irq_work: Add generic hardirq context callbacks

Provide a mechanism that allows running code in IRQ context. It is
most useful for NMI code that needs to interact with the rest of the
system -- like wakeup a task to drain buffers.

Perf currently has such a mechanism, so extract that and provide it as
a generic feature, independent of perf so that others may also
benefit.

The IRQ context callback is generated through self-IPIs where
possible, or on architectures like powerpc the decrementer (the
built-in timer facility) is set to generate an interrupt immediately.

Architectures that don't have anything like this get to do with a
callback from the timer tick. These architectures can call
irq_work_run() at the tail of any IRQ handlers that might enqueue such
work (like the perf IRQ handler) to avoid undue latencies in
processing the work.

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: Kyle McMartin <kyle@mcmartin.ca>
Acked-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
[ various fixes ]
Signed-off-by: Huang Ying <ying.huang@intel.com>
LKML-Reference: <1287036094.7768.291.camel@yhuang-dev>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

arch/alpha/Kconfig

@@ -9,6 +9,7 @@ config ALPHA
 	select HAVE_IDE
 	select HAVE_OPROFILE
 	select HAVE_SYSCALL_WRAPPERS
+	select HAVE_IRQ_WORK
 	select HAVE_PERF_EVENTS
 	select HAVE_DMA_ATTRS
 	help

arch/alpha/include/asm/perf_event.h

 #ifndef __ASM_ALPHA_PERF_EVENT_H
 #define __ASM_ALPHA_PERF_EVENT_H
 
-/* Alpha only supports software events through this interface. */
-extern void set_perf_event_pending(void);
-
-#define PERF_EVENT_INDEX_OFFSET 0
-
 #ifdef CONFIG_PERF_EVENTS
 extern void init_hw_perf_events(void);
 #else

arch/alpha/kernel/time.c

@@ -41,7 +41,7 @@
 #include <linux/init.h>
 #include <linux/bcd.h>
 #include <linux/profile.h>
-#include <linux/perf_event.h>
+#include <linux/irq_work.h>
 
 #include <asm/uaccess.h>
 #include <asm/io.h>
@@ -83,25 +83,25 @@ static struct {
 
 unsigned long est_cycle_freq;
 
-#ifdef CONFIG_PERF_EVENTS
+#ifdef CONFIG_IRQ_WORK
 
-DEFINE_PER_CPU(u8, perf_event_pending);
+DEFINE_PER_CPU(u8, irq_work_pending);
 
-#define set_perf_event_pending_flag()  __get_cpu_var(perf_event_pending) = 1
-#define test_perf_event_pending()      __get_cpu_var(perf_event_pending)
-#define clear_perf_event_pending()     __get_cpu_var(perf_event_pending) = 0
+#define set_irq_work_pending_flag()  __get_cpu_var(irq_work_pending) = 1
+#define test_irq_work_pending()      __get_cpu_var(irq_work_pending)
+#define clear_irq_work_pending()     __get_cpu_var(irq_work_pending) = 0
 
-void set_perf_event_pending(void)
+void set_irq_work_pending(void)
 {
-	set_perf_event_pending_flag();
+	set_irq_work_pending_flag();
 }
 
-#else  /* CONFIG_PERF_EVENTS */
+#else  /* CONFIG_IRQ_WORK */
 
-#define test_perf_event_pending()      0
-#define clear_perf_event_pending()
+#define test_irq_work_pending()      0
+#define clear_irq_work_pending()
 
-#endif /* CONFIG_PERF_EVENTS */
+#endif /* CONFIG_IRQ_WORK */
 
 
 static inline __u32 rpcc(void)
@@ -191,9 +191,9 @@ irqreturn_t timer_interrupt(int irq, void *dev)
 
 	write_sequnlock(&xtime_lock);
 
-	if (test_perf_event_pending()) {
-		clear_perf_event_pending();
-		perf_event_do_pending();
+	if (test_irq_work_pending()) {
+		clear_irq_work_pending();
+		irq_work_run();
 	}
 
