From 85974bc77e38946c64e5d361d78341fda82e1bbb Mon Sep 17 00:00:00 2001
From: Tim Northover <tnorthover@apple.com>
Date: Tue, 22 Apr 2014 10:10:14 +0000
Subject: [PATCH] AArch64/ARM64: mark fmul intrinsic as commutative.

This gives DAG patterns matching indexed patterns where either side is an
indexed vector.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206875 91177308-0d34-0410-b5e6-96231b3b80d8
---
 include/llvm/IR/IntrinsicsARM64.td            |   4 +-
 .../AArch64/neon-scalar-by-elem-mul.ll        |   2 +-
 .../ARM64/aarch64-neon-scalar-by-elem-mul.ll  | 124 ++++++++++++++++++
 3 files changed, 128 insertions(+), 2 deletions(-)
 create mode 100644 test/CodeGen/ARM64/aarch64-neon-scalar-by-elem-mul.ll

diff --git a/include/llvm/IR/IntrinsicsARM64.td b/include/llvm/IR/IntrinsicsARM64.td
index abdb655b122..146ea5d970c 100644
--- a/include/llvm/IR/IntrinsicsARM64.td
+++ b/include/llvm/IR/IntrinsicsARM64.td
@@ -189,7 +189,9 @@ let Properties = [IntrNoMem] in {
         Intrinsic<[llvm_v16i8_ty], [llvm_i64_ty, llvm_i64_ty], [IntrNoMem]>;
 
   // Vector Extending Multiply
-  def int_arm64_neon_fmulx : AdvSIMD_2FloatArg_Intrinsic;
+  def int_arm64_neon_fmulx : AdvSIMD_2FloatArg_Intrinsic {
+    let Properties = [IntrNoMem, Commutative];
+  }
 
