Hello ALL,

Someone knows why at line 149 in .../kamailio-3.1.2/modules_k/db_text/dbt_tb.c the shm_malloc function sometimes allocates memory to dtp->prev but sometimes doesn't?
I'm asking this question because when the subscriber table is being loading to shared memory in db_text module, dtp->prev is allocated but on version, dispatcher, location table it doesn't... Analysing the source code I can't see any reason to this behavior.

The allocation of dtp->prev to subscriber table is generating a core at db_text module.

Best Regards

2011/9/27 Bruno Bresciani <bruno.bresciani@gmail.com>
Hello Daniel,

I'm still analyzing the source code of db_text module and I can't get find the reason of core when variable _tbc->prev is accessed. Did you obtain some conclusion about this core?

PS: In kamailio 1.5.0 I can't get replicate this core

Best Regards


2011/9/26 Bruno Bresciani <bruno.bresciani@gmail.com>
Hello,

Follows the output of p *_tbc->prev:

(gdb) p *_tbc->prev
Cannot access memory at address 0x4c58586e


Best Regards




2011/9/24 Daniel-Constantin Mierla <miconda@gmail.com>
Hello,

thanks, the only possibility then is that the previous item in the list of tables has something wrong in it, send also the output for:

p *_tbc->prev

Actually, from now I see the _tmb->prev is 0x4c58586e and _tbc is 0xb6175d20, so the pointers are in different memory zones, it could be pkg one and shm the other. But lets see first the output of p *_tbc->prev

Cheers,
Daniel


On 9/23/11 2:11 PM, Bruno Bresciani wrote:
Follows the output:

p _tbc
$1 = (dbt_table_p) 0xb6175d20

(gdb) p *_tbc
$312 = {dbname = {s = 0xb616d780 "/axs/cfg/db_text", len = 16}, name = {s = 0xb616e288 "subscriber", len = 10}, hash = 6, mark = 1316722939, flag = 0,
  auto_col = -1, auto_val = 0, nrcols = 10, cols = 0xb616f310, colv = 0xb6175e70, nrrows = 58, rows = 0xb616e940, mt = 1316722962, next = 0x0,
  prev = 0x4c58586e}




2011/9/23 Daniel-Constantin Mierla <miconda@gmail.com>


On 9/23/11 1:57 PM, Bruno Bresciani wrote:
Sorry... but I didn't understand how I get the output of:

p _tbc
p *_tbc
run these commands inside gdb.

Daniel




Best Regards


2011/9/23 Daniel-Constantin Mierla <miconda@gmail.com>
Hello,

interesting ... give also the output of:

p _tbc
p *_tbc

Cheers,
Daniel


On 9/23/11 1:27 PM, Bruno Bresciani wrote:
Hello,

Doesn't exist a exact often that the table subscriber is changed, it can be changed at any time and almost always the core is generated. I was thinking the same thing that you, the deletion of the table is done without a lock, but the function dbt_db_get_table that call dbt_db_del_table does this lock previously...

Follows the output of 'bt full' from gdb:

