Re: [Mesa-dev] [PATCH 1/2] gallium/sw: allow for negative strides in some places

2013-03-19 Thread Dave Airlie
On Tue, Mar 19, 2013 at 8:54 PM, Jose Fonseca wrote: > I think this is fine in principle, but I believe it's better to be exhaustive > now than to waste time debugging unsigned/signed stride mismatches later. > > Especially all src/gallium/auxiliary/util modules should be updated: > > - src/galli

Re: [Mesa-dev] [PATCH 1/2] gallium/sw: allow for negative strides in some places

2013-03-19 Thread Jose Fonseca
I think this is fine in principle, but I believe it's better to be exhaustive now than to waste time debugging unsigned/signed stride mismatches later. Especially all src/gallium/auxiliary/util modules should be updated: - src/gallium/auxiliary/util/u_format_* - src/gallium/auxiliary/util/u_surf

[Mesa-dev] [PATCH 1/2] gallium/sw: allow for negative strides in some places

2013-03-18 Thread Dave Airlie
From: Dave Airlie This is for a specific use-case for the QXL driver (which wraps llvmpipe/softpipe). Texture it get via TFP are inverted, so we need to use a negative stride to sample from them. So this converts a bunch of cases where we use unsigned stride to a signed value. I can't see the sin