Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[clang][wasm] Replace the target integer sub saturate intrinsics with the equivalent generic __builtin_elementwise_sub_sat intrinsics #109405

Merged
merged 2 commits into from
Sep 22, 2024
Merged

[clang][wasm] Replace the target integer sub saturate intrinsics with the equivalent generic __builtin_elementwise_sub_sat intrinsics #109405

merged 2 commits into from
Sep 22, 2024

Conversation

RKSimon
Copy link
Collaborator

@RKSimon RKSimon commented Sep 20, 2024

Remove the Intrinsic::wasm_sub_sat_signed/wasm_sub_sat_unsigned entries and just use sub_sat_s/sub_sat_u directly

@RKSimon RKSimon requested a review from tlively September 20, 2024 11:37
@llvmbot llvmbot added clang Clang issues not falling into any other category backend:WebAssembly backend:X86 clang:frontend Language frontend issues, e.g. anything involving "Sema" clang:headers Headers provided by Clang, e.g. for intrinsics clang:codegen IR generation bugs: mangling, exceptions, etc. llvm:ir labels Sep 20, 2024
@llvmbot
Copy link
Member

llvmbot commented Sep 20, 2024 •
edited
Loading

@llvm/pr-subscribers-clang

@llvm/pr-subscribers-clang-codegen

Author: Simon Pilgrim (RKSimon)

Changes

Remove the Intrinsic::wasm_sub_sat_signed/wasm_sub_sat_unsigned entries and just use sub_sat_s/sub_sat_u directly

Full diff: https://github.com/llvm/llvm-project/pull/109405.diff

8 Files Affected:

  • (modified) clang/include/clang/Basic/BuiltinsWebAssembly.def (-5)
  • (modified) clang/lib/CodeGen/CGBuiltin.cpp (-22)
  • (modified) clang/lib/Headers/wasm_simd128.h (+4-4)
  • (modified) clang/test/CodeGen/builtins-wasm.c (+1-29)
  • (modified) llvm/include/llvm/IR/IntrinsicsWebAssembly.td (-8)
  • (modified) llvm/lib/Target/WebAssembly/WebAssemblyISelLowering.cpp (+2-2)
  • (modified) llvm/lib/Target/WebAssembly/WebAssemblyInstrSIMD.td (+2-4)
  • (modified) llvm/test/CodeGen/WebAssembly/simd-intrinsics.ll (+8-8)
diff --git a/clang/include/clang/Basic/BuiltinsWebAssembly.def b/clang/include/clang/Basic/BuiltinsWebAssembly.def
index 90441a5d500120..ab480369b3820e 100644
--- a/clang/include/clang/Basic/BuiltinsWebAssembly.def
+++ b/clang/include/clang/Basic/BuiltinsWebAssembly.def
@@ -68,11 +68,6 @@ TARGET_BUILTIN(__builtin_wasm_trunc_saturate_u_i64_f64, "LLid", "nc", "nontrappi
 // SIMD builtins
 TARGET_BUILTIN(__builtin_wasm_swizzle_i8x16, "V16ScV16ScV16Sc", "nc", "simd128")
 
-TARGET_BUILTIN(__builtin_wasm_sub_sat_s_i8x16, "V16ScV16ScV16Sc", "nc", "simd128")
-TARGET_BUILTIN(__builtin_wasm_sub_sat_u_i8x16, "V16UcV16UcV16Uc", "nc", "simd128")
-TARGET_BUILTIN(__builtin_wasm_sub_sat_s_i16x8, "V8sV8sV8s", "nc", "simd128")
-TARGET_BUILTIN(__builtin_wasm_sub_sat_u_i16x8, "V8UsV8UsV8Us", "nc", "simd128")
-
 TARGET_BUILTIN(__builtin_wasm_abs_i8x16, "V16ScV16Sc", "nc", "simd128")
 TARGET_BUILTIN(__builtin_wasm_abs_i16x8, "V8sV8s", "nc", "simd128")
 TARGET_BUILTIN(__builtin_wasm_abs_i32x4, "V4iV4i", "nc", "simd128")
diff --git a/clang/lib/CodeGen/CGBuiltin.cpp b/clang/lib/CodeGen/CGBuiltin.cpp
index 3d1138b7773853..78b432474ba3b7 100644
--- a/clang/lib/CodeGen/CGBuiltin.cpp
+++ b/clang/lib/CodeGen/CGBuiltin.cpp
@@ -21443,28 +21443,6 @@ Value *CodeGenFunction::EmitWebAssemblyBuiltinExpr(unsigned BuiltinID,
     Function *Callee = CGM.getIntrinsic(Intrinsic::wasm_swizzle);
     return Builder.CreateCall(Callee, {Src, Indices});
   }
-  case WebAssembly::BI__builtin_wasm_sub_sat_s_i8x16:
-  case WebAssembly::BI__builtin_wasm_sub_sat_u_i8x16:
-  case WebAssembly::BI__builtin_wasm_sub_sat_s_i16x8:
-  case WebAssembly::BI__builtin_wasm_sub_sat_u_i16x8: {
-    unsigned IntNo;
-    switch (BuiltinID) {
-    case WebAssembly::BI__builtin_wasm_sub_sat_s_i8x16:
-    case WebAssembly::BI__builtin_wasm_sub_sat_s_i16x8:
-      IntNo = Intrinsic::wasm_sub_sat_signed;
-      break;
-    case WebAssembly::BI__builtin_wasm_sub_sat_u_i8x16:
-    case WebAssembly::BI__builtin_wasm_sub_sat_u_i16x8:
-      IntNo = Intrinsic::wasm_sub_sat_unsigned;
-      break;
-    default:
-      llvm_unreachable("unexpected builtin ID");
-    }
-    Value *LHS = EmitScalarExpr(E->getArg(0));
-    Value *RHS = EmitScalarExpr(E->getArg(1));
-    Function *Callee = CGM.getIntrinsic(IntNo, ConvertType(E->getType()));
-    return Builder.CreateCall(Callee, {LHS, RHS});
-  }
   case WebAssembly::BI__builtin_wasm_abs_i8x16:
   case WebAssembly::BI__builtin_wasm_abs_i16x8:
   case WebAssembly::BI__builtin_wasm_abs_i32x4:
diff --git a/clang/lib/Headers/wasm_simd128.h b/clang/lib/Headers/wasm_simd128.h
index b1bef7097800b9..08e39bf1a79b4f 100644
--- a/clang/lib/Headers/wasm_simd128.h
+++ b/clang/lib/Headers/wasm_simd128.h
@@ -997,12 +997,12 @@ static __inline__ v128_t __DEFAULT_FN_ATTRS wasm_i8x16_sub(v128_t __a,
 
 static __inline__ v128_t __DEFAULT_FN_ATTRS wasm_i8x16_sub_sat(v128_t __a,
                                                                v128_t __b) {
-  return (v128_t)__builtin_wasm_sub_sat_s_i8x16((__i8x16)__a, (__i8x16)__b);
+  return (v128_t)__builtin_elementwise_sub_sat((__i8x16)__a, (__i8x16)__b);
 }
 
 static __inline__ v128_t __DEFAULT_FN_ATTRS wasm_u8x16_sub_sat(v128_t __a,
                                                                v128_t __b) {
-  return (v128_t)__builtin_wasm_sub_sat_u_i8x16((__u8x16)__a, (__u8x16)__b);
+  return (v128_t)__builtin_elementwise_sub_sat((__u8x16)__a, (__u8x16)__b);
 }
 
 static __inline__ v128_t __DEFAULT_FN_ATTRS wasm_i8x16_min(v128_t __a,
@@ -1083,12 +1083,12 @@ static __inline__ v128_t __DEFAULT_FN_ATTRS wasm_i16x8_sub(v128_t __a,
 
