[ovs-dev] [PATCH v2] ovn-controller: Fix crash when sending GARP when openflow disconnection.

[Guoshuai Li]

This is call stack:
Program received signal SIGABRT, Aborted.
1  0x00007ffff6a4f8e8 in __GI_abort () at abort.c:90
2  0x00000000004765d6 in ofputil_protocol_to_ofp_version (protocol=<optimized out>) at lib/ofp-util.c:769
3  0x000000000047c19e in ofputil_encode_packet_out (po=po at entry=0x7fffffffa0e0, protocol=<optimized out>) at lib/ofp-util.c:7060
4  0x0000000000410870 in send_garp (garp=0x83cfe0, current_time=current_time at entry=1200375400) at ovn/controller/pinctrl.c:1738
5  0x000000000041430f in send_garp_run (active_tunnels=<optimized out>, local_datapaths=0x7fffffffc0a0, chassis_index=<optimized out>, chassis=0x8194d0, br_int=<optimized out>, ctx=0x7fffffffc080) at ovn/controller/pinctrl.c:2069

Signed-off-by: Guoshuai Li <ligs at dtdream.com>
Acked-by: Mark Michelson <mmichels at redhat.com>
---
 ovn/controller/pinctrl.c | 38 ++++++++++++++++++++------------------
 tests/ovn.at             | 13 +++++++++++++
 2 files changed, 33 insertions(+), 18 deletions(-)

diff --git a/ovn/controller/pinctrl.c b/ovn/controller/pinctrl.c
index 8a11672a1..e34916f5c 100644
--- a/ovn/controller/pinctrl.c
+++ b/ovn/controller/pinctrl.c
@@ -1073,27 +1073,29 @@ pinctrl_run(struct controller_ctx *ctx,
 
     rconn_run(swconn);
 
-    if (rconn_is_connected(swconn)) {
-        if (conn_seq_no != rconn_get_connection_seqno(swconn)) {
-            pinctrl_setup(swconn);
-            conn_seq_no = rconn_get_connection_seqno(swconn);
-            flush_put_mac_bindings();
-        }
-
-        /* Process a limited number of messages per call. */
-        for (int i = 0; i < 50; i++) {
-            struct ofpbuf *msg = rconn_recv(swconn);
-            if (!msg) {
-                break;
-            }
+    if (!rconn_is_connected(swconn)) {
+        return;
+    }
 
-            const struct ofp_header *oh = msg->data;
-            enum ofptype type;
+    if (conn_seq_no != rconn_get_connection_seqno(swconn)) {
+        pinctrl_setup(swconn);
+        conn_seq_no = rconn_get_connection_seqno(swconn);
+        flush_put_mac_bindings();
+    }
 
-            ofptype_decode(&type, oh);
-            pinctrl_recv(oh, type, ctx);
-            ofpbuf_delete(msg);
+    /* Process a limited number of messages per call. */
+    for (int i = 0; i < 50; i++) {
+        struct ofpbuf *msg = rconn_recv(swconn);
+        if (!msg) {
+            break;
         }
+
+        const struct ofp_header *oh = msg->data;
+        enum ofptype type;
+
+        ofptype_decode(&type, oh);
+        pinctrl_recv(oh, type, ctx);
+        ofpbuf_delete(msg);
     }
 
     run_put_mac_bindings(ctx);
diff --git a/tests/ovn.at b/tests/ovn.at
index 00d26e757..8ee3bf0b5 100644
--- a/tests/ovn.at
+++ b/tests/ovn.at
@@ -3566,6 +3566,19 @@ AT_CHECK([ovs-vsctl add-port br-int localvif1 -- set Interface localvif1 externa
 echo "fffffffffffff0000000000108060001080006040001f00000000001c0a80102000000000000c0a80102" > expected
 OVN_CHECK_PACKETS([hv/snoopvif-tx.pcap], [expected])
 
+# Check GARP packet when restart openflow connection.
+as hv
+OVS_APP_EXIT_AND_WAIT([ovs-vswitchd])
+
+OVS_WAIT_UNTIL([grep -c "waiting 4 seconds before reconnect" hv/ovn-controller.log])
+
+as hv
+start_daemon ovs-vswitchd --enable-dummy=system -vvconn -vofproto_dpif -vunixctl
+
+# Wait for packet to be received.
+echo "fffffffffffff0000000000108060001080006040001f00000000001c0a80102000000000000c0a80102" > expected
+OVN_CHECK_PACKETS([hv/snoopvif-tx.pcap], [expected])
+
 # Delete the localnet ports.
 AT_CHECK([ovs-vsctl del-port localvif1])
 AT_CHECK([ovn-nbctl lsp-del ln_port])
-- 
2.13.2.windows.1