From: Chandan Babu R <chandanrlinux@xxxxxxxxx>
Source kernel commit: 0dbc5cb1a91cc8c44b1c75429f5b9351837114fd
Directory entry removal must always succeed; Hence XFS does the
following during low disk space scenario:
1. Data/Free blocks linger until a future remove operation.
2. Dabtree blocks would be swapped with the last block in the leaf space
and then the new last block will be unmapped.
This facility is reused during low inode extent count scenario i.e. this
that the above mentioned behaviour is exercised causing no change to the
directory's extent count.
Signed-off-by: Chandan Babu R <chandanrlinux@xxxxxxxxx>
Reviewed-by: Darrick J. Wong <darrick.wong@xxxxxxxxxx>
Signed-off-by: Darrick J. Wong <darrick.wong@xxxxxxxxxx>
Signed-off-by: Allison Henderson <allison.henderson@xxxxxxxxxx>
---
libxfs/xfs_bmap.c | 18 ++++++++++++++++++
1 file changed, 18 insertions(+)
diff --git a/libxfs/xfs_bmap.c b/libxfs/xfs_bmap.c
index 336c6d6..e3c6b0b 100644
--- a/libxfs/xfs_bmap.c
+++ b/libxfs/xfs_bmap.c
@@ -5144,6 +5144,24 @@ xfs_bmap_del_extent_real(
/*
* Deleting the middle of the extent.
*/
+
+ /*
+ * For directories, -ENOSPC is returned since a directory entry
+ * remove operation must not fail due to low extent count
+ * availability. -ENOSPC will be handled by higher layers of XFS
+ * by letting the corresponding empty Data/Free blocks to linger
+ * until a future remove operation. Dabtree blocks would be
+ * swapped with the last block in the leaf space and then the
+ * new last block will be unmapped.
+ */
+ error = xfs_iext_count_may_overflow(ip, whichfork, 1);
+ if (error) {
+ ASSERT(S_ISDIR(VFS_I(ip)->i_mode) &&
+ whichfork == XFS_DATA_FORK);
+ error = -ENOSPC;
+ goto done;
+ }
+
old = got;
got.br_blockcount = del->br_startoff - got.br_startoff;
--
2.7.4
