Merge commit 'b3201c5cf1e183d840f7c99ff779d57f1549d8e5' into merge_20130226

Conflicts: include/llvm/Support/ELF.h lib/Support/DeltaAlgorithm.cpp Change-Id: I24a4fbce62eb39d924efee3c687b55e1e17b30cd
author: Stephen Hines <srhines@google.com> 2013-03-05 23:27:24 -0800
committer: Stephen Hines <srhines@google.com> 2013-03-05 23:27:24 -0800
commit: 5adb136be579e8fff3734461580cb34d1d2983b8 (patch)
tree: bff1a422e9c9789df563aaf9a7e91e63e8ec0384 /test/Transforms
parent: 227a4a4ade38716ba9eb3205f48b52910f3b955e (diff)
parent: b3201c5cf1e183d840f7c99ff779d57f1549d8e5 (diff)
download: external_llvm-5adb136be579e8fff3734461580cb34d1d2983b8.zip
external_llvm-5adb136be579e8fff3734461580cb34d1d2983b8.tar.gz
external_llvm-5adb136be579e8fff3734461580cb34d1d2983b8.tar.bz2
139 files changed, 4078 insertions, 499 deletions
diff --git a/test/Transforms/ArgumentPromotion/2008-02-01-ReturnAttrs.ll b/test/Transforms/ArgumentPromotion/2008-02-01-ReturnAttrs.ll
index e740b29..1226b98 100644
--- a/test/Transforms/ArgumentPromotion/2008-02-01-ReturnAttrs.ll
+++ b/test/Transforms/ArgumentPromotion/2008-02-01-ReturnAttrs.ll
@@ -1,15 +1,19 @@
-; RUN: opt < %s -argpromotion -S | grep nounwind | count 2
+; RUN: opt < %s -argpromotion -S | FileCheck %s
 
+; CHECK: define internal i32 @deref(i32 %x.val) #0 {
 define internal i32 @deref(i32* %x) nounwind {
 entry:
-	%tmp2 = load i32* %x, align 4		; <i32> [#uses=1]
-	ret i32 %tmp2
+  %tmp2 = load i32* %x, align 4
+  ret i32 %tmp2
 }
 
 define i32 @f(i32 %x) {
 entry:
-	%x_addr = alloca i32		; <i32*> [#uses=2]
-	store i32 %x, i32* %x_addr, align 4
-	%tmp1 = call i32 @deref( i32* %x_addr ) nounwind 		; <i32> [#uses=1]
-	ret i32 %tmp1
+  %x_addr = alloca i32
+  store i32 %x, i32* %x_addr, align 4
+; CHECK: %tmp1 = call i32 @deref(i32 %x_addr.val) [[NUW:#[0-9]+]]
+  %tmp1 = call i32 @deref( i32* %x_addr ) nounwind
+  ret i32 %tmp1
 }
+
+; CHECK: attributes [[NUW]] = { nounwind }
diff --git a/test/Transforms/BBVectorize/X86/pr15289.ll b/test/Transforms/BBVectorize/X86/pr15289.ll
new file mode 100644
index 0000000..07cc5d8
--- /dev/null
+++ b/test/Transforms/BBVectorize/X86/pr15289.ll
@@ -0,0 +1,98 @@
+; RUN: opt < %s -basicaa -bb-vectorize -disable-output
+; This is a bugpoint-reduced test case. It did not always assert, but does reproduce the bug
+; and running under valgrind (or some similar tool) will catch the error.
+
+target datalayout = "e-p:64:64:64-S128-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f16:16:16-f32:32:32-f64:64:64-f128:128:128-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64"
+target triple = "x86_64-apple-darwin12.2.0"
+
+%0 = type { [10 x { float, float }], [10 x { float, float }], [10 x { float, float }], [10 x { float, float }], [10 x { float, float }] }
+%1 = type { [10 x [8 x i8]] }
+%2 = type { i64, i64 }
+%3 = type { [10 x i64], i64, i64, i64, i64, i64 }
+%4 = type { i64, i64, i64, i64, i64, i64 }
+%5 = type { [10 x i64] }
+%6 = type { [10 x float], [10 x float], [10 x float], [10 x float] }
+%struct.__st_parameter_dt.1.3.5.7 = type { %struct.__st_parameter_common.0.2.4.6, i64, i64*, i64*, i8*, i8*, i32, i32, i8*, i8*, i32, i32, i8*, [256 x i8], i32*, i64, i8*, i32, i32, i8*, i8*, i32, i32, i8*, i8*, i32, i32, i8*, i8*, i32, [4 x i8] }
+%struct.__st_parameter_common.0.2.4.6 = type { i32, i32, i8*, i32, i32, i8*, i32* }
+
+@cctenso_ = external unnamed_addr global %0, align 32
+@ctenso_ = external unnamed_addr global %1, align 32
+@i_dim_ = external unnamed_addr global %2, align 16
+@itenso1_ = external unnamed_addr global %3, align 32
+@itenso2_ = external unnamed_addr global %4, align 32
+@ltenso_ = external unnamed_addr global %5, align 32
+@rtenso_ = external unnamed_addr global %6, align 32
+@.cst = external unnamed_addr constant [8 x i8], align 8
+@.cst1 = external unnamed_addr constant [3 x i8], align 8
+@.cst2 = external unnamed_addr constant [29 x i8], align 8
+@.cst3 = external unnamed_addr constant [32 x i8], align 64
+
+define void @cart_to_dc2y_(double* noalias nocapture %xx, double* noalias nocapture %yy, double* noalias nocapture %zz, [5 x { double, double }]* noalias nocapture %c2ten) nounwind uwtable {
+entry:
+  %0 = fmul double undef, undef
+  %1 = fmul double undef, undef
+  %2 = fadd double undef, undef
+  %3 = fmul double undef, 0x3FE8B8B76E3E9919
+  %4 = fsub double %0, %1
+  %5 = fsub double -0.000000e+00, undef
+  %6 = fmul double undef, undef
+  %7 = fmul double %4, %6
+  %8 = fmul double undef, 2.000000e+00
+  %9 = fmul double %8, undef
+  %10 = fmul double undef, %9
+  %11 = fmul double %10, undef
+  %12 = fsub double undef, %7
+  %13 = fmul double %3, %12
+  %14 = fmul double %3, undef
+  %15 = getelementptr inbounds [5 x { double, double }]* %c2ten, i64 0, i64 0, i32 0
+  store double %13, double* %15, align 8, !tbaa !0
+  %16 = getelementptr inbounds [5 x { double, double }]* %c2ten, i64 0, i64 0, i32 1
+  %17 = fmul double undef, %8
+  %18 = fmul double %17, undef
+  %19 = fmul double undef, %18
+  %20 = fadd double undef, undef
+  %21 = fmul double %3, %19
+  %22 = fsub double -0.000000e+00, %21
+  %23 = getelementptr inbounds [5 x { double, double }]* %c2ten, i64 0, i64 1, i32 0
+  store double %22, double* %23, align 8, !tbaa !0
+  %24 = getelementptr inbounds [5 x { double, double }]* %c2ten, i64 0, i64 1, i32 1
+  %25 = fmul double undef, 0x3FE42F601A8C6794
+  %26 = fmul double undef, 2.000000e+00
+  %27 = fsub double %26, %0
+  %28 = fmul double %6, undef
+  %29 = fsub double undef, %28
+  %30 = getelementptr inbounds [5 x { double, double }]* %c2ten, i64 0, i64 2, i32 0
+  store double undef, double* %30, align 8, !tbaa !0
+  %31 = getelementptr inbounds [5 x { double, double }]* %c2ten, i64 0, i64 2, i32 1
+  %32 = fmul double undef, %17
+  %33 = fmul double undef, %17
+  %34 = fmul double undef, %32
+  %35 = fmul double undef, %33
+  %36 = fsub double undef, %35
+  %37 = fmul double %3, %34
+  %38 = getelementptr inbounds [5 x { double, double }]* %c2ten, i64 0, i64 3, i32 0
+  store double %37, double* %38, align 8, !tbaa !0
+  %39 = getelementptr inbounds [5 x { double, double }]* %c2ten, i64 0, i64 3, i32 1
+  %40 = fmul double undef, %8
+  %41 = fmul double undef, %40
+  %42 = fmul double undef, %41
+  %43 = fsub double undef, %42
+  %44 = fmul double %3, %43
+  %45 = getelementptr inbounds [5 x { double, double }]* %c2ten, i64 0, i64 4, i32 0
+  store double %13, double* %45, align 8, !tbaa !0
+  %46 = getelementptr inbounds [5 x { double, double }]* %c2ten, i64 0, i64 4, i32 1
+  %47 = fsub double -0.000000e+00, %14
+  store double %47, double* %16, align 8, !tbaa !0
+  store double undef, double* %24, align 8, !tbaa !0
+  store double -0.000000e+00, double* %31, align 8, !tbaa !0
+  store double undef, double* %39, align 8, !tbaa !0
+  store double undef, double* %46, align 8, !tbaa !0
+  ret void
+}
+
+attributes #0 = { nounwind uwtable }
+attributes #1 = { nounwind readnone }
+attributes #2 = { nounwind }
+
+!0 = metadata !{metadata !"alias set 17: real(kind=8)", metadata !1}
+!1 = metadata !{metadata !1}
diff --git a/test/Transforms/BBVectorize/simple-int.ll b/test/Transforms/BBVectorize/simple-int.ll
index d7b7d6b..e4d5152 100644
--- a/test/Transforms/BBVectorize/simple-int.ll
+++ b/test/Transforms/BBVectorize/simple-int.ll
@@ -124,8 +124,10 @@ define double @test4(double %A1, double %A2, double %B1, double %B2, i32 %P) {
 ; CHECK: ret double %R
 }
 
-; CHECK: declare <2 x double> @llvm.fma.v2f64(<2 x double>, <2 x double>, <2 x double>) nounwind readnone
-; CHECK: declare <2 x double> @llvm.fmuladd.v2f64(<2 x double>, <2 x double>, <2 x double>) nounwind readnone
-; CHECK: declare <2 x double> @llvm.cos.v2f64(<2 x double>) nounwind readonly
-; CHECK: declare <2 x double> @llvm.powi.v2f64(<2 x double>, i32) nounwind readonly
+; CHECK: declare <2 x double> @llvm.fma.v2f64(<2 x double>, <2 x double>, <2 x double>) #0
+; CHECK: declare <2 x double> @llvm.fmuladd.v2f64(<2 x double>, <2 x double>, <2 x double>) #0
+; CHECK: declare <2 x double> @llvm.cos.v2f64(<2 x double>) #1
+; CHECK: declare <2 x double> @llvm.powi.v2f64(<2 x double>, i32) #1
 
+; CHECK: attributes #0 = { nounwind readnone }
+; CHECK: attributes #1 = { nounwind readonly }
diff --git a/test/Transforms/ConstProp/2007-11-23-cttz.ll b/test/Transforms/ConstProp/2007-11-23-cttz.ll
index 6d34cb1..c5ee70c 100644
--- a/test/Transforms/ConstProp/2007-11-23-cttz.ll
+++ b/test/Transforms/ConstProp/2007-11-23-cttz.ll
@@ -3,6 +3,6 @@
 declare i13 @llvm.cttz.i13(i13, i1)
 
 define i13 @test() {
-	%X = call i13 @llvm.cttz.i13(i13 0, i1 true)
+	%X = call i13 @llvm.cttz.i13(i13 0, i1 false)
 	ret i13 %X
 }
diff --git a/test/Transforms/CorrelatedValuePropagation/basic.ll b/test/Transforms/CorrelatedValuePropagation/basic.ll
index 475cd8d..39c437c 100644
--- a/test/Transforms/CorrelatedValuePropagation/basic.ll
+++ b/test/Transforms/CorrelatedValuePropagation/basic.ll
@@ -81,6 +81,26 @@ LessThanOrEqualToTwo:
   ret i32 0
 }
 
+declare i32* @f(i32*)
+define void @test5(i32* %x, i32* %y) {
+; CHECK: @test5
+entry:
+  %pre = icmp eq i32* %x, null
+  br i1 %pre, label %return, label %loop
+
+loop:
+  %phi = phi i32* [ %sel, %loop ], [ %x, %entry ]
+; CHECK: %phi = phi i32* [ %f, %loop ], [ %x, %entry ]
+  %f = tail call i32* @f(i32* %phi)
+  %cmp1 = icmp ne i32* %f, %y
+  %sel = select i1 %cmp1, i32* %f, i32* null
+  %cmp2 = icmp eq i32* %sel, null
+  br i1 %cmp2, label %return, label %loop
+
+return:
+  ret void
+}
+
 define i32 @switch1(i32 %s) {
 ; CHECK: @switch1
 entry:
@@ -105,7 +125,7 @@ negative:
   ]
 
 out:
-  %p = phi i32 [ 1, %entry ], [ -1, %negative ], [ -1, %negative ], [ -1, %negative ], [ -1, %negative ]
+  %p = phi i32 [ 1, %entry ], [ -1, %negative ], [ -1, %negative ], [ -1, %negative ], [ -1, %negative ], [ -1, %negative ]
   ret i32 %p
 
 next:
diff --git a/test/Transforms/DeadArgElim/2007-12-20-ParamAttrs.ll b/test/Transforms/DeadArgElim/2007-12-20-ParamAttrs.ll
index 7c6c575..f049265 100644
--- a/test/Transforms/DeadArgElim/2007-12-20-ParamAttrs.ll
+++ b/test/Transforms/DeadArgElim/2007-12-20-ParamAttrs.ll
@@ -1,20 +1,20 @@
-; RUN: opt < %s -deadargelim -S > %t
-; RUN: cat %t | grep nounwind | count 2
-; RUN: cat %t | grep signext | count 2
-; RUN: cat %t | not grep inreg
-; RUN: cat %t | not grep zeroext
-; RUN: cat %t | not grep byval
+; RUN: opt < %s -deadargelim -S | FileCheck %s
 
-	%struct = type { }
+%struct = type { }
 
 @g = global i8 0
 
+; CHECK: define internal void @foo(i8 signext %y) [[NUW:#[0-9]+]]
+
 define internal zeroext i8 @foo(i8* inreg %p, i8 signext %y, ... )  nounwind {
-	store i8 %y, i8* @g
-	ret i8 0
+  store i8 %y, i8* @g
+  ret i8 0
 }
 
 define i32 @bar() {
-	%A = call zeroext i8(i8*, i8, ...)* @foo(i8* inreg null, i8 signext 1, %struct* byval null ) nounwind
-	ret i32 0
+; CHECK: call void @foo(i8 signext 1) [[NUW]]
+  %A = call zeroext i8(i8*, i8, ...)* @foo(i8* inreg null, i8 signext 1, %struct* byval null ) nounwind
+  ret i32 0
 }
+
+; CHECK: attributes [[NUW]] = { nounwind }
diff --git a/test/Transforms/DeadArgElim/2010-04-30-DbgInfo.ll b/test/Transforms/DeadArgElim/2010-04-30-DbgInfo.ll
index 2f820ba..f5d2588 100644
--- a/test/Transforms/DeadArgElim/2010-04-30-DbgInfo.ll
+++ b/test/Transforms/DeadArgElim/2010-04-30-DbgInfo.ll
@@ -8,14 +8,14 @@ entry:
   call void @llvm.dbg.value(metadata !{i32 %len}, i64 0, metadata !10)
   call void @llvm.dbg.value(metadata !{i32 %hash}, i64 0, metadata !11)
   call void @llvm.dbg.value(metadata !{i32 %flags}, i64 0, metadata !12)
-; CHECK:  call fastcc i8* @add_name_internal(i8* %name, i32 %hash) nounwind, !dbg !13
+; CHECK:  call fastcc i8* @add_name_internal(i8* %name, i32 %hash) [[NUW:#[0-9]+]], !dbg !13
   %0 = call fastcc i8* @add_name_internal(i8* %name, i32 %len, i32 %hash, i8 zeroext 0, i32 %flags) nounwind, !dbg !13 ; <i8*> [#uses=1]
   ret i8* %0, !dbg !13
 }
 
 declare void @llvm.dbg.declare(metadata, metadata) nounwind readnone
 
-define internal fastcc i8* @add_name_internal(i8* %name, i32 %len, i32 %hash, i8 zeroext %extra, i32 %flags) nounwind noinline ssp {
+define internal fastcc i8* @add_name_internal(i8* %name, i32 %len, i32 %hash, i8 zeroext %extra, i32 %flags) noinline nounwind ssp {
 entry:
   call void @llvm.dbg.value(metadata !{i8* %name}, i64 0, metadata !15)
   call void @llvm.dbg.value(metadata !{i32 %len}, i64 0, metadata !20)
@@ -38,6 +38,11 @@ bb2:                                              ; preds = %bb1, %bb
 
 declare void @llvm.dbg.value(metadata, i64, metadata) nounwind readnone
 
+; CHECK: attributes #0 = { nounwind ssp }
+; CHECK: attributes #1 = { nounwind readnone }
+; CHECK: attributes #2 = { noinline nounwind ssp }
+; CHECK: attributes [[NUW]] = { nounwind }
+
 !0 = metadata !{i32 524545, metadata !1, metadata !"name", metadata !2, i32 8, metadata !6} ; [ DW_TAG_arg_variable ]
 !1 = metadata !{i32 524334, i32 0, metadata !2, metadata !"vfs_addname", metadata !"vfs_addname", metadata !"vfs_addname", metadata !2, i32 12, metadata !4, i1 false, i1 true, i32 0, i32 0, null, i1 false} ; [ DW_TAG_subprogram ]
 !2 = metadata !{i32 524329, metadata !"tail.c", metadata !"/Users/echeng/LLVM/radars/r7927803/", metadata !3} ; [ DW_TAG_file_type ]
diff --git a/test/Transforms/DeadArgElim/dbginfo.ll b/test/Transforms/DeadArgElim/dbginfo.ll
index b07b60d..59eb458 100644
--- a/test/Transforms/DeadArgElim/dbginfo.ll
+++ b/test/Transforms/DeadArgElim/dbginfo.ll
@@ -37,13 +37,11 @@ entry:
 !llvm.dbg.cu = !{!0}
 
 !0 = metadata !{i32 786449, i32 0, i32 4, metadata !"test.cc", metadata !"/home/samsonov/tmp/clang-di", metadata !"clang version 3.2 (trunk 165305)", i1 true, i1 false, metadata !"", i32 0, metadata !1, metadata !1, metadata !3, metadata !1} ; [ DW_TAG_compile_unit ] [/home/samsonov/tmp/clang-di/test.cc] [DW_LANG_C_plus_plus]
-!1 = metadata !{metadata !2}
-!2 = metadata !{i32 0}
-!3 = metadata !{metadata !4}
-!4 = metadata !{metadata !5, metadata !8, metadata !9}
+!1 = metadata !{i32 0}
+!3 = metadata !{metadata !5, metadata !8, metadata !9}
 !5 = metadata !{i32 786478, i32 0, metadata !6, metadata !"run", metadata !"run", metadata !"", metadata !6, i32 8, metadata !7, i1 false, i1 true, i32 0, i32 0, null, i32 256, i1 false, void ()* @_Z3runv, null, null, metadata !1, i32 8} ; [ DW_TAG_subprogram ] [line 8] [def] [run]
 !6 = metadata !{i32 786473, metadata !"test.cc", metadata !"/home/samsonov/tmp/clang-di", null} ; [ DW_TAG_file_type ]
-!7 = metadata !{i32 786453, i32 0, metadata !"", i32 0, i32 0, i64 0, i64 0, i64 0, i32 0, null, metadata !2, i32 0, i32 0} ; [ DW_TAG_subroutine_type ] [line 0, size 0, align 0, offset 0] [from ]
+!7 = metadata !{i32 786453, i32 0, metadata !"", i32 0, i32 0, i64 0, i64 0, i64 0, i32 0, null, metadata !1, i32 0, i32 0} ; [ DW_TAG_subroutine_type ] [line 0, size 0, align 0, offset 0] [from ]
 !8 = metadata !{i32 786478, i32 0, metadata !6, metadata !"dead_vararg", metadata !"dead_vararg", metadata !"", metadata !6, i32 5, metadata !7, i1 true, i1 true, i32 0, i32 0, null, i32 256, i1 false, void (...)* @_ZN12_GLOBAL__N_111dead_varargEz, null, null, metadata !1, i32 5} ; [ DW_TAG_subprogram ] [line 5] [local] [def] [dead_vararg]
 
 ; CHECK: metadata !"dead_vararg"{{.*}}void ()* @_ZN12_GLOBAL__N_111dead_varargEz
diff --git a/test/Transforms/DeadArgElim/keepalive.ll b/test/Transforms/DeadArgElim/keepalive.ll
index dc92dc9..e41110c 100644
--- a/test/Transforms/DeadArgElim/keepalive.ll
+++ b/test/Transforms/DeadArgElim/keepalive.ll
@@ -1,6 +1,4 @@
-; RUN: opt < %s -deadargelim -S > %t
-; RUN: grep "define internal zeroext i32 @test1() nounwind" %t
-; RUN: grep "define internal <{ i32, i32 }> @test2" %t
+; RUN: opt < %s -deadargelim -S | FileCheck %s
 
 %Ty = type <{ i32, i32 }>
 
@@ -9,11 +7,13 @@
 ; the function and then changing too much.
 
 ; This checks if the return value attributes are not removed
+; CHECK: define internal zeroext i32 @test1() #0
 define internal zeroext i32 @test1(i32 %DEADARG1) nounwind {
         ret i32 1
 }
 
 ; This checks if the struct doesn't get non-packed
+; CHECK: define internal <{ i32, i32 }> @test2
 define internal <{ i32, i32 }> @test2(i32 %DEADARG1) {
         ret <{ i32, i32 }> <{ i32 1, i32 2 }>
 }
@@ -28,3 +28,4 @@ define void @caller() {
         ret void
 }
 
+; CHECK: attributes #0 = { nounwind }
diff --git a/test/Transforms/FunctionAttrs/2008-09-03-ReadNone.ll b/test/Transforms/FunctionAttrs/2008-09-03-ReadNone.ll
index 946453f..36a7658 100644
--- a/test/Transforms/FunctionAttrs/2008-09-03-ReadNone.ll
+++ b/test/Transforms/FunctionAttrs/2008-09-03-ReadNone.ll
@@ -1,18 +1,24 @@
-; RUN: opt < %s -basicaa -functionattrs -S | grep readnone | count 4
+; RUN: opt < %s -basicaa -functionattrs -S | FileCheck %s
 @x = global i32 0
 
+; CHECK: declare i32 @e() #0
 declare i32 @e() readnone
 
+; CHECK: define i32 @f() #0
 define i32 @f() {
 	%tmp = call i32 @e( )		; <i32> [#uses=1]
 	ret i32 %tmp
 }
 
+; CHECK: define i32 @g() #0
 define i32 @g() readonly {
 	ret i32 0
 }
 
+; CHECK: define i32 @h() #0
 define i32 @h() readnone {
 	%tmp = load i32* @x		; <i32> [#uses=1]
 	ret i32 %tmp
 }
+
+; CHECK: attributes #0 = { readnone }
diff --git a/test/Transforms/FunctionAttrs/2008-09-03-ReadOnly.ll b/test/Transforms/FunctionAttrs/2008-09-03-ReadOnly.ll
index 22eca13..d8256ae 100644
--- a/test/Transforms/FunctionAttrs/2008-09-03-ReadOnly.ll
+++ b/test/Transforms/FunctionAttrs/2008-09-03-ReadOnly.ll
@@ -1,9 +1,13 @@
-; RUN: opt < %s -basicaa -functionattrs -S | grep readonly | count 2
+; RUN: opt < %s -basicaa -functionattrs -S | FileCheck %s
 
+; CHECK: define i32 @f() #0
 define i32 @f() {
 entry:
-	%tmp = call i32 @e( )		; <i32> [#uses=1]
-	ret i32 %tmp
+  %tmp = call i32 @e( )
+  ret i32 %tmp
 }
 
+; CHECK: declare i32 @e() #0
 declare i32 @e() readonly
+
+; CHECK: attributes #0 = { readonly }
diff --git a/test/Transforms/FunctionAttrs/atomic.ll b/test/Transforms/FunctionAttrs/atomic.ll
index 7c2bff7..027ee0f 100644
--- a/test/Transforms/FunctionAttrs/atomic.ll
+++ b/test/Transforms/FunctionAttrs/atomic.ll
@@ -3,7 +3,7 @@
 ; Atomic load/store to local doesn't affect whether a function is
 ; readnone/readonly.
 define i32 @test1(i32 %x) uwtable ssp {
-; CHECK: define i32 @test1(i32 %x) uwtable readnone ssp {
+; CHECK: define i32 @test1(i32 %x) #0 {
 entry:
   %x.addr = alloca i32, align 4
   store atomic i32 %x, i32* %x.addr seq_cst, align 4
@@ -13,9 +13,11 @@ entry:
 
 ; A function with an Acquire load is not readonly.
 define i32 @test2(i32* %x) uwtable ssp {
-; CHECK: define i32 @test2(i32* nocapture %x) uwtable ssp {
+; CHECK: define i32 @test2(i32* nocapture %x) #1 {
 entry:
   %r = load atomic i32* %x seq_cst, align 4
   ret i32 %r
 }
 
+; CHECK: attributes #0 = { readnone ssp uwtable }
+; CHECK: attributes #1 = { ssp uwtable }
diff --git a/test/Transforms/FunctionAttrs/noreturn.ll b/test/Transforms/FunctionAttrs/noreturn.ll
new file mode 100644
index 0000000..470ebcb
--- /dev/null
+++ b/test/Transforms/FunctionAttrs/noreturn.ll
@@ -0,0 +1,18 @@
+; RUN: opt < %s -functionattrs -instcombine -S | FileCheck %s
+
+define void @endless_loop() noreturn nounwind readnone ssp uwtable {
+entry:
+  br label %while.body
+
+while.body:
+  br label %while.body
+}
+;CHECK: @main
+;CHECK: endless_loop
+;CHECK: ret
+define i32 @main() noreturn nounwind ssp uwtable {
+entry:
+  tail call void @endless_loop()
+  unreachable
+}
+
diff --git a/test/Transforms/GlobalOpt/externally-initialized-global-ctr.ll b/test/Transforms/GlobalOpt/externally-initialized-global-ctr.ll
new file mode 100644
index 0000000..9295c20
--- /dev/null
+++ b/test/Transforms/GlobalOpt/externally-initialized-global-ctr.ll
@@ -0,0 +1,35 @@
+; RUN: opt < %s -globalopt -S | FileCheck %s
+; rdar://12580965.
+; ObjC++ test case.
+
+%struct.ButtonInitData = type { i8* }
+
+@_ZL14buttonInitData = internal global [1 x %struct.ButtonInitData] zeroinitializer, align 4
+
+@"\01L_OBJC_METH_VAR_NAME_40" = internal global [7 x i8] c"print:\00", section "__TEXT,__objc_methname,cstring_literals", align 1
+@"\01L_OBJC_SELECTOR_REFERENCES_41" = internal externally_initialized  global i8* getelementptr inbounds ([7 x i8]* @"\01L_OBJC_METH_VAR_NAME_40", i32 0, i32 0), section "__DATA, __objc_selrefs, literal_pointers, no_dead_strip"
+
+@llvm.global_ctors = appending global [1 x { i32, void ()* }] [{ i32, void ()* } { i32 65535, void ()* @_GLOBAL__I_a }]
+@llvm.used = appending global [2 x i8*] [i8* getelementptr inbounds ([7 x i8]* @"\01L_OBJC_METH_VAR_NAME_40", i32 0, i32 0),  i8* bitcast (i8** @"\01L_OBJC_SELECTOR_REFERENCES_41" to i8*)]
+
+define internal void @__cxx_global_var_init() section "__TEXT,__StaticInit,regular,pure_instructions" {
+  %1 = load i8** @"\01L_OBJC_SELECTOR_REFERENCES_41", !invariant.load !2009
+  store i8* %1, i8** getelementptr inbounds ([1 x %struct.ButtonInitData]* @_ZL14buttonInitData, i32 0, i32 0, i32 0), align 4
+  ret void
+}
+
+define internal void @_GLOBAL__I_a() section "__TEXT,__StaticInit,regular,pure_instructions" {
+  call void @__cxx_global_var_init()
+  ret void
+}
+
+declare void @test(i8*)
+
+define void @print() {
+; CHECK: %1 = load i8** getelementptr inbounds ([1 x %struct.ButtonInitData]* @_ZL14buttonInitData, i32 0, i32 0, i32 0), align 4
+  %1 = load i8** getelementptr inbounds ([1 x %struct.ButtonInitData]* @_ZL14buttonInitData, i32 0, i32 0, i32 0), align 4
+  call void @test(i8* %1)
+  ret void
+}
+
+!2009 = metadata !{}
diff --git a/test/Transforms/GlobalOpt/integer-bool.ll b/test/Transforms/GlobalOpt/integer-bool.ll
index 5a34a9c..cf025ec 100644
--- a/test/Transforms/GlobalOpt/integer-bool.ll
+++ b/test/Transforms/GlobalOpt/integer-bool.ll
@@ -1,23 +1,28 @@
-; RUN: opt < %s -globalopt -instcombine | \
-; RUN:    llvm-dis | grep "ret i1 true"
-
+; RUN: opt < %s -S -globalopt -instcombine | FileCheck %s
 ;; check that global opt turns integers that only hold 0 or 1 into bools.
 
-@G = internal global i32 0              ; <i32*> [#uses=3]
+@G = internal addrspace(1) global i32 0
+; CHECK @G.b
+; CHECK addrspace(1)
+; CHECK global i1 0
 
 define void @set1() {
-        store i32 0, i32* @G
-        ret void
+  store i32 0, i32 addrspace(1)* @G
+; CHECK: store i1 false
+  ret void
 }
 
 define void @set2() {
-        store i32 1, i32* @G
-        ret void
+  store i32 1, i32 addrspace(1)* @G
+; CHECK: store i1 true
+  ret void
 }
 
 define i1 @get() {
-        %A = load i32* @G               ; <i32> [#uses=1]
-        %C = icmp slt i32 %A, 2         ; <i1> [#uses=1]
-        ret i1 %C
+; CHECK @get
+  %A = load i32 addrspace(1) * @G
+  %C = icmp slt i32 %A, 2
+  ret i1 %C
+; CHECK: ret i1 true
 }
 
diff --git a/test/Transforms/IPConstantProp/user-with-multiple-uses.ll b/test/Transforms/IPConstantProp/user-with-multiple-uses.ll
index 402ea41..9687180 100644
--- a/test/Transforms/IPConstantProp/user-with-multiple-uses.ll
+++ b/test/Transforms/IPConstantProp/user-with-multiple-uses.ll
@@ -4,9 +4,9 @@
 ; IPSCCP should propagate the 0 argument, eliminate the switch, and propagate
 ; the result.
 
-; CHECK: define i32 @main() noreturn nounwind {
+; CHECK: define i32 @main() #0 {
 ; CHECK-NEXT: entry:
-; CHECK-NEXT: %call2 = tail call i32 @wwrite(i64 0) nounwind
+; CHECK-NEXT: %call2 = tail call i32 @wwrite(i64 0) [[NUW:#[0-9]+]]
 ; CHECK-NEXT: ret i32 123
 
 define i32 @main() noreturn nounwind {
@@ -28,3 +28,7 @@ sw.default:
 return:
   ret i32 0
 }
+
+; CHECK: attributes #0 = { noreturn nounwind }
+; CHECK: attributes #1 = { nounwind readnone }
+; CHECK: attributes [[NUW]] = { nounwind }
diff --git a/test/Transforms/Inline/inline_invoke.ll b/test/Transforms/Inline/inline_invoke.ll
index 9f5f670..c53bb5a 100644
--- a/test/Transforms/Inline/inline_invoke.ll
+++ b/test/Transforms/Inline/inline_invoke.ll
@@ -330,7 +330,7 @@ terminate:
 ; CHECK-NEXT: br label %[[JOIN]]
 ; CHECK:    [[JOIN]]:
 ; CHECK-NEXT: phi { i8*, i32 }
-; CHECK-NEXT: call void @opaque() nounwind
+; CHECK-NEXT: call void @opaque() [[NUW:#[0-9]+]]
 ; CHECK-NEXT: br label %[[FIX:[^\s]+]]
 ; CHECK:    lpad:
 ; CHECK-NEXT: landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
@@ -340,3 +340,8 @@ terminate:
 ; CHECK-NEXT: [[T1:%.*]] = phi i32 [ 0, %[[JOIN]] ], [ 1, %lpad ]
 ; CHECK-NEXT: call void @use(i32 [[T1]])
 ; CHECK-NEXT: call void @_ZSt9terminatev()
+
+; CHECK: attributes [[NUW]] = { nounwind }
+; CHECK: attributes #1 = { nounwind readnone }
+; CHECK: attributes #2 = { ssp uwtable }
+; CHECK: attributes #3 = { noreturn nounwind }
diff --git a/test/Transforms/Inline/inline_ssp.ll b/test/Transforms/Inline/inline_ssp.ll
new file mode 100644
index 0000000..a4b43a7
--- /dev/null
+++ b/test/Transforms/Inline/inline_ssp.ll
@@ -0,0 +1,160 @@
+; RUN: opt -inline %s -S | FileCheck %s
+; Ensure SSP attributes are propagated correctly when inlining.
+
+@.str = private unnamed_addr constant [11 x i8] c"fun_nossp\0A\00", align 1
+@.str1 = private unnamed_addr constant [9 x i8] c"fun_ssp\0A\00", align 1
+@.str2 = private unnamed_addr constant [15 x i8] c"fun_sspstrong\0A\00", align 1
+@.str3 = private unnamed_addr constant [12 x i8] c"fun_sspreq\0A\00", align 1
+
+; These first four functions (@fun_sspreq, @fun_sspstrong, @fun_ssp, @fun_nossp)
+; are used by the remaining functions to ensure that the SSP attributes are
+; propagated correctly.  The caller should have its SSP attribute set as:
+; strictest(caller-ssp-attr, callee-ssp-attr), where strictness is ordered as:
+;  sspreq > sspstrong > ssp > [no ssp]
+define internal void @fun_sspreq() nounwind sspreq uwtable {
+entry:
+  %call = call i32 (i8*, ...)* @printf(i8* getelementptr inbounds ([12 x i8]* @.str3, i32 0, i32 0))
+  ret void
+}
+
+define internal void @fun_sspstrong() nounwind sspstrong uwtable {
+entry:
+  %call = call i32 (i8*, ...)* @printf(i8* getelementptr inbounds ([15 x i8]* @.str2, i32 0, i32 0))
+  ret void
+}
+
+define internal void @fun_ssp() nounwind ssp uwtable {
+entry:
+  %call = call i32 (i8*, ...)* @printf(i8* getelementptr inbounds ([9 x i8]* @.str1, i32 0, i32 0))
+  ret void
+}
+
+define internal void @fun_nossp() nounwind uwtable {
+entry:
+  %call = call i32 (i8*, ...)* @printf(i8* getelementptr inbounds ([11 x i8]* @.str, i32 0, i32 0))
+  ret void
+}
+
+; Tests start below 
+
+define void @inline_req_req() nounwind sspreq uwtable {
+entry:
+; CHECK: @inline_req_req() #0
+  call void @fun_sspreq()
+  ret void
+}
+
+define void @inline_req_strong() nounwind sspstrong uwtable {
+entry:
+; CHECK: @inline_req_strong() #0
+  call void @fun_sspreq()
+  ret void
+}
+
+define void @inline_req_ssp() nounwind ssp uwtable {
+entry:
+; CHECK: @inline_req_ssp() #0
+  call void @fun_sspreq()
+  ret void
+}
+
+define void @inline_req_nossp() nounwind uwtable {
+entry:
+; CHECK: @inline_req_nossp() #0
+  call void @fun_sspreq()
+  ret void
+}
+
+define void @inline_strong_req() nounwind sspreq uwtable {
+entry:
+; CHECK: @inline_strong_req() #0
+  call void @fun_sspstrong()
+  ret void
+}
+
+
+define void @inline_strong_strong() nounwind sspstrong uwtable {
+entry:
+; CHECK: @inline_strong_strong() #1
+  call void @fun_sspstrong()
+  ret void
+}
+
+define void @inline_strong_ssp() nounwind ssp uwtable {
+entry:
+; CHECK: @inline_strong_ssp() #1
+  call void @fun_sspstrong()
+  ret void
+}
+
+define void @inline_strong_nossp() nounwind uwtable {
+entry:
+; CHECK: @inline_strong_nossp() #1
+  call void @fun_sspstrong()
+  ret void
+}
+
+define void @inline_ssp_req() nounwind sspreq uwtable {
+entry:
+; CHECK: @inline_ssp_req() #0
+  call void @fun_ssp()
+  ret void
+}
+
+
+define void @inline_ssp_strong() nounwind sspstrong uwtable {
+entry:
+; CHECK: @inline_ssp_strong() #1
+  call void @fun_ssp()
+  ret void
+}
+
+define void @inline_ssp_ssp() nounwind ssp uwtable {
+entry:
+; CHECK: @inline_ssp_ssp() #2
+  call void @fun_ssp()
+  ret void
+}
+
+define void @inline_ssp_nossp() nounwind uwtable {
+entry:
+; CHECK: @inline_ssp_nossp() #2
+  call void @fun_ssp()
+  ret void
+}
+
+define void @inline_nossp_req() nounwind uwtable sspreq {
+entry:
+; CHECK: @inline_nossp_req() #0
+  call void @fun_nossp()
+  ret void
+}
+
+
+define void @inline_nossp_strong() nounwind sspstrong uwtable {
+entry:
+; CHECK: @inline_nossp_strong() #1
+  call void @fun_nossp()
+  ret void
+}
+
+define void @inline_nossp_ssp() nounwind ssp uwtable {
+entry:
+; CHECK: @inline_nossp_ssp() #2
+  call void @fun_nossp()
+  ret void
+}
+
+define void @inline_nossp_nossp() nounwind uwtable {
+entry:
+; CHECK: @inline_nossp_nossp() #3
+  call void @fun_nossp()
+  ret void
+}
+
+declare i32 @printf(i8*, ...)
+
+; CHECK: attributes #0 = { nounwind sspreq uwtable }
+; CHECK: attributes #1 = { nounwind sspstrong uwtable }
+; CHECK: attributes #2 = { nounwind ssp uwtable }
+; CHECK: attributes #3 = { nounwind uwtable }
diff --git a/test/Transforms/InstCombine/2012-04-23-Neon-Intrinsics.ll b/test/Transforms/InstCombine/2012-04-23-Neon-Intrinsics.ll
index 0907c49..2dedd44 100644
--- a/test/Transforms/InstCombine/2012-04-23-Neon-Intrinsics.ll
+++ b/test/Transforms/InstCombine/2012-04-23-Neon-Intrinsics.ll
@@ -50,7 +50,7 @@ entry:
   %b = add <4 x i32> zeroinitializer, %a
   ret <4 x i32> %b
 ; CHECK: entry:
-; CHECK-NEXT: %a = tail call <4 x i32> @llvm.arm.neon.vmulls.v4i32(<4 x i16> <i16 2, i16 2, i16 2, i16 2>, <4 x i16> %x) nounwind
+; CHECK-NEXT: %a = tail call <4 x i32> @llvm.arm.neon.vmulls.v4i32(<4 x i16> <i16 2, i16 2, i16 2, i16 2>, <4 x i16> %x) [[NUW:#[0-9]+]]
 ; CHECK-NEXT: ret <4 x i32> %a
 }
 
@@ -66,3 +66,7 @@ entry:
 
 declare <4 x i32> @llvm.arm.neon.vmulls.v4i32(<4 x i16>, <4 x i16>) nounwind readnone
 declare <4 x i32> @llvm.arm.neon.vmullu.v4i32(<4 x i16>, <4 x i16>) nounwind readnone
+
+; CHECK: attributes #0 = { nounwind readnone ssp }
+; CHECK: attributes #1 = { nounwind readnone }
+; CHECK: attributes [[NUW]] = { nounwind }
diff --git a/test/Transforms/InstCombine/bitcast-vector-fold.ll b/test/Transforms/InstCombine/bitcast-vector-fold.ll
index 8feec22..8fd7f35 100644
--- a/test/Transforms/InstCombine/bitcast-vector-fold.ll
+++ b/test/Transforms/InstCombine/bitcast-vector-fold.ll
@@ -31,3 +31,8 @@ define <4 x i32> @test6() {
 	%tmp3 = bitcast <2 x double> <double 0.5, double 1.0> to <4 x i32>
 	ret <4 x i32> %tmp3
 }
+
+define i32 @test7() {
+       %tmp3 = bitcast <2 x half> <half 0xH1100, half 0xH0011> to i32
+       ret i32 %tmp3
+}
+\ No newline at end of file
diff --git a/test/Transforms/InstCombine/bitcast.ll b/test/Transforms/InstCombine/bitcast.ll
index 8f6ae7d..1e61132 100644
--- a/test/Transforms/InstCombine/bitcast.ll
+++ b/test/Transforms/InstCombine/bitcast.ll
@@ -11,7 +11,7 @@ define i32 @test1(i64 %a) {
         %t3 = xor <2 x i32> %t1, %t2
         %t4 = extractelement <2 x i32> %t3, i32 0
         ret i32 %t4
-        
+
 ; CHECK: @test1
 ; CHECK: ret i32 0
 }
@@ -30,7 +30,7 @@ define float @test2(<2 x float> %A, <2 x i32> %B) {
 
   %add = fadd float %tmp24, %tmp4
   ret float %add
-  
+
 ; CHECK: @test2
 ; CHECK-NEXT:  %tmp24 = extractelement <2 x float> %A, i32 0
 ; CHECK-NEXT:  bitcast <2 x i32> %B to <2 x float>
@@ -55,7 +55,7 @@ define float @test3(<2 x float> %A, <2 x i64> %B) {
 
   %add = fadd float %tmp24, %tmp4
   ret float %add
-  
+
 ; CHECK: @test3
 ; CHECK-NEXT:  %tmp24 = extractelement <2 x float> %A, i32 1
 ; CHECK-NEXT:  bitcast <2 x i64> %B to <4 x float>
@@ -75,7 +75,7 @@ define <2 x i32> @test4(i32 %A, i32 %B){
   ; CHECK: @test4
   ; CHECK-NEXT: insertelement <2 x i32> undef, i32 %A, i32 0
   ; CHECK-NEXT: insertelement <2 x i32> {{.*}}, i32 %B, i32 1
-  ; CHECK-NEXT: ret <2 x i32> 
+  ; CHECK-NEXT: ret <2 x i32>
 
 }
 
@@ -92,7 +92,7 @@ define <2 x float> @test5(float %A, float %B) {
   ; CHECK: @test5
   ; CHECK-NEXT: insertelement <2 x float> undef, float %A, i32 0
   ; CHECK-NEXT: insertelement <2 x float> {{.*}}, float %B, i32 1
-  ; CHECK-NEXT: ret <2 x float> 
+  ; CHECK-NEXT: ret <2 x float>
 }
 
 define <2 x float> @test6(float %A){
@@ -123,7 +123,7 @@ define i64 @Vec2(i64 %in) {
 }
 
 define i64 @All11(i64 %in) {
-  %out = and i64 %in, xor (i64 bitcast (<2 x float> bitcast (i64 -1 to <2 x float>) to i64), i64 -1) 
+  %out = and i64 %in, xor (i64 bitcast (<2 x float> bitcast (i64 -1 to <2 x float>) to i64), i64 -1)
   ret i64 %out
 ; CHECK: @All11
 ; CHECK: ret i64 0
@@ -131,9 +131,16 @@ define i64 @All11(i64 %in) {
 
 
 define i32 @All111(i32 %in) {
-  %out = and i32 %in, xor (i32 bitcast (<1 x float> bitcast (i32 -1 to <1 x float>) to i32), i32 -1) 
+  %out = and i32 %in, xor (i32 bitcast (<1 x float> bitcast (i32 -1 to <1 x float>) to i32), i32 -1)
   ret i32 %out
 ; CHECK: @All111
 ; CHECK: ret i32 0
 }
 
