aboutsummaryrefslogtreecommitdiff
path: root/test/Transforms/LoopVectorize
diff options
context:
space:
mode:
authorDimitry Andric <dim@FreeBSD.org>2015-01-15 22:30:16 +0000
committerDimitry Andric <dim@FreeBSD.org>2015-01-15 22:30:16 +0000
commit9f61947910e6ab40de38e6b4034751ef1513200f (patch)
tree3231b7529d89052b2edb92bb5ddc6a9e960e5161 /test/Transforms/LoopVectorize
parent5ca98fd98791947eba83a1ed3f2c8191ef7afa6c (diff)
Notes
Diffstat (limited to 'test/Transforms/LoopVectorize')
-rw-r--r--test/Transforms/LoopVectorize/incorrect-dom-info.ll142
-rw-r--r--test/Transforms/LoopVectorize/loop-form.ll31
-rw-r--r--test/Transforms/LoopVectorize/runtime-check-address-space.ll154
-rw-r--r--test/Transforms/LoopVectorize/runtime-check-readonly-address-space.ll120
-rw-r--r--test/Transforms/LoopVectorize/unsized-pointee-crash.ll24
-rw-r--r--test/Transforms/LoopVectorize/vect.stats.ll54
6 files changed, 346 insertions, 179 deletions
diff --git a/test/Transforms/LoopVectorize/incorrect-dom-info.ll b/test/Transforms/LoopVectorize/incorrect-dom-info.ll
new file mode 100644
index 000000000000..624ee7e50597
--- /dev/null
+++ b/test/Transforms/LoopVectorize/incorrect-dom-info.ll
@@ -0,0 +1,142 @@
+; This test is based on one of benchmarks from SPEC2006. It exposes a bug with
+; incorrect updating of the dom-tree.
+; RUN: opt < %s -loop-vectorize -verify-dom-info
+target datalayout = "e-m:o-i64:64-f80:128-n8:16:32:64-S128"
+
+@PL_utf8skip = external constant [0 x i8]
+
+; Function Attrs: nounwind ssp uwtable
+define void @Perl_pp_quotemeta() #0 {
+ %len = alloca i64, align 8
+ br i1 undef, label %2, label %1
+
+; <label>:1 ; preds = %0
+ br label %3
+
+; <label>:2 ; preds = %0
+ br label %3
+
+; <label>:3 ; preds = %2, %1
+ br i1 undef, label %34, label %4
+
+; <label>:4 ; preds = %3
+ br i1 undef, label %5, label %6
+
+; <label>:5 ; preds = %4
+ br label %6
+
+; <label>:6 ; preds = %5, %4
+ br i1 undef, label %7, label %8
+
+; <label>:7 ; preds = %6
+ br label %8
+
+; <label>:8 ; preds = %7, %6
+ br i1 undef, label %.preheader, label %9
+
+.preheader: ; preds = %9, %8
+ br i1 undef, label %.loopexit, label %.lr.ph
+
+; <label>:9 ; preds = %8
+ br i1 undef, label %thread-pre-split.preheader, label %.preheader
+
+thread-pre-split.preheader: ; preds = %9
+ br i1 undef, label %thread-pre-split._crit_edge, label %.lr.ph21
+
+.thread-pre-split.loopexit_crit_edge: ; preds = %19
+ %scevgep.sum = xor i64 %umax, -1
+ %scevgep45 = getelementptr i8* %d.020, i64 %scevgep.sum
+ br label %thread-pre-split.loopexit
+
+thread-pre-split.loopexit: ; preds = %11, %.thread-pre-split.loopexit_crit_edge
+ %d.1.lcssa = phi i8* [ %scevgep45, %.thread-pre-split.loopexit_crit_edge ], [ %d.020, %11 ]
+ br i1 false, label %thread-pre-split._crit_edge, label %.lr.ph21
+
+.lr.ph21: ; preds = %26, %thread-pre-split.loopexit, %thread-pre-split.preheader
+ %d.020 = phi i8* [ undef, %26 ], [ %d.1.lcssa, %thread-pre-split.loopexit ], [ undef, %thread-pre-split.preheader ]
