Hi all,
I meet a crash problem in openvswitch-2.3.0.
The ovs run pretty well for long time. But I add a packet_in limit setting
'ovs-vsctl set controller 60ef551e-6dab-49a6-803b-80b880f6ff1a
controller_burst_limit=200 controller_rate_limit=200' the ovs crashed as
followed
(gdb) bt
#0 0x000000000047b154 in ofpmsg_update_length (buf=0x2a7b2e0) at
lib/ofp-msgs.c:836
#1 0x00000000004b27eb in do_send (vconn=0x1e05a70, msg=0x2a7b2e0) at
lib/vconn.c:679
#2 0x00000000004a27fe in try_send (rc=0x1c6b300) at lib/rconn.c:1149
#3 0x00000000004a2dd4 in rconn_send__ (rc=0x1c6b300, b=0x2a7b2e0,
counter=0x1c6b5d0) at lib/rconn.c:767
#4 0x00000000004a2e8e in rconn_send_with_limit (rc=0x1c6b300, b=0x2a7b2e0,
counter=0x1c6b5d0, queue_limit=100) at lib/rconn.c:823
#5 0x000000000043771b in do_send_packet_ins (ofconn=0x1c6b480,
txq=0x7ffff052c700) at ofproto/connmgr.c:1756
#6 0x000000000043941b in ofconn_run (mgr=0x1c1ec80, handle_openflow=0x41e370
<handle_openflow>) at ofproto/connmgr.c:1387
#7 connmgr_run (mgr=0x1c1ec80, handle_openflow=0x41e370 <handle_openflow>) at
ofproto/connmgr.c:339
#8 0x000000000041aa31 in ofproto_run (p=0x1c0dba0) at ofproto/ofproto.c:1543
#9 0x00000000004059dc in bridge_run__ () at vswitchd/bridge.c:2255
#10 0x000000000040c967 in bridge_run () at vswitchd/bridge.c:2307
#11 0x000000000040e025 in main (argc=11, argv=0x7ffff052cb98) at
vswitchd/ovs-vswitchd.c:116
void ofpmsg_update_length(struct ofpbuf *buf)
{
struct ofp_header *oh = ofpbuf_at_assert(buf, 0, sizeof *oh);
oh->length = htons(ofpbuf_size(buf)); //crashed on this line
}
(gdb) p *buf
$33 = {base_ = 0x20, data_ = 0x51, size_ = 47931072, allocated = 0, frame =
0x0, l2_5_ofs = 65535, l3_ofs = 65535, l4_ofs = 65535,
source = OFPBUF_MALLOC, list_node = {prev = 0x1c6b360, next = 0x1c6b360}}
I think there is a memory write-overflow problem in controller_rate_limit mode.
Is there a bug for this?
BR
Xu Wen
_______________________________________________
discuss mailing list
[email protected]
http://openvswitch.org/mailman/listinfo/discuss
