Running "xfs_admin -c 0" on a V5 fs has unfortunate results.
Before:
versionnum [0xbca5+0x18a] = ... LAZYSBCOUNT,PROJID32BIT,CRC, ...
After:
versionnum [0xbca5+0x188] = ... LAZYSBCOUNT,PROJID32BIT,CRC, ...
The on-disk feature bit is in fact changed, but xfs_db reports that the
feature is still present. This is because xfs_sb_version_haslazysbcount()
ignores the feature bit, and always reports that the lazy-count feature is
present on a V5 fs.
i.e. the intent is that this feature is not optional on a V5 fs, yet we
still flip the bit on disk when asked. Not good.
So, document this, and error out if the user tries to disable lazy-count
on a version 5 filesystem.
Reported-by: Richard W.M. Jones <rjones@redhat.com>
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
.TP
.BR "\-c 0" | 1
Enable (1) or disable (0) lazy-counters in the filesystem.
+.IP
+Lazy-counters may not be disabled on Version 5 superblock filesystems
+(i.e. those with metadata CRCs enabled).
+.IP
This operation may take quite a bit of time on large filesystems as the
entire filesystem needs to be scanned when this option is changed.
.IP
sb->sb_bad_features2 |= XFS_SB_VERSION2_LAZYSBCOUNTBIT;
primary_sb_modified = 1;
printf(_("Enabling lazy-counters\n"));
- } else
- if (!lazy_count && xfs_sb_version_haslazysbcount(sb)) {
+ } else if (!lazy_count && xfs_sb_version_haslazysbcount(sb)) {
+ if (XFS_SB_VERSION_NUM(sb) == XFS_SB_VERSION_5) {
+ printf(
+_("Cannot disable lazy-counters on V5 fs\n"));
+ exit(1);
+ }
sb->sb_features2 &= ~XFS_SB_VERSION2_LAZYSBCOUNTBIT;
sb->sb_bad_features2 &= ~XFS_SB_VERSION2_LAZYSBCOUNTBIT;
printf(_("Disabling lazy-counters\n"));