gfs2: Fix length of holes reported at end-of-file
authorAndreas Gruenbacher <agruenba@redhat.com>
Fri, 5 Nov 2021 23:18:56 +0000 (00:18 +0100)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 8 Dec 2021 08:04:37 +0000 (09:04 +0100)
[ Upstream commit f3506eee81d1f700d9ee2d2f4a88fddb669ec032 ]

Fix the length of holes reported at the end of a file: the length is
relative to the beginning of the extent, not the seek position which is
rounded down to the filesystem block size.

This bug went unnoticed for some time, but is now caught by the
following assertion in iomap_iter_done():

  WARN_ON_ONCE(iter->iomap.offset + iter->iomap.length <= iter->pos)

Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
fs/gfs2/bmap.c

index 5414c2c..fba3214 100644 (file)
@@ -940,7 +940,7 @@ do_alloc:
                else if (height == ip->i_height)
                        ret = gfs2_hole_size(inode, lblock, len, mp, iomap);
                else
-                       iomap->length = size - pos;
+                       iomap->length = size - iomap->offset;
        } else if (flags & IOMAP_WRITE) {
                u64 alloc_size;