[sr-dev] Crash bug

[Daniel-Constantin Mierla]

Hello,

On 11/05/15 09:37, Alex Hermann wrote:
> On Sunday 10 May 2015, Alex Balashov wrote:
>>> Can you provide details what version are you running now?
>> I am running git 4.2:2421e0b987119c083bd5cebe754ae8c97ebfa867.
> That one doesn't seem to exist in the public repository.
>
> Make sure you have at least 2a0fab2d66b299cb6003cead79a986324d16560d in your 
> branch. This fixes a timer bug. There might be a lot more bugs like this as 
> the compiler warnings for this type of errors are squelched, resulting in the 
> hiding of tons of warnings.
>
I pushed a patch over the weekend that is related to the report of a
crash when removing from timer a suspended transaction:

 -
https://github.com/kamailio/kamailio/commit/96b84ba4d7f5ee4a763fe7db81bbf4f9a6a3ef77

I haven't had the time for deep analyze, but there were two different
mutexes (async and reply loacks) used for handling the suspended
transaction, and that may cause problems in some corner cases. The patch
changes to use of reply lock which is re-entrant, so the process that
acquired the transaction can handle it properly, no matter it does
continue or reply/failure route handling.

It will be backported as soon as more testing is performed.

Cheers,
Daniel

-- 
Daniel-Constantin Mierla
http://twitter.com/#!/miconda - http://www.linkedin.com/in/miconda
Kamailio World Conference, May 27-29, 2015
Berlin, Germany - http://www.kamailioworld.com