 static __inline__ v128_t __DEFAULT_FN_ATTRS wasm_i16x8_sub_sat(v128_t __a,
                                                                v128_t __b) {
-  return (v128_t)__builtin_wasm_sub_sat_s_i16x8((__i16x8)__a, (__i16x8)__b);
+  return (v128_t)__builtin_elementwise_sub_sat((__i16x8)__a, (__i16x8)__b);
 }
 
 static __inline__ v128_t __DEFAULT_FN_ATTRS wasm_u16x8_sub_sat(v128_t __a,
                                                                v128_t __b) {
-  return (v128_t)__builtin_wasm_sub_sat_u_i16x8((__u16x8)__a, (__u16x8)__b);
+  return (v128_t)__builtin_elementwise_sub_sat((__u16x8)__a, (__u16x8)__b);
 }
 
 static __inline__ v128_t __DEFAULT_FN_ATTRS wasm_i16x8_mul(v128_t __a,
diff --git a/clang/test/CodeGen/builtins-wasm.c b/clang/test/CodeGen/builtins-wasm.c
index ff3fe27a29345a..7ee2ac5de3eac7 100644
--- a/clang/test/CodeGen/builtins-wasm.c
+++ b/clang/test/CodeGen/builtins-wasm.c
@@ -190,23 +190,9 @@ double max_f64(double x, double y) {
   // WEBASSEMBLY-NEXT: ret
 }
 
-i8x16 sub_sat_s_i8x16(i8x16 x, i8x16 y) {
-  return __builtin_wasm_sub_sat_s_i8x16(x, y);
-  // MISSING-SIMD: error: '__builtin_wasm_sub_sat_s_i8x16' needs target feature simd128
-  // WEBASSEMBLY: call <16 x i8> @llvm.wasm.sub.sat.signed.v16i8(
-  // WEBASSEMBLY-SAME: <16 x i8> %x, <16 x i8> %y)
-  // WEBASSEMBLY-NEXT: ret
-}
-
-u8x16 sub_sat_u_i8x16(u8x16 x, u8x16 y) {
-  return __builtin_wasm_sub_sat_u_i8x16(x, y);
-  // WEBASSEMBLY: call <16 x i8> @llvm.wasm.sub.sat.unsigned.v16i8(
-  // WEBASSEMBLY-SAME: <16 x i8> %x, <16 x i8> %y)
-  // WEBASSEMBLY-NEXT: ret
-}
-
 i8x16 abs_i8x16(i8x16 v) {
   return __builtin_wasm_abs_i8x16(v);
+  // MISSING-SIMD: error: '__builtin_wasm_abs_i8x16' needs target feature simd128
   // WEBASSEMBLY: call <16 x i8> @llvm.abs.v16i8(<16 x i8> %v, i1 false)
   // WEBASSEMBLY-NEXT: ret
 }
@@ -229,20 +215,6 @@ i64x2 abs_i64x2(i64x2 v) {
   // WEBASSEMBLY-NEXT: ret
 }
 
-i16x8 sub_sat_s_i16x8(i16x8 x, i16x8 y) {
-  return __builtin_wasm_sub_sat_s_i16x8(x, y);
-  // WEBASSEMBLY: call <8 x i16> @llvm.wasm.sub.sat.signed.v8i16(
-  // WEBASSEMBLY-SAME: <8 x i16> %x, <8 x i16> %y)
-  // WEBASSEMBLY-NEXT: ret
-}
-
-u16x8 sub_sat_u_i16x8(u16x8 x, u16x8 y) {
-  return __builtin_wasm_sub_sat_u_i16x8(x, y);
-  // WEBASSEMBLY: call <8 x i16> @llvm.wasm.sub.sat.unsigned.v8i16(
-  // WEBASSEMBLY-SAME: <8 x i16> %x, <8 x i16> %y)
-  // WEBASSEMBLY-NEXT: ret
-}
-
 u8x16 avgr_u_i8x16(u8x16 x, u8x16 y) {
   return __builtin_wasm_avgr_u_i8x16(x, y);
   // WEBASSEMBLY: call <16 x i8> @llvm.wasm.avgr.unsigned.v16i8(
diff --git a/llvm/include/llvm/IR/IntrinsicsWebAssembly.td b/llvm/include/llvm/IR/IntrinsicsWebAssembly.td
index d03e532964f287..f592ff287a0e30 100644
--- a/llvm/include/llvm/IR/IntrinsicsWebAssembly.td
+++ b/llvm/include/llvm/IR/IntrinsicsWebAssembly.td
@@ -202,14 +202,6 @@ def int_wasm_shuffle :
                          ImmArg<ArgIndex<12>>, ImmArg<ArgIndex<13>>,
                          ImmArg<ArgIndex<14>>, ImmArg<ArgIndex<15>>,
                          ImmArg<ArgIndex<16>>, ImmArg<ArgIndex<17>>]>;
-def int_wasm_sub_sat_signed :
-  DefaultAttrsIntrinsic<[llvm_anyvector_ty],
-                        [LLVMMatchType<0>, LLVMMatchType<0>],
-                        [IntrNoMem, IntrSpeculatable]>;
-def int_wasm_sub_sat_unsigned :
-  DefaultAttrsIntrinsic<[llvm_anyvector_ty],
-                        [LLVMMatchType<0>, LLVMMatchType<0>],
-                        [IntrNoMem, IntrSpeculatable]>;
 def int_wasm_avgr_unsigned :
   DefaultAttrsIntrinsic<[llvm_anyvector_ty],
                         [LLVMMatchType<0>, LLVMMatchType<0>],
diff --git a/llvm/lib/Target/WebAssembly/WebAssemblyISelLowering.cpp b/llvm/lib/Target/WebAssembly/WebAssemblyISelLowering.cpp
index 1875a8fd4c4404..fa78bf38f426cd 100644
--- a/llvm/lib/Target/WebAssembly/WebAssemblyISelLowering.cpp
+++ b/llvm/lib/Target/WebAssembly/WebAssemblyISelLowering.cpp
@@ -198,8 +198,8 @@ WebAssemblyTargetLowering::WebAssemblyTargetLowering(
 
     setTargetDAGCombine(ISD::TRUNCATE);
 
-    // Support saturating add for i8x16 and i16x8
-    for (auto Op : {ISD::SADDSAT, ISD::UADDSAT})
+    // Support saturating add/sub for i8x16 and i16x8
+    for (auto Op : {ISD::SADDSAT, ISD::UADDSAT, ISD::SSUBSAT, ISD::USUBSAT})
       for (auto T : {MVT::v16i8, MVT::v8i16})
         setOperationAction(Op, T, Legal);
 
diff --git a/llvm/lib/Target/WebAssembly/WebAssemblyInstrSIMD.td b/llvm/lib/Target/WebAssembly/WebAssemblyInstrSIMD.td
index b652ee98cef107..60b3294b5f0bd0 100644
--- a/llvm/lib/Target/WebAssembly/WebAssemblyInstrSIMD.td
+++ b/llvm/lib/Target/WebAssembly/WebAssemblyInstrSIMD.td
@@ -1112,10 +1112,8 @@ defm ADD_SAT_U : SIMDBinaryIntSmall<uaddsat, "add_sat_u", 112>;
 
