updated patch for the ARM fused multiply add/sub

In this update:
- I assumed neon2 does not imply vfpv4, but neon and vfpv4 imply neon2.
- I kept setting .fpu=neon-vfpv4 code attribute because that is what the
assembler understands.

Patch by Ana Pazos <apazos@codeaurora.org>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@152036 91177308-0d34-0410-b5e6-96231b3b80d8

1 files changed, 41 insertions, 9 deletions

diff --git a/test/CodeGen/ARM/fusedMAC.ll b/test/CodeGen/ARM/fusedMAC.ll
index d7b8ed1..40e8bb2 100644
--- a/test/CodeGen/ARM/fusedMAC.ll
+++ b/test/CodeGen/ARM/fusedMAC.ll
@@ -1,7 +1,7 @@
-; RUN: llc < %s -march=arm -mattr=+neon-vfpv4 | FileCheck %s
+; RUN: llc < %s -march=arm -mattr=+neon,+vfp4 | FileCheck %s
 ; Check generated fused MAC and MLS.
 
-define double @fusedMACTest1(double %d1, double %d2, double %d3) nounwind readnone noinline {
+define double @fusedMACTest1(double %d1, double %d2, double %d3) {
 ;CHECK: fusedMACTest1:
 ;CHECK: vfma.f64
   %1 = fmul double %d1, %d2
@@ -9,7 +9,7 @@ define double @fusedMACTest1(double %d1, double %d2, double %d3) nounwind readno
   ret double %2
 }
 
-define float @fusedMACTest2(float %f1, float %f2, float %f3) nounwind readnone noinline {
+define float @fusedMACTest2(float %f1, float %f2, float %f3) {
 ;CHECK: fusedMACTest2:
 ;CHECK: vfma.f32
   %1 = fmul float %f1, %f2
@@ -17,7 +17,7 @@ define float @fusedMACTest2(float %f1, float %f2, float %f3) nounwind readnone n
   ret float %2
 }
 
-define double @fusedMACTest3(double %d1, double %d2, double %d3) nounwind readnone noinline {
+define double @fusedMACTest3(double %d1, double %d2, double %d3) {
 ;CHECK: fusedMACTest3:
 ;CHECK: vfms.f64
   %1 = fmul double %d2, %d3
@@ -25,7 +25,7 @@ define double @fusedMACTest3(double %d1, double %d2, double %d3) nounwind readno
   ret double %2
 }
 
-define float @fusedMACTest4(float %f1, float %f2, float %f3) nounwind readnone noinline {
+define float @fusedMACTest4(float %f1, float %f2, float %f3) {
 ;CHECK: fusedMACTest4:
 ;CHECK: vfms.f32
   %1 = fmul float %f2, %f3
@@ -33,7 +33,7 @@ define float @fusedMACTest4(float %f1, float %f2, float %f3) nounwind readnone n
   ret float %2
 }
 
-define double @fusedMACTest5(double %d1, double %d2, double %d3) nounwind readnone noinline {
+define double @fusedMACTest5(double %d1, double %d2, double %d3) {
 ;CHECK: fusedMACTest5:
 ;CHECK: vfnma.f64
   %1 = fmul double %d1, %d2
@@ -42,7 +42,7 @@ define double @fusedMACTest5(double %d1, double %d2, double %d3) nounwind readno
   ret double %3
 }
 
-define float @fusedMACTest6(float %f1, float %f2, float %f3) nounwind {
+define float @fusedMACTest6(float %f1, float %f2, float %f3) {
 ;CHECK: fusedMACTest6:
 ;CHECK: vfnma.f32
   %1 = fmul float %f1, %f2
@@ -51,7 +51,7 @@ define float @fusedMACTest6(float %f1, float %f2, float %f3) nounwind {
   ret float %3
 }
 
-define double @fusedMACTest7(double %d1, double %d2, double %d3) nounwind {
+define double @fusedMACTest7(double %d1, double %d2, double %d3) {
 ;CHECK: fusedMACTest7:
 ;CHECK: vfnms.f64
   %1 = fmul double %d1, %d2
@@ -59,10 +59,42 @@ define double @fusedMACTest7(double %d1, double %d2, double %d3) nounwind {
   ret double %2
 }
 
-define float @fusedMACTest8(float %f1, float %f2, float %f3) nounwind {
+define float @fusedMACTest8(float %f1, float %f2, float %f3) {
 ;CHECK: fusedMACTest8:
 ;CHECK: vfnms.f32
   %1 = fmul float %f1, %f2
   %2 = fsub float %1, %f3
   ret float %2
 }
+
+define <2 x float> @fusedMACTest9(<2 x float> %a, <2 x float> %b) {
+;CHECK: fusedMACTest9:
+;CHECK: vfma.f32
+  %mul = fmul <2 x float> %a, %b
+  %add = fadd <2 x float> %mul, %a
+  ret <2 x float> %add
+}
+
+define <2 x float> @fusedMACTest10(<2 x float> %a, <2 x float> %b) {
+;CHECK: fusedMACTest10:
+;CHECK: vfms.f32
+  %mul = fmul <2 x float> %a, %b
+  %sub = fsub <2 x float> %a, %mul
+  ret <2 x float> %sub
+}
+
+define <4 x float> @fusedMACTest11(<4 x float> %a, <4 x float> %b) {
+;CHECK: fusedMACTest11:
+;CHECK: vfma.f32
+  %mul = fmul <4 x float> %a, %b
+  %add = fadd <4 x float> %mul, %a
+  ret <4 x float> %add
+}
+
+define <4 x float> @fusedMACTest12(<4 x float> %a, <4 x float> %b) {
+;CHECK: fusedMACTest12:
+;CHECK: vfms.f32
+  %mul = fmul <4 x float> %a, %b
+  %sub = fsub <4 x float> %a, %mul
+  ret <4 x float> %sub
+}