7 files changed, 251 insertions, 6 deletions
diff --git a/test/Transforms/ScalarRepl/2008-01-29-PromoteBug.ll b/test/Transforms/ScalarRepl/2008-01-29-PromoteBug.ll
index d799bd77e458..8bc4ff0b3ffc 100644
--- a/test/Transforms/ScalarRepl/2008-01-29-PromoteBug.ll
+++ b/test/Transforms/ScalarRepl/2008-01-29-PromoteBug.ll
@@ -1,6 +1,6 @@
 ; RUN: opt < %s -scalarrepl -instcombine -S | grep {ret i8 17}
 ; rdar://5707076
-target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128"
+target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128-n8:16:32"
 target triple = "i386-apple-darwin9.1.0"
 	%struct.T = type <{ i8, [3 x i8] }>
 
diff --git a/test/Transforms/ScalarRepl/2008-06-05-loadstore-agg.ll b/test/Transforms/ScalarRepl/2008-06-05-loadstore-agg.ll
index 87a08b7eaaf2..ce70a1b13b81 100644
--- a/test/Transforms/ScalarRepl/2008-06-05-loadstore-agg.ll
+++ b/test/Transforms/ScalarRepl/2008-06-05-loadstore-agg.ll
@@ -13,7 +13,7 @@ define i32 @foo() {
 	%res2 = insertvalue { i32, i32 } %res1, i32 2, 1		; <{ i32, i32 }> [#uses=1]
         ; And store it
 	store { i32, i32 } %res2, { i32, i32 }* %target
-        ; Actually use %target, so it doesn't get removed alltogether
+        ; Actually use %target, so it doesn't get removed altogether
         %ptr = getelementptr { i32, i32 }* %target, i32 0, i32 0
         %val = load i32* %ptr
 	ret i32 %val
@@ -26,7 +26,7 @@ define i32 @bar() {
 	%res2 = insertvalue [ 2 x i32 ] %res1, i32 2, 1		; <{ i32, i32 }> [#uses=1]
         ; And store it
 	store [ 2 x i32 ] %res2, [ 2 x i32 ]* %target
-        ; Actually use %target, so it doesn't get removed alltogether
+        ; Actually use %target, so it doesn't get removed altogether
         %ptr = getelementptr [ 2 x i32 ]* %target, i32 0, i32 0
         %val = load i32* %ptr
 	ret i32 %val
diff --git a/test/Transforms/ScalarRepl/dg.exp b/test/Transforms/ScalarRepl/dg.exp
index f2005891a59a..39954d8a498d 100644
--- a/test/Transforms/ScalarRepl/dg.exp
+++ b/test/Transforms/ScalarRepl/dg.exp
@@ -1,3 +1,3 @@
 load_lib llvm.exp
 
-RunLLVMTests [lsort [glob -nocomplain $srcdir/$subdir/*.{ll,c,cpp}]]
+RunLLVMTests [lsort [glob -nocomplain $srcdir/$subdir/*.{ll}]]
diff --git a/test/Transforms/ScalarRepl/inline-vector.ll b/test/Transforms/ScalarRepl/inline-vector.ll
new file mode 100644
index 000000000000..2f51cc7cf59c
--- /dev/null
+++ b/test/Transforms/ScalarRepl/inline-vector.ll
@@ -0,0 +1,53 @@
+; RUN: opt < %s -scalarrepl -S | FileCheck %s
+; RUN: opt < %s -scalarrepl-ssa -S | FileCheck %s
+target datalayout = "e-p:32:32:32-i1:8:32-i8:8:32-i16:16:32-i32:32:32-i64:32:32-f32:32:32-f64:32:32-v64:64:64-v128:128:128-a0:0:32-n32"
+target triple = "thumbv7-apple-darwin10.0.0"
+
+%struct.Vector4 = type { float, float, float, float }
+@f.vector = internal constant %struct.Vector4 { float 1.000000e+00, float 2.000000e+00, float 3.000000e+00, float 4.000000e+00 }, align 16
+
+; CHECK: define void @f
+; CHECK-NOT: alloca
+; CHECK: phi <4 x float>
+
