear is one call, and a no-op stub is provided when CONFIG_NET_EGRESS is disabled. skb->nf_skip_egress is compiled under CONFIG_NET_EGRESS rather than CONFIG_NETFILTER_SKIP_EGRESS, so skb_at_tc_egress() is valid whenever the egress path is built. A local user in a network namespace can redirect a packet from a device with more TX queues to one with fewer after setting a mapping valid only on the larger device. That reaches these reads and, under KASAN, faults with "slab-out-of-bounds in taprio_enqueue". Conditions to recreate the bug: the report's own trigger is a local user with CAP_NET_ADMIN in a network namespace, so no eBPF program is needed. With CONFIG_NET_SCH_TAPRIO=y, CONFIG_NET_ACT_SKBEDIT=y, CONFIG_NET_ACT_MIRRED=y, CONFIG_NET_CLS_MATCHALL=y, CONFIG_NET_SCH_PRIO=y and KASAN enabled, create qa (3 queues), qb (2 queues) and qc (1 queue) as dummy devices; put a taprio root on qb (num_tc 1, queues 2@0) and clsact on all three; then add an egress matchall filter on every device. On qa: "action skbedit queue_mapping 2 pipe action mirred egress redirect dev qb". On qb: "action mirred egress mirror dev qc". On qc: "action skbedit queue_mapping 0 pipe". Send one packet out qa. qc's skbedit sets the flag while qb's outer xmit is in flight; without the fix qb consumes it and reads its two-entry taprio child array with the forwarded packet's stale mapping. A qc whose skbedit is instead installed in a transmit-qdisc classifier (a matchall filter on the qc root qdisc) reaches the same code path the same way without the fix. Testing: on a KASAN build with panic_on_warn=1 the unfixed kernel panics with "BUG: KASAN: slab-out-of-bounds in taprio_enqueue", a read 0 bytes past a 16-byte taprio_init() allocation, for the clsact-setter and the transmit-qdisc-classifier reproducers and for a clsact skbedit-then-tc-BPF store; the fixed kernel runs all three with no report, and the BPF store variant additionally shows the expected "selects TX queue" clamp notice from the write-back. Fixes: 2f1e85b1aee4 ("net: sched: use queue_mapping to pick tx queue") Reported-by: Zero Day Initiative Link: https://lore.kernel.org/netdev/CANn89iLwYx8nCVf0pCEk_MmEiyC6kQaMwCQT9WkQVeeNzNQHqQ@mail.gmail.com/ Link: https://lore.kernel.org/netdev/179008581937.2160803.7117814290574262942@kernel.org/ Link: https://lore.kernel.org/netdev/179033713973.2160803.4914570693994398206@kernel.org/ Link: https://lore.kernel.org/netdev/20260925180407.63647514@kernel.org/ Link: https://lore.kernel.org/netdev/CANn89i+k-mZKDQVtvws_MEXeuMTAdaCcOXFZE-RfhcGTu90sjA@mail.gmail.com/ Suggested-by: Eric Dumazet Suggested-by: Jakub Kicinski Tested-by: hybris Signed-off-by: Jamal Hadi Salim Reviewed-by: Eric Dumazet Link: https://patch.msgid.link/QDISC-9R8V.v4.20260928081529@mojatatu.com Signed-off-by: Jakub Kicinski --- include/linux/netfilter_netdev.h | 2 +- include/linux/rtnetlink.h | 7 ++++++- include/linux/skbuff.h | 2 +- include/net/sch_generic.h | 9 +++++++++ net/core/dev.c | 31 ++++++++++++++++++++++++------- net/sched/act_skbedit.c | 5 ++--- 6 files changed, 43 insertions(+), 13 deletions(-) diff --git a/include/linux/netfilter_netdev.h b/include/linux/netfilter_netdev.h index 3175073a66ba..2a854a0bbd4b 100644 --- a/include/linux/netfilter_netdev.h +++ b/include/linux/netfilter_netdev.h @@ -133,7 +133,7 @@ static inline struct sk_buff *nf_hook_egress(struct sk_buff *skb, int *rc, static inline void nf_skip_egress(struct sk_buff *skb, bool skip) { -#ifdef CONFIG_NETFILTER_SKIP_EGRESS +#ifdef CONFIG_NET_EGRESS skb->nf_skip_egress = skip; #endif } diff --git a/include/linux/rtnetlink.h b/include/linux/rtnetlink.h index 95729339e7a5..a54ec40d095c 100644 --- a/include/linux/rtnetlink.h +++ b/include/linux/rtnetlink.h @@ -186,7 +186,12 @@ void net_dec_ingress_queue(void); #ifdef CONFIG_NET_EGRESS void net_inc_egress_queue(void); void net_dec_egress_queue(void); -void netdev_xmit_skip_txqueue(bool skip); +bool netdev_xmit_skip_txqueue(bool skip); +#else +static inline bool netdev_xmit_skip_txqueue(bool skip) +{ + return false; +} #endif void rtnetlink_init(void); diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h index 84308498a3a8..a3ff380d43f0 100644 --- a/include/linux/skbuff.h +++ b/include/linux/skbuff.h @@ -1020,7 +1020,7 @@ struct sk_buff { #ifdef CONFIG_NET_REDIRECT __u8 from_ingress:1; #endif -#ifdef CONFIG_NETFILTER_SKIP_EGRESS +#ifdef CONFIG_NET_EGRESS __u8 nf_skip_egress:1; #endif #ifdef CONFIG_SKB_DECRYPTED diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h index f35bd06a6bad..1acaadb3e2cf 100644 --- a/include/net/sch_generic.h +++ b/include/net/sch_generic.h @@ -810,6 +810,15 @@ static inline bool skb_at_tc_ingress(const struct sk_buff *skb) #endif } +static inline bool skb_at_tc_egress(const struct sk_buff *skb) +{ +#ifdef CONFIG_NET_EGRESS + return skb->nf_skip_egress && !skb_at_tc_ingress(skb); +#else + return false; +#endif +} + static inline bool skb_skip_tc_classify(struct sk_buff *skb) { #ifdef CONFIG_NET_CLS_ACT diff --git a/net/core/dev.c b/net/core/dev.c index f660fccfc0db..8638c994bbcb 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -4404,9 +4404,14 @@ EXPORT_SYMBOL(dev_loopback_xmit); static struct netdev_queue * netdev_tx_queue_mapping(struct net_device *dev, struct sk_buff *skb) { - int qm = skb_get_queue_mapping(skb); + int queue = skb_get_queue_mapping(skb); + int capped; - return netdev_get_tx_queue(dev, netdev_cap_txqueue(dev, qm)); + capped = netdev_cap_txqueue(dev, queue); + if (unlikely(capped != queue)) + skb_set_queue_mapping(skb, capped); + + return netdev_get_tx_queue(dev, capped); } #ifndef CONFIG_PREEMPT_RT @@ -4415,9 +4420,13 @@ static bool netdev_xmit_txqueue_skipped(void) return __this_cpu_read(softnet_data.xmit.skip_txqueue); } -void netdev_xmit_skip_txqueue(bool skip) +bool netdev_xmit_skip_txqueue(bool skip) { + bool prev = netdev_xmit_txqueue_skipped(); + __this_cpu_write(softnet_data.xmit.skip_txqueue, skip); + + return prev; } EXPORT_SYMBOL_GPL(netdev_xmit_skip_txqueue); @@ -4427,9 +4436,13 @@ static bool netdev_xmit_txqueue_skipped(void) return current->net_xmit.skip_txqueue; } -void netdev_xmit_skip_txqueue(bool skip) +bool netdev_xmit_skip_txqueue(bool skip) { + bool prev = netdev_xmit_txqueue_skipped(); + current->net_xmit.skip_txqueue = skip; + + return prev; } EXPORT_SYMBOL_GPL(netdev_xmit_skip_txqueue); #endif @@ -4848,21 +4861,25 @@ int __dev_queue_xmit(struct sk_buff *skb, struct net_device *sb_dev) tcx_set_ingress(skb, false); #ifdef CONFIG_NET_EGRESS if (static_branch_unlikely(&egress_needed_key)) { + bool skip_txq; + if (nf_hook_egress_active()) { skb = nf_hook_egress(skb, &rc, dev); if (!skb) goto out; } - netdev_xmit_skip_txqueue(false); + skip_txq = netdev_xmit_skip_txqueue(false); nf_skip_egress(skb, true); skb = sch_handle_egress(skb, &rc, dev); - if (!skb) + if (!skb) { + netdev_xmit_skip_txqueue(skip_txq); goto out; + } nf_skip_egress(skb, false); - if (netdev_xmit_txqueue_skipped()) + if (netdev_xmit_skip_txqueue(skip_txq)) txq = netdev_tx_queue_mapping(dev, skb); } #endif diff --git a/net/sched/act_skbedit.c b/net/sched/act_skbedit.c index bfec6b668410..26258b353cde 100644 --- a/net/sched/act_skbedit.c +++ b/net/sched/act_skbedit.c @@ -72,9 +72,8 @@ TC_INDIRECT_SCOPE int tcf_skbedit_act(struct sk_buff *skb, } if (params->flags & SKBEDIT_F_QUEUE_MAPPING && skb->dev->real_num_tx_queues > params->queue_mapping) { -#ifdef CONFIG_NET_EGRESS - netdev_xmit_skip_txqueue(true); -#endif + if (skb_at_tc_egress(skb)) + netdev_xmit_skip_txqueue(true); skb_set_queue_mapping(skb, tcf_skbedit_hash(params, skb)); } if (params->flags & SKBEDIT_F_MARK) { From 5f0764cb1c81a9712bce605d38efe6d844a70b29 Mon Sep 17 00:00:00 2001 From: Willem de Bruijn Date: Sat, 26 Sep 2026 10:04:54 -0400 Subject: [PATCH 341/365] net: extend IPv6 exthdr detection of tunneled packets Commit c4336a07eb6b ("net: correctly handle tunneled traffic on IPV6_CSUM GSO fallback") split skb_gso_has_extension_hdr() into mutually exclusive branches on skb->encapsulation. This did not yet address all paths: 1. With skb->encapsulation set, the outer header is not checked. IPv6 tunnels such as ip6_gre and ip6_tunnel add an outer Destination Options header by default (encap_limit). Their GSO packets skip software GSO, then skb_csum_hwoffload_help() sees the outer extension header and calls skb_checksum_help() on the GSO skb, which warns and drops it. 2. Tunnels over IPv6 without an outer transport header, such as ip6_tunnel, leave skb->transport_header at the inner transport header. skb_network_header_len() then spans the outer IPv6, tunnel and inner IP headers, a false positive. 3. UDP tunnels without an inner network header, such as SCTP-in-UDP or PSP, have no inner IPv6 header to check. Decide on the outer header alone. 