From: Andreas Gruenbacher <agruenba@xxxxxxxxxx>
commit 6bd1c7bd4ee7b17980cdc347522dcb76feac9b98 upstream.
Right now, we can end up calling cancel_delayed_work_sync from within
delete_work_func via gfs2_lookup_by_inum -> gfs2_inode_lookup ->
gfs2_cancel_delete_work. When that happens, it will result in a
deadlock. Instead, gfs2_inode_lookup should skip the call to
gfs2_cancel_delete_work when called from delete_work_func (blktype ==
GFS2_BLKST_UNLINKED).
Reported-by: Alexander Ahring Oder Aring <aahringo@xxxxxxxxxx>
Fixes: a0e3cc65fa29 ("gfs2: Turn gl_delete into a delayed work")
Cc: stable@xxxxxxxxxxxxxxx # v5.8+
Signed-off-by: Andreas Gruenbacher <agruenba@xxxxxxxxxx>
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
---
fs/gfs2/inode.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
--- a/fs/gfs2/inode.c
+++ b/fs/gfs2/inode.c
@@ -180,7 +180,8 @@ struct inode *gfs2_inode_lookup(struct s
error = gfs2_glock_nq_init(io_gl, LM_ST_SHARED, GL_EXACT, &ip->i_iopen_gh);
if (unlikely(error))
goto fail;
- gfs2_cancel_delete_work(ip->i_iopen_gh.gh_gl);
+ if (blktype != GFS2_BLKST_UNLINKED)
+ gfs2_cancel_delete_work(ip->i_iopen_gh.gh_gl);
glock_set_object(ip->i_iopen_gh.gh_gl, ip);
gfs2_glock_put(io_gl);
io_gl = NULL;
