Skip to contentSkip to conversation

Make ksoftirqd a per-cpu variable

PATCH New Rusty Russell 1 participant 1 message Jul 1, 2003

lore
Share & tools
lore

Rusty Russell

author
#0 lore raw

PATCH 1/1 Make ksoftirqd a per-cpu variable

Linus, please apply.

Moves the ksoftirqd pointers out of the irq_stat struct, and use a normal per-cpu variable. It's not that time critical, nor referenced in assembler. This moves us closer to making irq_stat a per-cpu variable.

Because some archs have hardcoded asm references to offsets in this structure, I haven't touched non-x86. The __ksoftirqd_task field is unused in other archs, too.

Name: ksoftirqds in per-cpu variable
Author: Rusty Russell
Status: Tested on 2.5.73-bk8
Depends: Percpu/irq_syscall_count_removal.patch.gz

D: Moves the ksoftirqd pointers out of the irq_stat struct, and use a
D: normal per-cpu variable. It's not that time critical, nor referenced
D: in assembler. This moves us closer to making irq_stat a per-cpu variable. D:
D: Because some archs have hardcoded asm references to offsets in this
D: structure, I haven't touched non-x86. The __ksoftirqd_task field
D: is unused in other archs, too.

include/asm-i386/hardirq.h +0 -1
diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal .5097-2.5.73-bk8-ksoftirqd_percpu.pre/include/asm-i386/hardirq.h .5097-2.5.73-bk8-ksoftirqd_percpu/include/asm-i386/hardirq.h
@@ -7,7 +7,6 @@
 
 typedef struct {
 	unsigned int __softirq_pending;
-	struct task_struct * __ksoftirqd_task; /* waitqueue is too large */
 	unsigned long idle_timestamp;
 	unsigned int __nmi_count;	/* arch dependent */
 	unsigned int apic_timer_irqs;	/* arch dependent */
include/linux/irq_cpustat.h +0 -2
diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal .5097-2.5.73-bk8-ksoftirqd_percpu.pre/include/linux/irq_cpustat.h .5097-2.5.73-bk8-ksoftirqd_percpu/include/linux/irq_cpustat.h
@@ -29,8 +29,6 @@ extern irq_cpustat_t irq_stat[];		/* def
   /* arch independent irq_stat fields */
 #define softirq_pending(cpu)	__IRQ_STAT((cpu), __softirq_pending)
 #define local_softirq_pending()	softirq_pending(smp_processor_id())
-#define ksoftirqd_task(cpu)	__IRQ_STAT((cpu), __ksoftirqd_task)
-#define local_ksoftirqd_task()	ksoftirqd_task(smp_processor_id())
 
   /* arch dependent irq_stat fields */
 #define nmi_count(cpu)		__IRQ_STAT((cpu), __nmi_count)	/* i386 */
kernel/softirq.c +10 -6
diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal .5097-2.5.73-bk8-ksoftirqd_percpu.pre/kernel/softirq.c .5097-2.5.73-bk8-ksoftirqd_percpu/kernel/softirq.c
@@ -14,6 +14,7 @@
 #include <linux/init.h>
 #include <linux/mm.h>
 #include <linux/notifier.h>
+#include <linux/percpu.h>
 #include <linux/cpu.h>
 
 /*

@@ -41,15 +42,18 @@ EXPORT_SYMBOL(irq_stat);
 
 static struct softirq_action softirq_vec[32] __cacheline_aligned_in_smp;
 
+static DEFINE_PER_CPU(struct task_struct *, ksoftirqd);
+
 /*
  * we cannot loop indefinitely here to avoid userspace starvation,
  * but we also don't want to introduce a worst case 1/HZ latency
  * to the pending events, so lets the scheduler to balance
  * the softirq load for us.
  */
-static inline void wakeup_softirqd(unsigned cpu)
+static inline void wakeup_softirqd(void)
 {
-	struct task_struct * tsk = ksoftirqd_task(cpu);
+	/* Interrupts are disabled: no need to stop preemption */
+	struct task_struct *tsk = __get_cpu_var(ksoftirqd);
 
 	if (tsk && tsk->state != TASK_RUNNING)
 		wake_up_process(tsk);

@@ -96,7 +100,7 @@ restart:
 			goto restart;
 		}
 		if (pending)
-			wakeup_softirqd(smp_processor_id());
+			wakeup_softirqd();
 		__local_bh_enable();
 	}
 

@@ -131,7 +135,7 @@ inline void cpu_raise_softirq(unsigned i
 	 * schedule the softirq soon.
 	 */
 	if (!in_interrupt())
-		wakeup_softirqd(cpu);
+		wakeup_softirqd();
 }
 
 void raise_softirq(unsigned int nr)

@@ -325,7 +329,7 @@ static int ksoftirqd(void * __bind_cpu)
 	__set_current_state(TASK_INTERRUPTIBLE);
 	mb();
 
-	local_ksoftirqd_task() = current;
+	__get_cpu_var(ksoftirqd) = current;
 
 	for (;;) {
 		if (!local_softirq_pending())

@@ -354,7 +358,7 @@ static int __devinit cpu_callback(struct
 			return NOTIFY_BAD;
 		}
 
-		while (!ksoftirqd_task(hotcpu))
+		while (!per_cpu(ksoftirqd, hotcpu))
 			yield();
  	}
 	return NOTIFY_OK;

--
Anyone who quotes me in their sig is an idiot. -- Rusty Russell.