 // Integer subtraction: sub / sub_sat_s / sub_sat_u
 defm SUB : SIMDBinaryInt<sub, "sub", 113>;
-defm SUB_SAT_S :
-  SIMDBinaryIntSmall<int_wasm_sub_sat_signed, "sub_sat_s", 114>;
-defm SUB_SAT_U :
-  SIMDBinaryIntSmall<int_wasm_sub_sat_unsigned, "sub_sat_u", 115>;
+defm SUB_SAT_S : SIMDBinaryIntSmall<ssubsat, "sub_sat_s", 114>;
+defm SUB_SAT_U : SIMDBinaryIntSmall<usubsat, "sub_sat_u", 115>;
 
 // Integer multiplication: mul
 let isCommutable = 1 in
diff --git a/llvm/test/CodeGen/WebAssembly/simd-intrinsics.ll b/llvm/test/CodeGen/WebAssembly/simd-intrinsics.ll
index a033d2de79c877..feaea1c29e5a70 100644
--- a/llvm/test/CodeGen/WebAssembly/simd-intrinsics.ll
+++ b/llvm/test/CodeGen/WebAssembly/simd-intrinsics.ll
@@ -44,9 +44,9 @@ define <16 x i8> @add_sat_u_v16i8(<16 x i8> %x, <16 x i8> %y) {
 ; CHECK-NEXT: .functype sub_sat_s_v16i8 (v128, v128) -> (v128){{$}}
 ; CHECK-NEXT: i8x16.sub_sat_s $push[[R:[0-9]+]]=, $0, $1{{$}}
 ; CHECK-NEXT: return $pop[[R]]{{$}}
-declare <16 x i8> @llvm.wasm.sub.sat.signed.v16i8(<16 x i8>, <16 x i8>)
+declare <16 x i8> @llvm.ssub.sat.v16i8(<16 x i8>, <16 x i8>)
 define <16 x i8> @sub_sat_s_v16i8(<16 x i8> %x, <16 x i8> %y) {
-  %a = call <16 x i8> @llvm.wasm.sub.sat.signed.v16i8(
+  %a = call <16 x i8> @llvm.ssub.sat.v16i8(
     <16 x i8> %x, <16 x i8> %y
   )
   ret <16 x i8> %a
@@ -56,9 +56,9 @@ define <16 x i8> @sub_sat_s_v16i8(<16 x i8> %x, <16 x i8> %y) {
 ; CHECK-NEXT: .functype sub_sat_u_v16i8 (v128, v128) -> (v128){{$}}
 ; CHECK-NEXT: i8x16.sub_sat_u $push[[R:[0-9]+]]=, $0, $1{{$}}
 ; CHECK-NEXT: return $pop[[R]]{{$}}
-declare <16 x i8> @llvm.wasm.sub.sat.unsigned.v16i8(<16 x i8>, <16 x i8>)
+declare <16 x i8> @llvm.usub.sat.v16i8(<16 x i8>, <16 x i8>)
 define <16 x i8> @sub_sat_u_v16i8(<16 x i8> %x, <16 x i8> %y) {
-  %a = call <16 x i8> @llvm.wasm.sub.sat.unsigned.v16i8(
+  %a = call <16 x i8> @llvm.usub.sat.v16i8(
     <16 x i8> %x, <16 x i8> %y
   )
   ret <16 x i8> %a
@@ -216,9 +216,9 @@ define <8 x i16> @add_sat_u_v8i16(<8 x i16> %x, <8 x i16> %y) {
 ; CHECK-NEXT: .functype sub_sat_s_v8i16 (v128, v128) -> (v128){{$}}
 ; CHECK-NEXT: i16x8.sub_sat_s $push[[R:[0-9]+]]=, $0, $1{{$}}
 ; CHECK-NEXT: return $pop[[R]]{{$}}
-declare <8 x i16> @llvm.wasm.sub.sat.signed.v8i16(<8 x i16>, <8 x i16>)
+declare <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16>, <8 x i16>)
 define <8 x i16> @sub_sat_s_v8i16(<8 x i16> %x, <8 x i16> %y) {
-  %a = call <8 x i16> @llvm.wasm.sub.sat.signed.v8i16(
+  %a = call <8 x i16> @llvm.ssub.sat.v8i16(
     <8 x i16> %x, <8 x i16> %y
   )
   ret <8 x i16> %a
@@ -228,9 +228,9 @@ define <8 x i16> @sub_sat_s_v8i16(<8 x i16> %x, <8 x i16> %y) {
 ; CHECK-NEXT: .functype sub_sat_u_v8i16 (v128, v128) -> (v128){{$}}
 ; CHECK-NEXT: i16x8.sub_sat_u $push[[R:[0-9]+]]=, $0, $1{{$}}
 ; CHECK-NEXT: return $pop[[R]]{{$}}
-declare <8 x i16> @llvm.wasm.sub.sat.unsigned.v8i16(<8 x i16>, <8 x i16>)
+declare <8 x i16> @llvm.usub.sat.v8i16(<8 x i16>, <8 x i16>)
 define <8 x i16> @sub_sat_u_v8i16(<8 x i16> %x, <8 x i16> %y) {
-  %a = call <8 x i16> @llvm.wasm.sub.sat.unsigned.v8i16(
+  %a = call <8 x i16> @llvm.usub.sat.v8i16(
     <8 x i16> %x, <8 x i16> %y
   )
   ret <8 x i16> %a

@llvmbot
Copy link
Member

llvmbot commented Sep 20, 2024

@llvm/pr-subscribers-backend-x86

Author: Simon Pilgrim (RKSimon)

Changes

Remove the Intrinsic::wasm_sub_sat_signed/wasm_sub_sat_unsigned entries and just use sub_sat_s/sub_sat_u directly

Full diff: https://github.com/llvm/llvm-project/pull/109405.diff

8 Files Affected:

  • (modified) clang/include/clang/Basic/BuiltinsWebAssembly.def (-5)
  • (modified) clang/lib/CodeGen/CGBuiltin.cpp (-22)
  • (modified) clang/lib/Headers/wasm_simd128.h (+4-4)
  • (modified) clang/test/CodeGen/builtins-wasm.c (+1-29)
  • (modified) llvm/include/llvm/IR/IntrinsicsWebAssembly.td (-8)
  • (modified) llvm/lib/Target/WebAssembly/WebAssemblyISelLowering.cpp (+2-2)
  • (modified) llvm/lib/Target/WebAssembly/WebAssemblyInstrSIMD.td (+2-4)
  • (modified) llvm/test/CodeGen/WebAssembly/simd-intrinsics.ll (+8-8)
diff --git a/clang/include/clang/Basic/BuiltinsWebAssembly.def b/clang/include/clang/Basic/BuiltinsWebAssembly.def
index 90441a5d500120..ab480369b3820e 100644
--- a/clang/include/clang/Basic/BuiltinsWebAssembly.def
+++ b/clang/include/clang/Basic/BuiltinsWebAssembly.def
@@ -68,11 +68,6 @@ TARGET_BUILTIN(__builtin_wasm_trunc_saturate_u_i64_f64, "LLid", "nc", "nontrappi
 // SIMD builtins
 TARGET_BUILTIN(__builtin_wasm_swizzle_i8x16, "V16ScV16ScV16Sc", "nc", "simd128")
 