Program terminated with signal 11, Segmentation fault.
#0  0x008b3603 in dbt_db_del_table (_dc=0xb60fde60, _s=0xbfc94380, sync=0) at dbt_lib.c:238
238     dbt_lib.c: Arquivo ou diretório não encontrado.
        in dbt_lib.c
(gdb) bt full
#0  0x008b3603 in dbt_db_del_table (_dc=0xb60fde60, _s=0xbfc94380, sync=0) at dbt_lib.c:238
        _tbc = (dbt_table_p) 0xb6175d20
        hash = 6
#1  0x008b3f38 in dbt_db_get_table (_dc=0xb60fde60, _s=0xbfc94380) at dbt_lib.c:300
        _tbc = (dbt_table_p) 0xb6175d20
        hash = 6
#2  0x008ae12b in dbt_query (_h=0x82e3c90, _k=0xbfc94378, _op=0x0, _v=0xbfc94348, _c=0x82e3e50, _n=1, _nc=2, _o=0x0, _r=0xbfc94390) at dbt_base.c:200
        _tbc = <value optimized out>
        _drp = <value optimized out>
        _dres = <value optimized out>
        lkey = <value optimized out>
        lres = (int *) 0x0
        _o_k = (db_key_t *) 0x0
        _o_op = 0x0
        _o_n = <value optimized out>
        _o_l = <value optimized out>
        _o_nc = <value optimized out>
#3  0x001ae088 in digest_authenticate (msg=0x82df8f8, realm=<value optimized out>, tname=<value optimized out>, hftype=HDR_AUTHORIZATION_T) at authorize.c:98
        cred = <value optimized out>
        keys = {0x1b14c8, 0x1b14d0}
        vals = {{type = DB1_STR, nul = 0, free = 165777064, val = {int_val = 136948130, ll_val = 17316817314, double_val = 8.5556445301562903e-314,
      time_val = 136948130,
      string_val = 0x829a9a2 "3022\", realm=\"192.168.166.199\", nonce=\"TnuaJ057mPskvCOYDgyAURuuottiYy7cIAQKLUA=\", cnonce=\"4VwQFl/7Ei+IWecbrZ9rug\", algorithm=MD5, uri=\"sip:192.168.166.199\", response=\"45b23bc0e48592fb57b1ebc87f0e3dbc\""..., str_val = {
        s = 0x829a9a2 "3022\", realm=\"192.168.166.199\", nonce=\"TnuaJ057mPskvCOYDgyAURuuottiYy7cIAQKLUA=\", cnonce=\"4VwQFl/7Ei+IWecbrZ9rug\", algorithm=MD5, uri=\"sip:192.168.166.199\", response=\"45b23bc0e48592fb57b1ebc87f0e3dbc\""..., len = 4}, blob_val = {
        s = 0x829a9a2 "3022\", realm=\"192.168.166.199\", nonce=\"TnuaJ057mPskvCOYDgyAURuuottiYy7cIAQKLUA=\", cnonce=\"4VwQFl/7Ei+IWecbrZ9rug\", algorithm=MD5, uri=\"sip:192.168.166.199\", response=\"45b23bc0e48592fb57b1ebc87f0e3dbc\""..., len = 4}, bitmap_val = 136948130}}, {type = DB1_STR, nul = 0, free = 8200,
    val = {int_val = 137215560, ll_val = 64561725000, double_val = 3.1897730358749953e-313, time_val = 137215560, string_val = 0x82dbe48 "192.168.166.199",
      str_val = {s = 0x82dbe48 "192.168.166.199", len = 15}, blob_val = {s = 0x82dbe48 "192.168.166.199", len = 15}, bitmap_val = 137215560}}}
        result = {s = 0x0, len = 0}
        n = <value optimized out>
        ha1 = "\000\000\000\000\220GÉ¿\000\000\000\000ï=\000\000ÒBÉ¿L\215á\t\000\000\000\000(Óà\tÔBÉ¿F\000\000\000yv¯\000L\215á\t", '\0' <repeats 12 times>, "ÿÿÿÿ\a\000\000\000a{¯\000^{¯\000\000\000\000\000\000\000\000\000ÿÿÿÿ+\000\000\000\000\000\000\000\001\000\000\0008DÉ¿\210\033-\bÈVÉ¿HDÉ¿&\205\017\bÈVÉ¿øø-\b8DÉ¿P\020-\b", '\0' <repeats 32 times>, " e¯\000TDÉ¿`ñ\000\000ÐÑ\000\000\\\222¯d\030\000\000\000\000\000\000\000»¬£\000;\216á\t 1±\000\000 \000\000ÔCÉ¿8\216á\t\bÔ\r¶"...
        h = (struct hdr_field *) 0x82e0398
        domain = {s = 0x82dbe48 "192.168.166.199", len = 15}
        table = {s = 0x82d8f60 "subscriber", len = 10}
        result = (db1_res_t *) 0x0
        ret = 0


Best Regards

2011/9/23 Daniel-Constantin Mierla <miconda@gmail.com>
Hello,

can you send the output of 'bt full' from gdb?

