[PATCH] D131959: [AMDGPU] Fix SDST operand of V_DIV_SCALE to always be VCC

[Pierre van Houtryve via Phabricator via llvm-commits]

Pierre-vh updated this revision to Diff 456362.
Pierre-vh added a comment.

- Fix Parser/Printer

There is still a test (gfx10_asm_vop3.s) that I'm not sure how to update (so I deleted the checks for now)
I'd say the encoding needs to be updated everywhere, but should I duplicate the tests for w32/w64 then?
Or just use GFX10 check instead of W32 <https://reviews.llvm.org/W32>/W64 <https://reviews.llvm.org/W64>?
Moreover, there is no other test in that file that has a wavesize-dependent vcc/vcc_lo register usage so it'd
seem strange to leave div_scale there unless there's a pattern I'm not seeing.

Do I just not check for the "operands are not valid for this GPU or mode" errors and use GFX10 check, then update
all encodings as needed?

I feel like since another test already checks "operands are not valid for this GPU or mode", leaving the instructions
to just use VCC (or do half vcc_lo/half vcc) + using GFX10 check should be enough. What do you think?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D131959/new/

https://reviews.llvm.org/D131959

Files:
  llvm/lib/Target/AMDGPU/AMDGPUISelDAGToDAG.cpp
  llvm/lib/Target/AMDGPU/AMDGPUInstructionSelector.cpp
  llvm/lib/Target/AMDGPU/AsmParser/AMDGPUAsmParser.cpp
  llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPUInstPrinter.cpp
  llvm/lib/Target/AMDGPU/VOP3Instructions.td
  llvm/lib/Target/AMDGPU/VOPInstructions.td
  llvm/test/CodeGen/AMDGPU/GlobalISel/constant-bus-restriction.ll
  llvm/test/CodeGen/AMDGPU/GlobalISel/fdiv.f16.ll
  llvm/test/CodeGen/AMDGPU/GlobalISel/fdiv.f32.ll
  llvm/test/CodeGen/AMDGPU/GlobalISel/fdiv.f64.ll
  llvm/test/CodeGen/AMDGPU/GlobalISel/frem.ll
  llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.div.scale.ll
  llvm/test/CodeGen/AMDGPU/fdiv-nofpexcept.ll
  llvm/test/CodeGen/AMDGPU/fdiv.f64.ll
  llvm/test/CodeGen/AMDGPU/frem.ll
  llvm/test/CodeGen/AMDGPU/inserted-wait-states.mir
  llvm/test/CodeGen/AMDGPU/llvm.amdgcn.div.scale.ll
  llvm/test/CodeGen/AMDGPU/llvm.powi.ll
  llvm/test/CodeGen/AMDGPU/sched-crash-dbg-value.mir
  llvm/test/CodeGen/AMDGPU/wave32.ll
  llvm/test/MC/AMDGPU/gfx10_asm_vop3.s
  llvm/test/MC/AMDGPU/gfx7_asm_vop3_e64.s
  llvm/test/MC/AMDGPU/vop3.s
  llvm/test/MC/AMDGPU/wave32.s
  llvm/test/MC/AMDGPU/wave_any.s

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D131959.456362.patch
Type: text/x-patch
Size: 443405 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20220829/e5eab850/attachment-0001.bin>