+define <2 x i16> @BitcastInsert(i32 %a) {
+  %v = insertelement <1 x i32> undef, i32 %a, i32 0
+  %r = bitcast <1 x i32> %v to <2 x i16>
+  ret <2 x i16> %r
+; CHECK: @BitcastInsert
+; CHECK: bitcast i32 %a to <2 x i16>
+}
diff --git a/test/Transforms/InstCombine/constant-expr-datalayout.ll b/test/Transforms/InstCombine/constant-expr-datalayout.ll
new file mode 100644
index 0000000..9a72c77
--- /dev/null
+++ b/test/Transforms/InstCombine/constant-expr-datalayout.ll
@@ -0,0 +1,12 @@
+; RUN: opt -instcombine %s -S -o - | 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-S128"
+target triple = "x86_64-unknown-linux-gnu"
+
+%test1.struct = type { i32, i32 }
+@test1.aligned_glbl = global %test1.struct zeroinitializer, align 4
+define void @test1(i64 *%ptr) {
+  store i64 and (i64 ptrtoint (i32* getelementptr (%test1.struct* @test1.aligned_glbl, i32 0, i32 1) to i64), i64 3), i64* %ptr
+; CHECK: store i64 0, i64* %ptr
+  ret void
+}
diff --git a/test/Transforms/InstCombine/exact.ll b/test/Transforms/InstCombine/exact.ll
index 14741e3..88ca88c 100644
--- a/test/Transforms/InstCombine/exact.ll
+++ b/test/Transforms/InstCombine/exact.ll
@@ -99,9 +99,9 @@ define i1 @ashr_icmp2(i64 %X) nounwind {
 ; PR9998
 ; Make sure we don't transform the ashr here into an sdiv
 ; CHECK: @pr9998
-; CHECK: = and i32 %V, 1
-; CHECK: %Z = icmp ne
-; CHECK: ret i1 %Z
+; CHECK:      [[BIT:%[A-Za-z0-9.]+]] = and i32 %V, 1
+; CHECK-NEXT: [[CMP:%[A-Za-z0-9.]+]] = icmp ne i32 [[BIT]], 0
+; CHECK-NEXT: ret i1 [[CMP]]
 define i1 @pr9998(i32 %V) nounwind {
 entry:
   %W = shl i32 %V, 31
@@ -112,6 +112,7 @@ entry:
 }
 
 
+
 ; CHECK: @udiv_icmp1
 ; CHECK: icmp ne i64 %X, 0
 define i1 @udiv_icmp1(i64 %X) nounwind {
diff --git a/test/Transforms/InstCombine/fast-math.ll b/test/Transforms/InstCombine/fast-math.ll
index 5d40d71..c97bd28 100644
--- a/test/Transforms/InstCombine/fast-math.ll
+++ b/test/Transforms/InstCombine/fast-math.ll
@@ -7,7 +7,7 @@ define float @fold(float %a) {
   %mul1 = fmul fast float %mul, 0x4002666660000000
   ret float %mul1
 ; CHECK: @fold
-; CHECK: fmul float %a, 0x4006147AE0000000
+; CHECK: fmul fast float %a, 0x4006147AE0000000
 }
 
 ; Same testing-case as the one used in fold() except that the operators have
@@ -22,7 +22,7 @@ define float @notfold(float %a) {
 
 define float @fold2(float %a) {
 ; CHECK: @fold2
-; CHECK: fmul float %a, 0x4006147AE0000000
+; CHECK: fmul fast float %a, 0x4006147AE0000000
   %mul = fmul float %a, 0x3FF3333340000000
   %mul1 = fmul fast float %mul, 0x4002666660000000
   ret float %mul1
@@ -54,7 +54,7 @@ define float @fold5(float %f1, float %f2) {
   %add1 = fadd fast float %add, 5.000000e+00
   ret float %add1
 ; CHECK: @fold5
-; CHECK: fadd float %f1, 9.000000e+00
+; CHECK: fadd fast float %f1, 9.000000e+00
 }
 
 ; (X + X) + X => 3.0 * X
@@ -97,17 +97,17 @@ define float @fold9(float %f1, float %f2) {
 }
 
 ; Let C3 = C1 + C2. (f1 + C1) + (f2 + C2) => (f1 + f2) + C3 instead of
-; "(f1 + C3) + f2" or "(f2 + C3) + f1". Placing constant-addend at the 
+; "(f1 + C3) + f2" or "(f2 + C3) + f1". Placing constant-addend at the
 ; top of resulting simplified expression tree may potentially reveal some
 ; optimization opportunities in the super-expression trees.
-; 
+;
 define float @fold10(float %f1, float %f2) {
   %t1 = fadd fast float 2.000000e+00, %f1
   %t2 = fsub fast float %f2, 3.000000e+00
   %t3 = fadd fast float %t1, %t2
   ret float %t3
 ; CHECK: @fold10
-; CHECK: %t3 = fadd float %t2, -1.000000e+00
+; CHECK: %t3 = fadd fast float %t2, -1.000000e+00
 ; CHECK: ret float %t3
 }
 
@@ -130,37 +130,6 @@ define double @fail2(double %f1, double %f2) {
 ; CHECK: ret
 }
 
-; rdar://12753946:  x * cond ? 1.0 : 0.0 => cond ? x : 0.0
-define double @select1(i32 %cond, double %x, double %y) {
-  %tobool = icmp ne i32 %cond, 0
-  %cond1 = select i1 %tobool, double 1.000000e+00, double 0.000000e+00
-  %mul = fmul nnan nsz double %cond1, %x
-  %add = fadd double %mul, %y
-  ret double %add
-; CHECK: @select1
-; CHECK: select i1 %tobool, double %x, double 0.000000e+00
-}
-
-define double @select2(i32 %cond, double %x, double %y) {
-  %tobool = icmp ne i32 %cond, 0
-  %cond1 = select i1 %tobool, double 0.000000e+00, double 1.000000e+00
-  %mul = fmul nnan nsz double %cond1, %x
-  %add = fadd double %mul, %y
-  ret double %add
-; CHECK: @select2
-; CHECK: select i1 %tobool, double 0.000000e+00, double %x
-}
-
-define double @select3(i32 %cond, double %x, double %y) {
-  %tobool = icmp ne i32 %cond, 0
-  %cond1 = select i1 %tobool, double 0.000000e+00, double 2.000000e+00
-  %mul = fmul nnan nsz double %cond1, %x
-  %add = fadd double %mul, %y
-  ret double %add
-; CHECK: @select3
-; CHECK: fmul nnan nsz double %cond1, %x
-}
-
 ; =========================================================================
 ;
 ;   Testing-cases about fmul begin
@@ -172,7 +141,7 @@ define float @fmul_distribute1(float %f1) {
   %t1 = fmul float %f1, 6.0e+3
   %t2 = fadd float %t1, 2.0e+3
   %t3 = fmul fast float %t2, 5.0e+3
-  ret float %t3 
+  ret float %t3
 ; CHECK: @fmul_distribute1
 ; CHECK: %1 = fmul fast float %f1, 3.000000e+07
 ; CHECK: %t3 = fadd fast float %1, 1.000000e+07
@@ -205,9 +174,9 @@ define double @fmul_distribute3(double %f1) {
 
 ; C1/X * C2 => (C1*C2) / X
 define float @fmul2(float %f1) {
-  %t1 = fdiv float 2.0e+3, %f1 
+  %t1 = fdiv float 2.0e+3, %f1
   %t3 = fmul fast float %t1, 6.0e+3
-  ret float %t3 
+  ret float %t3
 ; CHECK: @fmul2
 ; CHECK: fdiv fast float 1.200000e+07, %f1
 }
@@ -216,7 +185,7 @@ define float @fmul2(float %f1) {
 define float @fmul3(float %f1, float %f2) {
   %t1 = fdiv float %f1, 2.0e+3
   %t3 = fmul fast float %t1, 6.0e+3
-  ret float %t3 
+  ret float %t3
 ; CHECK: @fmul3
 ; CHECK: fmul fast float %f1, 3.000000e+00
 }
@@ -227,21 +196,146 @@ define float @fmul3(float %f1, float %f2) {
 define float @fmul4(float %f1, float %f2) {
   %t1 = fdiv float %f1, 2.0e+3
   %t3 = fmul fast float %t1, 0x3810000000000000
-  ret float %t3 
+  ret float %t3
 ; CHECK: @fmul4
 ; CHECK: fmul fast float %t1, 0x3810000000000000
 }
 
-; X / C1 * C2 => X / (C2/C1) if  C1/C2 is either a special value of a denormal, 
+; X / C1 * C2 => X / (C2/C1) if  C1/C2 is either a special value of a denormal,
 ;  and C2/C1 is a normal value.
-; 
+;
 define float @fmul5(float %f1, float %f2) {
   %t1 = fdiv float %f1, 3.0e+0
   %t3 = fmul fast float %t1, 0x3810000000000000
-  ret float %t3 
+  ret float %t3
 ; CHECK: @fmul5
 ; CHECK: fdiv fast float %f1, 0x47E8000000000000
 }
 
+; (X*Y) * X => (X*X) * Y
+define float @fmul6(float %f1, float %f2) {
+  %mul = fmul float %f1, %f2
+  %mul1 = fmul fast float %mul, %f1
+  ret float %mul1
+; CHECK: @fmul6
+; CHECK: fmul fast float %f1, %f1
+}
+
+; "(X*Y) * X => (X*X) * Y" is disabled if "X*Y" has multiple uses
+define float @fmul7(float %f1, float %f2) {
+  %mul = fmul float %f1, %f2
+  %mul1 = fmul fast float %mul, %f1
+  %add = fadd float %mul1, %mul
+  ret float %add
+; CHECK: @fmul7
+; CHECK: fmul fast float %mul, %f1
+}
+
+; =========================================================================
+;
+;   Testing-cases about negation
+;
+; =========================================================================
+define float @fneg1(float %f1, float %f2) {
+  %sub = fsub float -0.000000e+00, %f1
+  %sub1 = fsub nsz float 0.000000e+00, %f2
+  %mul = fmul float %sub, %sub1
+  ret float %mul
+; CHECK: @fneg1
+; CHECK: fmul float %f1, %f2
+}
+
+; =========================================================================
+;
+;   Testing-cases about div
+;
+; =========================================================================
+
+; X/C1 / C2 => X * (1/(C2*C1))
+define float @fdiv1(float %x) {
+  %div = fdiv float %x, 0x3FF3333340000000
+  %div1 = fdiv fast float %div, 0x4002666660000000
+  ret float %div1
+; 0x3FF3333340000000 = 1.2f
+; 0x4002666660000000 = 2.3f
+; 0x3FD7303B60000000 = 0.36231884057971014492
+; CHECK: @fdiv1
+; CHECK: fmul fast float %x, 0x3FD7303B60000000
+}
+
+; X*C1 / C2 => X * (C1/C2)
+define float @fdiv2(float %x) {
+  %mul = fmul float %x, 0x3FF3333340000000
+  %div1 = fdiv fast float %mul, 0x4002666660000000
+  ret float %div1
+
+; 0x3FF3333340000000 = 1.2f
+; 0x4002666660000000 = 2.3f
+; 0x3FE0B21660000000 = 0.52173918485641479492
+; CHECK: @fdiv2
+; CHECK: fmul fast float %x, 0x3FE0B21660000000
+}
+
+; "X/C1 / C2 => X * (1/(C2*C1))" is disabled (for now) is C2/C1 is a denormal
+;
+define float @fdiv3(float %x) {
+  %div = fdiv float %x, 0x47EFFFFFE0000000
+  %div1 = fdiv fast float %div, 0x4002666660000000
+  ret float %div1
+; CHECK: @fdiv3
+; CHECK: fdiv float %x, 0x47EFFFFFE0000000
+}
+
+; "X*C1 / C2 => X * (C1/C2)" is disabled if C1/C2 is a denormal
+define float @fdiv4(float %x) {
+  %mul = fmul float %x, 0x47EFFFFFE0000000
+  %div = fdiv float %mul, 0x3FC99999A0000000
+  ret float %div
+; CHECK: @fdiv4
+; CHECK: fmul float %x, 0x47EFFFFFE0000000
+}
+
+; (X/Y)/Z = > X/(Y*Z)
+define float @fdiv5(float %f1, float %f2, float %f3) {
+  %t1 = fdiv float %f1, %f2
+  %t2 = fdiv fast float %t1, %f3
+  ret float %t2
+; CHECK: @fdiv5
+; CHECK: fmul float %f2, %f3
+}
 
+; Z/(X/Y) = > (Z*Y)/X
+define float @fdiv6(float %f1, float %f2, float %f3) {
+  %t1 = fdiv float %f1, %f2
+  %t2 = fdiv fast float %f3, %t1
+  ret float %t2
+; CHECK: @fdiv6
+; CHECK: fmul float %f3, %f2
+}
 
+; C1/(X*C2) => (C1/C2) / X
+define float @fdiv7(float %x) {
+  %t1 = fmul float %x, 3.0e0
+  %t2 = fdiv fast float 15.0e0, %t1
+  ret float %t2
+; CHECK: @fdiv7
+; CHECK: fdiv fast float 5.000000e+00, %x
+}
+
+; C1/(X/C2) => (C1*C2) / X
+define float @fdiv8(float %x) {
+  %t1 = fdiv float %x, 3.0e0
+  %t2 = fdiv fast float 15.0e0, %t1
+  ret float %t2
+; CHECK: @fdiv8
+; CHECK: fdiv fast float 4.500000e+01, %x
+}
+
+; C1/(C2/X) => (C1/C2) * X
+define float @fdiv9(float %x) {
+  %t1 = fdiv float 3.0e0, %x
+  %t2 = fdiv fast float 15.0e0, %t1
+  ret float %t2
+; CHECK: @fdiv9
+; CHECK: fmul fast float %x, 5.000000e+00
+}
diff --git a/test/Transforms/InstCombine/fmul.ll b/test/Transforms/InstCombine/fmul.ll
new file mode 100644
index 0000000..3671b4c
--- /dev/null
+++ b/test/Transforms/InstCombine/fmul.ll
@@ -0,0 +1,72 @@
+; RUN: opt -S -instcombine < %s | FileCheck %s
+
+; (-0.0 - X) * C => X * -C
+define float @test1(float %x) {
+  %sub = fsub float -0.000000e+00, %x
+  %mul = fmul float %sub, 2.0e+1
+  ret float %mul
+
+; CHECK: @test1
+; CHECK: fmul float %x, -2.000000e+01
+}
+
+; (0.0 - X) * C => X * -C
+define float @test2(float %x) {
+  %sub = fsub nsz float 0.000000e+00, %x
+  %mul = fmul float %sub, 2.0e+1
+  ret float %mul
+
+; CHECK: @test2
+; CHECK: fmul float %x, -2.000000e+01
+}
+
+; (-0.0 - X) * (-0.0 - Y) => X * Y
+define float @test3(float %x, float %y) {
+  %sub1 = fsub float -0.000000e+00, %x
+  %sub2 = fsub float -0.000000e+00, %y
+  %mul = fmul float %sub1, %sub2
+  ret float %mul
+; CHECK: @test3
+; CHECK: fmul float %x, %y
+}
+
+; (0.0 - X) * (0.0 - Y) => X * Y
+define float @test4(float %x, float %y) {
+  %sub1 = fsub nsz float 0.000000e+00, %x
+  %sub2 = fsub nsz float 0.000000e+00, %y
+  %mul = fmul float %sub1, %sub2
+  ret float %mul
+; CHECK: @test4
+; CHECK: fmul float %x, %y
+}
+
+; (-0.0 - X) * Y => -0.0 - (X * Y)
+define float @test5(float %x, float %y) {
+  %sub1 = fsub float -0.000000e+00, %x
+  %mul = fmul float %sub1, %y
+  ret float %mul
+; CHECK: @test5
+; CHECK: %1 = fmul float %x, %y
+; CHECK: %mul = fsub float -0.000000e+00, %1
+}
+
+; (0.0 - X) * Y => 0.0 - (X * Y)
+define float @test6(float %x, float %y) {
+  %sub1 = fsub nsz float 0.000000e+00, %x
+  %mul = fmul float %sub1, %y
+  ret float %mul
+; CHECK: @test6
+; CHECK: %1 = fmul float %x, %y
+; CHECK: %mul = fsub float -0.000000e+00, %1
+}
+
+; "(-0.0 - X) * Y => -0.0 - (X * Y)" is disabled if expression "-0.0 - X"
+; has multiple uses.
+define float @test7(float %x, float %y) {
+  %sub1 = fsub float -0.000000e+00, %x
+  %mul = fmul float %sub1, %y
+  %mul2 = fmul float %mul, %sub1
+  ret float %mul2
+; CHECK: @test7
+; CHECK: fsub float -0.000000e+00, %x
+}
diff --git a/test/Transforms/InstCombine/fpcast.ll b/test/Transforms/InstCombine/fpcast.ll
index bc6aa0a..09f0532 100644
--- a/test/Transforms/InstCombine/fpcast.ll
+++ b/test/Transforms/InstCombine/fpcast.ll
@@ -13,3 +13,22 @@ define i8 @test2() {
 ; CHECK: ret i8 -1
 }
 
+; CHECK: test3
+define half @test3(float %a) {
+; CHECK: fptrunc
+; CHECK: llvm.fabs.f16
+  %b = call float @llvm.fabs.f32(float %a)
+  %c = fptrunc float %b to half
+  ret half %c
+}
+
+; CHECK: test4
+define half @test4(float %a) {
+; CHECK: fptrunc
+; CHECK: fsub
+  %b = fsub float -0.0, %a
+  %c = fptrunc float %b to half
+  ret half %c
+}
+
+declare float @llvm.fabs.f32(float) nounwind readonly
diff --git a/test/Transforms/InstCombine/getelementptr.ll b/test/Transforms/InstCombine/getelementptr.ll
index 1c120ec..bb07736 100644
--- a/test/Transforms/InstCombine/getelementptr.ll
+++ b/test/Transforms/InstCombine/getelementptr.ll
@@ -424,7 +424,7 @@ define i32 @test35() nounwind {
              i8* getelementptr (%t1* bitcast (%t0* @s to %t1*), i32 0, i32 1, i32 0)) nounwind
   ret i32 0
 ; CHECK: @test35
-; CHECK: call i32 (i8*, ...)* @printf(i8* getelementptr inbounds ([17 x i8]* @"\01LC8", i64 0, i64 0), i8* getelementptr inbounds (%t0* @s, i64 0, i32 1, i64 0)) nounwind
+; CHECK: call i32 (i8*, ...)* @printf(i8* getelementptr inbounds ([17 x i8]* @"\01LC8", i64 0, i64 0), i8* getelementptr inbounds (%t0* @s, i64 0, i32 1, i64 0)) [[NUW:#[0-9]+]]
 }
 
 ; Instcombine should constant-fold the GEP so that indices that have
@@ -492,3 +492,21 @@ define void @three_gep_f(%three_gep_t2* %x) {
 
 declare void @three_gep_g(i32*)
 declare void @three_gep_h(%three_gep_t2*)
+
+%struct.ham = type { i32, %struct.zot*, %struct.zot*, %struct.zot* }
+%struct.zot = type { i64, i8 }
+
+define void @test39(%struct.ham* %arg, i8 %arg1) nounwind {
+  %tmp = getelementptr inbounds %struct.ham* %arg, i64 0, i32 2
+  %tmp2 = load %struct.zot** %tmp, align 8
+  %tmp3 = bitcast %struct.zot* %tmp2 to i8*
+  %tmp4 = getelementptr inbounds i8* %tmp3, i64 -8
+  store i8 %arg1, i8* %tmp4, align 8
+  ret void
+
+; CHECK: @test39
+; CHECK: getelementptr inbounds %struct.ham* %arg, i64 0, i32 2
+; CHECK: getelementptr inbounds i8* %tmp3, i64 -8
+}
+
+; CHECK: attributes [[NUW]] = { nounwind }
diff --git a/test/Transforms/InstCombine/icmp.ll b/test/Transforms/InstCombine/icmp.ll
index 8fb6144..331eb3f 100644
--- a/test/Transforms/InstCombine/icmp.ll
+++ b/test/Transforms/InstCombine/icmp.ll
@@ -706,3 +706,41 @@ define i1 @test69(i32 %c) nounwind uwtable {
   %3 = or i1 %1, %2
   ret i1 %3
 }
+
+; CHECK: @icmp_sext16trunc
+; CHECK-NEXT: %1 = trunc i32 %x to i16
+; CHECK-NEXT: %cmp = icmp slt i16 %1, 36
+define i1 @icmp_sext16trunc(i32 %x) {
+  %trunc = trunc i32 %x to i16
+  %sext = sext i16 %trunc to i32
+  %cmp = icmp slt i32 %sext, 36
+  ret i1 %cmp
+}
+
+; CHECK: @icmp_sext8trunc
+; CHECK-NEXT: %1 = trunc i32 %x to i8
+; CHECK-NEXT: %cmp = icmp slt i8 %1, 36
+define i1 @icmp_sext8trunc(i32 %x) {
+  %trunc = trunc i32 %x to i8
+  %sext = sext i8 %trunc to i32
+  %cmp = icmp slt i32 %sext, 36
+  ret i1 %cmp
+}
+
+; CHECK: @icmp_shl16
+; CHECK-NEXT: %1 = trunc i32 %x to i16
+; CHECK-NEXT: %cmp = icmp slt i16 %1, 36
+define i1 @icmp_shl16(i32 %x) {
+  %shl = shl i32 %x, 16
+  %cmp = icmp slt i32 %shl, 2359296
+  ret i1 %cmp
+}
+
+; CHECK: @icmp_shl24
+; CHECK-NEXT: %1 = trunc i32 %x to i8
+; CHECK-NEXT: %cmp = icmp slt i8 %1, 36
+define i1 @icmp_shl24(i32 %x) {
+  %shl = shl i32 %x, 24
+  %cmp = icmp slt i32 %shl, 603979776
+  ret i1 %cmp
+}
diff --git a/test/Transforms/InstCombine/intrinsics.ll b/test/Transforms/InstCombine/intrinsics.ll
index 93f0a95..f334b3b 100644
--- a/test/Transforms/InstCombine/intrinsics.ll
+++ b/test/Transforms/InstCombine/intrinsics.ll
@@ -220,3 +220,39 @@ define i32 @cttz_simplify1b(i32 %x) nounwind readnone ssp {
 ; CHECK: @cttz_simplify1b
 ; CHECK-NEXT: ret i32 0
 }
+
+define i32 @ctlz_undef(i32 %Value) nounwind {
+  %ctlz = call i32 @llvm.ctlz.i32(i32 0, i1 true)
+  ret i32 %ctlz
+
+; CHECK: @ctlz_undef
+; CHECK-NEXT: ret i32 undef
+}
+
+define i32 @cttz_undef(i32 %Value) nounwind {
+  %cttz = call i32 @llvm.cttz.i32(i32 0, i1 true)
+  ret i32 %cttz
+
+; CHECK: @cttz_undef
+; CHECK-NEXT: ret i32 undef
+}
+
+define i32 @ctlz_select(i32 %Value) nounwind {
+  %tobool = icmp ne i32 %Value, 0
+  %ctlz = call i32 @llvm.ctlz.i32(i32 %Value, i1 true)
+  %s = select i1 %tobool, i32 %ctlz, i32 32
+  ret i32 %s
+
+; CHECK: @ctlz_select
+; CHECK: select i1 %tobool, i32 %ctlz, i32 32
+}
+
+define i32 @cttz_select(i32 %Value) nounwind {
+  %tobool = icmp ne i32 %Value, 0
+  %cttz = call i32 @llvm.cttz.i32(i32 %Value, i1 true)
+  %s = select i1 %tobool, i32 %cttz, i32 32
+  ret i32 %s
+
+; CHECK: @cttz_select
+; CHECK: select i1 %tobool, i32 %cttz, i32 32
+}
diff --git a/test/Transforms/InstCombine/load3.ll b/test/Transforms/InstCombine/load3.ll
index 35398e1..db74426 100644
--- a/test/Transforms/InstCombine/load3.ll
+++ b/test/Transforms/InstCombine/load3.ll
@@ -1,6 +1,6 @@
 ; RUN: opt < %s -instcombine -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"
+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-S128"
+target triple = "i386-apple-macosx10.0.0"
 
 ; Instcombine should be able to do trivial CSE of loads.
 
@@ -24,4 +24,23 @@ define float @test2() {
   
 ; CHECK: @test2
 ; CHECK: ret float 0x3806965600000000
-}
-\ No newline at end of file
+}
+
+@rslts32 = global [36 x i32] zeroinitializer, align 4
+
+@expect32 = internal constant [36 x i32][ i32 1, i32 2, i32 0, i32 100, i32 3,
+i32 4, i32 0, i32 -7, i32 4, i32 4, i32 8, i32 8, i32 1, i32 3, i32 8, i32 3,
+i32 4, i32 -2, i32 2, i32 8, i32 83, i32 77, i32 8, i32 17, i32 77, i32 88, i32
+22, i32 33, i32 44, i32 88, i32 77, i32 4, i32 4, i32 7, i32 -7, i32 -8] ,
+align 4
+
+; PR14986
+define void @test3() nounwind {
+; This is a weird way of computing zero.
+  %l = load i32* getelementptr ([36 x i32]* @expect32, i32 29826161, i32 28), align 4
+  store i32 %l, i32* getelementptr ([36 x i32]* @rslts32, i32 29826161, i32 28), align 4
+  ret void
+
+; CHECK: @test3
+; CHECK: store i32 1, i32* getelementptr inbounds ([36 x i32]* @rslts32, i32 0, i32 0)
+}
diff --git a/test/Transforms/InstCombine/logical-select.ll b/test/Transforms/InstCombine/logical-select.ll
index bb59817..f8c0676 100644
--- a/test/Transforms/InstCombine/logical-select.ll
+++ b/test/Transforms/InstCombine/logical-select.ll
@@ -10,8 +10,8 @@ define i32 @foo(i32 %a, i32 %b, i32 %c, i32 %d) nounwind {
   %j = or i32 %g, %i
   ret i32 %j
 ; CHECK: %e = icmp slt i32 %a, %b
-; CHECK: %j = select i1 %e, i32 %c, i32 %d
-; CHECK: ret i32 %j
+; CHECK-NEXT: [[result:%.*]] = select i1 %e, i32 %c, i32 %d
+; CHECK-NEXT: ret i32 [[result]]
 }
 define i32 @bar(i32 %a, i32 %b, i32 %c, i32 %d) nounwind {
   %e = icmp slt i32 %a, %b
@@ -22,8 +22,8 @@ define i32 @bar(i32 %a, i32 %b, i32 %c, i32 %d) nounwind {
   %j = or i32 %i, %g
   ret i32 %j
 ; CHECK: %e = icmp slt i32 %a, %b
-; CHECK: %j = select i1 %e, i32 %c, i32 %d
-; CHECK: ret i32 %j
+; CHECK-NEXT: [[result:%.*]] = select i1 %e, i32 %c, i32 %d
+; CHECK-NEXT: ret i32 [[result]]
 }
 
 define i32 @goo(i32 %a, i32 %b, i32 %c, i32 %d) nounwind {
@@ -36,8 +36,8 @@ entry:
   %3 = or i32 %1, %2
   ret i32 %3
 ; CHECK: %0 = icmp slt i32 %a, %b
-; CHECK: %1 = select i1 %0, i32 %c, i32 %d
-; CHECK: ret i32 %1
+; CHECK-NEXT: [[result:%.*]] = select i1 %0, i32 %c, i32 %d
+; CHECK-NEXT: ret i32 [[result]]
 }
 define i32 @poo(i32 %a, i32 %b, i32 %c, i32 %d) nounwind {
 entry:
@@ -49,8 +49,8 @@ entry:
   %3 = or i32 %1, %2
   ret i32 %3
 ; CHECK: %0 = icmp slt i32 %a, %b
-; CHECK: %1 = select i1 %0, i32 %c, i32 %d
-; CHECK: ret i32 %1
+; CHECK-NEXT: [[result:%.*]] = select i1 %0, i32 %c, i32 %d
+; CHECK-NEXT: ret i32 [[result]]
 }
 
 define i32 @par(i32 %a, i32 %b, i32 %c, i32 %d) nounwind {
@@ -63,6 +63,6 @@ entry:
   %3 = or i32 %1, %2
   ret i32 %3
 ; CHECK: %0 = icmp slt i32 %a, %b
-; CHECK: %1 = select i1 %0, i32 %c, i32 %d
-; CHECK: ret i32 %1
+; CHECK-NEXT: [[result:%.*]] = select i1 %0, i32 %c, i32 %d
+; CHECK-NEXT: ret i32 [[result]]
 }
diff --git a/test/Transforms/InstCombine/mul.ll b/test/Transforms/InstCombine/mul.ll
index bbc70fe..16213b8 100644
--- a/test/Transforms/InstCombine/mul.ll
+++ b/test/Transforms/InstCombine/mul.ll
@@ -138,10 +138,8 @@ define i32 @test16(i32 %b, i1 %c) {
         ; e = b & (a >> 31)
         %e = mul i32 %d, %b             ; <i32> [#uses=1]
         ret i32 %e
-; CHECK: [[TEST16:%.*]] = zext i1 %c to i32
-; CHECK-NEXT: %1 = sub i32 0, [[TEST16]]
-; CHECK-NEXT: %e = and i32 %1, %b
-; CHECK-NEXT: ret i32 %e
+; CHECK: [[TEST16:%.*]] = select i1 %c, i32 %b, i32 0
+; CHECK-NEXT: ret i32 [[TEST16]]
 }
 
 ; X * Y (when Y is 0 or 1) --> x & (0-Y)
diff --git a/test/Transforms/InstCombine/objsize.ll b/test/Transforms/InstCombine/objsize.ll
index 0ead9d1..31a3cb4 100644
--- a/test/Transforms/InstCombine/objsize.ll
+++ b/test/Transforms/InstCombine/objsize.ll
@@ -256,131 +256,3 @@ xpto:
 return:
   ret i32 7
 }
-
-declare noalias i8* @valloc(i32) nounwind
-
-; CHECK: @test14
-; CHECK: ret i32 6
-define i32 @test14(i32 %a) nounwind {
-  switch i32 %a, label %sw.default [
-    i32 1, label %sw.bb
-    i32 2, label %sw.bb1
-  ]
-
-sw.bb:
-  %call = tail call noalias i8* @malloc(i32 6) nounwind
-  br label %sw.epilog
-
-sw.bb1:
-  %call2 = tail call noalias i8* @calloc(i32 3, i32 2) nounwind
-  br label %sw.epilog
-
-sw.default:
-  %call3 = tail call noalias i8* @valloc(i32 6) nounwind
-  br label %sw.epilog
-
-sw.epilog:
-  %b.0 = phi i8* [ %call3, %sw.default ], [ %call2, %sw.bb1 ], [ %call, %sw.bb ]
-  %1 = tail call i32 @llvm.objectsize.i32(i8* %b.0, i1 false)
-  ret i32 %1
-}
-
-; CHECK: @test15
-; CHECK: llvm.objectsize
-define i32 @test15(i32 %a) nounwind {
-  switch i32 %a, label %sw.default [
-    i32 1, label %sw.bb
-    i32 2, label %sw.bb1
-  ]
-
-sw.bb:
-  %call = tail call noalias i8* @malloc(i32 3) nounwind
-  br label %sw.epilog
-
-sw.bb1:
-  %call2 = tail call noalias i8* @calloc(i32 2, i32 1) nounwind
-  br label %sw.epilog
-
-sw.default:
-  %call3 = tail call noalias i8* @valloc(i32 3) nounwind
-  br label %sw.epilog
-
-sw.epilog:
-  %b.0 = phi i8* [ %call3, %sw.default ], [ %call2, %sw.bb1 ], [ %call, %sw.bb ]
-  %1 = tail call i32 @llvm.objectsize.i32(i8* %b.0, i1 false)
-  ret i32 %1
-}
-
-; CHECK: @test16
-; CHECK: llvm.objectsize
-define i32 @test16(i8* %a, i32 %n) nounwind {
-  %b = alloca [5 x i8], align 1
-  %c = alloca [5 x i8], align 1
-  switch i32 %n, label %sw.default [
-    i32 1, label %sw.bb
-    i32 2, label %sw.bb1
-  ]
-
-sw.bb:
-  %bp = bitcast [5 x i8]* %b to i8*
-  br label %sw.epilog
-
-sw.bb1:
-  %cp = bitcast [5 x i8]* %c to i8*
-  br label %sw.epilog
-
-sw.default:
-  br label %sw.epilog
-
-sw.epilog:
-  %phi = phi i8* [ %a, %sw.default ], [ %cp, %sw.bb1 ], [ %bp, %sw.bb ]
-  %sz = call i32 @llvm.objectsize.i32(i8* %phi, i1 false)
-  ret i32 %sz
-}
-
-; CHECK: @test17
-; CHECK: ret i32 5
-define i32 @test17(i32 %n) nounwind {
-  %b = alloca [5 x i8], align 1
-  %c = alloca [5 x i8], align 1
-  %bp = bitcast [5 x i8]* %b to i8*
-  switch i32 %n, label %sw.default [
-    i32 1, label %sw.bb
-    i32 2, label %sw.bb1
-  ]
-
-sw.bb:
-  br label %sw.epilog
-
-sw.bb1:
-  %cp = bitcast [5 x i8]* %c to i8*
-  br label %sw.epilog
-
-sw.default:
-  br label %sw.epilog
-
-sw.epilog:
-  %phi = phi i8* [ %bp, %sw.default ], [ %cp, %sw.bb1 ], [ %bp, %sw.bb ]
-  %sz = call i32 @llvm.objectsize.i32(i8* %phi, i1 false)
-  ret i32 %sz
-}
-
-@globalalias = alias internal [60 x i8]* @a
-
-; CHECK: @test18
-; CHECK-NEXT: ret i32 60
-define i32 @test18() {
-  %bc = bitcast [60 x i8]* @globalalias to i8*
-  %1 = call i32 @llvm.objectsize.i32(i8* %bc, i1 false)
-  ret i32 %1
-}
-
-@globalalias2 = alias weak [60 x i8]* @a
-
-; CHECK: @test19
-; CHECK: llvm.objectsize
-define i32 @test19() {
-  %bc = bitcast [60 x i8]* @globalalias2 to i8*
-  %1 = call i32 @llvm.objectsize.i32(i8* %bc, i1 false)
-  ret i32 %1
-}
diff --git a/test/Transforms/InstCombine/or.ll b/test/Transforms/InstCombine/or.ll
index c0bb28d..bde2a54 100644
--- a/test/Transforms/InstCombine/or.ll
+++ b/test/Transforms/InstCombine/or.ll
@@ -344,10 +344,9 @@ define <4 x i32> @test32(<4 x i1> %and.i1352, <4 x i32> %vecinit6.i176, <4 x i32
   %and.i = and <4 x i32> %vecinit6.i191, %neg.i   ; <<4 x i32>> [#uses=1]
   %or.i = or <4 x i32> %and.i, %and.i129          ; <<4 x i32>> [#uses=1]
   ret <4 x i32> %or.i
-; Don't turn this into a vector select until codegen matures to handle them
-; better.
+; codegen is mature enough to handle vector selects.
 ; CHECK: @test32
-; CHECK: or <4 x i32> %and.i, %and.i129
+; CHECK: select <4 x i1> %and.i1352, <4 x i32> %vecinit6.i176, <4 x i32> %vecinit6.i191
 }
 
 define i1 @test33(i1 %X, i1 %Y) {
diff --git a/test/Transforms/InstCombine/pow-1.ll b/test/Transforms/InstCombine/pow-1.ll
index c8e5f38..8a311f0 100644
--- a/test/Transforms/InstCombine/pow-1.ll
+++ b/test/Transforms/InstCombine/pow-1.ll
@@ -30,7 +30,7 @@ define double @test_simplify2(double %x) {
 define float @test_simplify3(float %x) {
 ; CHECK: @test_simplify3
   %retval = call float @powf(float 2.0, float %x)
-; CHECK-NEXT: [[EXP2F:%[a-z0-9]+]] = call float @exp2f(float %x) nounwind readonly
+; CHECK-NEXT: [[EXP2F:%[a-z0-9]+]] = call float @exp2f(float %x) [[NUW_RO:#[0-9]+]]
   ret float %retval
 ; CHECK-NEXT: ret float [[EXP2F]]
 }
@@ -38,7 +38,7 @@ define float @test_simplify3(float %x) {
 define double @test_simplify4(double %x) {
 ; CHECK: @test_simplify4
   %retval = call double @pow(double 2.0, double %x)
-; CHECK-NEXT: [[EXP2:%[a-z0-9]+]] = call double @exp2(double %x) nounwind readonly
+; CHECK-NEXT: [[EXP2:%[a-z0-9]+]] = call double @exp2(double %x) [[NUW_RO]]
   ret double %retval
 ; CHECK-NEXT: ret double [[EXP2]]
 }
@@ -64,8 +64,8 @@ define double @test_simplify6(double %x) {
 define float @test_simplify7(float %x) {
 ; CHECK: @test_simplify7
   %retval = call float @powf(float %x, float 0.5)
-; CHECK-NEXT: [[SQRTF:%[a-z0-9]+]] = call float @sqrtf(float %x) nounwind readonly
-; CHECK-NEXT: [[FABSF:%[a-z0-9]+]] = call float @fabsf(float [[SQRTF]]) nounwind readonly
+; CHECK-NEXT: [[SQRTF:%[a-z0-9]+]] = call float @sqrtf(float %x) [[NUW_RO]]
+; CHECK-NEXT: [[FABSF:%[a-z0-9]+]] = call float @fabsf(float [[SQRTF]]) [[NUW_RO]]
 ; CHECK-NEXT: [[FCMP:%[a-z0-9]+]] = fcmp oeq float %x, 0xFFF0000000000000
 ; CHECK-NEXT: [[SELECT:%[a-z0-9]+]] = select i1 [[FCMP]], float 0x7FF0000000000000, float [[FABSF]]
   ret float %retval
@@ -75,8 +75,8 @@ define float @test_simplify7(float %x) {
 define double @test_simplify8(double %x) {
 ; CHECK: @test_simplify8
   %retval = call double @pow(double %x, double 0.5)
-; CHECK-NEXT: [[SQRT:%[a-z0-9]+]] = call double @sqrt(double %x) nounwind readonly
-; CHECK-NEXT: [[FABS:%[a-z0-9]+]] = call double @fabs(double [[SQRT]]) nounwind readonly
+; CHECK-NEXT: [[SQRT:%[a-z0-9]+]] = call double @sqrt(double %x) [[NUW_RO]]
+; CHECK-NEXT: [[FABS:%[a-z0-9]+]] = call double @fabs(double [[SQRT]]) [[NUW_RO]]
 ; CHECK-NEXT: [[FCMP:%[a-z0-9]+]] = fcmp oeq double %x, 0xFFF0000000000000
 ; CHECK-NEXT: [[SELECT:%[a-z0-9]+]] = select i1 [[FCMP]], double 0x7FF0000000000000, double [[FABS]]
   ret double %retval
@@ -150,3 +150,5 @@ define double @test_simplify16(double %x) {
   ret double %retval
 ; CHECK-NEXT: ret double [[RECIPROCAL]]
 }
+
+; CHECK: attributes [[NUW_RO]] = { nounwind readonly }
diff --git a/test/Transforms/InstCombine/ptr-int-cast.ll b/test/Transforms/InstCombine/ptr-int-cast.ll
index 9524d44..7a6ecff 100644
--- a/test/Transforms/InstCombine/ptr-int-cast.ll
+++ b/test/Transforms/InstCombine/ptr-int-cast.ll
@@ -27,3 +27,34 @@ define i64 @f0(i32 %a0) nounwind {
        ret i64 %t1
 }
 
+define <4 x i32> @test4(<4 x i8*> %arg) nounwind {
+; CHECK: @test4
+; CHECK: ptrtoint <4 x i8*> %arg to <4 x i64>
+; CHECK: trunc <4 x i64> %1 to <4 x i32>
+  %p1 = ptrtoint <4 x i8*> %arg to <4 x i32>
+  ret <4 x i32> %p1
+}
+
+define <4 x i128> @test5(<4 x i8*> %arg) nounwind {
+; CHECK: @test5
+; CHECK: ptrtoint <4 x i8*> %arg to <4 x i64>
+; CHECK: zext <4 x i64> %1 to <4 x i128>
+  %p1 = ptrtoint <4 x i8*> %arg to <4 x i128>
+  ret <4 x i128> %p1
+}
+
+define <4 x i8*> @test6(<4 x i32> %arg) nounwind {
+; CHECK: @test6
+; CHECK: zext <4 x i32> %arg to <4 x i64>
+; CHECK: inttoptr <4 x i64> %1 to <4 x i8*>
+  %p1 = inttoptr <4 x i32> %arg to <4 x i8*>
+  ret <4 x i8*> %p1
+}
+
+define <4 x i8*> @test7(<4 x i128> %arg) nounwind {
+; CHECK: @test7
+; CHECK: trunc <4 x i128> %arg to <4 x i64>
+; CHECK: inttoptr <4 x i64> %1 to <4 x i8*>
+  %p1 = inttoptr <4 x i128> %arg to <4 x i8*>
+  ret <4 x i8*> %p1
+}
diff --git a/test/Transforms/InstCombine/sext.ll b/test/Transforms/InstCombine/sext.ll
index f198797..968f37c 100644
--- a/test/Transforms/InstCombine/sext.ll
+++ b/test/Transforms/InstCombine/sext.ll
@@ -184,3 +184,12 @@ define i32 @test16(i16 %x) nounwind {
 ; CHECK-NEXT: %ext = sext i16 %sext to i32
 ; CHECK-NEXT: ret i32 %ext
 }
+
+define i32 @test17(i1 %x) nounwind {
+  %c1 = sext i1 %x to i32
+  %c2 = sub i32 0, %c1
+  ret i32 %c2
+; CHECK: @test17
+; CHECK-NEXT: [[TEST17:%.*]] = zext i1 %x to i32
+; CHECK-NEXT: ret i32 [[TEST17]]
+}
diff --git a/test/Transforms/InstCombine/vec_extract_elt.ll b/test/Transforms/InstCombine/vec_extract_elt.ll
index 63e4ee2..166066a 100644
--- a/test/Transforms/InstCombine/vec_extract_elt.ll
+++ b/test/Transforms/InstCombine/vec_extract_elt.ll
@@ -7,3 +7,13 @@ define i32 @test(float %f) {
         ret i32 %tmp19
 }
 
+define i64 @test2(i64 %in) {
+  %vec = insertelement <8 x i64> undef, i64 %in, i32 0
+  %splat = shufflevector <8 x i64> %vec, <8 x i64> undef, <8 x i32> zeroinitializer
+  %add = add <8 x i64> %splat, <i64 0, i64 1, i64 2, i64 3, i64 4, i64 5, i64 6, i64 7>
+  %scl1 = extractelement <8 x i64> %add, i32 0
+  %scl2 = extractelement <8 x i64> %add, i32 0
+  %r = add i64 %scl1, %scl2
+  ret i64 %r
+}
+
diff --git a/test/Transforms/InstCombine/vector-casts.ll b/test/Transforms/InstCombine/vector-casts.ll
index 7bbf53c..2f2990b 100644
--- a/test/Transforms/InstCombine/vector-casts.ll
+++ b/test/Transforms/InstCombine/vector-casts.ll
@@ -64,7 +64,8 @@ entry:
         
