[RISCV] Change RISCVMCExpr::VK_RISCV_None to RISCVMCExpr::VK_None

[hchandel]

Fix RISCVMCExpr::VK_RISCV_None which were added in #130779

[llvmbot]

@llvm/pr-subscribers-backend-risc-v

Author: quic_hchandel (hchandel)

Changes

Fix RISCVMCExpr::VK_RISCV_None which were added in #130779

---

Full diff: https://github.com/llvm/llvm-project/pull/131774.diff

1 Files Affected:

• (modified) llvm/lib/Target/RISCV/AsmParser/RISCVAsmParser.cpp (+6-6)

diff --git a/llvm/lib/Target/RISCV/AsmParser/RISCVAsmParser.cpp b/llvm/lib/Target/RISCV/AsmParser/RISCVAsmParser.cpp
index 267d8bb1d67ba..5ac7779240148 100644
--- a/llvm/lib/Target/RISCV/AsmParser/RISCVAsmParser.cpp
+++ b/llvm/lib/Target/RISCV/AsmParser/RISCVAsmParser.cpp
@@ -823,12 +823,12 @@ struct RISCVOperand final : public MCParsedAsmOperand {
   bool isSImm5NonZero() const {
     if (!isImm())
       return false;
-    RISCVMCExpr::VariantKind VK = RISCVMCExpr::VK_RISCV_None;
+    RISCVMCExpr::VariantKind VK = RISCVMCExpr::VK_None;
     int64_t Imm;
     bool IsConstantImm = evaluateConstantImm(getImm(), Imm, VK);
     return IsConstantImm && Imm != 0 &&
            isInt<5>(fixImmediateForRV32(Imm, isRV64Imm())) &&
-           VK == RISCVMCExpr::VK_RISCV_None;
+           VK == RISCVMCExpr::VK_None;
   }
 
   bool isSImm6() const {
@@ -1021,22 +1021,22 @@ struct RISCVOperand final : public MCParsedAsmOperand {
   bool isSImm16NonZero() const {
     if (!isImm())
       return false;
-    RISCVMCExpr::VariantKind VK = RISCVMCExpr::VK_RISCV_None;
+    RISCVMCExpr::VariantKind VK = RISCVMCExpr::VK_None;
     int64_t Imm;
     bool IsConstantImm = evaluateConstantImm(getImm(), Imm, VK);
     return IsConstantImm && Imm != 0 &&
            isInt<16>(fixImmediateForRV32(Imm, isRV64Imm())) &&
-           VK == RISCVMCExpr::VK_RISCV_None;
+           VK == RISCVMCExpr::VK_None;
   }
 
   bool isUImm16NonZero() const {
     if (!isImm())
       return false;
     int64_t Imm;
-    RISCVMCExpr::VariantKind VK = RISCVMCExpr::VK_RISCV_None;
+    RISCVMCExpr::VariantKind VK = RISCVMCExpr::VK_None;
     bool IsConstantImm = evaluateConstantImm(getImm(), Imm, VK);
     return IsConstantImm && isUInt<16>(Imm) && (Imm != 0) &&
-           VK == RISCVMCExpr::VK_RISCV_None;
+           VK == RISCVMCExpr::VK_None;
   }
 
   bool isUImm20LUI() const {

[kito-cheng]

LGTM

[RKSimon]

LGTM - please either get this committed asap or revert #130779 as you are breaking most buildbots at the moment

[hchandel]

LGTM - please either get this committed asap or revert #130779 as you are breaking most buildbots at the moment

@RKSimon I am waiting for the checks to pass. Only one check is remaining. Should I merge this without waiting for that check?

[RKSimon]

If either of the windows or linux build has already passed, a fixup like this should be good to go.

[hchandel]

If either of the windows or linux build has already passed, a fixup like this should be good to go.

I didn't know that. Thanks for the clarification.

[RKSimon]

If either of the windows or linux build has already passed, a fixup like this should be good to go.

I didn't know that. Thanks for the clarification.

Its not ideal, but either you get a minimal fix in or you should revert and start again when everything is green again. The CI checks on the original PR should have caught all of this though, which is strange.

[hchandel]

Yeah, That's what I am also thinking about. #131489 which does this change landed before my PR #130779. CI checks should have caught this, but all those checks passed and I merged after approvals. Maybe my branch didn't have #131489 which somehow led to passing checks. I am not sure though.

[david-arm]

@hchandel I have observed the same problem before. The checks run in github do not necessarily reflect reality because they are based on whatever commit your PR is based on. To work around this I normally rebase the PR downstream and run "make check-all" locally as I do not trust the github results.