]> git.ipfire.org Git - thirdparty/xfsprogs-dev.git/commitdiff
xfs: update superblock counters correctly for !lazysbcount
authorDave Chinner <dchinner@redhat.com>
Wed, 30 Jun 2021 22:38:58 +0000 (18:38 -0400)
committerEric Sandeen <sandeen@sandeen.net>
Wed, 30 Jun 2021 22:38:58 +0000 (18:38 -0400)
Source kernel commit: 6543990a168acf366f4b6174d7bd46ba15a8a2a6

Keep the mount superblock counters up to date for !lazysbcount
filesystems so that when we log the superblock they do not need
updating in any way because they are already correct.

It's found by what Zorro reported:
1. mkfs.xfs -f -l lazy-count=0 -m crc=0 $dev
2. mount $dev $mnt
3. fsstress -d $mnt -p 100 -n 1000 (maybe need more or less io load)
4. umount $mnt
5. xfs_repair -n $dev
and I've seen no problem with this patch.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reported-by: Zorro Lang <zlang@redhat.com>
Reviewed-by: Gao Xiang <hsiangkao@redhat.com>
Signed-off-by: Gao Xiang <hsiangkao@redhat.com>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Eric Sandeen <sandeen@sandeen.net>
libxfs/xfs_sb.c

index 8037b369a9acdfeebe9f102a15e22268ec1f5f1d..345635aa35ae1f3ef526b8131cb67f496dcbb8ef 100644 (file)
@@ -923,9 +923,19 @@ xfs_log_sb(
        struct xfs_mount        *mp = tp->t_mountp;
        struct xfs_buf          *bp = xfs_trans_getsb(tp);
 
-       mp->m_sb.sb_icount = percpu_counter_sum(&mp->m_icount);
-       mp->m_sb.sb_ifree = percpu_counter_sum(&mp->m_ifree);
-       mp->m_sb.sb_fdblocks = percpu_counter_sum(&mp->m_fdblocks);
+       /*
+        * Lazy sb counters don't update the in-core superblock so do that now.
+        * If this is at unmount, the counters will be exactly correct, but at
+        * any other time they will only be ballpark correct because of
+        * reservations that have been taken out percpu counters. If we have an
+        * unclean shutdown, this will be corrected by log recovery rebuilding
+        * the counters from the AGF block counts.
+        */
+       if (xfs_sb_version_haslazysbcount(&mp->m_sb)) {
+               mp->m_sb.sb_icount = percpu_counter_sum(&mp->m_icount);
+               mp->m_sb.sb_ifree = percpu_counter_sum(&mp->m_ifree);
+               mp->m_sb.sb_fdblocks = percpu_counter_sum(&mp->m_fdblocks);
+       }
 
        xfs_sb_to_disk(bp->b_addr, &mp->m_sb);
        xfs_trans_buf_set_type(tp, bp, XFS_BLFT_SB_BUF);