================
@@ -3622,6 +3622,15 @@ RValue CodeGenFunction::EmitBuiltinExpr(const GlobalDecl
GD, unsigned BuiltinID,
Builder.CreateArithmeticFence(ArgValue, ConvertType(ArgType)));
return RValue::get(ArgValue);
}
+ case Builtin::BI__builtin_bswapg: {
+ Value *ArgValue = EmitScalarExpr(E->getArg(0));
+ llvm::IntegerType *IntTy = cast<llvm::IntegerType>(ArgValue->getType());
+ assert(IntTy && "LLVM's __builtin_bswapg only supports integer variants");
----------------
ojhunt wrote:
I don't think mod 16 is right either, swapping `_BitInt(24)` (etc) is in
principle just as reasonable a swap, though it would be a novel choice. And at
the same time as disallowing _BitInt(24), the mod 16 rule allows the
equivalently odd _BitInt(48) to be swapped.
Would it be more reasonable to simply say it must be a power of 2 number of
bytes (N=8*2^^N for some positive N)? @philnik777 does that sound reasonable?
https://github.com/llvm/llvm-project/pull/162433
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
