[all-commits] [llvm/llvm-project] de5585: RegisterCoalescer: Correctly set valid lanes when ...

Matt Arsenault via All-commits all-commits at lists.llvm.org
Mon Sep 11 23:28:50 PDT 2023


  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: de5585078e910d8b31369d027a0bea97e31664a2
      https://github.com/llvm/llvm-project/commit/de5585078e910d8b31369d027a0bea97e31664a2
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2023-09-12 (Tue, 12 Sep 2023)

  Changed paths:
    M llvm/lib/CodeGen/RegisterCoalescer.cpp
    A llvm/test/CodeGen/X86/pr38795-verifier-error-pr38788.mir

  Log Message:
  -----------
  RegisterCoalescer: Correctly set valid lanes when keeping live out implicit defs

This fixes some verifier errors when live out implicit defs are
coalesced with identity copies. Fixes some reduced testcases from
issue #38788 but doesn't solve the original failure.

I was surprised this seems to obviate the special casing in
analyzeValue that's been there since the subregister liveness support
went in.

https://reviews.llvm.org/D158850


  Commit: cd4b906e183ffa6b78b2f2d062626ee5f54ceba3
      https://github.com/llvm/llvm-project/commit/cd4b906e183ffa6b78b2f2d062626ee5f54ceba3
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2023-09-12 (Tue, 12 Sep 2023)

  Changed paths:
    M llvm/lib/CodeGen/RegisterCoalescer.cpp
    M llvm/test/CodeGen/X86/pr38795-verifier-error-pr38788.mir
    M llvm/test/CodeGen/X86/pr38795.ll

  Log Message:
  -----------
  RegisterCoalescer: Don't delete IMPLICIT_DEF if it's live into the same block

Live out implicit_defs need to be kept, but the check for this only
checked if the block parent was the same. This doesn't work if the
parent blocks are the same but the value is live. Fixes verifier error
"Instruction ending live segment doesn't read the register", which
would appear at the coalesced non-implicit_def def.

Fixes #38788

https://reviews.llvm.org/D158882


Compare: https://github.com/llvm/llvm-project/compare/38e9006896ed...cd4b906e183f


More information about the All-commits mailing list