+define void @f() nounwind ssp {
+entry:
+  %i = alloca i32, align 4
+  %vector = alloca %struct.Vector4, align 16
+  %agg.tmp = alloca %struct.Vector4, align 16
+  %tmp = bitcast %struct.Vector4* %vector to i8*
+  call void @llvm.memcpy.p0i8.p0i8.i32(i8* %tmp, i8* bitcast (%struct.Vector4* @f.vector to i8*), i32 16, i32 16, i1 false)
+  br label %for.cond
+
+for.cond:                                         ; preds = %for.body, %entry
+  %storemerge = phi i32 [ 0, %entry ], [ %inc, %for.body ]
+  store i32 %storemerge, i32* %i, align 4
+  %cmp = icmp slt i32 %storemerge, 1000000
+  br i1 %cmp, label %for.body, label %for.end
+
+for.body:                                         ; preds = %for.cond
+  %tmp2 = bitcast %struct.Vector4* %agg.tmp to i8*
+  %tmp3 = bitcast %struct.Vector4* %vector to i8*
+  call void @llvm.memcpy.p0i8.p0i8.i32(i8* %tmp2, i8* %tmp3, i32 16, i32 16, i1 false)
+  %0 = bitcast %struct.Vector4* %agg.tmp to [2 x i64]*
+  %1 = load [2 x i64]* %0, align 16
+  %tmp2.i = extractvalue [2 x i64] %1, 0
+  %tmp3.i = zext i64 %tmp2.i to i128
+  %tmp10.i = bitcast i128 %tmp3.i to <4 x float>
+  %sub.i.i = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %tmp10.i
+  %2 = bitcast %struct.Vector4* %vector to <4 x float>*
+  store <4 x float> %sub.i.i, <4 x float>* %2, align 16
+  %tmp4 = load i32* %i, align 4
+  %inc = add nsw i32 %tmp4, 1
+  br label %for.cond
+
+for.end:                                          ; preds = %for.cond
+  %x = getelementptr inbounds %struct.Vector4* %vector, i32 0, i32 0
+  %tmp5 = load float* %x, align 16
+  %conv = fpext float %tmp5 to double
+  %call = call i32 (...)* @printf(double %conv) nounwind
+  ret void
+}
+
+declare void @llvm.memcpy.p0i8.p0i8.i32(i8* nocapture, i8* nocapture, i32, i32, i1) nounwind
+declare i32 @printf(...)
diff --git a/test/Transforms/ScalarRepl/only-memcpy-uses.ll b/test/Transforms/ScalarRepl/only-memcpy-uses.ll
new file mode 100644
index 000000000000..cfb88bd80d60
--- /dev/null
+++ b/test/Transforms/ScalarRepl/only-memcpy-uses.ll
@@ -0,0 +1,27 @@
+; RUN: opt < %s -scalarrepl -S | FileCheck %s
+target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64"
+target triple = "x86_64-apple-darwin10.0.0"
+
+%struct.S = type { [12 x i32] }
+
+; CHECK: @bar4
+define void @bar4(%struct.S* byval %s) nounwind ssp {
+entry:
+; CHECK: alloca
+; CHECK-NOT: load
+; CHECK: memcpy
+  %t = alloca %struct.S, align 4
+  %agg.tmp = alloca %struct.S, align 4
+  %tmp = bitcast %struct.S* %t to i8*
+  %tmp1 = bitcast %struct.S* %s to i8*
+  call void @llvm.memcpy.p0i8.p0i8.i64(i8* %tmp, i8* %tmp1, i64 48, i32 4, i1 false)
+  %tmp2 = bitcast %struct.S* %agg.tmp to i8*
+  %tmp3 = bitcast %struct.S* %t to i8*
+  call void @llvm.memcpy.p0i8.p0i8.i64(i8* %tmp2, i8* %tmp3, i64 48, i32 4, i1 false)
+  %call = call i32 (...)* @bazz(%struct.S* byval %agg.tmp)
+  ret void
+}
+
+declare void @llvm.memcpy.p0i8.p0i8.i64(i8* nocapture, i8* nocapture, i64, i32, i1) nounwind
+
+declare i32 @bazz(...)
