Age | Commit message (Collapse) | Author |
|
|
|
When detiling, the RS requires dimensions for the window
size that are a multiple of 4. With two pixel pipes this
amounts to 8.
Hence pad the height of the output buffers to the next multiple
of 8.
|
|
Signed-off-by: Christian Gmeiner <christian.gmeiner@gmail.com>
|
|
Signed-off-by: Christian Gmeiner <christian.gmeiner@gmail.com>
|
|
Signed-off-by: Christian Gmeiner <christian.gmeiner@gmail.com>
|
|
small steps to get gc2000 up and running
|
|
Signed-off-by: Christian Gmeiner <christian.gmeiner@gmail.com>
|
|
gc2000: fix RS_PIPE_OFFSET and RS_WINDOW_SIZE
|
|
Signed-off-by: Christian Gmeiner <christian.gmeiner@gmail.com>
|
|
Pass buffer object here instead of raw physical address.
|
|
No longer cache logical and GPU address for resource levels
but store offsets and query only when needed.
This is the first step toward submitting bo's with relocations
(just search for `etna_bo_gpu_address` to find all places where GPU
addresses are used).
|
|
This is a bit of a hack. FOURCC support should be done properly if
more modes are ever added, but it's not worth it for just the one.
|
|
This makes fence handles into globally increasing uint32_t's
and thus simplifies the handling in the driver, and makes it possible
to compare fence handles to see what happened before.
The scheme emulates a saner kernel API (closer to adreno drm driver)
|
|
This is more consistent with other projects.
|