[PATCH 0/5] yet more perf counter patches

[PATCH 0/5] yet more perf counter patches

[Peter Zijlstra]

-- 

[PATCH 1/5] perf_counter: fix perf-$cmd invokation

[Peter Zijlstra]

[-- Attachment #1: perf_counter-fix-perf-cmd.patch --]
[-- Type: text/plain, Size: 735 bytes --]

Fix:

  $ perf-top
  fatal: cannot handle -top internally

LKML-Reference: <new-submission>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
---
 Documentation/perf_counter/perf.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Index: linux-2.6/Documentation/perf_counter/perf.c
===================================================================
--- linux-2.6.orig/Documentation/perf_counter/perf.c
+++ linux-2.6/Documentation/perf_counter/perf.c
@@ -357,7 +357,7 @@ int main(int argc, const char **argv)
 	 * die if that one cannot handle it.
 	 */
 	if (!prefixcmp(cmd, "perf-")) {
-		cmd += 4;
+		cmd += 5;
 		argv[0] = cmd;
 		handle_internal_command(argc, argv);
 		die("cannot handle %s internally", cmd);

-- 

[tip:perfcounters/core] perf_counter: Fix perf-$cmd invokation

[tip-bot for Peter Zijlstra]

Commit-ID:  266dfb0b58bc4181b6158ee63a0069abaa9f3a98
Gitweb:     http://git.kernel.org/tip/266dfb0b58bc4181b6158ee63a0069abaa9f3a98
Author:     Peter Zijlstra <a.p.zijlstra@chello.nl>
AuthorDate: Mon, 25 May 2009 14:45:24 +0200
Committer:  Ingo Molnar <mingo@elte.hu>
CommitDate: Mon, 25 May 2009 14:54:59 +0200

perf_counter: Fix perf-$cmd invokation

Fix:

  $ perf-top
  fatal: cannot handle -top internally

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <20090525124559.995591577@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>


---
 Documentation/perf_counter/perf.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/Documentation/perf_counter/perf.c b/Documentation/perf_counter/perf.c
index 594d270..1d6d7aa 100644
--- a/Documentation/perf_counter/perf.c
+++ b/Documentation/perf_counter/perf.c
@@ -357,7 +357,7 @@ int main(int argc, const char **argv)
 	 * die if that one cannot handle it.
 	 */
 	if (!prefixcmp(cmd, "perf-")) {
-		cmd += 4;
+		cmd += 5;
 		argv[0] = cmd;
 		handle_internal_command(argc, argv);
 		die("cannot handle %s internally", cmd);

[PATCH 2/5] perf_counter: remove unused ABI bits.

[Peter Zijlstra]

[-- Attachment #1: perf_counter-cleanup-abi.patch --]
[-- Type: text/plain, Size: 721 bytes --]

extra_config_len isn't used for anything, remove it.

LKML-Reference: <new-submission>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
---
 include/linux/perf_counter.h |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

Index: linux-2.6/include/linux/perf_counter.h
===================================================================
--- linux-2.6.orig/include/linux/perf_counter.h
+++ linux-2.6/include/linux/perf_counter.h
@@ -154,11 +154,11 @@ struct perf_counter_hw_event {
 
 				__reserved_1   : 51;
 
-	__u32			extra_config_len;
 	__u32			wakeup_events;	/* wakeup every n events */
+	__u32			__reserved_2;
 
-	__u64			__reserved_2;
 	__u64			__reserved_3;
+	__u64			__reserved_4;
 };
 
 /*

-- 

[tip:perfcounters/core] perf_counter: Remove unused ABI bits

[tip-bot for Peter Zijlstra]

Commit-ID:  e527ea312f31e88a7fa5472b71db71c565b0d44f
Gitweb:     http://git.kernel.org/tip/e527ea312f31e88a7fa5472b71db71c565b0d44f
Author:     Peter Zijlstra <a.p.zijlstra@chello.nl>
AuthorDate: Mon, 25 May 2009 14:45:25 +0200
Committer:  Ingo Molnar <mingo@elte.hu>
CommitDate: Mon, 25 May 2009 14:55:00 +0200

perf_counter: Remove unused ABI bits

extra_config_len isn't used for anything, remove it.

