[Mesa-dev] [PATCH 01/10] glsl: don't let an 'if' then-branch kill const propagation for else-branch

2018-06-27 Thread Caio Marcelo de Oliveira Filho
When handling 'if' in constant propagation, if a certain variable was killed when processing the first branch of the 'if', then the second would get any propagation from previous nodes. This is similar to the change done for copy propagation code. x = 1; if (...) { z = x; // This

Re: [Mesa-dev] [PATCH 01/10] glsl: don't let an 'if' then-branch kill const propagation for else-branch

2018-07-05 Thread Eric Anholt
Caio Marcelo de Oliveira Filho writes: > When handling 'if' in constant propagation, if a certain variable was > killed when processing the first branch of the 'if', then the second > would get any propagation from previous nodes. This is similar to the > change done for copy propagation code. >

Re: [Mesa-dev] [PATCH 01/10] glsl: don't let an 'if' then-branch kill const propagation for else-branch

2018-07-09 Thread Caio Marcelo de Oliveira Filho
> Possible follow-up change for someone looking at compiler perf: > kill_entry doesn't seem to need to be a list since > 4654439fdd766f79a78fe0d812fd916f5815e7e6, and we could probably just > store the write_mask in the entry->data field and not have struct > kill_entry at all! That makes sense. I