[ARM] Fix MachO CPU Subtype selection

[oota-llvm.git] / test / CodeGen / ARM / vpadd.ll

diff --git a/test/CodeGen/ARM/vpadd.ll b/test/CodeGen/ARM/vpadd.ll
index f84721f996cd2b040b97da1041060c33b7274c97..e362ce36f8ba1dbd4879fc0d39c94aa135efa902 100644 (file)
--- a/test/CodeGen/ARM/vpadd.ll
+++ b/test/CodeGen/ARM/vpadd.ll
@@ -1,10 +1,10 @@
-; RUN: llc < %s -march=arm -mattr=+neon | FileCheck %s
+; RUN: llc -mtriple=arm-eabi -mattr=+neon %s -o - | FileCheck %s
 
 define <8 x i8> @vpaddi8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
 ;CHECK-LABEL: vpaddi8:
 ;CHECK: vpadd.i8
-       %tmp1 = load <8 x i8>* %A
-       %tmp2 = load <8 x i8>* %B
+       %tmp1 = load <8 x i8>, <8 x i8>* %A
+       %tmp2 = load <8 x i8>, <8 x i8>* %B
        %tmp3 = call <8 x i8> @llvm.arm.neon.vpadd.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2)
        ret <8 x i8> %tmp3
 }
@@ -12,8 +12,8 @@ define <8 x i8> @vpaddi8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
 define <4 x i16> @vpaddi16(<4 x i16>* %A, <4 x i16>* %B) nounwind {
 ;CHECK-LABEL: vpaddi16:
 ;CHECK: vpadd.i16
-       %tmp1 = load <4 x i16>* %A
-       %tmp2 = load <4 x i16>* %B
+       %tmp1 = load <4 x i16>, <4 x i16>* %A
+       %tmp2 = load <4 x i16>, <4 x i16>* %B
        %tmp3 = call <4 x i16> @llvm.arm.neon.vpadd.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
        ret <4 x i16> %tmp3
 }
@@ -21,8 +21,8 @@ define <4 x i16> @vpaddi16(<4 x i16>* %A, <4 x i16>* %B) nounwind {
 define <2 x i32> @vpaddi32(<2 x i32>* %A, <2 x i32>* %B) nounwind {
 ;CHECK-LABEL: vpaddi32:
 ;CHECK: vpadd.i32
-       %tmp1 = load <2 x i32>* %A
-       %tmp2 = load <2 x i32>* %B
+       %tmp1 = load <2 x i32>, <2 x i32>* %A
+       %tmp2 = load <2 x i32>, <2 x i32>* %B
        %tmp3 = call <2 x i32> @llvm.arm.neon.vpadd.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
        ret <2 x i32> %tmp3
 }
@@ -30,8 +30,8 @@ define <2 x i32> @vpaddi32(<2 x i32>* %A, <2 x i32>* %B) nounwind {
 define <2 x float> @vpaddf32(<2 x float>* %A, <2 x float>* %B) nounwind {
 ;CHECK-LABEL: vpaddf32:
 ;CHECK: vpadd.f32
-       %tmp1 = load <2 x float>* %A
-       %tmp2 = load <2 x float>* %B
+       %tmp1 = load <2 x float>, <2 x float>* %A
+       %tmp2 = load <2 x float>, <2 x float>* %B
        %tmp3 = call <2 x float> @llvm.arm.neon.vpadd.v2f32(<2 x float> %tmp1, <2 x float> %tmp2)
        ret <2 x float> %tmp3
 }
@@ -45,7 +45,7 @@ declare <2 x float> @llvm.arm.neon.vpadd.v2f32(<2 x float>, <2 x float>) nounwin
 define <4 x i16> @vpaddls8(<8 x i8>* %A) nounwind {
 ;CHECK-LABEL: vpaddls8:
 ;CHECK: vpaddl.s8
-       %tmp1 = load <8 x i8>* %A
+       %tmp1 = load <8 x i8>, <8 x i8>* %A
        %tmp2 = call <4 x i16> @llvm.arm.neon.vpaddls.v4i16.v8i8(<8 x i8> %tmp1)
