Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph...
authorLinus Torvalds <torvalds@linux-foundation.org>
Thu, 3 Jan 2013 01:32:49 +0000 (17:32 -0800)
committerLinus Torvalds <torvalds@linux-foundation.org>
Thu, 3 Jan 2013 01:32:49 +0000 (17:32 -0800)
Pull Ceph fixes from Sage Weil:
 "Two of Alex's patches deal with a race when reseting server
  connections for open RBD images, one demotes some non-fatal BUGs to
  WARNs, and my patch fixes a protocol feature bit failure path."

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph-client:
  libceph: fix protocol feature mismatch failure path
  libceph: WARN, don't BUG on unexpected connection states
  libceph: always reset osds when kicking
  libceph: move linger requests sooner in kick_requests()

net/ceph/messenger.c
net/ceph/osd_client.c

index 4d111fd2b4923f7fe7cd287de00029b19c61f274..5ccf87ed8d688820a23ba1267439abe149398ec2 100644 (file)
@@ -506,6 +506,7 @@ static void reset_connection(struct ceph_connection *con)
 {
        /* reset connection, out_queue, msg_ and connect_seq */
        /* discard existing out_queue and msg_seq */
+       dout("reset_connection %p\n", con);
        ceph_msg_remove_list(&con->out_queue);
        ceph_msg_remove_list(&con->out_sent);
 
@@ -561,7 +562,7 @@ void ceph_con_open(struct ceph_connection *con,
        mutex_lock(&con->mutex);
        dout("con_open %p %s\n", con, ceph_pr_addr(&addr->in_addr));
 
-       BUG_ON(con->state != CON_STATE_CLOSED);
+       WARN_ON(con->state != CON_STATE_CLOSED);
        con->state = CON_STATE_PREOPEN;
 
        con->peer_name.type = (__u8) entity_type;
@@ -1506,13 +1507,6 @@ static int process_banner(struct ceph_connection *con)
        return 0;
 }
 
-static void fail_protocol(struct ceph_connection *con)
-{
-       reset_connection(con);
-       BUG_ON(con->state != CON_STATE_NEGOTIATING);
-       con->state = CON_STATE_CLOSED;
-}
-
 static int process_connect(struct ceph_connection *con)
 {
        u64 sup_feat = con->msgr->supported_features;
@@ -1530,7 +1524,7 @@ static int process_connect(struct ceph_connection *con)
                       ceph_pr_addr(&con->peer_addr.in_addr),
                       sup_feat, server_feat, server_feat & ~sup_feat);
                con->error_msg = "missing required protocol features";
-               fail_protocol(con);
+               reset_connection(con);
                return -1;
 
        case CEPH_MSGR_TAG_BADPROTOVER:
@@ -1541,7 +1535,7 @@ static int process_connect(struct ceph_connection *con)
                       le32_to_cpu(con->out_connect.protocol_version),
                       le32_to_cpu(con->in_reply.protocol_version));
                con->error_msg = "protocol version mismatch";
-               fail_protocol(con);
+               reset_connection(con);
                return -1;
 
        case CEPH_MSGR_TAG_BADAUTHORIZER:
@@ -1631,11 +1625,11 @@ static int process_connect(struct ceph_connection *con)
                               ceph_pr_addr(&con->peer_addr.in_addr),
                               req_feat, server_feat, req_feat & ~server_feat);
                        con->error_msg = "missing required protocol features";
-                       fail_protocol(con);
+                       reset_connection(con);
                        return -1;
                }
 
-               BUG_ON(con->state != CON_STATE_NEGOTIATING);
+               WARN_ON(con->state != CON_STATE_NEGOTIATING);
                con->state = CON_STATE_OPEN;
 
                con->peer_global_seq = le32_to_cpu(con->in_reply.global_seq);
@@ -2132,7 +2126,6 @@ more:
                if (ret < 0)
                        goto out;
 
-               BUG_ON(con->state != CON_STATE_CONNECTING);
                con->state = CON_STATE_NEGOTIATING;
 
