[Users] record_route_preset(): No memory left

[Bastian Schern]

Hi,

I call record_route_present() directly from the script:

--- snip ---
#----------------------------------------------------------------
# record route section
#----------------------------------------------------------------
if( method == "INVITE" && client_nat_test( "3" ) ) {
     xlog( "L_INFO", "$Tf: <$fu>;tag=$ft : record_route_preset( 
\"217.160.188.74:5060;nat=yes\" )\n" );
     record_route_preset( "217.160.188.74:5060;nat=yes" );
} else if( method != "REGISTER" ) {
     xlog( "L_INFO", "$Tf: <$fu>;tag=$ft : record_route()\n" );
     record_route();
}
--- snap ---

I don't see what's going wrong.

Regards
	Bastian

Bogdan-Andrei Iancu schrieb:
> Hi Bastian,
> 
> I made a fast overview/testing and I found no problem with 
> record_route_preset() function. Are you calling it directly from script 
> or from another module....the bogus length may by due an invalid param.
> 
> regards,
> bogdan
> 
> 
> Bastian Schern wrote:
> 
>> Hi Bogdan,
>>
>> I compiled OpenSER with memory debugging and all other settings you 
>> recommend to me. After that I got the following debug output:
>>
>> --- snip ---
>> [...]
>>  0(31223) xl_printf: final buffer length 125
>>  0(31223) Wed Feb  1 00:31:43 2006: 
>> <sip:00045977408 at sipbase.de>;tag=77lh74wsus : record_route_preset( 
>> "217.160.188.74:5060;nat=yes" )
>>  0(31223) qm_malloc(0x8114d40, 32) called from data_lump.c: 
>> anchor_lump(351)
>>  0(31223) qm_malloc(0x8114d40, 32) returns address 0x815a390 frag. 
>> 0x815a378 (size=32) on 1 -th hit
>>  0(31223) qm_malloc(0x8114d40, 1074638763) called from record.c: 
>> record_route_preset(373)
>>  0(31223) record_route_preset(): No memory left
>>  0(31223) parse_headers: flags=200
>> [...]
>> --- snip ---
>>
>> It looks like qm_malloc tries to allocate 1074638763 Bytes (1024 MB).
>> I think this is really strange!
>>
>> In the memory dump I don't see something strange.
>>
>> Regards
>>     Bastian
>>
>>
>> Bogdan-Andrei Iancu schrieb:
>>
>>> Hi,
>>>
>>> look like something is not right - please compile the memory debug 
>>> support (in Makefile.defs set DBG_QM_MALLOC define and remove 
>>> F_MALLOC - Note: you need to recompile everything).
>>>
>>> run in no fork mode with debug=9 and wait for the error to appear.
>>> first, see how much memory the malloc tries to grab.
>>> second, hit the process with SIGUSR1 to force memory dump - see if 
>>> there is something strange there.
>>>
>>> regards,
>>> bogdan
>>>
>>> Bastian Schern wrote:
>>>
>>>> Bogdan-Andrei Iancu schrieb:
>>>>
>>>>> Hi Bastian,
>>>>>
>>>>> looks like you openser runs out of private memory (pkg memory) - 
>>>>> this may happen either because of insufficient  mem, either due a 
>>>>> mem leak.
>>>>>
>>>>> Does is happens after running for a long time? or quite soon after 
>>>>> start?
>>>>>
>>>>
>>>> It will occur quite soon after starting OpenSER.
>>>>
>>>>
>>>>> first try to increase the pkg memory  - see config.h file, the 
>>>>> PKG_MEM_POOL_SIZE define. By default, the size is of 1 M. If the 
>>>>> problem persists, please report back.
>>>>>
>>>>
>>>> I changed PKG_MEM_POOL_SIZE to 2*1024*1024 but it is still the same. 
>>>> Is it still to low?
>>>>
>>>> Regards
>>>>     Bastian
>>>>
>>>>
>>>>> regards,
>>>>> bogdan
>>>>>
>>>>> Bastian Schern wrote:
>>>>>
>>>>>> Hello to all,
>>>>>>
>>>>>> I'm using the latest CVS 1_0_0 and I got the following message in 
>>>>>> the log after a "record_route_preset( 
>>>>>> "213.191.xxx.xxx:5060;nat=yes" )":
>>>>>>
>>>>>> record_route_preset(): No memory left
>>>>>>
>>>>>> Why this comes?
>>>>>>
>>>>>> Regards
>>>>>>     Bastian
>>>>>


____________
Virus checked by G DATA AntiVirusKit
Version: AVK 16.5183 from 30.01.2006
Virus news: www.antiviruslab.com