diff --git a/test/Transforms/ScalarRepl/union-pointer.ll b/test/Transforms/ScalarRepl/union-pointer.ll
index fe702fa21772..ea4ec14e5621 100644
--- a/test/Transforms/ScalarRepl/union-pointer.ll
+++ b/test/Transforms/ScalarRepl/union-pointer.ll
@@ -3,7 +3,7 @@
 ; RUN:   not grep alloca
 ; RUN: opt < %s -scalarrepl -S | grep {ret i8}
 
-target datalayout = "e-p:32:32"
+target datalayout = "e-p:32:32-n8:16:32"
 target triple = "i686-apple-darwin8.7.2"
 	%struct.Val = type { i32*, i32 }
 
diff --git a/test/Transforms/ScalarRepl/vector_promote.ll b/test/Transforms/ScalarRepl/vector_promote.ll
index 37cb49f539d6..c51ef109c0b2 100644
--- a/test/Transforms/ScalarRepl/vector_promote.ll
+++ b/test/Transforms/ScalarRepl/vector_promote.ll
@@ -94,7 +94,172 @@ define i64 @test6(<2 x float> %X) {
 	%tmp = load i64* %P
 	ret i64 %tmp
 ; CHECK: @test6
-; CHECK: bitcast <2 x float> %X to <1 x i64>
+; CHECK: bitcast <2 x float> %X to i64
 ; CHECK: ret i64
 }
 
+define float @test7(<4 x float> %x) {
+	%a = alloca <4 x float>
+	store <4 x float> %x, <4 x float>* %a
+	%p = bitcast <4 x float>* %a to <2 x float>*
+	%b = load <2 x float>* %p
+	%q = getelementptr <4 x float>* %a, i32 0, i32 2
+	%c = load float* %q
+	ret float %c
+; CHECK: @test7
+; CHECK-NOT: alloca
+; CHECK: bitcast <4 x float> %x to <2 x double>
+; CHECK-NEXT: extractelement <2 x double>
+; CHECK-NEXT: bitcast double %tmp4 to <2 x float>
+; CHECK-NEXT: extractelement <4 x float>
+}
+
+define void @test8(<4 x float> %x, <2 x float> %y) {
+	%a = alloca <4 x float>
+	store <4 x float> %x, <4 x float>* %a
+	%p = bitcast <4 x float>* %a to <2 x float>*
+	store <2 x float> %y, <2 x float>* %p
+	ret void
+; CHECK: @test8
+; CHECK-NOT: alloca
+; CHECK: bitcast <4 x float> %x to <2 x double>
+; CHECK-NEXT: bitcast <2 x float> %y to double
+; CHECK-NEXT: insertelement <2 x double>
+; CHECK-NEXT: bitcast <2 x double> %tmp2 to <4 x float>
+}
+
+define i256 @test9(<4 x i256> %x) {
+	%a = alloca <4 x i256>
+	store <4 x i256> %x, <4 x i256>* %a
+	%p = bitcast <4 x i256>* %a to <2 x i256>*
+	%b = load <2 x i256>* %p
+	%q = getelementptr <4 x i256>* %a, i32 0, i32 2
+	%c = load i256* %q
+	ret i256 %c
+; CHECK: @test9
+; CHECK-NOT: alloca
+; CHECK: bitcast <4 x i256> %x to <2 x i512>
+; CHECK-NEXT: extractelement <2 x i512>
+; CHECK-NEXT: bitcast i512 %tmp4 to <2 x i256>
+; CHECK-NEXT: extractelement <4 x i256>
+}
+
+define void @test10(<4 x i256> %x, <2 x i256> %y) {
+	%a = alloca <4 x i256>
+	store <4 x i256> %x, <4 x i256>* %a
+	%p = bitcast <4 x i256>* %a to <2 x i256>*
+	store <2 x i256> %y, <2 x i256>* %p
+	ret void
+; CHECK: @test10
+; CHECK-NOT: alloca
+; CHECK: bitcast <4 x i256> %x to <2 x i512>
+; CHECK-NEXT: bitcast <2 x i256> %y to i512
+; CHECK-NEXT: insertelement <2 x i512>
+; CHECK-NEXT: bitcast <2 x i512> %tmp2 to <4 x i256>
+}
+
+%union.v = type { <2 x i64> }
+
+define void @test11(<2 x i64> %x) {
+  %a = alloca %union.v
+  %p = getelementptr inbounds %union.v* %a, i32 0, i32 0
+  store <2 x i64> %x, <2 x i64>* %p, align 16
+  %q = getelementptr inbounds %union.v* %a, i32 0, i32 0
+  %r = bitcast <2 x i64>* %q to <4 x float>*
+  %b = load <4 x float>* %r, align 16
+  ret void
+; CHECK: @test11
+; CHECK-NOT: alloca
+}
+
+define void @test12() {
+entry:
+  %a = alloca <64 x i8>, align 64
+  store <64 x i8> undef, <64 x i8>* %a, align 64
+  %p = bitcast <64 x i8>* %a to <16 x i8>*
+  %0 = load <16 x i8>* %p, align 64
+  store <16 x i8> undef, <16 x i8>* %p, align 64
