From mboxrd@z Thu Jan 1 00:00:00 1970 From: Bob Peterson Date: Fri, 8 Mar 2019 08:02:16 -0700 Subject: [Cluster-devel] [PATCH 3/3] gfs2: Fix missed wakeups in find_insert_glock In-Reply-To: <20190308150216.8732-1-rpeterso@redhat.com> References: <20190308150216.8732-1-rpeterso@redhat.com> Message-ID: <20190308150216.8732-4-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 From: Andreas Gruenbacher Mark Syms has reported seeing tasks that are stuck waiting in find_insert_glock. It turns out that struct lm_lockname contains four padding bytes on 64-bit architectures that function glock_waitqueue doesn't skip when hashing the glock name. As a result, we can end up waking up the wrong waitqueue, and the waiting tasks may be stuck forever. Fix that by using ht_parms.key_len instead of sizeof(struct lm_lockname) for the key length. Reported-by: Mark Syms Signed-off-by: Andreas Gruenbacher Signed-off-by: Bob Peterson --- fs/gfs2/glock.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/gfs2/glock.c b/fs/gfs2/glock.c index f66773c71bcd..d32964cd1117 100644 --- a/fs/gfs2/glock.c +++ b/fs/gfs2/glock.c @@ -107,7 +107,7 @@ static int glock_wake_function(wait_queue_entry_t *wait, unsigned int mode, static wait_queue_head_t *glock_waitqueue(struct lm_lockname *name) { - u32 hash = jhash2((u32 *)name, sizeof(*name) / 4, 0); + u32 hash = jhash2((u32 *)name, ht_parms.key_len / 4, 0); return glock_wait_table + hash_32(hash, GLOCK_WAIT_TABLE_BITS); } -- 2.20.1