 ; CHECK: @test5
 ; CHECK:   sext <4 x i1> %cmp to <4 x i32>	
-; CHECK:   sext <4 x i1> %cmp4 to <4 x i32>	
+; The sext-and pair is canonicalized to a select.
+; CHECK:   select <4 x i1> %cmp4, <4 x i32>	%sext, <4 x i32> zeroinitializer
 }
 
 
diff --git a/test/Transforms/InstCombine/zext-bool-add-sub.ll b/test/Transforms/InstCombine/zext-bool-add-sub.ll
index 78bcedb..b531057 100644
--- a/test/Transforms/InstCombine/zext-bool-add-sub.ll
+++ b/test/Transforms/InstCombine/zext-bool-add-sub.ll
@@ -4,9 +4,9 @@
 define i32 @a(i1 zeroext %x, i1 zeroext %y) {
 entry:
 ; CHECK: @a
-; CHECK: [[TMP1:%.*]] = zext i1 %y to i32
+; CHECK: [[TMP1:%.*]] = sext i1 %y to i32
 ; CHECK: [[TMP2:%.*]] = select i1 %x, i32 2, i32 1
-; CHECK-NEXT: sub i32 [[TMP2]], [[TMP1]]
+; CHECK-NEXT: add i32 [[TMP2]], [[TMP1]]
   %conv = zext i1 %x to i32
   %conv3 = zext i1 %y to i32
   %conv3.neg = sub i32 0, %conv3
diff --git a/test/Transforms/InstSimplify/call-callconv.ll b/test/Transforms/InstSimplify/call-callconv.ll
new file mode 100644
index 0000000..e475be7
--- /dev/null
+++ b/test/Transforms/InstSimplify/call-callconv.ll
@@ -0,0 +1,48 @@
+; RUN: opt < %s -instcombine -S | FileCheck %s
+; Verify that the non-default calling conv doesn't prevent the libcall simplification
+
+@.str = private unnamed_addr constant [4 x i8] c"abc\00", align 1
+
+define arm_aapcscc i32 @_abs(i32 %i) nounwind readnone {
+; CHECK: _abs
+  %call = tail call arm_aapcscc i32 @abs(i32 %i) nounwind readnone
+  ret i32 %call
+; CHECK: %[[ISPOS:.*]] = icmp sgt i32 %i, -1
+; CHECK: %[[NEG:.*]] = sub i32 0, %i
+; CHECK: %[[RET:.*]] = select i1 %[[ISPOS]], i32 %i, i32 %[[NEG]]
+; CHECK: ret i32 %[[RET]]
+}
+
+declare arm_aapcscc i32 @abs(i32) nounwind readnone
+
+define arm_aapcscc i32 @_labs(i32 %i) nounwind readnone {
+; CHECK: _labs
+  %call = tail call arm_aapcscc i32 @labs(i32 %i) nounwind readnone
+  ret i32 %call
+; CHECK: %[[ISPOS:.*]] = icmp sgt i32 %i, -1
+; CHECK: %[[NEG:.*]] = sub i32 0, %i
+; CHECK: %[[RET:.*]] = select i1 %[[ISPOS]], i32 %i, i32 %[[NEG]]
+; CHECK: ret i32 %[[RET]]
+}
+
+declare arm_aapcscc i32 @labs(i32) nounwind readnone
+
+define arm_aapcscc i32 @_strlen1() {
+; CHECK: _strlen1
+  %call = tail call arm_aapcscc i32 @strlen(i8* getelementptr inbounds ([4 x i8]* @.str, i32 0, i32 0))
+  ret i32 %call
+; CHECK: ret i32 3
+}
+
+declare arm_aapcscc i32 @strlen(i8*)
+
+define arm_aapcscc zeroext i1 @_strlen2(i8* %str) {
+; CHECK: _strlen2
+  %call = tail call arm_aapcscc i32 @strlen(i8* %str)
+  %cmp = icmp ne i32 %call, 0
+  ret i1 %cmp
+
+; CHECK: %[[STRLENFIRST:.*]] = load i8* %str
+; CHECK: %[[CMP:.*]] = icmp ne i8 %[[STRLENFIRST]], 0
+; CHECK: ret i1 %[[CMP]]
+}
diff --git a/test/Transforms/InstSimplify/call.ll b/test/Transforms/InstSimplify/call.ll
index 1a8d0c2..cf2f847 100644
--- a/test/Transforms/InstSimplify/call.ll
+++ b/test/Transforms/InstSimplify/call.ll
@@ -50,3 +50,54 @@ define float @test_fabs_libcall() {
   ret float %x
 ; CHECK-NEXT: ret float 4.2{{0+}}e+01
 }
+
+
+declare float @llvm.fabs.f32(float) nounwind readnone
+declare float @llvm.floor.f32(float) nounwind readnone
+declare float @llvm.ceil.f32(float) nounwind readnone
+declare float @llvm.trunc.f32(float) nounwind readnone
+declare float @llvm.rint.f32(float) nounwind readnone
+declare float @llvm.nearbyint.f32(float) nounwind readnone
+
+; Test idempotent intrinsics
+define float @test_idempotence(float %a) {
+; CHECK: @test_idempotence
+
+; CHECK: fabs
+; CHECK-NOT: fabs
+  %a0 = call float @llvm.fabs.f32(float %a)
+  %a1 = call float @llvm.fabs.f32(float %a0)
+
+; CHECK: floor
+; CHECK-NOT: floor
+  %b0 = call float @llvm.floor.f32(float %a)
+  %b1 = call float @llvm.floor.f32(float %b0)
+
+; CHECK: ceil
+; CHECK-NOT: ceil
+  %c0 = call float @llvm.ceil.f32(float %a)
+  %c1 = call float @llvm.ceil.f32(float %c0)
+
+; CHECK: trunc
+; CHECK-NOT: trunc
+  %d0 = call float @llvm.trunc.f32(float %a)
+  %d1 = call float @llvm.trunc.f32(float %d0)
+
+; CHECK: rint
+; CHECK-NOT: rint
+  %e0 = call float @llvm.rint.f32(float %a)
+  %e1 = call float @llvm.rint.f32(float %e0)
+
+; CHECK: nearbyint
+; CHECK-NOT: nearbyint
+  %f0 = call float @llvm.nearbyint.f32(float %a)
+  %f1 = call float @llvm.nearbyint.f32(float %f0)
+
+  %r0 = fadd float %a1, %b1
+  %r1 = fadd float %r0, %c1
+  %r2 = fadd float %r1, %d1
+  %r3 = fadd float %r2, %e1
+  %r4 = fadd float %r3, %f1
+
+  ret float %r4
+}
diff --git a/test/Transforms/InstSimplify/compare.ll b/test/Transforms/InstSimplify/compare.ll
index 56627b9..0ecfb1f 100644
--- a/test/Transforms/InstSimplify/compare.ll
+++ b/test/Transforms/InstSimplify/compare.ll
@@ -647,3 +647,38 @@ unreachableblock:
   %Y = icmp eq i32* %X, null
   ret i1 %Y
 }
+
+; It's not valid to fold a comparison of an argument with an alloca, even though
+; that's tempting. An argument can't *alias* an alloca, however the aliasing rule
+; relies on restrictions against guessing an object's address and dereferencing.
+; There are no restrictions against guessing an object's address and comparing.
+
+define i1 @alloca_argument_compare(i64* %arg) {
+  %alloc = alloca i64
+  %cmp = icmp eq i64* %arg, %alloc
+  ret i1 %cmp
+  ; CHECK: alloca_argument_compare
+  ; CHECK: ret i1 %cmp
+}
+
+; As above, but with the operands reversed.
+
+define i1 @alloca_argument_compare_swapped(i64* %arg) {
+  %alloc = alloca i64
+  %cmp = icmp eq i64* %alloc, %arg
+  ret i1 %cmp
+  ; CHECK: alloca_argument_compare_swapped
+  ; CHECK: ret i1 %cmp
+}
+
+; Don't assume that a noalias argument isn't equal to a global variable's
+; address. This is an example where AliasAnalysis' NoAlias concept is
+; different from actual pointer inequality.
+
+@y = external global i32
+define zeroext i1 @external_compare(i32* noalias %x) {
+  %cmp = icmp eq i32* %x, @y
+  ret i1 %cmp
+  ; CHECK: external_compare
+  ; CHECK: ret i1 %cmp
+}
diff --git a/test/Transforms/InstSimplify/past-the-end.ll b/test/Transforms/InstSimplify/past-the-end.ll
new file mode 100644
index 0000000..075da4a
--- /dev/null
+++ b/test/Transforms/InstSimplify/past-the-end.ll
@@ -0,0 +1,77 @@
+; RUN: opt < %s -instsimplify -S | FileCheck %s
+target datalayout = "p:32:32"
+
+; Check some past-the-end subtleties.
+
+@opte_a = global i32 0
+@opte_b = global i32 0
+
+; Comparing base addresses of two distinct globals. Never equal.
+
+define zeroext i1 @no_offsets() {
+  %t = icmp eq i32* @opte_a, @opte_b
+  ret i1 %t
+  ; CHECK: no_offsets(
+  ; CHECK: ret i1 false
+}
+
+; Comparing past-the-end addresses of two distinct globals. Never equal.
+
+define zeroext i1 @both_past_the_end() {
+  %x = getelementptr i32* @opte_a, i32 1
+  %y = getelementptr i32* @opte_b, i32 1
+  %t = icmp eq i32* %x, %y
+  ret i1 %t
+  ; CHECK: both_past_the_end(
+  ; CHECK-NOT: ret i1 true
+  ; TODO: refine this
+}
+
+; Comparing past-the-end addresses of one global to the base address
+; of another. Can't fold this.
+
+define zeroext i1 @just_one_past_the_end() {
+  %x = getelementptr i32* @opte_a, i32 1
+  %t = icmp eq i32* %x, @opte_b
+  ret i1 %t
+  ; CHECK: just_one_past_the_end(
+  ; CHECK: ret i1 icmp eq (i32* getelementptr inbounds (i32* @opte_a, i32 1), i32* @opte_b)
+}
+
+; Comparing base addresses of two distinct allocas. Never equal.
+
+define zeroext i1 @no_alloca_offsets() {
+  %m = alloca i32
+  %n = alloca i32
+  %t = icmp eq i32* %m, %n
+  ret i1 %t
+  ; CHECK: no_alloca_offsets(
+  ; CHECK: ret i1 false
+}
+
+; Comparing past-the-end addresses of two distinct allocas. Never equal.
+
+define zeroext i1 @both_past_the_end_alloca() {
+  %m = alloca i32
+  %n = alloca i32
+  %x = getelementptr i32* %m, i32 1
+  %y = getelementptr i32* %n, i32 1
+  %t = icmp eq i32* %x, %y
+  ret i1 %t
+  ; CHECK: both_past_the_end_alloca(
+  ; CHECK-NOT: ret i1 true
+  ; TODO: refine this
+}
+
+; Comparing past-the-end addresses of one alloca to the base address
+; of another. Can't fold this.
+
+define zeroext i1 @just_one_past_the_end_alloca() {
+  %m = alloca i32
+  %n = alloca i32
+  %x = getelementptr i32* %m, i32 1
+  %t = icmp eq i32* %x, %n
+  ret i1 %t
+  ; CHECK: just_one_past_the_end_alloca(
+  ; CHECK: ret i1 %t
+}
diff --git a/test/Transforms/InstSimplify/ptr_diff.ll b/test/Transforms/InstSimplify/ptr_diff.ll
index 1eb1fd4..8b4aa79 100644
--- a/test/Transforms/InstSimplify/ptr_diff.ll
+++ b/test/Transforms/InstSimplify/ptr_diff.ll
@@ -46,3 +46,33 @@ define i64 @ptrdiff3(i8* %ptr) {
   %diff = sub i64 %last.int, %first.int
   ret i64 %diff
 }
+
+define <4 x i32> @ptrdiff4(<4 x i8*> %arg) nounwind {
+; Handle simple cases of vectors of pointers.
+; CHECK: @ptrdiff4
+; CHECK: ret <4 x i32> zeroinitializer
+  %p1 = ptrtoint <4 x i8*> %arg to <4 x i32>
+  %bc = bitcast <4 x i8*> %arg to <4 x i32*>
+  %p2 = ptrtoint <4 x i32*> %bc to <4 x i32>
+  %sub = sub <4 x i32> %p1, %p2
+  ret <4 x i32> %sub
+}
+
+%struct.ham = type { i32, [2 x [2 x i32]] }
+
+@global = internal global %struct.ham zeroinitializer, align 4
+
+define i32 @ptrdiff5() nounwind {
+bb:
+  %tmp = getelementptr inbounds %struct.ham* @global, i32 0, i32 1
+  %tmp1 = getelementptr inbounds [2 x [2 x i32]]* %tmp, i32 0, i32 0
+  %tmp2 = bitcast [2 x i32]* %tmp1 to i32*
+  %tmp3 = ptrtoint i32* %tmp2 to i32
+  %tmp4 = getelementptr inbounds %struct.ham* @global, i32 0, i32 1
+  %tmp5 = getelementptr inbounds [2 x [2 x i32]]* %tmp4, i32 0, i32 0
+  %tmp6 = ptrtoint [2 x i32]* %tmp5 to i32
+  %tmp7 = sub i32 %tmp3, %tmp6
+  ret i32 %tmp7
+; CHECK: @ptrdiff5
+; CHECK: ret i32 0
+}
diff --git a/test/Transforms/JumpThreading/basic.ll b/test/Transforms/JumpThreading/basic.ll
index 93fa29b..fe3dc77 100644
--- a/test/Transforms/JumpThreading/basic.ll
+++ b/test/Transforms/JumpThreading/basic.ll
@@ -497,8 +497,8 @@ l2:
   br label %l3
 
 l3:
-; CHECK: call void @g() noduplicate
-; CHECK-NOT: call void @g() noduplicate
+; CHECK: call void @g() [[NOD:#[0-9]+]]
+; CHECK-NOT: call void @g() [[NOD]]
   call void @g() noduplicate
   %y = icmp ult i32 %p, 5
   br i1 %y, label %l4, label %l5
@@ -512,3 +512,5 @@ l5:
   ret void
 ; CHECK: }
 }
+
+; CHECK: attributes [[NOD]] = { noduplicate }
diff --git a/test/Transforms/LICM/2003-12-11-SinkingToPHI.ll b/test/Transforms/LICM/2003-12-11-SinkingToPHI.ll
index fe8d445..2bf2604 100644
--- a/test/Transforms/LICM/2003-12-11-SinkingToPHI.ll
+++ b/test/Transforms/LICM/2003-12-11-SinkingToPHI.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s -licm | lli %defaultjit
+; RUN: opt < %s -licm | lli -force-interpreter
 
 define i32 @main() {
 entry:
diff --git a/test/Transforms/LICM/hoisting.ll b/test/Transforms/LICM/hoisting.ll
index 98f9334..1ca377e 100644
--- a/test/Transforms/LICM/hoisting.ll
+++ b/test/Transforms/LICM/hoisting.ll
@@ -90,3 +90,29 @@ for.end:                                          ; preds = %for.body
 
 declare void @foo_may_call_exit(i32)
 
+; PR14854
+; CHECK: @test5
+; CHECK: extractvalue
+; CHECK: br label %tailrecurse
+; CHECK: tailrecurse:
+; CHECK: ifend:
+; CHECK: insertvalue
+define { i32*, i32 } @test5(i32 %i, { i32*, i32 } %e) {
+entry:
+  br label %tailrecurse
+
+tailrecurse:                                      ; preds = %then, %entry
+  %i.tr = phi i32 [ %i, %entry ], [ %cmp2, %then ]
+  %out = extractvalue { i32*, i32 } %e, 1
+  %d = insertvalue { i32*, i32 } %e, i32* null, 0
+  %cmp1 = icmp sgt i32 %out, %i.tr
+  br i1 %cmp1, label %then, label %ifend
+
+then:                                             ; preds = %tailrecurse
+  call void @foo()
+  %cmp2 = add i32 %i.tr, 1
+  br label %tailrecurse
+
+ifend:                                            ; preds = %tailrecurse
+  ret { i32*, i32 } %d
+}
diff --git a/test/Transforms/LoopDeletion/simplify-then-delete.ll b/test/Transforms/LoopDeletion/simplify-then-delete.ll
index 5a21672..4278ef1 100644
--- a/test/Transforms/LoopDeletion/simplify-then-delete.ll
+++ b/test/Transforms/LoopDeletion/simplify-then-delete.ll
@@ -4,7 +4,7 @@
 ; Indvars and loop deletion should be able to eliminate all looping
 ; in this testcase.
 
-; CHECK:      define i32 @pmat(i32 %m, i32 %n, double* %y) nounwind {
+; CHECK:      define i32 @pmat(i32 %m, i32 %n, double* %y) #0 {
 ; CHECK-NEXT: entry:
 ; CHECK-NEXT:   ret i32 0
 ; CHECK-NEXT: }
@@ -63,3 +63,5 @@ w.e:
 w.e12:
   ret i32 0
 }
+
+; CHECK: attributes #0 = { nounwind }
diff --git a/test/Transforms/LoopIdiom/X86/popcnt.ll b/test/Transforms/LoopIdiom/X86/popcnt.ll
index 2f458fb..25df93d 100644
--- a/test/Transforms/LoopIdiom/X86/popcnt.ll
+++ b/test/Transforms/LoopIdiom/X86/popcnt.ll
@@ -118,3 +118,23 @@ while.end:                                        ; preds = %while.body, %entry
   %c.0.lcssa = phi i32 [ 0, %entry ], [ %inc, %while.body ]
   ret i32 %c.0.lcssa
 }
+
+define i32 @PopCntCrash3(i64 %a, i32 %x) {
+entry:
+  %tobool3 = icmp eq i64 %a, 0
+  %cmp = icmp eq i32 %x, 0
+  br i1 %tobool3, label %while.end, label %while.body
+
+while.body:                                       ; preds = %entry, %while.body
+  %c.05 = phi i32 [ %inc, %while.body ], [ 0, %entry ]
+  %a.addr.04 = phi i64 [ %and, %while.body ], [ %a, %entry ]
+  %inc = add nsw i32 %c.05, 1
+  %sub = add i64 %a.addr.04, -1
+  %and = and i64 %sub, %a.addr.04
+  %tobool = icmp eq i64 %and, 0
+  br i1 %cmp, label %while.end, label %while.body
+
+while.end:                                        ; preds = %while.body, %entry
+  %c.0.lcssa = phi i32 [ 0, %entry ], [ %inc, %while.body ]
+  ret i32 %c.0.lcssa
+}
diff --git a/test/Transforms/LoopStrengthReduce/2012-07-18-LimitReassociate.ll b/test/Transforms/LoopStrengthReduce/2012-07-18-LimitReassociate.ll
index 53da462..9524be3 100644
--- a/test/Transforms/LoopStrengthReduce/2012-07-18-LimitReassociate.ll
+++ b/test/Transforms/LoopStrengthReduce/2012-07-18-LimitReassociate.ll
@@ -5,7 +5,7 @@
 ; PR13361: LSR + SCEV "hangs" on reasonably sized test with sequence of loops
 ;
 ; Without limits on CollectSubexpr, we have thousands of formulae for
-; the use that crosses loops. With limits we have six.
+; the use that crosses loops. With limits we have five.
 ; CHECK: LSR on loop %bb221:
 ; CHECK: After generating reuse formulae:
 ; CHECK: LSR is examining the following uses:
@@ -15,11 +15,8 @@
 ; CHECK: {{.*reg\(\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{}}
 ; CHECK: {{.*reg\(\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{}}
 ; CHECK: {{.*reg\(\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{}}
-; CHECK: {{.*reg\(\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{}}
 ; CHECK-NOT:reg
 ; CHECK: Filtering for use
-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-S128"
-target triple = "x86_64-unknown-freebsd9"
 
 %struct.snork = type { %struct.fuga, i32, i32, i32, i32, i32, i32 }
 %struct.fuga = type { %struct.gork, i64 }
diff --git a/test/Transforms/LoopStrengthReduce/2013-01-14-ReuseCast.ll b/test/Transforms/LoopStrengthReduce/2013-01-14-ReuseCast.ll
new file mode 100644
index 0000000..8fbddf8
--- /dev/null
+++ b/test/Transforms/LoopStrengthReduce/2013-01-14-ReuseCast.ll
@@ -0,0 +1,84 @@
+; RUN: opt -loop-reduce -S < %s | FileCheck %s
+;
+; LTO of clang, which mistakenly uses no TargetLoweringInfo, causes a
+; miscompile. ReuseOrCreateCast replace ptrtoint operand with undef.
+; Reproducing the miscompile requires no triple, hence no "TTI".
+; rdar://13007381
+
+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-S128"
+
+; Verify that nothing uses the "dead" ptrtoint from "undef".
+; CHECK: @VerifyDiagnosticConsumerTest
+; CHECK: bb:
+; CHECK: %0 = ptrtoint i8* undef to i64
+; CHECK-NOT: %0
+; CHECK: .lr.ph
+; CHECK-NOT: %0
+; CHECK: sub i64 %7, %tmp6
+; CHECK-NOT: %0
+; CHECK: ret void
+define void @VerifyDiagnosticConsumerTest() unnamed_addr nounwind uwtable align 2 {
+bb:
+  %tmp3 = call i8* @getCharData() nounwind
+  %tmp4 = call i8* @getCharData() nounwind
+  %tmp5 = ptrtoint i8* %tmp4 to i64
+  %tmp6 = ptrtoint i8* %tmp3 to i64
+  %tmp7 = sub i64 %tmp5, %tmp6
+  br i1 undef, label %bb87, label %.preheader
+
+.preheader:                                       ; preds = %bb10, %bb
+  br i1 undef, label %_ZNK4llvm9StringRef4findEcm.exit42.thread, label %bb10
+
+bb10:                                             ; preds = %.preheader
+  br i1 undef, label %_ZNK4llvm9StringRef4findEcm.exit42, label %.preheader
+
+_ZNK4llvm9StringRef4findEcm.exit42:               ; preds = %bb10
+  br i1 undef, label %_ZNK4llvm9StringRef4findEcm.exit42.thread, label %.lr.ph
+
+_ZNK4llvm9StringRef4findEcm.exit42.thread:        ; preds = %_ZNK4llvm9StringRef4findEcm.exit42, %.preheader
+  unreachable
+
+.lr.ph:                                           ; preds = %_ZNK4llvm9StringRef4findEcm.exit42
+  br label %bb36
+
+_ZNK4llvm9StringRef4findEcm.exit.loopexit:        ; preds = %bb63
+  %tmp21 = icmp eq i64 %i.0.i, -1
+  br i1 %tmp21, label %_ZNK4llvm9StringRef4findEcm.exit._crit_edge, label %bb36
+
+_ZNK4llvm9StringRef4findEcm.exit._crit_edge:      ; preds = %bb61, %_ZNK4llvm9StringRef4findEcm.exit.loopexit
+  unreachable
+
+bb36:                                             ; preds = %_ZNK4llvm9StringRef4findEcm.exit.loopexit, %.lr.ph
+  %loc.063 = phi i64 [ undef, %.lr.ph ], [ %i.0.i, %_ZNK4llvm9StringRef4findEcm.exit.loopexit ]
+  switch i8 undef, label %bb57 [
+    i8 10, label %bb48
+    i8 13, label %bb48
+  ]
+
+bb48:                                             ; preds = %bb36, %bb36
+  br label %bb58
+
+bb57:                                             ; preds = %bb36
+  br label %bb58
+
+bb58:                                             ; preds = %bb57, %bb48
+  %tmp59 = icmp ugt i64 %tmp7, undef
+  %tmp60 = select i1 %tmp59, i64 undef, i64 %tmp7
+  br label %bb61
+
+bb61:                                             ; preds = %bb63, %bb58
+  %i.0.i = phi i64 [ %tmp60, %bb58 ], [ %tmp67, %bb63 ]
+  %tmp62 = icmp eq i64 %i.0.i, %tmp7
+  br i1 %tmp62, label %_ZNK4llvm9StringRef4findEcm.exit._crit_edge, label %bb63
+
+bb63:                                             ; preds = %bb61
+  %tmp64 = getelementptr inbounds i8* %tmp3, i64 %i.0.i
+  %tmp65 = load i8* %tmp64, align 1
+  %tmp67 = add i64 %i.0.i, 1
+  br i1 undef, label %_ZNK4llvm9StringRef4findEcm.exit.loopexit, label %bb61
+
+bb87:                                             ; preds = %bb
+  ret void
+}
+
+declare i8* @getCharData()
diff --git a/test/Transforms/LoopStrengthReduce/ARM/ivchain-ARM.ll b/test/Transforms/LoopStrengthReduce/ARM/ivchain-ARM.ll
index 9189d79..ee3cc4d 100644
--- a/test/Transforms/LoopStrengthReduce/ARM/ivchain-ARM.ll
+++ b/test/Transforms/LoopStrengthReduce/ARM/ivchain-ARM.ll
@@ -205,18 +205,18 @@ for.end:                                          ; preds = %for.body
 ; post-increment addressing, no add's or add.w's beyond the three
 ; mentioned. Most importantly, there should be no spills or reloads!
 ;
-; CHECK: testNeon:
-; CHECK: %.lr.ph
-; CHECK-NOT: lsl.w
-; CHECK-NOT: {{ldr|str|adds|add r}}
-; CHECK: add.w r
-; CHECK-NOT: {{ldr|str|adds|add r}}
-; CHECK: add.w r
-; CHECK-NOT: {{ldr|str|adds|add r}}
-; CHECK: add.w r
-; CHECK-NOT: {{ldr|str|adds|add r}}
-; CHECK-NOT: add.w r
-; CHECK: bne
+; A9: testNeon:
+; A9: %.lr.ph
+; A9-NOT: lsl.w
+; A9-NOT: {{ldr|str|adds|add r}}
+; A9: add.w r
+; A9-NOT: {{ldr|str|adds|add r}}
+; A9: add.w r
+; A9-NOT: {{ldr|str|adds|add r}}
+; A9: add.w r
+; A9-NOT: {{ldr|str|adds|add r}}
+; A9-NOT: add.w r
+; A9: bne
 define hidden void @testNeon(i8* %ref_data, i32 %ref_stride, i32 %limit, <16 x i8>* nocapture %data) nounwind optsize {
   %1 = icmp sgt i32 %limit, 0
   br i1 %1, label %.lr.ph, label %45
@@ -290,3 +290,80 @@ define hidden void @testNeon(i8* %ref_data, i32 %ref_stride, i32 %limit, <16 x i
 }
 
 declare <1 x i64> @llvm.arm.neon.vld1.v1i64(i8*, i32) nounwind readonly
+
+; Handle chains in which the same offset is used for both loads and
+; stores to the same array.
+; rdar://11410078.
+;
+; A9: @testReuse
+; A9: %for.body
+; A9: vld1.8 {d{{[0-9]+}}}, [[BASE:[r[0-9]+]]], [[INC:r[0-9]]]
+; A9: vld1.8 {d{{[0-9]+}}}, [[BASE]], [[INC]]
+; A9: vld1.8 {d{{[0-9]+}}}, [[BASE]], [[INC]]
+; A9: vld1.8 {d{{[0-9]+}}}, [[BASE]], [[INC]]
+; A9: vld1.8 {d{{[0-9]+}}}, [[BASE]], [[INC]]
+; A9: vld1.8 {d{{[0-9]+}}}, [[BASE]], [[INC]]
+; A9: vld1.8 {d{{[0-9]+}}}, [[BASE]], [[INC]]
+; A9: vld1.8 {d{{[0-9]+}}}, [[BASE]], {{r[0-9]}}
+; A9: vst1.8 {d{{[0-9]+}}}, [[BASE]], [[INC]]
+; A9: vst1.8 {d{{[0-9]+}}}, [[BASE]], [[INC]]
+; A9: vst1.8 {d{{[0-9]+}}}, [[BASE]], [[INC]]
+; A9: vst1.8 {d{{[0-9]+}}}, [[BASE]], [[INC]]
+; A9: vst1.8 {d{{[0-9]+}}}, [[BASE]], [[INC]]
+; A9: vst1.8 {d{{[0-9]+}}}, [[BASE]]
+; A9: bne
+define void @testReuse(i8* %src, i32 %stride) nounwind ssp {
+entry:
+  %mul = shl nsw i32 %stride, 2
+  %idx.neg = sub i32 0, %mul
+  %mul1 = mul nsw i32 %stride, 3
+  %idx.neg2 = sub i32 0, %mul1
+  %mul5 = shl nsw i32 %stride, 1
+  %idx.neg6 = sub i32 0, %mul5
+  %idx.neg10 = sub i32 0, %stride
+  br label %for.body
+
+for.body:                                         ; preds = %for.body, %entry
+  %i.0110 = phi i32 [ 0, %entry ], [ %inc, %for.body ]
+  %src.addr = phi i8* [ %src, %entry ], [ %add.ptr45, %for.body ]
+  %add.ptr = getelementptr inbounds i8* %src.addr, i32 %idx.neg
+  %vld1 = tail call <8 x i8> @llvm.arm.neon.vld1.v8i8(i8* %add.ptr, i32 1)
+  %add.ptr3 = getelementptr inbounds i8* %src.addr, i32 %idx.neg2
+  %vld2 = tail call <8 x i8> @llvm.arm.neon.vld1.v8i8(i8* %add.ptr3, i32 1)
+  %add.ptr7 = getelementptr inbounds i8* %src.addr, i32 %idx.neg6
+  %vld3 = tail call <8 x i8> @llvm.arm.neon.vld1.v8i8(i8* %add.ptr7, i32 1)
+  %add.ptr11 = getelementptr inbounds i8* %src.addr, i32 %idx.neg10
+  %vld4 = tail call <8 x i8> @llvm.arm.neon.vld1.v8i8(i8* %add.ptr11, i32 1)
+  %vld5 = tail call <8 x i8> @llvm.arm.neon.vld1.v8i8(i8* %src.addr, i32 1)
+  %add.ptr17 = getelementptr inbounds i8* %src.addr, i32 %stride
+  %vld6 = tail call <8 x i8> @llvm.arm.neon.vld1.v8i8(i8* %add.ptr17, i32 1)
+  %add.ptr20 = getelementptr inbounds i8* %src.addr, i32 %mul5
+  %vld7 = tail call <8 x i8> @llvm.arm.neon.vld1.v8i8(i8* %add.ptr20, i32 1)
+  %add.ptr23 = getelementptr inbounds i8* %src.addr, i32 %mul1
+  %vld8 = tail call <8 x i8> @llvm.arm.neon.vld1.v8i8(i8* %add.ptr23, i32 1)
+  %vadd1 = tail call <8 x i8> @llvm.arm.neon.vhaddu.v8i8(<8 x i8> %vld1, <8 x i8> %vld2) nounwind
+  %vadd2 = tail call <8 x i8> @llvm.arm.neon.vhaddu.v8i8(<8 x i8> %vld2, <8 x i8> %vld3) nounwind
+  %vadd3 = tail call <8 x i8> @llvm.arm.neon.vhaddu.v8i8(<8 x i8> %vld3, <8 x i8> %vld4) nounwind
+  %vadd4 = tail call <8 x i8> @llvm.arm.neon.vhaddu.v8i8(<8 x i8> %vld4, <8 x i8> %vld5) nounwind
+  %vadd5 = tail call <8 x i8> @llvm.arm.neon.vhaddu.v8i8(<8 x i8> %vld5, <8 x i8> %vld6) nounwind
+  %vadd6 = tail call <8 x i8> @llvm.arm.neon.vhaddu.v8i8(<8 x i8> %vld6, <8 x i8> %vld7) nounwind
+  tail call void @llvm.arm.neon.vst1.v8i8(i8* %add.ptr3, <8 x i8> %vadd1, i32 1)
+  tail call void @llvm.arm.neon.vst1.v8i8(i8* %add.ptr7, <8 x i8> %vadd2, i32 1)
+  tail call void @llvm.arm.neon.vst1.v8i8(i8* %add.ptr11, <8 x i8> %vadd3, i32 1)
+  tail call void @llvm.arm.neon.vst1.v8i8(i8* %src.addr, <8 x i8> %vadd4, i32 1)
+  tail call void @llvm.arm.neon.vst1.v8i8(i8* %add.ptr17, <8 x i8> %vadd5, i32 1)
+  tail call void @llvm.arm.neon.vst1.v8i8(i8* %add.ptr20, <8 x i8> %vadd6, i32 1)
+  %inc = add nsw i32 %i.0110, 1
+  %add.ptr45 = getelementptr inbounds i8* %src.addr, i32 8
+  %exitcond = icmp eq i32 %inc, 4
+  br i1 %exitcond, label %for.end, label %for.body
+
+for.end:                                          ; preds = %for.body
+  ret void
+}
+
+declare <8 x i8> @llvm.arm.neon.vld1.v8i8(i8*, i32) nounwind readonly
+
+declare void @llvm.arm.neon.vst1.v8i8(i8*, <8 x i8>, i32) nounwind
+
+declare <8 x i8> @llvm.arm.neon.vhaddu.v8i8(<8 x i8>, <8 x i8>) nounwind readnone
diff --git a/test/Transforms/LoopStrengthReduce/X86/2011-12-04-loserreg.ll b/test/Transforms/LoopStrengthReduce/X86/2011-12-04-loserreg.ll
index 5108650..eedfc20 100644
--- a/test/Transforms/LoopStrengthReduce/X86/2011-12-04-loserreg.ll
+++ b/test/Transforms/LoopStrengthReduce/X86/2011-12-04-loserreg.ll
@@ -1,4 +1,4 @@
-; RUN: llc < %s | FileCheck %s
+; RUN: opt < %s -loop-reduce -S | FileCheck %s
 ;
 ; Test LSR's ability to prune formulae that refer to nonexistant
 ; AddRecs in other loops.
@@ -15,13 +15,10 @@ target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f3
 target triple = "x86_64-apple-darwin"
 
 ; CHECK: @test
-; CHECK: # %for.body{{$}}
-; dummyiv copy should be removed
-; CHECK-NOT: movq
-; CHECK: # %for.cond19.preheader
-; dummycnt should be removed
-; CHECK-NOT: incq
-; CHECK: # %for.body23{{$}}
+; CHECK: for.body:
+; CHECK: %lsr.iv
+; CHECK-NOT: %dummyout
+; CHECK: ret
 define i64 @test(i64 %count, float* nocapture %srcrow, i32* nocapture %destrow) nounwind uwtable ssp {
 entry:
   %cmp34 = icmp eq i64 %count, 0
diff --git a/test/Transforms/LoopStrengthReduce/post-inc-icmpzero.ll b/test/Transforms/LoopStrengthReduce/post-inc-icmpzero.ll
index 9e02d92..45aeb4e 100644
--- a/test/Transforms/LoopStrengthReduce/post-inc-icmpzero.ll
+++ b/test/Transforms/LoopStrengthReduce/post-inc-icmpzero.ll
@@ -4,18 +4,17 @@
 ; LSR should properly handle the post-inc offset when folding the
 ; non-IV operand of an icmp into the IV.
 
-; CHECK:   %3 = sub i64 %sub.ptr.lhs.cast, %sub.ptr.rhs.cast
-; CHECK:   %4 = lshr i64 %3, 1
-; CHECK:   %5 = mul i64 %4, 2
+; CHECK:   [[r1:%[a-z0-9]+]] = sub i64 %sub.ptr.lhs.cast, %sub.ptr.rhs.cast
+; CHECK:   [[r2:%[a-z0-9]+]] = lshr i64 [[r1]], 1
+; CHECK:   [[r3:%[a-z0-9]+]] = mul i64 [[r2]], 2
 ; CHECK:   br label %for.body
 ; CHECK: for.body:
-; CHECK:   %lsr.iv2 = phi i64 [ %lsr.iv.next, %for.body ], [ %5, %for.body.lr.ph ]
+; CHECK:   %lsr.iv2 = phi i64 [ %lsr.iv.next, %for.body ], [ [[r3]], %for.body.lr.ph ]
 ; CHECK:   %lsr.iv.next = add i64 %lsr.iv2, -2
 ; CHECK:   %lsr.iv.next3 = inttoptr i64 %lsr.iv.next to i16*
 ; CHECK:   %cmp27 = icmp eq i16* %lsr.iv.next3, null
 
 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-unknown-linux-gnu"
 
 %struct.Vector2 = type { i16*, [64 x i16], i32 }
 
diff --git a/test/Transforms/LoopUnswitch/2011-11-18-SimpleSwitch.ll b/test/Transforms/LoopUnswitch/2011-11-18-SimpleSwitch.ll
index 59a8236..bde52da 100644
--- a/test/Transforms/LoopUnswitch/2011-11-18-SimpleSwitch.ll
+++ b/test/Transforms/LoopUnswitch/2011-11-18-SimpleSwitch.ll
@@ -19,7 +19,7 @@
 ; CHECK-NEXT:     i32 1, label %inc.us
 
 ; CHECK:      inc.us:                                           ; preds = %loop_begin.us
-; CHECK-NEXT:   call void @incf() noreturn nounwind
+; CHECK-NEXT:   call void @incf() [[NOR_NUW:#[0-9]+]]
 ; CHECK-NEXT:   br label %loop_begin.backedge.us
 
 ; CHECK:      .split:                                           ; preds = %..split_crit_edge
@@ -40,7 +40,7 @@
 ; CHECK-NEXT:   ]
 
 ; CHECK:      dec.us3:                                          ; preds = %loop_begin.us1
-; CHECK-NEXT:   call void @decf() noreturn nounwind
+; CHECK-NEXT:   call void @decf() [[NOR_NUW]]
 ; CHECK-NEXT:   br label %loop_begin.backedge.us5
 
 ; CHECK:      .split.split:                                     ; preds = %.split..split.split_crit_edge
@@ -89,3 +89,6 @@ loop_exit:
 
 declare void @incf() noreturn
 declare void @decf() noreturn
+
+; CHECK: attributes #0 = { noreturn }
+; CHECK: attributes [[NOR_NUW]] = { noreturn nounwind }
diff --git a/test/Transforms/LoopUnswitch/2011-11-18-TwoSwitches-Threshold.ll b/test/Transforms/LoopUnswitch/2011-11-18-TwoSwitches-Threshold.ll
index 67982fe..c3bf596 100644
--- a/test/Transforms/LoopUnswitch/2011-11-18-TwoSwitches-Threshold.ll
+++ b/test/Transforms/LoopUnswitch/2011-11-18-TwoSwitches-Threshold.ll
@@ -25,7 +25,7 @@
 ; CHECK-NEXT:   ]
 
 ; CHECK:      inc.us:                                           ; preds = %second_switch.us, %loop_begin.us
-; CHECK-NEXT:   call void @incf() noreturn nounwind
+; CHECK-NEXT:   call void @incf() [[NOR_NUW:#[0-9]+]]
 ; CHECK-NEXT:   br label %loop_begin.backedge.us
 
 ; CHECK:      .split:                                           ; preds = %..split_crit_edge
@@ -45,7 +45,7 @@
 ; CHECK-NEXT:   ]
 
 ; CHECK:      inc:                                              ; preds = %loop_begin.inc_crit_edge, %second_switch
-; CHECK-NEXT:   call void @incf() noreturn nounwind
+; CHECK-NEXT:   call void @incf() [[NOR_NUW]]
 ; CHECK-NEXT:   br label %loop_begin.backedge
 
 define i32 @test(i32* %var) {
@@ -82,3 +82,6 @@ loop_exit:
 
 declare void @incf() noreturn
 declare void @decf() noreturn
+
+; CHECK: attributes #0 = { noreturn }
+; CHECK: attributes [[NOR_NUW]] = { noreturn nounwind }
diff --git a/test/Transforms/LoopUnswitch/2011-11-18-TwoSwitches.ll b/test/Transforms/LoopUnswitch/2011-11-18-TwoSwitches.ll
index 36b7eff..9530333 100644
--- a/test/Transforms/LoopUnswitch/2011-11-18-TwoSwitches.ll
+++ b/test/Transforms/LoopUnswitch/2011-11-18-TwoSwitches.ll
@@ -30,7 +30,7 @@
 ; CHECK-NEXT:     i32 1, label %inc.us.us
 
 ; CHECK:      inc.us.us:                                        ; preds = %second_switch.us.us, %loop_begin.us.us
-; CHECK-NEXT:   call void @incf() noreturn nounwind
+; CHECK-NEXT:   call void @incf() [[NOR_NUW:#[0-9]+]]
 ; CHECK-NEXT:   br label %loop_begin.backedge.us.us
 
 ; CHECK:      .split.us.split:                                  ; preds = %.split.us..split.us.split_crit_edge
@@ -50,7 +50,7 @@
 ; CHECK-NEXT:   br i1 true, label %us-unreachable8, label %inc.us
 
 ; CHECK:      inc.us:                                           ; preds = %second_switch.us.inc.us_crit_edge, %loop_begin.us
-; CHECK-NEXT:   call void @incf() noreturn nounwind
+; CHECK-NEXT:   call void @incf() [[NOR_NUW]]
 ; CHECK-NEXT:   br label %loop_begin.backedge.us
 
 ; CHECK:      .split:                                           ; preds = %..split_crit_edge
@@ -75,7 +75,7 @@
 ; CHECK-NEXT:   ]
 
 ; CHECK:      inc.us4:                                          ; preds = %loop_begin.inc_crit_edge.us, %second_switch.us3
-; CHECK-NEXT:   call void @incf() noreturn nounwind
+; CHECK-NEXT:   call void @incf() [[NOR_NUW]]
 ; CHECK-NEXT:   br label %loop_begin.backedge.us6
 
 ; CHECK:      loop_begin.inc_crit_edge.us:                      ; preds = %loop_begin.us1
@@ -136,3 +136,6 @@ loop_exit:
 
 declare void @incf() noreturn
 declare void @decf() noreturn
+
+; CHECK: attributes #0 = { noreturn }
+; CHECK: attributes [[NOR_NUW]] = { noreturn nounwind }
diff --git a/test/Transforms/LoopUnswitch/infinite-loop.ll b/test/Transforms/LoopUnswitch/infinite-loop.ll
index 73391ca..f3fba64 100644
--- a/test/Transforms/LoopUnswitch/infinite-loop.ll
+++ b/test/Transforms/LoopUnswitch/infinite-loop.ll
@@ -21,11 +21,11 @@
 ; CHECK-NEXT: br label %cond.end.us
 
 ; CHECK: abort0.split:
-; CHECK-NEXT: call void @end0() noreturn nounwind
+; CHECK-NEXT: call void @end0() [[NOR_NUW:#[0-9]+]]
 ; CHECK-NEXT: unreachable
 
 ; CHECK: abort1:
-; CHECK-NEXT: call void @end1() noreturn nounwind
+; CHECK-NEXT: call void @end1() [[NOR_NUW]]
 ; CHECK-NEXT: unreachable
 
