Re: [PATCH] blk queue io tracing support

[Nathan Scott <nathans@sgi.com>]

Hi Jens,

On Wed, Aug 24, 2005 at 11:28:39AM +0200, Jens Axboe wrote:
> Patch attached is against 2.6.13-rc6-mm2. Still a good idea to apply the
> relayfs read update from the previous mail [*] as well.

There's a small memory leak there on one of the start-tracing
error paths (relay_open failure)... this should plug it up.

cheers.

-- 
Nathan


Index: 2.6.x-xfs/drivers/block/blktrace.c
===================================================================
--- 2.6.x-xfs.orig/drivers/block/blktrace.c
+++ 2.6.x-xfs/drivers/block/blktrace.c
@@ -73,9 +73,9 @@ int blk_start_trace(struct block_device 
 {
 	request_queue_t *q = bdev_get_queue(bdev);
 	struct blk_user_trace_setup buts;
-	struct blk_trace *bt;
+	struct blk_trace *bt = NULL;
 	char b[BDEVNAME_SIZE];
-	int ret = 0;
+	int ret;
 
 	if (!q)
 		return -ENXIO;
@@ -116,9 +116,14 @@ int blk_start_trace(struct block_device 
 	spin_lock_irq(q->queue_lock);
 	q->blk_trace = bt;
 	spin_unlock_irq(q->queue_lock);
-	ret = 0;
+
+	up(&bdev->bd_sem);
+	return 0;
+
 err:
 	up(&bdev->bd_sem);
+	if (bt)
+		kfree(bt);
 	return ret;
 }