-TARGET_BUILTIN(__builtin_wasm_sub_sat_s_i8x16, "V16ScV16ScV16Sc", "nc", "simd128")
-TARGET_BUILTIN(__builtin_wasm_sub_sat_u_i8x16, "V16UcV16UcV16Uc", "nc", "simd128")
-TARGET_BUILTIN(__builtin_wasm_sub_sat_s_i16x8, "V8sV8sV8s", "nc", "simd128")
-TARGET_BUILTIN(__builtin_wasm_sub_sat_u_i16x8, "V8UsV8UsV8Us", "nc", "simd128")
-
 TARGET_BUILTIN(__builtin_wasm_abs_i8x16, "V16ScV16Sc", "nc", "simd128")
 TARGET_BUILTIN(__builtin_wasm_abs_i16x8, "V8sV8s", "nc", "simd128")
 TARGET_BUILTIN(__builtin_wasm_abs_i32x4, "V4iV4i", "nc", "simd128")
diff --git a/clang/lib/CodeGen/CGBuiltin.cpp b/clang/lib/CodeGen/CGBuiltin.cpp
index 3d1138b7773853..78b432474ba3b7 100644
--- a/clang/lib/CodeGen/CGBuiltin.cpp
+++ b/clang/lib/CodeGen/CGBuiltin.cpp
@@ -21443,28 +21443,6 @@ Value *CodeGenFunction::EmitWebAssemblyBuiltinExpr(unsigned BuiltinID,
     Function *Callee = CGM.getIntrinsic(Intrinsic::wasm_swizzle);
     return Builder.CreateCall(Callee, {Src, Indices});
   }
-  case WebAssembly::BI__builtin_wasm_sub_sat_s_i8x16:
-  case WebAssembly::BI__builtin_wasm_sub_sat_u_i8x16:
-  case WebAssembly::BI__builtin_wasm_sub_sat_s_i16x8:
-  case WebAssembly::BI__builtin_wasm_sub_sat_u_i16x8: {
-    unsigned IntNo;
-    switch (BuiltinID) {
-    case WebAssembly::BI__builtin_wasm_sub_sat_s_i8x16:
-    case WebAssembly::BI__builtin_wasm_sub_sat_s_i16x8:
-      IntNo = Intrinsic::wasm_sub_sat_signed;
-      break;
-    case WebAssembly::BI__builtin_wasm_sub_sat_u_i8x16:
-    case WebAssembly::BI__builtin_wasm_sub_sat_u_i16x8:
-      IntNo = Intrinsic::wasm_sub_sat_unsigned;
-      break;
-    default:
-      llvm_unreachable("unexpected builtin ID");
-    }
-    Value *LHS = EmitScalarExpr(E->getArg(0));
-    Value *RHS = EmitScalarExpr(E->getArg(1));
-    Function *Callee = CGM.getIntrinsic(IntNo, ConvertType(E->getType()));
-    return Builder.CreateCall(Callee, {LHS, RHS});
-  }
   case WebAssembly::BI__builtin_wasm_abs_i8x16:
   case WebAssembly::BI__builtin_wasm_abs_i16x8:
   case WebAssembly::BI__builtin_wasm_abs_i32x4:
diff --git a/clang/lib/Headers/wasm_simd128.h b/clang/lib/Headers/wasm_simd128.h
index b1bef7097800b9..08e39bf1a79b4f 100644
--- a/clang/lib/Headers/wasm_simd128.h
+++ b/clang/lib/Headers/wasm_simd128.h
@@ -997,12 +997,12 @@ static __inline__ v128_t __DEFAULT_FN_ATTRS wasm_i8x16_sub(v128_t __a,
 
 static __inline__ v128_t __DEFAULT_FN_ATTRS wasm_i8x16_sub_sat(v128_t __a,
                                                                v128_t __b) {
-  return (v128_t)__builtin_wasm_sub_sat_s_i8x16((__i8x16)__a, (__i8x16)__b);
+  return (v128_t)__builtin_elementwise_sub_sat((__i8x16)__a, (__i8x16)__b);
 }
 
 static __inline__ v128_t __DEFAULT_FN_ATTRS wasm_u8x16_sub_sat(v128_t __a,
                                                                v128_t __b) {
-  return (v128_t)__builtin_wasm_sub_sat_u_i8x16((__u8x16)__a, (__u8x16)__b);
+  return (v128_t)__builtin_elementwise_sub_sat((__u8x16)__a, (__u8x16)__b);
 }
 
 static __inline__ v128_t __DEFAULT_FN_ATTRS wasm_i8x16_min(v128_t __a,
@@ -1083,12 +1083,12 @@ static __inline__ v128_t __DEFAULT_FN_ATTRS wasm_i16x8_sub(v128_t __a,
 
 static __inline__ v128_t __DEFAULT_FN_ATTRS wasm_i16x8_sub_sat(v128_t __a,
                                                                v128_t __b) {
-  return (v128_t)__builtin_wasm_sub_sat_s_i16x8((__i16x8)__a, (__i16x8)__b);
+  return (v128_t)__builtin_elementwise_sub_sat((__i16x8)__a, (__i16x8)__b);
 }
 
 static __inline__ v128_t __DEFAULT_FN_ATTRS wasm_u16x8_sub_sat(v128_t __a,
                                                                v128_t __b) {
-  return (v128_t)__builtin_wasm_sub_sat_u_i16x8((__u16x8)__a, (__u16x8)__b);
+  return (v128_t)__builtin_elementwise_sub_sat((__u16x8)__a, (__u16x8)__b);
 }
 
 static __inline__ v128_t __DEFAULT_FN_ATTRS wasm_i16x8_mul(v128_t __a,
diff --git a/clang/test/CodeGen/builtins-wasm.c b/clang/test/CodeGen/builtins-wasm.c
index ff3fe27a29345a..7ee2ac5de3eac7 100644
--- a/clang/test/CodeGen/builtins-wasm.c
+++ b/clang/test/CodeGen/builtins-wasm.c
@@ -190,23 +190,9 @@ double max_f64(double x, double y) {
   // WEBASSEMBLY-NEXT: ret
 }
 
-i8x16 sub_sat_s_i8x16(i8x16 x, i8x16 y) {
-  return __builtin_wasm_sub_sat_s_i8x16(x, y);
-  // MISSING-SIMD: error: '__builtin_wasm_sub_sat_s_i8x16' needs target feature simd128
-  // WEBASSEMBLY: call <16 x i8> @llvm.wasm.sub.sat.signed.v16i8(
-  // WEBASSEMBLY-SAME: <16 x i8> %x, <16 x i8> %y)
-  // WEBASSEMBLY-NEXT: ret
-}
-
-u8x16 sub_sat_u_i8x16(u8x16 x, u8x16 y) {
-  return __builtin_wasm_sub_sat_u_i8x16(x, y);
-  // WEBASSEMBLY: call <16 x i8> @llvm.wasm.sub.sat.unsigned.v16i8(
-  // WEBASSEMBLY-SAME: <16 x i8> %x, <16 x i8> %y)
-  // WEBASSEMBLY-NEXT: ret
-}
-
 i8x16 abs_i8x16(i8x16 v) {
   return __builtin_wasm_abs_i8x16(v);
+  // MISSING-SIMD: error: '__builtin_wasm_abs_i8x16' needs target feature simd128
   // WEBASSEMBLY: call <16 x i8> @llvm.abs.v16i8(<16 x i8> %v, i1 false)
   // WEBASSEMBLY-NEXT: ret
 }
@@ -229,20 +215,6 @@ i64x2 abs_i64x2(i64x2 v) {
   // WEBASSEMBLY-NEXT: ret
 }
 