 #ifndef CONFIG_SMP

arch/arm/Kconfig

@@ -23,6 +23,7 @@ config ARM
 	select HAVE_KERNEL_GZIP
 	select HAVE_KERNEL_LZO
 	select HAVE_KERNEL_LZMA
+	select HAVE_IRQ_WORK
 	select HAVE_PERF_EVENTS
 	select PERF_USE_VMALLOC
 	select HAVE_REGS_AND_STACK_ACCESS_API

arch/arm/include/asm/perf_event.h

@@ -12,18 +12,6 @@
 #ifndef __ARM_PERF_EVENT_H__
 #define __ARM_PERF_EVENT_H__
 
-/*
- * NOP: on *most* (read: all supported) ARM platforms, the performance
- * counter interrupts are regular interrupts and not an NMI. This
- * means that when we receive the interrupt we can call
- * perf_event_do_pending() that handles all of the work with
- * interrupts disabled.
- */
-static inline void
-set_perf_event_pending(void)
-{
-}
-
 /* ARM performance counters start from 1 (in the cp15 accesses) so use the
  * same indexes here for consistency. */
 #define PERF_EVENT_INDEX_OFFSET 1

arch/arm/kernel/perf_event.c

@@ -1092,7 +1092,7 @@ armv6pmu_handle_irq(int irq_num,
 	 * platforms that can have the PMU interrupts raised as an NMI, this
 	 * will not work.
 	 */
-	perf_event_do_pending();
+	irq_work_run();
 
 	return IRQ_HANDLED;
 }
@@ -2068,7 +2068,7 @@ static irqreturn_t armv7pmu_handle_irq(int irq_num, void *dev)
 	 * platforms that can have the PMU interrupts raised as an NMI, this
 	 * will not work.
 	 */
-	perf_event_do_pending();
+	irq_work_run();
 
 	return IRQ_HANDLED;
 }
@@ -2436,7 +2436,7 @@ xscale1pmu_handle_irq(int irq_num, void *dev)
 			armpmu->disable(hwc, idx);
 	}
 
-	perf_event_do_pending();
+	irq_work_run();
 
 	/*
 	 * Re-enable the PMU.
@@ -2763,7 +2763,7 @@ xscale2pmu_handle_irq(int irq_num, void *dev)
 			armpmu->disable(hwc, idx);
 	}
 
-	perf_event_do_pending();
+	irq_work_run();
 
 	/*
 	 * Re-enable the PMU.

arch/frv/Kconfig

@@ -7,6 +7,7 @@ config FRV
 	default y
 	select HAVE_IDE
 	select HAVE_ARCH_TRACEHOOK
+	select HAVE_IRQ_WORK
 	select HAVE_PERF_EVENTS
 
 config ZONE_DMA

arch/frv/lib/Makefile

@@ -5,4 +5,4 @@
 lib-y := \
 	__ashldi3.o __lshrdi3.o __muldi3.o __ashrdi3.o __negdi2.o __ucmpdi2.o \
 	checksum.o memcpy.o memset.o atomic-ops.o atomic64-ops.o \
-	outsl_ns.o outsl_sw.o insl_ns.o insl_sw.o cache.o perf_event.o
+	outsl_ns.o outsl_sw.o insl_ns.o insl_sw.o cache.o

arch/frv/lib/perf_event.c

-/* Performance event handling
- *
- * Copyright (C) 2009 Red Hat, Inc. All Rights Reserved.
- * Written by David Howells (dhowells@redhat.com)
- *
- * This program is free software; you can redistribute it and/or
- * modify it under the terms of the GNU General Public Licence
- * as published by the Free Software Foundation; either version
- * 2 of the Licence, or (at your option) any later version.
- */
-
-#include <linux/perf_event.h>
-
-/*
- * mark the performance event as pending
- */
-void set_perf_event_pending(void)
-{
-}

arch/parisc/Kconfig

@@ -16,6 +16,7 @@ config PARISC
 	select RTC_DRV_GENERIC
 	select INIT_ALL_POSSIBLE
 	select BUG
+	select HAVE_IRQ_WORK
 	select HAVE_PERF_EVENTS
 	select GENERIC_ATOMIC64 if !64BIT
 	help

arch/parisc/include/asm/perf_event.h

 #ifndef __ASM_PARISC_PERF_EVENT_H
 #define __ASM_PARISC_PERF_EVENT_H
 
-/* parisc only supports software events through this interface. */
-static inline void set_perf_event_pending(void) { }
+/* Empty, just to avoid compiling error */
 
