Bug in wsp_unit.c ?

Hi, list.

Yury Mikhienko wrote:

>  
> I changed kannel source for dumping the next information:
> 
> in wapbox.log now:
> 
> 2002-11-12 09:21:45 [1] DEBUG: WSP 227/4150: New method state REPLYING
> 2002-11-12 09:21:45 [1] DEBUG: WSP 227: New state CONNECTED
> 2002-11-12 09:21:45 [2] WARNING: WSP PDU with unknown type -1
> 2002-11-12 09:21:45 [2] WARNING: Incoming Octetstr has been dumped!!!
> 2002-11-12 09:21:45 [2] DEBUG:  Octet string at 0x89a77c8:
> 2002-11-12 09:21:45 [2] DEBUG:    len:  74
> 2002-11-12 09:21:45 [2] DEBUG:    size: 76
> 2002-11-12 09:21:45 [2] DEBUG:    immutable: 0
> 2002-11-12 09:21:45 [2] DEBUG:    data: 73 5a 12 01 10 13 30 04   sZ....0.
> 2002-11-12 09:21:45 [2] DEBUG:    data: 80 86 a0 00 04 81 92 a7   ........
> 2002-11-12 09:21:45 [2] DEBUG:    data: 60 02 82 00 02 83 05 02   `.......
> 2002-11-12 09:21:45 [2] DEBUG:    data: 84 01 80 80 83 99 81 ea   ........
> 2002-11-12 09:21:45 [2] DEBUG:    data: a9 4e 6f 6b 69 61 37 36   .Nokia76
> 2002-11-12 09:21:45 [2] DEBUG:    data: 35 30 2f 31 2e 30 20 53   50/1.0 S
> 2002-11-12 09:21:45 [2] DEBUG:    data: 79 6d 62 69 61 6e 4f 53   ymbianOS
> 2002-11-12 09:21:45 [2] DEBUG:    data: 2f 36 2e 31 20 53 65 72   /6.1 Ser
> 2002-11-12 09:21:45 [2] DEBUG:    data: 69 65 73 36 30 2f 30 2e   ies60/0.
> 2002-11-12 09:21:45 [2] DEBUG:    data: 39 00                     9.
> 2002-11-12 09:21:45 [2] DEBUG:  Octet string dump ends.
> ################################################################################
> 2002-11-12 09:21:45 [6] DEBUG: Thread 6 (gw/wap-appl.c:main_thread) terminates.
> ################################################################################
> 
> How can I solve this? May be I shall apply another SAR patch (which one?)?
> Or this problen can be solved another way?
> 
> Now I solved the problem with dirty hack the  gwthread-pthread.c:
> in:
> static void *new_thread(void *arg)
> {
>     int ret;
>     struct new_thread_args *p = arg;
> ....
>     (p->func)(p->arg);
> 
> /* Changed by me */
>    do
>    {
>      lock();
> 
>      
>      debug("gwlib.gwthread", 0, "Thread %ld (%s) terminates. ",
>           p->ti->number, p->ti->name);
>      debug("gwlib.gwthread", 0, "Try up Thread %ld (%s). ",
>           p->ti->number, p->ti->name);
>      
>      unlock();
>      
>      ret = pthread_setspecific(tsd_key, p->ti);
>      if (ret != 0) {
>         panic(ret, "gwthread-pthread: pthread_setspecific failed");
>      }
>      
>      (p->func)(p->arg);
>      
>    }while(p->ti->number==6 && !strcmp(p->ti->name,"gw/wap-appl.c:main_thread"));
> //******************************
> ....
> };
> 
> It works, but it is not a good way :((
> 
> Thanx.
> 
> 


Seems that there is bug in wsp_unit.c

static void main_thread(void *arg) {
         WAPEvent *e;
         WAPEvent *newevent;

         while (run_status == running && (e = list_consume(queue)) != NULL) {
                 wap_event_assert(e);
                 switch (e->type) {
                 case T_DUnitdata_Ind:
                         newevent = unpack_datagram(e);
/* MUST BE INSERTED  if (newevent != NULL) */
	                	dispatch_to_appl(newevent);

unpack_datagram returns NULL
(> 2002-11-12 09:21:45 [2] WARNING: WSP PDU with unknown type -1)
and dispatch NULL to wap-appl.c

So later in wap-appl.c we have (main_thread):
while (run_status == running && (ind = list_consume(queue)) != NULL) {
}

and list_consume breaks the loop and

 > 2002-11-12 09:21:45 [6] DEBUG: Thread 6 (gw/wap-appl.c:main_thread) terminates.

So any following requests will not be processed.

I think we must insert check for NULL (see above, line /* MUST BE INSERTED)

-- 
Vjacheslav Chekushin                                mailto:[email protected]
Latvian Mobile Phone Company                        http://www.lmt.lv