[Openal-devel] Trunk + 0.0.8 (was: Yet another source queue race condition)

[Elias Naur]

On Friday 08 April 2005 16:23, Adam D. Moss wrote:
> Elias Naur wrote:
> > P.S. Please do not hesitate to comment or propose a correct fix on my
> > earlier reported queue race condition too.
>

[...]

> I'm happy enough that the earlier race fix doesn't cause
> more bogosity than it fixes, and when this latter fix boils
> down to acquiring a lock I think that sounds at least harmless
> enough to let into the trunk for a few weeks (please submit a
> patch if you can).
>
> --Adam

Here you go. I've changed the fix for the earlier race condition to removing 
the assignment to new_readindex and new_soundpos on underflow instead of 
setting new_readindex to the queue size, which seems better than the old fix. 
I've also added a source lock to alSourceUnqueueBuffers for the other patch 
since I don't think the existing context lock is enough to protect the queue.

The patches against HEAD are attached. Please forgive me if this is not the 
correct way to create or submit patches, as I'm not used to the process.

 - elias
-------------- next part --------------
A non-text attachment was scrubbed...
Name: race_condition_on_source_buffer_queue.patch
Type: text/x-diff
Size: 1600 bytes
Desc: not available
Url : http://localhost.localdomain/pipermail/openal-devel/attachments/20050408/683a6292/race_condition_on_source_buffer_queue.bin
-------------- next part --------------
A non-text attachment was scrubbed...
Name: race_condition_on_underflow.patch
Type: text/x-diff
Size: 571 bytes
Desc: not available
Url : http://localhost.localdomain/pipermail/openal-devel/attachments/20050408/683a6292/race_condition_on_underflow.bin