On Thu, Feb 8, 2018 at 9:05 AM, Daniel Stone <dani...@collabora.com> wrote:
> When Ilia asked for suggestions on where to look to add 10bpc BGR
> support to the GBM and Wayland EGL backends, I had a quick look; the
> number of places the formats were duplicated and lossily/inconsistently
> translated between rapidly annoyed me enough to just type this series
> This gives both the Wayland and GBM backends a single source of truth
> for conversion between their native format tokens
> (wl_drm/drm_fourcc/wl_shm for the three Wayland protocols, GBM format
> tokens for that), DRIImage formats, bpp, and per-channel masks used to
> match/derive EGLConfigs.
> The very first patch is a harmless cosmetic fix, but a fix nonetheless;
> the others all just move the two EGL backends towards using the single
> table. The last two commits actually add the 10bpc BGR formats - by now
> trivial - which obviously depends on Ilia's series to add them.
> I've not actually tested 10bpc BGR, but Ilia tested it on kmscube and
Tested-by: Ilia Mirkin <imir...@alum.mit.edu>
I (first fixed then) tested on kmscube, and started weston. Weston's
log showed that a RGBA 10 10 10 2 config was used, and the colors in
weston-terminal seemed not-wrong (although the background changed
compared to what it was in 8bpp mode).
mesa-dev mailing list