 #endif /* __ASM_PARISC_PERF_EVENT_H */

arch/powerpc/Kconfig

@@ -138,6 +138,7 @@ config PPC
 	select HAVE_OPROFILE
 	select HAVE_SYSCALL_WRAPPERS if PPC64
 	select GENERIC_ATOMIC64 if PPC32
+	select HAVE_IRQ_WORK
 	select HAVE_PERF_EVENTS
 	select HAVE_REGS_AND_STACK_ACCESS_API
 	select HAVE_HW_BREAKPOINT if PERF_EVENTS && PPC_BOOK3S_64

arch/powerpc/include/asm/paca.h

@@ -129,7 +129,7 @@ struct paca_struct {
 	u8 soft_enabled;		/* irq soft-enable flag */
 	u8 hard_enabled;		/* set if irqs are enabled in MSR */
 	u8 io_sync;			/* writel() needs spin_unlock sync */
-	u8 perf_event_pending;		/* PM interrupt while soft-disabled */
+	u8 irq_work_pending;		/* IRQ_WORK interrupt while soft-disable */
 
 	/* Stuff for accurate time accounting */
 	u64 user_time;			/* accumulated usermode TB ticks */

arch/powerpc/kernel/time.c

@@ -53,7 +53,7 @@
 #include <linux/posix-timers.h>
 #include <linux/irq.h>
 #include <linux/delay.h>
-#include <linux/perf_event.h>
+#include <linux/irq_work.h>
 #include <asm/trace.h>
 
 #include <asm/io.h>
@@ -493,60 +493,60 @@ void __init iSeries_time_init_early(void)
 }
 #endif /* CONFIG_PPC_ISERIES */
 
-#ifdef CONFIG_PERF_EVENTS
+#ifdef CONFIG_IRQ_WORK
 
 /*
  * 64-bit uses a byte in the PACA, 32-bit uses a per-cpu variable...
  */
 #ifdef CONFIG_PPC64
-static inline unsigned long test_perf_event_pending(void)
+static inline unsigned long test_irq_work_pending(void)
 {
 	unsigned long x;
 
 	asm volatile("lbz %0,%1(13)"
 		: "=r" (x)
-		: "i" (offsetof(struct paca_struct, perf_event_pending)));
+		: "i" (offsetof(struct paca_struct, irq_work_pending)));
 	return x;
 }
 
-static inline void set_perf_event_pending_flag(void)
+static inline void set_irq_work_pending_flag(void)
 {
 	asm volatile("stb %0,%1(13)" : :
 		"r" (1),
-		"i" (offsetof(struct paca_struct, perf_event_pending)));
+		"i" (offsetof(struct paca_struct, irq_work_pending)));
 }
 
-static inline void clear_perf_event_pending(void)
+static inline void clear_irq_work_pending(void)
 {
 	asm volatile("stb %0,%1(13)" : :
 		"r" (0),
-		"i" (offsetof(struct paca_struct, perf_event_pending)));
+		"i" (offsetof(struct paca_struct, irq_work_pending)));
 }
 
 #else /* 32-bit */
 
-DEFINE_PER_CPU(u8, perf_event_pending);
+DEFINE_PER_CPU(u8, irq_work_pending);
 
-#define set_perf_event_pending_flag()	__get_cpu_var(perf_event_pending) = 1
-#define test_perf_event_pending()	__get_cpu_var(perf_event_pending)
-#define clear_perf_event_pending()	__get_cpu_var(perf_event_pending) = 0
+#define set_irq_work_pending_flag()	__get_cpu_var(irq_work_pending) = 1
+#define test_irq_work_pending()		__get_cpu_var(irq_work_pending)
+#define clear_irq_work_pending()	__get_cpu_var(irq_work_pending) = 0
 
