[PATCH] quota: Fix possible infinite loop in quota code

[PATCH] quota: Fix possible infinite loop in quota code

[Jan Kara]

When quota structure is going to be dropped and it is dirty, quota code
tries to write it. If the write fails for some reason (e. g. transaction
cannot be started because the journal is aborted), we try writing again
and again and again... Fix the problem by clearing the dirty bit even if
the write failed.

Signed-off-by: Jan Kara <jack@suse.cz>
Reported-by: dingdinghua <dingdinghua85@gmail.com>
---
 fs/dquot.c |   16 +++++++++++++++-
 1 files changed, 15 insertions(+), 1 deletions(-)

diff --git a/fs/dquot.c b/fs/dquot.c
index 5ac77da..ad88cf6 100644
--- a/fs/dquot.c
+++ b/fs/dquot.c
@@ -562,6 +562,8 @@ static struct shrinker dqcache_shrinker = {
  */
 static void dqput(struct dquot *dquot)
 {
+	int ret;
+
 	if (!dquot)
 		return;
 #ifdef __DQUOT_PARANOIA
@@ -594,7 +596,19 @@ we_slept:
 	if (test_bit(DQ_ACTIVE_B, &dquot->dq_flags) && dquot_dirty(dquot)) {
 		spin_unlock(&dq_list_lock);
 		/* Commit dquot before releasing */
-		dquot->dq_sb->dq_op->write_dquot(dquot);
+		ret = dquot->dq_sb->dq_op->write_dquot(dquot);
+		if (ret < 0) {
+			printk(KERN_ERR "VFS: cannot write quota structure on "
+				"device %s (error %d). Quota may get out of "
+				"sync!\n", dquot->dq_sb->s_id, ret);
+			/*
+			 * We clear dirty bit anyway, so that we avoid
+			 * infinite loop here
+			 */
+			spin_lock(&dq_list_lock);
+			clear_dquot_dirty(dquot);
+			spin_unlock(&dq_list_lock);
+		}
 		goto we_slept;
 	}
 	/* Clear flag in case dquot was inactive (something bad happened) */
-- 
1.5.2.4

Re: [PATCH] quota: Fix possible infinite loop in quota code

[Andrew Morton]

On Wed,  9 Jul 2008 19:30:41 +0200 Jan Kara <jack@suse.cz> wrote:

> When quota structure is going to be dropped and it is dirty, quota code
> tries to write it. If the write fails for some reason (e. g. transaction
> cannot be started because the journal is aborted), we try writing again
> and again and again... Fix the problem by clearing the dirty bit even if
> the write failed.

What do you consider the priority is here?  Needed in 2.6.26?  Backport to
2.6.25.x?  Merge into 2.6.27 with a backport to 2.6.26.x and 2.6.25.x?

Lots of options ;)

Re: [PATCH] quota: Fix possible infinite loop in quota code

[Jan Kara]

On Wed 09-07-08 15:12:54, Andrew Morton wrote:
> On Wed,  9 Jul 2008 19:30:41 +0200 Jan Kara <jack@suse.cz> wrote:
> 
> > When quota structure is going to be dropped and it is dirty, quota code
> > tries to write it. If the write fails for some reason (e. g. transaction
> > cannot be started because the journal is aborted), we try writing again
> > and again and again... Fix the problem by clearing the dirty bit even if
> > the write failed.
> 
> What do you consider the priority is here?  Needed in 2.6.26?  Backport to
> 2.6.25.x?  Merge into 2.6.27 with a backport to 2.6.26.x and 2.6.25.x?
> 
> Lots of options ;)
  Since nobody has really seen it in the wild, I don't think it's urgent.
The bug has been there for 5 years or so without notice... I'd just let it
be included in 2.6.26.x maybe 2.6.25.x and push into 2.6.27.

									Honza
-- 
Jan Kara <jack@suse.cz>
SUSE Labs, CR

Re: [PATCH] quota: Fix possible infinite loop in quota code

[Andrew Morton]

On Thu, 10 Jul 2008 11:05:53 +0200 Jan Kara <jack@suse.cz> wrote:

> On Wed 09-07-08 15:12:54, Andrew Morton wrote:
> > On Wed,  9 Jul 2008 19:30:41 +0200 Jan Kara <jack@suse.cz> wrote:
> > 
> > > When quota structure is going to be dropped and it is dirty, quota code
> > > tries to write it. If the write fails for some reason (e. g. transaction
> > > cannot be started because the journal is aborted), we try writing again
> > > and again and again... Fix the problem by clearing the dirty bit even if
> > > the write failed.
> > 
> > What do you consider the priority is here?  Needed in 2.6.26?  Backport to
> > 2.6.25.x?  Merge into 2.6.27 with a backport to 2.6.26.x and 2.6.25.x?
> > 
> > Lots of options ;)
>   Since nobody has really seen it in the wild, I don't think it's urgent.
> The bug has been there for 5 years or so without notice... I'd just let it
> be included in 2.6.26.x maybe 2.6.25.x and push into 2.6.27.
> 

So noted, thanks.