[llvm] [clang] [PowerPC][X86] Make cpu id builtins target independent and lower for PPC (PR #68919)
Nemanja Ivanovic via cfe-commits
cfe-commits at lists.llvm.org
Thu Jan 25 11:29:28 PST 2024
================
@@ -16086,6 +16086,41 @@ Value *CodeGenFunction::EmitPPCBuiltinExpr(unsigned BuiltinID,
switch (BuiltinID) {
default: return nullptr;
+ case Builtin::BI__builtin_cpu_is: {
+ const Expr *CPUExpr = E->getArg(0)->IgnoreParenCasts();
+ StringRef CPUStr = cast<clang::StringLiteral>(CPUExpr)->getString();
+ unsigned NumCPUID = StringSwitch<unsigned>(CPUStr)
+#define PPC_LNX_CPU(Name, NumericID) .Case(Name, NumericID)
+#include "llvm/TargetParser/PPCTargetParser.def"
+ .Default(-1U);
+ Value *Op0 = llvm::ConstantInt::get(Int32Ty, PPC_FAWORD_CPUID);
+ llvm::Function *F = CGM.getIntrinsic(Intrinsic::ppc_fixed_addr_ld);
+ Value *TheCall = Builder.CreateCall(F, {Op0}, "cpu_is");
+ return Builder.CreateICmpEQ(TheCall,
+ llvm::ConstantInt::get(Int32Ty, NumCPUID));
+ }
+ case Builtin::BI__builtin_cpu_supports: {
+ unsigned FeatureWord;
+ unsigned BitMask;
+ const Expr *CPUExpr = E->getArg(0)->IgnoreParenCasts();
+ StringRef CPUStr = cast<clang::StringLiteral>(CPUExpr)->getString();
+ std::tie(FeatureWord, BitMask) =
+ StringSwitch<std::pair<unsigned, unsigned>>(CPUStr)
+#define PPC_LNX_FEATURE(Name, Description, EnumName, Bitmask, FA_WORD) \
+ .Case(Name, {FA_WORD, Bitmask})
+#include "llvm/TargetParser/PPCTargetParser.def"
+ .Default({0, 0});
----------------
nemanjai wrote:
I think this is a good point. If we somehow have a string argument that would produce the `default` case and it has made it past Sema checking, it would be good to crash here rather than produce an invalid call to the intrinsic (or an invalid mask).
Of course, that assert is not all that friendly and it might be better to assert within this function.
What do you think about keeping the `default` and adding an assert below such as:
```
assert(BitMask && "Invalid target feature string. Missed by SemaChecking?");
```
https://github.com/llvm/llvm-project/pull/68919
More information about the cfe-commits
mailing list