-i16x8 sub_sat_s_i16x8(i16x8 x, i16x8 y) {
-  return __builtin_wasm_sub_sat_s_i16x8(x, y);
-  // WEBASSEMBLY: call <8 x i16> @llvm.wasm.sub.sat.signed.v8i16(
-  // WEBASSEMBLY-SAME: <8 x i16> %x, <8 x i16> %y)
-  // WEBASSEMBLY-NEXT: ret
-}
-
-u16x8 sub_sat_u_i16x8(u16x8 x, u16x8 y) {
-  return __builtin_wasm_sub_sat_u_i16x8(x, y);
-  // WEBASSEMBLY: call <8 x i16> @llvm.wasm.sub.sat.unsigned.v8i16(
-  // WEBASSEMBLY-SAME: <8 x i16> %x, <8 x i16> %y)
-  // WEBASSEMBLY-NEXT: ret
-}
-
 u8x16 avgr_u_i8x16(u8x16 x, u8x16 y) {
   return __builtin_wasm_avgr_u_i8x16(x, y);
   // WEBASSEMBLY: call <16 x i8> @llvm.wasm.avgr.unsigned.v16i8(
diff --git a/llvm/include/llvm/IR/IntrinsicsWebAssembly.td b/llvm/include/llvm/IR/IntrinsicsWebAssembly.td
index d03e532964f287..f592ff287a0e30 100644
--- a/llvm/include/llvm/IR/IntrinsicsWebAssembly.td
+++ b/llvm/include/llvm/IR/IntrinsicsWebAssembly.td
@@ -202,14 +202,6 @@ def int_wasm_shuffle :
                          ImmArg<ArgIndex<12>>, ImmArg<ArgIndex<13>>,
                          ImmArg<ArgIndex<14>>, ImmArg<ArgIndex<15>>,
                          ImmArg<ArgIndex<16>>, ImmArg<ArgIndex<17>>]>;
-def int_wasm_sub_sat_signed :
-  DefaultAttrsIntrinsic<[llvm_anyvector_ty],
-                        [LLVMMatchType<0>, LLVMMatchType<0>],
-                        [IntrNoMem, IntrSpeculatable]>;
-def int_wasm_sub_sat_unsigned :
-  DefaultAttrsIntrinsic<[llvm_anyvector_ty],
-                        [LLVMMatchType<0>, LLVMMatchType<0>],
-                        [IntrNoMem, IntrSpeculatable]>;
 def int_wasm_avgr_unsigned :
   DefaultAttrsIntrinsic<[llvm_anyvector_ty],
                         [LLVMMatchType<0>, LLVMMatchType<0>],
diff --git a/llvm/lib/Target/WebAssembly/WebAssemblyISelLowering.cpp b/llvm/lib/Target/WebAssembly/WebAssemblyISelLowering.cpp
index 1875a8fd4c4404..fa78bf38f426cd 100644
--- a/llvm/lib/Target/WebAssembly/WebAssemblyISelLowering.cpp
+++ b/llvm/lib/Target/WebAssembly/WebAssemblyISelLowering.cpp
@@ -198,8 +198,8 @@ WebAssemblyTargetLowering::WebAssemblyTargetLowering(
 
     setTargetDAGCombine(ISD::TRUNCATE);
 
-    // Support saturating add for i8x16 and i16x8
-    for (auto Op : {ISD::SADDSAT, ISD::UADDSAT})
+    // Support saturating add/sub for i8x16 and i16x8
+    for (auto Op : {ISD::SADDSAT, ISD::UADDSAT, ISD::SSUBSAT, ISD::USUBSAT})
       for (auto T : {MVT::v16i8, MVT::v8i16})
         setOperationAction(Op, T, Legal);
 
diff --git a/llvm/lib/Target/WebAssembly/WebAssemblyInstrSIMD.td b/llvm/lib/Target/WebAssembly/WebAssemblyInstrSIMD.td
index b652ee98cef107..60b3294b5f0bd0 100644
--- a/llvm/lib/Target/WebAssembly/WebAssemblyInstrSIMD.td
+++ b/llvm/lib/Target/WebAssembly/WebAssemblyInstrSIMD.td
@@ -1112,10 +1112,8 @@ defm ADD_SAT_U : SIMDBinaryIntSmall<uaddsat, "add_sat_u", 112>;
 
 // Integer subtraction: sub / sub_sat_s / sub_sat_u
 defm SUB : SIMDBinaryInt<sub, "sub", 113>;
-defm SUB_SAT_S :
-  SIMDBinaryIntSmall<int_wasm_sub_sat_signed, "sub_sat_s", 114>;
-defm SUB_SAT_U :
-  SIMDBinaryIntSmall<int_wasm_sub_sat_unsigned, "sub_sat_u", 115>;
+defm SUB_SAT_S : SIMDBinaryIntSmall<ssubsat, "sub_sat_s", 114>;
+defm SUB_SAT_U : SIMDBinaryIntSmall<usubsat, "sub_sat_u", 115>;
 
 // Integer multiplication: mul
 let isCommutable = 1 in
diff --git a/llvm/test/CodeGen/WebAssembly/simd-intrinsics.ll b/llvm/test/CodeGen/WebAssembly/simd-intrinsics.ll
index a033d2de79c877..feaea1c29e5a70 100644
--- a/llvm/test/CodeGen/WebAssembly/simd-intrinsics.ll
+++ b/llvm/test/CodeGen/WebAssembly/simd-intrinsics.ll
@@ -44,9 +44,9 @@ define <16 x i8> @add_sat_u_v16i8(<16 x i8> %x, <16 x i8> %y) {
 ; CHECK-NEXT: .functype sub_sat_s_v16i8 (v128, v128) -> (v128){{$}}
 ; CHECK-NEXT: i8x16.sub_sat_s $push[[R:[0-9]+]]=, $0, $1{{$}}
 ; CHECK-NEXT: return $pop[[R]]{{$}}
-declare <16 x i8> @llvm.wasm.sub.sat.signed.v16i8(<16 x i8>, <16 x i8>)
+declare <16 x i8> @llvm.ssub.sat.v16i8(<16 x i8>, <16 x i8>)
 define <16 x i8> @sub_sat_s_v16i8(<16 x i8> %x, <16 x i8> %y) {
-  %a = call <16 x i8> @llvm.wasm.sub.sat.signed.v16i8(
+  %a = call <16 x i8> @llvm.ssub.sat.v16i8(
     <16 x i8> %x, <16 x i8> %y
   )
   ret <16 x i8> %a
@@ -56,9 +56,9 @@ define <16 x i8> @sub_sat_s_v16i8(<16 x i8> %x, <16 x i8> %y) {
 ; CHECK-NEXT: .functype sub_sat_u_v16i8 (v128, v128) -> (v128){{$}}
 ; CHECK-NEXT: i8x16.sub_sat_u $push[[R:[0-9]+]]=, $0, $1{{$}}
 ; CHECK-NEXT: return $pop[[R]]{{$}}
-declare <16 x i8> @llvm.wasm.sub.sat.unsigned.v16i8(<16 x i8>, <16 x i8>)
+declare <16 x i8> @llvm.usub.sat.v16i8(<16 x i8>, <16 x i8>)
 define <16 x i8> @sub_sat_u_v16i8(<16 x i8> %x, <16 x i8> %y) {
-  %a = call <16 x i8> @llvm.wasm.sub.sat.unsigned.v16i8(
+  %a = call <16 x i8> @llvm.usub.sat.v16i8(
     <16 x i8> %x, <16 x i8> %y
   )
   ret <16 x i8> %a
@@ -216,9 +216,9 @@ define <8 x i16> @add_sat_u_v8i16(<8 x i16> %x, <8 x i16> %y) {
 ; CHECK-NEXT: .functype sub_sat_s_v8i16 (v128, v128) -> (v128){{$}}
 ; CHECK-NEXT: i16x8.sub_sat_s $push[[R:[0-9]+]]=, $0, $1{{$}}
 ; CHECK-NEXT: return $pop[[R]]{{$}}
-declare <8 x i16> @llvm.wasm.sub.sat.signed.v8i16(<8 x i16>, <8 x i16>)
+declare <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16>, <8 x i16>)
 define <8 x i16> @sub_sat_s_v8i16(<8 x i16> %x, <8 x i16> %y) {
-  %a = call <8 x i16> @llvm.wasm.sub.sat.signed.v8i16(
+  %a = call <8 x i16> @llvm.ssub.sat.v8i16(
     <8 x i16> %x, <8 x i16> %y
   )
   ret <8 x i16> %a
@@ -228,9 +228,9 @@ define <8 x i16> @sub_sat_s_v8i16(<8 x i16> %x, <8 x i16> %y) {
 ; CHECK-NEXT: .functype sub_sat_u_v8i16 (v128, v128) -> (v128){{$}}
 ; CHECK-NEXT: i16x8.sub_sat_u $push[[R:[0-9]+]]=, $0, $1{{$}}
 ; CHECK-NEXT: return $pop[[R]]{{$}}
-declare <8 x i16> @llvm.wasm.sub.sat.unsigned.v8i16(<8 x i16>, <8 x i16>)
+declare <8 x i16> @llvm.usub.sat.v8i16(<8 x i16>, <8 x i16>)
 define <8 x i16> @sub_sat_u_v8i16(<8 x i16> %x, <8 x i16> %y) {
-  %a = call <8 x i16> @llvm.wasm.sub.sat.unsigned.v8i16(
+  %a = call <8 x i16> @llvm.usub.sat.v8i16(
     <8 x i16> %x, <8 x i16> %y
   )
   ret <8 x i16> %a