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <20090525124600.116035832@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>


---
 include/linux/perf_counter.h |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/include/linux/perf_counter.h b/include/linux/perf_counter.h
index 2ddf5e3..b1f2bac 100644
--- a/include/linux/perf_counter.h
+++ b/include/linux/perf_counter.h
@@ -154,11 +154,11 @@ struct perf_counter_hw_event {
 
 				__reserved_1   : 51;
 
-	__u32			extra_config_len;
 	__u32			wakeup_events;	/* wakeup every n events */
+	__u32			__reserved_2;
 
-	__u64			__reserved_2;
 	__u64			__reserved_3;
+	__u64			__reserved_4;
 };
 
 /*

[PATCH 3/5] perf_counter: make pctrl() affect inherited counters too

[Peter Zijlstra]

[-- Attachment #1: perf_counter-pctrl-child.patch --]
[-- Type: text/plain, Size: 2182 bytes --]

Paul noted that the new ptcrl() didn't work on child counters.

LKML-Reference: <new-submission>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
---
 kernel/perf_counter.c |   48 ++++++++++++++++++++++++------------------------
 1 file changed, 24 insertions(+), 24 deletions(-)

Index: linux-2.6/kernel/perf_counter.c
===================================================================
--- linux-2.6.orig/kernel/perf_counter.c
+++ linux-2.6/kernel/perf_counter.c
@@ -1067,30 +1067,6 @@ static void perf_counter_cpu_sched_in(st
 	__perf_counter_sched_in(ctx, cpuctx, cpu);
 }
 
-int perf_counter_task_enable(void)
-{
-	struct perf_counter *counter;
-
-	mutex_lock(&current->perf_counter_mutex);
-	list_for_each_entry(counter, &current->perf_counter_list, owner_entry)
-		perf_counter_enable(counter);
-	mutex_unlock(&current->perf_counter_mutex);
-
-	return 0;
-}
-
-int perf_counter_task_disable(void)
-{
-	struct perf_counter *counter;
-
-	mutex_lock(&current->perf_counter_mutex);
-	list_for_each_entry(counter, &current->perf_counter_list, owner_entry)
-		perf_counter_disable(counter);
-	mutex_unlock(&current->perf_counter_mutex);
-
-	return 0;
-}
-
 static void perf_log_period(struct perf_counter *counter, u64 period);
 
 static void perf_adjust_freq(struct perf_counter_context *ctx)
@@ -1505,6 +1481,30 @@ static long perf_ioctl(struct file *file
 	return 0;
 }
 
+int perf_counter_task_enable(void)
+{
+	struct perf_counter *counter;
+
+	mutex_lock(&current->perf_counter_mutex);
+	list_for_each_entry(counter, &current->perf_counter_list, owner_entry)
+		perf_counter_for_each_child(counter, perf_counter_enable);
+	mutex_unlock(&current->perf_counter_mutex);
+
+	return 0;
+}
+
+int perf_counter_task_disable(void)
+{
+	struct perf_counter *counter;
+
+	mutex_lock(&current->perf_counter_mutex);
+	list_for_each_entry(counter, &current->perf_counter_list, owner_entry)
+		perf_counter_for_each_child(counter, perf_counter_disable);
+	mutex_unlock(&current->perf_counter_mutex);
+
+	return 0;
+}
+
 /*
  * Callers need to ensure there can be no nesting of this function, otherwise
  * the seqlock logic goes bad. We can not serialize this because the arch

-- 

[tip:perfcounters/core] perf_counter: Make pctrl() affect inherited counters too

[tip-bot for Peter Zijlstra]

Commit-ID:  771d7cde144d87f2d1fbee4da3c6234d61f7e42a
Gitweb:     http://git.kernel.org/tip/771d7cde144d87f2d1fbee4da3c6234d61f7e42a
Author:     Peter Zijlstra <a.p.zijlstra@chello.nl>
AuthorDate: Mon, 25 May 2009 14:45:26 +0200
Committer:  Ingo Molnar <mingo@elte.hu>
CommitDate: Mon, 25 May 2009 14:55:00 +0200

perf_counter: Make pctrl() affect inherited counters too

Paul noted that the new ptcrl() didn't work on child counters.

