Joseph Qi
2014-Oct-08 05:44 UTC
[Ocfs2-devel] [PATCH] ocfs2/dlm: let sender retry if dlm_dispatch_assert_master failed with -ENOMEM
Do not BUG() if GFP_ATOMIC allocation fails in
dlm_dispatch_assert_master.
Instead, return -ENOMEM to the sender and then retry.
Signed-off-by: Joseph Qi <joseph.qi at huawei.com>
Reviewed-by: Alex Chen <alex.chen at huawei.com>
---
fs/ocfs2/dlm/dlmrecovery.c | 18 +++++++++++++-----
1 file changed, 13 insertions(+), 5 deletions(-)
diff --git a/fs/ocfs2/dlm/dlmrecovery.c b/fs/ocfs2/dlm/dlmrecovery.c
index 3365839..79b5af5 100644
--- a/fs/ocfs2/dlm/dlmrecovery.c
+++ b/fs/ocfs2/dlm/dlmrecovery.c
@@ -1656,14 +1656,18 @@ int dlm_do_master_requery(struct dlm_ctxt *dlm, struct
dlm_lock_resource *res,
req.namelen = res->lockname.len;
memcpy(req.name, res->lockname.name, res->lockname.len);
+resend:
ret = o2net_send_message(DLM_MASTER_REQUERY_MSG, dlm->key,
&req, sizeof(req), nodenum, &status);
- /* XXX: negative status not handled properly here. */
if (ret < 0)
mlog(ML_ERROR, "Error %d when sending message %u (key "
"0x%x) to node %u\n", ret, DLM_MASTER_REQUERY_MSG,
dlm->key, nodenum);
- else {
+ else if (status == -ENOMEM) {
+ mlog_errno(status);
+ msleep(50);
+ goto resend;
+ } else {
BUG_ON(status < 0);
BUG_ON(status > DLM_LOCK_RES_OWNER_UNKNOWN);
*real_master = (u8) (status & 0xff);
@@ -1705,9 +1709,13 @@ int dlm_master_requery_handler(struct o2net_msg *msg, u32
len, void *data,
int ret = dlm_dispatch_assert_master(dlm, res,
0, 0, flags);
if (ret < 0) {
- mlog_errno(-ENOMEM);
- /* retry!? */
- BUG();
+ mlog_errno(ret);
+ spin_unlock(&res->spinlock);
+ dlm_lockres_put(res);
+ spin_unlock(&dlm->spinlock);
+ dlm_put(dlm);
+ /* sender will take care of this and retry */
+ return ret;
} else
__dlm_lockres_grab_inflight_worker(dlm, res);
spin_unlock(&res->spinlock);
--
1.8.4.3