4. Directly dereferencing inner_ip_hdr(skb)->version without skb_header_pointer() is unsafe. Instead, check ipv6_ext_hdr(nexthdr) on the outer IPv6 header and, if set, on the inner IPv6 header. Read the headers with skb_header_pointer(). Keep the skb_network_header_len() check when !skb->encapsulation to also catch encapsulated packets without skb->encapsulation (e.g., virtio). Use the same helper in skb_csum_hwoffload_help(). Its open coded test has the false positive of (2) and ignores the inner header. Background: checksum offload of tunneled packets invariants: Non-GSO skb: - If the inner packet is CHECKSUM_PARTIAL, Local Checksum Offload computes the outer checksum in software and the device offloads only the inner L4 checksum. - If the inner packet is CHECKSUM_NONE (e.g., SCTP-in-UDP, ESP-in-UDP, Remote Checksum Offload), the device offloads the outer UDP or GRE checksum instead. GSO skb: - A device with NETIF_F_GSO_UDP_TUNNEL_CSUM or NETIF_F_GSO_GRE_CSUM computes both inner and outer checksums per segment. The outer checksum is seeded with only the pseudo-header checksum. A NETIF_F_IPV6_CSUM device must parse through the outer headers to reach the inner ones. Fixes: c4336a07eb6b ("net: correctly handle tunneled traffic on IPV6_CSUM GSO fallback") Cc: stable@vger.kernel.org Signed-off-by: Willem de Bruijn Link: https://patch.msgid.link/20260926140506.2335137-1-willemdebruijn.kernel@gmail.com Signed-off-by: Jakub Kicinski --- net/core/dev.c | 47 +++++++++++++++++++++++++++++++---------------- 1 file changed, 31 insertions(+), 16 deletions(-) diff --git a/net/core/dev.c b/net/core/dev.c index 8638c994bbcb..18dc88990510 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -3818,20 +3818,36 @@ static netdev_features_t dflt_features_check(struct sk_buff *skb, return vlan_features_check(skb, features); } -static bool skb_gso_has_extension_hdr(const struct sk_buff *skb) +static bool __skb_has_ipv6_ext_hdr(const struct sk_buff *skb, int nhoff) { - if (!skb->encapsulation) - return ((skb_shinfo(skb)->gso_type & SKB_GSO_TCPV6 || - (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4 && - vlan_get_protocol(skb) == htons(ETH_P_IPV6))) && - skb_transport_header_was_set(skb) && - skb_network_header_len(skb) != sizeof(struct ipv6hdr)); - else - return (!skb_inner_network_header_was_set(skb) || - ((skb_shinfo(skb)->gso_type & SKB_GSO_TCPV6 || - (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4 && - inner_ip_hdr(skb)->version == 6)) && - skb_inner_network_header_len(skb) != sizeof(struct ipv6hdr))); + const struct ipv6hdr *ip6h; + struct ipv6hdr _ip6h; + + ip6h = skb_header_pointer(skb, nhoff, sizeof(_ip6h), &_ip6h); + return ip6h && ip6h->version == 6 && ipv6_ext_hdr(ip6h->nexthdr); +} + +static bool skb_has_ipv6_extension_hdr(const struct sk_buff *skb) +{ + if (vlan_get_protocol(skb) == htons(ETH_P_IPV6)) { + if (__skb_has_ipv6_ext_hdr(skb, skb_network_offset(skb))) + return true; + + /* Catch tunnels without skb->encapsulation (e.g., virtio). */ + if (!skb->encapsulation && + skb_transport_header_was_set(skb) && + skb_network_header_len(skb) != sizeof(struct ipv6hdr)) + return true; + } + + /* Tunnels without an inner network header, such as SCTP-in-UDP or + * PSP, have no inner IP header and thus no inner extension header. + */ + if (skb->encapsulation && skb_inner_network_header_was_set(skb) && + __skb_has_ipv6_ext_hdr(skb, skb_inner_network_offset(skb))) + return true; + + return false; } static netdev_features_t gso_features_check(const struct sk_buff *skb, @@ -3886,7 +3902,7 @@ static netdev_features_t gso_features_check(const struct sk_buff *skb, * so neither does TSO that depends on it. */ if (features & NETIF_F_IPV6_CSUM && - skb_gso_has_extension_hdr(skb)) + skb_has_ipv6_extension_hdr(skb)) features &= ~(NETIF_F_IPV6_CSUM | NETIF_F_TSO6 | NETIF_F_GSO_UDP_L4); return features; @@ -3988,8 +4004,7 @@ int skb_csum_hwoffload_help(struct sk_buff *skb, return 0; if (features & (NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM)) { - if (vlan_get_protocol(skb) == htons(ETH_P_IPV6) && - skb_network_header_len(skb) != sizeof(struct ipv6hdr)) + if (skb_has_ipv6_extension_hdr(skb)) goto sw_checksum; switch (skb->csum_offset) { From 6f0c2c4f5e7143eba75153ecc8af3a03f7b6a98b Mon Sep 17 00:00:00 2001 From: Sandeep Haemoon Date: Thu, 24 Sep 2026 23:51:37 +0300 Subject: [PATCH 342/365] net: ethernet: mtk_eth_soc: allocate dummy netdev sooner The RX rings and the shared NAPI are carried by an internal "dummy" netdev (eth->dummy_dev). mtk_probe() creates that dummy device, and adds the shared tx/rx NAPI to it, only after the MAC netdevs have been registered. That order has been there since the MT7623 support was added and leaves a window in which the netdevs are visible to the network core before the shared NAPI and the dummy device exist. If anything brings the first netdev up during that window (e.g. netifd opening the WAN link the instant the interface is registered), the first-open path runs without them. On NETSYS v2 and later SoCs mtk_rx_alloc() -> __xdp_rxq_info_reg() then dereferences eth->dummy_dev == NULL: WARNING: CPU: ... Missing net_device from driver in net/core/xdp.c and mtk_open() fails with -ENODEV. On NETSYS v1 SoCs (MT7621/MT7622/MT7623/MT7629) there is no page pool, so the same open instead calls napi_enable() on a napi_struct that netif_napi_add() has not initialised yet and dereferences a NULL napi->dev. Allocate the dummy netdev and add the shared tx/rx NAPI to it before the register_netdev() loop. Both the dummy-allocation failure and a register_netdev() failure now unwind through mtk_unreg_dev(), which unregisters the net_device notifiers and the netdevs registered so far, stopping a concurrently-opened netdev (ndo_stop) and its NAPI before the dummy carrier is freed. A netdev that opened during the window can also have queued the frame-engine reset worker (via the tx watchdog or phylink), so the probe error path now cancels eth->pending_work before freeing the shared state, as mtk_cleanup() already does on remove. Fixes: 656e705243fd ("net-next: mediatek: add support for MT7623 ethernet") Signed-off-by: Sandeep Haemoon Reviewed-by: Simon Horman Link: https://patch.msgid.link/patch-e52c13380b0768fec63af04c80e4b487@loljews.com Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/mediatek/mtk_eth_soc.c | 38 ++++++++++++++------- 1 file changed, 26 insertions(+), 12 deletions(-) diff --git a/drivers/net/ethernet/mediatek/mtk_eth_soc.c b/drivers/net/ethernet/mediatek/mtk_eth_soc.c index 2ea5dfe85539..ad0fdc4133fb 100644 --- a/drivers/net/ethernet/mediatek/mtk_eth_soc.c +++ b/drivers/net/ethernet/mediatek/mtk_eth_soc.c @@ -5341,6 +5341,22 @@ static int mtk_probe(struct platform_device *pdev) } } + /* we run 2 devices on the same DMA ring so we need a dummy device + * for NAPI to work. Allocate it before registering the netdevs so a + * concurrent ndo_open (e.g. netifd bringing the first netdev up the + * instant it is registered) never observes eth->dummy_dev == NULL in + * mtk_dma_init() -> mtk_rx_alloc() -> __xdp_rxq_info_reg() (net/core/xdp.c + * "Missing net_device from driver") and fails the first open. + */ + eth->dummy_dev = alloc_netdev_dummy(0); + if (!eth->dummy_dev) { + err = -ENOMEM; + dev_err(eth->dev, "failed to allocated dummy device\n"); + goto err_unreg_netdev; + } + netif_napi_add(eth->dummy_dev, ð->tx_napi, mtk_napi_tx); + netif_napi_add(eth->dummy_dev, ð->rx_napi, mtk_napi_rx); + for (i = 0; i < MTK_MAX_DEVS; i++) { if (!eth->netdev[i]) continue; @@ -5355,18 +5371,6 @@ static int mtk_probe(struct platform_device *pdev) eth->netdev[i]->base_addr, eth->irq[MTK_FE_IRQ_SHARED]); } - /* we run 2 devices on the same DMA ring so we need a dummy device - * for NAPI to work - */ - eth->dummy_dev = alloc_netdev_dummy(0); - if (!eth->dummy_dev) { - err = -ENOMEM; - dev_err(eth->dev, "failed to allocated dummy device\n"); - goto err_unreg_netdev; - } - netif_napi_add(eth->dummy_dev, ð->tx_napi, mtk_napi_tx); - netif_napi_add(eth->dummy_dev, ð->rx_napi, mtk_napi_rx); - platform_set_drvdata(pdev, eth); schedule_delayed_work(ð->reset.monitor_work, MTK_DMA_MONITOR_TIMEOUT); @@ -5376,6 +5380,16 @@ static int mtk_probe(struct platform_device *pdev) err_unreg_netdev: mtk_unreg_dev(eth); err_deinit_ppe: + /* A netdev opened while probe was still registering can have queued + * the frame-engine reset worker, which takes rtnl and drives the + * hardware; cancel it before anything below is torn down. + */ + cancel_work_sync(ð->pending_work); + if (eth->dummy_dev) { + netif_napi_del(ð->tx_napi); + netif_napi_del(ð->rx_napi); + free_netdev(eth->dummy_dev); + } mtk_ppe_deinit(eth); mtk_mdio_cleanup(eth); err_free_dev: From 20f3599d85b416fd292a199364cb1248dcd9c780 Mon Sep 17 00:00:00 2001 From: Yongzhao Chen Date: Sun, 27 Sep 2026 17:51:36 +0200 Subject: [PATCH 343/365] net: phy: qcom: at803x: Fix IPQ5018 short-cable DAC values When "qcom,dac-preset-short-cable" is set, ipq5018_config_init() programs the MDAC (MMD1 0x8100) and EDAC (debug 0x4380) fields. Both fields occupy bits 15:8 (IPQ5018_PHY_DAC_MASK), but the value 0x10 is passed unshifted as the set argument of phy_modify_mmd() and at803x_debug_reg_mask(). Neither helper shifts or masks that argument, so both fields are cleared to 0x00 instead of being set to 0x10, and bit 4 of the low byte, which is outside the field, is set. Use FIELD_PREP() to place the value in the field. This matches the vendor SDK, which clears bits 15:8 and ORs in the value shifted left by 8. On a Redmi AX5400 board, where the IPQ5018 internal PHY connects to a QCA8337 switch PHY without a cable, MDAC and EDAC read 0x6868 and 0x7800 before the write. With this change they read back 0x1068 and 0x1000, with the low byte preserved. Without it, the same writes would leave 0x0078 and 0x0010. No in-tree DTS sets this property yet, but it is documented in qca,ar803x.yaml and used by several IPQ5018 boards in OpenWrt. Fixes: d46502279a11 ("net: phy: qcom: at803x: Add Qualcomm IPQ5018 Internal PHY support") Signed-off-by: Yongzhao Chen Reviewed-by: Andrew Lunn Link: https://patch.msgid.link/20260927155136.2489-1-yongzhao.derek@gmail.com Signed-off-by: Jakub Kicinski --- drivers/net/phy/qcom/at803x.