We need to splice COW blocks we've completed in xfs_end_io_direct_write
into the data fork before converting unwritten extents. Otherwise
xfs_bmapi_write might first allocate blocks for any holes in the data
fork, which isn't only not needed but also harmful as it might cause
reserved block underruns in the transaction.
Signed-off-by: Christoph Hellwig <hch@xxxxxx>
---
fs/xfs/xfs_aops.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
index 0e2a931..3e57a56 100644
--- a/fs/xfs/xfs_aops.c
+++ b/fs/xfs/xfs_aops.c
@@ -1547,13 +1547,13 @@ xfs_end_io_direct_write(
i_size_write(inode, offset + size);
spin_unlock(&ip->i_flags_lock);
+ if (flags & XFS_DIO_FLAG_COW)
+ error = xfs_reflink_end_cow(ip, offset, size);
if (flags & XFS_DIO_FLAG_UNWRITTEN) {
trace_xfs_end_io_direct_write_unwritten(ip, offset, size);
error = xfs_iomap_write_unwritten(ip, offset, size);
}
- if (flags & XFS_DIO_FLAG_COW)
- error = xfs_reflink_end_cow(ip, offset, size);
if (flags & XFS_DIO_FLAG_APPEND) {
trace_xfs_end_io_direct_write_append(ip, offset, size);
--
2.10.1.382.ga23ca1b
--
To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