Reported-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <20090525124600.203151469@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>


---
 kernel/perf_counter.c |   48 ++++++++++++++++++++++++------------------------
 1 files changed, 24 insertions(+), 24 deletions(-)

diff --git a/kernel/perf_counter.c b/kernel/perf_counter.c
index 6cdf824..217dbcc 100644
--- a/kernel/perf_counter.c
+++ b/kernel/perf_counter.c
@@ -1067,30 +1067,6 @@ static void perf_counter_cpu_sched_in(struct perf_cpu_context *cpuctx, int cpu)
 	__perf_counter_sched_in(ctx, cpuctx, cpu);
 }
 
-int perf_counter_task_enable(void)
-{
-	struct perf_counter *counter;
-
-	mutex_lock(&current->perf_counter_mutex);
-	list_for_each_entry(counter, &current->perf_counter_list, owner_entry)
-		perf_counter_enable(counter);
-	mutex_unlock(&current->perf_counter_mutex);
-
-	return 0;
-}
-
-int perf_counter_task_disable(void)
-{
-	struct perf_counter *counter;
-
-	mutex_lock(&current->perf_counter_mutex);
-	list_for_each_entry(counter, &current->perf_counter_list, owner_entry)
-		perf_counter_disable(counter);
-	mutex_unlock(&current->perf_counter_mutex);
-
-	return 0;
-}
-
 static void perf_log_period(struct perf_counter *counter, u64 period);
 
 static void perf_adjust_freq(struct perf_counter_context *ctx)
@@ -1505,6 +1481,30 @@ static long perf_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
 	return 0;
 }
 
+int perf_counter_task_enable(void)
+{
+	struct perf_counter *counter;
+
+	mutex_lock(&current->perf_counter_mutex);
+	list_for_each_entry(counter, &current->perf_counter_list, owner_entry)
+		perf_counter_for_each_child(counter, perf_counter_enable);
+	mutex_unlock(&current->perf_counter_mutex);
+
+	return 0;
+}
+
+int perf_counter_task_disable(void)
+{
+	struct perf_counter *counter;
+
+	mutex_lock(&current->perf_counter_mutex);
+	list_for_each_entry(counter, &current->perf_counter_list, owner_entry)
+		perf_counter_for_each_child(counter, perf_counter_disable);
+	mutex_unlock(&current->perf_counter_mutex);
+
+	return 0;
+}
+
 /*
  * Callers need to ensure there can be no nesting of this function, otherwise
  * the seqlock logic goes bad. We can not serialize this because the arch

[PATCH 4/5] perf_counter: propagate inheritance failures down the fork() path

[Peter Zijlstra]

[-- Attachment #1: perf_counter-init-fail.patch --]
[-- Type: text/plain, Size: 2732 bytes --]

Fail fork() when we fail inheritance for some reason (-ENOMEM most likely).

LKML-Reference: <new-submission>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
---
 kernel/fork.c         |    5 ++++-
 kernel/perf_counter.c |   20 ++++++++++++--------
 2 files changed, 16 insertions(+), 9 deletions(-)

Index: linux-2.6/kernel/fork.c
===================================================================
--- linux-2.6.orig/kernel/fork.c
+++ linux-2.6/kernel/fork.c
@@ -983,7 +983,9 @@ static struct task_struct *copy_process(
 		goto fork_out;
 
 	rt_mutex_init_task(p);
-	perf_counter_init_task(p);
+	retval = perf_counter_init_task(p);
+	if (retval)
+		goto bad_fork_free;
 
