Updating branches/google/stable to r176857

git-svn-id: https://llvm.org/svn/llvm-project/llvm/branches/google/stable@177040 91177308-0d34-0410-b5e6-96231b3b80d8

14 files changed, 747 insertions, 10 deletions

diff --git a/test/Transforms/LoopVectorize/X86/avx1.ll b/test/Transforms/LoopVectorize/X86/avx1.ll
index a2d176a534..6c0366eae9 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 <2 x i64>
 ;CHECK: ret i32
 define i32 @read_mod_i64(i64* nocapture %a, i32 %n) nounwind uwtable ssp {
   %1 = icmp sgt i32 %n, 0
@@ -37,7 +37,7 @@ define i32 @read_mod_i64(i64* nocapture %a, i32 %n) nounwind uwtable ssp {
   %indvars.iv = phi i64 [ %indvars.iv.next, %.lr.ph ], [ 0, %0 ]
   %2 = getelementptr inbounds i64* %a, i64 %indvars.iv
   %3 = load i64* %2, align 4
-  %4 = mul i64 %3, 3
+  %4 = add i64 %3, 3
   store i64 %4, i64* %2, align 4
   %indvars.iv.next = add i64 %indvars.iv, 1
   %lftr.wideiv = trunc i64 %indvars.iv.next to i32
diff --git a/test/Transforms/LoopVectorize/X86/conversion-cost.ll b/test/Transforms/LoopVectorize/X86/conversion-cost.ll
index 8f1bb545fa..23d9233544 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 <2 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/cost-model.ll b/test/Transforms/LoopVectorize/X86/cost-model.ll
index 628f9912c8..b7f479acf9 100644
--- a/test/Transforms/LoopVectorize/X86/cost-model.ll
+++ b/test/Transforms/LoopVectorize/X86/cost-model.ll
@@ -8,8 +8,11 @@ target triple = "x86_64-apple-macosx10.8.0"
 @d = common global [2048 x i32] zeroinitializer, align 16
 @a = common global [2048 x i32] zeroinitializer, align 16
 
+; The program below gathers and scatters data. We better not vectorize it.
 ;CHECK: cost_model_1
-;CHECK: <4 x i32>
+;CHECK-NOT: <2 x i32>
+;CHECK-NOT: <4 x i32>
+;CHECK-NOT: <8 x i32>
 ;CHECK: ret void
 define void @cost_model_1() nounwind uwtable noinline ssp {
 entry:
diff --git a/test/Transforms/LoopVectorize/X86/gcc-examples.ll b/test/Transforms/LoopVectorize/X86/gcc-examples.ll
index 574c529834..d2d0eac305 100644
--- a/test/Transforms/LoopVectorize/X86/gcc-examples.ll
+++ b/test/Transforms/LoopVectorize/X86/gcc-examples.ll
@@ -1,4 +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 -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"
@@ -9,10 +10,19 @@ target triple = "x86_64-apple-macosx10.8.0"
 
 ; Select VF = 8;
 ;CHECK: @example1
-;CHECK: load <8 x i32>
-;CHECK: add nsw <8 x i32>
-;CHECK: store <8 x i32>
+;CHECK: load <4 x i32>
+;CHECK: add nsw <4 x i32>
+;CHECK: store <4 x i32>
 ;CHECK: ret void
+
+;UNROLL: @example1
+;UNROLL: load <4 x i32>
+;UNROLL: load <4 x i32>
+;UNROLL: add nsw <4 x i32>
+;UNROLL: add nsw <4 x i32>
+;UNROLL: store <4 x i32>
+;UNROLL: store <4 x i32>
+;UNROLL: ret void
 define void @example1() nounwind uwtable ssp {
   br label %1
 
@@ -34,13 +44,18 @@ define void @example1() nounwind uwtable ssp {
   ret void
 }
 
-
-; Select VF=4 because sext <8 x i1> to <8 x i32> is expensive. 
+; Select VF=4 because sext <8 x i1> to <8 x i32> is expensive.
 ;CHECK: @example10b
 ;CHECK: load <4 x i16>
 ;CHECK: sext <4 x i16>
 ;CHECK: store <4 x i32>
 ;CHECK: ret void
+;UNROLL: @example10b
+;UNROLL: load <4 x i16>
+;UNROLL: load <4 x i16>
+;UNROLL: store <4 x i32>
+;UNROLL: store <4 x i32>
+;UNROLL: 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
 