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/drivers/net/phy/qcom/at803x.c b/drivers/net/phy/qcom/at803x.c index 6872dbf77856..cacbadf1f484 100644 --- a/drivers/net/phy/qcom/at803x.c +++ b/drivers/net/phy/qcom/at803x.c @@ -1051,11 +1051,15 @@ static int ipq5018_config_init(struct phy_device *phydev) if (priv->set_short_cable_dac) { /* setting MDAC (Multi-level Digital-to-Analog Converter) in MMD1 */ phy_modify_mmd(phydev, MDIO_MMD_PMAPMD, IPQ5018_PHY_MMD1_MDAC, - IPQ5018_PHY_DAC_MASK, IPQ5018_PHY_MMD1_MDAC_VAL); + IPQ5018_PHY_DAC_MASK, + FIELD_PREP(IPQ5018_PHY_DAC_MASK, + IPQ5018_PHY_MMD1_MDAC_VAL)); /* setting EDAC (Error-detection and Correction) in debug register */ at803x_debug_reg_mask(phydev, IPQ5018_PHY_DEBUG_EDAC, - IPQ5018_PHY_DAC_MASK, IPQ5018_PHY_DEBUG_EDAC_VAL); + IPQ5018_PHY_DAC_MASK, + FIELD_PREP(IPQ5018_PHY_DAC_MASK, + IPQ5018_PHY_DEBUG_EDAC_VAL)); } return 0; From 0d2c49915bc3aea4521cda5e4d42f24f63b94585 Mon Sep 17 00:00:00 2001 From: Quentin Freimanis Date: Sun, 27 Sep 2026 21:45:42 -0700 Subject: [PATCH 344/365] net: sparx5: skip ptp deinit if init was skipped sparx5_ptp_init() returns early on the base lan969x variants because they do not have the SPX5_FEATURE_PTP flag. It also returns early when no "ptp" interrupt is described. Unbinding the driver then causes a NULL pointer dereference when cleaning up uninitialized tx_skbs queues: Unable to handle kernel NULL pointer dereference at virtual address 0000000000000008 Call trace: skb_queue_purge_reason+0x68/0x120 (P) sparx5_ptp_deinit+0x64/0xc0 mchp_sparx5_remove+0x38/0x70 platform_remove+0x20/0x30 device_remove+0x4c/0x80 Before per-port tx_skbs were added, a NULL dereference would have happened when calling ptp_clock_unregister() on the never-registered PTP clocks. Fix by returning early in sparx5_ptp_deinit() if PTP is not used. Fixes: 0933bd04047c ("net: sparx5: Add support for ptp clocks") Signed-off-by: Quentin Freimanis Reviewed-by: Daniel Machon Link: https://patch.msgid.link/20260928044542.29287-1-quentin@q-lab.dev Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/microchip/sparx5/sparx5_ptp.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/drivers/net/ethernet/microchip/sparx5/sparx5_ptp.c b/drivers/net/ethernet/microchip/sparx5/sparx5_ptp.c index a16ec8136d6d..5953e04ef14e 100644 --- a/drivers/net/ethernet/microchip/sparx5/sparx5_ptp.c +++ b/drivers/net/ethernet/microchip/sparx5/sparx5_ptp.c @@ -673,6 +673,9 @@ void sparx5_ptp_deinit(struct sparx5 *sparx5) struct sparx5_port *port; int i; + if (!sparx5->ptp) + return; + if (sparx5->ptp_irq >= 0) { disable_irq(sparx5->ptp_irq); sparx5->ptp_irq = -ENXIO; From 1b92780ae4fc3be64f05d97ed8833c445e73b0cf Mon Sep 17 00:00:00 2001 From: Jeff Jo Date: Thu, 24 Sep 2026 15:44:58 -0700 Subject: [PATCH 345/365] tcp: refresh TS.Recent for accepted old ACKs A TCP packet can carry new data while acknowledging traffic in the opposite direction. With overlapping traffic in both directions, a delayed packet's acknowledgment can be older than one Linux has already accepted, even when that packet fills a gap in the received data. Linux accepts the data, but tcp_ack() takes the old_ack path and skips updating TS.Recent, the timestamp saved for outgoing acknowledgments. The reply therefore echoes an older timestamp. If the sender uses this echo to measure round-trip time after a long idle period, its estimate includes the idle time and can reduce its sending rate. Update TS.Recent in old_ack using tcp_replace_ts_recent(), before SACK processing can trigger a transmission. This reuses the existing timestamp and sequence checks, including PAWS protection against old duplicate packets. ACK validation already rejects old ACKs in SYN_RECV before this path, so no additional state check is needed. Echoing the timestamp of the packet that fills the receive gap follows RFC 7323 section 4.3. In a socket reproduction with 300 seconds idle, controlled reordering and retransmission to exercise timestamp-based RTT sampling, the sender's smoothed round-trip time was 37.5 seconds without the fix and 15.5 ms with it. Fixes: 12fb3dd9dc3c ("tcp: call tcp_replace_ts_recent() from tcp_ack()") Assisted-by: LLM sparse Signed-off-by: Jeff Jo Reviewed-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 92bc60716f33..99baf14afdfd 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -4465,6 +4465,12 @@ static int tcp_ack(struct sock *sk, const struct sk_buff *skb, int flag) return 1; old_ack: + /* An old ACK can carry new data. Update TS.Recent before SACK + * processing can trigger a retransmission. + */ + if (flag & FLAG_UPDATE_TS_RECENT) + tcp_replace_ts_recent(tp, TCP_SKB_CB(skb)->seq); + /* If data was SACKed, tag it and see if we should send more data. * If data was DSACKed, see if we can undo a cwnd reduction. */ From 81fe707ae58e67e5a029a9778c4b0c1b478bcb97 Mon Sep 17 00:00:00 2001 From: Jeff Jo Date: Thu, 24 Sep 2026 15:44:59 -0700 Subject: [PATCH 346/365] selftests: net: check timestamp echo after an old ACK Add a regression test for a gap-filling packet whose acknowledgment has become old. Linux first sends data to the peer. Deliver two peer data packets out of order: the later packet acknowledges Linux's data, while the delayed packet still carries the earlier acknowledgment. Require the ACK that closes the receive gap to echo the delayed packet's timestamp, 301000. Without the fix, Linux accepts the data but still echoes the previously saved timestamp, 1000. Also check that the application can read all 34 bytes. Use a large jump in peer timestamps to represent the idle interval, with no real wait, loss or retransmission. The test directly checks the outgoing timestamp echo. It requires packetdrill's merged TSecr verification fix (linked below); older tools incorrectly pass on an unfixed kernel. Use the existing packetdrill selftest runner for IPv4, IPv6 and IPv4-mapped IPv6. Link: https://github.com/google/packetdrill/commit/83f72d3f9085d0e26eb4d206fe4d7cfab5b6d872 Assisted-by: LLM Signed-off-by: Jeff Jo Reviewed-by: Eric Dumazet Signed-off-by: David S. Miller --- .../net/packetdrill/tcp_old_ack_ts.pkt | 22 +++++++++++++++++++ 1 file changed, 22 insertions(+) create mode 100644 tools/testing/selftests/net/packetdrill/tcp_old_ack_ts.pkt diff --git a/tools/testing/selftests/net/packetdrill/tcp_old_ack_ts.pkt b/tools/testing/selftests/net/packetdrill/tcp_old_ack_ts.pkt new file mode 100644 index 000000000000..f072e5e4c449 --- /dev/null +++ b/tools/testing/selftests/net/packetdrill/tcp_old_ack_ts.pkt @@ -0,0 +1,22 @@ +// SPDX-License-Identifier: GPL-2.0 +// An out-of-order segment advances SND.UNA. The gap filler still carries +// its original, older ACK. The reply must echo the gap filler's timestamp. +--tcp_ts_tick_usecs=1000 + +`./defaults.sh` + +0.000 socket(..., SOCK_STREAM, IPPROTO_TCP) = 3 ++0 setsockopt(3, SOL_SOCKET, SO_REUSEADDR, [1], 4) = 0 ++0 bind(3, ..., ...) = 0 ++0 listen(3, 1) = 0 +0.100 < S 0:0(0) win 20000 ++0 > S. 0:0(0) ack 1 +0.200 < . 1:1(0) ack 1 win 20000 ++0 accept(3, ..., ...) = 4 +0.300 write(4, ..., 17) = 17 ++0 > P. 1:18(17) ack 1 +0.310 < P. 18:35(17) ack 18 win 20000 ++0 > . 18:18(0) ack 1 +0.320 < P. 1:18(17) ack 1 win 20000 ++0 > . 