+ %10 = phi i64 [ %28, %26 ], [ undef, %thread-pre-split.loopexit ], [ undef, %thread-pre-split.preheader ]
+ br i1 undef, label %11, label %22
+
+; <label>:11 ; preds = %.lr.ph21
+ %12 = getelementptr inbounds [0 x i8]* @PL_utf8skip, i64 0, i64 undef
+ %13 = load i8* %12, align 1
+ %14 = zext i8 %13 to i64
+ %15 = icmp ugt i64 %14, %10
+ %. = select i1 %15, i64 %10, i64 %14
+ br i1 undef, label %thread-pre-split.loopexit, label %.lr.ph28
+
+.lr.ph28: ; preds = %11
+ %16 = xor i64 %10, -1
+ %17 = xor i64 %14, -1
+ %18 = icmp ugt i64 %16, %17
+ %umax = select i1 %18, i64 %16, i64 %17
+ br label %19
+
+; <label>:19 ; preds = %19, %.lr.ph28
+ %ulen.126 = phi i64 [ %., %.lr.ph28 ], [ %20, %19 ]
+ %20 = add i64 %ulen.126, -1
+ %21 = icmp eq i64 %20, 0
+ br i1 %21, label %.thread-pre-split.loopexit_crit_edge, label %19
+
+; <label>:22 ; preds = %.lr.ph21
+ br i1 undef, label %26, label %23
+
+; <label>:23 ; preds = %22
+ br i1 undef, label %26, label %24
+
+; <label>:24 ; preds = %23
+ br i1 undef, label %26, label %25
+
+; <label>:25 ; preds = %24
+ br label %26
+
+; <label>:26 ; preds = %25, %24, %23, %22
+ %27 = load i64* %len, align 8
+ %28 = add i64 %27, -1
+ br i1 undef, label %thread-pre-split._crit_edge, label %.lr.ph21
+
+thread-pre-split._crit_edge: ; preds = %26, %thread-pre-split.loopexit, %thread-pre-split.preheader
+ br label %.loopexit
+
+.lr.ph: ; preds = %33, %.preheader
+ br i1 undef, label %29, label %thread-pre-split5
+
+; <label>:29 ; preds = %.lr.ph
+ br i1 undef, label %33, label %30
+
+; <label>:30 ; preds = %29
+ br i1 undef, label %33, label %31
+
+thread-pre-split5: ; preds = %.lr.ph
+ br i1 undef, label %33, label %31
+
+; <label>:31 ; preds = %thread-pre-split5, %30
+ br i1 undef, label %33, label %32
+
+; <label>:32 ; preds = %31
+ br label %33
+
+; <label>:33 ; preds = %32, %31, %thread-pre-split5, %30, %29
+ br i1 undef, label %.loopexit, label %.lr.ph
+
+.loopexit: ; preds = %33, %thread-pre-split._crit_edge, %.preheader
+ br label %35
+
+; <label>:34 ; preds = %3
+ br label %35
+
+; <label>:35 ; preds = %34, %.loopexit
+ br i1 undef, label %37, label %36
+
+; <label>:36 ; preds = %35
+ br label %37
+
+; <label>:37 ; preds = %36, %35
+ ret void
+}
+
+attributes #0 = { nounwind ssp uwtable "less-precise-fpmad"="false" "no-frame-pointer-elim"="true" "no-frame-pointer-elim-non-leaf" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "unsafe-fp-math"="false" "use-soft-float"="false" }
+
+!llvm.ident = !{!0}
+
+!0 = metadata !{metadata !"clang version 3.6.0 "}
diff --git a/test/Transforms/LoopVectorize/loop-form.ll b/test/Transforms/LoopVectorize/loop-form.ll
new file mode 100644
index 000000000000..138df1d96138
--- /dev/null
+++ b/test/Transforms/LoopVectorize/loop-form.ll
@@ -0,0 +1,31 @@
+; RUN: opt -S -loop-vectorize < %s | FileCheck %s
+target datalayout = "e-m:o-i64:64-f80:128-n8:16:32:64-S128"
+
+; Check that we vectorize only bottom-tested loops.