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 0000000000..186fba87d6
--- /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/no-vector.ll b/test/Transforms/LoopVectorize/X86/no-vector.ll
new file mode 100644
index 0000000000..692eec9895
--- /dev/null
+++ b/test/Transforms/LoopVectorize/X86/no-vector.ll
@@ -0,0 +1,22 @@
+; RUN: opt -S -mtriple=i386-unknown-freebsd -mcpu=i486 -loop-vectorize < %s
+
+define i32 @PR14639(i8* nocapture %s, i32 %len) nounwind {
+entry:
+  %cmp4 = icmp sgt i32 %len, 0
+  br i1 %cmp4, label %for.body, label %for.end
+
+for.body:                                         ; preds = %entry, %for.body
+  %i.06 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+  %r.05 = phi i32 [ %xor, %for.body ], [ 0, %entry ]
+  %arrayidx = getelementptr inbounds i8* %s, i32 %i.06
+  %0 = load i8* %arrayidx, align 1
+  %conv = sext i8 %0 to i32
+  %xor = xor i32 %conv, %r.05
+  %inc = add nsw i32 %i.06, 1
+  %exitcond = icmp eq i32 %inc, %len
+  br i1 %exitcond, label %for.end, label %for.body
+
+for.end:                                          ; preds = %for.body, %entry
+  %r.0.lcssa = phi i32 [ 0, %entry ], [ %xor, %for.body ]
+  ret i32 %r.0.lcssa
+}
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 0000000000..452d0df133
--- /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 0000000000..f648722734
--- /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/X86/reduction-crash.ll b/test/Transforms/LoopVectorize/X86/reduction-crash.ll
new file mode 100644
index 0000000000..f580846a02
--- /dev/null
+++ b/test/Transforms/LoopVectorize/X86/reduction-crash.ll
@@ -0,0 +1,35 @@
+; RUN: opt -S -loop-vectorize -mcpu=prescott < %s | FileCheck %s
+
+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-f80:32:32-n8:16:32-S128"
+target triple = "i386-apple-darwin"
+
+; PR15344
+define void @test1(float* nocapture %arg, i32 %arg1) nounwind {
+; CHECK: @test1
+; CHECK: preheader
+; CHECK: insertelement <2 x double> zeroinitializer, double %tmp, i32 0
+; CHECK: vector.memcheck
+
+bb:
+  br label %bb2
+
+bb2:                                              ; preds = %bb
+  %tmp = load double* null, align 8
+  br i1 undef, label %bb3, label %bb12
+
+bb3:                                              ; preds = %bb3, %bb2
+  %tmp4 = phi double [ %tmp9, %bb3 ], [ %tmp, %bb2 ]
+  %tmp5 = phi i32 [ %tmp8, %bb3 ], [ 0, %bb2 ]
+  %tmp6 = getelementptr inbounds [16 x double]* undef, i32 0, i32 %tmp5
+  %tmp7 = load double* %tmp6, align 4
+  %tmp8 = add nsw i32 %tmp5, 1
+  %tmp9 = fadd fast double %tmp4, undef
+  %tmp10 = getelementptr inbounds float* %arg, i32 %tmp5
+  store float undef, float* %tmp10, align 4
+  %tmp11 = icmp eq i32 %tmp8, %arg1
+  br i1 %tmp11, label %bb12, label %bb3
+
+bb12:                                             ; preds = %bb3, %bb2
+  %tmp13 = phi double [ %tmp, %bb2 ], [ %tmp9, %bb3 ]
+  ret void
+}
diff --git a/test/Transforms/LoopVectorize/X86/small-size.ll b/test/Transforms/LoopVectorize/X86/small-size.ll
new file mode 100644
index 0000000000..f390b33c03
--- /dev/null
+++ b/test/Transforms/LoopVectorize/X86/small-size.ll
@@ -0,0 +1,170 @@
+; 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"
+
+@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
+@G = common global [32 x [1024 x i32]] zeroinitializer, align 16
+@ub = common global [1024 x i32] zeroinitializer, align 16
+@uc = common global [1024 x i32] zeroinitializer, align 16
+@d = common global [2048 x i32] zeroinitializer, align 16
+@fa = common global [1024 x float] zeroinitializer, align 16
+@fb = common global [1024 x float] zeroinitializer, align 16
+@ic = common global [1024 x i32] zeroinitializer, align 16
+@da = common global [1024 x float] zeroinitializer, align 16
+@db = common global [1024 x float] zeroinitializer, align 16
+@dc = common global [1024 x float] zeroinitializer, align 16