 #ifdef CONFIG_PROVE_LOCKING
 	DEBUG_LOCKS_WARN_ON(!p->hardirqs_enabled);
@@ -1309,6 +1311,7 @@ bad_fork_cleanup_count:
 	put_cred(p->real_cred);
 	put_cred(p->cred);
 bad_fork_free:
+	perf_counter_exit_task(p);
 	free_task(p);
 fork_out:
 	return ERR_PTR(retval);
Index: linux-2.6/kernel/perf_counter.c
===================================================================
--- linux-2.6.orig/kernel/perf_counter.c
+++ linux-2.6/kernel/perf_counter.c
@@ -3434,18 +3434,23 @@ again:
 /*
  * Initialize the perf_counter context in task_struct
  */
-void perf_counter_init_task(struct task_struct *child)
+int perf_counter_init_task(struct task_struct *child)
 {
 	struct perf_counter_context *child_ctx, *parent_ctx;
 	struct perf_counter *counter;
 	struct task_struct *parent = current;
 	int inherited_all = 1;
+	int ret = 0;
 
 	child->perf_counter_ctxp = NULL;
 
 	mutex_init(&child->perf_counter_mutex);
 	INIT_LIST_HEAD(&child->perf_counter_list);
 
+	parent_ctx = parent->perf_counter_ctxp;
+	if (likely(!parent_ctx || !parent_ctx->nr_counters))
+		return 0;
+
 	/*
 	 * This is executed from the parent task context, so inherit
 	 * counters that have been marked for cloning.
@@ -3454,11 +3459,7 @@ void perf_counter_init_task(struct task_
 
 	child_ctx = kmalloc(sizeof(struct perf_counter_context), GFP_KERNEL);
 	if (!child_ctx)
-		return;
-
-	parent_ctx = parent->perf_counter_ctxp;
-	if (likely(!parent_ctx || !parent_ctx->nr_counters))
-		return;
+		return -ENOMEM;
 
 	__perf_counter_init_context(child_ctx, child);
 	child->perf_counter_ctxp = child_ctx;
@@ -3482,8 +3483,9 @@ void perf_counter_init_task(struct task_
 			continue;
 		}
 
-		if (inherit_group(counter, parent,
-				  parent_ctx, child, child_ctx)) {
+		ret = inherit_group(counter, parent, parent_ctx,
+					     child, child_ctx);
+		if (ret) {
 			inherited_all = 0;
 			break;
 		}
@@ -3505,6 +3507,8 @@ void perf_counter_init_task(struct task_
 	}
 
 	mutex_unlock(&parent_ctx->mutex);
+
+	return ret;
 }
 
 static void __cpuinit perf_counter_init_cpu(int cpu)

-- 

[tip:perfcounters/core] perf_counter: Propagate inheritance failures down the fork() path

[tip-bot for Peter Zijlstra]

Commit-ID:  6ab423e0eaca827fbd201ca4ae7d4f8573a366b2
Gitweb:     http://git.kernel.org/tip/6ab423e0eaca827fbd201ca4ae7d4f8573a366b2
Author:     Peter Zijlstra <a.p.zijlstra@chello.nl>
AuthorDate: Mon, 25 May 2009 14:45:27 +0200
Committer:  Ingo Molnar <mingo@elte.hu>
CommitDate: Mon, 25 May 2009 14:55:01 +0200

perf_counter: Propagate inheritance failures down the fork() path

Fail fork() when we fail inheritance for some reason (-ENOMEM most likely).