 #endif /* 32 vs 64 bit */
 
-void set_perf_event_pending(void)
+void set_irq_work_pending(void)
 {
 	preempt_disable();
-	set_perf_event_pending_flag();
+	set_irq_work_pending_flag();
 	set_dec(1);
 	preempt_enable();
 }
 
-#else  /* CONFIG_PERF_EVENTS */
+#else  /* CONFIG_IRQ_WORK */
 
-#define test_perf_event_pending()	0
-#define clear_perf_event_pending()
+#define test_irq_work_pending()	0
+#define clear_irq_work_pending()
 
-#endif /* CONFIG_PERF_EVENTS */
+#endif /* CONFIG_IRQ_WORK */
 
 /*
  * For iSeries shared processors, we have to let the hypervisor
@@ -587,9 +587,9 @@ void timer_interrupt(struct pt_regs * regs)
 
 	calculate_steal_time();
 
-	if (test_perf_event_pending()) {
-		clear_perf_event_pending();
-		perf_event_do_pending();
+	if (test_irq_work_pending()) {
+		clear_irq_work_pending();
+		irq_work_run();
 	}
 
 #ifdef CONFIG_PPC_ISERIES

arch/s390/Kconfig

@@ -95,6 +95,7 @@ config S390
 	select HAVE_KVM if 64BIT
 	select HAVE_ARCH_TRACEHOOK
 	select INIT_ALL_POSSIBLE
+	select HAVE_IRQ_WORK
 	select HAVE_PERF_EVENTS
 	select HAVE_KERNEL_GZIP
 	select HAVE_KERNEL_BZIP2

arch/s390/include/asm/perf_event.h

@@ -4,7 +4,6 @@
  * Copyright 2009 Martin Schwidefsky, IBM Corporation.
  */
 
-static inline void set_perf_event_pending(void) {}
-static inline void clear_perf_event_pending(void) {}
+/* Empty, just to avoid compiling error */
 
 #define PERF_EVENT_INDEX_OFFSET 0

arch/sh/Kconfig

@@ -16,6 +16,7 @@ config SUPERH
 	select HAVE_ARCH_TRACEHOOK
 	select HAVE_DMA_API_DEBUG
 	select HAVE_DMA_ATTRS
+	select HAVE_IRQ_WORK
 	select HAVE_PERF_EVENTS
 	select PERF_USE_VMALLOC
 	select HAVE_KERNEL_GZIP

arch/sh/include/asm/perf_event.h

@@ -26,11 +26,4 @@ extern int register_sh_pmu(struct sh_pmu *);
 extern int reserve_pmc_hardware(void);
 extern void release_pmc_hardware(void);
 
-static inline void set_perf_event_pending(void)
-{
-	/* Nothing to see here, move along. */
-}
-
-#define PERF_EVENT_INDEX_OFFSET	0
-
 #endif /* __ASM_SH_PERF_EVENT_H */

arch/sparc/Kconfig

@@ -26,6 +26,7 @@ config SPARC
 	select ARCH_WANT_OPTIONAL_GPIOLIB
 	select RTC_CLASS
 	select RTC_DRV_M48T59
+	select HAVE_IRQ_WORK
 	select HAVE_PERF_EVENTS
 	select PERF_USE_VMALLOC
 	select HAVE_DMA_ATTRS
@@ -54,6 +55,7 @@ config SPARC64
 	select RTC_DRV_BQ4802
 	select RTC_DRV_SUN4V
 	select RTC_DRV_STARFIRE
+	select HAVE_IRQ_WORK
 	select HAVE_PERF_EVENTS
 	select PERF_USE_VMALLOC
 

arch/sparc/include/asm/perf_event.h

 #ifndef __ASM_SPARC_PERF_EVENT_H
 #define __ASM_SPARC_PERF_EVENT_H
 
-extern void set_perf_event_pending(void);
-
-#define	PERF_EVENT_INDEX_OFFSET	0
-
 #ifdef CONFIG_PERF_EVENTS
 #include <asm/ptrace.h>