[all-commits] [llvm/llvm-project] da33c9: [AMDGPU] Make SGPR spills exec mask agnostic

[Carl Ritson via All-commits]

  Branch: refs/heads/master
  Home:   https://github.com/llvm/llvm-project
  Commit: da33c96d4762e0e59f1eff16f60d5c1575490331
      https://github.com/llvm/llvm-project/commit/da33c96d4762e0e59f1eff16f60d5c1575490331
  Author: Carl Ritson <carl.ritson at amd.com>
  Date:   2020-06-03 (Wed, 03 Jun 2020)

  Changed paths:
    M llvm/lib/Target/AMDGPU/SIRegisterInfo.cpp
    M llvm/lib/Target/AMDGPU/SIRegisterInfo.h
    M llvm/test/CodeGen/AMDGPU/control-flow-fastregalloc.ll
    M llvm/test/CodeGen/AMDGPU/partial-sgpr-to-vgpr-spills.ll
    A llvm/test/CodeGen/AMDGPU/sgpr-spill.mir
    M llvm/test/CodeGen/AMDGPU/si-spill-sgpr-stack.ll
    M llvm/test/CodeGen/AMDGPU/spill-m0.ll
    M llvm/test/CodeGen/AMDGPU/spill-scavenge-offset.ll
    M llvm/test/CodeGen/AMDGPU/spill-wide-sgpr.ll

  Log Message:
  -----------
  [AMDGPU] Make SGPR spills exec mask agnostic

Explicitly set the exec mask for SGPR spills and reloads.
This fixes a bug where SGPR spills to memory could be incorrect
if the exec mask was 0 (or differed between spill and reload).

Additionally pack scalar subregisters (upto 16/32 per VGPR),
so that the majority of scalar types can be spilt or reloaded
with a simple memory access.  This should amortize some of the
additional overhead of manipulating the exec mask.

Reviewed By: arsenm

Differential Revision: https://reviews.llvm.org/D80282