[PATCH] D68308: AMDGPU/GlobalISel: Widen 16-bit G_MERGE_VALUEs sources

[Matt Arsenault via Phabricator via llvm-commits]

arsenm created this revision.
arsenm added reviewers: tstellar, nhaehnle, kerbowa.
Herald added subscribers: Petar.Avramovic, t-tye, tpr, dstuttard, rovka, yaxunl, wdng, jvesely, kzhuravl.

Continue making a mess of merge/unmerge legality.


https://reviews.llvm.org/D68308

Files:
  lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp
  test/CodeGen/AMDGPU/GlobalISel/legalize-load-constant-32bit.mir
  test/CodeGen/AMDGPU/GlobalISel/legalize-load-constant.mir
  test/CodeGen/AMDGPU/GlobalISel/legalize-load-flat.mir
  test/CodeGen/AMDGPU/GlobalISel/legalize-load-global.mir
  test/CodeGen/AMDGPU/GlobalISel/legalize-load-local.mir
  test/CodeGen/AMDGPU/GlobalISel/legalize-load-private.mir
  test/CodeGen/AMDGPU/GlobalISel/legalize-merge-values.mir