+; This is a reduced testcase from PR21302.
+;
+; rdar://problem/18886083
+
+%struct.X = type { i32, i16 }
+; CHECK-LABEL: @foo(
+; CHECK-NOT: vector.body
+
+define void @foo(i32 %n) {
+entry:
+ br label %for.cond
+
+for.cond:
+ %i = phi i32 [ 0, %entry ], [ %inc, %for.body ]
+ %cmp = icmp slt i32 %i, %n
+ br i1 %cmp, label %for.body, label %if.end
+
+for.body:
+ %iprom = sext i32 %i to i64
+ %b = getelementptr inbounds %struct.X* undef, i64 %iprom, i32 1
+ store i16 0, i16* %b, align 4
+ %inc = add nsw i32 %i, 1
+ br label %for.cond
+
+if.end:
+ ret void
+}
diff --git a/test/Transforms/LoopVectorize/runtime-check-address-space.ll b/test/Transforms/LoopVectorize/runtime-check-address-space.ll
index 5bf7020a475a..848ffc428c5f 100644
--- a/test/Transforms/LoopVectorize/runtime-check-address-space.ll
+++ b/test/Transforms/LoopVectorize/runtime-check-address-space.ll
@@ -31,25 +31,23 @@ define void @foo(i32 addrspace(1)* %a, i32 addrspace(1)* %b, i32 %n) #0 {
; CHECK: ret
entry:
- br label %for.cond
+ %cmp1 = icmp slt i32 0, %n
+ br i1 %cmp1, label %for.body, label %for.end
-for.cond: ; preds = %for.body, %entry
- %i.0 = phi i32 [ 0, %entry ], [ %inc, %for.body ]
- %cmp = icmp slt i32 %i.0, %n
- br i1 %cmp, label %for.body, label %for.end
-
-for.body: ; preds = %for.cond
- %idxprom = sext i32 %i.0 to i64
+for.body: ; preds = %entry, %for.body
+ %i.02 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %idxprom = sext i32 %i.02 to i64
%arrayidx = getelementptr inbounds i32 addrspace(1)* %b, i64 %idxprom
%0 = load i32 addrspace(1)* %arrayidx, align 4
%mul = mul nsw i32 %0, 3
- %idxprom1 = sext i32 %i.0 to i64
+ %idxprom1 = sext i32 %i.02 to i64
%arrayidx2 = getelementptr inbounds i32 addrspace(1)* %a, i64 %idxprom1
store i32 %mul, i32 addrspace(1)* %arrayidx2, align 4
- %inc = add nsw i32 %i.0, 1
- br label %for.cond
+ %inc = add nsw i32 %i.02, 1
+ %cmp = icmp slt i32 %inc, %n
+ br i1 %cmp, label %for.body, label %for.end
-for.end: ; preds = %for.cond
+for.end: ; preds = %for.body, %entry
ret void
}
@@ -60,25 +58,23 @@ define void @bar0(i32* %a, i32 addrspace(1)* %b, i32 %n) #0 {
; CHECK: ret
entry:
- br label %for.cond
+ %cmp1 = icmp slt i32 0, %n
+ br i1 %cmp1, label %for.body, label %for.end
-for.cond: ; preds = %for.body, %entry
- %i.0 = phi i32 [ 0, %entry ], [ %inc, %for.body ]
- %cmp = icmp slt i32 %i.0, %n
- br i1 %cmp, label %for.body, label %for.end
-
-for.body: ; preds = %for.cond
- %idxprom = sext i32 %i.0 to i64
+for.body: ; preds = %entry, %for.body
+ %i.02 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %idxprom = sext i32 %i.02 to i64
%arrayidx = getelementptr inbounds i32 addrspace(1)* %b, i64 %idxprom
%0 = load i32 addrspace(1)* %arrayidx, align 4
%mul = mul nsw i32 %0, 3
- %idxprom1 = sext i32 %i.0 to i64
+ %idxprom1 = sext i32 %i.02 to i64
