Enable lowering ZERO_EXTEND/ANY_EXTEND to PMOVZX from SSE4.1

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@166486 91177308-0d34-0410-b5e6-96231b3b80d8

8 files changed, 16 insertions, 21 deletions

diff --git a/test/CodeGen/X86/2012-01-18-vbitcast.ll b/test/CodeGen/X86/2012-01-18-vbitcast.ll
index 8a3ccc8..3ce7db6 100644
--- a/test/CodeGen/X86/2012-01-18-vbitcast.ll
+++ b/test/CodeGen/X86/2012-01-18-vbitcast.ll
@@ -2,8 +2,8 @@
 
 ;CHECK: vcast
 define <2 x i32> @vcast(<2 x float> %a, <2 x float> %b) {
-;CHECK: pshufd
-;CHECK: pshufd
+;CHECK: pmovzxdq
+;CHECK: pmovzxdq
   %af = bitcast <2 x float> %a to <2 x i32>
   %bf = bitcast <2 x float> %b to <2 x i32>
   %x = sub <2 x i32> %af, %bf
diff --git a/test/CodeGen/X86/2012-03-15-build_vector_wl.ll b/test/CodeGen/X86/2012-03-15-build_vector_wl.ll
index fec17e9..c4b307e 100644
--- a/test/CodeGen/X86/2012-03-15-build_vector_wl.ll
+++ b/test/CodeGen/X86/2012-03-15-build_vector_wl.ll
@@ -4,7 +4,7 @@
 define <4 x i8> @build_vector_again(<16 x i8> %in) nounwind readnone {
 entry:
   %out = shufflevector <16 x i8> %in, <16 x i8> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-; CHECK: shufb
+; CHECK: pmovzxbd
   ret <4 x i8> %out
 ; CHECK: ret
 }
diff --git a/test/CodeGen/X86/2012-07-10-extload64.ll b/test/CodeGen/X86/2012-07-10-extload64.ll
index 906b748..4abdded 100644
--- a/test/CodeGen/X86/2012-07-10-extload64.ll
+++ b/test/CodeGen/X86/2012-07-10-extload64.ll
@@ -3,7 +3,7 @@
 ; CHECK: load_store
 define void @load_store(<4 x i16>* %in) {
 entry:
-; CHECK: movsd
+; CHECK: pmovzxwd
   %A27 = load <4 x i16>* %in, align 4
   %A28 = add <4 x i16> %A27, %A27
 ; CHECK: movlpd
@@ -27,6 +27,6 @@ define <2 x i32> @load_64(<2 x i32>* %ptr) {
 BB:
   %t = load <2 x i32>* %ptr
   ret <2 x i32> %t
-;CHECK: movsd
+;CHECK: pmovzxdq
 ;CHECK: ret
 }
diff --git a/test/CodeGen/X86/pointer-vector.ll b/test/CodeGen/X86/pointer-vector.ll
index 800fbed..58423d1 100644
--- a/test/CodeGen/X86/pointer-vector.ll
+++ b/test/CodeGen/X86/pointer-vector.ll
@@ -81,8 +81,7 @@ define <4 x i32*> @INT2PTR1(<4 x i8>* %p) nounwind {
 entry:
   %G = load <4 x i8>* %p
 ;CHECK: movl
-;CHECK: movd
-;CHECK: pshufb
+;CHECK: pmovzxbd
 ;CHECK: pand
   %K = inttoptr <4 x i8> %G to <4 x i32*>
 ;CHECK: ret
@@ -105,7 +104,7 @@ define <2 x i32*> @BITCAST1(<2 x i8*>* %p) nounwind {
 entry:
   %G = load <2 x i8*>* %p
 ;CHECK: movl
-;CHECK: movsd
+;CHECK: pmovzxdq
   %T = bitcast <2 x i8*> %G to <2 x i32*>
 ;CHECK: ret
   ret <2 x i32*> %T
diff --git a/test/CodeGen/X86/promote.ll b/test/CodeGen/X86/promote.ll
index 8b30dc7..283f48c 100644
--- a/test/CodeGen/X86/promote.ll
+++ b/test/CodeGen/X86/promote.ll
@@ -20,7 +20,7 @@ entry:
 ; CHECK: shuff_f
 define i32 @shuff_f(<4 x i8>* %A) {
 entry:
-; CHECK: pshufb
+; CHECK: pmovzxbd
 ; CHECK: paddd
 ; CHECK: pshufb
   %0 = load <4 x i8>* %A, align 8
diff --git a/test/CodeGen/X86/trunc-ext-ld-st.ll b/test/CodeGen/X86/trunc-ext-ld-st.ll
index 9877d7b..1d22a18 100644
--- a/test/CodeGen/X86/trunc-ext-ld-st.ll
+++ b/test/CodeGen/X86/trunc-ext-ld-st.ll
@@ -2,8 +2,7 @@
 
