On Wed, Jan 15, 2025 at 07:18:32AM +0100, cem@xxxxxxxxxx wrote:
> From: Carlos Maiolino <cem@xxxxxxxxxx>
>
> Mounting a filesystem that requires quota state changing will generate a
> transaction.
>
> We already check for a read-only device; we should do that for
> norecovery too.
>
> A quotacheck on a norecovery mount, and with the right log size, will cause
> the mount process to hang on:
>
> [<0>] xlog_grant_head_wait+0x5d/0x2a0 [xfs]
> [<0>] xlog_grant_head_check+0x112/0x180 [xfs]
> [<0>] xfs_log_reserve+0xe3/0x260 [xfs]
> [<0>] xfs_trans_reserve+0x179/0x250 [xfs]
> [<0>] xfs_trans_alloc+0x101/0x260 [xfs]
> [<0>] xfs_sync_sb+0x3f/0x80 [xfs]
> [<0>] xfs_qm_mount_quotas+0xe3/0x2f0 [xfs]
> [<0>] xfs_mountfs+0x7ad/0xc20 [xfs]
> [<0>] xfs_fs_fill_super+0x762/0xa50 [xfs]
> [<0>] get_tree_bdev_flags+0x131/0x1d0
> [<0>] vfs_get_tree+0x26/0xd0
> [<0>] vfs_cmd_create+0x59/0xe0
> [<0>] __do_sys_fsconfig+0x4e3/0x6b0
> [<0>] do_syscall_64+0x82/0x160
> [<0>] entry_SYSCALL_64_after_hwframe+0x76/0x7e
>
> This is caused by a transaction running with bogus initialized head/tail
>
> I initially hit this while running generic/050, with random log
> sizes, but I managed to reproduce it reliably here with the steps
> below:
>
> mkfs.xfs -f -lsize=1025M -f -b size=4096 -m crc=1,reflink=1,rmapbt=1, -i
> sparse=1 /dev/vdb2 > /dev/null
> mount -o usrquota,grpquota,prjquota /dev/vdb2 /mnt
> xfs_io -x -c 'shutdown -f' /mnt
> umount /mnt
> mount -o ro,norecovery,usrquota,grpquota,prjquota /dev/vdb2 /mnt
>
> Last mount hangs up
>
> Signed-off-by: Carlos Maiolino <cmaiolino@xxxxxxxxxx>
> ---
> fs/xfs/xfs_qm_bhv.c | 12 +++++++-----
> 1 file changed, 7 insertions(+), 5 deletions(-)
>
> diff --git a/fs/xfs/xfs_qm_bhv.c b/fs/xfs/xfs_qm_bhv.c
> index 37f1230e7584..eae106ca7e1b 100644
> --- a/fs/xfs/xfs_qm_bhv.c
> +++ b/fs/xfs/xfs_qm_bhv.c
> @@ -97,10 +97,11 @@ xfs_qm_newmount(
> }
>
> /*
> - * If the device itself is read-only, we can't allow
> - * the user to change the state of quota on the mount -
> - * this would generate a transaction on the ro device,
> - * which would lead to an I/O error and shutdown
> + * If the device itself is read-only and/or in norecovery
> + * mode, we can't allow the user to change the state of
> + * quota on the mount - this would generate a transaction
> + * on the ro device, which would lead to an I/O error and
> + * shutdown.
> */
>
> if (((uquotaondisk && !XFS_IS_UQUOTA_ON(mp)) ||
> @@ -109,7 +110,8 @@ xfs_qm_newmount(
> (!gquotaondisk && XFS_IS_GQUOTA_ON(mp)) ||
> (pquotaondisk && !XFS_IS_PQUOTA_ON(mp)) ||
> (!pquotaondisk && XFS_IS_PQUOTA_ON(mp))) &&
> - xfs_dev_is_read_only(mp, "changing quota state")) {
> + (xfs_dev_is_read_only(mp, "changing quota state") ||
> + xfs_has_norecovery(mp))) {
> xfs_warn(mp, "please mount with%s%s%s%s.",
> (!quotaondisk ? "out quota" : ""),
> (uquotaondisk ? " usrquota" : ""),
The logic seems ok, but (as I mentioned in office hours this morning) I
wonder if we shouldn't just ignore quota entirely for a norecovery
mount?
I guess for metadir we also could change the message to say "don't mount
with any quota options" since the quota mount options are persistent now.
--D
> --
> 2.47.1
>