 ; CHECK: }
@@ -51,3 +51,7 @@ abort1:
 
 declare void @end0() noreturn
 declare void @end1() noreturn
+
+; CHECK: attributes #0 = { nounwind }
+; CHECK: attributes #1 = { noreturn }
+; CHECK: attributes [[NOR_NUW]] = { noreturn nounwind }
diff --git a/test/Transforms/LoopVectorize/12-12-11-if-conv.ll b/test/Transforms/LoopVectorize/12-12-11-if-conv.ll
index f285887..2dd7fe3 100644
--- a/test/Transforms/LoopVectorize/12-12-11-if-conv.ll
+++ b/test/Transforms/LoopVectorize/12-12-11-if-conv.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -enable-if-conversion -dce -instcombine -licm -S | FileCheck %s
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -enable-if-conversion -dce -instcombine -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-S128"
 target triple = "x86_64-apple-macosx10.8.0"
diff --git a/test/Transforms/LoopVectorize/ARM/arm-unroll.ll b/test/Transforms/LoopVectorize/ARM/arm-unroll.ll
new file mode 100644
index 0000000..c8d307f
--- /dev/null
+++ b/test/Transforms/LoopVectorize/ARM/arm-unroll.ll
@@ -0,0 +1,32 @@
+; RUN: opt < %s  -loop-vectorize -mtriple=thumbv7-apple-ios3.0.0 -S | FileCheck %s
+; RUN: opt < %s  -loop-vectorize -mtriple=thumbv7-apple-ios3.0.0 -mcpu=swift -S | FileCheck %s --check-prefix=SWIFT
+
+target datalayout = "e-p:32:32:32-i1:8:32-i8:8:32-i16:16:32-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:32:64-v128:32:128-a0:0:32-n32-S32"
+target triple = "thumbv7-apple-ios3.0.0"
+
+;CHECK: @foo
+;CHECK: load <4 x i32>
+;CHECK-NOT: load <4 x i32>
+;CHECK: ret
+;SWIFT: @foo
+;SWIFT: load <4 x i32>
+;SWIFT: load <4 x i32>
+;SWIFT: ret
+define i32 @foo(i32* nocapture %A, i32 %n) nounwind readonly ssp {
+  %1 = icmp sgt i32 %n, 0
+  br i1 %1, label %.lr.ph, label %._crit_edge
+
+.lr.ph:                                           ; preds = %0, %.lr.ph
+  %i.02 = phi i32 [ %5, %.lr.ph ], [ 0, %0 ]
+  %sum.01 = phi i32 [ %4, %.lr.ph ], [ 0, %0 ]
+  %2 = getelementptr inbounds i32* %A, i32 %i.02
+  %3 = load i32* %2, align 4
+  %4 = add nsw i32 %3, %sum.01
+  %5 = add nsw i32 %i.02, 1
+  %exitcond = icmp eq i32 %5, %n
+  br i1 %exitcond, label %._crit_edge, label %.lr.ph
+
+._crit_edge:                                      ; preds = %.lr.ph, %0
+  %sum.0.lcssa = phi i32 [ 0, %0 ], [ %4, %.lr.ph ]
+  ret i32 %sum.0.lcssa
+}
diff --git a/test/Transforms/LoopVectorize/ARM/gcc-examples.ll b/test/Transforms/LoopVectorize/ARM/gcc-examples.ll
new file mode 100644
index 0000000..6a68e81
--- /dev/null
+++ b/test/Transforms/LoopVectorize/ARM/gcc-examples.ll
@@ -0,0 +1,60 @@
+; RUN: opt < %s  -loop-vectorize -mtriple=thumbv7-apple-ios3.0.0 -mcpu=swift -S -dce | FileCheck %s
+
+target datalayout = "e-p:32:32:32-i1:8:32-i8:8:32-i16:16:32-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:32:64-v128:32:128-a0:0:32-n32-S32"
+target triple = "thumbv7-apple-ios3.0.0"
+
+@b = common global [2048 x i32] zeroinitializer, align 16
+@c = common global [2048 x i32] zeroinitializer, align 16
+@a = common global [2048 x i32] zeroinitializer, align 16
+
+; Select VF = 8;
+;CHECK: @example1
+;CHECK: load <4 x i32>
+;CHECK: add nsw <4 x i32>
+;CHECK: store <4 x i32>
+;CHECK: ret void
+define void @example1() nounwind uwtable ssp {
+  br label %1
+
+; <label>:1                                       ; preds = %1, %0
+  %indvars.iv = phi i64 [ 0, %0 ], [ %indvars.iv.next, %1 ]
+  %2 = getelementptr inbounds [2048 x i32]* @b, i64 0, i64 %indvars.iv
+  %3 = load i32* %2, align 4
+  %4 = getelementptr inbounds [2048 x i32]* @c, i64 0, i64 %indvars.iv
+  %5 = load i32* %4, align 4
+  %6 = add nsw i32 %5, %3
+  %7 = getelementptr inbounds [2048 x i32]* @a, i64 0, i64 %indvars.iv
+  store i32 %6, i32* %7, align 4
+  %indvars.iv.next = add i64 %indvars.iv, 1
+  %lftr.wideiv = trunc i64 %indvars.iv.next to i32
+  %exitcond = icmp eq i32 %lftr.wideiv, 256
+  br i1 %exitcond, label %8, label %1
+
+; <label>:8                                       ; preds = %1
+  ret void
+}
+
+;CHECK: @example10b
+;CHECK: load <4 x i16>
+;CHECK: sext <4 x i16>
+;CHECK: store <4 x i32>
+;CHECK: ret void
+define void @example10b(i16* noalias nocapture %sa, i16* noalias nocapture %sb, i16* noalias nocapture %sc, i32* noalias nocapture %ia, i32* noalias nocapture %ib, i32* noalias nocapture %ic) nounwind uwtable ssp {
+  br label %1
+
+; <label>:1                                       ; preds = %1, %0
+  %indvars.iv = phi i64 [ 0, %0 ], [ %indvars.iv.next, %1 ]
+  %2 = getelementptr inbounds i16* %sb, i64 %indvars.iv
+  %3 = load i16* %2, align 2
+  %4 = sext i16 %3 to i32
+  %5 = getelementptr inbounds i32* %ia, i64 %indvars.iv
+  store i32 %4, i32* %5, align 4
+  %indvars.iv.next = add i64 %indvars.iv, 1
+  %lftr.wideiv = trunc i64 %indvars.iv.next to i32
+  %exitcond = icmp eq i32 %lftr.wideiv, 1024
+  br i1 %exitcond, label %6, label %1
+
+; <label>:6                                       ; preds = %1
+  ret void
+}
+
diff --git a/test/Transforms/LoopVectorize/ARM/lit.local.cfg b/test/Transforms/LoopVectorize/ARM/lit.local.cfg
new file mode 100644
index 0000000..cb77b09
--- /dev/null
+++ b/test/Transforms/LoopVectorize/ARM/lit.local.cfg
@@ -0,0 +1,6 @@
+config.suffixes = ['.ll', '.c', '.cpp']
+
+targets = set(config.root.targets_to_build.split())
+if not 'ARM' in targets:
+    config.unsupported = True
+
diff --git a/test/Transforms/LoopVectorize/ARM/mul-cast-vect.ll b/test/Transforms/LoopVectorize/ARM/mul-cast-vect.ll
new file mode 100644
index 0000000..d2e3de2
--- /dev/null
+++ b/test/Transforms/LoopVectorize/ARM/mul-cast-vect.ll
@@ -0,0 +1,114 @@
+; RUN: opt < %s  -cost-model -analyze -mtriple=armv7-linux-gnueabihf -mcpu=cortex-a9 | FileCheck --check-prefix=COST %s
+; To see the assembly output: llc -mcpu=cortex-a9 < %s | FileCheck --check-prefix=ASM %s
+; ASM lines below are only for reference, tests on that direction should go to tests/CodeGen/ARM
+
+; ModuleID = 'arm.ll'
+target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:64:128-a0:0:64-n32-S64"
+target triple = "armv7--linux-gnueabihf"
+
+%T216 = type <2 x i16>
+%T232 = type <2 x i32>
+%T264 = type <2 x i64>
+
+%T416 = type <4 x i16>
+%T432 = type <4 x i32>
+%T464 = type <4 x i64>
+
+define void @direct(%T432* %loadaddr, %T432* %loadaddr2, %T432* %storeaddr) {
+; COST: function 'direct':
+  %v0 = load %T432* %loadaddr
+; ASM: vld1.64
+  %v1 = load %T432* %loadaddr2
+; ASM: vld1.64
+  %r3 = mul %T432 %v0, %v1 
+; COST: cost of 2 for instruction: {{.*}} mul <4 x i32>
+; ASM: vmul.i32
+  store %T432 %r3, %T432* %storeaddr
+; ASM: vst1.64
+  ret void
+}
+
+define void @ups1632(%T416* %loadaddr, %T416* %loadaddr2, %T432* %storeaddr) {
+; COST: function 'ups1632':
+  %v0 = load %T416* %loadaddr
+; ASM: vldr
+  %v1 = load %T416* %loadaddr2
+; ASM: vldr
+  %r1 = sext %T416 %v0 to %T432
+  %r2 = sext %T416 %v1 to %T432
+; COST: cost of 0 for instruction: {{.*}} sext <4 x i16> {{.*}} to <4 x i32>
+  %r3 = mul %T432 %r1, %r2 
+; COST: cost of 2 for instruction: {{.*}} mul <4 x i32>
+; ASM: vmull.s16
+  store %T432 %r3, %T432* %storeaddr
+; ASM: vst1.64
+  ret void
+}
+
+define void @upu1632(%T416* %loadaddr, %T416* %loadaddr2, %T432* %storeaddr) {
+; COST: function 'upu1632':
+  %v0 = load %T416* %loadaddr
+; ASM: vldr
+  %v1 = load %T416* %loadaddr2
+; ASM: vldr
+  %r1 = zext %T416 %v0 to %T432
+  %r2 = zext %T416 %v1 to %T432
+; COST: cost of 0 for instruction: {{.*}} zext <4 x i16> {{.*}} to <4 x i32>
+  %r3 = mul %T432 %r1, %r2 
+; COST: cost of 2 for instruction: {{.*}} mul <4 x i32>
+; ASM: vmull.u16
+  store %T432 %r3, %T432* %storeaddr
+; ASM: vst1.64
+  ret void
+}
+
+define void @ups3264(%T232* %loadaddr, %T232* %loadaddr2, %T264* %storeaddr) {
+; COST: function 'ups3264':
+  %v0 = load %T232* %loadaddr
+; ASM: vldr
+  %v1 = load %T232* %loadaddr2
+; ASM: vldr
+  %r3 = mul %T232 %v0, %v1 
+; ASM: vmul.i32
+; COST: cost of 1 for instruction: {{.*}} mul <2 x i32>
+  %st = sext %T232 %r3 to %T264
+; ASM: vmovl.s32
+; COST: cost of 1 for instruction: {{.*}} sext <2 x i32> {{.*}} to <2 x i64>
+  store %T264 %st, %T264* %storeaddr
+; ASM: vst1.64
+  ret void
+}
+
+define void @upu3264(%T232* %loadaddr, %T232* %loadaddr2, %T264* %storeaddr) {
+; COST: function 'upu3264':
+  %v0 = load %T232* %loadaddr
+; ASM: vldr
+  %v1 = load %T232* %loadaddr2
+; ASM: vldr
+  %r3 = mul %T232 %v0, %v1 
+; ASM: vmul.i32
+; COST: cost of 1 for instruction: {{.*}} mul <2 x i32>
+  %st = zext %T232 %r3 to %T264
+; ASM: vmovl.u32
+; COST: cost of 1 for instruction: {{.*}} zext <2 x i32> {{.*}} to <2 x i64>
+  store %T264 %st, %T264* %storeaddr
+; ASM: vst1.64
+  ret void
+}
+
+define void @dn3216(%T432* %loadaddr, %T432* %loadaddr2, %T416* %storeaddr) {
+; COST: function 'dn3216':
+  %v0 = load %T432* %loadaddr
+; ASM: vld1.64
+  %v1 = load %T432* %loadaddr2
+; ASM: vld1.64
+  %r3 = mul %T432 %v0, %v1 
+; ASM: vmul.i32
+; COST: cost of 2 for instruction: {{.*}} mul <4 x i32>
+  %st = trunc %T432 %r3 to %T416
+; ASM: vmovn.i32
+; COST: cost of 1 for instruction: {{.*}} trunc <4 x i32> {{.*}} to <4 x i16>
+  store %T416 %st, %T416* %storeaddr
+; ASM: vstr
+  ret void
+}
diff --git a/test/Transforms/LoopVectorize/ARM/width-detect.ll b/test/Transforms/LoopVectorize/ARM/width-detect.ll
new file mode 100644
index 0000000..c0795b6
--- /dev/null
+++ b/test/Transforms/LoopVectorize/ARM/width-detect.ll
@@ -0,0 +1,52 @@
+; RUN: opt < %s  -loop-vectorize -mtriple=thumbv7-apple-ios3.0.0 -S | FileCheck %s
+
+target datalayout = "e-p:32:32:32-i1:8:32-i8:8:32-i16:16:32-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:32:64-v128:32:128-a0:0:32-n32-S32"
+target triple = "thumbv7-apple-ios3.0.0"
+
+;CHECK:foo_F64
+;CHECK: <2 x double>
+;CHECK:ret
+define double @foo_F64(double* nocapture %A, i32 %n) nounwind uwtable readonly ssp {
+  %1 = icmp sgt i32 %n, 0
+  br i1 %1, label %.lr.ph, label %._crit_edge
+
+.lr.ph:                                           ; preds = %0, %.lr.ph
+  %indvars.iv = phi i64 [ %indvars.iv.next, %.lr.ph ], [ 0, %0 ]
+  %prod.01 = phi double [ %4, %.lr.ph ], [ 0.000000e+00, %0 ]
+  %2 = getelementptr inbounds double* %A, i64 %indvars.iv
+  %3 = load double* %2, align 8
+  %4 = fmul fast double %prod.01, %3
+  %indvars.iv.next = add i64 %indvars.iv, 1
+  %lftr.wideiv = trunc i64 %indvars.iv.next to i32
+  %exitcond = icmp eq i32 %lftr.wideiv, %n
+  br i1 %exitcond, label %._crit_edge, label %.lr.ph
+
+._crit_edge:                                      ; preds = %.lr.ph, %0
+  %prod.0.lcssa = phi double [ 0.000000e+00, %0 ], [ %4, %.lr.ph ]
+  ret double %prod.0.lcssa
+}
+
+;CHECK:foo_I8
+;CHECK: xor <16 x i8>
+;CHECK:ret
+define signext i8 @foo_I8(i8* nocapture %A, i32 %n) nounwind uwtable readonly ssp {
+  %1 = icmp sgt i32 %n, 0
+  br i1 %1, label %.lr.ph, label %._crit_edge
+
+.lr.ph:                                           ; preds = %0, %.lr.ph
+  %indvars.iv = phi i64 [ %indvars.iv.next, %.lr.ph ], [ 0, %0 ]
+  %red.01 = phi i8 [ %4, %.lr.ph ], [ 0, %0 ]
+  %2 = getelementptr inbounds i8* %A, i64 %indvars.iv
+  %3 = load i8* %2, align 1
+  %4 = xor i8 %3, %red.01
+  %indvars.iv.next = add i64 %indvars.iv, 1
+  %lftr.wideiv = trunc i64 %indvars.iv.next to i32
+  %exitcond = icmp eq i32 %lftr.wideiv, %n
+  br i1 %exitcond, label %._crit_edge, label %.lr.ph
+
+._crit_edge:                                      ; preds = %.lr.ph, %0
+  %red.0.lcssa = phi i8 [ 0, %0 ], [ %4, %.lr.ph ]
+  ret i8 %red.0.lcssa
+}
+
+
diff --git a/test/Transforms/LoopVectorize/X86/avx1.ll b/test/Transforms/LoopVectorize/X86/avx1.ll
index a2d176a..a85c6fe 100644
--- a/test/Transforms/LoopVectorize/X86/avx1.ll
+++ b/test/Transforms/LoopVectorize/X86/avx1.ll
@@ -27,7 +27,7 @@ define i32 @read_mod_write_single_ptr(float* nocapture %a, i32 %n) nounwind uwta
 
 
 ;CHECK: @read_mod_i64
-;CHECK: load <8 x i64>
+;CHECK: load <4 x i64>
 ;CHECK: ret i32
 define i32 @read_mod_i64(i64* nocapture %a, i32 %n) nounwind uwtable ssp {
   %1 = icmp sgt i32 %n, 0
diff --git a/test/Transforms/LoopVectorize/X86/conversion-cost.ll b/test/Transforms/LoopVectorize/X86/conversion-cost.ll
index 60c742e..23d9233 100644
--- a/test/Transforms/LoopVectorize/X86/conversion-cost.ll
+++ b/test/Transforms/LoopVectorize/X86/conversion-cost.ll
@@ -4,7 +4,7 @@ target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f3
 target triple = "x86_64-apple-macosx10.8.0"
 
 ;CHECK: @conversion_cost1
-;CHECK: store <8 x i8>
+;CHECK: store <32 x i8>
 ;CHECK: ret
 define i32 @conversion_cost1(i32 %n, i8* nocapture %A, float* nocapture %B) nounwind uwtable ssp {
   %1 = icmp sgt i32 %n, 3
diff --git a/test/Transforms/LoopVectorize/X86/gcc-examples.ll b/test/Transforms/LoopVectorize/X86/gcc-examples.ll
index 0f21ba6..d2d0eac 100644
--- a/test/Transforms/LoopVectorize/X86/gcc-examples.ll
+++ b/test/Transforms/LoopVectorize/X86/gcc-examples.ll
@@ -1,5 +1,5 @@
-; RUN: opt < %s  -loop-vectorize -mtriple=x86_64-apple-macosx10.8.0 -mcpu=corei7 -dce -instcombine -licm -S | FileCheck %s
-; RUN: opt < %s  -loop-vectorize -mtriple=x86_64-apple-macosx10.8.0 -mcpu=corei7 -force-vector-unroll=0 -dce -instcombine -licm -S | FileCheck %s -check-prefix=UNROLL
+; RUN: opt < %s  -loop-vectorize -mtriple=x86_64-apple-macosx10.8.0 -mcpu=corei7 -dce -instcombine -S | FileCheck %s
+; RUN: opt < %s  -loop-vectorize -mtriple=x86_64-apple-macosx10.8.0 -mcpu=corei7 -force-vector-unroll=0 -dce -instcombine -S | FileCheck %s -check-prefix=UNROLL
 
 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-S128"
 target triple = "x86_64-apple-macosx10.8.0"
@@ -53,8 +53,6 @@ define void @example1() nounwind uwtable ssp {
 ;UNROLL: @example10b
 ;UNROLL: load <4 x i16>
 ;UNROLL: load <4 x i16>
-;UNROLL: load <4 x i16>
-;UNROLL: store <4 x i32>
 ;UNROLL: store <4 x i32>
 ;UNROLL: store <4 x i32>
 ;UNROLL: ret void
diff --git a/test/Transforms/LoopVectorize/X86/min-trip-count-switch.ll b/test/Transforms/LoopVectorize/X86/min-trip-count-switch.ll
new file mode 100644
index 0000000..186fba8
--- /dev/null
+++ b/test/Transforms/LoopVectorize/X86/min-trip-count-switch.ll
@@ -0,0 +1,28 @@
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -vectorizer-min-trip-count=1 -force-vector-width=4 -dce -instcombine -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-S128"
+target triple = "x86_64-unknown-linux-gnu"
+
+; CHECK: <4 x float>
+define void @trivial_loop(float* nocapture %a) nounwind uwtable optsize {
+entry:
+  br label %for.body
+
+for.body:                                         ; preds = %for.body, %entry
+  %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
+  %arrayidx = getelementptr inbounds float* %a, i64 %indvars.iv
+  %0 = load float* %arrayidx, align 4, !tbaa !0
+  %add = fadd float %0, 1.000000e+00
+  store float %add, float* %arrayidx, align 4, !tbaa !0
+  %indvars.iv.next = add i64 %indvars.iv, 1
+  %lftr.wideiv = trunc i64 %indvars.iv.next to i32
+  %exitcond = icmp eq i32 %lftr.wideiv, 8
+  br i1 %exitcond, label %for.end, label %for.body
+
+for.end:                                          ; preds = %for.body
+  ret void
+}
+
+!0 = metadata !{metadata !"float", metadata !1}
+!1 = metadata !{metadata !"omnipotent char", metadata !2}
+!2 = metadata !{metadata !"Simple C/C++ TBAA"}
diff --git a/test/Transforms/LoopVectorize/X86/parallel-loops-after-reg2mem.ll b/test/Transforms/LoopVectorize/X86/parallel-loops-after-reg2mem.ll
new file mode 100644
index 0000000..452d0df
--- /dev/null
+++ b/test/Transforms/LoopVectorize/X86/parallel-loops-after-reg2mem.ll
@@ -0,0 +1,52 @@
+; RUN: opt < %s -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -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-S128"
+target triple = "x86_64-unknown-linux-gnu"
+
+; The parallel loop has been invalidated by the new memory accesses introduced
+; by reg2mem (Loop::isParallel() starts to return false). Ensure the loop is
+; now non-vectorizable.
+
+;CHECK-NOT: <4 x i32>
+define void @parallel_loop(i32* nocapture %a, i32* nocapture %b) nounwind uwtable {
+entry:
+  %indvars.iv.next.reg2mem = alloca i64
+  %indvars.iv.reg2mem = alloca i64
+  %"reg2mem alloca point" = bitcast i32 0 to i32
+  store i64 0, i64* %indvars.iv.reg2mem
+  br label %for.body
+
+for.body:                                         ; preds = %for.body.for.body_crit_edge, %entry
+  %indvars.iv.reload = load i64* %indvars.iv.reg2mem
+  %arrayidx = getelementptr inbounds i32* %b, i64 %indvars.iv.reload
+  %0 = load i32* %arrayidx, align 4, !tbaa !0, !llvm.mem.parallel_loop_access !3
+  %arrayidx2 = getelementptr inbounds i32* %a, i64 %indvars.iv.reload
+  %1 = load i32* %arrayidx2, align 4, !tbaa !0, !llvm.mem.parallel_loop_access !3
+  %idxprom3 = sext i32 %1 to i64
+  %arrayidx4 = getelementptr inbounds i32* %a, i64 %idxprom3
+  store i32 %0, i32* %arrayidx4, align 4, !tbaa !0, !llvm.mem.parallel_loop_access !3
+  %indvars.iv.next = add i64 %indvars.iv.reload, 1
+  ; A new store without the parallel metadata here:
+  store i64 %indvars.iv.next, i64* %indvars.iv.next.reg2mem
+  %indvars.iv.next.reload1 = load i64* %indvars.iv.next.reg2mem
+  %arrayidx6 = getelementptr inbounds i32* %b, i64 %indvars.iv.next.reload1
+  %2 = load i32* %arrayidx6, align 4, !tbaa !0, !llvm.mem.parallel_loop_access !3
+  store i32 %2, i32* %arrayidx2, align 4, !tbaa !0, !llvm.mem.parallel_loop_access !3
+  %indvars.iv.next.reload = load i64* %indvars.iv.next.reg2mem
+  %lftr.wideiv = trunc i64 %indvars.iv.next.reload to i32
+  %exitcond = icmp eq i32 %lftr.wideiv, 512
+  br i1 %exitcond, label %for.end, label %for.body.for.body_crit_edge, !llvm.loop.parallel !3
+
+for.body.for.body_crit_edge:                      ; preds = %for.body
+  %indvars.iv.next.reload2 = load i64* %indvars.iv.next.reg2mem
+  store i64 %indvars.iv.next.reload2, i64* %indvars.iv.reg2mem
+  br label %for.body
+
+for.end:                                          ; preds = %for.body
+  ret void
+}
+
+!0 = metadata !{metadata !"int", metadata !1}
+!1 = metadata !{metadata !"omnipotent char", metadata !2}
+!2 = metadata !{metadata !"Simple C/C++ TBAA"}
+!3 = metadata !{metadata !3}
diff --git a/test/Transforms/LoopVectorize/X86/parallel-loops.ll b/test/Transforms/LoopVectorize/X86/parallel-loops.ll
new file mode 100644
index 0000000..f648722
--- /dev/null
+++ b/test/Transforms/LoopVectorize/X86/parallel-loops.ll
@@ -0,0 +1,114 @@
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -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-S128"
+target triple = "x86_64-unknown-linux-gnu"
+
+; A tricky loop:
+;
+; void loop(int *a, int *b) {
+;    for (int i = 0; i < 512; ++i) {
+;        a[a[i]] = b[i];
+;        a[i] = b[i+1];
+;    }
+;}
+
+;CHECK: @loop
+;CHECK-NOT: <4 x i32>
+define void @loop(i32* nocapture %a, i32* nocapture %b) nounwind uwtable {
+entry:
+  br label %for.body
+
+for.body:                                         ; preds = %for.body, %entry
+  %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
+  %arrayidx = getelementptr inbounds i32* %b, i64 %indvars.iv
+  %0 = load i32* %arrayidx, align 4, !tbaa !0
+  %arrayidx2 = getelementptr inbounds i32* %a, i64 %indvars.iv
+  %1 = load i32* %arrayidx2, align 4, !tbaa !0
+  %idxprom3 = sext i32 %1 to i64
+  %arrayidx4 = getelementptr inbounds i32* %a, i64 %idxprom3
+  store i32 %0, i32* %arrayidx4, align 4, !tbaa !0
+  %indvars.iv.next = add i64 %indvars.iv, 1
+  %arrayidx6 = getelementptr inbounds i32* %b, i64 %indvars.iv.next
+  %2 = load i32* %arrayidx6, align 4, !tbaa !0
+  store i32 %2, i32* %arrayidx2, align 4, !tbaa !0
+  %lftr.wideiv = trunc i64 %indvars.iv.next to i32
+  %exitcond = icmp eq i32 %lftr.wideiv, 512
+  br i1 %exitcond, label %for.end, label %for.body
+
+for.end:                                          ; preds = %for.body
+  ret void
+}
+
+; The same loop with parallel loop metadata added to the loop branch
+; and the memory instructions.
+
+;CHECK: @parallel_loop
+;CHECK: <4 x i32>
+define void @parallel_loop(i32* nocapture %a, i32* nocapture %b) nounwind uwtable {
+entry:
+  br label %for.body
+
+for.body:                                         ; preds = %for.body, %entry
+  %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
+  %arrayidx = getelementptr inbounds i32* %b, i64 %indvars.iv
+  %0 = load i32* %arrayidx, align 4, !tbaa !0, !llvm.mem.parallel_loop_access !3
+  %arrayidx2 = getelementptr inbounds i32* %a, i64 %indvars.iv
+  %1 = load i32* %arrayidx2, align 4, !tbaa !0, !llvm.mem.parallel_loop_access !3
+  %idxprom3 = sext i32 %1 to i64
+  %arrayidx4 = getelementptr inbounds i32* %a, i64 %idxprom3
+  ; This store might have originated from inlining a function with a parallel
+  ; loop. Refers to a list with the "original loop reference" (!4) also included.
+  store i32 %0, i32* %arrayidx4, align 4, !tbaa !0, !llvm.mem.parallel_loop_access !5
+  %indvars.iv.next = add i64 %indvars.iv, 1
+  %arrayidx6 = getelementptr inbounds i32* %b, i64 %indvars.iv.next
+  %2 = load i32* %arrayidx6, align 4, !tbaa !0, !llvm.mem.parallel_loop_access !3
+  store i32 %2, i32* %arrayidx2, align 4, !tbaa !0, !llvm.mem.parallel_loop_access !3
+  %lftr.wideiv = trunc i64 %indvars.iv.next to i32
+  %exitcond = icmp eq i32 %lftr.wideiv, 512
+  br i1 %exitcond, label %for.end, label %for.body, !llvm.loop.parallel !3
+
+for.end:                                          ; preds = %for.body
+  ret void
+}
+
+; The same loop with an illegal parallel loop metadata: the memory
+; accesses refer to a different loop's identifier.
+
+;CHECK: @mixed_metadata
+;CHECK-NOT: <4 x i32>
+
+define void @mixed_metadata(i32* nocapture %a, i32* nocapture %b) nounwind uwtable {
+entry:
+  br label %for.body
+
+for.body:                                         ; preds = %for.body, %entry
+  %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
+  %arrayidx = getelementptr inbounds i32* %b, i64 %indvars.iv
+  %0 = load i32* %arrayidx, align 4, !tbaa !0, !llvm.mem.parallel_loop_access !6
+  %arrayidx2 = getelementptr inbounds i32* %a, i64 %indvars.iv
+  %1 = load i32* %arrayidx2, align 4, !tbaa !0, !llvm.mem.parallel_loop_access !6
+  %idxprom3 = sext i32 %1 to i64
+  %arrayidx4 = getelementptr inbounds i32* %a, i64 %idxprom3
+  ; This refers to the loop marked with !7 which we are not in at the moment.
+  ; It should prevent detecting as a parallel loop.
+  store i32 %0, i32* %arrayidx4, align 4, !tbaa !0, !llvm.mem.parallel_loop_access !7
+  %indvars.iv.next = add i64 %indvars.iv, 1
+  %arrayidx6 = getelementptr inbounds i32* %b, i64 %indvars.iv.next
+  %2 = load i32* %arrayidx6, align 4, !tbaa !0, !llvm.mem.parallel_loop_access !6
+  store i32 %2, i32* %arrayidx2, align 4, !tbaa !0, !llvm.mem.parallel_loop_access !6
+  %lftr.wideiv = trunc i64 %indvars.iv.next to i32
+  %exitcond = icmp eq i32 %lftr.wideiv, 512
+  br i1 %exitcond, label %for.end, label %for.body, !llvm.loop.parallel !6
+
+for.end:                                          ; preds = %for.body
+  ret void
+}
+
+!0 = metadata !{metadata !"int", metadata !1}
+!1 = metadata !{metadata !"omnipotent char", metadata !2}
+!2 = metadata !{metadata !"Simple C/C++ TBAA"}
+!3 = metadata !{metadata !3}
+!4 = metadata !{metadata !4}
+!5 = metadata !{metadata !3, metadata !4}
+!6 = metadata !{metadata !6}
+!7 = metadata !{metadata !7}
diff --git a/test/Transforms/LoopVectorize/small-size.ll b/test/Transforms/LoopVectorize/X86/small-size.ll
index 35b91bb..f390b33 100644
--- a/test/Transforms/LoopVectorize/small-size.ll
+++ b/test/Transforms/LoopVectorize/X86/small-size.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -licm -S | FileCheck %s
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -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-S128"
 target triple = "x86_64-apple-macosx10.8.0"
diff --git a/test/Transforms/LoopVectorize/X86/unroll-small-loops.ll b/test/Transforms/LoopVectorize/X86/unroll-small-loops.ll
index 2075986..ef63a14 100644
--- a/test/Transforms/LoopVectorize/X86/unroll-small-loops.ll
+++ b/test/Transforms/LoopVectorize/X86/unroll-small-loops.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s  -loop-vectorize -mtriple=x86_64-apple-macosx10.8.0 -mcpu=corei7-avx2 -force-vector-width=4 -force-vector-unroll=0 -dce -S | FileCheck %s
+; RUN: opt < %s  -loop-vectorize -mtriple=x86_64-apple-macosx10.8.0 -mcpu=corei7-avx -force-vector-width=4 -force-vector-unroll=0 -dce -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-S128"
 target triple = "x86_64-apple-macosx10.8.0"
diff --git a/test/Transforms/LoopVectorize/X86/unroll_selection.ll b/test/Transforms/LoopVectorize/X86/unroll_selection.ll
new file mode 100644
index 0000000..2d7b663
--- /dev/null
+++ b/test/Transforms/LoopVectorize/X86/unroll_selection.ll
@@ -0,0 +1,71 @@
+; RUN: opt < %s  -loop-vectorize -mtriple=x86_64-apple-macosx10.8.0 -mcpu=corei7-avx -force-vector-width=4 -force-vector-unroll=0 -dce -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-S128"
+target triple = "x86_64-apple-macosx10.8.0"
+
+; Don't unroll when we have register pressure.
+;CHECK: reg_pressure
+;CHECK: load <4 x double>
+;CHECK-NOT: load  <4 x double>
+;CHECK: store <4 x double>
+;CHECK-NOT: store <4 x double>
+;CHECK: ret
+define void @reg_pressure(double* nocapture %A, i32 %n) nounwind uwtable ssp {
+  %1 = sext i32 %n to i64
+  br label %2
+
+; <label>:2                                       ; preds = %2, %0
+  %indvars.iv = phi i64 [ %indvars.iv.next, %2 ], [ %1, %0 ]
+  %3 = getelementptr inbounds double* %A, i64 %indvars.iv
+  %4 = load double* %3, align 8
+  %5 = fadd double %4, 3.000000e+00
+  %6 = fmul double %4, 2.000000e+00
+  %7 = fadd double %5, %6
+  %8 = fadd double %7, 2.000000e+00
+  %9 = fmul double %8, 5.000000e-01
+  %10 = fadd double %6, %9
+  %11 = fsub double %10, %5
+  %12 = fadd double %4, %11
+  %13 = fdiv double %8, %12
+  %14 = fmul double %13, %8
+  %15 = fmul double %6, %14
+  %16 = fmul double %5, %15
+  %17 = fadd double %16, -3.000000e+00
+  %18 = fsub double %4, %5
+  %19 = fadd double %6, %18
+  %20 = fadd double %13, %19
+  %21 = fadd double %20, %17
+  %22 = fadd double %21, 3.000000e+00
+  %23 = fmul double %4, %22
+  store double %23, double* %3, align 8
+  %indvars.iv.next = add i64 %indvars.iv, -1
+  %24 = trunc i64 %indvars.iv to i32
+  %25 = icmp eq i32 %24, 0
+  br i1 %25, label %26, label %2
+
+; <label>:26                                      ; preds = %2
+  ret void
+}
+
+; This is a small loop. Unroll it twice. 
+;CHECK: small_loop
+;CHECK: xor
+;CHECK: xor
+;CHECK: ret
+define void @small_loop(i16* nocapture %A, i64 %n) nounwind uwtable ssp {
+  %1 = icmp eq i64 %n, 0
+  br i1 %1, label %._crit_edge, label %.lr.ph
+
+.lr.ph:                                           ; preds = %0, %.lr.ph
+  %i.01 = phi i64 [ %5, %.lr.ph ], [ 0, %0 ]
+  %2 = getelementptr inbounds i16* %A, i64 %i.01
+  %3 = load i16* %2, align 2
+  %4 = xor i16 %3, 3
+  store i16 %4, i16* %2, align 2
+  %5 = add i64 %i.01, 1
+  %exitcond = icmp eq i64 %5, %n
+  br i1 %exitcond, label %._crit_edge, label %.lr.ph
+
+._crit_edge:                                      ; preds = %.lr.ph, %0
+  ret void
+}
diff --git a/test/Transforms/LoopVectorize/X86/vector_ptr_load_store.ll b/test/Transforms/LoopVectorize/X86/vector_ptr_load_store.ll
new file mode 100644
index 0000000..59bb8d0
--- /dev/null
+++ b/test/Transforms/LoopVectorize/X86/vector_ptr_load_store.ll
@@ -0,0 +1,150 @@
+; RUN: opt -loop-vectorize -mcpu=corei7-avx -debug -S < %s 2>&1 | FileCheck %s
+; REQUIRES: asserts
+
+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-S128"
+target triple = "x86_64-apple-macosx10.8.0"
+
+%0 = type { %0*, %1 }
+%1 = type { i8*, i32 }
+
+@p = global [2048 x [8 x i32*]] zeroinitializer, align 16
+@q = global [2048 x i16] zeroinitializer, align 16
+@r = global [2048 x i16] zeroinitializer, align 16
+
+; Tests for widest type
+; Ensure that we count the pointer store in the first test case. We have a
+; consecutive vector of pointers store, therefore we should count it towards the
+; widest vector count.
+;
+; CHECK: test_consecutive_store
+; CHECK: The Widest type: 64 bits
+define void @test_consecutive_store(%0**, %0**, %0** nocapture) nounwind ssp uwtable align 2 {
+  %4 = load %0** %2, align 8
+  %5 = icmp eq %0** %0, %1
+  br i1 %5, label %12, label %6
+
+; <label>:6                                       ; preds = %3
+  br label %7
+
+; <label>:7                                       ; preds = %7, %6
+  %8 = phi %0** [ %0, %6 ], [ %9, %7 ]
+  store %0* %4, %0** %8, align 8
+  %9 = getelementptr inbounds %0** %8, i64 1
+  %10 = icmp eq %0** %9, %1
+  br i1 %10, label %11, label %7
+
+; <label>:11                                      ; preds = %7
+  br label %12
+
+; <label>:12                                      ; preds = %11, %3
+  ret void
+}
+
+; However, if the store of a set of pointers is not to consecutive memory we do
+; NOT count the store towards the widest vector type.
+; In the test case below we add i16 types to store it in an array of pointer,
+; therefore the widest type should be i16.
+; int* p[2048][8];
+; short q[2048];
+;   for (int y = 0; y < 8; ++y)
+;     for (int i = 0; i < 1024; ++i) {
+;       p[i][y] = (int*) (1 + q[i]);
+;     }
+; CHECK: test_nonconsecutive_store
+; CHECK: The Widest type: 16 bits
+define void @test_nonconsecutive_store() nounwind ssp uwtable {
+  br label %1
+
+; <label>:1                                       ; preds = %14, %0
+  %2 = phi i64 [ 0, %0 ], [ %15, %14 ]
+  br label %3
+
+; <label>:3                                       ; preds = %3, %1
+  %4 = phi i64 [ 0, %1 ], [ %11, %3 ]
+  %5 = getelementptr inbounds [2048 x i16]* @q, i64 0, i64 %4
+  %6 = load i16* %5, align 2
+  %7 = sext i16 %6 to i64
+  %8 = add i64 %7, 1
+  %9 = inttoptr i64 %8 to i32*
+  %10 = getelementptr inbounds [2048 x [8 x i32*]]* @p, i64 0, i64 %4, i64 %2
+  store i32* %9, i32** %10, align 8
+  %11 = add i64 %4, 1
+  %12 = trunc i64 %11 to i32
+  %13 = icmp ne i32 %12, 1024
+  br i1 %13, label %3, label %14
+
+; <label>:14                                      ; preds = %3
+  %15 = add i64 %2, 1
+  %16 = trunc i64 %15 to i32
+  %17 = icmp ne i32 %16, 8
+  br i1 %17, label %1, label %18
+
+; <label>:18                                      ; preds = %14
+  ret void
+}
+
+
+@ia = global [1024 x i32*] zeroinitializer, align 16
+@ib = global [1024 x i32] zeroinitializer, align 16
+@ic = global [1024 x i8] zeroinitializer, align 16
+@p2 = global [2048 x [8 x i32*]] zeroinitializer, align 16
+@q2 = global [2048 x i16] zeroinitializer, align 16
+
+;; Now we check the same rules for loads. We should take consecutive loads of
+;; pointer types into account.
+; CHECK: test_consecutive_ptr_load
+; CHECK: The Widest type: 64 bits
+define i8 @test_consecutive_ptr_load() nounwind readonly ssp uwtable {
+  br label %1
+
+; <label>:1                                       ; preds = %1, %0
+  %2 = phi i64 [ 0, %0 ], [ %10, %1 ]
+  %3 = phi i8 [ 0, %0 ], [ %9, %1 ]
+  %4 = getelementptr inbounds [1024 x i32*]* @ia, i32 0, i64 %2
+  %5 = load i32** %4, align 4
+  %6 = ptrtoint i32* %5 to i64
+  %7 = trunc i64 %6 to i8
+  %8 = add i8 %3, 1
+  %9 = add i8 %7, %8
+  %10 = add i64 %2, 1
+  %11 = icmp ne i64 %10, 1024
+  br i1 %11, label %1, label %12
+
+; <label>:12                                      ; preds = %1
+  %13 = phi i8 [ %9, %1 ]
+  ret i8 %13
+}
+
+;; However, we should not take unconsecutive loads of pointers into account.
+; CHECK: test_nonconsecutive_ptr_load
+; CHECK: The Widest type: 16 bits
+define void @test_nonconsecutive_ptr_load() nounwind ssp uwtable {
+  br label %1
+
+; <label>:1                                       ; preds = %13, %0
+  %2 = phi i64 [ 0, %0 ], [ %14, %13 ]
+  br label %3
+
+; <label>:3                                       ; preds = %3, %1
+  %4 = phi i64 [ 0, %1 ], [ %10, %3 ]
+  %5 = getelementptr inbounds [2048 x [8 x i32*]]* @p2, i64 0, i64 %4, i64 %2
+  %6 = getelementptr inbounds [2048 x i16]* @q2, i64 0, i64 %4
+  %7 = load i32** %5, align 2
+  %8 = ptrtoint i32* %7 to i64
+  %9 = trunc i64 %8 to i16
+  store i16 %9, i16* %6, align 8
+  %10 = add i64 %4, 1
+  %11 = trunc i64 %10 to i32
+  %12 = icmp ne i32 %11, 1024
+  br i1 %12, label %3, label %13
+
+; <label>:13                                      ; preds = %3
+  %14 = add i64 %2, 1
+  %15 = trunc i64 %14 to i32
+  %16 = icmp ne i32 %15, 8
+  br i1 %16, label %1, label %17
+
+; <label>:17                                      ; preds = %13
+  ret void
+}
+
diff --git a/test/Transforms/LoopVectorize/bzip_reverse_loops.ll b/test/Transforms/LoopVectorize/bzip_reverse_loops.ll
new file mode 100644
index 0000000..431e422
--- /dev/null
+++ b/test/Transforms/LoopVectorize/bzip_reverse_loops.ll
@@ -0,0 +1,71 @@
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -S -enable-if-conversion | 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-S128"
+target triple = "x86_64-apple-macosx10.8.0"
+
+;CHECK: fc
+;CHECK: load <4 x i16>
+;CHECK-NEXT: shufflevector <4 x i16>
+;CHECK: select <4 x i1>
+;CHECK: store <4 x i16>
+;CHECK: ret
+define void @fc(i16* nocapture %p, i32 %n, i32 %size) nounwind uwtable ssp {
+entry:
+  br label %do.body
+
+do.body:                                          ; preds = %cond.end, %entry
+  %n.addr.0 = phi i32 [ %n, %entry ], [ %dec, %cond.end ]
+  %p.addr.0 = phi i16* [ %p, %entry ], [ %incdec.ptr, %cond.end ]
+  %incdec.ptr = getelementptr inbounds i16* %p.addr.0, i64 -1
+  %0 = load i16* %incdec.ptr, align 2, !tbaa !0
+  %conv = zext i16 %0 to i32
+  %cmp = icmp ult i32 %conv, %size
+  br i1 %cmp, label %cond.end, label %cond.true
+
+cond.true:                                        ; preds = %do.body
+  %sub = sub i32 %conv, %size
+  %phitmp = trunc i32 %sub to i16
+  br label %cond.end
+
+cond.end:                                         ; preds = %do.body, %cond.true
+  %cond = phi i16 [ %phitmp, %cond.true ], [ 0, %do.body ]
+  store i16 %cond, i16* %incdec.ptr, align 2, !tbaa !0
+  %dec = add i32 %n.addr.0, -1
+  %tobool = icmp eq i32 %dec, 0
+  br i1 %tobool, label %do.end, label %do.body
+
+do.end:                                           ; preds = %cond.end
+  ret void
+}
+
+;CHECK: example1
+;CHECK: load <4 x i32>
+;CHECK-NEXT: shufflevector <4 x i32>
+;CHECK: select <4 x i1>
+;CHECK: store <4 x i32>
+;CHECK: ret
+define void @example1(i32* nocapture %a, i32 %n, i32 %wsize) nounwind uwtable ssp {
+entry:
+  br label %do.body
+
+do.body:                                          ; preds = %do.body, %entry
+  %n.addr.0 = phi i32 [ %n, %entry ], [ %dec, %do.body ]
+  %p.0 = phi i32* [ %a, %entry ], [ %incdec.ptr, %do.body ]
+  %incdec.ptr = getelementptr inbounds i32* %p.0, i64 -1
+  %0 = load i32* %incdec.ptr, align 4, !tbaa !3
+  %cmp = icmp slt i32 %0, %wsize
+  %sub = sub nsw i32 %0, %wsize
+  %cond = select i1 %cmp, i32 0, i32 %sub
+  store i32 %cond, i32* %incdec.ptr, align 4, !tbaa !3
+  %dec = add nsw i32 %n.addr.0, -1
+  %tobool = icmp eq i32 %dec, 0
+  br i1 %tobool, label %do.end, label %do.body
+
+do.end:                                           ; preds = %do.body
+  ret void
+}
+
+!0 = metadata !{metadata !"short", metadata !1}
+!1 = metadata !{metadata !"omnipotent char", metadata !2}
+!2 = metadata !{metadata !"Simple C/C++ TBAA"}
+!3 = metadata !{metadata !"int", metadata !1}
diff --git a/test/Transforms/LoopVectorize/calloc.ll b/test/Transforms/LoopVectorize/calloc.ll
index 55c1eba..08c84ef 100644
--- a/test/Transforms/LoopVectorize/calloc.ll
+++ b/test/Transforms/LoopVectorize/calloc.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -licm -S | FileCheck %s
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -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-S128"
 target triple = "x86_64-apple-macosx10.9.0"
diff --git a/test/Transforms/LoopVectorize/cast-induction.ll b/test/Transforms/LoopVectorize/cast-induction.ll
index 5c090aa..2aa29ed 100644
--- a/test/Transforms/LoopVectorize/cast-induction.ll
+++ b/test/Transforms/LoopVectorize/cast-induction.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -licm -S | FileCheck %s
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -S | FileCheck %s
 
