linux-kernel - [patch V2 2/3] x86/irq: Cleanup posted MSI code

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20251125214631.108458942@linutronix.de>
Date: Tue, 25 Nov 2025 22:50:47 +0100 (CET)
From: Thomas Gleixner <tglx@...utronix.de>
To: LKML <linux-kernel@...r.kernel.org>
Cc: x86@...nel.org,
 Luigi Rizzo <lrizzo@...gle.com>
Subject: [patch V2 2/3] 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 <tglx@...utronix.de>
---
V2: Use __this_cpu...() - Luigi
---
 arch/x86/kernel/irq.c |   31 +++++++++++++------------------
 1 file changed, 13 insertions(+), 18 deletions(-)
--- a/arch/x86/kernel/irq.c
+++ b/arch/x86/kernel/irq.c
@@ -400,11 +400,9 @@ static DEFINE_PER_CPU_CACHE_HOT(bool, po
 
 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
@@ -448,8 +446,8 @@ static __always_inline bool handle_pendi
 }
 
 /*
- * 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
 
@@ -459,11 +457,8 @@ static __always_inline bool handle_pendi
  */
 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);
@@ -471,25 +466,25 @@ DEFINE_IDTENTRY_SYSVEC(sysvec_posted_msi
 	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);