* [PATCH] blk-cgroup: fix wrong unlock order when parse per blkg config
@ 2017-04-24 10:52 Joseph Qi
[not found] ` <be8d9709-84fe-d2df-f294-a12701675e27-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
0 siblings, 1 reply; 4+ messages in thread
From: Joseph Qi @ 2017-04-24 10:52 UTC (permalink / raw)
To: tj-DgEjT+Ai2ygdnm+yROfE0A
Cc: cgroups-u79uwXL29TY76Z2rM5mHXA, wenqing.lz-3b8fjiQLQpfQT0dZR+AlfA,
boyu.mt-3b8fjiQLQpfQT0dZR+AlfA, qijiang.qj-gPhfCIXyaqCqndwCJWfcng
From: Joseph Qi <qijiang.qj-gPhfCIXyaqCqndwCJWfcng@public.gmane.org>
In case of error blkg in blkg_conf_prep, we should unlock
queue->queue_lock first and then rcu, just like blkg_conf_finish.
Signed-off-by: Joseph Qi <qijiang.qj-gPhfCIXyaqCqndwCJWfcng@public.gmane.org>
Signed-off-by: Joseph Qi <jiangqi903-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
---
block/blk-cgroup.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index bbe7ee0..cc017b2 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -823,8 +823,8 @@ int blkg_conf_prep(struct blkcg *blkcg, const struct blkcg_policy *pol,
if (IS_ERR(blkg)) {
ret = PTR_ERR(blkg);
- rcu_read_unlock();
spin_unlock_irq(disk->queue->queue_lock);
+ rcu_read_unlock();
owner = disk->fops->owner;
put_disk(disk);
module_put(owner);
--
1.9.4
^ permalink raw reply related [flat|nested] 4+ messages in thread[parent not found: <be8d9709-84fe-d2df-f294-a12701675e27-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>]
* Re: [PATCH] blk-cgroup: fix wrong unlock order when parse per blkg config [not found] ` <be8d9709-84fe-d2df-f294-a12701675e27-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> @ 2017-04-28 21:22 ` Tejun Heo [not found] ` <20170428212233.GE22354-piEFEHQLUPpN0TnZuCh8vA@public.gmane.org> 0 siblings, 1 reply; 4+ messages in thread From: Tejun Heo @ 2017-04-28 21:22 UTC (permalink / raw) To: Joseph Qi Cc: cgroups-u79uwXL29TY76Z2rM5mHXA, wenqing.lz-3b8fjiQLQpfQT0dZR+AlfA, boyu.mt-3b8fjiQLQpfQT0dZR+AlfA, qijiang.qj-gPhfCIXyaqCqndwCJWfcng Hello, On Mon, Apr 24, 2017 at 06:52:05PM +0800, Joseph Qi wrote: > From: Joseph Qi <qijiang.qj-gPhfCIXyaqCqndwCJWfcng@public.gmane.org> > > In case of error blkg in blkg_conf_prep, we should unlock > queue->queue_lock first and then rcu, just like blkg_conf_finish. I get that this is inconsistent but why is this wrong? If this is for consistency, can you please explain so in the commit so that people don't confuse it for an actual bug fix? Thanks. -- tejun ^ permalink raw reply [flat|nested] 4+ messages in thread
[parent not found: <20170428212233.GE22354-piEFEHQLUPpN0TnZuCh8vA@public.gmane.org>]
* Re: [PATCH] blk-cgroup: fix wrong unlock order when parse per blkg config [not found] ` <20170428212233.GE22354-piEFEHQLUPpN0TnZuCh8vA@public.gmane.org> @ 2017-05-02 0:41 ` Joseph Qi [not found] ` <c39a752d-034e-02a0-3c46-892e42c6616a-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> 0 siblings, 1 reply; 4+ messages in thread From: Joseph Qi @ 2017-05-02 0:41 UTC (permalink / raw) To: Tejun Heo Cc: cgroups-u79uwXL29TY76Z2rM5mHXA, wenqing.lz-3b8fjiQLQpfQT0dZR+AlfA, boyu.mt-3b8fjiQLQpfQT0dZR+AlfA, qijiang.qj-gPhfCIXyaqCqndwCJWfcng Fine, I will explain so and resend it later. Thanks, Joseph On 17/4/29 05:22, Tejun Heo wrote: > Hello, > > On Mon, Apr 24, 2017 at 06:52:05PM +0800, Joseph Qi wrote: >> From: Joseph Qi <qijiang.qj-gPhfCIXyaqCqndwCJWfcng@public.gmane.org> >> >> In case of error blkg in blkg_conf_prep, we should unlock >> queue->queue_lock first and then rcu, just like blkg_conf_finish. > > I get that this is inconsistent but why is this wrong? If this is for > consistency, can you please explain so in the commit so that people > don't confuse it for an actual bug fix? > > Thanks. > ^ permalink raw reply [flat|nested] 4+ messages in thread
[parent not found: <c39a752d-034e-02a0-3c46-892e42c6616a-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>]
* Re: [PATCH] blk-cgroup: fix wrong unlock order when parse per blkg config [not found] ` <c39a752d-034e-02a0-3c46-892e42c6616a-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> @ 2017-05-02 1:10 ` Joseph Qi 0 siblings, 0 replies; 4+ messages in thread From: Joseph Qi @ 2017-05-02 1:10 UTC (permalink / raw) To: Tejun Heo Cc: cgroups-u79uwXL29TY76Z2rM5mHXA, wenqing.lz-3b8fjiQLQpfQT0dZR+AlfA, boyu.mt-3b8fjiQLQpfQT0dZR+AlfA, qijiang.qj-gPhfCIXyaqCqndwCJWfcng Hi Tejun, I've checked the upstream mainline and found that this is already fixed as part of your commit 457e490f2b74 ("blkcg: allocate struct blkcg_gq outside request queue spinlock"). So I do not have to resend it any more, thanks. Thanks, Joseph On 17/5/2 08:41, Joseph Qi wrote: > Fine, I will explain so and resend it later. > > Thanks, > Joseph > > On 17/4/29 05:22, Tejun Heo wrote: >> Hello, >> >> On Mon, Apr 24, 2017 at 06:52:05PM +0800, Joseph Qi wrote: >>> From: Joseph Qi <qijiang.qj-gPhfCIXyaqCqndwCJWfcng@public.gmane.org> >>> >>> In case of error blkg in blkg_conf_prep, we should unlock >>> queue->queue_lock first and then rcu, just like blkg_conf_finish. >> >> I get that this is inconsistent but why is this wrong? If this is for >> consistency, can you please explain so in the commit so that people >> don't confuse it for an actual bug fix? >> >> Thanks. >> ^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2017-05-02 1:10 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2017-04-24 10:52 [PATCH] blk-cgroup: fix wrong unlock order when parse per blkg config Joseph Qi
[not found] ` <be8d9709-84fe-d2df-f294-a12701675e27-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
2017-04-28 21:22 ` Tejun Heo
[not found] ` <20170428212233.GE22354-piEFEHQLUPpN0TnZuCh8vA@public.gmane.org>
2017-05-02 0:41 ` Joseph Qi
[not found] ` <c39a752d-034e-02a0-3c46-892e42c6616a-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
2017-05-02 1:10 ` Joseph Qi
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).