 ; rdar://problem/12848162
 
diff --git a/test/Transforms/LoopVectorize/cpp-new-array.ll b/test/Transforms/LoopVectorize/cpp-new-array.ll
index 7cd608d..da0fb05 100644
--- a/test/Transforms/LoopVectorize/cpp-new-array.ll
+++ b/test/Transforms/LoopVectorize/cpp-new-array.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -licm -S | FileCheck %s
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -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-S128"
 target triple = "x86_64-apple-macosx10.8.0"
diff --git a/test/Transforms/LoopVectorize/flags.ll b/test/Transforms/LoopVectorize/flags.ll
index b7f3815..656912e 100644
--- a/test/Transforms/LoopVectorize/flags.ll
+++ b/test/Transforms/LoopVectorize/flags.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -licm -S | FileCheck %s
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -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-S128"
 target triple = "x86_64-apple-macosx10.8.0"
diff --git a/test/Transforms/LoopVectorize/gcc-examples.ll b/test/Transforms/LoopVectorize/gcc-examples.ll
index b8b125f..f335557 100644
--- a/test/Transforms/LoopVectorize/gcc-examples.ll
+++ b/test/Transforms/LoopVectorize/gcc-examples.ll
@@ -1,5 +1,5 @@
-; RUN: opt < %s  -loop-vectorize -force-vector-width=4 -force-vector-unroll=1 -dce -instcombine -licm -S | FileCheck %s
-; RUN: opt < %s  -loop-vectorize -force-vector-width=4 -force-vector-unroll=4 -dce -instcombine -licm -S | FileCheck %s -check-prefix=UNROLL
+; RUN: opt < %s  -loop-vectorize -force-vector-width=4 -force-vector-unroll=1 -dce -instcombine -S | FileCheck %s
+; RUN: opt < %s  -loop-vectorize -force-vector-width=4 -force-vector-unroll=4 -dce -instcombine -S | FileCheck %s -check-prefix=UNROLL
 
 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-S128"
 target triple = "x86_64-apple-macosx10.8.0"
diff --git a/test/Transforms/LoopVectorize/global_alias.ll b/test/Transforms/LoopVectorize/global_alias.ll
new file mode 100644
index 0000000..24e698b
--- /dev/null
+++ b/test/Transforms/LoopVectorize/global_alias.ll
@@ -0,0 +1,1078 @@
+; RUN: opt < %s -O3 -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -S | FileCheck %s
+
+target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:64:128-a0:0:64-n32-S64"
+
+%struct.anon = type { [100 x i32], i32, [100 x i32] }
+%struct.anon.0 = type { [100 x [100 x i32]], i32, [100 x [100 x i32]] }
+
+@Foo = common global %struct.anon zeroinitializer, align 4
+@Bar = common global %struct.anon.0 zeroinitializer, align 4
+
+@PB = external global i32*
+@PA = external global i32*
+
+
+;; === First, the tests that should always vectorize, wither statically or by adding run-time checks ===
+
+
+; /// Different objects, positive induction, constant distance
+; int noAlias01 (int a) {
+;   int i;
+;   for (i=0; i<SIZE; i++)
+;     Foo.A[i] = Foo.B[i] + a;
+;   return Foo.A[a];
+; }
+; CHECK: define i32 @noAlias01
+; CHECK: add nsw <4 x i32>
+; CHECK ret
+
+define i32 @noAlias01(i32 %a) nounwind {
+entry:
+  %a.addr = alloca i32, align 4
+  %i = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  store i32 0, i32* %i, align 4
+  br label %for.cond
+
+for.cond:                                         ; preds = %for.inc, %entry
+  %0 = load i32* %i, align 4
+  %cmp = icmp slt i32 %0, 100
+  br i1 %cmp, label %for.body, label %for.end
+
+for.body:                                         ; preds = %for.cond
+  %1 = load i32* %i, align 4
+  %arrayidx = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 2), i32 0, i32 %1
+  %2 = load i32* %arrayidx, align 4
+  %3 = load i32* %a.addr, align 4
+  %add = add nsw i32 %2, %3
+  %4 = load i32* %i, align 4
+  %arrayidx1 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %4
+  store i32 %add, i32* %arrayidx1, align 4
+  br label %for.inc
+
+for.inc:                                          ; preds = %for.body
+  %5 = load i32* %i, align 4
+  %inc = add nsw i32 %5, 1
+  store i32 %inc, i32* %i, align 4
+  br label %for.cond
+
+for.end:                                          ; preds = %for.cond
+  %6 = load i32* %a.addr, align 4
+  %arrayidx2 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %6
+  %7 = load i32* %arrayidx2, align 4
+  ret i32 %7
+}
+
+; /// Different objects, positive induction with widening slide
+; int noAlias02 (int a) {
+;   int i;
+;   for (i=0; i<SIZE-10; i++)
+;     Foo.A[i] = Foo.B[i+10] + a;
+;   return Foo.A[a];
+; }
+; CHECK: define i32 @noAlias02
+; CHECK: add nsw <4 x i32>
+; CHECK ret
+
+define i32 @noAlias02(i32 %a) {
+entry:
+  %a.addr = alloca i32, align 4
+  %i = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  store i32 0, i32* %i, align 4
+  br label %for.cond
+
+for.cond:                                         ; preds = %for.inc, %entry
+  %0 = load i32* %i, align 4
+  %cmp = icmp slt i32 %0, 90
+  br i1 %cmp, label %for.body, label %for.end
+
+for.body:                                         ; preds = %for.cond
+  %1 = load i32* %i, align 4
+  %add = add nsw i32 %1, 10
+  %arrayidx = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 2), i32 0, i32 %add
+  %2 = load i32* %arrayidx, align 4
+  %3 = load i32* %a.addr, align 4
+  %add1 = add nsw i32 %2, %3
+  %4 = load i32* %i, align 4
+  %arrayidx2 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %4
+  store i32 %add1, i32* %arrayidx2, align 4
+  br label %for.inc
+
+for.inc:                                          ; preds = %for.body
+  %5 = load i32* %i, align 4
+  %inc = add nsw i32 %5, 1
+  store i32 %inc, i32* %i, align 4
+  br label %for.cond
+
+for.end:                                          ; preds = %for.cond
+  %6 = load i32* %a.addr, align 4
+  %arrayidx3 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %6
+  %7 = load i32* %arrayidx3, align 4
+  ret i32 %7
+}
+
+; /// Different objects, positive induction with shortening slide
+; int noAlias03 (int a) {
+;   int i;
+;   for (i=0; i<SIZE; i++)
+;     Foo.A[i+10] = Foo.B[i] + a;
+;   return Foo.A[a];
+; }
+; CHECK: define i32 @noAlias03
+; CHECK: add nsw <4 x i32>
+; CHECK ret
+
+define i32 @noAlias03(i32 %a) {
+entry:
+  %a.addr = alloca i32, align 4
+  %i = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  store i32 0, i32* %i, align 4
+  br label %for.cond
+
+for.cond:                                         ; preds = %for.inc, %entry
+  %0 = load i32* %i, align 4
+  %cmp = icmp slt i32 %0, 100
+  br i1 %cmp, label %for.body, label %for.end
+
+for.body:                                         ; preds = %for.cond
+  %1 = load i32* %i, align 4
+  %arrayidx = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 2), i32 0, i32 %1
+  %2 = load i32* %arrayidx, align 4
+  %3 = load i32* %a.addr, align 4
+  %add = add nsw i32 %2, %3
+  %4 = load i32* %i, align 4
+  %add1 = add nsw i32 %4, 10
+  %arrayidx2 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %add1
+  store i32 %add, i32* %arrayidx2, align 4
+  br label %for.inc
+
+for.inc:                                          ; preds = %for.body
+  %5 = load i32* %i, align 4
+  %inc = add nsw i32 %5, 1
+  store i32 %inc, i32* %i, align 4
+  br label %for.cond
+
+for.end:                                          ; preds = %for.cond
+  %6 = load i32* %a.addr, align 4
+  %arrayidx3 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %6
+  %7 = load i32* %arrayidx3, align 4
+  ret i32 %7
+}
+
+; /// Pointer access, positive stride, run-time check added
+; int noAlias04 (int a) {
+;   int i;
+;   for (i=0; i<SIZE; i++)
+;     *(PA+i) = *(PB+i) + a;
+;   return *(PA+a);
+; }
+; CHECK: define i32 @noAlias04
+; CHECK-NOT: add nsw <4 x i32>
+; CHECK ret
+;
+; TODO: This test vectorizes (with run-time check) on real targets with -O3)
+; Check why it's not being vectorized even when forcing vectorization
+
+define i32 @noAlias04(i32 %a) #0 {
+entry:
+  %a.addr = alloca i32, align 4
+  %i = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  store i32 0, i32* %i, align 4
+  br label %for.cond
+
+for.cond:                                         ; preds = %for.inc, %entry
+  %0 = load i32* %i, align 4
+  %cmp = icmp slt i32 %0, 100
+  br i1 %cmp, label %for.body, label %for.end
+
+for.body:                                         ; preds = %for.cond
+  %1 = load i32** @PB, align 4
+  %2 = load i32* %i, align 4
+  %add.ptr = getelementptr inbounds i32* %1, i32 %2
+  %3 = load i32* %add.ptr, align 4
+  %4 = load i32* %a.addr, align 4
+  %add = add nsw i32 %3, %4
+  %5 = load i32** @PA, align 4
+  %6 = load i32* %i, align 4
+  %add.ptr1 = getelementptr inbounds i32* %5, i32 %6
+  store i32 %add, i32* %add.ptr1, align 4
+  br label %for.inc
+
+for.inc:                                          ; preds = %for.body
+  %7 = load i32* %i, align 4
+  %inc = add nsw i32 %7, 1
+  store i32 %inc, i32* %i, align 4
+  br label %for.cond
+
+for.end:                                          ; preds = %for.cond
+  %8 = load i32** @PA, align 4
+  %9 = load i32* %a.addr, align 4
+  %add.ptr2 = getelementptr inbounds i32* %8, i32 %9
+  %10 = load i32* %add.ptr2, align 4
+  ret i32 %10
+}
+
+; /// Different objects, positive induction, multi-array
+; int noAlias05 (int a) {
+;   int i, N=10;
+;   for (i=0; i<SIZE; i++)
+;     Bar.A[N][i] = Bar.B[N][i] + a;
+;   return Bar.A[N][a];
+; }
+; CHECK: define i32 @noAlias05
+; CHECK: add nsw <4 x i32>
+; CHECK ret
+
+define i32 @noAlias05(i32 %a) #0 {
+entry:
+  %a.addr = alloca i32, align 4
+  %i = alloca i32, align 4
+  %N = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  store i32 10, i32* %N, align 4
+  store i32 0, i32* %i, align 4
+  br label %for.cond
+
+for.cond:                                         ; preds = %for.inc, %entry
+  %0 = load i32* %i, align 4
+  %cmp = icmp slt i32 %0, 100
+  br i1 %cmp, label %for.body, label %for.end
+
+for.body:                                         ; preds = %for.cond
+  %1 = load i32* %i, align 4
+  %2 = load i32* %N, align 4
+  %arrayidx = getelementptr inbounds [100 x [100 x i32]]* getelementptr inbounds (%struct.anon.0* @Bar, i32 0, i32 2), i32 0, i32 %2
+  %arrayidx1 = getelementptr inbounds [100 x i32]* %arrayidx, i32 0, i32 %1
+  %3 = load i32* %arrayidx1, align 4
+  %4 = load i32* %a.addr, align 4
+  %add = add nsw i32 %3, %4
+  %5 = load i32* %i, align 4
+  %6 = load i32* %N, align 4
+  %arrayidx2 = getelementptr inbounds [100 x [100 x i32]]* getelementptr inbounds (%struct.anon.0* @Bar, i32 0, i32 0), i32 0, i32 %6
+  %arrayidx3 = getelementptr inbounds [100 x i32]* %arrayidx2, i32 0, i32 %5
+  store i32 %add, i32* %arrayidx3, align 4
+  br label %for.inc
+
+for.inc:                                          ; preds = %for.body
+  %7 = load i32* %i, align 4
+  %inc = add nsw i32 %7, 1
+  store i32 %inc, i32* %i, align 4
+  br label %for.cond
+
+for.end:                                          ; preds = %for.cond
+  %8 = load i32* %a.addr, align 4
+  %9 = load i32* %N, align 4
+  %arrayidx4 = getelementptr inbounds [100 x [100 x i32]]* getelementptr inbounds (%struct.anon.0* @Bar, i32 0, i32 0), i32 0, i32 %9
+  %arrayidx5 = getelementptr inbounds [100 x i32]* %arrayidx4, i32 0, i32 %8
+  %10 = load i32* %arrayidx5, align 4
+  ret i32 %10
+}
+
+; /// Same objects, positive induction, multi-array, different sub-elements
+; int noAlias06 (int a) {
+;   int i, N=10;
+;   for (i=0; i<SIZE; i++)
+;     Bar.A[N][i] = Bar.A[N+1][i] + a;
+;   return Bar.A[N][a];
+; }
+; CHECK: define i32 @noAlias06
+; CHECK: add nsw <4 x i32>
+; CHECK ret
+
+define i32 @noAlias06(i32 %a) #0 {
+entry:
+  %a.addr = alloca i32, align 4
+  %i = alloca i32, align 4
+  %N = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  store i32 10, i32* %N, align 4
+  store i32 0, i32* %i, align 4
+  br label %for.cond
+
+for.cond:                                         ; preds = %for.inc, %entry
+  %0 = load i32* %i, align 4
+  %cmp = icmp slt i32 %0, 100
+  br i1 %cmp, label %for.body, label %for.end
+
+for.body:                                         ; preds = %for.cond
+  %1 = load i32* %i, align 4
+  %2 = load i32* %N, align 4
+  %add = add nsw i32 %2, 1
+  %arrayidx = getelementptr inbounds [100 x [100 x i32]]* getelementptr inbounds (%struct.anon.0* @Bar, i32 0, i32 0), i32 0, i32 %add
+  %arrayidx1 = getelementptr inbounds [100 x i32]* %arrayidx, i32 0, i32 %1
+  %3 = load i32* %arrayidx1, align 4
+  %4 = load i32* %a.addr, align 4
+  %add2 = add nsw i32 %3, %4
+  %5 = load i32* %i, align 4
+  %6 = load i32* %N, align 4
+  %arrayidx3 = getelementptr inbounds [100 x [100 x i32]]* getelementptr inbounds (%struct.anon.0* @Bar, i32 0, i32 0), i32 0, i32 %6
+  %arrayidx4 = getelementptr inbounds [100 x i32]* %arrayidx3, i32 0, i32 %5
+  store i32 %add2, i32* %arrayidx4, align 4
+  br label %for.inc
+
+for.inc:                                          ; preds = %for.body
+  %7 = load i32* %i, align 4
+  %inc = add nsw i32 %7, 1
+  store i32 %inc, i32* %i, align 4
+  br label %for.cond
+
+for.end:                                          ; preds = %for.cond
+  %8 = load i32* %a.addr, align 4
+  %9 = load i32* %N, align 4
+  %arrayidx5 = getelementptr inbounds [100 x [100 x i32]]* getelementptr inbounds (%struct.anon.0* @Bar, i32 0, i32 0), i32 0, i32 %9
+  %arrayidx6 = getelementptr inbounds [100 x i32]* %arrayidx5, i32 0, i32 %8
+  %10 = load i32* %arrayidx6, align 4
+  ret i32 %10
+}
+
+; /// Different objects, negative induction, constant distance
+; int noAlias07 (int a) {
+;   int i;
+;   for (i=0; i<SIZE; i++)
+;     Foo.A[SIZE-i-1] = Foo.B[SIZE-i-1] + a;
+;   return Foo.A[a];
+; }
+; CHECK: define i32 @noAlias07
+; CHECK: sub nsw <4 x i32>
+; CHECK ret
+
+define i32 @noAlias07(i32 %a) #0 {
+entry:
+  %a.addr = alloca i32, align 4
+  %i = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  store i32 0, i32* %i, align 4
+  br label %for.cond
+
+for.cond:                                         ; preds = %for.inc, %entry
+  %0 = load i32* %i, align 4
+  %cmp = icmp slt i32 %0, 100
+  br i1 %cmp, label %for.body, label %for.end
+
+for.body:                                         ; preds = %for.cond
+  %1 = load i32* %i, align 4
+  %sub = sub nsw i32 100, %1
+  %sub1 = sub nsw i32 %sub, 1
+  %arrayidx = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 2), i32 0, i32 %sub1
+  %2 = load i32* %arrayidx, align 4
+  %3 = load i32* %a.addr, align 4
+  %add = add nsw i32 %2, %3
+  %4 = load i32* %i, align 4
+  %sub2 = sub nsw i32 100, %4
+  %sub3 = sub nsw i32 %sub2, 1
+  %arrayidx4 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %sub3
+  store i32 %add, i32* %arrayidx4, align 4
+  br label %for.inc
+
+for.inc:                                          ; preds = %for.body
+  %5 = load i32* %i, align 4
+  %inc = add nsw i32 %5, 1
+  store i32 %inc, i32* %i, align 4
+  br label %for.cond
+
+for.end:                                          ; preds = %for.cond
+  %6 = load i32* %a.addr, align 4
+  %arrayidx5 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %6
+  %7 = load i32* %arrayidx5, align 4
+  ret i32 %7
+}
+
+; /// Different objects, negative induction, shortening slide
+; int noAlias08 (int a) {
+;   int i;
+;   for (i=0; i<SIZE-10; i++)
+;     Foo.A[SIZE-i-1] = Foo.B[SIZE-i-10] + a;
+;   return Foo.A[a];
+; }
+; CHECK: define i32 @noAlias08
+; CHECK: sub nsw <4 x i32>
+; CHECK ret
+
+define i32 @noAlias08(i32 %a) #0 {
+entry:
+  %a.addr = alloca i32, align 4
+  %i = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  store i32 0, i32* %i, align 4
+  br label %for.cond
+
+for.cond:                                         ; preds = %for.inc, %entry
+  %0 = load i32* %i, align 4
+  %cmp = icmp slt i32 %0, 90
+  br i1 %cmp, label %for.body, label %for.end
+
+for.body:                                         ; preds = %for.cond
+  %1 = load i32* %i, align 4
+  %sub = sub nsw i32 100, %1
+  %sub1 = sub nsw i32 %sub, 10
+  %arrayidx = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 2), i32 0, i32 %sub1
+  %2 = load i32* %arrayidx, align 4
+  %3 = load i32* %a.addr, align 4
+  %add = add nsw i32 %2, %3
+  %4 = load i32* %i, align 4
+  %sub2 = sub nsw i32 100, %4
+  %sub3 = sub nsw i32 %sub2, 1
+  %arrayidx4 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %sub3
+  store i32 %add, i32* %arrayidx4, align 4
+  br label %for.inc
+
+for.inc:                                          ; preds = %for.body
+  %5 = load i32* %i, align 4
+  %inc = add nsw i32 %5, 1
+  store i32 %inc, i32* %i, align 4
+  br label %for.cond
+
+for.end:                                          ; preds = %for.cond
+  %6 = load i32* %a.addr, align 4
+  %arrayidx5 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %6
+  %7 = load i32* %arrayidx5, align 4
+  ret i32 %7
+}
+
+; /// Different objects, negative induction, widening slide
+; int noAlias09 (int a) {
+;   int i;
+;   for (i=0; i<SIZE; i++)
+;     Foo.A[SIZE-i-10] = Foo.B[SIZE-i-1] + a;
+;   return Foo.A[a];
+; }
+; CHECK: define i32 @noAlias09
+; CHECK: sub nsw <4 x i32>
+; CHECK ret
+
+define i32 @noAlias09(i32 %a) #0 {
+entry:
+  %a.addr = alloca i32, align 4
+  %i = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  store i32 0, i32* %i, align 4
+  br label %for.cond
+
+for.cond:                                         ; preds = %for.inc, %entry
+  %0 = load i32* %i, align 4
+  %cmp = icmp slt i32 %0, 100
+  br i1 %cmp, label %for.body, label %for.end
+
+for.body:                                         ; preds = %for.cond
+  %1 = load i32* %i, align 4
+  %sub = sub nsw i32 100, %1
+  %sub1 = sub nsw i32 %sub, 1
+  %arrayidx = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 2), i32 0, i32 %sub1
+  %2 = load i32* %arrayidx, align 4
+  %3 = load i32* %a.addr, align 4
+  %add = add nsw i32 %2, %3
+  %4 = load i32* %i, align 4
+  %sub2 = sub nsw i32 100, %4
+  %sub3 = sub nsw i32 %sub2, 10
+  %arrayidx4 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %sub3
+  store i32 %add, i32* %arrayidx4, align 4
+  br label %for.inc
+
+for.inc:                                          ; preds = %for.body
+  %5 = load i32* %i, align 4
+  %inc = add nsw i32 %5, 1
+  store i32 %inc, i32* %i, align 4
+  br label %for.cond
+
+for.end:                                          ; preds = %for.cond
+  %6 = load i32* %a.addr, align 4
+  %arrayidx5 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %6
+  %7 = load i32* %arrayidx5, align 4
+  ret i32 %7
+}
+
+; /// Pointer access, negative stride, run-time check added
+; int noAlias10 (int a) {
+;   int i;
+;   for (i=0; i<SIZE; i++)
+;     *(PA+SIZE-i-1) = *(PB+SIZE-i-1) + a;
+;   return *(PA+a);
+; }
+; CHECK: define i32 @noAlias10
+; CHECK-NOT: sub nsw <4 x i32>
+; CHECK ret
+;
+; TODO: This test vectorizes (with run-time check) on real targets with -O3)
+; Check why it's not being vectorized even when forcing vectorization
+
+define i32 @noAlias10(i32 %a) #0 {
+entry:
+  %a.addr = alloca i32, align 4
+  %i = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  store i32 0, i32* %i, align 4
+  br label %for.cond
+
+for.cond:                                         ; preds = %for.inc, %entry
+  %0 = load i32* %i, align 4
+  %cmp = icmp slt i32 %0, 100
+  br i1 %cmp, label %for.body, label %for.end
+
+for.body:                                         ; preds = %for.cond
+  %1 = load i32** @PB, align 4
+  %add.ptr = getelementptr inbounds i32* %1, i32 100
+  %2 = load i32* %i, align 4
+  %idx.neg = sub i32 0, %2
+  %add.ptr1 = getelementptr inbounds i32* %add.ptr, i32 %idx.neg
+  %add.ptr2 = getelementptr inbounds i32* %add.ptr1, i32 -1
+  %3 = load i32* %add.ptr2, align 4
+  %4 = load i32* %a.addr, align 4
+  %add = add nsw i32 %3, %4
+  %5 = load i32** @PA, align 4
+  %add.ptr3 = getelementptr inbounds i32* %5, i32 100
+  %6 = load i32* %i, align 4
+  %idx.neg4 = sub i32 0, %6
+  %add.ptr5 = getelementptr inbounds i32* %add.ptr3, i32 %idx.neg4
+  %add.ptr6 = getelementptr inbounds i32* %add.ptr5, i32 -1
+  store i32 %add, i32* %add.ptr6, align 4
+  br label %for.inc
+
+for.inc:                                          ; preds = %for.body
+  %7 = load i32* %i, align 4
+  %inc = add nsw i32 %7, 1
+  store i32 %inc, i32* %i, align 4
+  br label %for.cond
+
+for.end:                                          ; preds = %for.cond
+  %8 = load i32** @PA, align 4
+  %9 = load i32* %a.addr, align 4
+  %add.ptr7 = getelementptr inbounds i32* %8, i32 %9
+  %10 = load i32* %add.ptr7, align 4
+  ret i32 %10
+}
+
+; /// Different objects, negative induction, multi-array
+; int noAlias11 (int a) {
+;   int i, N=10;
+;   for (i=0; i<SIZE; i++)
+;     Bar.A[N][SIZE-i-1] = Bar.B[N][SIZE-i-1] + a;
+;   return Bar.A[N][a];
+; }
+; CHECK: define i32 @noAlias11
+; CHECK: sub nsw <4 x i32>
+; CHECK ret
+
+define i32 @noAlias11(i32 %a) #0 {
+entry:
+  %a.addr = alloca i32, align 4
+  %i = alloca i32, align 4
+  %N = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  store i32 10, i32* %N, align 4
+  store i32 0, i32* %i, align 4
+  br label %for.cond
+
+for.cond:                                         ; preds = %for.inc, %entry
+  %0 = load i32* %i, align 4
+  %cmp = icmp slt i32 %0, 100
+  br i1 %cmp, label %for.body, label %for.end
+
+for.body:                                         ; preds = %for.cond
+  %1 = load i32* %i, align 4
+  %sub = sub nsw i32 100, %1
+  %sub1 = sub nsw i32 %sub, 1
+  %2 = load i32* %N, align 4
+  %arrayidx = getelementptr inbounds [100 x [100 x i32]]* getelementptr inbounds (%struct.anon.0* @Bar, i32 0, i32 2), i32 0, i32 %2
+  %arrayidx2 = getelementptr inbounds [100 x i32]* %arrayidx, i32 0, i32 %sub1
+  %3 = load i32* %arrayidx2, align 4
+  %4 = load i32* %a.addr, align 4
+  %add = add nsw i32 %3, %4
+  %5 = load i32* %i, align 4
+  %sub3 = sub nsw i32 100, %5
+  %sub4 = sub nsw i32 %sub3, 1
+  %6 = load i32* %N, align 4
+  %arrayidx5 = getelementptr inbounds [100 x [100 x i32]]* getelementptr inbounds (%struct.anon.0* @Bar, i32 0, i32 0), i32 0, i32 %6
+  %arrayidx6 = getelementptr inbounds [100 x i32]* %arrayidx5, i32 0, i32 %sub4
+  store i32 %add, i32* %arrayidx6, align 4
+  br label %for.inc
+
+for.inc:                                          ; preds = %for.body
+  %7 = load i32* %i, align 4
+  %inc = add nsw i32 %7, 1
+  store i32 %inc, i32* %i, align 4
+  br label %for.cond
+
+for.end:                                          ; preds = %for.cond
+  %8 = load i32* %a.addr, align 4
+  %9 = load i32* %N, align 4
+  %arrayidx7 = getelementptr inbounds [100 x [100 x i32]]* getelementptr inbounds (%struct.anon.0* @Bar, i32 0, i32 0), i32 0, i32 %9
+  %arrayidx8 = getelementptr inbounds [100 x i32]* %arrayidx7, i32 0, i32 %8
+  %10 = load i32* %arrayidx8, align 4
+  ret i32 %10
+}
+
+; /// Same objects, negative induction, multi-array, different sub-elements
+; int noAlias12 (int a) {
+;   int i, N=10;
+;   for (i=0; i<SIZE; i++)
+;     Bar.A[N][SIZE-i-1] = Bar.A[N+1][SIZE-i-1] + a;
+;   return Bar.A[N][a];
+; }
+; CHECK: define i32 @noAlias12
+; CHECK: sub nsw <4 x i32>
+; CHECK ret
+
+define i32 @noAlias12(i32 %a) #0 {
+entry:
+  %a.addr = alloca i32, align 4
+  %i = alloca i32, align 4
+  %N = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  store i32 10, i32* %N, align 4
+  store i32 0, i32* %i, align 4
+  br label %for.cond
+
+for.cond:                                         ; preds = %for.inc, %entry
+  %0 = load i32* %i, align 4
+  %cmp = icmp slt i32 %0, 100
+  br i1 %cmp, label %for.body, label %for.end
+
+for.body:                                         ; preds = %for.cond
+  %1 = load i32* %i, align 4
+  %sub = sub nsw i32 100, %1
+  %sub1 = sub nsw i32 %sub, 1
+  %2 = load i32* %N, align 4
+  %add = add nsw i32 %2, 1
+  %arrayidx = getelementptr inbounds [100 x [100 x i32]]* getelementptr inbounds (%struct.anon.0* @Bar, i32 0, i32 0), i32 0, i32 %add
+  %arrayidx2 = getelementptr inbounds [100 x i32]* %arrayidx, i32 0, i32 %sub1
+  %3 = load i32* %arrayidx2, align 4
+  %4 = load i32* %a.addr, align 4
+  %add3 = add nsw i32 %3, %4
+  %5 = load i32* %i, align 4
+  %sub4 = sub nsw i32 100, %5
+  %sub5 = sub nsw i32 %sub4, 1
+  %6 = load i32* %N, align 4
+  %arrayidx6 = getelementptr inbounds [100 x [100 x i32]]* getelementptr inbounds (%struct.anon.0* @Bar, i32 0, i32 0), i32 0, i32 %6
+  %arrayidx7 = getelementptr inbounds [100 x i32]* %arrayidx6, i32 0, i32 %sub5
+  store i32 %add3, i32* %arrayidx7, align 4
+  br label %for.inc
+
+for.inc:                                          ; preds = %for.body
+  %7 = load i32* %i, align 4
+  %inc = add nsw i32 %7, 1
+  store i32 %inc, i32* %i, align 4
+  br label %for.cond
+
+for.end:                                          ; preds = %for.cond
+  %8 = load i32* %a.addr, align 4
+  %9 = load i32* %N, align 4
+  %arrayidx8 = getelementptr inbounds [100 x [100 x i32]]* getelementptr inbounds (%struct.anon.0* @Bar, i32 0, i32 0), i32 0, i32 %9
+  %arrayidx9 = getelementptr inbounds [100 x i32]* %arrayidx8, i32 0, i32 %8
+  %10 = load i32* %arrayidx9, align 4
+  ret i32 %10
+}
+
+; /// Same objects, positive induction, constant distance, just enough for vector size
+; int noAlias13 (int a) {
+;   int i;
+;   for (i=0; i<SIZE; i++)
+;     Foo.A[i] = Foo.A[i+4] + a;
+;   return Foo.A[a];
+; }
+; CHECK: define i32 @noAlias13
+; CHECK: add nsw <4 x i32>
+; CHECK ret
+
+define i32 @noAlias13(i32 %a) #0 {
+entry:
+  %a.addr = alloca i32, align 4
+  %i = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  store i32 0, i32* %i, align 4
+  br label %for.cond
+
+for.cond:                                         ; preds = %for.inc, %entry
+  %0 = load i32* %i, align 4
+  %cmp = icmp slt i32 %0, 100
+  br i1 %cmp, label %for.body, label %for.end
+
+for.body:                                         ; preds = %for.cond
+  %1 = load i32* %i, align 4
+  %add = add nsw i32 %1, 4
+  %arrayidx = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %add
+  %2 = load i32* %arrayidx, align 4
+  %3 = load i32* %a.addr, align 4
+  %add1 = add nsw i32 %2, %3
+  %4 = load i32* %i, align 4
+  %arrayidx2 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %4
+  store i32 %add1, i32* %arrayidx2, align 4
+  br label %for.inc
+
+for.inc:                                          ; preds = %for.body
+  %5 = load i32* %i, align 4
+  %inc = add nsw i32 %5, 1
+  store i32 %inc, i32* %i, align 4
+  br label %for.cond
+
+for.end:                                          ; preds = %for.cond
+  %6 = load i32* %a.addr, align 4
+  %arrayidx3 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %6
+  %7 = load i32* %arrayidx3, align 4
+  ret i32 %7
+}
+
+; /// Same objects, negative induction, constant distance, just enough for vector size
+; int noAlias14 (int a) {
+;   int i;
+;   for (i=0; i<SIZE; i++)
+;     Foo.A[SIZE-i-1] = Foo.A[SIZE-i-5] + a;
+;   return Foo.A[a];
+; }
+; CHECK: define i32 @noAlias14
+; CHECK: sub nsw <4 x i32>
+; CHECK ret
+
+define i32 @noAlias14(i32 %a) #0 {
+entry:
+  %a.addr = alloca i32, align 4
+  %i = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  store i32 0, i32* %i, align 4
+  br label %for.cond
+
+for.cond:                                         ; preds = %for.inc, %entry
+  %0 = load i32* %i, align 4
+  %cmp = icmp slt i32 %0, 100
+  br i1 %cmp, label %for.body, label %for.end
+
+for.body:                                         ; preds = %for.cond
+  %1 = load i32* %i, align 4
+  %sub = sub nsw i32 100, %1
+  %sub1 = sub nsw i32 %sub, 5
+  %arrayidx = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %sub1
+  %2 = load i32* %arrayidx, align 4
+  %3 = load i32* %a.addr, align 4
+  %add = add nsw i32 %2, %3
+  %4 = load i32* %i, align 4
+  %sub2 = sub nsw i32 100, %4
+  %sub3 = sub nsw i32 %sub2, 1
+  %arrayidx4 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %sub3
+  store i32 %add, i32* %arrayidx4, align 4
+  br label %for.inc
+
+for.inc:                                          ; preds = %for.body
+  %5 = load i32* %i, align 4
+  %inc = add nsw i32 %5, 1
+  store i32 %inc, i32* %i, align 4
+  br label %for.cond
+
+for.end:                                          ; preds = %for.cond
+  %6 = load i32* %a.addr, align 4
+  %arrayidx5 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %6
+  %7 = load i32* %arrayidx5, align 4
+  ret i32 %7
+}
+
+
+;; === Now, the tests that we could vectorize with induction changes or run-time checks ===
+
+
+; /// Different objects, swapped induction, alias at the end
+; int mayAlias01 (int a) {
+;   int i;
+;   for (i=0; i<SIZE; i++)
+;     Foo.A[i] = Foo.B[SIZE-i-1] + a;
+;   return Foo.A[a];
+; }
+; CHECK: define i32 @mayAlias01
+; CHECK-NOT: add nsw <4 x i32>
+; CHECK ret
+
+define i32 @mayAlias01(i32 %a) nounwind {
+entry:
+  %a.addr = alloca i32, align 4
+  %i = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  store i32 0, i32* %i, align 4
+  br label %for.cond
+
+for.cond:                                         ; preds = %for.inc, %entry
+  %0 = load i32* %i, align 4
+  %cmp = icmp slt i32 %0, 100
+  br i1 %cmp, label %for.body, label %for.end
+
+for.body:                                         ; preds = %for.cond
+  %1 = load i32* %i, align 4
+  %sub = sub nsw i32 100, %1
+  %sub1 = sub nsw i32 %sub, 1
+  %arrayidx = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 2), i32 0, i32 %sub1
+  %2 = load i32* %arrayidx, align 4
+  %3 = load i32* %a.addr, align 4
+  %add = add nsw i32 %2, %3
+  %4 = load i32* %i, align 4
+  %arrayidx2 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %4
+  store i32 %add, i32* %arrayidx2, align 4
+  br label %for.inc
+
+for.inc:                                          ; preds = %for.body
+  %5 = load i32* %i, align 4
+  %inc = add nsw i32 %5, 1
+  store i32 %inc, i32* %i, align 4
+  br label %for.cond
+
+for.end:                                          ; preds = %for.cond
+  %6 = load i32* %a.addr, align 4
+  %arrayidx3 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %6
+  %7 = load i32* %arrayidx3, align 4
+  ret i32 %7
+}
+
+; /// Different objects, swapped induction, alias at the beginning
+; int mayAlias02 (int a) {
+;   int i;
+;   for (i=0; i<SIZE; i++)
+;     Foo.A[SIZE-i-1] = Foo.B[i] + a;
+;   return Foo.A[a];
+; }
+; CHECK: define i32 @mayAlias02
+; CHECK-NOT: add nsw <4 x i32>
+; CHECK ret
+
+define i32 @mayAlias02(i32 %a) nounwind {
+entry:
+  %a.addr = alloca i32, align 4
+  %i = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  store i32 0, i32* %i, align 4
+  br label %for.cond
+
+for.cond:                                         ; preds = %for.inc, %entry
+  %0 = load i32* %i, align 4
+  %cmp = icmp slt i32 %0, 100
+  br i1 %cmp, label %for.body, label %for.end
+
+for.body:                                         ; preds = %for.cond
+  %1 = load i32* %i, align 4
+  %arrayidx = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 2), i32 0, i32 %1
+  %2 = load i32* %arrayidx, align 4
+  %3 = load i32* %a.addr, align 4
+  %add = add nsw i32 %2, %3
+  %4 = load i32* %i, align 4
+  %sub = sub nsw i32 100, %4
+  %sub1 = sub nsw i32 %sub, 1
+  %arrayidx2 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %sub1
+  store i32 %add, i32* %arrayidx2, align 4
+  br label %for.inc
+
+for.inc:                                          ; preds = %for.body
+  %5 = load i32* %i, align 4
+  %inc = add nsw i32 %5, 1
+  store i32 %inc, i32* %i, align 4
+  br label %for.cond
+
+for.end:                                          ; preds = %for.cond
+  %6 = load i32* %a.addr, align 4
+  %arrayidx3 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %6
+  %7 = load i32* %arrayidx3, align 4
+  ret i32 %7
+}
+
+; /// Pointer access, run-time check added
+; int mayAlias03 (int a) {
+;   int i;
+;   for (i=0; i<SIZE; i++)
+;     *(PA+i) = *(PB+SIZE-i-1) + a;
+;   return *(PA+a);
+; }
+; CHECK: define i32 @mayAlias03
+; CHECK-NOT: add nsw <4 x i32>
+; CHECK ret
+
+define i32 @mayAlias03(i32 %a) nounwind {
+entry:
+  %a.addr = alloca i32, align 4
+  %i = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  store i32 0, i32* %i, align 4
+  br label %for.cond
+
+for.cond:                                         ; preds = %for.inc, %entry
+  %0 = load i32* %i, align 4
+  %cmp = icmp slt i32 %0, 100
+  br i1 %cmp, label %for.body, label %for.end
+
+for.body:                                         ; preds = %for.cond
+  %1 = load i32** @PB, align 4
+  %add.ptr = getelementptr inbounds i32* %1, i32 100
+  %2 = load i32* %i, align 4
+  %idx.neg = sub i32 0, %2
+  %add.ptr1 = getelementptr inbounds i32* %add.ptr, i32 %idx.neg
+  %add.ptr2 = getelementptr inbounds i32* %add.ptr1, i32 -1
+  %3 = load i32* %add.ptr2, align 4
+  %4 = load i32* %a.addr, align 4
+  %add = add nsw i32 %3, %4
+  %5 = load i32** @PA, align 4
+  %6 = load i32* %i, align 4
+  %add.ptr3 = getelementptr inbounds i32* %5, i32 %6
+  store i32 %add, i32* %add.ptr3, align 4
+  br label %for.inc
+
+for.inc:                                          ; preds = %for.body
+  %7 = load i32* %i, align 4
+  %inc = add nsw i32 %7, 1
+  store i32 %inc, i32* %i, align 4
+  br label %for.cond
+
+for.end:                                          ; preds = %for.cond
+  %8 = load i32** @PA, align 4
+  %9 = load i32* %a.addr, align 4
+  %add.ptr4 = getelementptr inbounds i32* %8, i32 %9
+  %10 = load i32* %add.ptr4, align 4
+  ret i32 %10
+}
+
+
+;; === Finally, the tests that should only vectorize with care (or if we ignore undefined behaviour at all) ===
+
+
+; int mustAlias01 (int a) {
+;   int i;
+;   for (i=0; i<SIZE; i++)
+;     Foo.A[i+10] = Foo.B[SIZE-i-1] + a;
+;   return Foo.A[a];
+; }
+; CHECK: define i32 @mustAlias01
+; CHECK-NOT: add nsw <4 x i32>
+; CHECK ret
+
+define i32 @mustAlias01(i32 %a) nounwind {
+entry:
+  %a.addr = alloca i32, align 4
+  %i = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  store i32 0, i32* %i, align 4
+  br label %for.cond
+
+for.cond:                                         ; preds = %for.inc, %entry
+  %0 = load i32* %i, align 4
+  %cmp = icmp slt i32 %0, 100
+  br i1 %cmp, label %for.body, label %for.end
+
+for.body:                                         ; preds = %for.cond
+  %1 = load i32* %i, align 4
+  %sub = sub nsw i32 100, %1
+  %sub1 = sub nsw i32 %sub, 1
+  %arrayidx = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 2), i32 0, i32 %sub1
+  %2 = load i32* %arrayidx, align 4
+  %3 = load i32* %a.addr, align 4
+  %add = add nsw i32 %2, %3
+  %4 = load i32* %i, align 4
+  %add2 = add nsw i32 %4, 10
+  %arrayidx3 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %add2
+  store i32 %add, i32* %arrayidx3, align 4
+  br label %for.inc
+
+for.inc:                                          ; preds = %for.body
+  %5 = load i32* %i, align 4
+  %inc = add nsw i32 %5, 1
+  store i32 %inc, i32* %i, align 4
+  br label %for.cond
+
+for.end:                                          ; preds = %for.cond
+  %6 = load i32* %a.addr, align 4
+  %arrayidx4 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %6
+  %7 = load i32* %arrayidx4, align 4
+  ret i32 %7
+}
+
+; int mustAlias02 (int a) {
+;   int i;
+;   for (i=0; i<SIZE; i++)
+;     Foo.A[i] = Foo.B[SIZE-i-10] + a;
+;   return Foo.A[a];
+; }
+; CHECK: define i32 @mustAlias02
+; CHECK-NOT: add nsw <4 x i32>
+; CHECK ret
+
+define i32 @mustAlias02(i32 %a) nounwind {
+entry:
+  %a.addr = alloca i32, align 4
+  %i = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  store i32 0, i32* %i, align 4
+  br label %for.cond
+
+for.cond:                                         ; preds = %for.inc, %entry
+  %0 = load i32* %i, align 4
+  %cmp = icmp slt i32 %0, 100
+  br i1 %cmp, label %for.body, label %for.end
+
+for.body:                                         ; preds = %for.cond
+  %1 = load i32* %i, align 4
+  %sub = sub nsw i32 100, %1
+  %sub1 = sub nsw i32 %sub, 10
+  %arrayidx = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 2), i32 0, i32 %sub1
+  %2 = load i32* %arrayidx, align 4
+  %3 = load i32* %a.addr, align 4
+  %add = add nsw i32 %2, %3
+  %4 = load i32* %i, align 4
+  %arrayidx2 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %4
+  store i32 %add, i32* %arrayidx2, align 4
+  br label %for.inc
+
+for.inc:                                          ; preds = %for.body
+  %5 = load i32* %i, align 4
+  %inc = add nsw i32 %5, 1
+  store i32 %inc, i32* %i, align 4
+  br label %for.cond
+
+for.end:                                          ; preds = %for.cond
+  %6 = load i32* %a.addr, align 4
+  %arrayidx3 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %6
+  %7 = load i32* %arrayidx3, align 4
+  ret i32 %7
+}
+
+; int mustAlias03 (int a) {
+;   int i;
+;   for (i=0; i<SIZE; i++)
+;     Foo.A[i+10] = Foo.B[SIZE-i-10] + a;
+;   return Foo.A[a];
+; }
+; CHECK: define i32 @mustAlias03
+; CHECK-NOT: add nsw <4 x i32>
+; CHECK ret
+
+define i32 @mustAlias03(i32 %a) nounwind {
+entry:
+  %a.addr = alloca i32, align 4
+  %i = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  store i32 0, i32* %i, align 4
+  br label %for.cond
+
+for.cond:                                         ; preds = %for.inc, %entry
+  %0 = load i32* %i, align 4
+  %cmp = icmp slt i32 %0, 100
+  br i1 %cmp, label %for.body, label %for.end
+
+for.body:                                         ; preds = %for.cond
+  %1 = load i32* %i, align 4
+  %sub = sub nsw i32 100, %1
+  %sub1 = sub nsw i32 %sub, 10
+  %arrayidx = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 2), i32 0, i32 %sub1
+  %2 = load i32* %arrayidx, align 4
+  %3 = load i32* %a.addr, align 4
+  %add = add nsw i32 %2, %3
+  %4 = load i32* %i, align 4
+  %add2 = add nsw i32 %4, 10
+  %arrayidx3 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %add2
+  store i32 %add, i32* %arrayidx3, align 4
+  br label %for.inc
+
+for.inc:                                          ; preds = %for.body
+  %5 = load i32* %i, align 4
+  %inc = add nsw i32 %5, 1
+  store i32 %inc, i32* %i, align 4
+  br label %for.cond
+
+for.end:                                          ; preds = %for.cond
+  %6 = load i32* %a.addr, align 4
+  %arrayidx4 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %6
+  %7 = load i32* %arrayidx4, align 4
+  ret i32 %7
+}
diff --git a/test/Transforms/LoopVectorize/i8-induction.ll b/test/Transforms/LoopVectorize/i8-induction.ll
new file mode 100644
index 0000000..7759b70
--- /dev/null
+++ b/test/Transforms/LoopVectorize/i8-induction.ll
@@ -0,0 +1,35 @@
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -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-S128"
+target triple = "x86_64-apple-macosx10.8.0"
+
+@a = common global i8 0, align 1
+@b = common global i8 0, align 1
+
+define void @f() nounwind uwtable ssp {
+scalar.ph:
+  store i8 0, i8* inttoptr (i64 1 to i8*), align 1, !tbaa !0
+  %0 = load i8* @a, align 1, !tbaa !0
+  br label %for.body
+
+for.body:
+  %mul16 = phi i8 [ 0, %scalar.ph ], [ %mul, %for.body ]              ; <------- i8 induction var.
+  %c.015 = phi i8 [ undef, %scalar.ph ], [ %conv8, %for.body ]
+  %conv2 = sext i8 %c.015 to i32
+  %tobool = icmp ne i8 %c.015, 0
+  %.sink = select i1 %tobool, i8 %c.015, i8 %0
+  %mul = mul i8 %mul16, %.sink
+  %add = add nsw i32 %conv2, 1
+  %conv8 = trunc i32 %add to i8
+  %sext = shl i32 %add, 24
+  %phitmp14 = icmp slt i32 %sext, 268435456
+  br i1 %phitmp14, label %for.body, label %for.end
+
+for.end:                                          ; preds = %for.body
+  store i8 %mul, i8* @b, align 1, !tbaa !0
+  ret void
+}
+
+!0 = metadata !{metadata !"omnipotent char", metadata !1}
+!1 = metadata !{metadata !"Simple C/C++ TBAA"}
+
diff --git a/test/Transforms/LoopVectorize/if-conversion-reduction.ll b/test/Transforms/LoopVectorize/if-conversion-reduction.ll
index c6dc5d7..3a2d82e 100644
--- a/test/Transforms/LoopVectorize/if-conversion-reduction.ll
+++ b/test/Transforms/LoopVectorize/if-conversion-reduction.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -enable-if-conversion -dce -instcombine -licm -S | FileCheck %s
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -enable-if-conversion -dce -instcombine -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-S128"
 target triple = "x86_64-apple-macosx10.9.0"
diff --git a/test/Transforms/LoopVectorize/if-conversion.ll b/test/Transforms/LoopVectorize/if-conversion.ll
index 28407dc..6e7c03a 100644
--- a/test/Transforms/LoopVectorize/if-conversion.ll
+++ b/test/Transforms/LoopVectorize/if-conversion.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -enable-if-conversion -dce -instcombine -licm -S | FileCheck %s
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -enable-if-conversion -dce -instcombine -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-S128"
 target triple = "x86_64-apple-macosx10.9.0"
diff --git a/test/Transforms/LoopVectorize/increment.ll b/test/Transforms/LoopVectorize/increment.ll
index e24fb39..3fa6b19 100644
--- a/test/Transforms/LoopVectorize/increment.ll
+++ b/test/Transforms/LoopVectorize/increment.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -licm -S | FileCheck %s
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -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-S128"
 target triple = "x86_64-apple-macosx10.8.0"
diff --git a/test/Transforms/LoopVectorize/intrinsic.ll b/test/Transforms/LoopVectorize/intrinsic.ll
index 49c8ecb..7d5a5d7 100644
--- a/test/Transforms/LoopVectorize/intrinsic.ll
+++ b/test/Transforms/LoopVectorize/intrinsic.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -licm -S | FileCheck %s
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -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-S128"
 target triple = "x86_64-unknown-linux-gnu"
diff --git a/test/Transforms/LoopVectorize/no_int_induction.ll b/test/Transforms/LoopVectorize/no_int_induction.ll
index 6eab799..45aa8c7 100644
--- a/test/Transforms/LoopVectorize/no_int_induction.ll
+++ b/test/Transforms/LoopVectorize/no_int_induction.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -licm -S | FileCheck %s
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -S | FileCheck %s
 