+@dd = common global [1024 x float] zeroinitializer, align 16
+@dj = common global [1024 x i32] zeroinitializer, align 16
+
+; We can optimize this test without a tail.
+;CHECK: @example1
+;CHECK: load <4 x i32>
+;CHECK: add nsw <4 x i32>
+;CHECK: store <4 x i32>
+;CHECK: ret void
+define void @example1() optsize {
+  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
+}
+
+; Can't vectorize in 'optsize' mode because we need a tail.
+;CHECK: @example2
+;CHECK-NOT: store <4 x i32>
+;CHECK: ret void
+define void @example2(i32 %n, i32 %x) optsize {
+  %1 = icmp sgt i32 %n, 0
+  br i1 %1, label %.lr.ph5, label %.preheader
+
+..preheader_crit_edge:                            ; preds = %.lr.ph5
+  %phitmp = sext i32 %n to i64
+  br label %.preheader
+
+.preheader:                                       ; preds = %..preheader_crit_edge, %0
+  %i.0.lcssa = phi i64 [ %phitmp, %..preheader_crit_edge ], [ 0, %0 ]
+  %2 = icmp eq i32 %n, 0
+  br i1 %2, label %._crit_edge, label %.lr.ph
+
+.lr.ph5:                                          ; preds = %0, %.lr.ph5
+  %indvars.iv6 = phi i64 [ %indvars.iv.next7, %.lr.ph5 ], [ 0, %0 ]
+  %3 = getelementptr inbounds [2048 x i32]* @b, i64 0, i64 %indvars.iv6
+  store i32 %x, i32* %3, align 4
+  %indvars.iv.next7 = add i64 %indvars.iv6, 1
+  %lftr.wideiv = trunc i64 %indvars.iv.next7 to i32
+  %exitcond = icmp eq i32 %lftr.wideiv, %n
+  br i1 %exitcond, label %..preheader_crit_edge, label %.lr.ph5
+
+.lr.ph:                                           ; preds = %.preheader, %.lr.ph
+  %indvars.iv = phi i64 [ %indvars.iv.next, %.lr.ph ], [ %i.0.lcssa, %.preheader ]
+  %.02 = phi i32 [ %4, %.lr.ph ], [ %n, %.preheader ]
+  %4 = add nsw i32 %.02, -1
+  %5 = getelementptr inbounds [2048 x i32]* @b, i64 0, i64 %indvars.iv
+  %6 = load i32* %5, align 4
+  %7 = getelementptr inbounds [2048 x i32]* @c, i64 0, i64 %indvars.iv
+  %8 = load i32* %7, align 4
+  %9 = and i32 %8, %6
+  %10 = getelementptr inbounds [2048 x i32]* @a, i64 0, i64 %indvars.iv
+  store i32 %9, i32* %10, align 4
+  %indvars.iv.next = add i64 %indvars.iv, 1
+  %11 = icmp eq i32 %4, 0
+  br i1 %11, label %._crit_edge, label %.lr.ph
+
+._crit_edge:                                      ; preds = %.lr.ph, %.preheader
+  ret void
+}
+
+; N is unknown, we need a tail. Can't vectorize.
+;CHECK: @example3
+;CHECK-NOT: <4 x i32>
+;CHECK: ret void
+define void @example3(i32 %n, i32* noalias nocapture %p, i32* noalias nocapture %q) optsize {
+  %1 = icmp eq i32 %n, 0
+  br i1 %1, label %._crit_edge, label %.lr.ph
+
+.lr.ph:                                           ; preds = %0, %.lr.ph
+  %.05 = phi i32 [ %2, %.lr.ph ], [ %n, %0 ]
+  %.014 = phi i32* [ %5, %.lr.ph ], [ %p, %0 ]
+  %.023 = phi i32* [ %3, %.lr.ph ], [ %q, %0 ]
+  %2 = add nsw i32 %.05, -1
+  %3 = getelementptr inbounds i32* %.023, i64 1
+  %4 = load i32* %.023, align 16
+  %5 = getelementptr inbounds i32* %.014, i64 1
+  store i32 %4, i32* %.014, align 16
+  %6 = icmp eq i32 %2, 0
+  br i1 %6, label %._crit_edge, label %.lr.ph
+
+._crit_edge:                                      ; preds = %.lr.ph, %0
+  ret void
+}
+
+
+; We can't vectorize this one because we need a runtime ptr check.
+;CHECK: @example23
+;CHECK-NOT: <4 x i32>
+;CHECK: ret void
+define void @example23(i16* nocapture %src, i32* nocapture %dst) optsize {
+  br label %1
+
+; <label>:1                                       ; preds = %1, %0
+  %.04 = phi i16* [ %src, %0 ], [ %2, %1 ]
+  %.013 = phi i32* [ %dst, %0 ], [ %6, %1 ]
+  %i.02 = phi i32 [ 0, %0 ], [ %7, %1 ]
+  %2 = getelementptr inbounds i16* %.04, i64 1
+  %3 = load i16* %.04, align 2
+  %4 = zext i16 %3 to i32
+  %5 = shl nuw nsw i32 %4, 7
+  %6 = getelementptr inbounds i32* %.013, i64 1
+  store i32 %5, i32* %.013, align 4
+  %7 = add nsw i32 %i.02, 1
+  %exitcond = icmp eq i32 %7, 256
