i965: Always do NIR IO lowering at specialization time.
authorKenneth Graunke <kenneth@whitecape.org>
Thu, 25 Feb 2016 05:36:15 +0000 (21:36 -0800)
committerKenneth Graunke <kenneth@whitecape.org>
Fri, 26 Feb 2016 23:55:59 +0000 (15:55 -0800)
commitdcd4a841e9096b988ea3ca2779e7c8b1ca5e5747
tree13e1f6f30630ed5bd23ba863da877530e2b0e587
parentfa7135107f462fad8cda4731c037e393041e9369
i965: Always do NIR IO lowering at specialization time.

We've now hit literally every case other than geometry shaders (and
compute shaders, but those are a no-op).  So, let's just move geometry
shaders over too and be done with it.

The only advantage to doing this at link time was to save the expense
of running the pass on recompiles.  But we're already running a lot of
passes, and the extra code complexity isn't worth it.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
src/mesa/drivers/dri/i965/brw_nir.c
src/mesa/drivers/dri/i965/brw_vec4_gs_visitor.cpp