From: Bob Peterson Date: Thu, 28 Oct 2021 16:53:10 +0000 (-0500) Subject: gfs2: release iopen glock early in evict X-Git-Tag: accepted/tizen/unified/20230118.172025~4645 X-Git-Url: http://review.tizen.org/git/?a=commitdiff_plain;h=4b11e583193c694a2946f9bc9a719967cc70d062;p=platform%2Fkernel%2Flinux-rpi.git gfs2: release iopen glock early in evict [ Upstream commit 49462e2be119d38c5eb5759d0d1b712df3a41239 ] Before this patch, evict would clear the iopen glock's gl_object after releasing the inode glock. In the meantime, another process could reuse the same block and thus glocks for a new inode. It would lock the inode glock (exclusively), and then the iopen glock (shared). The shared locking mode doesn't provide any ordering against the evict, so by the time the iopen glock is reused, evict may not have gotten to setting gl_object to NULL. Fix that by releasing the iopen glock before the inode glock in gfs2_evict_inode. Signed-off-by: Bob Peterson gl_object Signed-off-by: Andreas Gruenbacher Signed-off-by: Sasha Levin --- diff --git a/fs/gfs2/super.c b/fs/gfs2/super.c index 6e00d15..cc51b5f 100644 --- a/fs/gfs2/super.c +++ b/fs/gfs2/super.c @@ -1402,13 +1402,6 @@ out: gfs2_ordered_del_inode(ip); clear_inode(inode); gfs2_dir_hash_inval(ip); - if (ip->i_gl) { - glock_clear_object(ip->i_gl, ip); - wait_on_bit_io(&ip->i_flags, GIF_GLOP_PENDING, TASK_UNINTERRUPTIBLE); - gfs2_glock_add_to_lru(ip->i_gl); - gfs2_glock_put_eventually(ip->i_gl); - ip->i_gl = NULL; - } if (gfs2_holder_initialized(&ip->i_iopen_gh)) { struct gfs2_glock *gl = ip->i_iopen_gh.gh_gl; @@ -1421,6 +1414,13 @@ out: gfs2_holder_uninit(&ip->i_iopen_gh); gfs2_glock_put_eventually(gl); } + if (ip->i_gl) { + glock_clear_object(ip->i_gl, ip); + wait_on_bit_io(&ip->i_flags, GIF_GLOP_PENDING, TASK_UNINTERRUPTIBLE); + gfs2_glock_add_to_lru(ip->i_gl); + gfs2_glock_put_eventually(ip->i_gl); + ip->i_gl = NULL; + } } static struct inode *gfs2_alloc_inode(struct super_block *sb)