@llvmbot
Copy link
Member

llvmbot commented Sep 20, 2024

@llvm/pr-subscribers-llvm-ir

Author: Simon Pilgrim (RKSimon)

Changes

Remove the Intrinsic::wasm_sub_sat_signed/wasm_sub_sat_unsigned entries and just use sub_sat_s/sub_sat_u directly

Full diff: https://github.com/llvm/llvm-project/pull/109405.diff

8 Files Affected:

  • (modified) clang/include/clang/Basic/BuiltinsWebAssembly.def (-5)
  • (modified) clang/lib/CodeGen/CGBuiltin.cpp (-22)
  • (modified) clang/lib/Headers/wasm_simd128.h (+4-4)
  • (modified) clang/test/CodeGen/builtins-wasm.c (+1-29)
  • (modified) llvm/include/llvm/IR/IntrinsicsWebAssembly.td (-8)
  • (modified) llvm/lib/Target/WebAssembly/WebAssemblyISelLowering.cpp (+2-2)
  • (modified) llvm/lib/Target/WebAssembly/WebAssemblyInstrSIMD.td (+2-4)
  • (modified) llvm/test/CodeGen/WebAssembly/simd-intrinsics.ll (+8-8)
diff --git a/clang/include/clang/Basic/BuiltinsWebAssembly.def b/clang/include/clang/Basic/BuiltinsWebAssembly.def
index 90441a5d500120..ab480369b3820e 100644
--- a/clang/include/clang/Basic/BuiltinsWebAssembly.def
+++ b/clang/include/clang/Basic/BuiltinsWebAssembly.def
@@ -68,11 +68,6 @@ TARGET_BUILTIN(__builtin_wasm_trunc_saturate_u_i64_f64, "LLid", "nc", "nontrappi
 // SIMD builtins
 TARGET_BUILTIN(__builtin_wasm_swizzle_i8x16, "V16ScV16ScV16Sc", "nc", "simd128")
 
-TARGET_BUILTIN(__builtin_wasm_sub_sat_s_i8x16, "V16ScV16ScV16Sc", "nc", "simd128")
-TARGET_BUILTIN(__builtin_wasm_sub_sat_u_i8x16, "V16UcV16UcV16Uc", "nc", "simd128")
-TARGET_BUILTIN(__builtin_wasm_sub_sat_s_i16x8, "V8sV8sV8s", "nc", "simd128")
-TARGET_BUILTIN(__builtin_wasm_sub_sat_u_i16x8, "V8UsV8UsV8Us", "nc", "simd128")
-
 TARGET_BUILTIN(__builtin_wasm_abs_i8x16, "V16ScV16Sc", "nc", "simd128")
 TARGET_BUILTIN(__builtin_wasm_abs_i16x8, "V8sV8s", "nc", "simd128")
 TARGET_BUILTIN(__builtin_wasm_abs_i32x4, "V4iV4i", "nc", "simd128")
diff --git a/clang/lib/CodeGen/CGBuiltin.cpp b/clang/lib/CodeGen/CGBuiltin.cpp
index 3d1138b7773853..78b432474ba3b7 100644
--- a/clang/lib/CodeGen/CGBuiltin.cpp
+++ b/clang/lib/CodeGen/CGBuiltin.cpp
@@ -21443,28 +21443,6 @@ Value *CodeGenFunction::EmitWebAssemblyBuiltinExpr(unsigned BuiltinID,
     Function *Callee = CGM.getIntrinsic(Intrinsic::wasm_swizzle);
     return Builder.CreateCall(Callee, {Src, Indices});
   }
-  case WebAssembly::BI__builtin_wasm_sub_sat_s_i8x16:
-  case WebAssembly::BI__builtin_wasm_sub_sat_u_i8x16:
-  case WebAssembly::BI__builtin_wasm_sub_sat_s_i16x8:
-  case WebAssembly::BI__builtin_wasm_sub_sat_u_i16x8: {
-    unsigned IntNo;
-    switch (BuiltinID) {
-    case WebAssembly::BI__builtin_wasm_sub_sat_s_i8x16:
-    case WebAssembly::BI__builtin_wasm_sub_sat_s_i16x8:
-      IntNo = Intrinsic::wasm_sub_sat_signed;
-      break;
-    case WebAssembly::BI__builtin_wasm_sub_sat_u_i8x16:
-    case WebAssembly::BI__builtin_wasm_sub_sat_u_i16x8:
-      IntNo = Intrinsic::wasm_sub_sat_unsigned;
-      break;
-    default:
-      llvm_unreachable("unexpected builtin ID");
-    }
-    Value *LHS = EmitScalarExpr(E->getArg(0));
-    Value *RHS = EmitScalarExpr(E->getArg(1));
-    Function *Callee = CGM.getIntrinsic(IntNo, ConvertType(E->getType()));
-    return Builder.CreateCall(Callee, {LHS, RHS});
-  }
   case WebAssembly::BI__builtin_wasm_abs_i8x16:
   case WebAssembly::BI__builtin_wasm_abs_i16x8:
   case WebAssembly::BI__builtin_wasm_abs_i32x4:
diff --git a/clang/lib/Headers/wasm_simd128.h b/clang/lib/Headers/wasm_simd128.h
index b1bef7097800b9..08e39bf1a79b4f 100644
--- a/clang/lib/Headers/wasm_simd128.h
+++ b/clang/lib/Headers/wasm_simd128.h
@@ -997,12 +997,12 @@ static __inline__ v128_t __DEFAULT_FN_ATTRS wasm_i8x16_sub(v128_t __a,
 
 static __inline__ v128_t __DEFAULT_FN_ATTRS wasm_i8x16_sub_sat(v128_t __a,
                                                                v128_t __b) {
-  return (v128_t)__builtin_wasm_sub_sat_s_i8x16((__i8x16)__a, (__i8x16)__b);
+  return (v128_t)__builtin_elementwise_sub_sat((__i8x16)__a, (__i8x16)__b);
 }
 
 static __inline__ v128_t __DEFAULT_FN_ATTRS wasm_u8x16_sub_sat(v128_t __a,
                                                                v128_t __b) {
-  return (v128_t)__builtin_wasm_sub_sat_u_i8x16((__u8x16)__a, (__u8x16)__b);
+  return (v128_t)__builtin_elementwise_sub_sat((__u8x16)__a, (__u8x16)__b);
 }
 
