From 4021a6dad720273a95ac3c0816fc48e35e77dace Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 25 Nov 2025 22:50:47 +0100 Subject: [PATCH 1/2] x86/irq: Cleanup posted MSI code Make code and comments readable and use __this_cpu..() as this is guaranteed to be invoked with interrupts disabled. Signed-off-by: Thomas Gleixner Link: https://patch.msgid.link/20251125214631.108458942@linutronix.de --- arch/x86/kernel/irq.c | 31 +++++++++++++------------------ 1 file changed, 13 insertions(+), 18 deletions(-) diff --git a/arch/x86/kernel/irq.c b/arch/x86/kernel/irq.c index b2fe6181960c..d817febfd4bc 100644 --- a/arch/x86/kernel/irq.c +++ b/arch/x86/kernel/irq.c @@ -401,11 +401,9 @@ static DEFINE_PER_CPU_CACHE_HOT(bool, posted_msi_handler_active); void intel_posted_msi_init(void) { - u32 destination; - u32 apic_id; + u32 destination, apic_id; this_cpu_write(posted_msi_pi_desc.nv, POSTED_MSI_NOTIFICATION_VECTOR); - /* * APIC destination ID is stored in bit 8:15 while in XAPIC mode. * VT-d spec. CH 9.11 @@ -449,8 +447,8 @@ static __always_inline bool handle_pending_pir(unsigned long *pir, struct pt_reg } /* - * Performance data shows that 3 is good enough to harvest 90+% of the benefit - * on high IRQ rate workload. + * Performance data shows that 3 is good enough to harvest 90+% of the + * benefit on high interrupt rate workloads. */ #define MAX_POSTED_MSI_COALESCING_LOOP 3 @@ -460,11 +458,8 @@ static __always_inline bool handle_pending_pir(unsigned long *pir, struct pt_reg */ DEFINE_IDTENTRY_SYSVEC(sysvec_posted_msi_notification) { + struct pi_desc *pid = this_cpu_ptr(&posted_msi_pi_desc); struct pt_regs *old_regs = set_irq_regs(regs); - struct pi_desc *pid; - int i = 0; - - pid = this_cpu_ptr(&posted_msi_pi_desc); /* Mark the handler active for intel_ack_posted_msi_irq() */ __this_cpu_write(posted_msi_handler_active, true); @@ -472,25 +467,25 @@ DEFINE_IDTENTRY_SYSVEC(sysvec_posted_msi_notification) irq_enter(); /* - * Max coalescing count includes the extra round of handle_pending_pir - * after clearing the outstanding notification bit. Hence, at most - * MAX_POSTED_MSI_COALESCING_LOOP - 1 loops are executed here. + * Loop only MAX_POSTED_MSI_COALESCING_LOOP - 1 times here to take + * the final handle_pending_pir() invocation after clearing the + * outstanding notification bit into account. */ - while (++i < MAX_POSTED_MSI_COALESCING_LOOP) { + for (int i = 1; i < MAX_POSTED_MSI_COALESCING_LOOP; i++) { if (!handle_pending_pir(pid->pir, regs)) break; } /* - * Clear outstanding notification bit to allow new IRQ notifications, - * do this last to maximize the window of interrupt coalescing. + * Clear the outstanding notification bit to rearm the notification + * mechanism. */ pi_clear_on(pid); /* - * There could be a race of PI notification and the clearing of ON bit, - * process PIR bits one last time such that handling the new interrupts - * are not delayed until the next IRQ. + * Clearing the ON bit can race with a notification. Process the + * PIR bits one last time so that handling the new interrupts is + * not delayed until the next notification happens. */ handle_pending_pir(pid->pir, regs); From d441e38a2c87824afc7e656e634e55141d015307 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 25 Nov 2025 22:50:49 +0100 Subject: [PATCH 2/2] x86/irq_remapping: Sanitize posted_msi_supported() posted_msi_supported() is a misnomer as it actually checks whether it is enabled or not. Aside of that this does not take CONFIG_X86_POSTED_MSI into account which is required to actually use it. Rename it to posted_msi_enabled() and make the return value depend on CONFIG_X86_POSTED_MSI, which allows the compiler to eliminate the related dead code and data if disabled: text data bss dec hex filename 10046 701 3296 14043 36db drivers/iommu/intel/irq_remapping.o 9904 413 3296 13613 352d drivers/iommu/intel/irq_remapping.o Signed-off-by: Thomas Gleixner Link: https://patch.msgid.link/20251125214631.170499997@linutronix.de --- arch/x86/include/asm/irq_remapping.h | 5 +++-- drivers/iommu/intel/irq_remapping.c | 4 ++-- 2 files changed, 5 insertions(+), 4 deletions(-) diff --git a/arch/x86/include/asm/irq_remapping.h b/arch/x86/include/asm/irq_remapping.h index 4e55d1755846..37b94f484ef3 100644 --- a/arch/x86/include/asm/irq_remapping.h +++ b/arch/x86/include/asm/irq_remapping.h @@ -67,9 +67,10 @@ static inline struct irq_domain *arch_get_ir_parent_domain(void) extern bool enable_posted_msi; -static inline bool posted_msi_supported(void) +static inline bool posted_msi_enabled(void) { - return enable_posted_msi && irq_remapping_cap(IRQ_POSTING_CAP); + return IS_ENABLED(CONFIG_X86_POSTED_MSI) && + enable_posted_msi && irq_remapping_cap(IRQ_POSTING_CAP); } #else /* CONFIG_IRQ_REMAP */ diff --git a/drivers/iommu/intel/irq_remapping.c b/drivers/iommu/intel/irq_remapping.c index 8bcbfe3d9c72..ecb591e98565 100644 --- a/drivers/iommu/intel/irq_remapping.c +++ b/drivers/iommu/intel/irq_remapping.c @@ -1368,7 +1368,7 @@ static void intel_irq_remapping_prepare_irte(struct intel_ir_data *data, break; case X86_IRQ_ALLOC_TYPE_PCI_MSI: case X86_IRQ_ALLOC_TYPE_PCI_MSIX: - if (posted_msi_supported()) { + if (posted_msi_enabled()) { prepare_irte_posted(irte); data->irq_2_iommu.posted_msi = 1; } @@ -1460,7 +1460,7 @@ static int intel_irq_remapping_alloc(struct irq_domain *domain, irq_data->hwirq = (index << 16) + i; irq_data->chip_data = ird; - if (posted_msi_supported() && + if (posted_msi_enabled() && ((info->type == X86_IRQ_ALLOC_TYPE_PCI_MSI) || (info->type == X86_IRQ_ALLOC_TYPE_PCI_MSIX))) irq_data->chip = &intel_ir_chip_post_msi;