+  %q = bitcast <16 x i8>* %p to <64 x i8>*
+  %1 = load <64 x i8>* %q, align 64
+  ret void
+; CHECK: @test12
+; CHECK-NOT: alloca
+; CHECK: extractelement <4 x i128>
+; CHECK: insertelement <4 x i128>
+}
+
+define float @test13(<4 x float> %x, <2 x i32> %y) {
+	%a = alloca <4 x float>
+	store <4 x float> %x, <4 x float>* %a
+	%p = bitcast <4 x float>* %a to <2 x float>*
+	%b = load <2 x float>* %p
+	%q = getelementptr <4 x float>* %a, i32 0, i32 2
+	%c = load float* %q
+	%r = bitcast <4 x float>* %a to <2 x i32>*
+	store <2 x i32> %y, <2 x i32>* %r
+	ret float %c
+; CHECK: @test13
+; CHECK-NOT: alloca
+; CHECK: bitcast <4 x float> %x to i128
+}
+
+define <3 x float> @test14(<3 x float> %x)  {
+entry:
+  %x.addr = alloca <3 x float>, align 16
+  %r = alloca <3 x i32>, align 16
+  %extractVec = shufflevector <3 x float> %x, <3 x float> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 undef>
+  %storetmp = bitcast <3 x float>* %x.addr to <4 x float>*
+  store <4 x float> %extractVec, <4 x float>* %storetmp, align 16
+  %tmp = load <3 x float>* %x.addr, align 16
+  %cmp = fcmp une <3 x float> %tmp, zeroinitializer
+  %sext = sext <3 x i1> %cmp to <3 x i32>
+  %and = and <3 x i32> <i32 1065353216, i32 1065353216, i32 1065353216>, %sext
+  %extractVec1 = shufflevector <3 x i32> %and, <3 x i32> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 undef>
+  %storetmp2 = bitcast <3 x i32>* %r to <4 x i32>*
+  store <4 x i32> %extractVec1, <4 x i32>* %storetmp2, align 16
+  %tmp3 = load <3 x i32>* %r, align 16
+  %0 = bitcast <3 x i32> %tmp3 to <3 x float>
+  %tmp4 = load <3 x float>* %x.addr, align 16
+  ret <3 x float> %tmp4
+; CHECK: @test14
+; CHECK-NOT: alloca
+; CHECK: shufflevector <4 x i32> %extractVec1, <4 x i32> undef, <3 x i32> <i32 0, i32 1, i32 2>
+}
+
+define void @test15(<3 x i64>* sret %agg.result, <3 x i64> %x, <3 x i64> %min) {
+entry:
+  %x.addr = alloca <3 x i64>, align 32
+  %min.addr = alloca <3 x i64>, align 32
+  %extractVec = shufflevector <3 x i64> %x, <3 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 undef>
+  %storetmp = bitcast <3 x i64>* %x.addr to <4 x i64>*
+  store <4 x i64> %extractVec, <4 x i64>* %storetmp, align 32
+  %extractVec1 = shufflevector <3 x i64> %min, <3 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 undef>
+  %storetmp2 = bitcast <3 x i64>* %min.addr to <4 x i64>*
+  store <4 x i64> %extractVec1, <4 x i64>* %storetmp2, align 32
+  %tmp = load <3 x i64>* %x.addr
+  %tmp5 = extractelement <3 x i64> %tmp, i32 0
+  %tmp11 = insertelement <3 x i64> %tmp, i64 %tmp5, i32 0
+  store <3 x i64> %tmp11, <3 x i64>* %x.addr
+  %tmp30 = load <3 x i64>* %x.addr, align 32
+  store <3 x i64> %tmp30, <3 x i64>* %agg.result
+  ret void
+; CHECK: @test15
+; CHECK-NOT: alloca
+; CHECK: shufflevector <4 x i64> %tmpV2, <4 x i64> undef, <3 x i32> <i32 0, i32 1, i32 2>
+}
+
+define <4 x float> @test16(<4 x float> %x, i64 %y0, i64 %y1) {
+entry:
+  %tmp8 = bitcast <4 x float> undef to <2 x double>
+  %tmp9 = bitcast i64 %y0 to double
+  %tmp10 = insertelement <2 x double> %tmp8, double %tmp9, i32 0
+  %tmp11 = bitcast <2 x double> %tmp10 to <4 x float>
+  %tmp3 = bitcast <4 x float> %tmp11 to <2 x double>
+  %tmp4 = bitcast i64 %y1 to double
+  %tmp5 = insertelement <2 x double> %tmp3, double %tmp4, i32 1
+  %tmp6 = bitcast <2 x double> %tmp5 to <4 x float>
+	ret <4 x float> %tmp6
+; CHECK: @test16
+; CHECK-NOT: alloca
+; CHECK: bitcast <4 x float> %tmp11 to <2 x double>
+}