[WebAssembly] Add builtin and intrinsic for v8x16.swizzle

Summary:
This clang builtin and corresponding LLVM intrinsic are necessary to
expose the exact semantics of the underlying WebAssembly instruction
to users. LLVM produces a poison value if the dynamic swizzle indices
are greater than the vector size, but the WebAssembly instruction sets
the corresponding output lane to zero. Users who depend on this
behavior can safely use this builtin.

Depends on D68527.

Reviewers: aheejin, dschuff

Subscribers: sbc100, jgravelle-google, hiraditya, sunfish, cfe-commits, llvm-commits

Tags: #clang, #llvm

Differential Revision: https://reviews.llvm.org/D68531

git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@374189 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/include/clang/Basic/BuiltinsWebAssembly.def b/include/clang/Basic/BuiltinsWebAssembly.def
index f825040d6142893cb161eadd476621bf50fc088a..7e80e91c567ff58edff1d18058078d1432ccadce 100644 (file)
--- a/include/clang/Basic/BuiltinsWebAssembly.def
+++ b/include/clang/Basic/BuiltinsWebAssembly.def
@@ -60,6 +60,8 @@ TARGET_BUILTIN(__builtin_wasm_trunc_saturate_s_i64_f64, "LLid", "nc", "nontrappi
 TARGET_BUILTIN(__builtin_wasm_trunc_saturate_u_i64_f64, "LLid", "nc", "nontrapping-fptoint")
 
 // SIMD builtins
+TARGET_BUILTIN(__builtin_wasm_swizzle_v8x16, "V16cV16cV16c", "nc", "unimplemented-simd128")
+
 TARGET_BUILTIN(__builtin_wasm_extract_lane_s_i8x16, "iV16cIi", "nc", "simd128")
 TARGET_BUILTIN(__builtin_wasm_extract_lane_u_i8x16, "iV16cIi", "nc", "unimplemented-simd128")
 TARGET_BUILTIN(__builtin_wasm_extract_lane_s_i16x8, "iV8sIi", "nc", "simd128")

diff --git a/lib/CodeGen/CGBuiltin.cpp b/lib/CodeGen/CGBuiltin.cpp
index d5389d74cb8da7b3e004e67a62f8bed2b4cdaeda..0562a623ae2b3eff62f42bf1447d71a53f87e43e 100644 (file)
--- a/lib/CodeGen/CGBuiltin.cpp
+++ b/lib/CodeGen/CGBuiltin.cpp
@@ -14066,6 +14066,12 @@ Value *CodeGenFunction::EmitWebAssemblyBuiltinExpr(unsigned BuiltinID,
                                      ConvertType(E->getType()));
     return Builder.CreateCall(Callee, {LHS, RHS});
   }
+  case WebAssembly::BI__builtin_wasm_swizzle_v8x16: {
+    Value *Src = EmitScalarExpr(E->getArg(0));
+    Value *Indices = EmitScalarExpr(E->getArg(1));
+    Function *Callee = CGM.getIntrinsic(Intrinsic::wasm_swizzle);
+    return Builder.CreateCall(Callee, {Src, Indices});
+  }
   case WebAssembly::BI__builtin_wasm_extract_lane_s_i8x16:
   case WebAssembly::BI__builtin_wasm_extract_lane_u_i8x16:
   case WebAssembly::BI__builtin_wasm_extract_lane_s_i16x8:

diff --git a/test/CodeGen/builtins-wasm.c b/test/CodeGen/builtins-wasm.c
index ca407ebc0ed4f1101a6c1f5d81dbe917ea5f32d2..436f47c7b444856d3fa1889e4d8418f889f47ecb 100644 (file)
--- a/test/CodeGen/builtins-wasm.c
+++ b/test/CodeGen/builtins-wasm.c
@@ -157,7 +157,6 @@ double max_f64(double x, double y) {
   // WEBASSEMBLY-NEXT: ret
 }
 
-
 int extract_lane_s_i8x16(i8x16 v) {
   return __builtin_wasm_extract_lane_s_i8x16(v, 13);
   // MISSING-SIMD: error: '__builtin_wasm_extract_lane_s_i8x16' needs target feature simd128
@@ -539,3 +538,9 @@ i32x4 widen_high_u_i32x4_i16x8(i16x8 v) {
   // WEBASSEMBLY: call <4 x i32> @llvm.wasm.widen.high.unsigned.v4i32.v8i16(<8 x i16> %v)
   // WEBASSEMBLY: ret
 }
+
+i8x16 swizzle_v8x16(i8x16 x, i8x16 y) {
+  return __builtin_wasm_swizzle_v8x16(x, y);
+  // WEBASSEMBLY: call <16 x i8> @llvm.wasm.swizzle(<16 x i8> %x, <16 x i8> %y)
+  // WEBASSEMBLY-NEXT: ret
+}