Timothy Arceri <t_arc...@yahoo.com.au> writes: > On 03/07/18 05:51, Eric Anholt wrote: > >> Eric Anholt <e...@anholt.net> writes: >> >>> [ Unknown signature status ] >>> Timothy Arceri <tarc...@itsqueeze.com> writes: >>> >>>> nir_compact_varyings() is meant to skip over varyings used by xfb: >>>> >>>> /* We can't repack xfb varyings. */ >>>> if (var->data.always_active_io) >>>> continue; >>>> >>>> Any idea why that isn't working in this case? >>> Looks like GLSL IR has that flag wrong. points.7 has v_var6,7,8,9 >>> transform feedback output, but the IR says: >> [...] >> >> Any feedback on this? This is my remaining TF issue for V3D. > Hmm. I think this still gets messed up because if packing the non-xfb > varyings results in freeing up a location then st_nir_assign_var_locations() > will end up assigning a new location for the xfb varyings. > > I think your patch to disable for xfb is the easiest way to work around > this for now. I still hope we might one day have a full NIR GLSL linker > to replace GLSL IR but my hopes of ever seeing one are slowly fading. > > Anyway for this patch: > > Reviewed-by: Timothy Arceri <tarc...@itsqueeze.com>
Thanks! I do wish for better linking as well, but state_tracker's program handling is so convoluted that I can only ever hold a fraction of it in my head at a time.
signature.asc
Description: PGP signature
_______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev
