On Thu 27-05-10 20:25:22, Dmitry Monakhov wrote:
> I've finally port my tests to common platform with respect to previous
> comments. Pathes agains Jen's xfstest tree
>
> *Tests state*
> I've executed it on ext4 on current ext4.git/next snapshot and
> with http://patchwork.ozlabs.org/patch/53720/
>
> 0001) Is now succeed finally.(i've found more than 5 bugs with that testcase)
>
> 0002) Is failed by now:
> Sometimes it failed due to quota leak (dont know where yet)
I sometimes observe that quota accounting is wrong on ext4 in my fsstress
test as well...
> Sometimes kernel hungs with following call trace:
> sys_quotactl
> ->do_quotactl
> ->vfs_quota_off
> ->vfs_quota_disable
> ->invalidate_dquots()
> /* Wait here for a dqput */
> Seems what we have dquot leakage.
> Still digging...
Hmm, interesting. I don't have a clue where that could happen.
Honza
--
Jan Kara <jack@xxxxxxx>
SUSE Labs, CR
|