 ; int __attribute__((noinline)) sum_array(int *A, int n) {
 ;  return std::accumulate(A, A + n, 0);
diff --git a/test/Transforms/LoopVectorize/nofloat.ll b/test/Transforms/LoopVectorize/nofloat.ll
index dbdec33..de23bf0 100644
--- a/test/Transforms/LoopVectorize/nofloat.ll
+++ b/test/Transforms/LoopVectorize/nofloat.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -licm -S | FileCheck %s
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -S | FileCheck %s
 
 ; Make sure that we don't vectorize functions with 'noimplicitfloat' attributes.
 
diff --git a/test/Transforms/LoopVectorize/non-const-n.ll b/test/Transforms/LoopVectorize/non-const-n.ll
index 7e4cee4..8262a18 100644
--- a/test/Transforms/LoopVectorize/non-const-n.ll
+++ b/test/Transforms/LoopVectorize/non-const-n.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -licm -S | FileCheck %s
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -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-S128"
 target triple = "x86_64-apple-macosx10.8.0"
diff --git a/test/Transforms/LoopVectorize/nsw-crash.ll b/test/Transforms/LoopVectorize/nsw-crash.ll
new file mode 100644
index 0000000..e5fad14
--- /dev/null
+++ b/test/Transforms/LoopVectorize/nsw-crash.ll
@@ -0,0 +1,25 @@
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4
+
+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-S128"
+target triple = "x86_64-apple-macosx10.7.0"
+
+define void @test() {
+entry:
+  br i1 undef, label %while.end, label %while.body.lr.ph
+
+while.body.lr.ph:
+  br label %while.body
+
+while.body:
+  %it.sroa.0.091 = phi i32* [ undef, %while.body.lr.ph ], [ %incdec.ptr.i, %while.body ]
+  %incdec.ptr.i = getelementptr inbounds i32* %it.sroa.0.091, i64 1
+  %inc32 = add i32 undef, 1                                        ; <------------- Make sure we don't set NSW flags to the undef.
+  %cmp.i11 = icmp eq i32* %incdec.ptr.i, undef
+  br i1 %cmp.i11, label %while.end, label %while.body
+
+while.end:
+  ret void
+}
+
+
diff --git a/test/Transforms/LoopVectorize/ptr_loops.ll b/test/Transforms/LoopVectorize/ptr_loops.ll
new file mode 100644
index 0000000..25599f8
--- /dev/null
+++ b/test/Transforms/LoopVectorize/ptr_loops.ll
@@ -0,0 +1,74 @@
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -S -enable-if-conversion | 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-S128"
+target triple = "x86_64-apple-macosx10.8.0"
+
+@A = global [36 x i32] [i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 32, i32 33, i32 34, i32 35], align 16
+@B = global [36 x i32] [i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 32, i32 33, i32 34, i32 35], align 16
+
+;CHECK:_Z5test1v
+;CHECK: load <4 x i32>
+;CHECK: shufflevector <4 x i32>
+;CHECK: store <4 x i32>
+;CHECK: ret
+define i32 @_Z5test1v() nounwind uwtable ssp {
+  br label %1
+
+; <label>:1                                       ; preds = %0, %1
+  %p.02 = phi i32* [ getelementptr inbounds ([36 x i32]* @A, i64 0, i64 18), %0 ], [ %4, %1 ]
+  %b.01 = phi i32* [ getelementptr inbounds ([36 x i32]* @B, i64 0, i64 0), %0 ], [ %5, %1 ]
+  %2 = load i32* %b.01, align 4
+  %3 = shl nsw i32 %2, 1
+  store i32 %3, i32* %p.02, align 4
+  %4 = getelementptr inbounds i32* %p.02, i64 -1
+  %5 = getelementptr inbounds i32* %b.01, i64 1
+  %6 = icmp eq i32* %4, getelementptr ([36 x i32]* @A, i64 128102389400760775, i64 3)
+  br i1 %6, label %7, label %1
+
+; <label>:7                                       ; preds = %1
+  ret i32 0
+}
+
+;CHECK:_Z5test2v
+;CHECK: load <4 x i32>
+;CHECK: shufflevector <4 x i32>
+;CHECK: store <4 x i32>
+;CHECK: ret
+define i32 @_Z5test2v() nounwind uwtable ssp {
+  br label %1
+
+; <label>:1                                       ; preds = %0, %1
+  %p.02 = phi i32* [ getelementptr inbounds ([36 x i32]* @A, i64 0, i64 25), %0 ], [ %3, %1 ]
+  %b.01 = phi i32* [ getelementptr inbounds ([36 x i32]* @B, i64 0, i64 2), %0 ], [ %4, %1 ]
+  %2 = load i32* %b.01, align 4
+  store i32 %2, i32* %p.02, align 4
+  %3 = getelementptr inbounds i32* %p.02, i64 -1
+  %4 = getelementptr inbounds i32* %b.01, i64 1
+  %5 = icmp eq i32* %4, getelementptr inbounds ([36 x i32]* @A, i64 0, i64 18)
+  br i1 %5, label %6, label %1
+
+; <label>:6                                       ; preds = %1
+  ret i32 0
+}
+
+;CHECK:_Z5test3v
+;CHECK: load <4 x i32>
+;CHECK: shufflevector <4 x i32>
+;CHECK: store <4 x i32>
+;CHECK: ret
+define i32 @_Z5test3v() nounwind uwtable ssp {
+  br label %1
+
+; <label>:1                                       ; preds = %0, %1
+  %p.02 = phi i32* [ getelementptr inbounds ([36 x i32]* @A, i64 0, i64 29), %0 ], [ %3, %1 ]
+  %b.01 = phi i32* [ getelementptr inbounds ([36 x i32]* @B, i64 0, i64 5), %0 ], [ %4, %1 ]
+  %2 = load i32* %b.01, align 4
+  store i32 %2, i32* %p.02, align 4
+  %3 = getelementptr inbounds i32* %p.02, i64 -1
+  %4 = getelementptr inbounds i32* %b.01, i64 1
+  %5 = icmp eq i32* %3, getelementptr ([36 x i32]* @A, i64 128102389400760775, i64 3)
+  br i1 %5, label %6, label %1
+
+; <label>:6                                       ; preds = %1
+  ret i32 0
+}
diff --git a/test/Transforms/LoopVectorize/read-only.ll b/test/Transforms/LoopVectorize/read-only.ll
index c3c9035..bfaa6d4 100644
--- a/test/Transforms/LoopVectorize/read-only.ll
+++ b/test/Transforms/LoopVectorize/read-only.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -licm -S | FileCheck %s
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -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-S128"
 target triple = "x86_64-apple-macosx10.8.0"
diff --git a/test/Transforms/LoopVectorize/reduction.ll b/test/Transforms/LoopVectorize/reduction.ll
index 129c20d..08b7b27 100644
--- a/test/Transforms/LoopVectorize/reduction.ll
+++ b/test/Transforms/LoopVectorize/reduction.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -licm -S | FileCheck %s
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -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-S128"
 target triple = "x86_64-apple-macosx10.8.0"
diff --git a/test/Transforms/LoopVectorize/runtime-check.ll b/test/Transforms/LoopVectorize/runtime-check.ll
index 2852684..86098a6 100644
--- a/test/Transforms/LoopVectorize/runtime-check.ll
+++ b/test/Transforms/LoopVectorize/runtime-check.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -licm -S | FileCheck %s
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -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-S128"
 target triple = "x86_64-apple-macosx10.9.0"
@@ -9,6 +9,10 @@ target triple = "x86_64-apple-macosx10.9.0"
 ;     a[i] = b[i] * 3;
 ; }
 
+;CHECK: for.body.preheader:
+;CHECK: br i1 %cmp.zero, label %middle.block, label %vector.memcheck
+;CHECK: vector.memcheck:
+;CHECK: br i1 %found.conflict, label %middle.block, label %vector.ph
 ;CHECK: load <4 x float>
 define i32 @foo(float* nocapture %a, float* nocapture %b, i32 %n) nounwind uwtable ssp {
 entry:
diff --git a/test/Transforms/LoopVectorize/same-base-access.ll b/test/Transforms/LoopVectorize/same-base-access.ll
index 2a1f19d..1573893 100644
--- a/test/Transforms/LoopVectorize/same-base-access.ll
+++ b/test/Transforms/LoopVectorize/same-base-access.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -licm -S -enable-if-conversion | FileCheck %s
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -S -enable-if-conversion | 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-S128"
 target triple = "x86_64-apple-macosx10.9.0"
diff --git a/test/Transforms/LoopVectorize/scalar-select.ll b/test/Transforms/LoopVectorize/scalar-select.ll
index d72cd14..7a14d24 100644
--- a/test/Transforms/LoopVectorize/scalar-select.ll
+++ b/test/Transforms/LoopVectorize/scalar-select.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -licm -S | FileCheck %s
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -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-S128"
 target triple = "x86_64-apple-macosx10.8.0"
diff --git a/test/Transforms/LoopVectorize/simple-unroll.ll b/test/Transforms/LoopVectorize/simple-unroll.ll
index 9825764..7e2dd5f 100644
--- a/test/Transforms/LoopVectorize/simple-unroll.ll
+++ b/test/Transforms/LoopVectorize/simple-unroll.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s  -loop-vectorize -force-vector-width=4 -force-vector-unroll=2 -dce -instcombine -licm -S | FileCheck %s
+; RUN: opt < %s  -loop-vectorize -force-vector-width=4 -force-vector-unroll=2 -dce -instcombine -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-S128"
 target triple = "x86_64-apple-macosx10.8.0"
diff --git a/test/Transforms/LoopVectorize/small-loop.ll b/test/Transforms/LoopVectorize/small-loop.ll
index ae784b3..fa83dba 100644
--- a/test/Transforms/LoopVectorize/small-loop.ll
+++ b/test/Transforms/LoopVectorize/small-loop.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -licm -S | FileCheck %s
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -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-S128"
 target triple = "x86_64-apple-macosx10.8.0"
diff --git a/test/Transforms/LoopVectorize/struct_access.ll b/test/Transforms/LoopVectorize/struct_access.ll
new file mode 100644
index 0000000..de65d0d
--- /dev/null
+++ b/test/Transforms/LoopVectorize/struct_access.ll
@@ -0,0 +1,50 @@
+; RUN: opt < %s -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -instcombine -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-S128"
+target triple = "x86_64-apple-macosx10.9.0"
+
+%struct.coordinate = type { i32, i32 }
+
+; Make sure that we don't generate a wide load when accessing the struct.
+; struct coordinate {
+;  int x;
+;  int y;
+; };
+;
+;
+; int foo(struct coordinate *A, int n) {
+;
+;   int sum = 0;
+;   for (int i = 0; i < n; ++i)
+;     sum += A[i].x;
+;
+;   return sum;
+; }
+
+;CHECK: @foo
+;CHECK-NOT: load <4 x i32>
+;CHECK: ret
+define i32 @foo(%struct.coordinate* nocapture %A, i32 %n) nounwind uwtable readonly ssp {
+entry:
+  %cmp4 = icmp sgt i32 %n, 0
+  br i1 %cmp4, label %for.body, label %for.end
+
+for.body:                                         ; preds = %entry, %for.body
+  %indvars.iv = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %entry ]
+  %sum.05 = phi i32 [ %add, %for.body ], [ 0, %entry ]
+  %x = getelementptr inbounds %struct.coordinate* %A, i64 %indvars.iv, i32 0
+  %0 = load i32* %x, align 4, !tbaa !0
+  %add = add nsw i32 %0, %sum.05
+  %indvars.iv.next = add i64 %indvars.iv, 1
+  %lftr.wideiv = trunc i64 %indvars.iv.next to i32
+  %exitcond = icmp eq i32 %lftr.wideiv, %n
+  br i1 %exitcond, label %for.end, label %for.body
+
+for.end:                                          ; preds = %for.body, %entry
+  %sum.0.lcssa = phi i32 [ 0, %entry ], [ %add, %for.body ]
+  ret i32 %sum.0.lcssa
+}
+
+!0 = metadata !{metadata !"int", metadata !1}
+!1 = metadata !{metadata !"omnipotent char", metadata !2}
+!2 = metadata !{metadata !"Simple C/C++ TBAA"}
diff --git a/test/Transforms/LoopVectorize/write-only.ll b/test/Transforms/LoopVectorize/write-only.ll
index b42122b..54cbe8d 100644
--- a/test/Transforms/LoopVectorize/write-only.ll
+++ b/test/Transforms/LoopVectorize/write-only.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -licm -S | FileCheck %s
+; RUN: opt < %s  -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -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-S128"
 target triple = "x86_64-apple-macosx10.8.0"
diff --git a/test/Transforms/MemCpyOpt/memcpy.ll b/test/Transforms/MemCpyOpt/memcpy.ll
index 3fa1628..582a57b 100644
--- a/test/Transforms/MemCpyOpt/memcpy.ll
+++ b/test/Transforms/MemCpyOpt/memcpy.ll
@@ -70,20 +70,20 @@ define void @test4(i8 *%P) {
   %A = alloca %1
   %a = bitcast %1* %A to i8*
   call void @llvm.memcpy.p0i8.p0i8.i64(i8* %a, i8* %P, i64 8, i32 4, i1 false)
-  call void @test4a(i8* byval align 1 %a)
+  call void @test4a(i8* align 1 byval %a)
   ret void
 ; CHECK: @test4
 ; CHECK-NEXT: call void @test4a(
 }
 
-declare void @test4a(i8* byval align 1)
+declare void @test4a(i8* align 1 byval)
 declare void @llvm.memcpy.p0i8.p0i8.i64(i8* nocapture, i8* nocapture, i64, i32, i1) nounwind
 
 %struct.S = type { i128, [4 x i8]}
 
 @sS = external global %struct.S, align 16
 
-declare void @test5a(%struct.S* byval align 16) nounwind ssp
+declare void @test5a(%struct.S* align 16 byval) nounwind ssp
 
 
 ; rdar://8713376 - This memcpy can't be eliminated.
@@ -94,7 +94,7 @@ entry:
   call void @llvm.memcpy.p0i8.p0i8.i64(i8* %tmp, i8* bitcast (%struct.S* @sS to i8*), i64 32, i32 16, i1 false)
   %a = getelementptr %struct.S* %y, i64 0, i32 1, i64 0
   store i8 4, i8* %a
-  call void @test5a(%struct.S* byval align 16 %y)
+  call void @test5a(%struct.S* align 16 byval %y)
   ret i32 0
   ; CHECK: @test5(
   ; CHECK: store i8 4
@@ -114,19 +114,19 @@ define void @test6(i8 *%P) {
 ; isn't itself 8 byte aligned.
 %struct.p = type { i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32 }
 
-define i32 @test7(%struct.p* nocapture byval align 8 %q) nounwind ssp {
+define i32 @test7(%struct.p* nocapture align 8 byval %q) nounwind ssp {
 entry:
   %agg.tmp = alloca %struct.p, align 4
   %tmp = bitcast %struct.p* %agg.tmp to i8*
   %tmp1 = bitcast %struct.p* %q to i8*
   call void @llvm.memcpy.p0i8.p0i8.i64(i8* %tmp, i8* %tmp1, i64 48, i32 4, i1 false)
-  %call = call i32 @g(%struct.p* byval align 8 %agg.tmp) nounwind
+  %call = call i32 @g(%struct.p* align 8 byval %agg.tmp) nounwind
   ret i32 %call
 ; CHECK: @test7
-; CHECK: call i32 @g(%struct.p* byval align 8 %q) nounwind
+; CHECK: call i32 @g(%struct.p* byval align 8 %q) [[NUW:#[0-9]+]]
 }
 
-declare i32 @g(%struct.p* byval align 8)
+declare i32 @g(%struct.p* align 8 byval)
 
 declare void @llvm.memcpy.p0i8.p0i8.i32(i8* nocapture, i8* nocapture, i32, i32, i1) nounwind
 
@@ -152,7 +152,7 @@ declare noalias i8* @malloc(i32)
 ; rdar://11341081
 %struct.big = type { [50 x i32] }
 
-define void @test9() nounwind uwtable ssp {
+define void @test9() nounwind ssp uwtable {
 entry:
 ; CHECK: test9
 ; CHECK: f1
@@ -170,3 +170,7 @@ entry:
 
 declare void @f1(%struct.big* sret)
 declare void @f2(%struct.big*)
+
+; CHECK: attributes [[NUW]] = { nounwind }
+; CHECK: attributes #1 = { nounwind ssp }
+; CHECK: attributes #2 = { nounwind ssp uwtable }
diff --git a/test/Transforms/MergeFunc/2013-01-10-MergeFuncAssert.ll b/test/Transforms/MergeFunc/2013-01-10-MergeFuncAssert.ll
new file mode 100644
index 0000000..3f6a5ba
--- /dev/null
+++ b/test/Transforms/MergeFunc/2013-01-10-MergeFuncAssert.ll
@@ -0,0 +1,36 @@
+; RUN: opt -mergefunc -disable-output < %s
+; This used to trigger a ConstantExpr::getBitCast assertion.
+
+define void @t1() unnamed_addr uwtable ssp align 2 {
+entry:
+  switch i32 undef, label %sw.bb12 [
+    i32 127, label %sw.bb
+    i32 126, label %sw.bb4
+  ]
+
+sw.bb:                                            ; preds = %entry
+  unreachable
+
+sw.bb4:                                           ; preds = %entry
+  unreachable
+
+sw.bb12:                                          ; preds = %entry
+  ret void
+}
+
+define void @t2() unnamed_addr uwtable ssp align 2 {
+entry:
+  switch i32 undef, label %sw.bb8 [
+    i32 4, label %sw.bb
+    i32 3, label %sw.bb4
+  ]
+
+sw.bb:                                            ; preds = %entry
+  unreachable
+
+sw.bb4:                                           ; preds = %entry
+  ret void
+
+sw.bb8:                                           ; preds = %entry
+  unreachable
+}
diff --git a/test/Transforms/ObjCARC/apelim.ll b/test/Transforms/ObjCARC/apelim.ll
index 8c7b5b1..4541b3f 100644
--- a/test/Transforms/ObjCARC/apelim.ll
+++ b/test/Transforms/ObjCARC/apelim.ll
@@ -38,8 +38,8 @@ entry:
 }
 
 ; CHECK: define internal void @_GLOBAL__I_y()
-; CHECK: %0 = call i8* @objc_autoreleasePoolPush() nounwind
-; CHECK: call void @objc_autoreleasePoolPop(i8* %0) nounwind
+; CHECK: %0 = call i8* @objc_autoreleasePoolPush() [[NUW:#[0-9]+]]
+; CHECK: call void @objc_autoreleasePoolPop(i8* %0) [[NUW]]
 ; CHECK: }
 define internal void @_GLOBAL__I_y() {
 entry:
@@ -51,3 +51,5 @@ entry:
 
 declare i8* @objc_autoreleasePoolPush()
 declare void @objc_autoreleasePoolPop(i8*)
+
+; CHECK: attributes #0 = { nounwind }
diff --git a/test/Transforms/ObjCARC/basic.ll b/test/Transforms/ObjCARC/basic.ll
index 7b64b1b..4c24ebf 100644
--- a/test/Transforms/ObjCARC/basic.ll
+++ b/test/Transforms/ObjCARC/basic.ll
@@ -92,10 +92,10 @@ alt_return:
 
 ; CHECK: define void @test1b(
 ; CHECK: entry:
-; CHECK:   tail call i8* @objc_retain(i8* %x) nounwind
+; CHECK:   tail call i8* @objc_retain(i8* %x) [[NUW:#[0-9]+]]
 ; CHECK-NOT: @objc_
 ; CHECK: if.end5:
-; CHECK:   tail call void @objc_release(i8* %x) nounwind, !clang.imprecise_release !0
+; CHECK:   tail call void @objc_release(i8* %x) [[NUW]], !clang.imprecise_release !0
 ; CHECK-NOT: @objc_
 ; CHECK: }
 define void @test1b(i8* %x, i1 %p, i1 %q) {
@@ -404,8 +404,8 @@ entry:
 ; a stack argument.
 
 ; CHECK: define void @test11(
-; CHECK: tail call i8* @objc_retain(i8* %x) nounwind
-; CHECK: tail call i8* @objc_autorelease(i8* %0) nounwind
+; CHECK: tail call i8* @objc_retain(i8* %x) [[NUW]]
+; CHECK: call i8* @objc_autorelease(i8* %0) [[NUW]]
 ; CHECK: }
 define void @test11(i8* %x) nounwind {
 entry:
@@ -431,8 +431,8 @@ entry:
 ; Same as test11 but the value is returned. Do an RV optimization.
 
 ; CHECK: define i8* @test11b(
-; CHECK: tail call i8* @objc_retain(i8* %x) nounwind
-; CHECK: tail call i8* @objc_autoreleaseReturnValue(i8* %0) nounwind
+; CHECK: tail call i8* @objc_retain(i8* %x) [[NUW]]
+; CHECK: tail call i8* @objc_autoreleaseReturnValue(i8* %0) [[NUW]]
 ; CHECK: }
 define i8* @test11b(i8* %x) nounwind {
 entry:
@@ -462,10 +462,10 @@ entry:
 ; Trivial retain,autorelease pair. Don't delete!
 
 ; CHECK: define void @test13(
-; CHECK: tail call i8* @objc_retain(i8* %x) nounwind
-; CHECK: tail call i8* @objc_retain(i8* %x) nounwind
+; CHECK: tail call i8* @objc_retain(i8* %x) [[NUW]]
+; CHECK: tail call i8* @objc_retain(i8* %x) [[NUW]]
 ; CHECK: @use_pointer(i8* %x)
-; CHECK: tail call i8* @objc_autorelease(i8* %x) nounwind
+; CHECK: call i8* @objc_autorelease(i8* %x) [[NUW]]
 ; CHECK: }
 define void @test13(i8* %x, i64 %n) {
 entry:
@@ -716,7 +716,7 @@ entry:
 ; Bitcast insertion
 
 ; CHECK: define void @test20(
-; CHECK: %tmp1 = tail call i8* @objc_retain(i8* %tmp) nounwind
+; CHECK: %tmp1 = tail call i8* @objc_retain(i8* %tmp) [[NUW]]
 ; CHECK-NEXT: invoke
 define void @test20(double* %self) {
 if.then12:
@@ -980,7 +980,7 @@ done:
 ; CHECK: call i8* @objc_retain(
 ; CHECK: call void @callee()
 ; CHECK: store
-; CHECK: call void @objc_release(i8* %p) nounwind, !clang.imprecise_release
+; CHECK: call void @objc_release(i8* %p) [[NUW]], !clang.imprecise_release
 ; CHECK: done:
 ; CHECK-NOT: @objc_
 ; CHECK: }
@@ -1450,9 +1450,9 @@ define void @test45(i8** %pp, i8** %qq) {
 ; Don't delete retain and autorelease here.
 
 ; CHECK: define void @test46(
-; CHECK: tail call i8* @objc_retain(i8* %p) nounwind
+; CHECK: tail call i8* @objc_retain(i8* %p) [[NUW]]
 ; CHECK: true:
-; CHECK: tail call i8* @objc_autorelease(i8* %p) nounwind
+; CHECK: call i8* @objc_autorelease(i8* %p) [[NUW]]
 define void @test46(i8* %p, i1 %a) {
 entry:
   call i8* @objc_retain(i8* %p)
@@ -1565,7 +1565,7 @@ define void @test53(void ()** %zz, i8** %pp) {
 
 ; CHECK: define void @test54(
 ; CHECK: call i8* @returner()
-; CHECK-NEXT: call void @objc_release(i8* %t) nounwind, !clang.imprecise_release !0
+; CHECK-NEXT: call void @objc_release(i8* %t) [[NUW]], !clang.imprecise_release !0
 ; CHECK-NEXT: ret void
 define void @test54() {
   %t = call i8* @returner()
@@ -1595,10 +1595,10 @@ entry:
 ; CHECK: define void @test56(
 ; CHECK-NOT: @objc
 ; CHECK: if.then:
-; CHECK-NEXT: %0 = tail call i8* @objc_retain(i8* %x) nounwind
+; CHECK-NEXT: %0 = tail call i8* @objc_retain(i8* %x) [[NUW]]
 ; CHECK-NEXT: tail call void @use_pointer(i8* %x)
 ; CHECK-NEXT: tail call void @use_pointer(i8* %x)
-; CHECK-NEXT: tail call void @objc_release(i8* %x) nounwind, !clang.imprecise_release !0
+; CHECK-NEXT: tail call void @objc_release(i8* %x) [[NUW]], !clang.imprecise_release !0
 ; CHECK-NEXT: br label %if.end
 ; CHECK-NOT: @objc
 ; CHECK: }
@@ -1630,10 +1630,10 @@ if.end:                                           ; preds = %entry, %if.then
 ; CHECK-NEXT: entry:
 ; CHECK-NEXT:   call void @use_pointer(i8* %x)
 ; CHECK-NEXT:   call void @use_pointer(i8* %x)
-; CHECK-NEXT:   %0 = tail call i8* @objc_retain(i8* %x) nounwind
+; CHECK-NEXT:   %0 = tail call i8* @objc_retain(i8* %x) [[NUW]]
 ; CHECK-NEXT:   call void @use_pointer(i8* %x)
 ; CHECK-NEXT:   call void @use_pointer(i8* %x)
-; CHECK-NEXT:   call void @objc_release(i8* %x) nounwind
+; CHECK-NEXT:   call void @objc_release(i8* %x) [[NUW]]
 ; CHECK-NEXT:   ret void
 ; CHECK-NEXT: }
 define void @test57(i8* %x) nounwind {
@@ -1673,10 +1673,10 @@ entry:
 
 ; CHECK:      define void @test59(
 ; CHECK-NEXT: entry:
-; CHECK-NEXT:   %0 = tail call i8* @objc_retain(i8* %x) nounwind
+; CHECK-NEXT:   %0 = tail call i8* @objc_retain(i8* %x) [[NUW]]
 ; CHECK-NEXT:   call void @use_pointer(i8* %x)
 ; CHECK-NEXT:   call void @use_pointer(i8* %x)
-; CHECK-NEXT:   call void @objc_release(i8* %x) nounwind
+; CHECK-NEXT:   call void @objc_release(i8* %x) [[NUW]]
 ; CHECK-NEXT:   ret void
 ; CHECK-NEXT: }
 define void @test59(i8* %x) nounwind {
@@ -1875,8 +1875,8 @@ return:                                           ; preds = %if.then, %entry
 ; rdar://11931823
 
 ; CHECK: define void @test66(
-; CHECK:   %tmp7 = tail call i8* @objc_retain(i8* %cond) nounwind
-; CHECK:   tail call void @objc_release(i8* %cond) nounwind
+; CHECK:   %tmp7 = tail call i8* @objc_retain(i8* %cond) [[NUW]]
+; CHECK:   tail call void @objc_release(i8* %cond) [[NUW]]
 ; CHECK: }
 define void @test66(i8* %tmp5, i8* %bar, i1 %tobool, i1 %tobool1, i8* %call) {
 entry:
@@ -2224,3 +2224,6 @@ end:                                              ; preds = %if.end125, %if.end1
 !0 = metadata !{}
 
 declare i32 @__gxx_personality_v0(...)
+
+; CHECK: attributes #0 = { nounwind readnone }
+; CHECK: attributes [[NUW]] = { nounwind }
diff --git a/test/Transforms/ObjCARC/cfg-hazards.ll b/test/Transforms/ObjCARC/cfg-hazards.ll
index 1519423..899298b 100644
--- a/test/Transforms/ObjCARC/cfg-hazards.ll
+++ b/test/Transforms/ObjCARC/cfg-hazards.ll
@@ -86,9 +86,9 @@ for.end:                                          ; preds = %for.body
 
 ; Delete nested retain+release pairs around loops.
 
-;      CHECK: define void @test3(i8* %a) nounwind {
+;      CHECK: define void @test3(i8* %a) #0 {
 ; CHECK-NEXT: entry:
-; CHECK-NEXT:   tail call i8* @objc_retain(i8* %a) nounwind
+; CHECK-NEXT:   tail call i8* @objc_retain(i8* %a) [[NUW:#[0-9]+]]
 ; CHECK-NEXT:   br label %loop
 ;  CHECK-NOT:   @objc_
 ;      CHECK: exit:
@@ -112,9 +112,9 @@ exit:
   ret void
 }
 
-;      CHECK: define void @test4(i8* %a) nounwind {
+;      CHECK: define void @test4(i8* %a) #0 {
 ; CHECK-NEXT: entry:
-; CHECK-NEXT:   tail call i8* @objc_retain(i8* %a) nounwind
+; CHECK-NEXT:   tail call i8* @objc_retain(i8* %a) [[NUW]]
 ; CHECK-NEXT:   br label %loop
 ;  CHECK-NOT:   @objc_
 ;      CHECK: exit:
@@ -142,9 +142,9 @@ exit:
   ret void
 }
 
-;      CHECK: define void @test5(i8* %a) nounwind {
+;      CHECK: define void @test5(i8* %a) #0 {
 ; CHECK-NEXT: entry:
-; CHECK-NEXT:   tail call i8* @objc_retain(i8* %a) nounwind
+; CHECK-NEXT:   tail call i8* @objc_retain(i8* %a) [[NUW]]
 ; CHECK-NEXT:   call void @callee()
 ; CHECK-NEXT:   br label %loop
 ;  CHECK-NOT:   @objc_
@@ -176,9 +176,9 @@ exit:
   ret void
 }
 
-;      CHECK: define void @test6(i8* %a) nounwind {
+;      CHECK: define void @test6(i8* %a) #0 {
 ; CHECK-NEXT: entry:
-; CHECK-NEXT:   tail call i8* @objc_retain(i8* %a) nounwind
+; CHECK-NEXT:   tail call i8* @objc_retain(i8* %a) [[NUW]]
 ; CHECK-NEXT:   br label %loop
 ;  CHECK-NOT:   @objc_
 ;      CHECK: exit:
@@ -209,9 +209,9 @@ exit:
   ret void
 }
 
-;      CHECK: define void @test7(i8* %a) nounwind {
+;      CHECK: define void @test7(i8* %a) #0 {
 ; CHECK-NEXT: entry:
-; CHECK-NEXT:   tail call i8* @objc_retain(i8* %a) nounwind
+; CHECK-NEXT:   tail call i8* @objc_retain(i8* %a) [[NUW]]
 ; CHECK-NEXT:   call void @callee()
 ; CHECK-NEXT:   br label %loop
 ;  CHECK-NOT:   @objc_
@@ -242,9 +242,9 @@ exit:
   ret void
 }
 
-;      CHECK: define void @test8(i8* %a) nounwind {
+;      CHECK: define void @test8(i8* %a) #0 {
 ; CHECK-NEXT: entry:
-; CHECK-NEXT:   tail call i8* @objc_retain(i8* %a) nounwind
+; CHECK-NEXT:   tail call i8* @objc_retain(i8* %a) [[NUW]]
 ; CHECK-NEXT:   br label %loop
 ;  CHECK-NOT:   @objc_
 ;      CHECK: exit:
@@ -274,7 +274,7 @@ exit:
   ret void
 }
 
-;      CHECK: define void @test9(i8* %a) nounwind {
+;      CHECK: define void @test9(i8* %a) #0 {
 ; CHECK-NEXT: entry:
 ; CHECK-NEXT:   br label %loop
 ;  CHECK-NOT:   @objc_
@@ -303,7 +303,7 @@ exit:
   ret void
 }
 
-;      CHECK: define void @test10(i8* %a) nounwind {
+;      CHECK: define void @test10(i8* %a) #0 {
 ; CHECK-NEXT: entry:
 ; CHECK-NEXT:   br label %loop
 ;  CHECK-NOT:   @objc_
@@ -332,7 +332,7 @@ exit:
   ret void
 }
 
-;      CHECK: define void @test11(i8* %a) nounwind {
+;      CHECK: define void @test11(i8* %a) #0 {
 ; CHECK-NEXT: entry:
 ; CHECK-NEXT:   br label %loop
 ;  CHECK-NOT:   @objc_
@@ -362,15 +362,15 @@ exit:
 
 ; Don't delete anything if they're not balanced.
 
