On 07/07/2008 13:47, Benny Prijono wrote:
> On Mon, Jul 7, 2008 at 12:32 PM, Filippo Zangheri <filippo.zangheri at yahoo.it>
> wrote:
>
>> Debugging info:
>>
>> (gdb) run
>> Starting program:
>> /home/filippo/tesi/pjsip/debug/pjproject-0.9.0/pjsip-apps/bin/samples/stateful_proxy-i686-pc-linux-gnu
>> [Thread debugging using libthread_db enabled]
>> [New Thread -1211930944 (LWP 23669)]
>> 13:29:34.620 os_core_unix.c pjlib 0.9.0-release for POSIX initialized
>> 13:29:34.621 pjlib select() I/O Queue created (0xb7bf4098)
>> 13:29:34.621 sip_endpoint.c Module "mod-msg-print" registered
>> 13:29:34.621 sip_endpoint.c Module "mod-tsx-layer" registered
>> 13:29:34.621 sip_endpoint.c Module "mod-stateful-util" registered
>> 13:29:34.651 udp0x80a2b60 SIP UDP transport started, published address
>> is 82.59.213.185:5060
>> 13:29:34.651 sip_endpoint.c Module "mod-msg-logger" registered
>>
>> Program received signal SIGSEGV, Segmentation fault.
>> [Switching to Thread -1211930944 (LWP 23669)]
>> 0x08077908 in if_enum_by_af (af=2, p_cnt=0xbfe7a298, ifs=0xbfe7a2e0) at
>> ../src/pj/ip_helper_generic.c:93
>> 93 if (ad->sa_family != af) {
>> (gdb)
>>
>
> That would be very strange. Is "ad" NULL or something? While you're there,
> it would be great if you could make sense on what's happening. The
> if_enum_by_af() is quite straightforward so I think it shouldn't be too
> difficult to understand.
Yes, I have evidence of 'ad' being NULL.
I added the following check right after the declaration of that pointer:
if (ad == NULL) continue;
And the segfault disappears. The only different behaviour is that loopback
address 127.0.0.1 is not listed in host alias addresses:
$ ./stateful_proxy-i686-pc-linux-gnu
14:36:53.997 os_core_unix.c pjlib 0.9.0-release for POSIX initialized
14:36:53.998 pjlib select() I/O Queue created (0xb7c1c098)
14:36:53.998 sip_endpoint.c Module "mod-msg-print" registered
14:36:53.998 sip_endpoint.c Module "mod-tsx-layer" registered
14:36:53.998 sip_endpoint.c Module "mod-stateful-util" registered
14:36:54.019 udp0x80a2b60 SIP UDP transport started, published address is 82.59.213.185:5060
14:36:54.019 sip_endpoint.c Module "mod-msg-logger" registered
14:36:54.039 stateful_proxy Proxy started, listening on port 5060
14:36:54.039 stateful_proxy Local host aliases:
14:36:54.039 stateful_proxy 82.59.213.185:5060
14:36:54.039 stateful_proxy 192.168.2.20:5060
14:36:54.039 stateful_proxy asus-laptop:5060
14:36:54.039 stateful_proxy localhost:5060
14:36:54.039 sip_endpoint.c Module "mod-stateful-proxy" registered
14:36:54.039 sip_endpoint.c Module "mod-transaction-user" registered
Menu:
q quit
d dump status
dd dump detailed status
q
14:36:58.025 sip_endpoint.c Module "mod-transaction-user" unregistered
14:36:58.025 sip_endpoint.c Module "mod-stateful-util" unregistered
14:36:58.025 sip_endpoint.c Module "mod-stateful-proxy" unregistered
14:36:58.025 sip_transactio Stopping transaction layer module
14:36:58.026 sip_transactio Transaction layer module destroyed
14:36:58.026 sip_endpoint.c Module "mod-tsx-layer" unregistered
14:36:58.026 sip_endpoint.c Module "mod-msg-print" unregistered
14:36:58.026 sip_endpoint.c Module "mod-msg-logger" unregistered
14:36:58.029 sip_endpoint.c Endpoint 0x808b584 destroyed
Now, facts are:
* getifaddrs(&ifap) = 0 (OK);
* it = ifap != NULL (OK);
* ad = it->ifa_addr = NULL (not OK);
So it seems that getifaddrs() fills up a struct ifaddrs whose ifa_addr
member is NULL. By the way, I can't find the implementation of that
function. Where is it?
Thanks!
--
Filippo Zangheri
GPG key ID: 0x6C1F2F2F
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: OpenPGP digital signature
Url : http://lists.pjsip.org/pipermail/pjsip_lists.pjsip.org/attachments/20080707/2964d45a/attachment.bin