%arrayidx2 = getelementptr inbounds i32* %a, i64 %idxprom1
store i32 %mul, i32* %arrayidx2, align 4
- %inc = add nsw i32 %i.0, 1
- br label %for.cond
+ %inc = add nsw i32 %i.02, 1
+ %cmp = icmp slt i32 %inc, %n
+ br i1 %cmp, label %for.body, label %for.end
-for.end: ; preds = %for.cond
+for.end: ; preds = %for.body, %entry
ret void
}
@@ -89,25 +85,23 @@ define void @bar1(i32 addrspace(1)* %a, i32* %b, i32 %n) #0 {
; CHECK: ret
entry:
- br label %for.cond
+ %cmp1 = icmp slt i32 0, %n
+ br i1 %cmp1, label %for.body, label %for.end
-for.cond: ; preds = %for.body, %entry
- %i.0 = phi i32 [ 0, %entry ], [ %inc, %for.body ]
- %cmp = icmp slt i32 %i.0, %n
- br i1 %cmp, label %for.body, label %for.end
-
-for.body: ; preds = %for.cond
- %idxprom = sext i32 %i.0 to i64
+for.body: ; preds = %entry, %for.body
+ %i.02 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %idxprom = sext i32 %i.02 to i64
%arrayidx = getelementptr inbounds i32* %b, i64 %idxprom
%0 = load i32* %arrayidx, align 4
%mul = mul nsw i32 %0, 3
- %idxprom1 = sext i32 %i.0 to i64
+ %idxprom1 = sext i32 %i.02 to i64
%arrayidx2 = getelementptr inbounds i32 addrspace(1)* %a, i64 %idxprom1
store i32 %mul, i32 addrspace(1)* %arrayidx2, align 4
- %inc = add nsw i32 %i.0, 1
- br label %for.cond
+ %inc = add nsw i32 %i.02, 1
+ %cmp = icmp slt i32 %inc, %n
+ br i1 %cmp, label %for.body, label %for.end
-for.end: ; preds = %for.cond
+for.end: ; preds = %for.body, %entry
ret void
}
@@ -119,25 +113,23 @@ define void @bar2(i32* noalias %a, i32 addrspace(1)* noalias %b, i32 %n) #0 {
; CHECK: ret
entry:
- br label %for.cond
-
-for.cond: ; preds = %for.body, %entry
- %i.0 = phi i32 [ 0, %entry ], [ %inc, %for.body ]
- %cmp = icmp slt i32 %i.0, %n
- br i1 %cmp, label %for.body, label %for.end
+ %cmp1 = icmp slt i32 0, %n
+ br i1 %cmp1, label %for.body, label %for.end
-for.body: ; preds = %for.cond
- %idxprom = sext i32 %i.0 to i64
+for.body: ; preds = %entry, %for.body
+ %i.02 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %idxprom = sext i32 %i.02 to i64
%arrayidx = getelementptr inbounds i32 addrspace(1)* %b, i64 %idxprom
%0 = load i32 addrspace(1)* %arrayidx, align 4
%mul = mul nsw i32 %0, 3
- %idxprom1 = sext i32 %i.0 to i64
+ %idxprom1 = sext i32 %i.02 to i64
%arrayidx2 = getelementptr inbounds i32* %a, i64 %idxprom1
store i32 %mul, i32* %arrayidx2, align 4
- %inc = add nsw i32 %i.0, 1
- br label %for.cond
+ %inc = add nsw i32 %i.02, 1
+ %cmp = icmp slt i32 %inc, %n
+ br i1 %cmp, label %for.body, label %for.end
-for.end: ; preds = %for.cond
+for.end: ; preds = %for.body, %entry
ret void
}
@@ -149,25 +141,23 @@ define void @arst0(i32* %b, i32 %n) #0 {
; CHECK: ret
entry:
- br label %for.cond
-
-for.cond: ; preds = %for.body, %entry
- %i.0 = phi i32 [ 0, %entry ], [ %inc, %for.body ]
- %cmp = icmp slt i32 %i.0, %n