   // Vector Saturating Doubling Long Multiply
   def int_arm64_neon_sqdmull : AdvSIMD_2VectorArg_Long_Intrinsic;
diff --git a/test/CodeGen/AArch64/neon-scalar-by-elem-mul.ll b/test/CodeGen/AArch64/neon-scalar-by-elem-mul.ll
index c9128e7e535..ff294132520 100644
--- a/test/CodeGen/AArch64/neon-scalar-by-elem-mul.ll
+++ b/test/CodeGen/AArch64/neon-scalar-by-elem-mul.ll
@@ -1,5 +1,5 @@
 ; RUN: llc < %s -verify-machineinstrs -mtriple=aarch64-none-linux-gnu -mattr=+neon -fp-contract=fast | FileCheck %s
-
+; arm64 has separate copy due to intrinsics (aarch64-neon-scalar-by-elem-mul.ll)
 define float @test_fmul_lane_ss2S(float %a, <2 x float> %v) {
   ; CHECK: test_fmul_lane_ss2S
   ; CHECK: fmul {{s[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}.s[1]
diff --git a/test/CodeGen/ARM64/aarch64-neon-scalar-by-elem-mul.ll b/test/CodeGen/ARM64/aarch64-neon-scalar-by-elem-mul.ll
new file mode 100644
index 00000000000..18e6e0fe8b6
--- /dev/null
+++ b/test/CodeGen/ARM64/aarch64-neon-scalar-by-elem-mul.ll
@@ -0,0 +1,124 @@
+; RUN: llc < %s -verify-machineinstrs -mtriple=arm64-none-linux-gnu -mattr=+neon -fp-contract=fast | FileCheck %s
+
+define float @test_fmul_lane_ss2S(float %a, <2 x float> %v) {
+  ; CHECK-LABEL: test_fmul_lane_ss2S
+  ; CHECK: fmul {{s[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}.s[1]
+  %tmp1 = extractelement <2 x float> %v, i32 1
+  %tmp2 = fmul float %a, %tmp1;
+  ret float %tmp2;
+}
+
+define float @test_fmul_lane_ss2S_swap(float %a, <2 x float> %v) {
+  ; CHECK-LABEL: test_fmul_lane_ss2S_swap
+  ; CHECK: fmul {{s[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}.s[1]
+  %tmp1 = extractelement <2 x float> %v, i32 1
+  %tmp2 = fmul float %tmp1, %a;
+  ret float %tmp2;
+}
+
+
+define float @test_fmul_lane_ss4S(float %a, <4 x float> %v) {
+  ; CHECK-LABEL: test_fmul_lane_ss4S
+  ; CHECK: fmul {{s[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}.s[3]
+  %tmp1 = extractelement <4 x float> %v, i32 3
+  %tmp2 = fmul float %a, %tmp1;
+  ret float %tmp2;
+}
+
+define float @test_fmul_lane_ss4S_swap(float %a, <4 x float> %v) {
+  ; CHECK-LABEL: test_fmul_lane_ss4S_swap
+  ; CHECK: fmul {{s[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}.s[3]
+  %tmp1 = extractelement <4 x float> %v, i32 3
+  %tmp2 = fmul float %tmp1, %a;
+  ret float %tmp2;
+}
+
+
+define double @test_fmul_lane_ddD(double %a, <1 x double> %v) {
+  ; CHECK-LABEL: test_fmul_lane_ddD
+  ; CHECK: fmul {{d[0-9]+}}, {{d[0-9]+}}, {{v[0-9]+.d\[0]|d[0-9]+}}
+  %tmp1 = extractelement <1 x double> %v, i32 0
+  %tmp2 = fmul double %a, %tmp1;
+  ret double %tmp2;
+}
+
+
+
+define double @test_fmul_lane_dd2D(double %a, <2 x double> %v) {
+  ; CHECK-LABEL: test_fmul_lane_dd2D
+  ; CHECK: fmul {{d[0-9]+}}, {{d[0-9]+}}, {{v[0-9]+}}.d[1]
+  %tmp1 = extractelement <2 x double> %v, i32 1
+  %tmp2 = fmul double %a, %tmp1;
+  ret double %tmp2;
+}
+
+
+define double @test_fmul_lane_dd2D_swap(double %a, <2 x double> %v) {
+  ; CHECK-LABEL: test_fmul_lane_dd2D_swap
+  ; CHECK: fmul {{d[0-9]+}}, {{d[0-9]+}}, {{v[0-9]+}}.d[1]
+  %tmp1 = extractelement <2 x double> %v, i32 1
+  %tmp2 = fmul double %tmp1, %a;
+  ret double %tmp2;
+}
+
+declare float @llvm.arm64.neon.fmulx.f32(float, float)
+
+define float @test_fmulx_lane_f32(float %a, <2 x float> %v) {
+  ; CHECK-LABEL: test_fmulx_lane_f32
+  ; CHECK: fmulx {{s[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}.s[1]
+  %tmp1 = extractelement <2 x float> %v, i32 1
+  %tmp2 = call float @llvm.arm64.neon.fmulx.f32(float %a, float %tmp1)
+  ret float %tmp2;
+}
+
+define float @test_fmulx_laneq_f32(float %a, <4 x float> %v) {
+  ; CHECK-LABEL: test_fmulx_laneq_f32
+  ; CHECK: fmulx {{s[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}.s[3]
+  %tmp1 = extractelement <4 x float> %v, i32 3
+  %tmp2 = call float @llvm.arm64.neon.fmulx.f32(float %a, float %tmp1)
+  ret float %tmp2;
+}
+
+define float @test_fmulx_laneq_f32_swap(float %a, <4 x float> %v) {
+  ; CHECK-LABEL: test_fmulx_laneq_f32_swap
+  ; CHECK: fmulx {{s[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}.s[3]
+  %tmp1 = extractelement <4 x float> %v, i32 3
+  %tmp2 = call float @llvm.arm64.neon.fmulx.f32(float %tmp1, float %a)
+  ret float %tmp2;
+}
+
+declare double @llvm.arm64.neon.fmulx.f64(double, double)
+
+define double @test_fmulx_lane_f64(double %a, <1 x double> %v) {
+  ; CHECK-LABEL: test_fmulx_lane_f64
+  ; CHECK: fmulx {{d[0-9]+}}, {{d[0-9]+}}, {{v[0-9]+.d\[0]|d[0-9]+}}
+  %tmp1 = extractelement <1 x double> %v, i32 0
+  %tmp2 = call double @llvm.arm64.neon.fmulx.f64(double %a, double %tmp1)
+  ret double %tmp2;
+}
+
+define double @test_fmulx_laneq_f64_0(double %a, <2 x double> %v) {
+  ; CHECK-LABEL: test_fmulx_laneq_f64_0
+  ; CHECK: fmulx {{d[0-9]+}}, {{d[0-9]+}}, {{v[0-9]+}}.d[0]
+  %tmp1 = extractelement <2 x double> %v, i32 0
+  %tmp2 = call double @llvm.arm64.neon.fmulx.f64(double %a, double %tmp1)
+  ret double %tmp2;
+}
+
+
+define double @test_fmulx_laneq_f64_1(double %a, <2 x double> %v) {
+  ; CHECK-LABEL: test_fmulx_laneq_f64_1
+  ; CHECK: fmulx {{d[0-9]+}}, {{d[0-9]+}}, {{v[0-9]+}}.d[1]
+  %tmp1 = extractelement <2 x double> %v, i32 1
+  %tmp2 = call double @llvm.arm64.neon.fmulx.f64(double %a, double %tmp1)
+  ret double %tmp2;
+}
+
+define double @test_fmulx_laneq_f64_1_swap(double %a, <2 x double> %v) {
+  ; CHECK-LABEL: test_fmulx_laneq_f64_1_swap
+  ; CHECK: fmulx {{d[0-9]+}}, {{d[0-9]+}}, {{v[0-9]+}}.d[1]
+  %tmp1 = extractelement <2 x double> %v, i32 1
+  %tmp2 = call double @llvm.arm64.neon.fmulx.f64(double %tmp1, double %a)
+  ret double %tmp2;
+}
+