[Mesa-dev] [PATCH 1/7] i965/fs_surface_builder: Explicitly handle FORMAT_NONE in num_image_coordinates

[Chad Versace]

On Mon 23 Nov 2015, Jason Ekstrand wrote:
> On Mon, Nov 23, 2015 at 6:15 AM, Francisco Jerez <currojerez at riseup.net> wrote:
> > Jason Ekstrand <jason at jlekstrand.net> writes:

> > For that reason 'has_matching_typed_format(MESA_FORMAT_NONE) = true'
> > makes sense to me, but 'has_matching_typed_format(BRW_SURFACEFORMAT_RAW)
> > = true' and the identification of MESA_FORMAT_NONE with
> > BRW_SURFACEFORMAT_RAW does not.
> 
> Yeah, that's a distinction I would like to keep.  Perhaps a
> BRW_SURFACEFORMAT_INVALID?  That's what we're doing in libisl right
> now.  Does that sound reasonable?

Yes, I believe BRW_SURFACEFORMAT_INVALID is very reasonable. To be
compatible with the new intel-surface-layout (isl) code I'm writing, you
should define BRW_SURFACEFORMAT_INVALID = 0xffff (UINT16_MAX).