- br i1 %cmp, label %for.body, label %for.end
+ %cmp1 = icmp slt i32 0, %n
+ br i1 %cmp1, label %for.body, label %for.end
-for.body: ; preds = %for.cond
- %idxprom = sext i32 %i.0 to i64
+for.body: ; preds = %entry, %for.body
+ %i.02 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %idxprom = sext i32 %i.02 to i64
%arrayidx = getelementptr inbounds i32* %b, i64 %idxprom
%0 = load i32* %arrayidx, align 4
%mul = mul nsw i32 %0, 3
- %idxprom1 = sext i32 %i.0 to i64
+ %idxprom1 = sext i32 %i.02 to i64
%arrayidx2 = getelementptr inbounds [1024 x i32] addrspace(1)* @g_as1, i64 0, i64 %idxprom1
store i32 %mul, i32 addrspace(1)* %arrayidx2, align 4
- %inc = add nsw i32 %i.0, 1
- br label %for.cond
+ %inc = add nsw i32 %i.02, 1
+ %cmp = icmp slt i32 %inc, %n
+ br i1 %cmp, label %for.body, label %for.end
-for.end: ; preds = %for.cond
+for.end: ; preds = %for.body, %entry
ret void
}
@@ -180,25 +170,23 @@ define void @arst1(i32* %b, i32 %n) #0 {
; CHECK: ret
entry:
- br label %for.cond
-
-for.cond: ; preds = %for.body, %entry
- %i.0 = phi i32 [ 0, %entry ], [ %inc, %for.body ]
- %cmp = icmp slt i32 %i.0, %n
- br i1 %cmp, label %for.body, label %for.end
+ %cmp1 = icmp slt i32 0, %n
+ br i1 %cmp1, label %for.body, label %for.end
-for.body: ; preds = %for.cond
- %idxprom = sext i32 %i.0 to i64
+for.body: ; preds = %entry, %for.body
+ %i.02 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %idxprom = sext i32 %i.02 to i64
%arrayidx = getelementptr inbounds [1024 x i32] addrspace(1)* @g_as1, i64 0, i64 %idxprom
%0 = load i32 addrspace(1)* %arrayidx, align 4
%mul = mul nsw i32 %0, 3
- %idxprom1 = sext i32 %i.0 to i64
+ %idxprom1 = sext i32 %i.02 to i64
%arrayidx2 = getelementptr inbounds i32* %b, i64 %idxprom1
store i32 %mul, i32* %arrayidx2, align 4
- %inc = add nsw i32 %i.0, 1
- br label %for.cond
+ %inc = add nsw i32 %i.02, 1
+ %cmp = icmp slt i32 %inc, %n
+ br i1 %cmp, label %for.body, label %for.end
-for.end: ; preds = %for.cond
+for.end: ; preds = %for.body, %entry
ret void
}
@@ -210,25 +198,23 @@ define void @aoeu(i32 %n) #0 {
; CHECK: ret
entry:
- br label %for.cond
+ %cmp1 = icmp slt i32 0, %n
+ br i1 %cmp1, label %for.body, label %for.end
-for.cond: ; preds = %for.body, %entry
- %i.0 = phi i32 [ 0, %entry ], [ %inc, %for.body ]
- %cmp = icmp slt i32 %i.0, %n
- br i1 %cmp, label %for.body, label %for.end
-
-for.body: ; preds = %for.cond
- %idxprom = sext i32 %i.0 to i64
+for.body: ; preds = %entry, %for.body
+ %i.02 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %idxprom = sext i32 %i.02 to i64
%arrayidx = getelementptr inbounds [1024 x i32] addrspace(2)* @q_as2, i64 0, i64 %idxprom
%0 = load i32 addrspace(2)* %arrayidx, align 4
%mul = mul nsw i32 %0, 3
- %idxprom1 = sext i32 %i.0 to i64
+ %idxprom1 = sext i32 %i.02 to i64
%arrayidx2 = getelementptr inbounds [1024 x i32] addrspace(1)* @g_as1, i64 0, i64 %idxprom1