18:18(0) ack 35 ++0 read(4, ..., 34) = 34 From 747928b5d4ff8c8de55138a54c13980e4bb88578 Mon Sep 17 00:00:00 2001 From: Aohan Mei Date: Thu, 10 Sep 2026 16:33:26 +0800 Subject: [PATCH 347/365] netfilter: nft_flow_offload: drop flowtable reference on init error path nft_flow_offload_init() bumps the flowtable use count with nft_use_inc() before calling nf_ct_netns_get(). When the latter fails, the error is returned as-is and the reference is leaked. The upper layers do not balance it either: nf_tables_newexpr() clears expr->ops when the expression init callback fails, so the nft_expr_more() iteration in nft_rule_expr_deactivate() and nf_tables_rule_destroy() stops right before the failed expression and its ->destroy callback, which would drop the reference, never runs. Each failed rule addition therefore leaks one flowtable reference and the flowtable can no longer be removed: NFT_MSG_DELFLOWTABLE keeps reporting -EBUSY even though no rule references it. Save the nf_ct_netns_get() return value and undo the nft_use_inc() when it fails, restoring the inc/dec pairing within nft_flow_offload_init() itself. Fixes: a3c90f7a2323 ("netfilter: nf_tables: flow offload expression") Reported-by: TencentOS Corvus AI Cc: stable@vger.kernel.org Assisted-by: CodeBuddy:Kimi-K3 Signed-off-by: Aohan Mei Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nft_flow_offload.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/net/netfilter/nft_flow_offload.c b/net/netfilter/nft_flow_offload.c index 32b4281038dd..d3c5651dd699 100644 --- a/net/netfilter/nft_flow_offload.c +++ b/net/netfilter/nft_flow_offload.c @@ -160,6 +160,7 @@ static int nft_flow_offload_init(const struct nft_ctx *ctx, struct nft_flow_offload *priv = nft_expr_priv(expr); u8 genmask = nft_genmask_next(ctx->net); struct nft_flowtable *flowtable; + int err; if (!tb[NFTA_FLOW_TABLE_NAME]) return -EINVAL; @@ -174,7 +175,11 @@ static int nft_flow_offload_init(const struct nft_ctx *ctx, priv->flowtable = flowtable; - return nf_ct_netns_get(ctx->net, ctx->family); + err = nf_ct_netns_get(ctx->net, ctx->family); + if (err < 0) + nft_use_dec(&flowtable->use); + + return err; } static void nft_flow_offload_deactivate(const struct nft_ctx *ctx, From bdac17779f46817f60103baf09f0370546c3f023 Mon Sep 17 00:00:00 2001 From: Julian Anastasov Date: Thu, 10 Sep 2026 13:08:31 +0300 Subject: [PATCH 348/365] ipvs: fix missing counter decrement in lblc LBLC may delete cache entries for destinations that are removed or overloaded and replace them with available ones. But ip_vs_lblc_new() forgets to decrement the tbl->entries counter after calling ip_vs_lblc_del(). This can lead to increased shrinking of the cache with every new garbage collection. Fixes: 2f3d771a35fe ("ipvs: do not use dest after ip_vs_dest_put in LBLC") Link: https://sashiko.dev/#/patchset/0bdd5abe9968ded7ca2b9cb6844ba83d94cc8d53.1787318053.git.zhilinz%40nebusec.ai Signed-off-by: Julian Anastasov Signed-off-by: Pablo Neira Ayuso --- net/netfilter/ipvs/ip_vs_lblc.c | 1 + 1 file changed, 1 insertion(+) diff --git a/net/netfilter/ipvs/ip_vs_lblc.c b/net/netfilter/ipvs/ip_vs_lblc.c index 693bcc82ccb7..a2b574904f13 100644 --- a/net/netfilter/ipvs/ip_vs_lblc.c +++ b/net/netfilter/ipvs/ip_vs_lblc.c @@ -203,6 +203,7 @@ ip_vs_lblc_new(struct ip_vs_lblc_table *tbl, const union nf_inet_addr *daddr, if (en->dest == dest) return en; ip_vs_lblc_del(en); + atomic_dec(&tbl->entries); } en = kmalloc_obj(*en, GFP_ATOMIC); if (!en) From 2a3c3de660f96865f303fd25f1285f2ffbfd521d Mon Sep 17 00:00:00 2001 From: Zhiling Zou Date: Thu, 10 Sep 2026 13:08:32 +0300 Subject: [PATCH 349/365] ipvs: bound LBLCR and LBLC cache growth ip_vs_lblcr_new() and ip_vs_lblc_new() create cache entries for every previously unseen destination address. The table max_size only tells the periodic collector to reclaim entries after the cache has already exceeded the limit. It does not reclaim entries that the attacker continues to use. Reject new cache entries once either table reaches max_size * 3 / 2. The extra headroom lets the periodic collector catch up while the existing scheduler fallback continues to use the selected destination when cache creation fails. New traffic therefore stays serviceable without growing the tables further. Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Cc: stable@vger.kernel.org Reported-by: Vega Suggested-by: Julian Anastasov Signed-off-by: Zhiling Zou Acked-by: Julian Anastasov Signed-off-by: Pablo Neira Ayuso --- net/netfilter/ipvs/ip_vs_lblc.c | 3 +++ net/netfilter/ipvs/ip_vs_lblcr.c | 3 +++ 2 files changed, 6 insertions(+) diff --git a/net/netfilter/ipvs/ip_vs_lblc.c b/net/netfilter/ipvs/ip_vs_lblc.c index a2b574904f13..55af77a56929 100644 --- a/net/netfilter/ipvs/ip_vs_lblc.c +++ b/net/netfilter/ipvs/ip_vs_lblc.c @@ -205,6 +205,9 @@ ip_vs_lblc_new(struct ip_vs_lblc_table *tbl, const union nf_inet_addr *daddr, ip_vs_lblc_del(en); atomic_dec(&tbl->entries); } + if (atomic_read(&tbl->entries) >= tbl->max_size * 3 / 2) + return NULL; + en = kmalloc_obj(*en, GFP_ATOMIC); if (!en) return NULL; diff --git a/net/netfilter/ipvs/ip_vs_lblcr.c b/net/netfilter/ipvs/ip_vs_lblcr.c index f53f05ceea36..858393b1d2d1 100644 --- a/net/netfilter/ipvs/ip_vs_lblcr.c +++ b/net/netfilter/ipvs/ip_vs_lblcr.c @@ -363,6 +363,9 @@ ip_vs_lblcr_new(struct ip_vs_lblcr_table *tbl, const union nf_inet_addr *daddr, en = ip_vs_lblcr_get(af, tbl, daddr); if (!en) { + if (atomic_read(&tbl->entries) >= tbl->max_size * 3 / 2) + return NULL; + en = kmalloc_obj(*en, GFP_ATOMIC); if (!en) return NULL; From 616cf5c06934364ab1002b420ebe975850b4b208 Mon Sep 17 00:00:00 2001 From: Julian Anastasov Date: Sat, 19 Sep 2026 21:07:32 +0300 Subject: [PATCH 350/365] ipvs: do not create invisible templates The IP_VS_CONN_F_ONE_PACKET flag was implemented for normal connections. When conn template inherits this flag from dest->conn_flags it will not be hashed. As result, we will create new template for every new normal connection. Fix it to allow one template to be used by many normal connections. Fixes: 26ec037f9841 ("IPVS: one-packet scheduling") Link: https://netdev-ai.bots.linux.dev/sashiko/#/patchset/20260916231652.127456-1-pablo%40netfilter.org Signed-off-by: Julian Anastasov Signed-off-by: Pablo Neira Ayuso --- net/netfilter/ipvs/ip_vs_conn.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/net/netfilter/ipvs/ip_vs_conn.c b/net/netfilter/ipvs/ip_vs_conn.c index 6fa3e1dc534c..cb009208826f 100644 --- a/net/netfilter/ipvs/ip_vs_conn.c +++ b/net/netfilter/ipvs/ip_vs_conn.c @@ -1102,6 +1102,9 @@ ip_vs_bind_dest(struct ip_vs_conn *cp, struct ip_vs_dest *dest) if (cp->protocol != IPPROTO_UDP) conn_flags &= ~IP_VS_CONN_F_ONE_PACKET; flags = cp->flags; + /* Only visible templates can control multiple connections */ + if (flags & IP_VS_CONN_F_TEMPLATE) + conn_flags &= ~IP_VS_CONN_F_ONE_PACKET; /* Bind with the destination and its corresponding transmitter */ if (flags & IP_VS_CONN_F_SYNC) { /* Synced conns are hashed, so they can not get this flag */ From f96e91748f6304668fe647e686d199d7207d36f1 Mon Sep 17 00:00:00 2001 From: Julian Anastasov Date: Sat, 19 Sep 2026 21:07:33 +0300 Subject: [PATCH 351/365] ipvs: filter some flags received in the backup server While the IPVS SYNC protocol is not secure by design we can still protect the backup server from messages that can wreak havoc. This commit addresses problems from received connection flags or their combinations. We now drop messages as follows: 1. the NO_CPORT+TEMPLATE combination allows lookups for normal connections to hit template which can break in many ways. While the master does not sync connections with NO_CPORT flag, i.e. before they are established, we still accept NO_CPORT without TEMPLATE. 2. ONE_PACKET: it is not sent by master, so we do not expect it in backup. Before now it was ignored by IP_VS_CONN_F_BACKUP_MASK for protocol v1 while protocol v0 created connections that are not hashed and dropped immediately. Better to apply the IP_VS_CONN_F_BACKUP_MASK also to the flags from v0 messages for consistency with v1. Fixes: 87375ab47cd0 ("[IPVS]: ip_vs_ftp breaks connections using persistence") Signed-off-by: Julian Anastasov Signed-off-by: Pablo Neira Ayuso --- net/netfilter/ipvs/ip_vs_sync.c | 33 ++++++++++++++++++++++++++++++--- 1 file changed, 30 insertions(+), 3 deletions(-) diff --git a/net/netfilter/ipvs/ip_vs_sync.c b/net/netfilter/ipvs/ip_vs_sync.c index 5383aeafb0ae..69dc28153ec1 100644 --- a/net/netfilter/ipvs/ip_vs_sync.c +++ b/net/netfilter/ipvs/ip_vs_sync.c @@ -949,6 +949,21 @@ static void ip_vs_proc_conn(struct netns_ipvs *ipvs, struct ip_vs_conn_param *pa ip_vs_conn_put(cp); } +/* Check for incompatible flags */ +static bool ip_vs_sync_validate_flags(u32 flags) +{ + /* We do not expect NO_CPORT, especially to allow lookups + * to hit templates + */ + if (flags & IP_VS_CONN_F_NO_CPORT) { + if (flags & IP_VS_CONN_F_TEMPLATE) + return false; + } + if (flags & IP_VS_CONN_F_ONE_PACKET) + return false; + return true; +} + /* * Process received multicast message for Version 0 */ @@ -972,8 +987,7 @@ static void ip_vs_process_message_v0(struct netns_ipvs *ipvs, const char *buffer return; } s = (struct ip_vs_sync_conn_v0 *) p; - flags = ntohs(s->flags) | IP_VS_CONN_F_SYNC; - flags &= ~IP_VS_CONN_F_HASHED; + flags = ntohs(s->flags); if (flags & IP_VS_CONN_F_SEQ_MASK) { opt = (struct ip_vs_sync_conn_options *)&s[1]; p += FULL_CONN_SIZE; @@ -986,6 +1000,13 @@ static void ip_vs_process_message_v0(struct netns_ipvs *ipvs, const char *buffer p += SIMPLE_CONN_SIZE; } + if (!ip_vs_sync_validate_flags(flags)) { + IP_VS_DBG(2, "BACKUP v0, Invalid flags 0x%X\n", flags); + continue; + } + flags &= IP_VS_CONN_F_BACKUP_MASK; + flags |= IP_VS_CONN_F_SYNC; + state = ntohs(s->state); if (!