Re: [libvirt] [PATCH 3/4] Release conn lock before reporting errors in vir(Get|Release)Interface
On 07/22/2009 10:21 AM, Daniel Veillard wrote:
On Wed, Jul 22, 2009 at 12:23:31PM +0100, Daniel P. Berrange wrote:
> On Wed, Jul 22, 2009 at 01:20:58AM -0400, Laine Stump wrote:
>
>> Since virRaiseErrorFull needs to lock the conn, we must take care to
>> call it with the lock *not* held.
>>
>> If this patch is approved, similar patches need to be done for other
>> types in datatypes.c
>>
> Doh, that's a annoying edge case ! ACK
>
I wonder why we didn't see deadlocks so far ...
Because you didn't write any bugridden code to inadvertantly exercise it
like I did! ;-)
> If you have time to do the same change to other methods in this
> file that'd be very helpful
>
I pushed the patch for the Interface, Laine do you think you will
fix the others or should I do it ?
I just submitted a patch for the rest (in datatypes.c anyway).