[Mesa-dev] [PATCH 06/34] gbm: Export a per plane getter for stride

[Daniel Stone]

Hi,

On 4 February 2017 at 05:29, Ben Widawsky <ben at bwidawsk.net> wrote:
> On 17-01-31 11:33:39, Jason Ekstrand wrote:
>> On Mon, Jan 23, 2017 at 10:19 PM, Ben Widawsky <ben at bwidawsk.net> wrote:
>>> +   /* Preserve legacy behavior if plane is 0 */
>>> +   if (plane == 0)
>>> +      return _bo->stride;
>>
>> This implies that, for multi-planar images, you have to have the BO stride
>> be the same as the image stride.  Do we want this or do we only want to
>> return bo->stride if we don't have enough DRI stuff (in the if below) to
>> use the image?  I think it's probably ok either way.
>>
>
> The biggest problem is if I change it (see hunk below) is that it
> potentially
> doesn't match the old behavior, and I don't really have a great way to test
> this
> outside of our environment. I went with the path of least risk, however you
> do
> raise a good point.
>
> Daniel, do you have an opinion here?

I think Jason's suggestion is the right thing to do. The main case I
worry about is dumb buffers, covered by !bo->image, which cannot be
multi-planar. If we ever have a single-planar image where the stride
doesn't match the DRIimage stride, that sounds like something to be
fixed anyway. So, let's go with this and fix up anything which breaks,
though I can't immediately see how that would happen.

Cheers,
Daniel