[asterisk-dev] need help debugging a deadlock in 1.6.2.13

[Antonio Goméz Soto]

Op 19-10-10 20:12, Russell Bryant schreef:
> On Tue, 2010-10-19 at 16:56 +0100, Steve Davies wrote:
>> On 6 October 2010 23:35, Russell Bryant<russell at digium.com>  wrote:
>>>
>>> No.  The patch I wrote was actually for a test system that demonstrated the problem while doing some load testing for a customer.
>>> I'm currently on a business trip, so if not sooner, I'll get the patch updated to 1.4 at least and make it available next week.
>>>
>>
>> While not wanting to be rude :)
>>
>> *BUMP*
>
> Further testing showed that my patch is causing some crashes due to
> improper locking to deal with dialplan reloads and hint state change
> processing happening at the same time.

As it happens I experienced this locking problem this morning.
It was at the end of ast_features_reload(). That was without your patch.
My temp fix will be acquiring the reloadlock  before locking the contexts
in handle_state_change(). Though that is very crude.

 > More work is needed.  It's assigned to someone else to finish up so
> there should be some progress soon.
>
> Has anyone opened a mantis issue that I should associate our work with?
>
>

https://issues.asterisk.org/view.php?id=18165

Antonio.