On 04/30/2013 03:36 PM, Dave Chinner wrote:
> On Mon, Apr 29, 2013 at 07:44:18PM +0800, Jeff Liu wrote:
>> From: Jie Liu <jeff.liu@xxxxxxxxxx>
>>
>> As per the mount man page, sunit and swidth can be changed via
>> mount options. For XFS, on the face of it, those options seems
>> works if the specified alignments is properly, e.g.
>> # mount -o sunit=4096,swidth=8192 /dev/sdb1 /mnt
>> # mount | grep sdb1
>> /dev/sdb1 on /mnt type xfs (rw,sunit=4096,swidth=8192)
>>
>> However, neither sunit nor swidth is shown from the xfs_info output.
>> # xfs_info /mnt
>> meta-data=/dev/sdb1 isize=256 agcount=4, agsize=262144 blks
>> = sectsz=512 attr=2
>> data = bsize=4096 blocks=1048576, imaxpct=25
>> = sunit=0 swidth=0 blks
>> ^^^^^^^^^^^^^^^^^^^^^^^^^^
>> naming =version 2 bsize=4096 ascii-ci=0
>> log =internal bsize=4096 blocks=2560, version=2
>> = sectsz=512 sunit=0 blks, lazy-count=1
>> realtime =none extsz=4096 blocks=0, rtextents=0
>>
>> The reason is that the alignment can only be changed if the relevant
>> super block is already configured with alignments, otherwise, the
>> given value will be silently ignored, so it's better to tell user
>> that the alignment-changing can not take affect in one way or another.
>>
>> With this fix, the attempt to mount a storage without strip alignment
>> setup on super block will failed if XFS_MOUNT_RETERR is enabled, or
>> just ignore the given alignment and drop a warning to indicate the
>> cause in syslog.
>>
>> # mount -o sunit=4096,swidth=8192 /dev/sdb1 /mnt
>> mount: wrong fs type, bad option, bad superblock on /dev/sdb1,
>> missing codepage or helper program, or other error
>> In some cases useful info is found in syslog - try
>> dmesg | tail or so
>>
>> # dmesg|tail
>> .......
>> XFS (sdb1): can not change alignment: no data alignment on superblock
>>
>> Signed-off-by: Jie Liu <jeff.liu@xxxxxxxxxx>
>>
>> ---
>> fs/xfs/xfs_mount.c | 7 +++++++
>> 1 file changed, 7 insertions(+)
>>
>> diff --git a/fs/xfs/xfs_mount.c b/fs/xfs/xfs_mount.c
>> index 3806088..bc7fdd4 100644
>> --- a/fs/xfs/xfs_mount.c
>> +++ b/fs/xfs/xfs_mount.c
>> @@ -924,6 +924,13 @@ xfs_update_alignment(xfs_mount_t *mp)
>> sbp->sb_width = mp->m_swidth;
>> mp->m_update_flags |= XFS_SB_WIDTH;
>> }
>> + } else {
>> + xfs_warn(mp, "can not change alignment: "
>> + "no data alignment on superblock");
>> + if (mp->m_flags & XFS_MOUNT_RETERR)
>> + return XFS_ERROR(EINVAL);
>> + mp->m_dalign = 0;
>> + mp->m_swidth = 0;
>
> Can someone tell me why the XFS_MOUNT_RETERR flag exists?
This is really a very opportune response because I also worked out
another tiny patch for removing XFS_MOUNT_RETERR a few minutes ago,
just hesitating if I missed anything or not.
>
> It looks like dead code to me as the only time mp->m_dalign is set
> prior to calling xfs_update_alignment() is the same code that sets
> XFS_MOUNT_RETERR in xfs_parseargs().
>
> IOWs, any time we enter this "if (mp->m_dalign)" branch in
> xfs_update_alignment(), XFS_MOUNT_RETERR is going to be set and so
> we should always be emitting a warning and returning an error.
Yes, I realized that as I can not trigger a warning only, it always
returning an error to me. :(
>
> If this is correct, Jeff, can you remove the XFS_MOUNT_RETERR flag
> and get rid of all the dead code in xfs_update_alignment() at the
> same time, please?
Sure, I'll post this patch tonight together with another initial patch
for fixing transaction space over-reservation we have discussed two
weeks ago, xfstests is running now.
Thanks,
-Jeff
|