[PATCH] D46799: [x86][eflags] Fix PR37431 by teaching the x86 backend to understand a setCC pseudo when trying to understand generic setCC behavior.

[Chandler Carruth via Phabricator via llvm-commits]

chandlerc planned changes to this revision.
chandlerc added a comment.

1. I need CHECK lines in this test case.

2. More importantly, this is the wrong approach. SETB_C* is more different from SETB than I realized. Props to Craig for pointing this out. I need to specially handle SETB_C* in the EFLAGS copy lowering to "promote" the way I emit SETB to use SETB_C* when appropriate, and then fix the registers afterward.


Repository:
  rL LLVM

https://reviews.llvm.org/D46799