[sr-dev] Fwd: crash & 100% CPU usage problem with kamailio 3.1.2

Zunnun zunnun at gmail.com
Mon May 2 23:13:27 CEST 2011


Hi Andrei,



Yes this is kamailio 3.1.2 code. I have put some DBG & LOG inside the tcp
code to debug this problem that’s why the line number does not match.



It happens frequently – you need to run stress for a day or more. We are
using Microsoft OCS tools for the stress.



Here’s a problem – it creates a circular linked list



0x00000000004e8d5b in _tcpconn_find (id=<value optimized out>,
ip=0x7fff6d7393c0, port=5061, l_ip=0x7fff6d739280, l_port=0) at
tcp_main.c:1491

(gdb) list

1486    #ifdef EXTRA_DEBUG

1487                            DBG("a=%p, c=%p, c->id=%d, alias port= %d
port=%d\n", a, a->parent,

1488                                            a->parent->id, a->port,
a->parent->rcv.src_port);

1489                            print_ip("ip=",&a->parent->rcv.src_ip,"\n");

1490    #endif



(gdb) print a

$1 = (struct tcp_conn_alias *) 0x2aed7ed00220

(gdb) print *a

$2 = {parent = 0x2aed7ed000e0, next = 0x2aed7ebfc568, prev = 0x2aed7ebfc568,
port = 5061, hash = 2739}

(gdb) print a->next

$3 = (struct tcp_conn_alias *) 0x2aed7ebfc568

(gdb) print *a->next

$4 = {parent = 0x2aed7ebfc428, next = 0x2aed7ed00220, prev = 0x0, port =
5061, hash = 2739}

(gdb) print *a->next->next

$5 = {parent = 0x2aed7ed000e0, next = 0x2aed7ebfc568, prev = 0x2aed7ebfc568,
port = 5061, hash = 2739}

(gdb) print *a->next->next->next

$6 = {parent = 0x2aed7ebfc428, next = 0x2aed7ed00220, prev = 0x0, port =
5061, hash = 2739}

(gdb) print *a->next->next->next->next

$7 = {parent = 0x2aed7ed000e0, next = 0x2aed7ebfc568, prev = 0x2aed7ebfc568,
port = 5061, hash = 2739}

(gdb) print *a->next->next->next->next->next

$8 = {parent = 0x2aed7ebfc428, next = 0x2aed7ed00220, prev = 0x0, port =
5061, hash = 2739}

(gdb) print *a->next->next->next->next->next->next

$9 = {parent = 0x2aed7ed000e0, next = 0x2aed7ebfc568, prev = 0x2aed7ebfc568,
port = 5061, hash = 2739}

(gdb) print c->aliases

$10 = 3



Hope this helps.



I have to check whether I have the core dumps . If I have I’ll send you.



Regards

-Zunnun
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.sip-router.org/pipermail/sr-dev/attachments/20110502/5f6281dd/attachment.htm>


More information about the sr-dev mailing list