diff options
author | Darrick J. Wong <darrick.wong@oracle.com> | 2017-12-14 15:42:59 -0800 |
---|---|---|
committer | Darrick J. Wong <darrick.wong@oracle.com> | 2017-12-21 08:47:28 -0800 |
commit | 363e59baa4f76d3f97c0133ff7014cba3d90a7c3 (patch) | |
tree | fee3a061d7ab180e6637399edd30377f5ad3efdd /fs | |
parent | 91aae6be4139b9e3902656d819e6af66e051bd7a (diff) | |
download | op-kernel-dev-363e59baa4f76d3f97c0133ff7014cba3d90a7c3.zip op-kernel-dev-363e59baa4f76d3f97c0133ff7014cba3d90a7c3.tar.gz |
xfs: don't be so eager to clear the cowblocks tag on truncate
Currently, xfs_itruncate_extents clears the cowblocks tag if i_cnextents
is zero. This is wrong, since i_cnextents only tracks real extents in
the CoW fork, which means that we could have some delayed CoW
reservations still in there that will now never get cleaned.
Fix a further bug where we /don't/ clear the reflink iflag if there are
any attribute blocks -- really, it's only safe to clear the reflink flag
if there are no data fork extents and no cow fork extents.
Found by adding clonerange to fsstress in xfs/017.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Diffstat (limited to 'fs')
-rw-r--r-- | fs/xfs/xfs_inode.c | 28 |
1 files changed, 19 insertions, 9 deletions
diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c index b41952a..6f95bdb 100644 --- a/fs/xfs/xfs_inode.c +++ b/fs/xfs/xfs_inode.c @@ -1487,6 +1487,24 @@ xfs_link( return error; } +/* Clear the reflink flag and the cowblocks tag if possible. */ +static void +xfs_itruncate_clear_reflink_flags( + struct xfs_inode *ip) +{ + struct xfs_ifork *dfork; + struct xfs_ifork *cfork; + + if (!xfs_is_reflink_inode(ip)) + return; + dfork = XFS_IFORK_PTR(ip, XFS_DATA_FORK); + cfork = XFS_IFORK_PTR(ip, XFS_COW_FORK); + if (dfork->if_bytes == 0 && cfork->if_bytes == 0) + ip->i_d.di_flags2 &= ~XFS_DIFLAG2_REFLINK; + if (cfork->if_bytes == 0) + xfs_inode_clear_cowblocks_tag(ip); +} + /* * Free up the underlying blocks past new_size. The new size must be smaller * than the current size. This routine can be used both for the attribute and @@ -1583,15 +1601,7 @@ xfs_itruncate_extents( if (error) goto out; - /* - * Clear the reflink flag if there are no data fork blocks and - * there are no extents staged in the cow fork. - */ - if (xfs_is_reflink_inode(ip) && ip->i_cnextents == 0) { - if (ip->i_d.di_nblocks == 0) - ip->i_d.di_flags2 &= ~XFS_DIFLAG2_REFLINK; - xfs_inode_clear_cowblocks_tag(ip); - } + xfs_itruncate_clear_reflink_flags(ip); /* * Always re-log the inode so that our permanent transaction can keep |