 ;CHECK: load_2_i8
 ; A single 16-bit load
-;CHECK: movzwl
-;CHECK: pshufb
+;CHECK: pmovzxbq
 ;CHECK: paddq
 ;CHECK: pshufb
 ; A single 16-bit store
@@ -19,8 +18,7 @@ define void @load_2_i8(<2 x i8>* %A)  {
 
 ;CHECK: load_2_i16
 ; Read 32-bits
-;CHECK: movd
-;CHECK: pshufb
+;CHECK: pmovzxwq
 ;CHECK: paddq
 ;CHECK: pshufb
 ;CHECK: movd
@@ -33,7 +31,7 @@ define void @load_2_i16(<2 x i16>* %A)  {
 } 
 
 ;CHECK: load_2_i32
-;CHECK: pshufd
+;CHECK: pmovzxdq
 ;CHECK: paddq
 ;CHECK: pshufd
 ;CHECK: ret
@@ -45,8 +43,7 @@ define void @load_2_i32(<2 x i32>* %A)  {
 } 
 
 ;CHECK: load_4_i8
-;CHECK: movd
-;CHECK: pshufb
+;CHECK: pmovzxbd
 ;CHECK: paddd
 ;CHECK: pshufb
 ;CHECK: ret
@@ -58,7 +55,7 @@ define void @load_4_i8(<4 x i8>* %A)  {
 } 
 
 ;CHECK: load_4_i16
-;CHECK: punpcklwd
+;CHECK: pmovzxwd
 ;CHECK: paddd
 ;CHECK: pshufb
 ;CHECK: ret
@@ -70,7 +67,7 @@ define void @load_4_i16(<4 x i16>* %A)  {
 } 
 
 ;CHECK: load_8_i8
-;CHECK: punpcklbw
+;CHECK: pmovzxbw
 ;CHECK: paddw
 ;CHECK: pshufb
 ;CHECK: ret
diff --git a/test/CodeGen/X86/vec_compare-2.ll b/test/CodeGen/X86/vec_compare-2.ll
index 46d6a23..4da7953 100644
--- a/test/CodeGen/X86/vec_compare-2.ll
+++ b/test/CodeGen/X86/vec_compare-2.ll
@@ -10,8 +10,7 @@ define void @blackDespeckle_wrapper(i8** %args_list, i64* %gtid, i64 %xend) {
 entry:
 ; CHECK: cfi_def_cfa_offset
 ; CHECK-NOT: set
-; CHECK: punpcklwd
-; CHECK: pshufd
+; CHECK: pmovzxwq
 ; CHECK: pshufb
   %shr.i = ashr <4 x i32> zeroinitializer, <i32 3, i32 3, i32 3, i32 3> ; <<4 x i32>> [#uses=1]
   %cmp318.i = sext <4 x i1> zeroinitializer to <4 x i32> ; <<4 x i32>> [#uses=1]
diff --git a/test/CodeGen/X86/widen_load-2.ll b/test/CodeGen/X86/widen_load-2.ll
index 79aa000..224898c 100644
--- a/test/CodeGen/X86/widen_load-2.ll
+++ b/test/CodeGen/X86/widen_load-2.ll
@@ -170,7 +170,7 @@ define void @add31i8(%i8vec31* nocapture sret %ret, %i8vec31* %ap, %i8vec31* %bp
 ; CHECK: rot
 %i8vec3pack = type { <3 x i8>, i8 }
 define %i8vec3pack  @rot() nounwind {
-; CHECK: movd {{-?[0-9]+}}(%rsp), {{%xmm[0-9]}}
+; CHECK: pmovzxbd {{-?[0-9]+}}(%rsp), {{%xmm[0-9]}}
 entry:
   %X = alloca %i8vec3pack, align 4
   %rot = alloca %i8vec3pack, align 4