+  br i1 %exitcond, label %8, label %1
+
+; <label>:8                                       ; preds = %1
+  ret void
+}
+
+
+; We CAN vectorize this example because the pointers are marked as noalias.
+;CHECK: @example23b
+;CHECK: <4 x i32>
+;CHECK: ret void
+define void @example23b(i16* noalias nocapture %src, i32* noalias nocapture %dst) optsize {
+  br label %1
+
+; <label>:1                                       ; preds = %1, %0
+  %.04 = phi i16* [ %src, %0 ], [ %2, %1 ]
+  %.013 = phi i32* [ %dst, %0 ], [ %6, %1 ]
+  %i.02 = phi i32 [ 0, %0 ], [ %7, %1 ]
+  %2 = getelementptr inbounds i16* %.04, i64 1
+  %3 = load i16* %.04, align 2
+  %4 = zext i16 %3 to i32
+  %5 = shl nuw nsw i32 %4, 7
+  %6 = getelementptr inbounds i32* %.013, i64 1
+  store i32 %5, i32* %.013, align 4
+  %7 = add nsw i32 %i.02, 1
+  %exitcond = icmp eq i32 %7, 256
+  br i1 %exitcond, label %8, label %1
+
+; <label>:8                                       ; preds = %1
+  ret void
+}
+
+
diff --git a/test/Transforms/LoopVectorize/X86/struct-store.ll b/test/Transforms/LoopVectorize/X86/struct-store.ll
new file mode 100644
index 0000000000..a995e43a5a
--- /dev/null
+++ b/test/Transforms/LoopVectorize/X86/struct-store.ll
@@ -0,0 +1,27 @@
+; RUN: opt < %s -loop-vectorize -mtriple=x86_64-unknown-linux-gnu -S
+
+; Make sure we are not crashing on this one.
+
+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"
+
+@glbl = external global [16 x { i64, i64 }], align 16
+
+declare void @fn()
+
+define void @test() {
+entry:
+  br label %loop
+
+loop:
+  %indvars.iv = phi i64 [ %indvars.iv.next, %loop ], [ 0, %entry ]
+  %tmp = getelementptr inbounds [16 x { i64, i64 }]* @glbl, i64 0, i64 %indvars.iv
+  store { i64, i64 } { i64 ptrtoint (void ()* @fn to i64), i64 0 }, { i64, i64 }* %tmp, align 16
+  %indvars.iv.next = add i64 %indvars.iv, 1
+  %lftr.wideiv = trunc i64 %indvars.iv.next to i32
+  %exitcond = icmp ne i32 %lftr.wideiv, 16
+  br i1 %exitcond, label %loop, label %exit
+
+exit:
+  ret void
+}
diff --git a/test/Transforms/LoopVectorize/X86/unroll-small-loops.ll b/test/Transforms/LoopVectorize/X86/unroll-small-loops.ll
new file mode 100644
index 0000000000..ef63a145d0
--- /dev/null
+++ b/test/Transforms/LoopVectorize/X86/unroll-small-loops.ll
@@ -0,0 +1,50 @@
+; 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"
+;CHECK: @foo
+;CHECK: load <4 x i32>
+;CHECK-NOT: load <4 x i32>
+;CHECK: store <4 x i32>
+;CHECK-NOT: store <4 x i32>
+;CHECK: ret
+define i32 @foo(i32* nocapture %A) nounwind uwtable ssp {
+  br label %1
+
+; <label>:1                                       ; preds = %1, %0
+  %indvars.iv = phi i64 [ 0, %0 ], [ %indvars.iv.next, %1 ]
+  %2 = getelementptr inbounds i32* %A, i64 %indvars.iv
+  %3 = load i32* %2, align 4
+  %4 = add nsw i32 %3, 6
+  store i32 %4, i32* %2, 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, 100
+  br i1 %exitcond, label %5, label %1
+
+; <label>:5                                       ; preds = %1
+  ret i32 undef
+}
+
+;CHECK: @bar
+;CHECK: store <4 x i32>
+;CHECK: store <4 x i32>
+;CHECK: ret
+define i32 @bar(i32* nocapture %A, i32 %n) nounwind uwtable 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 ]
+  %2 = getelementptr inbounds i32* %A, i64 %indvars.iv
+  %3 = load i32* %2, align 4
+  %4 = add nsw i32 %3, 6
+  store i32 %4, i32* %2, 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, %n
+  br i1 %exitcond, label %._crit_edge, label %.lr.ph
+
+._crit_edge:                                      ; preds = %.lr.ph, %0
+  ret i32 undef
+}
diff --git a/test/Transforms/LoopVectorize/X86/unroll_selection.ll b/test/Transforms/LoopVectorize/X86/unroll_selection.ll
new file mode 100644
index 0000000000..2d7b663804
--- /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 0000000000..59bb8d0054
--- /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
+}
+