-;      CHECK: define void @test12(i8* %a) nounwind {
+;      CHECK: define void @test12(i8* %a) #0 {
 ; CHECK-NEXT: entry:
-; CHECK-NEXT:   %outer = tail call i8* @objc_retain(i8* %a) nounwind
-; CHECK-NEXT:   %inner = tail call i8* @objc_retain(i8* %a) nounwind
+; CHECK-NEXT:   %outer = tail call i8* @objc_retain(i8* %a) [[NUW]]
+; CHECK-NEXT:   %inner = tail call i8* @objc_retain(i8* %a) [[NUW]]
 ; CHECK-NEXT:   br label %loop
 ;  CHECK-NOT:   @objc_
 ;      CHECK: exit:
-; CHECK-NEXT: call void @objc_release(i8* %a) nounwind
-; CHECK-NEXT: call void @objc_release(i8* %a) nounwind, !clang.imprecise_release !0
+; CHECK-NEXT: call void @objc_release(i8* %a) [[NUW]]
+; CHECK-NEXT: call void @objc_release(i8* %a) [[NUW]], !clang.imprecise_release !0
 ; CHECK-NEXT:   ret void
 ; CHECK-NEXT: }
 define void @test12(i8* %a) nounwind {
@@ -394,4 +394,6 @@ exit:
   ret void
 }
 
+; CHECK: attributes [[NUW]] = { nounwind }
+
 !0 = metadata !{}
diff --git a/test/Transforms/ObjCARC/contract-marker.ll b/test/Transforms/ObjCARC/contract-marker.ll
index 01d978a..01fd1e7 100644
--- a/test/Transforms/ObjCARC/contract-marker.ll
+++ b/test/Transforms/ObjCARC/contract-marker.ll
@@ -3,7 +3,7 @@
 ; CHECK:      %call = tail call i32* @qux()
 ; CHECK-NEXT: %tcall = bitcast i32* %call to i8*
 ; CHECK-NEXT: call void asm sideeffect "mov\09r7, r7\09\09@ marker for objc_retainAutoreleaseReturnValue", ""()
-; CHECK-NEXT: %0 = tail call i8* @objc_retainAutoreleasedReturnValue(i8* %tcall) nounwind
+; CHECK-NEXT: %0 = tail call i8* @objc_retainAutoreleasedReturnValue(i8* %tcall) [[NUW:#[0-9]+]]
 
 define void @foo() {
 entry:
@@ -21,3 +21,5 @@ declare void @bar(i8*)
 !clang.arc.retainAutoreleasedReturnValueMarker = !{!0}
 
 !0 = metadata !{metadata !"mov\09r7, r7\09\09@ marker for objc_retainAutoreleaseReturnValue"}
+
+; CHECK: attributes [[NUW]] = { nounwind }
diff --git a/test/Transforms/ObjCARC/contract-storestrong.ll b/test/Transforms/ObjCARC/contract-storestrong.ll
index 2922f81..6999237 100644
--- a/test/Transforms/ObjCARC/contract-storestrong.ll
+++ b/test/Transforms/ObjCARC/contract-storestrong.ll
@@ -10,7 +10,7 @@ declare void @use_pointer(i8*)
 
 ; CHECK: define void @test0(
 ; CHECK: entry:
-; CHECK-NEXT: tail call void @objc_storeStrong(i8** @x, i8* %p) nounwind
+; CHECK-NEXT: tail call void @objc_storeStrong(i8** @x, i8* %p) [[NUW:#[0-9]+]]
 ; CHECK-NEXT: ret void
 define void @test0(i8* %p) {
 entry:
@@ -25,10 +25,10 @@ entry:
 
 ;      CHECK: define void @test1(i8* %p) {
 ; CHECK-NEXT: entry:
-; CHECK-NEXT:   %0 = tail call i8* @objc_retain(i8* %p) nounwind
+; CHECK-NEXT:   %0 = tail call i8* @objc_retain(i8* %p) [[NUW]]
 ; CHECK-NEXT:   %tmp = load volatile i8** @x, align 8
 ; CHECK-NEXT:   store i8* %0, i8** @x, align 8
-; CHECK-NEXT:   tail call void @objc_release(i8* %tmp) nounwind
+; CHECK-NEXT:   tail call void @objc_release(i8* %tmp) [[NUW]]
 ; CHECK-NEXT:   ret void
 ; CHECK-NEXT: }
 define void @test1(i8* %p) {
@@ -44,10 +44,10 @@ entry:
 
 ;      CHECK: define void @test2(i8* %p) {
 ; CHECK-NEXT: entry:
-; CHECK-NEXT:   %0 = tail call i8* @objc_retain(i8* %p) nounwind
+; CHECK-NEXT:   %0 = tail call i8* @objc_retain(i8* %p) [[NUW]]
 ; CHECK-NEXT:   %tmp = load i8** @x, align 8
 ; CHECK-NEXT:   store volatile i8* %0, i8** @x, align 8
-; CHECK-NEXT:   tail call void @objc_release(i8* %tmp) nounwind
+; CHECK-NEXT:   tail call void @objc_release(i8* %tmp) [[NUW]]
 ; CHECK-NEXT:   ret void
 ; CHECK-NEXT: }
 define void @test2(i8* %p) {
@@ -64,11 +64,11 @@ entry:
 
 ; CHECK:      define void @test3(i8* %newValue) {
 ; CHECK-NEXT: entry:
-; CHECK-NEXT:   %x0 = tail call i8* @objc_retain(i8* %newValue) nounwind
+; CHECK-NEXT:   %x0 = tail call i8* @objc_retain(i8* %newValue) [[NUW]]
 ; CHECK-NEXT:   %x1 = load i8** @x, align 8
 ; CHECK-NEXT:   store i8* %x0, i8** @x, align 8
 ; CHECK-NEXT:   tail call void @use_pointer(i8* %x1), !clang.arc.no_objc_arc_exceptions !0
-; CHECK-NEXT:   tail call void @objc_release(i8* %x1) nounwind, !clang.imprecise_release !0
+; CHECK-NEXT:   tail call void @objc_release(i8* %x1) [[NUW]], !clang.imprecise_release !0
 ; CHECK-NEXT:   ret void
 ; CHECK-NEXT: }
 define void @test3(i8* %newValue) {
@@ -85,11 +85,11 @@ entry:
 
 ; CHECK:      define i1 @test4(i8* %newValue, i8* %foo) {
 ; CHECK-NEXT: entry:
-; CHECK-NEXT:   %x0 = tail call i8* @objc_retain(i8* %newValue) nounwind
+; CHECK-NEXT:   %x0 = tail call i8* @objc_retain(i8* %newValue) [[NUW]]
 ; CHECK-NEXT:   %x1 = load i8** @x, align 8
 ; CHECK-NEXT:   store i8* %x0, i8** @x, align 8
 ; CHECK-NEXT:   %t = icmp eq i8* %x1, %foo
-; CHECK-NEXT:   tail call void @objc_release(i8* %x1) nounwind, !clang.imprecise_release !0
+; CHECK-NEXT:   tail call void @objc_release(i8* %x1) [[NUW]], !clang.imprecise_release !0
 ; CHECK-NEXT:   ret i1 %t
 ; CHECK-NEXT: }
 define i1 @test4(i8* %newValue, i8* %foo) {
@@ -106,7 +106,7 @@ entry:
 
 ; CHECK: define i1 @test5(i8* %newValue, i8* %foo) {
 ; CHECK: %t = icmp eq i8* %x1, %foo
-; CHECK: tail call void @objc_storeStrong(i8** @x, i8* %newValue) nounwind
+; CHECK: tail call void @objc_storeStrong(i8** @x, i8* %newValue) [[NUW]]
 define i1 @test5(i8* %newValue, i8* %foo) {
 entry:
   %x0 = tail call i8* @objc_retain(i8* %newValue) nounwind
@@ -121,7 +121,7 @@ entry:
 
 ; CHECK: define i1 @test6(i8* %newValue, i8* %foo) {
 ; CHECK: %t = icmp eq i8* %x1, %foo
-; CHECK: tail call void @objc_storeStrong(i8** @x, i8* %newValue) nounwind
+; CHECK: tail call void @objc_storeStrong(i8** @x, i8* %newValue) [[NUW]]
 define i1 @test6(i8* %newValue, i8* %foo) {
 entry:
   %x0 = tail call i8* @objc_retain(i8* %newValue) nounwind
@@ -136,9 +136,9 @@ entry:
 
 ;      CHECK: define void @test7(
 ; CHECK-NEXT: entry:
-; CHECK-NEXT:   %0 = tail call i8* @objc_retain(i8* %p) nounwind
+; CHECK-NEXT:   %0 = tail call i8* @objc_retain(i8* %p) [[NUW]]
 ; CHECK-NEXT:   %tmp = load i8** @x, align 8
-; CHECK-NEXT:   tail call void @objc_release(i8* %tmp) nounwind
+; CHECK-NEXT:   tail call void @objc_release(i8* %tmp) [[NUW]]
 ; CHECK-NEXT:   ret void
 ; CHECK-NEXT: }
 define void @test7(i8* %p) {
@@ -155,7 +155,7 @@ entry:
 ; CHECK-NEXT: entry:
 ; CHECK-NEXT:   %tmp = load i8** @x, align 8
 ; CHECK-NEXT:   store i8* %p, i8** @x, align 8
-; CHECK-NEXT:   tail call void @objc_release(i8* %tmp) nounwind
+; CHECK-NEXT:   tail call void @objc_release(i8* %tmp) [[NUW]]
 ; CHECK-NEXT:   ret void
 ; CHECK-NEXT: }
 define void @test8(i8* %p) {
@@ -167,3 +167,5 @@ entry:
 }
 
 !0 = metadata !{}
+
+; CHECK: attributes [[NUW]] = { nounwind }
diff --git a/test/Transforms/ObjCARC/contract-testcases.ll b/test/Transforms/ObjCARC/contract-testcases.ll
index 1510ed0..85b03be 100644
--- a/test/Transforms/ObjCARC/contract-testcases.ll
+++ b/test/Transforms/ObjCARC/contract-testcases.ll
@@ -69,7 +69,7 @@ bb7:                                              ; preds = %bb6, %bb6, %bb5
 ; CHECK: define void @_Z6doTestP8NSString() {
 ; CHECK: invoke.cont:                                      ; preds = %entry
 ; CHECK-NEXT: call void asm sideeffect "mov\09r7, r7\09\09@ marker for objc_retainAutoreleaseReturnValue", ""()
-; CHECK-NEXT: %tmp = tail call i8* @objc_retainAutoreleasedReturnValue(i8* %call) nounwind
+; CHECK-NEXT: %tmp = tail call i8* @objc_retainAutoreleasedReturnValue(i8* %call) [[NUW:#[0-9]+]]
 define void @_Z6doTestP8NSString() {
 entry:
   %call = invoke i8* bitcast (i8* (i8*, i8*, ...)* @objc_msgSend to i8* ()*)()
@@ -88,3 +88,6 @@ lpad:                                             ; preds = %entry
 !clang.arc.retainAutoreleasedReturnValueMarker = !{!0}
 
 !0 = metadata !{metadata !"mov\09r7, r7\09\09@ marker for objc_retainAutoreleaseReturnValue"}
+
+; CHECK: attributes #0 = { optsize }
+; CHECK: attributes [[NUW]] = { nounwind }
diff --git a/test/Transforms/ObjCARC/contract.ll b/test/Transforms/ObjCARC/contract.ll
index c48f8a5..b6fba59 100644
--- a/test/Transforms/ObjCARC/contract.ll
+++ b/test/Transforms/ObjCARC/contract.ll
@@ -34,12 +34,12 @@ entry:
 ; Merge objc_retain and objc_autorelease into objc_retainAutorelease.
 
 ; CHECK: define void @test2(
-; CHECK: tail call i8* @objc_retainAutorelease(i8* %x) nounwind
+; CHECK: tail call i8* @objc_retainAutorelease(i8* %x) [[NUW:#[0-9]+]]
 ; CHECK: }
 define void @test2(i8* %x) nounwind {
 entry:
   %0 = tail call i8* @objc_retain(i8* %x) nounwind
-  tail call i8* @objc_autorelease(i8* %0) nounwind
+  call i8* @objc_autorelease(i8* %0) nounwind
   call void @use_pointer(i8* %x)
   ret void
 }
@@ -47,7 +47,7 @@ entry:
 ; Same as test2 but the value is returned. Do an RV optimization.
 
 ; CHECK: define i8* @test2b(
-; CHECK: tail call i8* @objc_retainAutoreleaseReturnValue(i8* %x) nounwind
+; CHECK: tail call i8* @objc_retainAutoreleaseReturnValue(i8* %x) [[NUW]]
 ; CHECK: }
 define i8* @test2b(i8* %x) nounwind {
 entry:
@@ -59,14 +59,14 @@ entry:
 ; Merge a retain,autorelease pair around a call.
 
 ; CHECK: define void @test3(
-; CHECK: tail call i8* @objc_retainAutorelease(i8* %x) nounwind
+; CHECK: tail call i8* @objc_retainAutorelease(i8* %x) [[NUW]]
 ; CHECK: @use_pointer(i8* %0)
 ; CHECK: }
 define void @test3(i8* %x, i64 %n) {
 entry:
   tail call i8* @objc_retain(i8* %x) nounwind
   call void @use_pointer(i8* %x)
-  tail call i8* @objc_autorelease(i8* %x) nounwind
+  call i8* @objc_autorelease(i8* %x) nounwind
   ret void
 }
 
@@ -75,7 +75,7 @@ entry:
 
 ; CHECK: define void @test4(
 ; CHECK-NEXT: entry:
-; CHECK-NEXT: @objc_retainAutorelease(i8* %x) nounwind
+; CHECK-NEXT: @objc_retainAutorelease(i8* %x) [[NUW]]
 ; CHECK-NEXT: @use_pointer
 ; CHECK-NEXT: @objc_release
 ; CHECK-NEXT: ret void
@@ -84,7 +84,7 @@ define void @test4(i8* %x, i64 %n) {
 entry:
   tail call i8* @objc_retain(i8* %x) nounwind
   call void @use_pointer(i8* %x)
-  tail call i8* @objc_autorelease(i8* %x) nounwind
+  call i8* @objc_autorelease(i8* %x) nounwind
   tail call void @objc_release(i8* %x) nounwind
   ret void
 }
@@ -92,9 +92,9 @@ entry:
 ; Don't merge retain and autorelease if they're not control-equivalent.
 
 ; CHECK: define void @test5(
-; CHECK: tail call i8* @objc_retain(i8* %p) nounwind
+; CHECK: tail call i8* @objc_retain(i8* %p) [[NUW]]
 ; CHECK: true:
-; CHECK: tail call i8* @objc_autorelease(i8* %0) nounwind
+; CHECK: call i8* @objc_autorelease(i8* %0) [[NUW]]
 ; CHECK: }
 define void @test5(i8* %p, i1 %a) {
 entry:
@@ -102,7 +102,7 @@ entry:
   br i1 %a, label %true, label %false
 
 true:
-  tail call i8* @objc_autorelease(i8* %p) nounwind
+  call i8* @objc_autorelease(i8* %p) nounwind
   call void @use_pointer(i8* %p)
   ret void
 
@@ -119,8 +119,8 @@ false:
 ; Those entrypoints don't exist yet though.
 
 ; CHECK: define i8* @test6(
-; CHECK: call i8* @objc_retainAutoreleasedReturnValue(i8* %p) nounwind
-; CHECK: %t = tail call i8* @objc_autoreleaseReturnValue(i8* %1) nounwind
+; CHECK: call i8* @objc_retainAutoreleasedReturnValue(i8* %p) [[NUW]]
+; CHECK: %t = tail call i8* @objc_autoreleaseReturnValue(i8* %1) [[NUW]]
 ; CHECK: }
 define i8* @test6() {
   %p = call i8* @returner()
@@ -161,3 +161,5 @@ return:                                           ; preds = %if.then, %entry
   %retval = phi i8* [ %c, %if.then ], [ null, %entry ]
   ret i8* %retval
 }
+
+; CHECK: attributes [[NUW]] = { nounwind }
diff --git a/test/Transforms/ObjCARC/dont-infinite-loop-during-block-escape-analysis.ll b/test/Transforms/ObjCARC/dont-infinite-loop-during-block-escape-analysis.ll
new file mode 100644
index 0000000..bdee2be
--- /dev/null
+++ b/test/Transforms/ObjCARC/dont-infinite-loop-during-block-escape-analysis.ll
@@ -0,0 +1,87 @@
+; RUN: opt -S -objc-arc < %s
+; bugzilla://14551
+; rdar://12851911
+
+; Make sure that we do not hang clang during escape analysis.
+
+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-S128"
+target triple = "x86_64-darwin"
+
+%struct.__block_descriptor = type { i64, i64 }
+%struct.__block_byref_foo = type { i8*, %struct.__block_byref_foo*, i32, i32, i32 }
+
+@_NSConcreteGlobalBlock = external global i8*
+@.str = private unnamed_addr constant [6 x i8] c"v8@?0\00", align 1
+@__block_descriptor_tmp = internal constant { i64, i64, i8*, i8* } { i64 0, i64 32, i8* getelementptr inbounds ([6 x i8]* @.str, i32 0, i32 0), i8* null }
+@__block_literal_global = internal constant { i8**, i32, i32, i8*, %struct.__block_descriptor* } { i8** @_NSConcreteGlobalBlock, i32 1342177280, i32 0, i8* bitcast (void (i8*)* @__hang_clang_block_invoke to i8*), %struct.__block_descriptor* bitcast ({ i64, i64, i8*, i8* }* @__block_descriptor_tmp to %struct.__block_descriptor*) }, align 8
+
+define void @hang_clang() uwtable optsize ssp {
+entry:
+  %foo = alloca %struct.__block_byref_foo, align 8
+  %byref.isa = getelementptr inbounds %struct.__block_byref_foo* %foo, i64 0, i32 0
+  store i8* null, i8** %byref.isa, align 8
+  %byref.forwarding = getelementptr inbounds %struct.__block_byref_foo* %foo, i64 0, i32 1
+  store %struct.__block_byref_foo* %foo, %struct.__block_byref_foo** %byref.forwarding, align 8
+  %byref.flags = getelementptr inbounds %struct.__block_byref_foo* %foo, i64 0, i32 2
+  store i32 536870912, i32* %byref.flags, align 8
+  %byref.size = getelementptr inbounds %struct.__block_byref_foo* %foo, i64 0, i32 3
+  store i32 32, i32* %byref.size, align 4
+  %foo1 = getelementptr inbounds %struct.__block_byref_foo* %foo, i64 0, i32 4
+  store i32 0, i32* %foo1, align 8, !tbaa !4
+  br label %for.body
+
+for.body:                                         ; preds = %for.inc.for.body_crit_edge, %entry
+  %0 = phi i1 [ true, %entry ], [ %phitmp, %for.inc.for.body_crit_edge ]
+  %i.06 = phi i32 [ 1, %entry ], [ %phitmp8, %for.inc.for.body_crit_edge ]
+  %block.05 = phi void (...)* [ null, %entry ], [ %block.1, %for.inc.for.body_crit_edge ]
+  br i1 %0, label %for.inc, label %if.then
+
+if.then:                                          ; preds = %for.body
+  %1 = call i8* @objc_retainBlock(i8* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor* }* @__block_literal_global to i8*)) nounwind, !clang.arc.copy_on_escape !7
+  %2 = bitcast i8* %1 to void (...)*
+  %3 = bitcast void (...)* %block.05 to i8*
+  call void @objc_release(i8* %3) nounwind, !clang.imprecise_release !7
+  br label %for.inc
+
+for.inc:                                          ; preds = %for.body, %if.then
+  %block.1 = phi void (...)* [ %2, %if.then ], [ %block.05, %for.body ]
+  %exitcond = icmp eq i32 %i.06, 10
+  br i1 %exitcond, label %for.end, label %for.inc.for.body_crit_edge
+
+for.inc.for.body_crit_edge:                       ; preds = %for.inc
+  %.pre = load %struct.__block_byref_foo** %byref.forwarding, align 8
+  %foo2.phi.trans.insert = getelementptr inbounds %struct.__block_byref_foo* %.pre, i64 0, i32 4
+  %.pre7 = load i32* %foo2.phi.trans.insert, align 4, !tbaa !4
+  %phitmp = icmp eq i32 %.pre7, 0
+  %phitmp8 = add i32 %i.06, 1
+  br label %for.body
+
+for.end:                                          ; preds = %for.inc
+  %4 = bitcast %struct.__block_byref_foo* %foo to i8*
+  call void @_Block_object_dispose(i8* %4, i32 8)
+  %5 = bitcast void (...)* %block.1 to i8*
+  call void @objc_release(i8* %5) nounwind, !clang.imprecise_release !7
+  ret void
+}
+
+define internal void @__hang_clang_block_invoke(i8* nocapture %.block_descriptor) nounwind uwtable readnone optsize ssp {
+entry:
+  ret void
+}
+
+declare i8* @objc_retainBlock(i8*)
+
+declare void @objc_release(i8*) nonlazybind
+
+declare void @_Block_object_dispose(i8*, i32)
+
+!llvm.module.flags = !{!0, !1, !2, !3}
+
+!0 = metadata !{i32 1, metadata !"Objective-C Version", i32 2}
+!1 = metadata !{i32 1, metadata !"Objective-C Image Info Version", i32 0}
+!2 = metadata !{i32 1, metadata !"Objective-C Image Info Section", metadata !"__DATA, __objc_imageinfo, regular, no_dead_strip"}
+!3 = metadata !{i32 4, metadata !"Objective-C Garbage Collection", i32 0}
+!4 = metadata !{metadata !"int", metadata !5}
+!5 = metadata !{metadata !"omnipotent char", metadata !6}
+!6 = metadata !{metadata !"Simple C/C++ TBAA"}
+!7 = metadata !{}
diff --git a/test/Transforms/ObjCARC/ensure-that-exception-unwind-path-is-visited.ll b/test/Transforms/ObjCARC/ensure-that-exception-unwind-path-is-visited.ll
new file mode 100644
index 0000000..05257d1
--- /dev/null
+++ b/test/Transforms/ObjCARC/ensure-that-exception-unwind-path-is-visited.ll
@@ -0,0 +1,174 @@
+; RUN: opt -objc-arc -S < %s | FileCheck %s
+; rdar://11744105
+; bugzilla://14584
+
+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-S128"
+target triple = "x86_64-apple-macosx10.9.0"
+
+%0 = type opaque
+%struct._class_t = type { %struct._class_t*, %struct._class_t*, %struct._objc_cache*, i8* (i8*, i8*)**, %struct._class_ro_t* }
+%struct._objc_cache = type opaque
+%struct._class_ro_t = type { i32, i32, i32, i8*, i8*, %struct.__method_list_t*, %struct._objc_protocol_list*, %struct._ivar_list_t*, i8*, %struct._prop_list_t* }
+%struct.__method_list_t = type { i32, i32, [0 x %struct._objc_method] }
+%struct._objc_method = type { i8*, i8*, i8* }
+%struct._objc_protocol_list = type { i64, [0 x %struct._protocol_t*] }
+%struct._protocol_t = type { i8*, i8*, %struct._objc_protocol_list*, %struct.__method_list_t*, %struct.__method_list_t*, %struct.__method_list_t*, %struct.__method_list_t*, %struct._prop_list_t*, i32, i32, i8** }
+%struct._prop_list_t = type { i32, i32, [0 x %struct._prop_t] }
+%struct._prop_t = type { i8*, i8* }
+%struct._ivar_list_t = type { i32, i32, [0 x %struct._ivar_t] }
+%struct._ivar_t = type { i64*, i8*, i8*, i32, i32 }
+%struct.NSConstantString = type { i32*, i32, i8*, i64 }
+
+@"OBJC_CLASS_$_NSObject" = external global %struct._class_t
+@"\01L_OBJC_CLASSLIST_REFERENCES_$_" = internal global %struct._class_t* @"OBJC_CLASS_$_NSObject", section "__DATA, __objc_classrefs, regular, no_dead_strip", align 8
+@"\01L_OBJC_METH_VAR_NAME_" = internal global [4 x i8] c"new\00", section "__TEXT,__objc_methname,cstring_literals", align 1
+@"\01L_OBJC_SELECTOR_REFERENCES_" = internal global i8* getelementptr inbounds ([4 x i8]* @"\01L_OBJC_METH_VAR_NAME_", i64 0, i64 0), section "__DATA, __objc_selrefs, literal_pointers, no_dead_strip"
+@__CFConstantStringClassReference = external global [0 x i32]
+@.str = linker_private unnamed_addr constant [11 x i8] c"Failed: %@\00", align 1
+@_unnamed_cfstring_ = private constant %struct.NSConstantString { i32* getelementptr inbounds ([0 x i32]* @__CFConstantStringClassReference, i32 0, i32 0), i32 1992, i8* getelementptr inbounds ([11 x i8]* @.str, i32 0, i32 0), i64 10 }, section "__DATA,__cfstring"
+@"OBJC_CLASS_$_NSException" = external global %struct._class_t
+@"\01L_OBJC_CLASSLIST_REFERENCES_$_1" = internal global %struct._class_t* @"OBJC_CLASS_$_NSException", section "__DATA, __objc_classrefs, regular, no_dead_strip", align 8
+@.str2 = linker_private unnamed_addr constant [4 x i8] c"Foo\00", align 1
+@_unnamed_cfstring_3 = private constant %struct.NSConstantString { i32* getelementptr inbounds ([0 x i32]* @__CFConstantStringClassReference, i32 0, i32 0), i32 1992, i8* getelementptr inbounds ([4 x i8]* @.str2, i32 0, i32 0), i64 3 }, section "__DATA,__cfstring"
+@"\01L_OBJC_METH_VAR_NAME_4" = internal global [14 x i8] c"raise:format:\00", section "__TEXT,__objc_methname,cstring_literals", align 1
+@"\01L_OBJC_SELECTOR_REFERENCES_5" = internal global i8* getelementptr inbounds ([14 x i8]* @"\01L_OBJC_METH_VAR_NAME_4", i64 0, i64 0), section "__DATA, __objc_selrefs, literal_pointers, no_dead_strip"
+@llvm.used = appending global [6 x i8*] [i8* bitcast (%struct._class_t** @"\01L_OBJC_CLASSLIST_REFERENCES_$_" to i8*), i8* getelementptr inbounds ([4 x i8]* @"\01L_OBJC_METH_VAR_NAME_", i32 0, i32 0), i8* bitcast (i8** @"\01L_OBJC_SELECTOR_REFERENCES_" to i8*), i8* bitcast (%struct._class_t** @"\01L_OBJC_CLASSLIST_REFERENCES_$_1" to i8*), i8* getelementptr inbounds ([14 x i8]* @"\01L_OBJC_METH_VAR_NAME_4", i32 0, i32 0), i8* bitcast (i8** @"\01L_OBJC_SELECTOR_REFERENCES_5" to i8*)], section "llvm.metadata"
+
+define i32 @main() uwtable ssp {
+entry:
+  %tmp = load %struct._class_t** @"\01L_OBJC_CLASSLIST_REFERENCES_$_", align 8, !dbg !37
+  %tmp1 = load i8** @"\01L_OBJC_SELECTOR_REFERENCES_", align 8, !dbg !37, !invariant.load !38
+  %tmp2 = bitcast %struct._class_t* %tmp to i8*, !dbg !37
+; CHECK: call i8* bitcast (i8* (i8*, i8*, ...)* @objc_msgSend to i8* (i8*, i8*)*)(i8* %tmp2, i8* %tmp1)
+  %call = call i8* bitcast (i8* (i8*, i8*, ...)* @objc_msgSend to i8* (i8*, i8*)*)(i8* %tmp2, i8* %tmp1), !dbg !37, !clang.arc.no_objc_arc_exceptions !38
+  call void @llvm.dbg.value(metadata !{i8* %call}, i64 0, metadata !12), !dbg !37
+; CHECK: call i8* @objc_retain(i8* %call) [[NUW:#[0-9]+]]
+  %tmp3 = call i8* @objc_retain(i8* %call) nounwind, !dbg !39
+  call void @llvm.dbg.value(metadata !{i8* %call}, i64 0, metadata !25), !dbg !39
+  invoke fastcc void @ThrowFunc(i8* %call)
+          to label %eh.cont unwind label %lpad, !dbg !40, !clang.arc.no_objc_arc_exceptions !38
+
+eh.cont:                                          ; preds = %entry
+; CHECK: call void @objc_release(i8* %call)
+  call void @objc_release(i8* %call) nounwind, !dbg !42, !clang.imprecise_release !38
+  br label %if.end, !dbg !43
+
+lpad:                                             ; preds = %entry
+  %tmp4 = landingpad { i8*, i32 } personality i8* bitcast (i32 (...)* @__objc_personality_v0 to i8*)
+          catch i8* null, !dbg !40
+  %tmp5 = extractvalue { i8*, i32 } %tmp4, 0, !dbg !40
+  %exn.adjusted = call i8* @objc_begin_catch(i8* %tmp5) nounwind, !dbg !44
+  call void @llvm.dbg.value(metadata !45, i64 0, metadata !21), !dbg !46
+  call void @objc_end_catch(), !dbg !49, !clang.arc.no_objc_arc_exceptions !38
+; CHECK: call void @objc_release(i8* %call)
+  call void @objc_release(i8* %call) nounwind, !dbg !42, !clang.imprecise_release !38
+  call void (i8*, ...)* @NSLog(i8* bitcast (%struct.NSConstantString* @_unnamed_cfstring_ to i8*), i8* %call), !dbg !50, !clang.arc.no_objc_arc_exceptions !38
+  br label %if.end, !dbg !52
+
+if.end:                                           ; preds = %lpad, %eh.cont
+  call void (i8*, ...)* @NSLog(i8* bitcast (%struct.NSConstantString* @_unnamed_cfstring_ to i8*), i8* %call), !dbg !53, !clang.arc.no_objc_arc_exceptions !38
+; CHECK: call void @objc_release(i8* %call)
+  call void @objc_release(i8* %call) nounwind, !dbg !54, !clang.imprecise_release !38
+  ret i32 0, !dbg !54
+}
+
+declare void @llvm.dbg.declare(metadata, metadata) nounwind readnone
+
+declare i8* @objc_msgSend(i8*, i8*, ...) nonlazybind
+
+declare i8* @objc_retain(i8*) nonlazybind
+
+declare i8* @objc_begin_catch(i8*)
+
+declare void @objc_end_catch()
+
+declare void @objc_exception_rethrow()
+
+define internal fastcc void @ThrowFunc(i8* %obj) uwtable noinline ssp {
+entry:
+  %tmp = call i8* @objc_retain(i8* %obj) nounwind
+  call void @llvm.dbg.value(metadata !{i8* %obj}, i64 0, metadata !32), !dbg !55
+  %tmp1 = load %struct._class_t** @"\01L_OBJC_CLASSLIST_REFERENCES_$_1", align 8, !dbg !56
+  %tmp2 = load i8** @"\01L_OBJC_SELECTOR_REFERENCES_5", align 8, !dbg !56, !invariant.load !38
+  %tmp3 = bitcast %struct._class_t* %tmp1 to i8*, !dbg !56
+  call void (i8*, i8*, %0*, %0*, ...)* bitcast (i8* (i8*, i8*, ...)* @objc_msgSend to void (i8*, i8*, %0*, %0*, ...)*)(i8* %tmp3, i8* %tmp2, %0* bitcast (%struct.NSConstantString* @_unnamed_cfstring_3 to %0*), %0* bitcast (%struct.NSConstantString* @_unnamed_cfstring_3 to %0*)), !dbg !56, !clang.arc.no_objc_arc_exceptions !38
+  call void @objc_release(i8* %obj) nounwind, !dbg !58, !clang.imprecise_release !38
+  ret void, !dbg !58
+}
+
+declare i32 @__objc_personality_v0(...)
+
+declare void @objc_release(i8*) nonlazybind
+
+declare void @NSLog(i8*, ...)
+
+declare void @llvm.dbg.value(metadata, i64, metadata) nounwind readnone
+
+; CHECK: attributes #0 = { ssp uwtable }
+; CHECK: attributes #1 = { nounwind readnone }
+; CHECK: attributes #2 = { nonlazybind }
+; CHECK: attributes #3 = { noinline ssp uwtable }
+; CHECK: attributes [[NUW]] = { nounwind }
+
+!llvm.dbg.cu = !{!0}
+!llvm.module.flags = !{!33, !34, !35, !36}
+
+!0 = metadata !{i32 786449, i32 0, i32 16, metadata !"test.m", metadata !"/Volumes/Files/gottesmmcab/Radar/12906997", metadata !"clang version 3.3 ", i1 true, i1 true, metadata !"", i32 2, metadata !1, metadata !1, metadata !3, metadata !1} ; [ DW_TAG_compile_unit ] [/Volumes/Files/gottesmmcab/Radar/12906997/test.m] [DW_LANG_ObjC]
+!1 = metadata !{metadata !2}
+!2 = metadata !{i32 0}
+!3 = metadata !{metadata !4}
+!4 = metadata !{metadata !5, metadata !27}
+!5 = metadata !{i32 786478, i32 0, metadata !6, metadata !"main", metadata !"main", metadata !"", metadata !6, i32 9, metadata !7, i1 false, i1 true, i32 0, i32 0, null, i32 0, i1 true, i32 ()* @main, null, null, metadata !10, i32 10} ; [ DW_TAG_subprogram ] [line 9] [def] [scope 10] [main]
+!6 = metadata !{i32 786473, metadata !"test.m", metadata !"/Volumes/Files/gottesmmcab/Radar/12906997", null} ; [ DW_TAG_file_type ]
+!7 = metadata !{i32 786453, i32 0, metadata !"", i32 0, i32 0, i64 0, i64 0, i64 0, i32 0, null, metadata !8, i32 0, i32 0} ; [ DW_TAG_subroutine_type ] [line 0, size 0, align 0, offset 0] [from ]
+!8 = metadata !{metadata !9}
+!9 = metadata !{i32 786468, null, metadata !"int", null, i32 0, i64 32, i64 32, i64 0, i32 0, i32 5} ; [ DW_TAG_base_type ] [int] [line 0, size 32, align 32, offset 0, enc DW_ATE_signed]
+!10 = metadata !{metadata !11}
+!11 = metadata !{metadata !12, metadata !21, metadata !25}
+!12 = metadata !{i32 786688, metadata !13, metadata !"obj", metadata !6, i32 11, metadata !14, i32 0, i32 0} ; [ DW_TAG_auto_variable ] [obj] [line 11]
+!13 = metadata !{i32 786443, metadata !5, i32 10, i32 0, metadata !6, i32 0} ; [ DW_TAG_lexical_block ] [/Volumes/Files/gottesmmcab/Radar/12906997/test.m]
+!14 = metadata !{i32 786454, null, metadata !"id", metadata !6, i32 11, i64 0, i64 0, i64 0, i32 0, metadata !15} ; [ DW_TAG_typedef ] [id] [line 11, size 0, align 0, offset 0] [from ]
+!15 = metadata !{i32 786447, null, metadata !"", null, i32 0, i64 64, i64 64, i64 0, i32 0, metadata !16} ; [ DW_TAG_pointer_type ] [line 0, size 64, align 64, offset 0] [from objc_object]
+!16 = metadata !{i32 786451, null, metadata !"objc_object", metadata !6, i32 0, i64 0, i64 0, i32 0, i32 0, null, metadata !17, i32 0, i32 0, i32 0} ; [ DW_TAG_structure_type ] [objc_object] [line 0, size 0, align 0, offset 0] [from ]
+!17 = metadata !{metadata !18}
+!18 = metadata !{i32 786445, metadata !16, metadata !"isa", metadata !6, i32 0, i64 64, i64 0, i64 0, i32 0, metadata !19} ; [ DW_TAG_member ] [isa] [line 0, size 64, align 0, offset 0] [from ]
+!19 = metadata !{i32 786447, null, metadata !"", null, i32 0, i64 64, i64 0, i64 0, i32 0, metadata !20} ; [ DW_TAG_pointer_type ] [line 0, size 64, align 0, offset 0] [from objc_class]
+!20 = metadata !{i32 786451, null, metadata !"objc_class", metadata !6, i32 0, i64 0, i64 0, i32 0, i32 4, null, null, i32 0} ; [ DW_TAG_structure_type ] [objc_class] [line 0, size 0, align 0, offset 0] [fwd] [from ]
+!21 = metadata !{i32 786688, metadata !22, metadata !"ok", metadata !6, i32 13, metadata !23, i32 0, i32 0} ; [ DW_TAG_auto_variable ] [ok] [line 13]
+!22 = metadata !{i32 786443, metadata !13, i32 12, i32 0, metadata !6, i32 1} ; [ DW_TAG_lexical_block ] [/Volumes/Files/gottesmmcab/Radar/12906997/test.m]
+!23 = metadata !{i32 786454, null, metadata !"BOOL", metadata !6, i32 62, i64 0, i64 0, i64 0, i32 0, metadata !24} ; [ DW_TAG_typedef ] [BOOL] [line 62, size 0, align 0, offset 0] [from signed char]
+!24 = metadata !{i32 786468, null, metadata !"signed char", null, i32 0, i64 8, i64 8, i64 0, i32 0, i32 6} ; [ DW_TAG_base_type ] [signed char] [line 0, size 8, align 8, offset 0, enc DW_ATE_signed_char]
+!25 = metadata !{i32 786688, metadata !26, metadata !"obj2", metadata !6, i32 15, metadata !14, i32 0, i32 0} ; [ DW_TAG_auto_variable ] [obj2] [line 15]
+!26 = metadata !{i32 786443, metadata !22, i32 14, i32 0, metadata !6, i32 2} ; [ DW_TAG_lexical_block ] [/Volumes/Files/gottesmmcab/Radar/12906997/test.m]
+!27 = metadata !{i32 786478, i32 0, metadata !6, metadata !"ThrowFunc", metadata !"ThrowFunc", metadata !"", metadata !6, i32 4, metadata !28, i1 true, i1 true, i32 0, i32 0, null, i32 256, i1 true, void (i8*)* @ThrowFunc, null, null, metadata !30, i32 5} ; [ DW_TAG_subprogram ] [line 4] [local] [def] [scope 5] [ThrowFunc]
+!28 = metadata !{i32 786453, i32 0, metadata !"", i32 0, i32 0, i64 0, i64 0, i64 0, i32 0, null, metadata !29, i32 0, i32 0} ; [ DW_TAG_subroutine_type ] [line 0, size 0, align 0, offset 0] [from ]
+!29 = metadata !{null, metadata !14}
+!30 = metadata !{metadata !31}
+!31 = metadata !{metadata !32}
+!32 = metadata !{i32 786689, metadata !27, metadata !"obj", metadata !6, i32 16777220, metadata !14, i32 0, i32 0} ; [ DW_TAG_arg_variable ] [obj] [line 4]
+!33 = metadata !{i32 1, metadata !"Objective-C Version", i32 2}
+!34 = metadata !{i32 1, metadata !"Objective-C Image Info Version", i32 0}
+!35 = metadata !{i32 1, metadata !"Objective-C Image Info Section", metadata !"__DATA, __objc_imageinfo, regular, no_dead_strip"}
+!36 = metadata !{i32 4, metadata !"Objective-C Garbage Collection", i32 0}
+!37 = metadata !{i32 11, i32 0, metadata !13, null}
+!38 = metadata !{}
+!39 = metadata !{i32 15, i32 0, metadata !26, null}
+!40 = metadata !{i32 17, i32 0, metadata !41, null}
+!41 = metadata !{i32 786443, metadata !26, i32 16, i32 0, metadata !6, i32 3} ; [ DW_TAG_lexical_block ] [/Volumes/Files/gottesmmcab/Radar/12906997/test.m]
+!42 = metadata !{i32 22, i32 0, metadata !26, null}
+!43 = metadata !{i32 23, i32 0, metadata !22, null}
+!44 = metadata !{i32 19, i32 0, metadata !41, null}
+!45 = metadata !{i8 0}
+!46 = metadata !{i32 20, i32 0, metadata !47, null}
+!47 = metadata !{i32 786443, metadata !48, i32 19, i32 0, metadata !6, i32 5} ; [ DW_TAG_lexical_block ] [/Volumes/Files/gottesmmcab/Radar/12906997/test.m]
+!48 = metadata !{i32 786443, metadata !26, i32 19, i32 0, metadata !6, i32 4} ; [ DW_TAG_lexical_block ] [/Volumes/Files/gottesmmcab/Radar/12906997/test.m]
+!49 = metadata !{i32 21, i32 0, metadata !47, null}
+!50 = metadata !{i32 24, i32 0, metadata !51, null}
+!51 = metadata !{i32 786443, metadata !22, i32 23, i32 0, metadata !6, i32 6} ; [ DW_TAG_lexical_block ] [/Volumes/Files/gottesmmcab/Radar/12906997/test.m]
+!52 = metadata !{i32 25, i32 0, metadata !51, null}
+!53 = metadata !{i32 27, i32 0, metadata !13, null}
+!54 = metadata !{i32 28, i32 0, metadata !13, null}
+!55 = metadata !{i32 4, i32 0, metadata !27, null}
+!56 = metadata !{i32 6, i32 0, metadata !57, null}
+!57 = metadata !{i32 786443, metadata !27, i32 5, i32 0, metadata !6, i32 7} ; [ DW_TAG_lexical_block ] [/Volumes/Files/gottesmmcab/Radar/12906997/test.m]
+!58 = metadata !{i32 7, i32 0, metadata !57, null}
diff --git a/test/Transforms/ObjCARC/escape.ll b/test/Transforms/ObjCARC/escape.ll
index 3f694cf..8f252a0 100644
--- a/test/Transforms/ObjCARC/escape.ll
+++ b/test/Transforms/ObjCARC/escape.ll
@@ -10,8 +10,8 @@ target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f3
 ; with the objc_storeWeak call.
 
 ; CHECK: define void @test0(
-; CHECK: %tmp7 = call i8* @objc_retainBlock(i8* %tmp6) nounwind, !clang.arc.copy_on_escape !0
-; CHECK: call void @objc_release(i8* %tmp7) nounwind, !clang.imprecise_release !0
+; CHECK: %tmp7 = call i8* @objc_retainBlock(i8* %tmp6) [[NUW:#[0-9]+]], !clang.arc.copy_on_escape !0
+; CHECK: call void @objc_release(i8* %tmp7) [[NUW]], !clang.imprecise_release !0
 ; CHECK: }
 define void @test0() nounwind {
 entry:
@@ -129,3 +129,6 @@ declare i8* @not_really_objc_storeWeak(i8**, i8*)
 declare void @objc_release(i8*)
 
 !0 = metadata !{}
+
+; CHECK: attributes [[NUW]] = { nounwind }
+; CHECK: attributes #1 = { nounwind ssp }
diff --git a/test/Transforms/ObjCARC/gvn.ll b/test/Transforms/ObjCARC/gvn.ll
index 6917b02..3648866 100644
--- a/test/Transforms/ObjCARC/gvn.ll
+++ b/test/Transforms/ObjCARC/gvn.ll
@@ -1,4 +1,4 @@
-; RUN: opt -S -basicaa -objc-arc -gvn < %s | FileCheck %s
+; RUN: opt -S -basicaa -objc-arc-aa -gvn < %s | FileCheck %s
 
 @x = common global i8* null, align 8
 
diff --git a/test/Transforms/ObjCARC/invoke.ll b/test/Transforms/ObjCARC/invoke.ll
index 1a58e34..f528b4a 100644
--- a/test/Transforms/ObjCARC/invoke.ll
+++ b/test/Transforms/ObjCARC/invoke.ll
@@ -12,10 +12,10 @@ declare i8* @returner()
 
 ; CHECK: define void @test0(
 ; CHECK: invoke.cont:
-; CHECK:   call void @objc_release(i8* %zipFile) nounwind, !clang.imprecise_release !0
+; CHECK:   call void @objc_release(i8* %zipFile) [[NUW:#[0-9]+]], !clang.imprecise_release !0
 ; CHECK:   ret void
 ; CHECK: lpad:
-; CHECK:   call void @objc_release(i8* %zipFile) nounwind, !clang.imprecise_release !0
+; CHECK:   call void @objc_release(i8* %zipFile) [[NUW]], !clang.imprecise_release !0
 ; CHECK:   ret void
 define void @test0(i8* %zipFile) {
 entry:
@@ -39,11 +39,11 @@ lpad:                                             ; preds = %entry
 
 ; CHECK: define void @test1(
 ; CHECK: invoke.cont:
-; CHECK:   call void @objc_release(i8* %zipFile) nounwind, !clang.imprecise_release !0
+; CHECK:   call void @objc_release(i8* %zipFile) [[NUW]], !clang.imprecise_release !0
 ; CHECK:   call void @callee()
 ; CHECK:   br label %done
 ; CHECK: lpad:
-; CHECK:   call void @objc_release(i8* %zipFile) nounwind, !clang.imprecise_release !0
+; CHECK:   call void @objc_release(i8* %zipFile) [[NUW]], !clang.imprecise_release !0
 ; CHECK:   call void @callee()
 ; CHECK:   br label %done
 ; CHECK: done:
@@ -108,7 +108,7 @@ finally.rethrow:                                  ; preds = %invoke.cont, %entry
 
 ; CHECK: define void @test3(
 ; CHECK: if.end:
-; CHECK-NEXT: call void @objc_release(i8* %p) nounwind
+; CHECK-NEXT: call void @objc_release(i8* %p) [[NUW]]
 ; CHECK-NEXT: ret void
 define void @test3(i8* %p, i1 %b) {
 entry:
@@ -140,10 +140,10 @@ if.end:
 ; CHECK: lpad:
 ; CHECK-NEXT: %r = landingpad { i8*, i32 } personality i8* bitcast (i32 (...)* @__objc_personality_v0 to i8*)
 ; CHECK-NEXT: cleanup
-; CHECK-NEXT: call void @objc_release(i8* %p) nounwind
+; CHECK-NEXT: call void @objc_release(i8* %p) [[NUW]]
 ; CHECK-NEXT: ret void
 ; CHECK: if.end:
-; CHECK-NEXT: call void @objc_release(i8* %p) nounwind
+; CHECK-NEXT: call void @objc_release(i8* %p) [[NUW]]
 ; CHECK-NEXT: ret void
 define void @test4(i8* %p, i1 %b) {
 entry:
@@ -215,4 +215,6 @@ if.end:
 declare i32 @__gxx_personality_v0(...)
 declare i32 @__objc_personality_v0(...)
 
+; CHECK: attributes [[NUW]] = { nounwind }
+
 !0 = metadata !{}
diff --git a/test/Transforms/ObjCARC/move-and-form-retain-autorelease.ll b/test/Transforms/ObjCARC/move-and-form-retain-autorelease.ll
index 170d0a9..5d05825 100644
--- a/test/Transforms/ObjCARC/move-and-form-retain-autorelease.ll
+++ b/test/Transforms/ObjCARC/move-and-form-retain-autorelease.ll
@@ -4,7 +4,7 @@
 ; and various scary looking things and fold it into an objc_retainAutorelease.
 
 ; CHECK: bb57:
-; CHECK: tail call i8* @objc_retainAutorelease(i8* %tmp71x) nounwind
+; CHECK: tail call i8* @objc_retainAutorelease(i8* %tmp71x) [[NUW:#[0-9]+]]
 ; CHECK: bb99:
 
 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"
@@ -212,10 +212,12 @@ bb99:                                             ; preds = %bb57
   br label %bb104
 
 bb104:                                            ; preds = %bb99, %bb57
-  %tmp105 = tail call i8* @objc_autorelease(i8* %tmp72) nounwind
+  %tmp105 = call i8* @objc_autorelease(i8* %tmp72) nounwind
   %tmp106 = bitcast i8* %tmp105 to %14*
   tail call void @objc_release(i8* %tmp85) nounwind
   %tmp107 = bitcast %18* %tmp47 to i8*
   tail call void @objc_release(i8* %tmp107) nounwind
   ret %14* %tmp106
 }
+
+; CHECK: attributes [[NUW]] = { nounwind }
diff --git a/test/Transforms/ObjCARC/nested.ll b/test/Transforms/ObjCARC/nested.ll
index 32be03e..ca9c58b 100644
--- a/test/Transforms/ObjCARC/nested.ll
+++ b/test/Transforms/ObjCARC/nested.ll
@@ -770,9 +770,9 @@ forcoll.empty:
 @__block_d_tmp5 = external hidden constant { i64, i64, i8*, i8*, i8*, i8* }
 
 ; CHECK: define void @test11(
-; CHECK: tail call i8* @objc_retain(i8* %call) nounwind
-; CHECK: tail call i8* @objc_retain(i8* %call) nounwind
-; CHECK: call void @objc_release(i8* %call) nounwind, !clang.imprecise_release !0
+; CHECK: tail call i8* @objc_retain(i8* %call) [[NUW:#[0-9]+]]
+; CHECK: tail call i8* @objc_retain(i8* %call) [[NUW]]
+; CHECK: call void @objc_release(i8* %call) [[NUW]], !clang.imprecise_release !0
 ; CHECK: }
 define void @test11() {
 entry:
@@ -820,3 +820,6 @@ entry:
   call void @objc_release(i8* %call) nounwind, !clang.imprecise_release !0
   ret void
 }
+
+; CHECK: attributes [[NUW]] = { nounwind }
+; CHECK: attributes #1 = { nonlazybind }
diff --git a/test/Transforms/ObjCARC/retain-block-alloca.ll b/test/Transforms/ObjCARC/retain-block-alloca.ll
index 01f2087..f40be23 100644
--- a/test/Transforms/ObjCARC/retain-block-alloca.ll
+++ b/test/Transforms/ObjCARC/retain-block-alloca.ll
@@ -9,7 +9,7 @@
 @"\01L_OBJC_SELECTOR_REFERENCES_" = external hidden global i8*, section "__DATA, __objc_selrefs, literal_pointers, no_dead_strip"
 
 ; CHECK: define void @test(
-; CHECK: %3 = call i8* @objc_retainBlock(i8* %2) nounwind
+; CHECK: %3 = call i8* @objc_retainBlock(i8* %2) [[NUW:#[0-9]+]]
 ; CHECK: @objc_msgSend
 ; CHECK-NEXT: @objc_release(i8* %3)
 define void @test(%0* %array) uwtable {
@@ -87,4 +87,8 @@ declare i8* @objc_msgSend(i8*, i8*, ...) nonlazybind
 
 declare void @objc_release(i8*)
 
+; CHECK: attributes #0 = { uwtable }
+; CHECK: attributes #1 = { nonlazybind }
+; CHECK: attributes [[NUW]] = { nounwind }
+
 !0 = metadata !{}
diff --git a/test/Transforms/ObjCARC/retain-block-side-effects.ll b/test/Transforms/ObjCARC/retain-block-side-effects.ll
index e84d48f..7fa73cb 100644
--- a/test/Transforms/ObjCARC/retain-block-side-effects.ll
+++ b/test/Transforms/ObjCARC/retain-block-side-effects.ll
@@ -4,7 +4,7 @@
 ; objc_retainBlock stores into %repeater so the load from after the
 ; call isn't forwardable from the store before the call.
 