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <20090525124600.324656474@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>


---
 include/linux/perf_counter.h |    4 ++--
 kernel/fork.c                |    6 +++++-
 kernel/perf_counter.c        |   20 ++++++++++++--------
 3 files changed, 19 insertions(+), 11 deletions(-)

diff --git a/include/linux/perf_counter.h b/include/linux/perf_counter.h
index b1f2bac..d3e85de 100644
--- a/include/linux/perf_counter.h
+++ b/include/linux/perf_counter.h
@@ -566,7 +566,7 @@ extern void perf_counter_task_sched_in(struct task_struct *task, int cpu);
 extern void perf_counter_task_sched_out(struct task_struct *task,
 					struct task_struct *next, int cpu);
 extern void perf_counter_task_tick(struct task_struct *task, int cpu);
-extern void perf_counter_init_task(struct task_struct *child);
+extern int perf_counter_init_task(struct task_struct *child);
 extern void perf_counter_exit_task(struct task_struct *child);
 extern void perf_counter_do_pending(void);
 extern void perf_counter_print_debug(void);
@@ -631,7 +631,7 @@ perf_counter_task_sched_out(struct task_struct *task,
 			    struct task_struct *next, int cpu)		{ }
 static inline void
 perf_counter_task_tick(struct task_struct *task, int cpu)		{ }
-static inline void perf_counter_init_task(struct task_struct *child)	{ }
+static inline int perf_counter_init_task(struct task_struct *child)	{ }
 static inline void perf_counter_exit_task(struct task_struct *child)	{ }
 static inline void perf_counter_do_pending(void)			{ }
 static inline void perf_counter_print_debug(void)			{ }
diff --git a/kernel/fork.c b/kernel/fork.c
index 675e01e..c07c333 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -1095,7 +1095,10 @@ static struct task_struct *copy_process(unsigned long clone_flags,
 
 	/* Perform scheduler related setup. Assign this task to a CPU. */
 	sched_fork(p, clone_flags);
-	perf_counter_init_task(p);
+
+	retval = perf_counter_init_task(p);
+	if (retval)
+		goto bad_fork_cleanup_policy;
 
 	if ((retval = audit_alloc(p)))
 		goto bad_fork_cleanup_policy;
@@ -1295,6 +1298,7 @@ bad_fork_cleanup_semundo:
 bad_fork_cleanup_audit:
 	audit_free(p);
 bad_fork_cleanup_policy:
+	perf_counter_exit_task(p);
 #ifdef CONFIG_NUMA
 	mpol_put(p->mempolicy);
 bad_fork_cleanup_cgroup:
diff --git a/kernel/perf_counter.c b/kernel/perf_counter.c
index 217dbcc..7a7a144 100644
--- a/kernel/perf_counter.c
+++ b/kernel/perf_counter.c
@@ -3434,18 +3434,23 @@ again:
 /*
  * Initialize the perf_counter context in task_struct
  */
-void perf_counter_init_task(struct task_struct *child)
+int perf_counter_init_task(struct task_struct *child)
 {
 	struct perf_counter_context *child_ctx, *parent_ctx;
 	struct perf_counter *counter;
 	struct task_struct *parent = current;
 	int inherited_all = 1;
+	int ret = 0;
 
 	child->perf_counter_ctxp = NULL;
 
 	mutex_init(&child->perf_counter_mutex);
 	INIT_LIST_HEAD(&child->perf_counter_list);
 
+	parent_ctx = parent->perf_counter_ctxp;
+	if (likely(!parent_ctx || !parent_ctx->nr_counters))
+		return 0;
+
 	/*
 	 * This is executed from the parent task context, so inherit
 	 * counters that have been marked for cloning.
@@ -3454,11 +3459,7 @@ void perf_counter_init_task(struct task_struct *child)
 
 	child_ctx = kmalloc(sizeof(struct perf_counter_context), GFP_KERNEL);
 	if (!child_ctx)
-		return;
-
-	parent_ctx = parent->perf_counter_ctxp;
-	if (likely(!parent_ctx || !parent_ctx->nr_counters))
-		return;
+		return -ENOMEM;
 
 	__perf_counter_init_context(child_ctx, child);
 	child->perf_counter_ctxp = child_ctx;
@@ -3482,8 +3483,9 @@ void perf_counter_init_task(struct task_struct *child)
 			continue;
 		}
 