(flags & IP_VS_CONN_F_TEMPLATE)) { pp = ip_vs_proto_get(s->protocol); @@ -1141,7 +1162,13 @@ static inline int ip_vs_proc_sync_conn(struct netns_ipvs *ipvs, __u8 *p, __u8 *m } /* Get flags and Mask off unsupported */ - flags = ntohl(s->v4.flags) & IP_VS_CONN_F_BACKUP_MASK; + flags = ntohl(s->v4.flags); + if (!ip_vs_sync_validate_flags(flags)) { + IP_VS_DBG(3, "BACKUP, Invalid flags 0x%X\n", flags); + retc = 25; + goto out; + } + flags &= IP_VS_CONN_F_BACKUP_MASK; flags |= IP_VS_CONN_F_SYNC; state = ntohs(s->v4.state); From 86b6471e690c67d4c5a1892993f34e080f1eae9d Mon Sep 17 00:00:00 2001 From: Weiming Shi Date: Mon, 21 Sep 2026 10:08:22 +0800 Subject: [PATCH 352/365] netfilter: nft_set_rbtree: skip transaction elements during GC Since nft_set_commit_update() runs set commit callbacks before processing NEWSETELEM transactions, nft_rbtree_gc_scan() can observe elements added by the transaction being committed. The scan records an interval end in rbe_end without checking the element's transaction state. A later, unrelated expired start then moves both elements to the expired list. The synchronous GC queue can free the new end element before the transaction subsequently activates it, causing a use-after-free. Only consider elements that are fully active in both generations. This keeps transaction-state elements out of the GC scan and preserves interval pairing across skipped elements. KASAN reports: BUG: KASAN: slab-use-after-free in nft_setelem_activate nft_setelem_activate net/netfilter/nf_tables_api.c:7047 nf_tables_commit net/netfilter/nf_tables_api.c:11137 Allocated by task 130: nft_set_elem_init net/netfilter/nf_tables_api.c:6794 nft_add_set_elem net/netfilter/nf_tables_api.c:7523 Freed by task 130: nft_trans_gc_trans_free net/netfilter/nf_tables_api.c:10506 rcu_core kernel/rcu/tree.c:2919 Fixes: 1e3b9e1c77fe ("netfilter: nf_tables: call set ops .commit when building new ruleset blob") Reported-by: Assisted-by: LLM Signed-off-by: Weiming Shi Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nft_set_rbtree.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/net/netfilter/nft_set_rbtree.c b/net/netfilter/nft_set_rbtree.c index 9894832281c4..12431b55752f 100644 --- a/net/netfilter/nft_set_rbtree.c +++ b/net/netfilter/nft_set_rbtree.c @@ -900,6 +900,8 @@ static void nft_rbtree_gc_scan(struct nft_set *set) next = rb_next(node); rbe = rb_entry(node, struct nft_rbtree_elem, node); + if (!nft_set_elem_active(&rbe->ext, NFT_GENMASK_ANY)) + continue; /* elements are reversed in the rbtree for historical reasons, * from highest to lowest value, that is why end element is From 16d464013ec2267b00a83f4bfbb97b3ecc63bfa7 Mon Sep 17 00:00:00 2001 From: Fernando Fernandez Mancera Date: Tue, 22 Sep 2026 21:18:52 +0200 Subject: [PATCH 353/365] netfilter: bpf: reject invalid NAT manipulation types As bpf_ct_set_nat_info() is not validating the NAT manipulation type a wrong value can be passed directly to nf_nat_setup_info(). This triggers the WARN_ON() at nf_nat_setup_info() and if panic_on_warn isn't set, then IPS_SRC_NAT_DONE is set without adding nat_bysource and conntrack cleanup tries to unlink an uninitialized hlist node. Fix this by checking that NAT manipulation type is correct before calling nf_nat_setup_info(). In addition, if the WARN_ON is hit, return NF_DROP instead of continuing with the processing to avoid similar situations in the future. Reported-by: VEGA Fixes: 0fabd2aa199f ("net: netfilter: add bpf_ct_set_nat_info kfunc helper") Signed-off-by: Fernando Fernandez Mancera Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nf_nat_bpf.c | 3 +++ net/netfilter/nf_nat_core.c | 5 +++-- 2 files changed, 6 insertions(+), 2 deletions(-) diff --git a/net/netfilter/nf_nat_bpf.c b/net/netfilter/nf_nat_bpf.c index f9dd85ccea01..7572b58c448b 100644 --- a/net/netfilter/nf_nat_bpf.c +++ b/net/netfilter/nf_nat_bpf.c @@ -39,6 +39,9 @@ __bpf_kfunc int bpf_ct_set_nat_info(struct nf_conn___init *nfct, if (proto != NFPROTO_IPV4 && proto != NFPROTO_IPV6) return -EINVAL; + if (manip != NF_NAT_MANIP_SRC && manip != NF_NAT_MANIP_DST) + return -EINVAL; + memset(&range, 0, sizeof(struct nf_nat_range2)); range.flags = NF_NAT_RANGE_MAP_IPS; range.min_addr = *addr; diff --git a/net/netfilter/nf_nat_core.c b/net/netfilter/nf_nat_core.c index a4858c2b2d65..cc8e1e81006d 100644 --- a/net/netfilter/nf_nat_core.c +++ b/net/netfilter/nf_nat_core.c @@ -767,8 +767,9 @@ nf_nat_setup_info(struct nf_conn *ct, if (nf_ct_is_confirmed(ct)) return NF_ACCEPT; - WARN_ON(maniptype != NF_NAT_MANIP_SRC && - maniptype != NF_NAT_MANIP_DST); + if (WARN_ON(maniptype != NF_NAT_MANIP_SRC && + maniptype != NF_NAT_MANIP_DST)) + return NF_DROP; if (WARN_ON(nf_nat_initialized(ct, maniptype))) return NF_DROP; From 7c549fb7eecd01fdba7c0d12c01da876ef8a3214 Mon Sep 17 00:00:00 2001 From: Pablo Neira Ayuso Date: Wed, 23 Sep 2026 10:53:02 +0200 Subject: [PATCH 354/365] netfilter: flowtable: generalize pending status bit Rename NF_FLOW_HW_PENDING to NF_FLOW_PENDING and use it to inhibit the flowtable GC worker until pending hw offload work has been completed. Apparently, nf_flow_offload_stats() can schedule work to retrieve stats while the flow is being removed by GC. And this bit can also be used in a follow up patch to disable GC until the flow has been fully added in both directions. Revert the reordering done in commit d644b23afe1e ("netfilter: flowtable: publish HW_DEAD after worker is done") to prevent a race between GC and hw offload handler. Fixes: 2c8897953f3b ("netfilter: flowtable: Add pending bit for offload work") Signed-off-by: Pablo Neira Ayuso --- include/net/netfilter/nf_flow_table.h | 2 +- net/netfilter/nf_flow_table_core.c | 7 ++++++- net/netfilter/nf_flow_table_offload.c | 14 +++++--------- net/sched/act_ct.c | 2 +- 4 files changed, 13 insertions(+), 12 deletions(-) diff --git a/include/net/netfilter/nf_flow_table.h b/include/net/netfilter/nf_flow_table.h index f2e2771f188f..5b611efaa3cd 100644 --- a/include/net/netfilter/nf_flow_table.h +++ b/include/net/netfilter/nf_flow_table.h @@ -183,10 +183,10 @@ enum nf_flow_flags { NF_FLOW_DNAT, NF_FLOW_CLOSING, NF_FLOW_TEARDOWN, + NF_FLOW_PENDING, NF_FLOW_HW, NF_FLOW_HW_DYING, NF_FLOW_HW_DEAD, - NF_FLOW_HW_PENDING, NF_FLOW_HW_BIDIRECTIONAL, NF_FLOW_HW_ESTABLISHED, }; diff --git a/net/netfilter/nf_flow_table_core.c b/net/netfilter/nf_flow_table_core.c index 934c6151f558..36bbc7be2f74 100644 --- a/net/netfilter/nf_flow_table_core.c +++ b/net/netfilter/nf_flow_table_core.c @@ -575,7 +575,12 @@ static void nf_flow_table_extend_ct_timeout(struct nf_conn *ct) static void nf_flow_offload_gc_step(struct nf_flowtable *flow_table, struct flow_offload *flow, void *data) { - bool teardown = test_bit(NF_FLOW_TEARDOWN, &flow->flags); + bool teardown; + + if (test_bit(NF_FLOW_PENDING, &flow->flags)) + return; + + teardown = test_bit(NF_FLOW_TEARDOWN, &flow->flags); if (nf_flow_has_expired(flow) || nf_ct_is_dying(flow->ct) || diff --git a/net/netfilter/nf_flow_table_offload.c b/net/netfilter/nf_flow_table_offload.c index 6757fd89c1f1..4365859220e6 100644 --- a/net/netfilter/nf_flow_table_offload.c +++ b/net/netfilter/nf_flow_table_offload.c @@ -995,6 +995,7 @@ static void flow_offload_work_del(struct flow_offload_work *offload) flow_offload_tuple_del(offload, FLOW_OFFLOAD_DIR_ORIGINAL); if (test_bit(NF_FLOW_HW_BIDIRECTIONAL, &offload->flow->flags)) flow_offload_tuple_del(offload, FLOW_OFFLOAD_DIR_REPLY); + set_bit(NF_FLOW_HW_DEAD, &offload->flow->flags); } static void flow_offload_tuple_stats(struct flow_offload_work *offload, @@ -1056,13 +1057,8 @@ static void flow_offload_work_handler(struct work_struct *work) default: WARN_ON_ONCE(1); } - - clear_bit(NF_FLOW_HW_PENDING, &offload->flow->flags); - if (offload->cmd == FLOW_CLS_DESTROY) { - /* Publish after the worker's last flow access. */ - smp_mb__before_atomic(); - set_bit(NF_FLOW_HW_DEAD, &offload->flow->flags); - } + smp_mb__before_atomic(); + clear_bit(NF_FLOW_PENDING, &offload->flow->flags); kfree(offload); } @@ -1089,12 +1085,12 @@ nf_flow_offload_work_alloc(struct nf_flowtable *flowtable, { struct flow_offload_work *offload; - if (test_and_set_bit(NF_FLOW_HW_PENDING, &flow->flags)) + if (test_and_set_bit(NF_FLOW_PENDING, &flow->flags)) return NULL; offload = kmalloc_obj(struct flow_offload_work, GFP_ATOMIC); if (!offload) { - clear_bit(NF_FLOW_HW_PENDING, &flow->flags); + clear_bit(NF_FLOW_PENDING, &flow->flags); return