-; CHECK: %tmp16 = call i8* @objc_retainBlock(i8* %tmp15) nounwind
+; CHECK: %tmp16 = call i8* @objc_retainBlock(i8* %tmp15) [[NUW:#[0-9]+]]
 ; CHECK: %tmp17 = bitcast i8* %tmp16 to void ()*
 ; CHECK: %tmp18 = load %struct.__block_byref_repeater** %byref.forwarding, align 8
 ; CHECK: %repeater12 = getelementptr inbounds %struct.__block_byref_repeater* %tmp18, i64 0, i32 6
@@ -37,3 +37,6 @@ entry:
 }
 
 declare i8* @objc_retainBlock(i8*)
+
+; CHECK: attributes #0 = { noreturn }
+; CHECK: attributes [[NUW]] = { nounwind }
diff --git a/test/Transforms/ObjCARC/retain-block.ll b/test/Transforms/ObjCARC/retain-block.ll
index b3b62d3..ee57049 100644
--- a/test/Transforms/ObjCARC/retain-block.ll
+++ b/test/Transforms/ObjCARC/retain-block.ll
@@ -28,8 +28,8 @@ entry:
 ; optimization possible.
 
 ; CHECK: define void @test0_no_metadata(i8* %tmp) {
-; CHECK: %tmp2 = tail call i8* @objc_retainBlock(i8* %tmp) nounwind
-; CHECK: tail call void @objc_release(i8* %tmp2) nounwind, !clang.imprecise_release !0
+; CHECK: %tmp2 = tail call i8* @objc_retainBlock(i8* %tmp) [[NUW:#[0-9]+]]
+; CHECK: tail call void @objc_release(i8* %tmp2) [[NUW]], !clang.imprecise_release !0
 ; CHECK: }
 define void @test0_no_metadata(i8* %tmp) {
 entry:
@@ -43,8 +43,8 @@ entry:
 ; optimization possible.
 
 ; CHECK: define void @test0_escape(i8* %tmp, i8** %z) {
-; CHECK: %tmp2 = tail call i8* @objc_retainBlock(i8* %tmp) nounwind, !clang.arc.copy_on_escape !0
-; CHECK: tail call void @objc_release(i8* %tmp2) nounwind, !clang.imprecise_release !0
+; CHECK: %tmp2 = tail call i8* @objc_retainBlock(i8* %tmp) [[NUW]], !clang.arc.copy_on_escape !0
+; CHECK: tail call void @objc_release(i8* %tmp2) [[NUW]], !clang.imprecise_release !0
 ; CHECK: }
 define void @test0_escape(i8* %tmp, i8** %z) {
 entry:
@@ -58,8 +58,8 @@ entry:
 ; Same as test0_escape, but there's no intervening call.
 
 ; CHECK: define void @test0_just_escape(i8* %tmp, i8** %z) {
-; CHECK: %tmp2 = tail call i8* @objc_retainBlock(i8* %tmp) nounwind, !clang.arc.copy_on_escape !0
-; CHECK: tail call void @objc_release(i8* %tmp2) nounwind, !clang.imprecise_release !0
+; CHECK: %tmp2 = tail call i8* @objc_retainBlock(i8* %tmp) [[NUW]], !clang.arc.copy_on_escape !0
+; CHECK: tail call void @objc_release(i8* %tmp2) [[NUW]], !clang.imprecise_release !0
 ; CHECK: }
 define void @test0_just_escape(i8* %tmp, i8** %z) {
 entry:
@@ -73,9 +73,9 @@ entry:
 
 ; CHECK: define void @test1(i8* %tmp) {
 ; CHECK-NOT: @objc
-; CHECK: tail call i8* @objc_retain(i8* %tmp) nounwind
+; CHECK: tail call i8* @objc_retain(i8* %tmp) [[NUW]]
 ; CHECK-NOT: @objc
-; CHECK: tail call void @objc_release(i8* %tmp) nounwind, !clang.imprecise_release !0
+; CHECK: tail call void @objc_release(i8* %tmp) [[NUW]], !clang.imprecise_release !0
 ; CHECK-NOT: @objc
 ; CHECK: }
 define void @test1(i8* %tmp) {
@@ -95,10 +95,10 @@ entry:
 
 ; CHECK: define void @test1_no_metadata(i8* %tmp) {
 ; CHECK-NEXT: entry:
-; CHECK-NEXT: tail call i8* @objc_retainBlock(i8* %tmp) nounwind
+; CHECK-NEXT: tail call i8* @objc_retainBlock(i8* %tmp) [[NUW]]
 ; CHECK-NEXT: @use_pointer(i8* %tmp2)
 ; CHECK-NEXT: @use_pointer(i8* %tmp2)
-; CHECK-NEXT: tail call void @objc_release(i8* %tmp) nounwind, !clang.imprecise_release !0
+; CHECK-NEXT: tail call void @objc_release(i8* %tmp) [[NUW]], !clang.imprecise_release !0
 ; CHECK-NOT: @objc
 ; CHECK: }
 define void @test1_no_metadata(i8* %tmp) {
@@ -118,11 +118,11 @@ entry:
 
 ; CHECK: define void @test1_escape(i8* %tmp, i8** %z) {
 ; CHECK-NEXT: entry:
-; CHECK-NEXT: %tmp2 = tail call i8* @objc_retainBlock(i8* %tmp) nounwind, !clang.arc.copy_on_escape !0
+; CHECK-NEXT: %tmp2 = tail call i8* @objc_retainBlock(i8* %tmp) [[NUW]], !clang.arc.copy_on_escape !0
 ; CHECK-NEXT: store i8* %tmp2, i8** %z
 ; CHECK-NEXT: @use_pointer(i8* %tmp2)
 ; CHECK-NEXT: @use_pointer(i8* %tmp2)
-; CHECK-NEXT: tail call void @objc_release(i8* %tmp) nounwind, !clang.imprecise_release !0
+; CHECK-NEXT: tail call void @objc_release(i8* %tmp) [[NUW]], !clang.imprecise_release !0
 ; CHECK-NOT: @objc
 ; CHECK: }
 define void @test1_escape(i8* %tmp, i8** %z) {
@@ -136,3 +136,5 @@ entry:
   tail call void @objc_release(i8* %tmp) nounwind, !clang.imprecise_release !0
   ret void
 }
+
+; CHECK: attributes [[NUW]] = { nounwind }
diff --git a/test/Transforms/ObjCARC/retain-not-declared.ll b/test/Transforms/ObjCARC/retain-not-declared.ll
index f876e51..e834179 100644
--- a/test/Transforms/ObjCARC/retain-not-declared.ll
+++ b/test/Transforms/ObjCARC/retain-not-declared.ll
@@ -13,7 +13,7 @@ declare void @objc_release(i8*)
 
 ; CHECK:      define i8* @test0(i8* %p) {
 ; CHECK-NEXT: entry:
-; CHECK-NEXT:   %0 = tail call i8* @objc_retainAutoreleaseReturnValue(i8* %p) nounwind
+; CHECK-NEXT:   %0 = tail call i8* @objc_retainAutoreleaseReturnValue(i8* %p) [[NUW:#[0-9]+]]
 ; CHECK-NEXT:   ret i8* %0
 ; CHECK-NEXT: }
 
@@ -65,3 +65,5 @@ lpad100:                                          ; preds = %invoke.cont93
 declare i32 @__gxx_personality_v0(...)
 
 !0 = metadata !{}
+
+; CHECK: attributes [[NUW]] = { nounwind }
diff --git a/test/Transforms/ObjCARC/rle-s2l.ll b/test/Transforms/ObjCARC/rle-s2l.ll
index 8f8d5c0..2865c94 100644
--- a/test/Transforms/ObjCARC/rle-s2l.ll
+++ b/test/Transforms/ObjCARC/rle-s2l.ll
@@ -57,7 +57,7 @@ define void @test2(i8** %p) {
 
 ; CHECK:      define void @test3(i8** %p) {
 ; CHECK-NEXT:   %x = call i8* @objc_loadWeak(i8** %p)
-; CHECK-NEXT:   call void @use_pointer(i8* %x) readonly
+; CHECK-NEXT:   call void @use_pointer(i8* %x) [[RO:#[0-9]+]]
 ; CHECK-NEXT:   %1 = tail call i8* @objc_retain(i8* %x)
 ; CHECK-NEXT:   call void @use_pointer(i8* %x)
 ; CHECK-NEXT:   ret void
@@ -74,7 +74,7 @@ define void @test3(i8** %p) {
 
 ; CHECK:      define void @test4(i8** %p) {
 ; CHECK-NEXT:   %x = call i8* @objc_loadWeak(i8** %p)
-; CHECK-NEXT:   call void @use_pointer(i8* %x) readonly
+; CHECK-NEXT:   call void @use_pointer(i8* %x) [[RO]]
 ; CHECK-NEXT:   call void @callee()
 ; CHECK-NEXT:   %y = call i8* @objc_loadWeak(i8** %p)
 ; CHECK-NEXT:   call void @use_pointer(i8* %y)
@@ -133,3 +133,6 @@ define void @test7(i8** %p, i8* %n, i8** %q, i8* %m) {
   call void @use_pointer(i8* %y)
   ret void
 }
+
+; CHECK: attributes #0 = { nounwind }
+; CHECK: attributes [[RO]] = { readonly }
diff --git a/test/Transforms/ObjCARC/rv.ll b/test/Transforms/ObjCARC/rv.ll
index 9353a19..a2fef96 100644
--- a/test/Transforms/ObjCARC/rv.ll
+++ b/test/Transforms/ObjCARC/rv.ll
@@ -29,7 +29,7 @@ declare i8* @returner()
 ; CHECK:      define void @test0(
 ; CHECK-NEXT: entry:
 ; CHECK-NEXT:   %x = call i8* @returner
-; CHECK-NEXT:   %0 = tail call i8* @objc_retainAutoreleasedReturnValue(i8* %x) nounwind
+; CHECK-NEXT:   %0 = tail call i8* @objc_retainAutoreleasedReturnValue(i8* %x) [[NUW:#[0-9]+]]
 ; CHECK: t:
 ; CHECK-NOT: @objc_
 ; CHECK: return:
@@ -150,7 +150,7 @@ define void @test8() {
 ; Don't apply the RV optimization to autorelease if there's no retain.
 
 ; CHECK: define i8* @test9(i8* %p)
-; CHECK: tail call i8* @objc_autorelease(i8* %p)
+; CHECK: call i8* @objc_autorelease(i8* %p)
 define i8* @test9(i8* %p) {
   call i8* @objc_autorelease(i8* %p)
   ret i8* %p
@@ -159,8 +159,8 @@ define i8* @test9(i8* %p) {
 ; Apply the RV optimization.
 
 ; CHECK: define i8* @test10(i8* %p)
-; CHECK: tail call i8* @objc_retain(i8* %p) nounwind
-; CHECK: tail call i8* @objc_autoreleaseReturnValue(i8* %p) nounwind
+; CHECK: tail call i8* @objc_retain(i8* %p) [[NUW]]
+; CHECK: tail call i8* @objc_autoreleaseReturnValue(i8* %p) [[NUW]]
 ; CHECK-NEXT: ret i8* %p
 define i8* @test10(i8* %p) {
   %1 = call i8* @objc_retain(i8* %p)
@@ -174,7 +174,7 @@ define i8* @test10(i8* %p) {
 ; CHECK: define i8* @test11(i8* %p)
 ; CHECK: tail call i8* @objc_retain(i8* %p)
 ; CHECK-NEXT: call void @use_pointer(i8* %p)
-; CHECK: tail call i8* @objc_autorelease(i8* %p)
+; CHECK: call i8* @objc_autorelease(i8* %p)
 ; CHECK-NEXT: ret i8* %p
 define i8* @test11(i8* %p) {
   %1 = call i8* @objc_retain(i8* %p)
@@ -201,7 +201,7 @@ define i8* @test12(i8* %p) {
 
 ; CHECK: define i8* @test13(
 ; CHECK: tail call i8* @objc_retainAutoreleasedReturnValue(i8* %p)
-; CHECK: tail call i8* @objc_autorelease(i8* %p)
+; CHECK: call i8* @objc_autorelease(i8* %p)
 ; CHECK: ret i8* %p
 define i8* @test13() {
   %p = call i8* @returner()
@@ -215,7 +215,7 @@ define i8* @test13() {
 ; argument is not a return value.
 
 ; CHECK: define void @test14(
-; CHECK-NEXT: tail call i8* @objc_retain(i8* %p) nounwind
+; CHECK-NEXT: tail call i8* @objc_retain(i8* %p) [[NUW]]
 ; CHECK-NEXT: ret void
 define void @test14(i8* %p) {
   call i8* @objc_retainAutoreleasedReturnValue(i8* %p)
@@ -227,7 +227,7 @@ define void @test14(i8* %p) {
 
 ; CHECK: define void @test15(
 ; CHECK-NEXT: %y = call i8* @returner()
-; CHECK-NEXT: tail call i8* @objc_retainAutoreleasedReturnValue(i8* %y) nounwind
+; CHECK-NEXT: tail call i8* @objc_retainAutoreleasedReturnValue(i8* %y) [[NUW]]
 ; CHECK-NEXT: ret void
 define void @test15() {
   %y = call i8* @returner()
@@ -240,7 +240,7 @@ define void @test15() {
 
 ; CHECK: define void @test16(
 ; CHECK-NEXT: %y = call i8* @returner()
-; CHECK-NEXT: tail call i8* @objc_retainAutoreleasedReturnValue(i8* %y) nounwind
+; CHECK-NEXT: tail call i8* @objc_retainAutoreleasedReturnValue(i8* %y) [[NUW]]
 ; CHECK-NEXT: ret void
 define void @test16() {
   %y = call i8* @returner()
@@ -252,7 +252,7 @@ define void @test16() {
 ; argument is not a return value.
 
 ; CHECK: define void @test17(
-; CHECK-NEXT: tail call i8* @objc_retain(i8* %y) nounwind
+; CHECK-NEXT: tail call i8* @objc_retain(i8* %y) [[NUW]]
 ; CHECK-NEXT: ret void
 define void @test17(i8* %y) {
   call i8* @objc_retain(i8* %y)
@@ -265,7 +265,7 @@ define void @test17(i8* %y) {
 ; CHECK: define void @test18(
 ; CHECK-NEXT: %y = call i8* @returner()
 ; CHECK-NEXT: call void @callee()
-; CHECK-NEXT: tail call i8* @objc_retain(i8* %y) nounwind
+; CHECK-NEXT: tail call i8* @objc_retain(i8* %y) [[NUW]]
 ; CHECK-NEXT: ret void
 define void @test18() {
   %y = call i8* @returner()
@@ -323,7 +323,7 @@ define i8* @test22(i8* %p) {
 ; Convert autoreleaseRV to autorelease.
 
 ; CHECK: define void @test23(
-; CHECK: tail call i8* @objc_autorelease(i8* %p) nounwind
+; CHECK: call i8* @objc_autorelease(i8* %p) [[NUW]]
 define void @test23(i8* %p) {
   store i8 0, i8* %p
   call i8* @objc_autoreleaseReturnValue(i8* %p)
@@ -340,3 +340,5 @@ define {}* @test24(i8* %p) {
   %s = bitcast i8* %p to {}*
   ret {}* %s
 }
+
+; CHECK: attributes [[NUW]] = { nounwind }
diff --git a/test/Transforms/ObjCARC/split-backedge.ll b/test/Transforms/ObjCARC/split-backedge.ll
index 08e2dce..5ac278a 100644
--- a/test/Transforms/ObjCARC/split-backedge.ll
+++ b/test/Transforms/ObjCARC/split-backedge.ll
@@ -4,12 +4,12 @@
 ; rdar://11256239
 
 ; CHECK: define void @test0
-; CHECK: call i8* @objc_retain(i8* %call) nounwind
-; CHECK: call i8* @objc_retain(i8* %call) nounwind
-; CHECK: call i8* @objc_retain(i8* %cond) nounwind
-; CHECK: call void @objc_release(i8* %call) nounwind
-; CHECK: call void @objc_release(i8* %call) nounwind
-; CHECK: call void @objc_release(i8* %cond) nounwind
+; CHECK: call i8* @objc_retain(i8* %call) [[NUW:#[0-9]+]]
+; CHECK: call i8* @objc_retain(i8* %call) [[NUW]]
+; CHECK: call i8* @objc_retain(i8* %cond) [[NUW]]
+; CHECK: call void @objc_release(i8* %call) [[NUW]]
+; CHECK: call void @objc_release(i8* %call) [[NUW]]
+; CHECK: call void @objc_release(i8* %cond) [[NUW]]
 define void @test0() {
 entry:
   br label %while.body
@@ -46,3 +46,5 @@ declare i8* @objc_retain(i8*)
 declare void @use_pointer(i8*)
 
 !0 = metadata !{}
+
+; CHECK: attributes [[NUW]] = { nounwind }
diff --git a/test/Transforms/ObjCARC/tail-call-invariant-enforcement.ll b/test/Transforms/ObjCARC/tail-call-invariant-enforcement.ll
new file mode 100644
index 0000000..74ac97c
--- /dev/null
+++ b/test/Transforms/ObjCARC/tail-call-invariant-enforcement.ll
@@ -0,0 +1,84 @@
+; RUN: opt -objc-arc -S < %s | FileCheck %s
+
+declare i8* @objc_release(i8* %x)
+declare i8* @objc_retain(i8* %x)
+declare i8* @objc_autorelease(i8* %x)
+declare i8* @objc_autoreleaseReturnValue(i8* %x)
+declare i8* @objc_retainAutoreleasedReturnValue(i8* %x)
+
+; Never tail call objc_autorelease.
+define i8* @test0(i8* %x) {
+entry:
+  ; CHECK: %tmp0 = call i8* @objc_autorelease(i8* %x)
+  %tmp0 = call i8* @objc_autorelease(i8* %x)
+  ; CHECK: %tmp1 = call i8* @objc_autorelease(i8* %x)
+  %tmp1 = tail call i8* @objc_autorelease(i8* %x)
+
+  ret i8* %x
+}
+
+; Always tail call autoreleaseReturnValue.
+define i8* @test1(i8* %x) {
+entry:
+  ; CHECK: %tmp0 = tail call i8* @objc_autoreleaseReturnValue(i8* %x)
+  %tmp0 = call i8* @objc_autoreleaseReturnValue(i8* %x)
+  ; CHECK: %tmp1 = tail call i8* @objc_autoreleaseReturnValue(i8* %x)
+  %tmp1 = tail call i8* @objc_autoreleaseReturnValue(i8* %x)
+  ret i8* %x
+}
+
+; Always tail call objc_retain.
+define i8* @test2(i8* %x) {
+entry:
+  ; CHECK: %tmp0 = tail call i8* @objc_retain(i8* %x)
+  %tmp0 = call i8* @objc_retain(i8* %x)
+  ; CHECK: %tmp1 = tail call i8* @objc_retain(i8* %x)
+  %tmp1 = tail call i8* @objc_retain(i8* %x)
+  ret i8* %x
+}
+
+define i8* @tmp(i8* %x) {
+  ret i8* %x
+}
+
+; Always tail call objc_retainAutoreleasedReturnValue.
+define i8* @test3(i8* %x) {
+entry:
+  %y = call i8* @tmp(i8* %x)
+  ; CHECK: %tmp0 = tail call i8* @objc_retainAutoreleasedReturnValue(i8* %y)
+  %tmp0 = call i8* @objc_retainAutoreleasedReturnValue(i8* %y)
+  %z = call i8* @tmp(i8* %x)
+  ; CHECK: %tmp1 = tail call i8* @objc_retainAutoreleasedReturnValue(i8* %z)
+  %tmp1 = tail call i8* @objc_retainAutoreleasedReturnValue(i8* %z)
+  ret i8* %x
+}
+
+; By itself, we should never change whether or not objc_release is tail called.
+define i8* @test4(i8* %x) {
+entry:
+  ; CHECK: %tmp0 = call i8* @objc_release(i8* %x)
+  %tmp0 = call i8* @objc_release(i8* %x)
+  ; CHECK: %tmp1 = tail call i8* @objc_release(i8* %x)
+  %tmp1 = tail call i8* @objc_release(i8* %x)
+  ret i8* %x
+}
+
+; If we convert a tail called @objc_autoreleaseReturnValue to an
+; @objc_autorelease, ensure that the tail call is removed.
+define i8* @test5(i8* %x) {
+entry:
+  ; CHECK: %tmp0 = call i8* @objc_autorelease(i8* %x)
+  %tmp0 = tail call i8* @objc_autoreleaseReturnValue(i8* %x)
+  ret i8* %tmp0
+}
+
+; If we convert a called @objc_autorelease to an @objc_autoreleaseReturnValue,
+; ensure that the tail call is added.
+define i8* @test6(i8* %x) {
+entry:
+  ; CHECK: %tmp0 = tail call i8* @objc_retain(i8* %x)
+  %tmp0 = tail call i8* @objc_retain(i8* %x)
+  ; CHECK: %tmp1 = tail call i8* @objc_autoreleaseReturnValue(i8* %x)
+  %tmp1 = call i8* @objc_autorelease(i8* %x)
+  ret i8* %x
+}
diff --git a/test/Transforms/ObjCARC/weak-copies.ll b/test/Transforms/ObjCARC/weak-copies.ll
index e1a94bb..5dab4e0 100644
--- a/test/Transforms/ObjCARC/weak-copies.ll
+++ b/test/Transforms/ObjCARC/weak-copies.ll
@@ -19,7 +19,7 @@ target triple = "x86_64-apple-darwin11.0.0"
 ; CHECK:      define void @foo() {
 ; CHECK-NEXT: entry:
 ; CHECK-NEXT:   %call = call i8* @bar()
-; CHECK-NEXT:   call void @use(i8* %call) nounwind
+; CHECK-NEXT:   call void @use(i8* %call) [[NUW:#[0-9]+]]
 ; CHECK-NEXT:   ret void
 ; CHECK-NEXT: }
 define void @foo() {
@@ -39,7 +39,7 @@ entry:
 
 ; Eliminate unnecessary weak pointer copies in a block initialization.
 
-; CHECK:      define void @qux(i8* %me) nounwind {
+; CHECK:      define void @qux(i8* %me) #0 {
 ; CHECK-NEXT: entry:
 ; CHECK-NEXT:   %block = alloca %1, align 8
 ; CHECK-NOT:    alloca
@@ -84,4 +84,6 @@ declare i8* @objc_loadWeak(i8**)
 declare void @use(i8*) nounwind
 declare void @objc_destroyWeak(i8**)
 
+; CHECK: attributes [[NUW]] = { nounwind }
+
 !0 = metadata !{}
diff --git a/test/Transforms/ScalarRepl/2003-09-12-IncorrectPromote.ll b/test/Transforms/ScalarRepl/2003-09-12-IncorrectPromote.ll
index 0b5e415..3f28cb1 100644
--- a/test/Transforms/ScalarRepl/2003-09-12-IncorrectPromote.ll
+++ b/test/Transforms/ScalarRepl/2003-09-12-IncorrectPromote.ll
@@ -1,7 +1,6 @@
 ; Scalar replacement was incorrectly promoting this alloca!!
 ;
-; RUN: opt < %s -scalarrepl -S | \
-; RUN:   sed "s/;.*//g" | grep "\["
+; RUN: opt < %s -scalarrepl -S | FileCheck %s
 
 define i8* @test() {
 	%A = alloca [30 x i8]		; <[30 x i8]*> [#uses=1]
@@ -10,4 +9,4 @@ define i8* @test() {
 	store i8 0, i8* %B
 	ret i8* %C
 }
-
+; CHECK: alloca [
diff --git a/test/Transforms/ScalarRepl/phi-cycle.ll b/test/Transforms/ScalarRepl/phi-cycle.ll
index cb5101c..05d9382 100644
--- a/test/Transforms/ScalarRepl/phi-cycle.ll
+++ b/test/Transforms/ScalarRepl/phi-cycle.ll
@@ -67,7 +67,7 @@ while.cond.backedge.i:                            ; preds = %if.end.i, %while.bo
 
 ; CHECK: func.exit:
 ; CHECK-NOT: load
-; CHECK: %call = call i32 (i8*, ...)* @printf(i8* getelementptr inbounds ([6 x i8]* @.str, i64 0, i64 0), i32 %tmp) nounwind
+; CHECK: %call = call i32 (i8*, ...)* @printf(i8* getelementptr inbounds ([6 x i8]* @.str, i64 0, i64 0), i32 %tmp) [[NUW:#[0-9]+]]
 func.exit:                                        ; preds = %while.body.i.func.exit_crit_edge, %while.cond.i.func.exit_crit_edge
   %tmp3 = load i32* %x.i, align 4
   %call = call i32 (i8*, ...)* @printf(i8* getelementptr inbounds ([6 x i8]* @.str, i64 0, i64 0), i32 %tmp3) nounwind
@@ -75,3 +75,6 @@ func.exit:                                        ; preds = %while.body.i.func.e
 }
 
 declare i32 @printf(i8* nocapture, ...) nounwind
+
+; CHECK: attributes #0 = { nounwind uwtable }
+; CHECK: attributes [[NUW]] = { nounwind }
diff --git a/test/Transforms/ScalarRepl/volatile.ll b/test/Transforms/ScalarRepl/volatile.ll
index 056526c..d506cdf 100644
--- a/test/Transforms/ScalarRepl/volatile.ll
+++ b/test/Transforms/ScalarRepl/volatile.ll
@@ -1,12 +1,13 @@
-; RUN: opt < %s -scalarrepl -S | grep "load volatile"
-; RUN: opt < %s -scalarrepl -S | grep "store volatile"
+; RUN: opt < %s -scalarrepl -S | FileCheck %s
 
 define i32 @voltest(i32 %T) {
 	%A = alloca {i32, i32}
 	%B = getelementptr {i32,i32}* %A, i32 0, i32 0
 	store volatile i32 %T, i32* %B
+; CHECK: store volatile
 
 	%C = getelementptr {i32,i32}* %A, i32 0, i32 1
 	%X = load volatile i32* %C
+; CHECK: load volatile
 	ret i32 %X
 }
diff --git a/test/Transforms/SimplifyCFG/2002-05-05-EmptyBlockMerge.ll b/test/Transforms/SimplifyCFG/EmptyBlockMerge.ll
index feffb4e..aba08dc 100644
--- a/test/Transforms/SimplifyCFG/2002-05-05-EmptyBlockMerge.ll
+++ b/test/Transforms/SimplifyCFG/EmptyBlockMerge.ll
@@ -1,8 +1,6 @@
 ; Basic block #2 should not be merged into BB #3!
 ;
-; RUN: opt < %s -simplifycfg -S | \
-; RUN:   grep "br label"
-;
+; RUN: opt < %s -simplifycfg -S | FileCheck %s
 
 declare void @foo()
 
@@ -13,6 +11,7 @@ bb0:
 	br i1 %cond218, label %bb3, label %bb2
 bb2:		; preds = %bb0
 	call void @foo( )
+; CHECK: br label %bb3
 	br label %bb3
 bb3:		; preds = %bb2, %bb0
 	%reg117 = phi i32 [ 110, %bb2 ], [ %reg108, %bb0 ]		; <i32> [#uses=1]
diff --git a/test/Transforms/SimplifyCFG/2002-06-24-PHINode.ll b/test/Transforms/SimplifyCFG/PHINode.ll
index 88f32bc..25a242a 100644
--- a/test/Transforms/SimplifyCFG/2002-06-24-PHINode.ll
+++ b/test/Transforms/SimplifyCFG/PHINode.ll
@@ -1,10 +1,11 @@
 ; -simplifycfg is not folding blocks if there is a PHI node involved.  This 
 ; should be fixed eventually
 
-; RUN: opt < %s -simplifycfg -S | not grep br
+; RUN: opt < %s -simplifycfg -S | FileCheck %s
 
 define i32 @main(i32 %argc) {
 ; <label>:0
+; CHECK-NOT: br label %InlinedFunctionReturnNode
 	br label %InlinedFunctionReturnNode
 InlinedFunctionReturnNode:		; preds = %0
 	%X = phi i32 [ 7, %0 ]		; <i32> [#uses=1]
diff --git a/test/Transforms/SimplifyCFG/SpeculativeExec.ll b/test/Transforms/SimplifyCFG/SpeculativeExec.ll
index a61867f..dd2e5d1 100644
--- a/test/Transforms/SimplifyCFG/SpeculativeExec.ll
+++ b/test/Transforms/SimplifyCFG/SpeculativeExec.ll
@@ -44,3 +44,44 @@ join:
   ret i8 %c
 }
 
+define i8* @test4(i1* %dummy, i8* %a, i8* %b) {
+; Test that we don't speculate an arbitrarily large number of unfolded constant
+; expressions.
+; CHECK: @test4
+
+entry:
+  %cond1 = load volatile i1* %dummy
+  br i1 %cond1, label %if, label %end
+
+if:
+  %cond2 = load volatile i1* %dummy
+  br i1 %cond2, label %then, label %end
+
+then:
+  br label %end
+
+end:
+  %x1 = phi i8* [ %a, %entry ], [ %b, %if ], [ inttoptr (i64 1 to i8*), %then ]
+  %x2 = phi i8* [ %a, %entry ], [ %b, %if ], [ inttoptr (i64 2 to i8*), %then ]
+  %x3 = phi i8* [ %a, %entry ], [ %b, %if ], [ inttoptr (i64 3 to i8*), %then ]
+  %x4 = phi i8* [ %a, %entry ], [ %b, %if ], [ inttoptr (i64 4 to i8*), %then ]
+  %x5 = phi i8* [ %a, %entry ], [ %b, %if ], [ inttoptr (i64 5 to i8*), %then ]
+  %x6 = phi i8* [ %a, %entry ], [ %b, %if ], [ inttoptr (i64 6 to i8*), %then ]
+  %x7 = phi i8* [ %a, %entry ], [ %b, %if ], [ inttoptr (i64 7 to i8*), %then ]
+  %x8 = phi i8* [ %a, %entry ], [ %b, %if ], [ inttoptr (i64 8 to i8*), %then ]
+  %x9 = phi i8* [ %a, %entry ], [ %b, %if ], [ inttoptr (i64 9 to i8*), %then ]
+  %x10 = phi i8* [ %a, %entry ], [ %b, %if ], [ inttoptr (i64 10 to i8*), %then ]
+; CHECK-NOT: select
+; CHECK: phi i8*
+; CHECK: phi i8*
+; CHECK: phi i8*
+; CHECK: phi i8*
+; CHECK: phi i8*
+; CHECK: phi i8*
+; CHECK: phi i8*
+; CHECK: phi i8*
+; CHECK: phi i8*
+; CHECK: phi i8*
+
+  ret i8* %x10
+}
diff --git a/test/Transforms/SimplifyCFG/switch-on-const-select.ll b/test/Transforms/SimplifyCFG/switch-on-const-select.ll
index 673a62b..9cd709f 100644
--- a/test/Transforms/SimplifyCFG/switch-on-const-select.ll
+++ b/test/Transforms/SimplifyCFG/switch-on-const-select.ll
@@ -35,7 +35,7 @@ define i32 @bar(i64 %x, i64 %y) nounwind {
 ; CHECK: @bar
 entry:
 ; CHECK-NEXT: entry:
-; CHECK-NEXT: tail call void @bees.a() nounwind
+; CHECK-NEXT: tail call void @bees.a() [[NUW:#[0-9]+]]
 ; CHECK-NEXT: ret i32 0
     %lt = icmp slt i64 %x, %y
     %qux = select i1 %lt, i32 0, i32 2
@@ -61,7 +61,7 @@ define void @bazz(i64 %x, i64 %y) nounwind {
 ; CHECK: @bazz
 entry:
 ; CHECK-NEXT: entry:
-; CHECK-NEXT: tail call void @bees.b() nounwind
+; CHECK-NEXT: tail call void @bees.b() [[NUW]]
 ; CHECK-NEXT: ret void
     %lt = icmp slt i64 %x, %y
     %qux = select i1 %lt, i32 10, i32 12
@@ -86,7 +86,7 @@ define void @quux(i64 %x, i64 %y) nounwind {
 ; CHECK: @quux
 entry:
 ; CHECK-NEXT: entry:
-; CHECK-NEXT: tail call void @bees.a() nounwind
+; CHECK-NEXT: tail call void @bees.a() [[NUW]]
 ; CHECK-NEXT: ret void
     %lt = icmp slt i64 %x, %y
     %qux = select i1 %lt, i32 0, i32 0
@@ -136,3 +136,6 @@ bees:
 declare void @llvm.trap() nounwind noreturn
 declare void @bees.a() nounwind
 declare void @bees.b() nounwind
+
+; CHECK: attributes [[NUW]] = { nounwind }
+; CHECK: attributes #1 = { noreturn nounwind }
diff --git a/test/Transforms/SimplifyLibCalls/2009-01-04-Annotate.ll b/test/Transforms/SimplifyLibCalls/2009-01-04-Annotate.ll
index 73eb05b..16791e2 100644
--- a/test/Transforms/SimplifyLibCalls/2009-01-04-Annotate.ll
+++ b/test/Transforms/SimplifyLibCalls/2009-01-04-Annotate.ll
@@ -1,12 +1,21 @@
-; RUN: opt < %s -simplify-libcalls -S > %t
-; RUN: grep noalias %t | count 2
-; RUN: grep nocapture %t | count 3
-; RUN: grep nounwind %t | count 3
-; RUN: grep readonly %t | count 1
+; RUN: opt < %s -simplify-libcalls -S | FileCheck %s
 
+; CHECK: declare noalias i8* @fopen(i8* nocapture, i8* nocapture) #0
 declare i8* @fopen(i8*, i8*)
+
+; CHECK: declare i8 @strlen(i8* nocapture) #1
 declare i8 @strlen(i8*)
+
+; CHECK: declare noalias i32* @realloc(i32* nocapture, i32) #0
 declare i32* @realloc(i32*, i32)
 
 ; Test deliberately wrong declaration
 declare i32 @strcpy(...)
+
+; CHECK-NOT: strcpy{{.*}}noalias
+; CHECK-NOT: strcpy{{.*}}nocapture
+; CHECK-NOT: strcpy{{.*}}nounwind
+; CHECK-NOT: strcpy{{.*}}readonly
+
+; CHECK: attributes #0 = { nounwind }
+; CHECK: attributes #1 = { nounwind readonly }
diff --git a/test/Transforms/TailCallElim/dont-tce-tail-marked-call.ll b/test/Transforms/TailCallElim/dont-tce-tail-marked-call.ll
index e4f8b48..97e67b2 100644
--- a/test/Transforms/TailCallElim/dont-tce-tail-marked-call.ll
+++ b/test/Transforms/TailCallElim/dont-tce-tail-marked-call.ll
@@ -1,5 +1,4 @@
-; RUN: opt < %s -tailcallelim -S | \
-; RUN:    grep "call i32 @foo"
+; RUN: opt < %s -tailcallelim -S | FileCheck %s
 
 declare void @bar(i32*)
 
@@ -7,6 +6,7 @@ define i32 @foo(i32 %N) {
 	%A = alloca i32, i32 %N		; <i32*> [#uses=2]
 	store i32 17, i32* %A
 	call void @bar( i32* %A )
+; CHECK: tail call i32 @foo
 	%X = tail call i32 @foo( i32 %N )		; <i32> [#uses=1]
 	ret i32 %X
 }
diff --git a/test/Transforms/TailCallElim/intervening-inst.ll b/test/Transforms/TailCallElim/intervening-inst.ll
index 0c40bd5..10dffbd 100644
--- a/test/Transforms/TailCallElim/intervening-inst.ll
+++ b/test/Transforms/TailCallElim/intervening-inst.ll
@@ -1,5 +1,5 @@
 ; This function contains intervening instructions which should be moved out of the way
-; RUN: opt < %s -tailcallelim -S | not grep call
+; RUN: opt < %s -tailcallelim -S | FileCheck %s
 
 define i32 @Test(i32 %X) {
 entry:
@@ -10,6 +10,7 @@ then.0:		; preds = %entry
 	ret i32 %tmp.4
 endif.0:		; preds = %entry
 	%tmp.10 = add i32 %X, -1		; <i32> [#uses=1]
+; CHECK-NOT: call
 	%tmp.8 = call i32 @Test( i32 %tmp.10 )		; <i32> [#uses=1]
 	%DUMMY = add i32 %X, 1		; <i32> [#uses=0]
 	ret i32 %tmp.8
diff --git a/test/Transforms/TailCallElim/reorder_load.ll b/test/Transforms/TailCallElim/reorder_load.ll
index 7f5c36e..53c65da 100644
--- a/test/Transforms/TailCallElim/reorder_load.ll
+++ b/test/Transforms/TailCallElim/reorder_load.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s -tailcallelim -S | not grep call
+; RUN: opt < %s -tailcallelim -S | FileCheck %s
 ; PR4323
 
 ; Several cases where tail call elimination should move the load above the call,
@@ -21,6 +21,7 @@ if:		; preds = %entry
 
 else:		; preds = %entry
 	%tmp7 = add i32 %start_arg, 1		; <i32> [#uses=1]
+; CHECK-NOT: call
 	%tmp8 = call fastcc i32 @raise_load_1(i32* %a_arg, i32 %a_len_arg, i32 %tmp7)		; <i32> [#uses=1]
 	%tmp9 = load i32* %a_arg		; <i32> [#uses=1]
 	%tmp10 = add i32 %tmp9, %tmp8		; <i32> [#uses=1]
@@ -47,6 +48,7 @@ unwind:		; preds = %else
 
 recurse:		; preds = %else
 	%tmp7 = add i32 %start_arg, 1		; <i32> [#uses=1]
+; CHECK-NOT: call
 	%tmp8 = call fastcc i32 @raise_load_2(i32* %a_arg, i32 %a_len_arg, i32 %tmp7)		; <i32> [#uses=1]
 	%tmp9 = load i32* @global		; <i32> [#uses=1]
 	%tmp10 = add i32 %tmp9, %tmp8		; <i32> [#uses=1]
@@ -66,6 +68,7 @@ if:		; preds = %entry
 
 else:		; preds = %entry
 	%tmp7 = add i32 %start_arg, 1		; <i32> [#uses=1]
+; CHECK-NOT: call
 	%tmp8 = call fastcc i32 @raise_load_3(i32* %a_arg, i32 %a_len_arg, i32 %tmp7)		; <i32> [#uses=1]
 	%tmp9 = load i32* @extern_weak_global		; <i32> [#uses=1]
 	%tmp10 = add i32 %tmp9, %tmp8		; <i32> [#uses=1]
@@ -94,6 +97,7 @@ unwind:		; preds = %else
 recurse:		; preds = %else
 	%tmp7 = add i32 %start_arg, 1		; <i32> [#uses=1]
 	%first = load i32* %a_arg		; <i32> [#uses=1]
+; CHECK-NOT: call
 	%tmp8 = call fastcc i32 @raise_load_4(i32* %a_arg, i32 %first, i32 %tmp7)		; <i32> [#uses=1]
 	%second = load i32* %a_arg		; <i32> [#uses=1]
 	%tmp10 = add i32 %second, %tmp8		; <i32> [#uses=1]
diff --git a/test/Transforms/TailCallElim/return_constant.ll b/test/Transforms/TailCallElim/return_constant.ll
index 48e5641..e99e57e 100644
--- a/test/Transforms/TailCallElim/return_constant.ll
+++ b/test/Transforms/TailCallElim/return_constant.ll
@@ -1,7 +1,7 @@
 ; Though this case seems to be fairly unlikely to occur in the wild, someone
 ; plunked it into the demo script, so maybe they care about it.
 ;
-; RUN: opt < %s -tailcallelim -S | not grep call
+; RUN: opt < %s -tailcallelim -S | FileCheck %s
 
 define i32 @aaa(i32 %c) {
 entry:
@@ -9,6 +9,7 @@ entry:
 	br i1 %tmp.1, label %return, label %else
 else:		; preds = %entry
 	%tmp.5 = add i32 %c, -1		; <i32> [#uses=1]
+; CHECK-NOT: call
 	%tmp.3 = call i32 @aaa( i32 %tmp.5 )		; <i32> [#uses=0]
 	ret i32 0
 return:		; preds = %entry
diff --git a/test/Transforms/TailCallElim/trivial_codegen_tailcall.ll b/test/Transforms/TailCallElim/trivial_codegen_tailcall.ll
index 3d01d17..7049e4d 100644
--- a/test/Transforms/TailCallElim/trivial_codegen_tailcall.ll
+++ b/test/Transforms/TailCallElim/trivial_codegen_tailcall.ll
@@ -1,11 +1,11 @@
-; RUN: opt < %s -tailcallelim -S | \
-; RUN:    grep "tail call void @foo"
+; RUN: opt < %s -tailcallelim -S | FileCheck %s
 
 
 declare void @foo()
 
 define void @bar() {
-	call void @foo( )
+; CHECK: tail call void @foo()
+	call void @foo()
 	ret void
 }
author	Stephen Hines <srhines@google.com>	2013-03-05 23:27:24 -0800
committer	Stephen Hines <srhines@google.com>	2013-03-05 23:27:24 -0800
commit	5adb136be579e8fff3734461580cb34d1d2983b8 (patch)
tree	bff1a422e9c9789df563aaf9a7e91e63e8ec0384 /test/Transforms
parent	227a4a4ade38716ba9eb3205f48b52910f3b955e (diff)
parent	b3201c5cf1e183d840f7c99ff779d57f1549d8e5 (diff)
download	external_llvm-5adb136be579e8fff3734461580cb34d1d2983b8.zip external_llvm-5adb136be579e8fff3734461580cb34d1d2983b8.tar.gz external_llvm-5adb136be579e8fff3734461580cb34d1d2983b8.tar.bz2