-		if (inherit_group(counter, parent,
-				  parent_ctx, child, child_ctx)) {
+		ret = inherit_group(counter, parent, parent_ctx,
+					     child, child_ctx);
+		if (ret) {
 			inherited_all = 0;
 			break;
 		}
@@ -3505,6 +3507,8 @@ void perf_counter_init_task(struct task_struct *child)
 	}
 
 	mutex_unlock(&parent_ctx->mutex);
+
+	return ret;
 }
 
 static void __cpuinit perf_counter_init_cpu(int cpu)

[PATCH 5/5] perf_counter: fix PERF_COUNTER_CONTEXT_SWITCHES for cpu counters

[Peter Zijlstra]

[-- Attachment #1: perf_counter-fix-swcounter-context-switches.patch --]
[-- Type: text/plain, Size: 1017 bytes --]

Ingo noticed that cpu counters had 0 context switches, even though there
was plenty scheduling on the cpu.

LKML-Reference: <new-submission>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
---
 kernel/perf_counter.c |    7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

Index: linux-2.6/kernel/perf_counter.c
===================================================================
--- linux-2.6.orig/kernel/perf_counter.c
+++ linux-2.6/kernel/perf_counter.c
@@ -932,14 +932,13 @@ void perf_counter_task_sched_out(struct 
 	struct perf_counter_context *next_ctx;
 	struct pt_regs *regs;
 
+	regs = task_pt_regs(task);
+	perf_swcounter_event(PERF_COUNT_CONTEXT_SWITCHES, 1, 1, regs, 0);
+
 	if (likely(!ctx || !cpuctx->task_ctx))
 		return;
 
 	update_context_time(ctx);
-
-	regs = task_pt_regs(task);
-	perf_swcounter_event(PERF_COUNT_CONTEXT_SWITCHES, 1, 1, regs, 0);
-
 	next_ctx = next->perf_counter_ctxp;
 	if (next_ctx && context_equiv(ctx, next_ctx)) {
 		task->perf_counter_ctxp = next_ctx;

-- 

[tip:perfcounters/core] perf_counter: Fix PERF_COUNTER_CONTEXT_SWITCHES for cpu counters

[tip-bot for Peter Zijlstra]

Commit-ID:  10989fb2451763fae6f42d85fa6106c8fd010cf5
Gitweb:     http://git.kernel.org/tip/10989fb2451763fae6f42d85fa6106c8fd010cf5
Author:     Peter Zijlstra <a.p.zijlstra@chello.nl>
AuthorDate: Mon, 25 May 2009 14:45:28 +0200
Committer:  Ingo Molnar <mingo@elte.hu>
CommitDate: Mon, 25 May 2009 14:55:01 +0200

perf_counter: Fix PERF_COUNTER_CONTEXT_SWITCHES for cpu counters

Ingo noticed that cpu counters had 0 context switches, even though
there was plenty scheduling on the cpu.

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <20090525124600.419025548@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>


---
 kernel/perf_counter.c |    7 +++----
 1 files changed, 3 insertions(+), 4 deletions(-)

diff --git a/kernel/perf_counter.c b/kernel/perf_counter.c
index 7a7a144..14b1fe9 100644
--- a/kernel/perf_counter.c
+++ b/kernel/perf_counter.c
@@ -924,14 +924,13 @@ void perf_counter_task_sched_out(struct task_struct *task,
 	struct perf_counter_context *next_ctx;
 	struct pt_regs *regs;
 
+	regs = task_pt_regs(task);
+	perf_swcounter_event(PERF_COUNT_CONTEXT_SWITCHES, 1, 1, regs, 0);
+
 	if (likely(!ctx || !cpuctx->task_ctx))
 		return;
 
 	update_context_time(ctx);
-
-	regs = task_pt_regs(task);
-	perf_swcounter_event(PERF_COUNT_CONTEXT_SWITCHES, 1, 1, regs, 0);
-
 	next_ctx = next->perf_counter_ctxp;
 	if (next_ctx && context_equiv(ctx, next_ctx)) {
 		task->perf_counter_ctxp = next_ctx;