OSDN Git Service

GFS2: Don't filter out I_FREEING inodes anymore
authorBob Peterson <rpeterso@redhat.com>
Fri, 27 Mar 2015 13:25:41 +0000 (08:25 -0500)
committerBob Peterson <rpeterso@redhat.com>
Tue, 15 Mar 2016 14:46:45 +0000 (10:46 -0400)
commitff34245d524a898eee6e013eb1ec165095277148
tree4c3c0abc8bd6cdaca2b0ab5667f0c72f0dbada0e
parenta4923865ea071b0bd708339df7a83c76732fa2db
GFS2: Don't filter out I_FREEING inodes anymore

This patch basically reverts a very old patch from 2008,
7a9f53b3c1875bef22ad4588e818bc046ef183da, with the title
"Alternate gfs2_iget to avoid looking up inodes being freed".
The original patch was designed to avoid a deadlock caused by lock
ordering with try_rgrp_unlink. The patch forced the function to not
find inodes that were being removed by VFS. The problem is, that
made it impossible for nodes to delete their own unlinked dinodes
after a certain point in time, because the inode needed was not found
by this filtering process. There is no longer a need for the patch,
since function try_rgrp_unlink no longer locks the inode: All it does
is queue the glock onto the delete work_queue, so there should be no
more deadlock.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
fs/gfs2/export.c
fs/gfs2/glock.c
fs/gfs2/inode.c
fs/gfs2/inode.h