From mboxrd@z Thu Jan 1 00:00:00 1970 From: Bob Peterson Date: Fri, 31 Jan 2020 12:07:21 -0600 Subject: [Cluster-devel] [GFS2 PATCH 23/25] gfs2: Don't demote a glock until its revokes are written In-Reply-To: <20200131180723.178863-1-rpeterso@redhat.com> References: <20200131180723.178863-1-rpeterso@redhat.com> Message-ID: <20200131180723.178863-24-rpeterso@redhat.com> List-Id: To: cluster-devel.redhat.com MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Before this patch, run_queue would demote glocks based on whether there are any more holders. But if the glock has pending revokes that haven't been written to the media, giving up the glock might end in file system corruption if the revokes never get written due to io errors, node crashes and fences, etc. In that case, another node will replay the metadata blocks associated with the glock, but because the revoke was never written, it could replay that block even though the glock had since been granted to another node who might have made changes. This patch changes the logic in run_queue so that it never demotes a glock until its count of pending revokes reaches zero. Signed-off-by: Bob Peterson Reviewed-by: Andreas Gruenbacher --- fs/gfs2/glock.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/fs/gfs2/glock.c b/fs/gfs2/glock.c index 70a97a40f3db..2b43a8488aec 100644 --- a/fs/gfs2/glock.c +++ b/fs/gfs2/glock.c @@ -708,6 +708,9 @@ __acquires(&gl->gl_lockref.lock) goto out_unlock; if (nonblock) goto out_sched; + smp_mb(); + if (atomic_read(&gl->gl_revokes) != 0) + goto out_sched; set_bit(GLF_DEMOTE_IN_PROGRESS, &gl->gl_flags); GLOCK_BUG_ON(gl, gl->gl_demote_state == LM_ST_EXCLUSIVE); gl->gl_target = gl->gl_demote_state; -- 2.24.1