NULL; } diff --git a/net/sched/act_ct.c b/net/sched/act_ct.c index 55f3521edb4c..626c9a5af0ef 100644 --- a/net/sched/act_ct.c +++ b/net/sched/act_ct.c @@ -289,7 +289,7 @@ static bool tcf_ct_flow_is_outdated(const struct flow_offload *flow) { return test_bit(IPS_SEEN_REPLY_BIT, &flow->ct->status) && test_bit(IPS_HW_OFFLOAD_BIT, &flow->ct->status) && - !test_bit(NF_FLOW_HW_PENDING, &flow->flags) && + !test_bit(NF_FLOW_PENDING, &flow->flags) && !test_bit(NF_FLOW_HW_ESTABLISHED, &flow->flags); } From 3ae37eafd36694bb2d3227f60ac98fbf602470a2 Mon Sep 17 00:00:00 2001 From: Pablo Neira Ayuso Date: Wed, 23 Sep 2026 20:46:18 +0200 Subject: [PATCH 355/365] netfilter: flowtable: restore ieee80211 forward path Before commit 871df5007eda ("netfilter: flowtable: bail out if forward path cannot be discovered"), there was a fallback to set up a forward path in case .ndo_fill_forward_path fails or DEV_PATH_MTK_WDMA was used. Such fallback was used by commit d787a3e38f01 ("mac80211: add support for .ndo_fill_forward_path"). One possibility is to handle DEV_PATH_MTK_WDMA from the flowtable forward path discovery. However, this is only used internally by drivers to retrieve mtk_wdma information to set up hardware offload. Felix decided to use the .fill_forward_path interface for this purpose due to the lack of a better interface at that time. Add a new DEV_PATH_IEEE80211 path which is offered if the new ieee80211 flag is set on in the struct net_device_path_ctx to restore the flowtable with a ieee80211 netdevice. Handle this new DEV_PATH_IEEE80211 path just like DEV_PATH_ETHERNET and DEV_PATH_DSA, ie. this is the last netdevice in the stack. This new ieee80211 flag is implicitly unset for mtk_ppe and airoha which call dev_fill_forward_path() to retrieve a DEV_PATH_MTK_WDMA path. Fixes: 871df5007eda ("netfilter: flowtable: bail out if forward path cannot be discovered") Signed-off-by: Pablo Neira Ayuso --- include/linux/netdevice.h | 3 +++ net/mac80211/iface.c | 7 +++++++ net/netfilter/nf_flow_table_path.c | 3 +++ 3 files changed, 13 insertions(+) diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h index 87cafc932e9e..3cff2174dc03 100644 --- a/include/linux/netdevice.h +++ b/include/linux/netdevice.h @@ -887,6 +887,7 @@ enum net_device_path_type { DEV_PATH_DSA, DEV_PATH_MTK_WDMA, DEV_PATH_TUN, + DEV_PATH_IEEE80211, }; struct net_device_path { @@ -953,6 +954,8 @@ struct net_device_path_ctx { u16 id; __be16 proto; } vlan[NET_DEVICE_PATH_VLAN_MAX]; + + bool ieee80211; }; enum tc_setup_type { diff --git a/net/mac80211/iface.c b/net/mac80211/iface.c index 889c32fd8de1..c5584435fde9 100644 --- a/net/mac80211/iface.c +++ b/net/mac80211/iface.c @@ -1023,6 +1023,13 @@ static int ieee80211_netdev_fill_forward_path(struct net_device_path_ctx *ctx, struct sta_info *sta; int ret = -ENOENT; + if (ctx->ieee80211) { + path->type = DEV_PATH_IEEE80211; + path->dev = ctx->dev; + ctx->dev = NULL; + return 0; + } + sdata = IEEE80211_DEV_TO_SUB_IF(ctx->dev); local = sdata->local; diff --git a/net/netfilter/nf_flow_table_path.c b/net/netfilter/nf_flow_table_path.c index 1e55644f2edb..d90013685bf1 100644 --- a/net/netfilter/nf_flow_table_path.c +++ b/net/netfilter/nf_flow_table_path.c @@ -53,6 +53,7 @@ static int nft_dev_fill_forward_path(const struct dst_entry *dst_cache, struct net_device_path_ctx ctx = { .dev = dev, .ether_type = ether_type, + .ieee80211 = true, }; struct neighbour *n; u8 nud_state; @@ -114,6 +115,7 @@ static int nft_dev_path_info(struct net_device_path_stack *stack, path = &stack->path[i]; switch (path->type) { case DEV_PATH_ETHERNET: + case DEV_PATH_IEEE80211: case DEV_PATH_DSA: case DEV_PATH_VLAN: case DEV_PATH_PPPOE: @@ -123,6 +125,7 @@ static int nft_dev_path_info(struct net_device_path_stack *stack, memcpy(info->h_source, path->dev->dev_addr, ETH_ALEN); if (path->type == DEV_PATH_ETHERNET || + path->type == DEV_PATH_IEEE80211 || path->type == DEV_PATH_DSA) break; From 99b43ede9e355ba35244cc9470bf1819774ce39d Mon Sep 17 00:00:00 2001 From: Harshit Mogalapalli Date: Thu, 24 Sep 2026 23:39:36 -0700 Subject: [PATCH 356/365] net: microchip: vcap: stop scanning after deleting key field The only call outside the KUnit tests is in sparx5_tc_add_rule_copy(), which passes a fixed key list with no duplicate entries. Therefore, the potential double-free is not currently reachable in tree. However, vcap_filter_rule_keys() is exported and does not require its key list entries to be unique. If a future or external caller supplies the same key at another position that the inner loop visits, it calls list_del() and kfree() a second time on the freed field, leading to a potential double-free. Break the inner loop after removing a matching field. The outer safe list traversal continues to process the remaining rule fields. Fixes: 465a38a269e9 ("net: microchip: sparx5: Support for copying and modifying rules in the API") Reported-by: Dan Carpenter Closes: https://lore.kernel.org/all/ahs-lCAXXikkaHky@stanley.mountain/ Signed-off-by: Harshit Mogalapalli Reviewed-by: Daniel Machon Signed-off-by: David S. Miller --- drivers/net/ethernet/microchip/vcap/vcap_api.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/net/ethernet/microchip/vcap/vcap_api.c b/drivers/net/ethernet/microchip/vcap/vcap_api.c index 788c0728d763..0f8d98f2094d 100644 --- a/drivers/net/ethernet/microchip/vcap/vcap_api.c +++ b/drivers/net/ethernet/microchip/vcap/vcap_api.c @@ -3527,8 +3527,8 @@ int vcap_filter_rule_keys(struct vcap_rule *rule, if (key == keylist[idx]) { list_del(&ckf->ctrl.list); kfree(ckf); - idx++; err = 0; + break; } } } From 5642b1648d5c4561d8ed7a410d0a7017f706604d Mon Sep 17 00:00:00 2001 From: Quentin Freimanis Date: Sun, 27 Sep 2026 21:50:12 -0700 Subject: [PATCH 357/365] net: sparx5: make ports inherit the switch base mac address type When the switch uses a random base MAC address it does not mark each port's address as random: $ dmesg | grep "MAC addr" sparx5-switch e00c0000.switch: MAC addr was not set, use random MAC $ cat /sys/class/net/eth0/addr_assign_type 0 0 is NET_ADDR_PERM, so userspace thinks this is a stable address. With a random MAC reported as NET_ADDR_PERM, systemd's MACAddressPolicy=persistent [1] leaves it unchanged, so a DHCP client gets a new IP address every boot and a static reservation can't be used. Reporting it as NET_ADDR_RANDOM makes systemd replace it with a stable address derived from the interface name and machine-id. Fixes: f3cad2611a77 ("net: sparx5: add hostmode with phylink support") Signed-off-by: Quentin Freimanis Reviewed-by: Simon Horman Reviewed-by: Daniel Machon Link: https://patch.msgid.link/20260928045012.30178-1-quentin@q-lab.dev Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/microchip/sparx5/sparx5_main.c | 1 + drivers/net/ethernet/microchip/sparx5/sparx5_main.h | 1 + drivers/net/ethernet/microchip/sparx5/sparx5_netdev.c | 1 + 3 files changed, 3 insertions(+) diff --git a/drivers/net/ethernet/microchip/sparx5/sparx5_main.c b/drivers/net/ethernet/microchip/sparx5/sparx5_main.c index dad713e9ddd5..b530fb6d3d0c 100644 --- a/drivers/net/ethernet/microchip/sparx5/sparx5_main.c +++ b/drivers/net/ethernet/microchip/sparx5/sparx5_main.c @@ -898,6 +898,7 @@ static int mchp_sparx5_probe(struct platform_device *pdev) dev_info(sparx5->dev, "MAC addr was not set, use random MAC\n"); eth_random_addr(sparx5->base_mac); sparx5->base_mac[5] = 0; + sparx5->base_mac_assign_type = NET_ADDR_RANDOM; } sparx5->fdma_irq = platform_get_irq_byname(sparx5->pdev, "fdma"); diff --git a/drivers/net/ethernet/microchip/sparx5/sparx5_main.h b/drivers/net/ethernet/microchip/sparx5/sparx5_main.h index eb57b86fbe22..0f7d8440f614 100644 --- a/drivers/net/ethernet/microchip/sparx5/sparx5_main.h +++ b/drivers/net/ethernet/microchip/sparx5/sparx5_main.h @@ -392,6 +392,7 @@ struct sparx5 { struct notifier_block switchdev_blocking_nb; /* Switch state */ u8 base_mac[ETH_ALEN]; + u8 base_mac_assign_type; /* Associated bridge device (when bridged) */ struct net_device *hw_bridge_dev; /* Bridged interfaces */ diff --git a/drivers/net/ethernet/microchip/sparx5/sparx5_netdev.c b/drivers/net/ethernet/microchip/sparx5/sparx5_netdev.c index 4dedf5d92641..b3725cf2dd44 100644 --- a/drivers/net/ethernet/microchip/sparx5/sparx5_netdev.c +++ b/drivers/net/ethernet/microchip/sparx5/sparx5_netdev.c @@ -294,6 +294,7 @@ struct net_device *sparx5_create_netdev(struct sparx5 *sparx5, u32 portno) ndev->ethtool_ops = &sparx5_ethtool_ops; eth_hw_addr_gen(ndev, sparx5->base_mac, portno + 1); + ndev->addr_assign_type = sparx5->base_mac_assign_type; return ndev; } From 4f1da630d13de0370e2f6361f961f1c8b384af1c Mon Sep 17 00:00:00 2001 From: Jun Yang Date: Sat, 26 Sep 2026 17:55:15 +0800 Subject: [PATCH 358/365] sctp: check RCV_SHUTDOWN after the sendmsg connect wait sctp_wait_for_connect() drops the socket lock while it sleeps. An out-of-the-blue ABORT can then be processed from the socket backlog and unlink the association. If a concurrent shutdown(fd, SHUT_RD) sets RCV_SHUTDOWN, the waiter breaks with err == 0 before checking asoc->base.dead. Its final sctp_association_put() can then free the association, leaving sctp_sendmsg_to_asoc() to continue with a dangling pointer. Check RCV_SHUTDOWN along with the wait error in sctp_sendmsg_to_asoc() before using the association again. The check only accesses the socket, so it needs no additional association reference. Return the existing -ESRCH so that sctp_sendmsg() skips freeing a new association that may already have been destroyed. Keep sctp_wait_for_connect() unchanged to preserve its behavior for the connect() caller. Fixes: 668c9beb9020 ("sctp: implement assign_number for sctp_stream_interleave") Cc: stable@vger.kernel.org Reported-by: TencentOS Corvus AI Signed-off-by: Jun Yang Acked-by: Xin Long Link: https://patch.msgid.link/20260926095606.68601-1-juny24602@gmail.com Signed-off-by: Jakub Kicinski --- net/sctp/socket.