Okay, stride question on this MP solved, production code was correct, but the tests were outdated a bit.
We used to have the problem that android was using geom::Stride as pixel-stride, and gbm was using byte-stride. The pattern checkers surrounding TestClientIPCRender were written with pixel-stride in mind. Update the pattern checkers to use byte-stride.
Its a bit confusing that android's alloc functions (and ANativeWindowBuffer) use pixel stride, so I added comments where appropriate. Frankly, pixel stride makes little sense to me, but those are the android interfaces we have. :) Some drivers, like qualcomm's, works around this by passing byte-stride in their opaque type on the ANativeWindowBuffer (handle).
Okay, stride question on this MP solved, production code was correct, but the tests were outdated a bit.
We used to have the problem that android was using geom::Stride as pixel-stride, and gbm was using byte-stride. The pattern checkers surrounding TestClientIPCRender were written with pixel-stride in mind. Update the pattern checkers to use byte-stride.
Its a bit confusing that android's alloc functions (and ANativeWindowBu ffer) use pixel stride, so I added comments where appropriate. Frankly, pixel stride makes little sense to me, but those are the android interfaces we have. :) Some drivers, like qualcomm's, works around this by passing byte-stride in their opaque type on the ANativeWindowBuffer (handle).