[JDEV] 'JCF' IRC transport dumping core on channel change

[Migs Paraz]

I managed to get the 'JCF' IRC daemon + the patched gabber to work...
I can go to one channel.  But when I switch to another, a core dump:

#0  0x401eafc7 in irc_parse_phase_two (m=0x81fd170, flag=1, arg=0x81fc118, 
    buffer=0x80d146c ":zimdude!zimdude at 203.176.75.45 JOIN :#plug\r\n:upcm.net 353 zimdude = #plug :zimdude @Triumph \r\n:upcm.net 366 zimdude #plug :End of /NAMES list.\r\n", bufsz=144) at irc_out.c:347


347	            u->id = jid_new(u->p, spools(p, channel->id, "/", shahash(u->user), p));

(gdb) p channel->id
Cannot access memory at address 0x14
(gdb) p channel
$3 = 0x0

Debug output:

Tue Nov 13 18:07:35 2001  init.c:31 [irc_phandler_receive] got <iq id='A7' to='#plug at irc.localhost' type='set' from='baloney at localhost/Gabber'><enter xmlns='jabber:iq:conference'><nickname>zimdude</nickname></enter></iq>
Tue Nov 13 18:07:35 2001  init.c:49 [irc_phandler_receive] handing off to mtq
Tue Nov 13 18:07:35 2001  mtq 8157238 leaving to pth
Tue Nov 13 18:07:35 2001  mtq 81674D8 entering from pth
Tue Nov 13 18:07:35 2001  mtq 81674D8 queue call 8201BD8
Tue Nov 13 18:07:35 2001  irc_in.c:190 [irc_parser] found a session for user baloney at localhost/Gabber
Tue Nov 13 18:07:35 2001  irc_in.c:284 [irc_parser] starting a new channel for user baloney at localhost/Gabber
Tue Nov 13 18:07:35 2001  mio.c:1032 mio_write called on x: 0 buffer: JOIN #plug 

Tue Nov 13 18:07:35 2001  mtq 81674D8 leaving to pth
Tue Nov 13 18:07:35 2001  mio.c:600 mio while loop top
Tue Nov 13 18:07:35 2001  mio.c:613 mio while loop, working
Tue Nov 13 18:07:35 2001  mio.c:275 write_dump writing data: JOIN #plug 

Tue Nov 13 18:07:35 2001  mio.c:600 mio while loop top
Tue Nov 13 18:07:35 2001  mio.c:613 mio while loop, working
Tue Nov 13 18:07:35 2001  mio.c:750 MIO read from socket 14: :zimdude!zimdude at 203.176.75.45 JOIN :#plug
:upcm.net 353 zimdude = #plug :zimdude @Triumph 
:upcm.net 366 zimdude #plug :End of /NAMES list.

Tue Nov 13 18:07:35 2001  irc_out.c:279 [irc_parse_phase_two] in phase two: m81FD170 flag:1, arg:81FC118, buffer: :zimdude!zimdude at 203.176.75.45 JOIN :#plug
:upcm.net 353 zimdude = #plug :zimdude @Triumph 
:upcm.net 366 zimdude #plug :End of /NAMES list.
, bufsz:144
Tue Nov 13 18:07:35 2001  log.c:105 <log type='alert' from='irc_out.c:337'>[irc_parse_phase_two] Unable to get channel #plug</log>
Tue Nov 13 18:07:35 2001  deliver.c:460 DELIVER 3:337 <log type='alert' from='irc_out.c:337'>[irc_parse_phase_two] Unable to get channel #plug</log>
Tue Nov 13 18:07:35 2001  deliver.c:651 delivering to instance 'elogger'
20011113T10:07:35: [alert] (irc_out.c:337): [irc_parse_phase_two] Unable to get channel #plug
Tue Nov 13 18:07:35 2001  dnsrv.c:143 dnsrv: Read error on coprocess(1): 2 No such file or directory
Tue Nov 13 18:07:35 2001  dnsrv.c:157 DNSRV CHILD: out of loop.. exiting normal
Segmentation fault (core dumped)


I looked at the code of irc_out.c:

            channel = xhash_get(s->channels, make_lowercase(icpr->next));

            if(icpr->from == NULL)
                continue;
            if (channel == NULL) {
                log_error(ZONE,"[irc_parse_phase_two] Unable to get channel %s", icpr->next);
            }
            ptr = strchr(++icpr->from, '!');
            ptr[0] = '\0';

            /* someone is joining the channel */
            p2 = pool_new();
            u = pmalloco(p2, sizeof(_user));
            u->p = p2;
            u->user = make_lowercase(pstrdup(u->p, icpr->from));
            u->id = jid_new(u->p, spools(p, channel->id, "/", shahash(u->user), p));


So, if channel == NULL, the error is printed but jid_new is called just the
same.  What causes channel to be null?