store i32 %mul, i32 addrspace(1)* %arrayidx2, align 4
- %inc = add nsw i32 %i.0, 1
- br label %for.cond
+ %inc = add nsw i32 %i.02, 1
+ %cmp = icmp slt i32 %inc, %n
+ br i1 %cmp, label %for.body, label %for.end
-for.end: ; preds = %for.cond
+for.end: ; preds = %for.body, %entry
ret void
}
diff --git a/test/Transforms/LoopVectorize/runtime-check-readonly-address-space.ll b/test/Transforms/LoopVectorize/runtime-check-readonly-address-space.ll
index 212b37cceab3..4e98ec504f5b 100644
--- a/test/Transforms/LoopVectorize/runtime-check-readonly-address-space.ll
+++ b/test/Transforms/LoopVectorize/runtime-check-readonly-address-space.ll
@@ -8,26 +8,24 @@ define void @add_ints_1_1_1(i32 addrspace(1)* %a, i32 addrspace(1)* %b, i32 addr
; CHECK-LABEL: @add_ints_1_1_1(
; CHECK: <4 x i32>
; CHECK: ret
-entry:
- br label %for.cond
-for.cond: ; preds = %for.body, %entry
- %i.0 = phi i64 [ 0, %entry ], [ %inc, %for.body ]
- %cmp = icmp ult i64 %i.0, 200
- br i1 %cmp, label %for.body, label %for.end
+entry:
+ br label %for.body
-for.body: ; preds = %for.cond
- %arrayidx = getelementptr inbounds i32 addrspace(1)* %b, i64 %i.0
+for.body: ; preds = %entry, %for.body
+ %i.01 = phi i64 [ 0, %entry ], [ %inc, %for.body ]
+ %arrayidx = getelementptr inbounds i32 addrspace(1)* %b, i64 %i.01
%0 = load i32 addrspace(1)* %arrayidx, align 4
- %arrayidx1 = getelementptr inbounds i32 addrspace(1)* %c, i64 %i.0
+ %arrayidx1 = getelementptr inbounds i32 addrspace(1)* %c, i64 %i.01
%1 = load i32 addrspace(1)* %arrayidx1, align 4
%add = add nsw i32 %0, %1
- %arrayidx2 = getelementptr inbounds i32 addrspace(1)* %a, i64 %i.0
+ %arrayidx2 = getelementptr inbounds i32 addrspace(1)* %a, i64 %i.01
store i32 %add, i32 addrspace(1)* %arrayidx2, align 4
- %inc = add i64 %i.0, 1
- br label %for.cond
+ %inc = add i64 %i.01, 1
+ %cmp = icmp ult i64 %inc, 200
+ br i1 %cmp, label %for.body, label %for.end
-for.end: ; preds = %for.cond
+for.end: ; preds = %for.body
ret void
}
@@ -35,26 +33,24 @@ define void @add_ints_as_1_0_0(i32 addrspace(1)* %a, i32* %b, i32* %c) #0 {
; CHECK-LABEL: @add_ints_as_1_0_0(
; CHECK-NOT: <4 x i32>
; CHECK: ret
-entry:
- br label %for.cond
-for.cond: ; preds = %for.body, %entry
- %i.0 = phi i64 [ 0, %entry ], [ %inc, %for.body ]
- %cmp = icmp ult i64 %i.0, 200
- br i1 %cmp, label %for.body, label %for.end
+entry:
+ br label %for.body
-for.body: ; preds = %for.cond
- %arrayidx = getelementptr inbounds i32* %b, i64 %i.0
+for.body: ; preds = %entry, %for.body
+ %i.01 = phi i64 [ 0, %entry ], [ %inc, %for.body ]
+ %arrayidx = getelementptr inbounds i32* %b, i64 %i.01
%0 = load i32* %arrayidx, align 4
- %arrayidx1 = getelementptr inbounds i32* %c, i64 %i.0
+ %arrayidx1 = getelementptr inbounds i32* %c, i64 %i.01
%1 = load i32* %arrayidx1, align 4
%add = add nsw i32 %0, %1