 static __inline__ v128_t __DEFAULT_FN_ATTRS wasm_i8x16_min(v128_t __a,
@@ -1083,12 +1083,12 @@ static __inline__ v128_t __DEFAULT_FN_ATTRS wasm_i16x8_sub(v128_t __a,
 
 static __inline__ v128_t __DEFAULT_FN_ATTRS wasm_i16x8_sub_sat(v128_t __a,
                                                                v128_t __b) {
-  return (v128_t)__builtin_wasm_sub_sat_s_i16x8((__i16x8)__a, (__i16x8)__b);
+  return (v128_t)__builtin_elementwise_sub_sat((__i16x8)__a, (__i16x8)__b);
 }
 
 static __inline__ v128_t __DEFAULT_FN_ATTRS wasm_u16x8_sub_sat(v128_t __a,
                                                                v128_t __b) {
-  return (v128_t)__builtin_wasm_sub_sat_u_i16x8((__u16x8)__a, (__u16x8)__b);
+  return (v128_t)__builtin_elementwise_sub_sat((__u16x8)__a, (__u16x8)__b);
 }
 
 static __inline__ v128_t __DEFAULT_FN_ATTRS wasm_i16x8_mul(v128_t __a,
diff --git a/clang/test/CodeGen/builtins-wasm.c b/clang/test/CodeGen/builtins-wasm.c
index ff3fe27a29345a..7ee2ac5de3eac7 100644
--- a/clang/test/CodeGen/builtins-wasm.c
+++ b/clang/test/CodeGen/builtins-wasm.c
@@ -190,23 +190,9 @@ double max_f64(double x, double y) {
   // WEBASSEMBLY-NEXT: ret
 }
 
-i8x16 sub_sat_s_i8x16(i8x16 x, i8x16 y) {
-  return __builtin_wasm_sub_sat_s_i8x16(x, y);
-  // MISSING-SIMD: error: '__builtin_wasm_sub_sat_s_i8x16' needs target feature simd128
-  // WEBASSEMBLY: call <16 x i8> @llvm.wasm.sub.sat.signed.v16i8(
-  // WEBASSEMBLY-SAME: <16 x i8> %x, <16 x i8> %y)
-  // WEBASSEMBLY-NEXT: ret
-}
-
-u8x16 sub_sat_u_i8x16(u8x16 x, u8x16 y) {
-  return __builtin_wasm_sub_sat_u_i8x16(x, y);
-  // WEBASSEMBLY: call <16 x i8> @llvm.wasm.sub.sat.unsigned.v16i8(
-  // WEBASSEMBLY-SAME: <16 x i8> %x, <16 x i8> %y)
-  // WEBASSEMBLY-NEXT: ret
-}
-
 i8x16 abs_i8x16(i8x16 v) {
   return __builtin_wasm_abs_i8x16(v);
+  // MISSING-SIMD: error: '__builtin_wasm_abs_i8x16' needs target feature simd128
   // WEBASSEMBLY: call <16 x i8> @llvm.abs.v16i8(<16 x i8> %v, i1 false)
   // WEBASSEMBLY-NEXT: ret
 }
@@ -229,20 +215,6 @@ i64x2 abs_i64x2(i64x2 v) {
   // WEBASSEMBLY-NEXT: ret
 }
 
-i16x8 sub_sat_s_i16x8(i16x8 x, i16x8 y) {
-  return __builtin_wasm_sub_sat_s_i16x8(x, y);
-  // WEBASSEMBLY: call <8 x i16> @llvm.wasm.sub.sat.signed.v8i16(
-  // WEBASSEMBLY-SAME: <8 x i16> %x, <8 x i16> %y)
-  // WEBASSEMBLY-NEXT: ret
-}
-
-u16x8 sub_sat_u_i16x8(u16x8 x, u16x8 y) {
-  return __builtin_wasm_sub_sat_u_i16x8(x, y);
-  // WEBASSEMBLY: call <8 x i16> @llvm.wasm.sub.sat.unsigned.v8i16(
-  // WEBASSEMBLY-SAME: <8 x i16> %x, <8 x i16> %y)
-  // WEBASSEMBLY-NEXT: ret
-}
-
 u8x16 avgr_u_i8x16(u8x16 x, u8x16 y) {
   return __builtin_wasm_avgr_u_i8x16(x, y);
   // WEBASSEMBLY: call <16 x i8> @llvm.wasm.avgr.unsigned.v16i8(
diff --git a/llvm/include/llvm/IR/IntrinsicsWebAssembly.td b/llvm/include/llvm/IR/IntrinsicsWebAssembly.td
index d03e532964f287..f592ff287a0e30 100644
--- a/llvm/include/llvm/IR/IntrinsicsWebAssembly.td
+++ b/llvm/include/llvm/IR/IntrinsicsWebAssembly.td
@@ -202,14 +202,6 @@ def int_wasm_shuffle :
                          ImmArg<ArgIndex<12>>, ImmArg<ArgIndex<13>>,
                          ImmArg<ArgIndex<14>>, ImmArg<ArgIndex<15>>,
                          ImmArg<ArgIndex<16>>, ImmArg<ArgIndex<17>>]>;
-def int_wasm_sub_sat_signed :
-  DefaultAttrsIntrinsic<[llvm_anyvector_ty],
-                        [LLVMMatchType<0>, LLVMMatchType<0>],
-                        [IntrNoMem, IntrSpeculatable]>;
-def int_wasm_sub_sat_unsigned :
-  DefaultAttrsIntrinsic<[llvm_anyvector_ty],
-                        [LLVMMatchType<0>, LLVMMatchType<0>],
-                        [IntrNoMem, IntrSpeculatable]>;
 def int_wasm_avgr_unsigned :
   DefaultAttrsIntrinsic<[llvm_anyvector_ty],
                         [LLVMMatchType<0>, LLVMMatchType<0>],
diff --git a/llvm/lib/Target/WebAssembly/WebAssemblyISelLowering.cpp b/llvm/lib/Target/WebAssembly/WebAssemblyISelLowering.cpp
index 1875a8fd4c4404..fa78bf38f426cd 100644
--- a/llvm/lib/Target/WebAssembly/WebAssemblyISelLowering.cpp
+++ b/llvm/lib/Target/WebAssembly/WebAssemblyISelLowering.cpp
@@ -198,8 +198,8 @@ WebAssemblyTargetLowering::WebAssemblyTargetLowering(
 
     setTargetDAGCombine(ISD::TRUNCATE);
 
-    // Support saturating add for i8x16 and i16x8
-    for (auto Op : {ISD::SADDSAT, ISD::UADDSAT})
+    // Support saturating add/sub for i8x16 and i16x8
+    for (auto Op : {ISD::SADDSAT, ISD::UADDSAT, ISD::SSUBSAT, ISD::USUBSAT})
       for (auto T : {MVT::v16i8, MVT::v8i16})
         setOperationAction(Op, T, Legal);
 
diff --git a/llvm/lib/Target/WebAssembly/WebAssemblyInstrSIMD.td b/llvm/lib/Target/WebAssembly/WebAssemblyInstrSIMD.td
index b652ee98cef107..60b3294b5f0bd0 100644
--- a/llvm/lib/Target/WebAssembly/WebAssemblyInstrSIMD.td
+++ b/llvm/lib/Target/WebAssembly/WebAssemblyInstrSIMD.td
@@ -1112,10 +1112,8 @@ defm ADD_SAT_U : SIMDBinaryIntSmall<uaddsat, "add_sat_u", 112>;
 
