regressions in xfs/633 and xfs/437?

regressions in xfs/633 and xfs/437?

[Darrick J. Wong]

Hey folks,

Have any of you noticed a recent regression in xfs/437 and xfs/633?

--- /run/fstests/bin/tests/xfs/437.out	2025-07-15 14:41:40.303420629 -0700
+++ /var/tmp/fstests/xfs/437.out.bad	2025-09-21 18:53:36.368250642 -0700
@@ -1,2 +1,3 @@
 QA output created by 437
 Silence is golden.
+mkfs/proto.c:1428:	error = libxfs_iget(mp, NULL, dst_ino, 0, &ip);

--- /var/tmp/fstests/xfs/633.out.bad
+++ /dev/fd/63	2025-09-21 17:45:58.431935255 -0700
@@ -1,6 +1,107468 @@
 QA output created by 633
 Format and populate
 Recover filesystem
+./common/xfs: line 335: 326611 Segmentation fault      (core dumped) $XFS_DB_PROG "$@" $(_scratch_xfs_db_options)
+./common/xfs: line 335: 326617 Segmentation fault      (core dumped) $XFS_DB_PROG "$@" $(_scratch_xfs_db_options)
 Check file contents
+--- /tmp/326192.stat.before	2025-09-21 17:36:10.071959154 -0700
++++ /tmp/326192.stat.after	2025-09-21 17:36:12.863959041 -0700
+@@ -1,23 +1,3 @@
+-11a4 0:0 0 0 fifo 0 1731556303 ./S_IFIFO
+-2000 0:0 0 0 character special file 0 1758501368 ./newfiles/p0/d1/c19
+-2000 0:0 0 0 character special file 0 1758501368 ./newfiles/p0/d1/c1a
<snip>

The xfs/437 failure is trivially fixable, not sure what's causing the
segfault here?  Probably something in the new file_{get,set}attr code in
xfs_db is my guess...?

MKFS_OPTIONS  -- -f -m metadir=1,autofsck=1,uquota,gquota,pquota, -n size=8k, /dev/sdf
MOUNT_OPTIONS -- /dev/sdf /opt

(note that both problems happen on a variety of different
configurations)

--D

Re: regressions in xfs/633 and xfs/437?

[Andrey Albershteyn]

On 2025-09-22 08:29:54, Darrick J. Wong wrote:
> Hey folks,
> 
> Have any of you noticed a recent regression in xfs/437 and xfs/633?
> 
> --- /run/fstests/bin/tests/xfs/437.out	2025-07-15 14:41:40.303420629 -0700
> +++ /var/tmp/fstests/xfs/437.out.bad	2025-09-21 18:53:36.368250642 -0700
> @@ -1,2 +1,3 @@
>  QA output created by 437
>  Silence is golden.
> +mkfs/proto.c:1428:	error = libxfs_iget(mp, NULL, dst_ino, 0, &ip);
> 
> --- /var/tmp/fstests/xfs/633.out.bad
> +++ /dev/fd/63	2025-09-21 17:45:58.431935255 -0700
> @@ -1,6 +1,107468 @@
>  QA output created by 633
>  Format and populate
>  Recover filesystem
> +./common/xfs: line 335: 326611 Segmentation fault      (core dumped) $XFS_DB_PROG "$@" $(_scratch_xfs_db_options)
> +./common/xfs: line 335: 326617 Segmentation fault      (core dumped) $XFS_DB_PROG "$@" $(_scratch_xfs_db_options)
>  Check file contents
> +--- /tmp/326192.stat.before	2025-09-21 17:36:10.071959154 -0700
> ++++ /tmp/326192.stat.after	2025-09-21 17:36:12.863959041 -0700
> +@@ -1,23 +1,3 @@
> +-11a4 0:0 0 0 fifo 0 1731556303 ./S_IFIFO
> +-2000 0:0 0 0 character special file 0 1758501368 ./newfiles/p0/d1/c19
> +-2000 0:0 0 0 character special file 0 1758501368 ./newfiles/p0/d1/c1a
> <snip>
> 
> The xfs/437 failure is trivially fixable, not sure what's causing the
> segfault here?  Probably something in the new file_{get,set}attr code in
> xfs_db is my guess...?
> 
> MKFS_OPTIONS  -- -f -m metadir=1,autofsck=1,uquota,gquota,pquota, -n size=8k, /dev/sdf
> MOUNT_OPTIONS -- /dev/sdf /opt
> 
> (note that both problems happen on a variety of different
> configurations)
> 
> --D
> 

I haven't seen this, thanks, will look into it

-- 
- Andrey

Re: regressions in xfs/633 and xfs/437?

[Darrick J. Wong]

On Mon, Sep 22, 2025 at 06:11:27PM +0200, Andrey Albershteyn wrote:
> On 2025-09-22 08:29:54, Darrick J. Wong wrote:
> > Hey folks,
> > 
> > Have any of you noticed a recent regression in xfs/437 and xfs/633?
> > 
> > --- /run/fstests/bin/tests/xfs/437.out	2025-07-15 14:41:40.303420629 -0700
> > +++ /var/tmp/fstests/xfs/437.out.bad	2025-09-21 18:53:36.368250642 -0700
> > @@ -1,2 +1,3 @@
> >  QA output created by 437
> >  Silence is golden.
> > +mkfs/proto.c:1428:	error = libxfs_iget(mp, NULL, dst_ino, 0, &ip);
> > 
> > --- /var/tmp/fstests/xfs/633.out.bad
> > +++ /dev/fd/63	2025-09-21 17:45:58.431935255 -0700
> > @@ -1,6 +1,107468 @@
> >  QA output created by 633
> >  Format and populate
> >  Recover filesystem
> > +./common/xfs: line 335: 326611 Segmentation fault      (core dumped) $XFS_DB_PROG "$@" $(_scratch_xfs_db_options)
> > +./common/xfs: line 335: 326617 Segmentation fault      (core dumped) $XFS_DB_PROG "$@" $(_scratch_xfs_db_options)
> >  Check file contents
> > +--- /tmp/326192.stat.before	2025-09-21 17:36:10.071959154 -0700
> > ++++ /tmp/326192.stat.after	2025-09-21 17:36:12.863959041 -0700
> > +@@ -1,23 +1,3 @@
> > +-11a4 0:0 0 0 fifo 0 1731556303 ./S_IFIFO
> > +-2000 0:0 0 0 character special file 0 1758501368 ./newfiles/p0/d1/c19
> > +-2000 0:0 0 0 character special file 0 1758501368 ./newfiles/p0/d1/c1a
> > <snip>
> > 
> > The xfs/437 failure is trivially fixable, not sure what's causing the
> > segfault here?  Probably something in the new file_{get,set}attr code in
> > xfs_db is my guess...?
> > 
> > MKFS_OPTIONS  -- -f -m metadir=1,autofsck=1,uquota,gquota,pquota, -n size=8k, /dev/sdf
> > MOUNT_OPTIONS -- /dev/sdf /opt
> > 
> > (note that both problems happen on a variety of different
> > configurations)
> > 
> > --D
> > 
> 
> I haven't seen this, thanks, will look into it

I'll send some fixes shortly.

--D

> -- 
> - Andrey
> 
>