[PATCH v3.5-rc1] blkcg: fix blkg_alloc() failure path

[PATCH v3.5-rc1] blkcg: fix blkg_alloc() failure path

[Tejun Heo]

When policy data allocation fails in the middle, blkg_alloc() invokes
blkg_free() to destroy the half constructed blkg.  This ends up
calling pd_exit_fn() on policy datas which didn't go through
pd_init_fn().  Fix it by making blkg_alloc() call pd_init_fn()
immediately after each policy data allocation.

Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Vivek Goyal <vgoyal@redhat.com>
---
Jens, this was posted as part of "block: implement per-blkg request
allocation" patchset[1] but should be applied as fix on v3.5-rc1.

[1] http://thread.gmane.org/gmane.linux.kernel.containers/23159/focus=2007

 block/blk-cgroup.c |    6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)

--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -125,12 +125,8 @@ static struct blkcg_gq *blkg_alloc(struc
 
 		blkg->pd[i] = pd;
 		pd->blkg = blkg;
-	}
-
-	/* invoke per-policy init */
-	for (i = 0; i < BLKCG_MAX_POLS; i++) {
-		struct blkcg_policy *pol = blkcg_policy[i];
 
+		/* invoke per-policy init */
 		if (blkcg_policy_enabled(blkg->q, pol))
 			pol->pd_init_fn(blkg);
 	}

Re: [PATCH v3.5-rc1] blkcg: fix blkg_alloc() failure path

[Jens Axboe]

On 06/04/2012 08:21 AM, Tejun Heo wrote:
> When policy data allocation fails in the middle, blkg_alloc() invokes
> blkg_free() to destroy the half constructed blkg.  This ends up
> calling pd_exit_fn() on policy datas which didn't go through
> pd_init_fn().  Fix it by making blkg_alloc() call pd_init_fn()
> immediately after each policy data allocation.
> 
> Signed-off-by: Tejun Heo <tj@kernel.org>
> Acked-by: Vivek Goyal <vgoyal@redhat.com>
> ---
> Jens, this was posted as part of "block: implement per-blkg request
> allocation" patchset[1] but should be applied as fix on v3.5-rc1.

Applied for 3.5 current, thanks.

-- 
Jens Axboe