Re: Possible race condition when clear_old_concat_parts is called

As no objections here,

Please commit to CVS.

Vincent.

--
Telemaque - 06560 SOPHIA-ANTIPOLIS - (FR)
Service Technique/Reseau - NOC
Developpement SMS/MMS/Kiosques
http://www.telemaque.fr/
[email protected]
Tel : +33 4 92 90 99 84 (fax 9142)
----- Original Message ----- 
From: "Vincent CHAVANIS" <[email protected]>
To: <[email protected]>
Sent: Monday, October 15, 2007 1:42 PM
Subject: Re: Possible race condition when clear_old_concat_parts is called


>
> Here is the patch to fix this issue (thanks Paul)
>
> diff -rauw /gateway-cvs/gw/bb_smscconn.c /gateway/gw/bb_smscconn.c
> --- /gateway-cvs/gw/bb_smscconn.c       2007-06-12 16:08:15.000000000 
> +0200
> +++ /gateway/gw/bb_smscconn.c   2007-10-15 10:40:57.000000000 +0200
> @@ -1254,7 +1254,7 @@
>     Octstr *key;
>
>     /* not initialised, go away */
> -    if (incoming_concat_msgs != NULL)
> +    if (incoming_concat_msgs == NULL)
>         return;
>
>     debug("bb.sms.splits", 0, "clear_old_concat_parts called");
>
>
> Vincent
>
>
> ----- Original Message ----- 
> From: "Vincent CHAVANIS" <[email protected]>
> To: <[email protected]>
> Cc: "Paul Bagyenda" <[email protected]>
> Sent: Monday, October 15, 2007 1:19 AM
> Subject: Possible race condition when clear_old_concat_parts is called
>
>
>> Only when sms-combine-concatenated-mo is set to 0,
>> I experienced a segfault in bearbox (bt full traces bellow)
>> The line where it crashes is :  keys = dict_keys(incoming_concat_msgs) 
>> (gw/bb_smscconn.c)
>> I've added a printf after it and it segfaults without showing my debug 
>> message,
>> so we can be sure dict_keys segfault.
>> Maybe we can have here a possible race condition.
>>
>> @Paul,Alex, have you experiencing that issue before ?
>>
>> Vincent.
>>
>> refs:
>> 2007-10-10 15:43:50.191 [22125] [67] DEBUG: HTTP[SMS_7]: Destroying 
>> client information
>> 2007-10-10 15:43:50.192 [22125] [101] DEBUG: boxc_receiver: got ack
>> 2007-10-10 15:43:51.001 [22125] [93] DEBUG: clear_old_concat_parts called
>> Erreur de segmentation (core dumped)
>>
>>
>> Program terminated with signal 11, Segmentation fault.
>> #0  0x000000000049622f in lock (dict=0x0) at gwlib/dict.c:127
>> 127         mutex_lock(dict->lock);
>> (gdb) bt full
>> #0  0x000000000049622f in lock (dict=0x0) at gwlib/dict.c:127
>>        __func__ = "lock"
>> #1  0x0000000000496990 in dict_keys (dict=0x0) at gwlib/dict.c:355
>>        list = (List *) 0x941a20
>>        item = (Item *) 0x470ce2a2
>>        i = 2049294384
>>        j = 4858217
>>        __func__ = "dict_keys"
>> #2  0x0000000000417b25 in clear_old_concat_parts () at 
>> gw/bb_smscconn.c:1263
>>        keys = (List *) 0x0
>>        key = (Octstr *) 0x7a25c0f0
>>        __func__ = "clear_old_concat_parts"
>> #3  0x0000000000415d17 in sms_router (arg=0x0) at gw/bb_smscconn.c:527
>>        msg = (Msg *) 0x0
>>        startmsg = (Msg *) 0x0
>>        newmsg = (Msg *) 0x0
>>        ret = 0
>>        concat_mo_check = 1192029069
>> #4  0x00000000004988e4 in new_thread (arg=0x98eb40) at 
>> gwlib/gwthread-pthread.c:362
>>        ret = 0
>>        p = (struct new_thread_args *) 0x98eb40
>> #5  0x0000003f312062f7 in start_thread () from /lib64/libpthread.so.0
>> No symbol table info available.
>> #6  0x0000003f306d0fbd in clone () from /lib64/libc.so.6
>> No symbol table info available.
>>
>> --
>> Telemaque - 06560 SOPHIA-ANTIPOLIS - (FR)
>> Service Technique/Reseau - NOC
>> Developpement SMS/MMS/Kiosques
>> http://www.telemaque.fr/
>> [email protected]
>> Tel : +33 4 92 90 99 84 (fax 9142)
>
>
>
>