<div dir="ltr"><div class="gmail_extra"><div class="gmail_quote">On Thu, Sep 8, 2016 at 2:39 PM, Francisco Jerez <<a href="mailto:currojerez@riseup.net" target="_blank">currojerez@riseup.net</a>> wrote: <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Jason Ekstrand <<a href="mailto:jason@jlekstrand.net">jason@jlekstrand.net</a>> writes: > Blorp doesn't handle spilling so we set allow_spilling to false in that > case. The blorp 16x MSAA resolve shader spills in 16-wide but not 8-wide. > This commit makes it so that we fail the 16-wide compile and successfully > fall back to 8-wide instead of just assert-failing when trying to compile > the 16-wide shader. > > Signed-off-by: Jason Ekstrand <<a href="mailto:jason@jlekstrand.net">jason@jlekstrand.net</a>> > --- > src/mesa/drivers/dri/i965/brw_fs.cpp | 7 +++++-- > 1 file changed, 5 insertions(+), 2 deletions(-) > > diff --git a/src/mesa/drivers/dri/i965/brw_fs.cpp b/src/mesa/drivers/dri/i965/brw_fs.cpp > index d0b55ae..73aa5d2 100644 > --- a/src/mesa/drivers/dri/i965/brw_fs.cpp > +++ b/src/mesa/drivers/dri/i965/brw_fs.cpp > @@ -5906,6 +5906,11 @@ fs_visitor::allocate_registers(bool allow_spilling) > } > > if (!allocated_without_spills) { > + if (!allow_spilling) { > + assert(dispatch_width > 8); Not sure it makes sense for the register allocator to set different requirements based on the build dispatch width, wouldn't it make sense to just fail() here consistently if allow_spilling is false regardless of the dispatch width, and let blorp assert fail if none of the dispatch widths it tries out compiles successfully? </blockquote><div> </div><div>Agreed. I'll drop the assert. </div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> > + fail("Failure to register allocate and spilling is not allowed."); > + } > + > /* We assume that any spilling is worse than just dropping back to > * SIMD8. There's probably actually some intermediate point where > * SIMD16 with a couple of spills is still better. > @@ -5930,8 +5935,6 @@ fs_visitor::allocate_registers(bool allow_spilling) > } > } > > - assert(last_scratch == 0 || allow_spilling); > - > /* This must come after all optimization and register allocation, since > * it inserts dead code that happens to have side effects, and it does > * so based on the actual physical registers in use. > -- > 2.5.0.400.gff86faf > > _______________________________________________ > mesa-dev mailing list > <a href="mailto:mesa-dev@lists.freedesktop.org">mesa-dev@lists.freedesktop.org</a> > <a href="https://lists.freedesktop.org/mailman/listinfo/mesa-dev" rel="noreferrer" target="_blank">https://lists.freedesktop.org/mailman/listinfo/mesa-dev</a> </blockquote></div> </div></div>