[patch] oprofile: fix memory ordering

[patch] oprofile: fix memory ordering

[Nick Piggin]

Regular bitops don't work as locks on all architectures.

Also: can use non-atomic unlock as no concurrent stores to the word.

Signed-off-by: Nick Piggin <npiggin@suse.de>
---
Index: linux-2.6/drivers/oprofile/event_buffer.c
===================================================================
--- linux-2.6.orig/drivers/oprofile/event_buffer.c
+++ linux-2.6/drivers/oprofile/event_buffer.c
@@ -105,7 +105,7 @@ static int event_buffer_open(struct inod
 	if (!capable(CAP_SYS_ADMIN))
 		return -EPERM;
 
-	if (test_and_set_bit(0, &buffer_opened))
+	if (test_and_set_bit_lock(0, &buffer_opened))
 		return -EBUSY;
 
 	/* Register as a user of dcookies
@@ -129,7 +129,7 @@ static int event_buffer_open(struct inod
 fail:
 	dcookie_unregister(file->private_data);
 out:
-	clear_bit(0, &buffer_opened);
+	__clear_bit_unlock(0, &buffer_opened);
 	return err;
 }
 
@@ -141,7 +141,7 @@ static int event_buffer_release(struct i
 	dcookie_unregister(file->private_data);
 	buffer_pos = 0;
 	atomic_set(&buffer_ready, 0);
-	clear_bit(0, &buffer_opened);
+	__clear_bit_unlock(0, &buffer_opened);
 	return 0;
 }
 

Re: [patch] oprofile: fix memory ordering

[Robert Richter]

On 23.10.08 16:25:54, Nick Piggin wrote:
> Regular bitops don't work as locks on all architectures.
> 
> Also: can use non-atomic unlock as no concurrent stores to the word.
> 
> Signed-off-by: Nick Piggin <npiggin@suse.de>

I applied your patch to oprofile/oprofile-for-tip. Thanks Nick.

-Robert

-- 
Advanced Micro Devices, Inc.
Operating System Research Center
email: robert.richter@amd.com