                /*
@@ -2160,7 +2153,7 @@ more:
                goto more;
        }
 
-       BUG_ON(con->state != CON_STATE_OPEN);
+       WARN_ON(con->state != CON_STATE_OPEN);
 
        if (con->in_base_pos < 0) {
                /*
@@ -2382,7 +2375,7 @@ static void ceph_fault(struct ceph_connection *con)
        dout("fault %p state %lu to peer %s\n",
             con, con->state, ceph_pr_addr(&con->peer_addr.in_addr));
 
-       BUG_ON(con->state != CON_STATE_CONNECTING &&
+       WARN_ON(con->state != CON_STATE_CONNECTING &&
               con->state != CON_STATE_NEGOTIATING &&
               con->state != CON_STATE_OPEN);
 
index 780caf6b049188cd20e043f2206b23600223b820..eb9a4447876481e9a4110a1e68ea351ce3ec86d9 100644 (file)
@@ -1270,7 +1270,7 @@ static void reset_changed_osds(struct ceph_osd_client *osdc)
  * Requeue requests whose mapping to an OSD has changed.  If requests map to
  * no osd, request a new map.
  *
- * Caller should hold map_sem for read and request_mutex.
+ * Caller should hold map_sem for read.
  */
 static void kick_requests(struct ceph_osd_client *osdc, int force_resend)
 {
@@ -1284,6 +1284,24 @@ static void kick_requests(struct ceph_osd_client *osdc, int force_resend)
        for (p = rb_first(&osdc->requests); p; ) {
                req = rb_entry(p, struct ceph_osd_request, r_node);
                p = rb_next(p);
+
+               /*
+                * For linger requests that have not yet been
+                * registered, move them to the linger list; they'll
+                * be sent to the osd in the loop below.  Unregister
+                * the request before re-registering it as a linger
+                * request to ensure the __map_request() below
+                * will decide it needs to be sent.
+                */
+               if (req->r_linger && list_empty(&req->r_linger_item)) {
+                       dout("%p tid %llu restart on osd%d\n",
+                            req, req->r_tid,
+                            req->r_osd ? req->r_osd->o_osd : -1);
+                       __unregister_request(osdc, req);
+                       __register_linger_request(osdc, req);
+                       continue;
+               }
+
                err = __map_request(osdc, req, force_resend);
                if (err < 0)
                        continue;  /* error */
@@ -1298,17 +1316,6 @@ static void kick_requests(struct ceph_osd_client *osdc, int force_resend)
                                req->r_flags |= CEPH_OSD_FLAG_RETRY;
                        }
                }
-               if (req->r_linger && list_empty(&req->r_linger_item)) {
-                       /*
-                        * register as a linger so that we will
-                        * re-submit below and get a new tid
-                        */
-                       dout("%p tid %llu restart on osd%d\n",
-                            req, req->r_tid,
-                            req->r_osd ? req->r_osd->o_osd : -1);
-                       __register_linger_request(osdc, req);
-                       __unregister_request(osdc, req);
-               }
        }
 
        list_for_each_entry_safe(req, nreq, &osdc->req_linger,
@@ -1316,6 +1323,7 @@ static void kick_requests(struct ceph_osd_client *osdc, int force_resend)
                dout("linger req=%p req->r_osd=%p\n", req, req->r_osd);
 
                err = __map_request(osdc, req, force_resend);
+               dout("__map_request returned %d\n", err);
                if (err == 0)
                        continue;  /* no change and no osd was specified */
                if (err < 0)
@@ -1337,6 +1345,7 @@ static void kick_requests(struct ceph_osd_client *osdc, int force_resend)
                dout("%d requests for down osds, need new map\n", needmap);
                ceph_monc_request_next_osdmap(&osdc->client->monc);
        }
+       reset_changed_osds(osdc);
 }
 
 
@@ -1393,7 +1402,6 @@ void ceph_osdc_handle_map(struct ceph_osd_client *osdc, struct ceph_msg *msg)
                                osdc->osdmap = newmap;
                        }
                        kick_requests(osdc, 0);
-                       reset_changed_osds(osdc);
                } else {
                        dout("ignoring incremental map %u len %d\n",
                             epoch, maplen);