        ret <4 x i16> %tmp2
 }
@@ -53,7 +53,7 @@ define <4 x i16> @vpaddls8(<8 x i8>* %A) nounwind {
 define <2 x i32> @vpaddls16(<4 x i16>* %A) nounwind {
 ;CHECK-LABEL: vpaddls16:
 ;CHECK: vpaddl.s16
-       %tmp1 = load <4 x i16>* %A
+       %tmp1 = load <4 x i16>, <4 x i16>* %A
        %tmp2 = call <2 x i32> @llvm.arm.neon.vpaddls.v2i32.v4i16(<4 x i16> %tmp1)
        ret <2 x i32> %tmp2
 }
@@ -61,7 +61,7 @@ define <2 x i32> @vpaddls16(<4 x i16>* %A) nounwind {
 define <1 x i64> @vpaddls32(<2 x i32>* %A) nounwind {
 ;CHECK-LABEL: vpaddls32:
 ;CHECK: vpaddl.s32
-       %tmp1 = load <2 x i32>* %A
+       %tmp1 = load <2 x i32>, <2 x i32>* %A
        %tmp2 = call <1 x i64> @llvm.arm.neon.vpaddls.v1i64.v2i32(<2 x i32> %tmp1)
        ret <1 x i64> %tmp2
 }
@@ -69,7 +69,7 @@ define <1 x i64> @vpaddls32(<2 x i32>* %A) nounwind {
 define <4 x i16> @vpaddlu8(<8 x i8>* %A) nounwind {
 ;CHECK-LABEL: vpaddlu8:
 ;CHECK: vpaddl.u8
-       %tmp1 = load <8 x i8>* %A
+       %tmp1 = load <8 x i8>, <8 x i8>* %A
        %tmp2 = call <4 x i16> @llvm.arm.neon.vpaddlu.v4i16.v8i8(<8 x i8> %tmp1)
        ret <4 x i16> %tmp2
 }
@@ -77,7 +77,7 @@ define <4 x i16> @vpaddlu8(<8 x i8>* %A) nounwind {
 define <2 x i32> @vpaddlu16(<4 x i16>* %A) nounwind {
 ;CHECK-LABEL: vpaddlu16:
 ;CHECK: vpaddl.u16
-       %tmp1 = load <4 x i16>* %A
+       %tmp1 = load <4 x i16>, <4 x i16>* %A
        %tmp2 = call <2 x i32> @llvm.arm.neon.vpaddlu.v2i32.v4i16(<4 x i16> %tmp1)
        ret <2 x i32> %tmp2
 }
@@ -85,7 +85,7 @@ define <2 x i32> @vpaddlu16(<4 x i16>* %A) nounwind {
 define <1 x i64> @vpaddlu32(<2 x i32>* %A) nounwind {
 ;CHECK-LABEL: vpaddlu32:
 ;CHECK: vpaddl.u32
-       %tmp1 = load <2 x i32>* %A
+       %tmp1 = load <2 x i32>, <2 x i32>* %A
        %tmp2 = call <1 x i64> @llvm.arm.neon.vpaddlu.v1i64.v2i32(<2 x i32> %tmp1)
        ret <1 x i64> %tmp2
 }
@@ -93,7 +93,7 @@ define <1 x i64> @vpaddlu32(<2 x i32>* %A) nounwind {
 define <8 x i16> @vpaddlQs8(<16 x i8>* %A) nounwind {
 ;CHECK-LABEL: vpaddlQs8:
 ;CHECK: vpaddl.s8
-       %tmp1 = load <16 x i8>* %A
+       %tmp1 = load <16 x i8>, <16 x i8>* %A
        %tmp2 = call <8 x i16> @llvm.arm.neon.vpaddls.v8i16.v16i8(<16 x i8> %tmp1)
        ret <8 x i16> %tmp2
 }
@@ -101,7 +101,7 @@ define <8 x i16> @vpaddlQs8(<16 x i8>* %A) nounwind {
 define <4 x i32> @vpaddlQs16(<8 x i16>* %A) nounwind {
 ;CHECK-LABEL: vpaddlQs16:
 ;CHECK: vpaddl.s16
-       %tmp1 = load <8 x i16>* %A
+       %tmp1 = load <8 x i16>, <8 x i16>* %A
        %tmp2 = call <4 x i32> @llvm.arm.neon.vpaddls.v4i32.v8i16(<8 x i16> %tmp1)
        ret <4 x i32> %tmp2
 }
@@ -109,7 +109,7 @@ define <4 x i32> @vpaddlQs16(<8 x i16>* %A) nounwind {
 define <2 x i64> @vpaddlQs32(<4 x i32>* %A) nounwind {
 ;CHECK-LABEL: vpaddlQs32:
 ;CHECK: vpaddl.s32
-       %tmp1 = load <4 x i32>* %A
+       %tmp1 = load <4 x i32>, <4 x i32>* %A
        %tmp2 = call <2 x i64> @llvm.arm.neon.vpaddls.v2i64.v4i32(<4 x i32> %tmp1)
        ret <2 x i64> %tmp2
 }
@@ -117,7 +117,7 @@ define <2 x i64> @vpaddlQs32(<4 x i32>* %A) nounwind {
 define <8 x i16> @vpaddlQu8(<16 x i8>* %A) nounwind {
 ;CHECK-LABEL: vpaddlQu8:
 ;CHECK: vpaddl.u8
-       %tmp1 = load <16 x i8>* %A
+       %tmp1 = load <16 x i8>, <16 x i8>* %A
        %tmp2 = call <8 x i16> @llvm.arm.neon.vpaddlu.v8i16.v16i8(<16 x i8> %tmp1)
        ret <8 x i16> %tmp2
 }
@@ -125,7 +125,7 @@ define <8 x i16> @vpaddlQu8(<16 x i8>* %A) nounwind {
 define <4 x i32> @vpaddlQu16(<8 x i16>* %A) nounwind {
 ;CHECK-LABEL: vpaddlQu16:
 ;CHECK: vpaddl.u16
-       %tmp1 = load <8 x i16>* %A
+       %tmp1 = load <8 x i16>, <8 x i16>* %A
        %tmp2 = call <4 x i32> @llvm.arm.neon.vpaddlu.v4i32.v8i16(<8 x i16> %tmp1)
        ret <4 x i32> %tmp2
 }
@@ -133,7 +133,7 @@ define <4 x i32> @vpaddlQu16(<8 x i16>* %A) nounwind {
 define <2 x i64> @vpaddlQu32(<4 x i32>* %A) nounwind {
 ;CHECK-LABEL: vpaddlQu32:
 ;CHECK: vpaddl.u32
-       %tmp1 = load <4 x i32>* %A
+       %tmp1 = load <4 x i32>, <4 x i32>* %A
        %tmp2 = call <2 x i64> @llvm.arm.neon.vpaddlu.v2i64.v4i32(<4 x i32> %tmp1)
        ret <2 x i64> %tmp2
 }
@@ -143,15 +143,26 @@ define void @addCombineToVPADDL() nounwind ssp {
 ; CHECK: vpaddl.s8
   %cbcr = alloca <16 x i8>, align 16
   %X = alloca <8 x i8>, align 8
-  %tmp = load <16 x i8>* %cbcr
+  %tmp = load <16 x i8>, <16 x i8>* %cbcr
   %tmp1 = shufflevector <16 x i8> %tmp, <16 x i8> undef, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>
-  %tmp2 = load <16 x i8>* %cbcr
+  %tmp2 = load <16 x i8>, <16 x i8>* %cbcr
   %tmp3 = shufflevector <16 x i8> %tmp2, <16 x i8> undef, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
   %add = add <8 x i8> %tmp3, %tmp1
   store <8 x i8> %add, <8 x i8>* %X, align 8
   ret void
 }
 
+; Legalization produces a EXTRACT_VECTOR_ELT DAG node which performs an extend from
+; i16 to i32. In this case the input for the formed VPADDL needs to be a vector of i16s.
+define <2 x i16> @fromExtendingExtractVectorElt(<4 x i16> %in) {
+;CHECK-LABEL: fromExtendingExtractVectorElt:
+;CHECK: vpaddl.s16
+  %tmp1 = shufflevector <4 x i16> %in, <4 x i16> undef, <2 x i32> <i32 0, i32 2>
+  %tmp2 = shufflevector <4 x i16> %in, <4 x i16> undef, <2 x i32> <i32 1, i32 3>
+  %x = add <2 x i16> %tmp2, %tmp1
+  ret <2 x i16> %x
+}
+
 declare <4 x i16> @llvm.arm.neon.vpaddls.v4i16.v8i8(<8 x i8>) nounwind readnone
 declare <2 x i32> @llvm.arm.neon.vpaddls.v2i32.v4i16(<4 x i16>) nounwind readnone
 declare <1 x i64> @llvm.arm.neon.vpaddls.v1i64.v2i32(<2 x i32>) nounwind readnone