 // Integer subtraction: sub / sub_sat_s / sub_sat_u
 defm SUB : SIMDBinaryInt<sub, "sub", 113>;
-defm SUB_SAT_S :
-  SIMDBinaryIntSmall<int_wasm_sub_sat_signed, "sub_sat_s", 114>;
-defm SUB_SAT_U :
-  SIMDBinaryIntSmall<int_wasm_sub_sat_unsigned, "sub_sat_u", 115>;
+defm SUB_SAT_S : SIMDBinaryIntSmall<ssubsat, "sub_sat_s", 114>;
+defm SUB_SAT_U : SIMDBinaryIntSmall<usubsat, "sub_sat_u", 115>;
 
 // Integer multiplication: mul
 let isCommutable = 1 in
diff --git a/llvm/test/CodeGen/WebAssembly/simd-intrinsics.ll b/llvm/test/CodeGen/WebAssembly/simd-intrinsics.ll
index a033d2de79c877..feaea1c29e5a70 100644
--- a/llvm/test/CodeGen/WebAssembly/simd-intrinsics.ll
+++ b/llvm/test/CodeGen/WebAssembly/simd-intrinsics.ll
@@ -44,9 +44,9 @@ define <16 x i8> @add_sat_u_v16i8(<16 x i8> %x, <16 x i8> %y) {
 ; CHECK-NEXT: .functype sub_sat_s_v16i8 (v128, v128) -> (v128){{$}}
 ; CHECK-NEXT: i8x16.sub_sat_s $push[[R:[0-9]+]]=, $0, $1{{$}}
 ; CHECK-NEXT: return $pop[[R]]{{$}}
-declare <16 x i8> @llvm.wasm.sub.sat.signed.v16i8(<16 x i8>, <16 x i8>)
+declare <16 x i8> @llvm.ssub.sat.v16i8(<16 x i8>, <16 x i8>)
 define <16 x i8> @sub_sat_s_v16i8(<16 x i8> %x, <16 x i8> %y) {
-  %a = call <16 x i8> @llvm.wasm.sub.sat.signed.v16i8(
+  %a = call <16 x i8> @llvm.ssub.sat.v16i8(
     <16 x i8> %x, <16 x i8> %y
   )
   ret <16 x i8> %a
@@ -56,9 +56,9 @@ define <16 x i8> @sub_sat_s_v16i8(<16 x i8> %x, <16 x i8> %y) {
 ; CHECK-NEXT: .functype sub_sat_u_v16i8 (v128, v128) -> (v128){{$}}
 ; CHECK-NEXT: i8x16.sub_sat_u $push[[R:[0-9]+]]=, $0, $1{{$}}
 ; CHECK-NEXT: return $pop[[R]]{{$}}
-declare <16 x i8> @llvm.wasm.sub.sat.unsigned.v16i8(<16 x i8>, <16 x i8>)
+declare <16 x i8> @llvm.usub.sat.v16i8(<16 x i8>, <16 x i8>)
 define <16 x i8> @sub_sat_u_v16i8(<16 x i8> %x, <16 x i8> %y) {
-  %a = call <16 x i8> @llvm.wasm.sub.sat.unsigned.v16i8(
+  %a = call <16 x i8> @llvm.usub.sat.v16i8(
     <16 x i8> %x, <16 x i8> %y
   )
   ret <16 x i8> %a
@@ -216,9 +216,9 @@ define <8 x i16> @add_sat_u_v8i16(<8 x i16> %x, <8 x i16> %y) {
 ; CHECK-NEXT: .functype sub_sat_s_v8i16 (v128, v128) -> (v128){{$}}
 ; CHECK-NEXT: i16x8.sub_sat_s $push[[R:[0-9]+]]=, $0, $1{{$}}
 ; CHECK-NEXT: return $pop[[R]]{{$}}
-declare <8 x i16> @llvm.wasm.sub.sat.signed.v8i16(<8 x i16>, <8 x i16>)
+declare <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16>, <8 x i16>)
 define <8 x i16> @sub_sat_s_v8i16(<8 x i16> %x, <8 x i16> %y) {
-  %a = call <8 x i16> @llvm.wasm.sub.sat.signed.v8i16(
+  %a = call <8 x i16> @llvm.ssub.sat.v8i16(
     <8 x i16> %x, <8 x i16> %y
   )
   ret <8 x i16> %a
@@ -228,9 +228,9 @@ define <8 x i16> @sub_sat_s_v8i16(<8 x i16> %x, <8 x i16> %y) {
 ; CHECK-NEXT: .functype sub_sat_u_v8i16 (v128, v128) -> (v128){{$}}
 ; CHECK-NEXT: i16x8.sub_sat_u $push[[R:[0-9]+]]=, $0, $1{{$}}
 ; CHECK-NEXT: return $pop[[R]]{{$}}
-declare <8 x i16> @llvm.wasm.sub.sat.unsigned.v8i16(<8 x i16>, <8 x i16>)
+declare <8 x i16> @llvm.usub.sat.v8i16(<8 x i16>, <8 x i16>)
 define <8 x i16> @sub_sat_u_v8i16(<8 x i16> %x, <8 x i16> %y) {
-  %a = call <8 x i16> @llvm.wasm.sub.sat.unsigned.v8i16(
+  %a = call <8 x i16> @llvm.usub.sat.v8i16(
     <8 x i16> %x, <8 x i16> %y
   )
   ret <8 x i16> %a

tlively
tlively approved these changes Sep 20, 2024
View reviewed changes
Copy link
Collaborator

@tlively tlively left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thank you!

@RKSimon
[clang][wasm] Replace the target integer sub saturate intrinsics with…
657f1c0 
… the equivalent generic `__builtin_elementwise_sub_sat` intrinsics

Remove the Intrinsic::wasm_sub_sat_signed/wasm_sub_sat_unsigned entries and just use sub_sat_s/sub_sat_u directly
@RKSimon
Fix header tests
d812627
@RKSimon RKSimon merged commit f8f0a26 into llvm:main Sep 22, 2024
8 checks passed
@RKSimon RKSimon deleted the wasm-sub-sat branch September 22, 2024 09:12
steven-johnson added a commit to halide/Halide that referenced this pull request Sep 23, 2024
@steven-johnson
Fix for top-of-tree LLVM
9eea28d 
The wasm backend should no longer specialize llvm.wasm.sub.sat* intrinsics, see llvm/llvm-project#109405
@steven-johnson steven-johnson mentioned this pull request Sep 23, 2024
Merged
steven-johnson added a commit to halide/Halide that referenced this pull request Sep 24, 2024
@steven-johnson
Fix for top-of-tree LLVM (#8425)
e55bd6f 
* Fix for top-of-tree LLVM

The wasm backend should no longer specialize llvm.wasm.sub.sat* intrinsics, see llvm/llvm-project#109405

* trigger buildbots
nikic added a commit to nikic/stdarch that referenced this pull request Feb 19, 2025
@nikic
Update wasm sub sat intrinsics for LLVM 20
7cfd8a2 
The wasm-specific intrinsics have been removed in
llvm/llvm-project#109405, use the generic
ones instead.
@nikic nikic mentioned this pull request Feb 19, 2025
Merged
github-merge-queue bot pushed a commit to rust-lang/stdarch that referenced this pull request Feb 19, 2025
@nikic @Amanieu
Update wasm sub sat intrinsics for LLVM 20
a70790e 
The wasm-specific intrinsics have been removed in
llvm/llvm-project#109405, use the generic
ones instead.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@tlively tlively tlively approved these changes

Assignees
No one assigned
Labels
backend:WebAssembly backend:X86 clang:codegen IR generation bugs: mangling, exceptions, etc. clang:frontend Language frontend issues, e.g. anything involving "Sema" clang:headers Headers provided by Clang, e.g. for intrinsics clang Clang issues not falling into any other category llvm:ir
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@RKSimon @llvmbot @tlively