[Serdev] Error with MTU size in SER with MIPv6

[José Ángel Calvo]

Thank you, Nils, but we think the "kernel panic" is caused by SER, 
because the MIPv6 daemon continues its execution and doesn't die. The 
computer doesn't die either, and the  SER is the only module that is 
died. We have tested in two different PCs, therefore, we already have 
trying with several networks card/driver. A E100 network card and a 
laptop network card.
When MIPv6 stack is not present, all is ok, but when it is present, 
then, the SER module dies.

P.D:There isn't a kernel panic in logs, it's a kernel bug. The computer 
doesn't die.

Thanks for your help.

Jose Angel Calvo.


Nils Ohlmeier wrote:
> Just to be sure: did you recognized that there is a kernel panic in your logs?
> Maybe that is the real reason for your problems. Just to be safe I would try 
> another network card/driver.
>
>   Nils
>
> On Wednesday 10 January 2007 11:49, José Ángel Calvo wrote:
>   
>> Thank you for your help and sorry for my English.
>>
>> I have a problem with mipv6 and a server SIP, SER (SIP Express Router).
>> The mipv6 path is  "mipv6-2.0.2" from http://www.mobile-ipv6.org/  ,in 
>> kernel  2.6.16.24-Ubuntu.
>>
>> The problem occurs when a packet is too big (it must be
>> fragmented), and go to the node what makes function of P-CSCF (Proxy in
>> SIP) and Correspondent Node.
>>
>> If the P-CSCF is not a CN, all is OK. But if the node is CN, then i have
>> the problem. The SER logs are:
>>
>> ##############################################################
>> ................
>>
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8051]: SER: new INVITE Nov 20
>> 18:28:25 pcscfD /usr/sbin/ser[8051]: parse_headers: flags=-1 Nov 20
>> 18:28:25 pcscfD /usr/sbin/ser[8051]:
>> check_via_address(2001:720:1500:1B:FCFD:FF:FE0C:103,
>> [2001:720:1500:1B:FCFD:FF:FE0C:103], 3) Nov 20 18:28:25 pcscfD
>> /usr/sbin/ser[8051]: ERROR: warning_builder: buffer size exceeded Nov
>> 20 18:28:25 pcscfD /usr/sbin/ser[8051]: WARNING: warning skipped -- too
>> big Nov 20 18:28:25 pcscfD /usr/sbin/ser[8051]: WARNING:vqm_resize:
>> resize(0) called Nov 20 18:28:25 pcscfD /usr/sbin/ser[8051]: DEBUG:
>> reply sent out.
>>
>> buf=0x8106270: SIP/2.0 1..., shmem=0xb5e8b330: SIP/2.0 1 Nov 20
>> 18:28:25 pcscfD /usr/sbin/ser[8051]: DEBUG: _reply_light: finished Nov
>> 20 18:28:25 pcscfD /usr/sbin/ser[8051]: DEBUG: mk_proxy: doing DNS
>> lookup... Nov 20 18:28:25 pcscfD /usr/sbin/ser[8051]:
>> check_via_address(2001:720:1500:1B:FCFD:FF:FE0C:103,
>> [2001:720:1500:1B:FCFD:FF:FE0C:103], 3) Nov 20 18:28:25 pcscfD
>> /usr/sbin/ser[8068]: DBG: tcp_main_loop: dead child 2 (shutting down?)
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8049]: child process 8051 exited
>> by a signal 11 Nov 20 18:28:25 pcscfD kernel: [4296472.851000]
>> <0>skb_under_panic: text:c02d2aad len:321 put:14 head:d62cfc00
>> data:d62cfbea tail:d62cfd2b end:d62cfd80 dev:eth0
>>
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8049]: core was not generated
>> Nov 20 18:28:25 pcscfD kernel: [4297162.639000] ------------[ cut here
>> ]------------
>>
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8049]: INFO: terminating due to
>> SIGCHLD Nov 20 18:28:25
>>
>> pcscfD kernel: [4297162.639000] kernel BUG at net/core/skbuff.c:112!
>>
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8059]: INFO: signal 15 received
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8052]: INFO: signal 15 received
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8053]: INFO: signal 15 received
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8054]: INFO: signal 15 received
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8055]: INFO: signal 15 received
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8056]: INFO: signal 15 received
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8057]: INFO: signal 15 received
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8058]: INFO: signal 15 received
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8060]: INFO: signal 15 received
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8061]: INFO: signal 15 received
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8062]: INFO: signal 15 received
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8063]: INFO: signal 15 received
>> Nov 20 18:28:25 pcscfD
>>
>> /usr/sbin/ser[8064]: INFO: signal 15 received
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8065]: INFO: signal 15 received
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8066]: INFO: signal 15 received
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8067]: INFO: signal 15 received
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8068]: INFO: signal 15 received
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8050]: INFO: signal 15 received
>> Nov 20 18:28:25 pcscfD kernel: [4297162.639000] invalid opcode: 0000 [#3]
>>
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8059]: Memory status (pkg):
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8052]: Memory status (pkg):
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8053]: Memory status (pkg):
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8054]: Memory status (pkg):
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8055]: Memory status (pkg):
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8056]: Memory status (pkg):
>> .......................
>>
>> Nov 20 18:28:25 pcscfD /usr/sbin/ser[8049]: hash =
>>
>> 2055 fragments no.:     1, unused:     0 ^I^I bucket size:
>> 1048576 -   2097152 (first   1572864) Nov 20 18:28:25 pcscfD
>> /usr/sbin/ser[8049]: hash = 2059 fragments no.:     1, unused:     0
>> ^I^I bucket size:  16777216 -  33554432 (first  31934160) Nov 20
>> 18:28:25 pcscfD /usr/sbin/ser[8049]: TOTAL:     26 free fragments =
>> 33537656 free bytes Nov 20 18:28:25 pcscfD /usr/sbin/ser[8049]:
>> ----------------------------- Nov 20 18:28:25 pcscfD
>> /usr/sbin/ser[8049]: shm_mem_destroy Nov 20 18:28:25 pcscfD
>> /usr/sbin/ser[8049]: destroying the shared memory lock Nov 20 18:28:25
>> pcscfD /usr/sbin/ser[8049]: terminating due to SIGCHLD
>> ##############################################################
>>
>> The problem is in the MTU size, because if the packet is not fragmented,
>> there is not a problem. But the INVITE packet
>>
>> needs to fragment, what it means, i always have the same problem.
>>
>> My SER version is version: ser 0.9.6 (i386/linux) and it's a vanilla
>> version. The packets is sent over UDP.
>>
>> My questions are:
>>
>> - Have anybody the same o similar problem?
>> - The MTU size is a problem in SER?
>> - Have anybody sent fragmented packets with SER?
>>
>> Thanks for your help.
>>
>> _______________________________________________
>> Serdev mailing list
>> Serdev at lists.iptel.org
>> http://lists.iptel.org/mailman/listinfo/serdev
>>