[PATCH] D138492: [AMDGPU][AsmParser] Refine parsing instruction operands.

[Ivan Kosarev via Phabricator via llvm-commits]

kosarev added inline comments.


================
Comment at: llvm/lib/Target/AMDGPU/AsmParser/AMDGPUAsmParser.cpp:379
+  bool isTFE() const {
+    return isImmTy(ImmTyTFE) || (isToken() && getToken() == "tfe");
+  }
----------------
foad wrote:
> Why does only tfe need this?
An earlier version of the patch used to parse `tfe` operands into both `ImmTyTFE`s and tokens, later changed to use `ImmTyTFE` in all cases for simplicity. I've removed this check and changed the rest of the code to employ `parseNamedBit()` for parsing hardcoded `tfe`s. Same is done for `gds`, another operand we use in both the hardcoded and `AsmOperand` forms.

There's a similar situation with `off`, parsed into tokens while also having `parseVReg32OrOff()` adding them as `ImmTyOff`s. I'd prefer to address this separately along with other things like `ImmTyIdxen` being effectively unused, etc.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D138492/new/

https://reviews.llvm.org/D138492