[llvm-branch-commits] [llvm] [AMDGPU] Error out on too long lit() constants (PR #197803)

[Jay Foad via llvm-branch-commits]

================
@@ -2540,8 +2540,12 @@ void AMDGPUOperand::addLiteralImmOperand(MCInst &Inst, int64_t Val, bool ApplyMo
     // truncated to uint32_t), if the target doesn't support 64-bit literals, or
     // the lit modifier is explicitly used, we need to truncate it to the 32
     // LSBs.
-    if (!AsmParser->has64BitLiterals() || Lit == LitModifier::Lit)
+    if (!AsmParser->has64BitLiterals() || Lit == LitModifier::Lit) {
+      if (!isInt<32>(Val) && !isUInt<32>(Val))
+        const_cast<AMDGPUAsmParser *>(AsmParser)->Error(
+            Inst.getLoc(), "invalid literal value");
----------------
jayfoad wrote:

The error message could be more helpful, maybe something like "literal value out of range"?

https://github.com/llvm/llvm-project/pull/197803