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/net/sctp/socket.c b/net/sctp/socket.c index c7b9e325ec1c..4652fd90d9a6 100644 --- a/net/sctp/socket.c +++ b/net/sctp/socket.c @@ -1844,7 +1844,7 @@ static int sctp_sendmsg_to_asoc(struct sctp_association *asoc, if (asoc->ep->intl_enable) { timeo = sock_sndtimeo(sk, 0); err = sctp_wait_for_connect(asoc, &timeo); - if (err) { + if (err || sk->sk_shutdown & RCV_SHUTDOWN) { err = -ESRCH; goto err; } From 9e792901418d814295674d1f53d6b615fb5c15c8 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Jari=20Sepp=C3=A4l=C3=A4?= Date: Mon, 28 Sep 2026 11:47:11 +0300 Subject: [PATCH 359/365] octeontx2-pf: Fix RSS indirection table size MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The conversion to the dedicated RSS context operations replaced the pointer to struct otx2_rss_ctx with an inline u32 ind_tbl[] array. However, otx2_rss_init() still uses sizeof(*rss->ind_tbl) to set rss_size. This now yields the size of one u32 (4), rather than the 256 entries in the indirection table. The RSS initialization and hardware programming loops use rss_size, so only four entries are initialized and programmed despite the NIX LF being allocated a 256-entry table. On an OCTEON CN102 with eight RX queues, the table contained 0, 1, 2, 3 followed by zeros. Traffic was concentrated on queue 0 and ethtool -X equal 8 did not correct the table. Use ARRAY_SIZE() to count the indirection table entries. On the same hardware, this restores the full table repeating queue numbers 0 through 7, and bidirectional multi-flow traffic increments multiple RX queues. Runtime testing on an Asterfusion ET2500 (OCTEON CN102 A0). Fixes: 62e01d8c4170 ("eth: otx2: migrate to the *_rxfh_context ops") Cc: stable@vger.kernel.org Closes: https://lore.kernel.org/netdev/361df563-e29e-43bf-8806-24bdfefed33d@nocloud.org/ Signed-off-by: Jari Seppälä Reviewed-by: Ratheesh Kannoth Link: https://patch.msgid.link/20260928084711.2462571-1-jase@nocloud.org Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c index 0f3f405d9f1b..b11389a841d1 100644 --- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c @@ -394,7 +394,7 @@ int otx2_rss_init(struct otx2_nic *pfvf) struct otx2_rss_info *rss = &pfvf->hw.rss_info; int idx, ret = 0; - rss->rss_size = sizeof(*rss->ind_tbl); + rss->rss_size = ARRAY_SIZE(rss->ind_tbl); /* Init RSS key if it is not setup already */ if (!rss->enable) From 3cdeaef1754ab8155cdd828c958ada3d36e9ad9a Mon Sep 17 00:00:00 2001 From: Oleksij Rempel Date: Fri, 25 Sep 2026 09:10:40 +0200 Subject: [PATCH 360/365] r8169: disable EEE on RTL8168h/8111h Force EEE off on the RTL8168h/8111h (RTL_GIGA_MAC_VER_46) at PHY connect with phy_disable_eee(). Commit 202fef9bbbf5 ("net: phy: realtek: fix EEE advertisement write on the internal PHY MMD path") made EEE actually advertise on the generic Realtek PHY; the write had been a no-op before. On the RTL8168h that un-masks a latent defect: once EEE negotiates, RX silently stalls after ~7-20 minutes. The carrier stays up, no counter or dmesg moves, and only "ip link set down/up" recovers it; disabling EEE keeps the link stable. Root-causing the RTL8168h LPI/RX path needs hardware not available now, so disable EEE for this version. Use phy_disable_eee() rather than dropping the version from rtl_supports_eee(): the latter also skips rtl_enable_tx_lpi(), whose disable branch clears the MAC TX-LPI bits (ERI 0x1b0[1:0]) on link up. rtl_hw_start_8168h_1() does not clear them (the RTL8402/RTL8106e init does), so a warm reboot from an EEE-active state could otherwise leave TX-LPI asserted while the PHY no longer negotiates EEE. Keeping the version EEE-capable preserves that clear path. This also covers the RTL8168M, which shares RTL_GIGA_MAC_VER_46. Reported-by: Dennis Piecha Closes: https://lore.kernel.org/all/353419280.954808.1790290283530@mail.yahoo.com/ Fixes: 202fef9bbbf5 ("net: phy: realtek: fix EEE advertisement write on the internal PHY MMD path") Cc: stable@vger.kernel.org Signed-off-by: Oleksij Rempel Link: https://patch.msgid.link/20260925071040.2137178-1-o.rempel@pengutronix.de Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/realtek/r8169_main.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/drivers/net/ethernet/realtek/r8169_main.c b/drivers/net/ethernet/realtek/r8169_main.c index ec4fc21fa21f..ee1bc9bd0447 100644 --- a/drivers/net/ethernet/realtek/r8169_main.c +++ b/drivers/net/ethernet/realtek/r8169_main.c @@ -5484,6 +5484,12 @@ static int r8169_mdio_register(struct rtl8169_private *tp) phy_disable_eee_mode(tp->phydev, ETHTOOL_LINK_MODE_2500baseT_Full_BIT); + /* Keep EEE off but the version stays EEE-capable so that link-up + * still clears the MAC TX-LPI bits. + */ + if (tp->mac_version == RTL_GIGA_MAC_VER_46) + phy_disable_eee(tp->phydev); + /* PHY will be woken up in rtl_open() */ phy_suspend(tp->phydev); From 7b97273cbebe9dd4fddc457f0c4f2a124d0e3996 Mon Sep 17 00:00:00 2001 From: Prathamesh Deshpande Date: Sat, 26 Sep 2026 17:53:57 +0100 Subject: [PATCH 361/365] net/mlx5e: Fix AF_XDP TX timestamp teardown NULL dereference During XSK TX queue teardown, outstanding descriptors are completed without a CQE. If one requested a TX timestamp, the completion path passes that NULL CQE to mlx5e_xsk_fill_timestamp(), which dereferences it. Return zero when no CQE is available, indicating that teardown did not produce a TX timestamp. Fixes: ec706a860eba ("net/mlx5e: Implement AF_XDP TX timestamp and checksum offload") Signed-off-by: Prathamesh Deshpande Reviewed-by: Tariq Toukan Link: https://patch.msgid.link/20260926165402.5902-1-prathameshdeshpande7@gmail.com Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/mellanox/mlx5/core/en/xdp.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/xdp.c b/drivers/net/ethernet/mellanox/mlx5/core/en/xdp.c index 77ea51bfbaae..2cca20e1b9b7 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en/xdp.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/xdp.c @@ -293,6 +293,9 @@ static u64 mlx5e_xsk_fill_timestamp(void *_priv) struct mlx5e_xsk_tx_complete *priv = _priv; u64 ts; + if (!priv->cqe) + return 0; + ts = get_cqe_ts(priv->cqe); if (mlx5_is_real_time_rq(priv->cq->mdev) || mlx5_is_real_time_sq(priv->cq->mdev)) From 8ec454dcd00ce3375119111c6394964455b36d0d Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Mon, 28 Sep 2026 19:45:24 +0000 Subject: [PATCH 362/365] ipv6: sr: use skb_get_hash_net() in seg6_make_flowlabel() Since commit d58e468b1112 ("flow_dissector: implements flow dissector BPF hook") __skb_flow_dissect() needs a net pointer, either from skb->dev, skb->sk, or since commit 3cbf4ffba5ee ("net: plumb network namespace into __skb_flow_dissect") a caller provided pointer. syzbot was able to reach seg6_make_flowlabel() with an skb having neither skb->dev nor skb->sk set: a TIPC UDP bearer sends a discovery message through an IPv4 route using seg6 encap, while net.ipv6.seg6_flowlabel is set to 1. seg6_make_flowlabel() already has a net pointer, use skb_get_hash_net(). WARNING: net/core/flow_dissector.c:1131 at __skb_flow_dissect+0x910/0x5368 net/core/flow_dissector.c:1126, CPU#0: syz.0.17/4930 Call trace: __skb_flow_dissect+0x910/0x5368 net/core/flow_dissector.c:1126 (P) __skb_get_hash_net+0xe0/0x29c net/core/flow_dissector.c:1903 skb_get_hash include/linux/skbuff.h:1663 [inline] seg6_make_flowlabel+0xcc/0x1ec net/ipv6/seg6_iptunnel.c:132 __seg6_do_srh_encap+0x320/0xbf4 net/ipv6/seg6_iptunnel.c:161 seg6_do_srh+0x4b4/0xa44 net/ipv6/seg6_iptunnel.c:431 seg6_output_core+0x164/0x688 net/ipv6/seg6_iptunnel.c:681 seg6_output+0x44/0x1ac net/ipv6/seg6_iptunnel.c:744 lwtunnel_output+0x3d0/0x664 net/core/lwtunnel.c:356 dst_output include/net/dst.h:470 [inline] ip_local_out+0x110/0x148 net/ipv4/ip_output.c:131 iptunnel_xmit+0x50c/0xd38 net/ipv4/ip_tunnel_core.c:97 udp_tunnel_xmit_skb+0x220/0x348 net/ipv4/udp_tunnel_core.c:187 tipc_udp_xmit+0x75c/0x9c0 net/tipc/udp_media.c:202 tipc_udp_send_msg+0x214/0x374 net/tipc/udp_media.c:274 tipc_bearer_xmit_skb+0x260/0x3b0 net/tipc/bearer.c:576 tipc_enable_bearer net/tipc/bearer.c:366 [inline] __tipc_nl_bearer_enable+0xc90/0xfb0 net/tipc/bearer.c:1048 tipc_nl_bearer_enable+0x2c/0x48 net/tipc/bearer.c:1057 genl_family_rcv_msg_doit+0x1e4/0x2d4 net/netlink/genetlink.c:1114 Fixes: d58e468b1112 ("flow_dissector: implements flow dissector BPF hook") Reported-by: syzbot+9408fbe0e6452a12e9ab@syzkaller.appspotmail.com Closes: https://lore.kernel.org/netdev/6abac2eb.3654fce1.1bec97.0000.GAE@google.com/ Signed-off-by: Eric Dumazet Reviewed-by: Xuanqiang Luo Reviewed-by: Ido Schimmel Link: https://patch.msgid.link/20260928194524.3617299-1-edumazet@kernel.org Signed-off-by: Jakub Kicinski --- net/ipv6/seg6_iptunnel.