How often the subscriber table is changed? I see in this case the deletion of the table is done without a lock.

Cheers,
Daniel


On 9/22/11 9:44 PM, Bruno Bresciani wrote:
Hi All,

Kamailio 3.1.2 generate a core in db_text module when subscriber table is updated and after this action some SIP message require authentication process...

Someone Can Help me to understand why this core is happening? Below  is part of kamailio's trace and gdb core too.


Kamailio's trace:

Sep 22 16:35:54 sswpst00 /home2/local/kamailio/sbin/kamailio[2327]: DEBUG: auth_db [authorize.c:239]: realm value [192.168.166.199]
Sep 22 16:35:54 sswpst00 /home2/local/kamailio/sbin/kamailio[2327]: DEBUG: auth [api.c:95]: auth: digest-algo: MD5 parsed value: 1
Sep 22 16:35:54 sswpst00 /home2/local/kamailio/sbin/kamailio[2327]: DEBUG: db_text [dbt_file.c:78]: [subscriber] was updated
Sep 22 16:35:54 sswpst00 /home2/local/kamailio/sbin/kamailio[2324]: ALERT: <core> [main.c:741]: child process 2327 exited by a signal 11
Sep 22 16:35:54 sswpst00 /home2/local/kamailio/sbin/kamailio[2324]: ALERT: <core> [main.c:744]: core was generated
Sep 22 16:35:54 sswpst00 /home2/local/kamailio/sbin/kamailio[2324]: INFO: <core> [main.c:756]: INFO: terminating due to SIGCHLD
Sep 22 16:35:54 sswpst00 /home2/local/kamailio/sbin/kamailio[2335]: INFO: <core> [main.c:807]: INFO: signal 15 received c
Sep 22 16:35:54 sswpst00 /home2/local/kamailio/sbin/kamailio[2335]: DEBUG: <core> [main.c:818]: Memory status (pkg):

gdb core trace:

Core was generated by `/home2/local/kamailio/sbin/kamailio -P /var/run/kamailio.pid'.
Program terminated with signal 11, Segmentation fault.
#0  0x00f21603 in dbt_db_del_table (_dc=0xb6071e60, _s=0xbf837040, sync=0) at dbt_lib.c:238
238     dbt_lib.c: Arquivo ou diretório não encontrado.
        in dbt_lib.c


Best Regards


_______________________________________________
SIP Express Router (SER) and Kamailio (OpenSER) - sr-users mailing list
sr-users@lists.sip-router.org
http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-users


-- 
Daniel-Constantin Mierla -- http://www.asipto.com
Kamailio Advanced Training, Oct 10-13, Berlin: http://asipto.com/u/kat
http://linkedin.com/in/miconda -- http://twitter.com/miconda



_______________________________________________
SIP Express Router (SER) and Kamailio (OpenSER) - sr-users mailing list
sr-users@lists.sip-router.org
http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-users


-- 
Daniel-Constantin Mierla -- http://www.asipto.com
Kamailio Advanced Training, Oct 10-13, Berlin: http://asipto.com/u/kat
http://linkedin.com/in/miconda -- http://twitter.com/miconda



_______________________________________________
SIP Express Router (SER) and Kamailio (OpenSER) - sr-users mailing list
sr-users@lists.sip-router.org
http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-users


-- 
Daniel-Constantin Mierla -- http://www.asipto.com
Kamailio Advanced Training, Oct 10-13, Berlin: http://asipto.com/u/kat
http://linkedin.com/in/miconda -- http://twitter.com/miconda



_______________________________________________
SIP Express Router (SER) and Kamailio (OpenSER) - sr-users mailing list
sr-users@lists.sip-router.org
http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-users


-- 
Daniel-Constantin Mierla -- http://www.asipto.com
Kamailio Advanced Training, Oct 10-13, Berlin: http://asipto.com/u/kat
http://linkedin.com/in/miconda -- http://twitter.com/miconda