From 4fe985292709eeb6a4653c71660f893e26c2f2dd Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Mon, 20 Apr 2026 20:03:26 -1000 Subject: [PATCH 1/2] rhashtable: Bounce deferred worker kick through irq_work Inserts past 75% load call schedule_work(&ht->run_work) to kick an async resize. If a caller holds a raw spinlock (e.g. an insecure_elasticity user), schedule_work() under that lock records caller_lock -> pool->lock -> pi_lock -> rq->__lock A cycle forms if any of these locks is acquired in the reverse direction elsewhere. sched_ext, the only current insecure_elasticity user, hits this: it holds scx_sched_lock across rhashtable inserts of sub-schedulers, while scx_bypass() takes rq->__lock -> scx_sched_lock. Exercising the resize path produces: Chain exists of: &pool->lock --> &rq->__lock --> scx_sched_lock Bounce the kick from the insert paths through irq_work so schedule_work() runs from hard IRQ context with the caller's lock no longer held. rht_deferred_worker()'s self-rearm on error stays on schedule_work(&ht->run_work) - the worker runs in process context with no caller lock held, and keeping the self-requeue on @run_work lets cancel_work_sync() in rhashtable_free_and_destroy() drain it. v3: Keep rht_deferred_worker()'s self-rearm on schedule_work(&run_work). Routing it through irq_work in v2 broke cancel_work_sync()'s self-requeue handling - an irq_work queued after irq_work_sync() returned but while cancel_work_sync() was still waiting could fire post-teardown. v2: Bounce unconditionally instead of gating on insecure_elasticity, as suggested by Herbert. Signed-off-by: Tejun Heo Acked-by: Herbert Xu --- include/linux/rhashtable-types.h | 3 +++ include/linux/rhashtable.h | 3 ++- lib/rhashtable.c | 31 ++++++++++++++++++++++++++++--- 3 files changed, 33 insertions(+), 4 deletions(-) diff --git a/include/linux/rhashtable-types.h b/include/linux/rhashtable-types.h index 72082428d6c6..fc2f596a6df1 100644 --- a/include/linux/rhashtable-types.h +++ b/include/linux/rhashtable-types.h @@ -12,6 +12,7 @@ #include #include #include +#include #include #include @@ -77,6 +78,7 @@ struct rhashtable_params { * @p: Configuration parameters * @rhlist: True if this is an rhltable * @run_work: Deferred worker to expand/shrink asynchronously + * @run_irq_work: Bounces the @run_work kick through hard IRQ context. * @mutex: Mutex to protect current/future table swapping * @lock: Spin lock to protect walker list * @nelems: Number of elements in table @@ -88,6 +90,7 @@ struct rhashtable { struct rhashtable_params p; bool rhlist; struct work_struct run_work; + struct irq_work run_irq_work; struct mutex mutex; spinlock_t lock; atomic_t nelems; diff --git a/include/linux/rhashtable.h b/include/linux/rhashtable.h index 7def3f0f556b..ef5230cece36 100644 --- a/include/linux/rhashtable.h +++ b/include/linux/rhashtable.h @@ -20,6 +20,7 @@ #include #include +#include #include #include #include @@ -847,7 +848,7 @@ static __always_inline void *__rhashtable_insert_fast( rht_assign_unlock(tbl, bkt, obj, flags); if (rht_grow_above_75(ht, tbl)) - schedule_work(&ht->run_work); + irq_work_queue(&ht->run_irq_work); data = NULL; out: diff --git a/lib/rhashtable.c b/lib/rhashtable.c index fb2b7bc137ba..7a67ef5b67b6 100644 --- a/lib/rhashtable.c +++ b/lib/rhashtable.c @@ -441,10 +441,33 @@ static void rht_deferred_worker(struct work_struct *work) mutex_unlock(&ht->mutex); + /* + * Re-arm via @run_work, not @run_irq_work. + * rhashtable_free_and_destroy() drains async work as irq_work_sync() + * followed by cancel_work_sync(). If this site queued irq_work while + * cancel_work_sync() was waiting for us, irq_work_sync() would already + * have returned and the stale irq_work could fire post-teardown. + * cancel_work_sync() natively handles self-requeue on @run_work. + */ if (err) schedule_work(&ht->run_work); } +/* + * Insert-path callers can run under a raw spinlock (e.g. an insecure_elasticity + * user). Calling schedule_work() under that lock records caller_lock -> + * pool->lock -> pi_lock -> rq->__lock, closing a locking cycle if any of + * these is acquired in the reverse direction elsewhere. Bounce through + * irq_work so the schedule_work() runs with the caller's lock no longer held. + */ +static void rht_deferred_irq_work(struct irq_work *irq_work) +{ + struct rhashtable *ht = container_of(irq_work, struct rhashtable, + run_irq_work); + + schedule_work(&ht->run_work); +} + static int