[clang] [Headers][X86] VectorExprEvaluator::VisitCallExpr - allow SSE/AVX2/AVX512 pack intrinsics to be used in constexpr (PR #156003)

[Simon Pilgrim via cfe-commits]

================
@@ -3435,6 +3524,22 @@ bool InterpretBuiltin(InterpState &S, CodePtr OpPC, 
const CallExpr *Call,
           }
           return LHS.lshr(RHS.getZExtValue());
         });
+  case clang::X86::BI__builtin_ia32_packsswb128:
+  case clang::X86::BI__builtin_ia32_packsswb256:
+  case clang::X86::BI__builtin_ia32_packsswb512:
+  case clang::X86::BI__builtin_ia32_packssdw128:
+  case clang::X86::BI__builtin_ia32_packssdw256:
+  case clang::X86::BI__builtin_ia32_packssdw512:
+    return interp__builtin_x86_pack(S, OpPC, Call, /*Unsat=*/false,
+                                    NarrowElement);
----------------
RKSimon wrote:

You should be able to create a very minimal wrapper for APInt::truncSSat for 
PACKSS:
```
[](const APSInt &Src) { 
  return Src.truncSSat(Src.getBitwidth() / 2);
}
```

https://github.com/llvm/llvm-project/pull/156003
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits