[Mesa-dev] [PATCH] mesa/bufferobj: fix atomic offset/size get

[Dave Airlie]

On 12 October 2017 at 18:22, Tapani Pälli <tapani.palli at intel.com> wrote:
>
>
> On 10/12/2017 11:14 AM, Dave Airlie wrote:
>>
>>
>> On 12 Oct. 2017 15:40, "Tapani Pälli" <tapani.palli at intel.com
>> <mailto:tapani.palli at intel.com>> wrote:
>>
>>
>>
>>     On 10/12/2017 02:34 AM, Dave Airlie wrote:
>>
>>         From: Dave Airlie <airlied at redhat.com <mailto:airlied at redhat.com>>
>>
>>         When I realigned the bufferobj code, I didn't see the getters
>>         were different, realign the getters to work the same as ssbo.
>>
>>
>>     Alternatively you could set these values as 0 in
>>     bind_buffer_base_atomic_buffer()? Not sure if it's any better but
>>     then value would match internally what it has been before these
>> changes.
>>
>>
>> Before these changes the ssbo and atomic code was gratuitously different,
>> this is just the last piece of making them consistent.
>
>
> Right .. what I mean is that before the refactoring Size and Offset values
> in the structure were stored as 0, now they are stored as -1 even though
> here we return different value. I haven't checked if anything in Mesa would
> assume 0 though .. so feel free to ignore my ramblings :) I just wanted to
> note this because I tried to fix this and it following change fixes the bug
> as well:
>

For atomics that is true, but for ssbo it was false. The idea of refactoring it
was to align all the code to be same for both, since there is no reason for
differences. Your change would reintroduce differences where none are needed.

My change to get.c aligns the gets
for
GL_ATOMIC_COUNTER_BUFFER_START
GL_ATOMIC_COUNTER_BUFFER_SIZE
with the ones above
for
GL_SHADER_STORAGE_BUFFER_START
GL_SHADER_STORAGE_BUFFER_SIZE

So the code is consistent across both types of buffer.

Dave.