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/net/ipv6/seg6_iptunnel.c b/net/ipv6/seg6_iptunnel.c index 61c6a27bf202..7cbe0685f8bb 100644 --- a/net/ipv6/seg6_iptunnel.c +++ b/net/ipv6/seg6_iptunnel.c @@ -129,7 +129,7 @@ static __be32 seg6_make_flowlabel(struct net *net, struct sk_buff *skb, u32 hash; if (do_flowlabel > 0) { - hash = skb_get_hash(skb); + hash = skb_get_hash_net(net, skb); hash = rol32(hash, 16); flowlabel = (__force __be32)hash & IPV6_FLOWLABEL_MASK; } else if (!do_flowlabel && skb->protocol == htons(ETH_P_IPV6)) { From 8f1c2a10500a84a621ff13073f96deea0753ed42 Mon Sep 17 00:00:00 2001 From: Lorenzo Bianconi Date: Tue, 29 Sep 2026 15:43:03 +0200 Subject: [PATCH 363/365] net: mvneta: clear XDP pfmemalloc flag between frames MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit mvneta_swbm_add_rx_fragment() sets XDP_FLAGS_FRAGS_PF_MEMALLOC on the xdp_buff when a fragment page is a pfmemalloc one (page under memory pressure). The xdp_buff is reused for the next frame, but only the XDP_FLAGS_HAS_FRAGS bit was cleared at frame start, so the pfmemalloc bit leaked from one frame into the following ones. mvneta_swbm_build_skb() propagates the flag to skb->pfmemalloc through xdp_update_skb_frags_info(), so the skb of a subsequent fragmented frame could be wrongly marked as pfmemalloc even if none of its pages are under pressure. Clear all the xdp_buff flags in mvneta_swbm_rx_frame(), which is invoked for each new frame, instead of just the XDP_FLAGS_HAS_FRAGS bit. Fixes: ed7a58cb40bd ("net: marvell: rely on xdp_update_skb_shared_info utility routine") Reviewed-by: Simon Horman Signed-off-by: Lorenzo Bianconi Reviewed-by: Toke Høiland-Jørgensen Link: https://patch.msgid.link/20260929-mvneta-xdp-clear-frag-fix-v4-1-1e63b25eeed8@oss.qualcomm.com Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/marvell/mvneta.c | 2 +- include/net/xdp.h | 5 +++++ 2 files changed, 6 insertions(+), 1 deletion(-) diff --git a/drivers/net/ethernet/marvell/mvneta.c b/drivers/net/ethernet/marvell/mvneta.c index 543e566425c1..4f479b6f731e 100644 --- a/drivers/net/ethernet/marvell/mvneta.c +++ b/drivers/net/ethernet/marvell/mvneta.c @@ -2340,7 +2340,7 @@ mvneta_swbm_rx_frame(struct mvneta_port *pp, /* Prefetch header */ prefetch(data); - xdp_buff_clear_frags_flag(xdp); + xdp_reinit_buff(xdp); xdp_prepare_buff(xdp, data, pp->rx_offset_correction + MVNETA_MH_SIZE, data_len, true); } diff --git a/include/net/xdp.h b/include/net/xdp.h index aa742f413c35..07231adfb5f8 100644 --- a/include/net/xdp.h +++ b/include/net/xdp.h @@ -106,6 +106,11 @@ struct xdp_buff { }; }; +static __always_inline void xdp_reinit_buff(struct xdp_buff *xdp) +{ + xdp->flags = 0; +} + static __always_inline bool xdp_buff_has_frags(const struct xdp_buff *xdp) { return !!(xdp->flags & XDP_FLAGS_HAS_FRAGS); From 7375d38364a9aa66fb31716bcefef38aecad75d8 Mon Sep 17 00:00:00 2001 From: Reinhard Speyerer Date: Fri, 25 Sep 2026 15:14:32 +0200 Subject: [PATCH 364/365] net: usb: qmi_wwan: add Rolling Wireless RN947R Add Rolling Wireless RN947R 0x9300 composition: DIAG + ADB + NMEA + MODEM + RMNET + QDSS + ADPL T: Bus=01 Lev=01 Prnt=01 Port=02 Cnt=02 Dev#= 6 Spd=480 MxCh= 0 D: Ver= 2.10 Cls=ef(misc ) Sub=02 Prot=01 MxPS=64 #Cfgs= 1 P: Vendor=33f8 ProdID=9300 Rev= 0.00 S: Manufacturer=Rolling Wireless S: Product=RN947R S: SerialNumber=xxxxxxxxxxx C:* #Ifs= 7 Cfg#= 1 Atr=a0 MxPwr=500mA I:* If#= 0 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=ff Prot=30 Driver=option E: Ad=01(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms E: Ad=81(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms I:* If#= 1 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=42 Prot=01 Driver=usbfs E: Ad=02(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms E: Ad=82(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms I:* If#= 2 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=ff Prot=40 Driver=option E: Ad=84(I) Atr=03(Int.) MxPS= 10 Ivl=32ms E: Ad=83(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms E: Ad=03(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms I:* If#= 3 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=ff Prot=40 Driver=option E: Ad=86(I) Atr=03(Int.) MxPS= 10 Ivl=32ms E: Ad=85(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms E: Ad=04(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms I:* If#= 8 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=ff Prot=50 Driver=qmi_wwan E: Ad=88(I) Atr=03(Int.) MxPS= 8 Ivl=32ms E: Ad=87(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms E: Ad=05(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms I:* If#=21 Alt= 0 #EPs= 1 Cls=ff(vend.) Sub=ff Prot=70 Driver=(none) E: Ad=89(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms I:* If#=22 Alt= 0 #EPs= 1 Cls=ff(vend.) Sub=ff Prot=80 Driver=(none) E: Ad=8a(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms Signed-off-by: Reinhard Speyerer Link: https://patch.msgid.link/arZzuOOvLUZtdOlv@arcor.de Signed-off-by: Jakub Kicinski --- drivers/net/usb/qmi_wwan.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/net/usb/qmi_wwan.c b/drivers/net/usb/qmi_wwan.c index 0e2ab567dd40..a44d1b9756a1 100644 --- a/drivers/net/usb/qmi_wwan.c +++ b/drivers/net/usb/qmi_wwan.c @@ -1455,6 +1455,7 @@ static const struct usb_device_id products[] = { {QMI_QUIRK_SET_DTR(0x1546, 0x1312, 4)}, /* u-blox LARA-R6 01B */ {QMI_QUIRK_SET_DTR(0x1546, 0x1342, 4)}, /* u-blox LARA-L6 */ {QMI_QUIRK_SET_DTR(0x33f8, 0x0104, 4)}, /* Rolling RW101 RMNET */ + {QMI_QUIRK_SET_DTR(0x33f8, 0x9300, 8)}, /* Rolling RN947R RMNET */ {QMI_FIXED_INTF(0x2dee, 0x4d22, 5)}, /* MeiG Smart SRM825L */ {QMI_QUIRK_SET_DTR(0x2dee, 0x4d63, 0)}, /* MeiG SRM813Q w/ Modem(PPP) */ {QMI_QUIRK_SET_DTR(0x2dee, 0x4d64, 0)}, /* MeiG SRM813Q w/ NMEA */ From 6e0022b5ae3dc5b833af0fcc1578dc20a1d6bc71 Mon Sep 17 00:00:00 2001 From: Bartosz Golaszewski Date: Wed, 23 Sep 2026 17:14:35 +0200 Subject: [PATCH 365/365] net: phy: aquantia: fix system interface type not updated in forced mode aqr_gen1_read_status() decodes the MDIO_PHYXS_VEND_IF_STATUS register to determine which SerDes interface the PHY is currently using on its system side and stores the result in phydev->interface. phylink relies on this value to configure the MAC. The autoneg == AUTONEG_DISABLE check is not correct: MDIO_PHYXS_VEND_IF_STATUS is set by the PHY firmware based on the negotiated link speed, not based on whether autoneg was used to reach it. When the link comes up at 1G in forced mode, the register correctly reads SGMII, but the early return prevents phydev->interface from being updated. It stays at whatever value it held before (typically 2500BASE-X from the initial autoneg run), so phylink configures the MAC for the wrong interface and the link cannot come up. Remove the autoneg guard so that the system interface type is always decoded when the link is up. Cc: stable@vger.kernel.org Fixes: 110a2432c520 ("net: phy: aquantia: add downshift support") Signed-off-by: Bartosz Golaszewski Link: https://patch.msgid.link/20260923-qcom-sa8255p-emac-v15-1-e82f33720737@oss.qualcomm.com Signed-off-by: Paolo Abeni --- drivers/net/phy/aquantia/aquantia_main.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/drivers/net/phy/aquantia/aquantia_main.c b/drivers/net/phy/aquantia/aquantia_main.c index e95d391a6dc2..096407bebd52 100644 --- a/drivers/net/phy/aquantia/aquantia_main.c +++ b/drivers/net/phy/aquantia/aquantia_main.c @@ -544,7 +544,7 @@ static int aqr_gen1_read_status(struct phy_device *phydev) if (ret) return ret; - if (!phydev->link || phydev->autoneg == AUTONEG_DISABLE) + if (!phydev->link) return 0; /* The status register is not immediately correct on line side link up. @@ -594,6 +594,9 @@ static int aqr_gen1_read_status(struct phy_device *phydev) phydev->interface = aqr_translate_interface(phydev, interface); + if (phydev->autoneg == AUTONEG_DISABLE) + return 0; + /* Read rate from vendor register */ return aqr_gen1_read_rate(phydev); }