rhashtable_insert_rehash(struct rhashtable *ht, struct bucket_table *tbl) { @@ -477,7 +500,7 @@ static int rhashtable_insert_rehash(struct rhashtable *ht, if (err == -EEXIST) err = 0; } else - schedule_work(&ht->run_work); + irq_work_queue(&ht->run_irq_work); return err; @@ -488,7 +511,7 @@ static int rhashtable_insert_rehash(struct rhashtable *ht, /* Schedule async rehash to retry allocation in process context. */ if (err == -ENOMEM) - schedule_work(&ht->run_work); + irq_work_queue(&ht->run_irq_work); return err; } @@ -630,7 +653,7 @@ static void *rhashtable_try_insert(struct rhashtable *ht, const void *key, rht_unlock(tbl, bkt, flags); if (inserted && rht_grow_above_75(ht, tbl)) - schedule_work(&ht->run_work); + irq_work_queue(&ht->run_irq_work); } } while (!IS_ERR_OR_NULL(new_tbl)); @@ -1085,6 +1108,7 @@ int rhashtable_init_noprof(struct rhashtable *ht, RCU_INIT_POINTER(ht->tbl, tbl); INIT_WORK(&ht->run_work, rht_deferred_worker); + init_irq_work(&ht->run_irq_work, rht_deferred_irq_work); return 0; } @@ -1150,6 +1174,7 @@ void rhashtable_free_and_destroy(struct rhashtable *ht, struct bucket_table *tbl, *next_tbl; unsigned int i; + irq_work_sync(&ht->run_irq_work); cancel_work_sync(&ht->run_work); mutex_lock(&ht->mutex); From 05909810a946222aca5d0611d37be82d18f95228 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Mon, 20 Apr 2026 21:17:11 -1000 Subject: [PATCH 2/2] tools/sched_ext: scx_qmap: Silence task_ctx lookup miss scx_fork() dispatches ops.init_task to exactly one scheduler - the one owning the forking task's cgroup. A task forked inside a sub-scheduler's cgroup is init'd into the sub only; the root scheduler has no task_ctx entry for it. When that task later appears as @prev in the root's qmap_dispatch() (or flows through core-sched comparison via task_qdist), the bpf_task_storage_get() legitimately misses. qmap treated those misses as fatal via scx_bpf_error("task_ctx lookup failed") and aborted the scheduler as soon as the first cross-sched task hit the root. Drop the error in the sites where the miss is legitimate: lookup_task_ctx() (helper; callers already check for NULL), qmap_dispatch()'s @prev branch (bookkeeping-only), task_qdist() (returns 0 which makes the comparison a no-op), and qmap_select_cpu() (returns prev_cpu as a no-op fallback instead of -ESRCH). The existing scx_error was a paranoid guard from the pre-sub-sched world where every task was owned by the one and only scheduler. v2: qmap_select_cpu() returns prev_cpu on NULL instead of -ESRCH, so the root scheduler doesn't error on cross-sched tasks that pass through it (Andrea Righi). Fixes: 4f8b122848db ("sched_ext: Add basic building blocks for nested sub-scheduler dispatching") Signed-off-by: Tejun Heo Reviewed-by: Andrea Righi Reviewed-by: Zhao Mengmeng --- tools/sched_ext/scx_qmap.bpf.c | 24 ++++++------------------ 1 file changed, 6 insertions(+), 18 deletions(-) diff --git a/tools/sched_ext/scx_qmap.bpf.c b/tools/sched_ext/scx_qmap.bpf.c index b68abb9e760b..aad698fe294b 100644 --- a/tools/sched_ext/scx_qmap.bpf.c +++ b/tools/sched_ext/scx_qmap.bpf.c @@ -159,13 +159,7 @@ static s32 pick_direct_dispatch_cpu(struct task_struct *p, s32 prev_cpu) static struct task_ctx *lookup_task_ctx(struct task_struct *p) { - struct task_ctx *tctx; - - if (!(tctx = bpf_task_storage_get(&task_ctx_stor, p, 0, 0))) { - scx_bpf_error("task_ctx lookup failed"); - return NULL; - } - return tctx; + return bpf_task_storage_get(&task_ctx_stor, p, 0, 0); } s32 BPF_STRUCT_OPS(qmap_select_cpu, struct task_struct *p, @@ -175,7 +169,7 @@ s32 BPF_STRUCT_OPS(qmap_select_cpu, struct task_struct *p, s32 cpu; if (!(tctx = lookup_task_ctx(p))) - return -ESRCH; + return prev_cpu; if (p->scx.weight < 2 && !(p->flags & PF_KTHREAD)) return prev_cpu; @@ -540,13 +534,9 @@ void BPF_STRUCT_OPS(qmap_dispatch, s32 cpu, struct task_struct *prev) */ if (prev) { tctx = bpf_task_storage_get(&task_ctx_stor, prev, 0, 0); - if (!tctx) { - scx_bpf_error("task_ctx lookup failed"); - return; - } - - tctx->core_sched_seq = - core_sched_tail_seqs[weight_to_idx(prev->scx.weight)]++; + if (tctx) + tctx->core_sched_seq = + core_sched_tail_seqs[weight_to_idx(prev->scx.weight)]++; } } @@ -584,10 +574,8 @@ static s64 task_qdist(struct task_struct *p) s64 qdist; tctx = bpf_task_storage_get(&task_ctx_stor, p, 0, 0); - if (!tctx) { - scx_bpf_error("task_ctx lookup failed"); + if (!tctx) return 0; - } qdist = tctx->core_sched_seq - core_sched_head_seqs[idx];