[PATCH v2.6.36-rc7] oprofile: don't use flush_scheduled_work()

[PATCH v2.6.36-rc7] oprofile: don't use flush_scheduled_work()

[Tejun Heo]

flush_scheduled_work() is deprecated and scheduled to be removed.
sync_stop() currently cancels cpu_buffer works inside buffer_mutex and
flushes the system workqueue outside.  Instead, split end_cpu_work()
into two parts - stopping further work enqueues and flushing works -
and do the former inside buffer_mutex and latter outside.

Signed-off-by: Tejun Heo <tj@kernel.org>
---
 drivers/oprofile/buffer_sync.c |    2 +-
 drivers/oprofile/cpu_buffer.c  |   10 +++++++---
 drivers/oprofile/cpu_buffer.h  |    1 +
 3 files changed, 9 insertions(+), 4 deletions(-)

Index: work/drivers/oprofile/buffer_sync.c
===================================================================
--- work.orig/drivers/oprofile/buffer_sync.c
+++ work/drivers/oprofile/buffer_sync.c
@@ -190,7 +190,7 @@ void sync_stop(void)
 	profile_event_unregister(PROFILE_TASK_EXIT, &task_exit_nb);
 	task_handoff_unregister(&task_free_nb);
 	mutex_unlock(&buffer_mutex);
-	flush_scheduled_work();
+	flush_cpu_work();

 	/* make sure we don't leak task structs */
 	process_task_mortuary();
Index: work/drivers/oprofile/cpu_buffer.c
===================================================================
--- work.orig/drivers/oprofile/cpu_buffer.c
+++ work/drivers/oprofile/cpu_buffer.c
@@ -111,14 +111,18 @@ void start_cpu_work(void)

 void end_cpu_work(void)
 {
-	int i;
-
 	work_enabled = 0;
+}
+
+void flush_cpu_work(void)
+{
+	int i;

 	for_each_online_cpu(i) {
 		struct oprofile_cpu_buffer *b = &per_cpu(op_cpu_buffer, i);

-		cancel_delayed_work(&b->work);
+		/* these works are per-cpu, no need for flush_sync */
+		flush_delayed_work(&b->work);
 	}
 }

Index: work/drivers/oprofile/cpu_buffer.h
===================================================================
--- work.orig/drivers/oprofile/cpu_buffer.h
+++ work/drivers/oprofile/cpu_buffer.h
@@ -25,6 +25,7 @@ void free_cpu_buffers(void);

 void start_cpu_work(void);
 void end_cpu_work(void);
+void flush_cpu_work(void);

 /* CPU buffer is composed of such entries (which are
  * also used for context switch notes)

Re: [PATCH v2.6.36-rc7] oprofile: don't use flush_scheduled_work()

[Tejun Heo]

On 10/15/2010 03:51 PM, Tejun Heo wrote:
> flush_scheduled_work() is deprecated and scheduled to be removed.
> sync_stop() currently cancels cpu_buffer works inside buffer_mutex and
> flushes the system workqueue outside.  Instead, split end_cpu_work()
> into two parts - stopping further work enqueues and flushing works -
> and do the former inside buffer_mutex and latter outside.
> 
> Signed-off-by: Tejun Heo <tj@kernel.org>

Ooh, BTW, the global symbols in cpu_buffer seem a bit too generic.  I
don't think oprofile should be defining alloc/free_cpu_buffers() and
start/end/flush_cpu_work().  Maybe it would be a good idea to add op_
prefix to them?

Thanks.

-- 
tejun

Re: [PATCH v2.6.36-rc7] oprofile: don't use flush_scheduled_work()

[Robert Richter]

On 15.10.10 09:51:08, Tejun Heo wrote:
> flush_scheduled_work() is deprecated and scheduled to be removed.
> sync_stop() currently cancels cpu_buffer works inside buffer_mutex and
> flushes the system workqueue outside.  Instead, split end_cpu_work()
> into two parts - stopping further work enqueues and flushing works -
> and do the former inside buffer_mutex and latter outside.
> 
> Signed-off-by: Tejun Heo <tj@kernel.org>

Patch applied to oprofile/urgent. Thanks Tejun.

-Robert

-- 
Advanced Micro Devices, Inc.
Operating System Research Center