[clang] [llvm] [HLSL][SPIR-V] Add SV_DispatchThreadID semantic support (PR #82536)

[Natalie Chouinard via cfe-commits]

================
@@ -342,8 +343,19 @@ llvm::Value *CGHLSLRuntime::emitInputSemantic(IRBuilder<> &B,
     return B.CreateCall(FunctionCallee(DxGroupIndex));
   }
   if (D.hasAttr<HLSLSV_DispatchThreadIDAttr>()) {
-    llvm::Function *DxThreadID = CGM.getIntrinsic(Intrinsic::dx_thread_id);
-    return buildVectorInput(B, DxThreadID, Ty);
+    llvm::Function *ThreadIDIntrinsic;
+    switch (CGM.getTarget().getTriple().getArch()) {
+    case llvm::Triple::dxil:
+      ThreadIDIntrinsic = CGM.getIntrinsic(Intrinsic::dx_thread_id);
+      break;
+    case llvm::Triple::spirv:
+      ThreadIDIntrinsic = CGM.getIntrinsic(Intrinsic::spv_thread_id);
+      break;
+    default:
+      llvm_unreachable("Input semantic not supported by target");
+      break;
+    }
+    return buildVectorInput(B, ThreadIDIntrinsic, Ty);
----------------
sudonatalie wrote:

Agreed this is not going to scale well. Another option I was looking into, but haven't yet figured out, is whether we could make dummy `ClangBuiltin`s to maintain the mapping. I think the appropriate `spv_create_handle` or `dx_create_handle` get generated without any special logic other than both having `ClangBuiltin<"__builtin_hlsl_create_handle">` in their `Intrinsics*.td`. I'll open an issue to follow up on this separately.

https://github.com/llvm/llvm-project/pull/82536