AArch64: Prefetch intrinsic

llvm-svn: 215569

clang/include/clang/Basic/BuiltinsAArch64.def

@@ -50,4 +50,7 @@ BUILTIN(__builtin_arm_dmb, "vUi", "nc")
 BUILTIN(__builtin_arm_dsb, "vUi", "nc")
 BUILTIN(__builtin_arm_isb, "vUi", "nc")
 
+// Prefetch
+BUILTIN(__builtin_arm_prefetch, "vvC*UiUiUiUi", "nc")
+
 #undef BUILTIN

clang/lib/CodeGen/CGBuiltin.cpp

@@ -3854,6 +3854,29 @@ Value *CodeGenFunction::EmitAArch64BuiltinExpr(unsigned BuiltinID,
     return Builder.CreateCall(F, llvm::ConstantInt::get(Int32Ty, HintID));
   }
 
+  if (BuiltinID == AArch64::BI__builtin_arm_prefetch) {
+    Value *Address         = EmitScalarExpr(E->getArg(0));
+    Value *RW              = EmitScalarExpr(E->getArg(1));
+    Value *CacheLevel      = EmitScalarExpr(E->getArg(2));
+    Value *RetentionPolicy = EmitScalarExpr(E->getArg(3));
+    Value *IsData          = EmitScalarExpr(E->getArg(4));
+
+    Value *Locality = nullptr;
+    if (cast<llvm::ConstantInt>(RetentionPolicy)->isZero()) {
+      // Temporal fetch, needs to convert cache level to locality.
+      Locality = llvm::ConstantInt::get(Int32Ty,
+        -cast<llvm::ConstantInt>(CacheLevel)->getValue() + 3);
+    } else {
+      // Streaming fetch.
+      Locality = llvm::ConstantInt::get(Int32Ty, 0);
+    }
+
+    // FIXME: We need AArch64 specific LLVM intrinsic if we want to specify
+    // PLDL3STRM or PLDL2STRM.
+    Value *F = CGM.getIntrinsic(Intrinsic::prefetch);
+    return Builder.CreateCall4(F, Address, RW, Locality, IsData);
+  }
+
   if (BuiltinID == AArch64::BI__builtin_arm_rbit) {
     assert((getContext().getTypeSize(E->getType()) == 32) &&
            "rbit of unusual size!");

clang/lib/Sema/SemaChecking.cpp

@@ -661,6 +661,13 @@ bool Sema::CheckAArch64BuiltinFunctionCall(unsigned BuiltinID,
     return CheckARMBuiltinExclusiveCall(BuiltinID, TheCall, 128);
   }
 
+  if (BuiltinID == AArch64::BI__builtin_arm_prefetch) {
+    return SemaBuiltinConstantArgRange(TheCall, 1, 0, 1) ||
+      SemaBuiltinConstantArgRange(TheCall, 2, 0, 2) ||
+      SemaBuiltinConstantArgRange(TheCall, 3, 0, 1) ||
+      SemaBuiltinConstantArgRange(TheCall, 4, 0, 1);
+  }
+
   if (CheckNeonBuiltinFunctionCall(BuiltinID, TheCall))
     return true;
 

clang/test/CodeGen/builtins-arm64.c

@@ -29,3 +29,17 @@ void barriers() {
   __builtin_arm_dsb(2);  //CHECK: call {{.*}} @llvm.aarch64.dsb(i32 2)
   __builtin_arm_isb(3);  //CHECK: call {{.*}} @llvm.aarch64.isb(i32 3)
 }
+
+void prefetch() {
+  __builtin_arm_prefetch(0, 1, 2, 0, 1); // pstl3keep
+// CHECK: call {{.*}} @llvm.prefetch(i8* null, i32 1, i32 1, i32 1)
+
+  __builtin_arm_prefetch(0, 0, 0, 1, 1); // pldl1keep
+// CHECK: call {{.*}} @llvm.prefetch(i8* null, i32 0, i32 0, i32 1)
+
+  __builtin_arm_prefetch(0, 0, 0, 1, 1); // pldl1strm
+// CHECK: call {{.*}} @llvm.prefetch(i8* null, i32 0, i32 0, i32 1)
+
+  __builtin_arm_prefetch(0, 0, 0, 0, 0); // plil1keep
+// CHECK: call {{.*}} @llvm.prefetch(i8* null, i32 0, i32 3, i32 0)
+}

clang/test/Sema/builtins-arm64.c

@@ -22,3 +22,10 @@ void test_memory_barriers() {
   __builtin_arm_dsb(17); // expected-error {{argument should be a value from 0 to 15}}
   __builtin_arm_isb(18); // expected-error {{argument should be a value from 0 to 15}}
 }
+
+void test_prefetch() {
+  __builtin_arm_prefetch(0, 2, 0, 0, 0); // expected-error {{argument should be a value from 0 to 1}}
+  __builtin_arm_prefetch(0, 0, 3, 0, 0); // expected-error {{argument should be a value from 0 to 2}}
+  __builtin_arm_prefetch(0, 0, 0, 2, 0); // expected-error {{argument should be a value from 0 to 1}}
+  __builtin_arm_prefetch(0, 0, 0, 0, 2); // expected-error {{argument should be a value from 0 to 1}}
+}