- %arrayidx2 = getelementptr inbounds i32 addrspace(1)* %a, i64 %i.0
+ %arrayidx2 = getelementptr inbounds i32 addrspace(1)* %a, i64 %i.01
store i32 %add, i32 addrspace(1)* %arrayidx2, align 4
- %inc = add i64 %i.0, 1
- br label %for.cond
+ %inc = add i64 %i.01, 1
+ %cmp = icmp ult i64 %inc, 200
+ br i1 %cmp, label %for.body, label %for.end
-for.end: ; preds = %for.cond
+for.end: ; preds = %for.body
ret void
}
@@ -62,26 +58,24 @@ define void @add_ints_as_0_1_0(i32* %a, i32 addrspace(1)* %b, i32* %c) #0 {
; CHECK-LABEL: @add_ints_as_0_1_0(
; CHECK-NOT: <4 x i32>
; CHECK: ret
-entry:
- br label %for.cond
-for.cond: ; preds = %for.body, %entry
- %i.0 = phi i64 [ 0, %entry ], [ %inc, %for.body ]
- %cmp = icmp ult i64 %i.0, 200
- br i1 %cmp, label %for.body, label %for.end
+entry:
+ br label %for.body
-for.body: ; preds = %for.cond
- %arrayidx = getelementptr inbounds i32 addrspace(1)* %b, i64 %i.0
+for.body: ; preds = %entry, %for.body
+ %i.01 = phi i64 [ 0, %entry ], [ %inc, %for.body ]
+ %arrayidx = getelementptr inbounds i32 addrspace(1)* %b, i64 %i.01
%0 = load i32 addrspace(1)* %arrayidx, align 4
- %arrayidx1 = getelementptr inbounds i32* %c, i64 %i.0
+ %arrayidx1 = getelementptr inbounds i32* %c, i64 %i.01
%1 = load i32* %arrayidx1, align 4
%add = add nsw i32 %0, %1
- %arrayidx2 = getelementptr inbounds i32* %a, i64 %i.0
+ %arrayidx2 = getelementptr inbounds i32* %a, i64 %i.01
store i32 %add, i32* %arrayidx2, align 4
- %inc = add i64 %i.0, 1
- br label %for.cond
+ %inc = add i64 %i.01, 1
+ %cmp = icmp ult i64 %inc, 200
+ br i1 %cmp, label %for.body, label %for.end
-for.end: ; preds = %for.cond
+for.end: ; preds = %for.body
ret void
}
@@ -89,26 +83,24 @@ define void @add_ints_as_0_1_1(i32* %a, i32 addrspace(1)* %b, i32 addrspace(1)*
; CHECK-LABEL: @add_ints_as_0_1_1(
; CHECK-NOT: <4 x i32>
; CHECK: ret
-entry:
- br label %for.cond
-for.cond: ; preds = %for.body, %entry
- %i.0 = phi i64 [ 0, %entry ], [ %inc, %for.body ]
- %cmp = icmp ult i64 %i.0, 200
- br i1 %cmp, label %for.body, label %for.end
+entry:
+ br label %for.body
-for.body: ; preds = %for.cond
- %arrayidx = getelementptr inbounds i32 addrspace(1)* %b, i64 %i.0
+for.body: ; preds = %entry, %for.body
+ %i.01 = phi i64 [ 0, %entry ], [ %inc, %for.body ]
+ %arrayidx = getelementptr inbounds i32 addrspace(1)* %b, i64 %i.01
%0 = load i32 addrspace(1)* %arrayidx, align 4
- %arrayidx1 = getelementptr inbounds i32 addrspace(1)* %c, i64 %i.0
+ %arrayidx1 = getelementptr inbounds i32 addrspace(1)* %c, i64 %i.01
%1 = load i32 addrspace(1)* %arrayidx1, align 4
%add = add nsw i32 %0, %1
- %arrayidx2 = getelementptr inbounds i32* %a, i64 %i.0
+ %arrayidx2 = getelementptr inbounds i32* %a, i64 %i.01
store i32 %add, i32* %arrayidx2, align 4
- %inc = add i64 %i.0, 1
- br label %for.cond
+ %inc = add i64 %i.01, 1
+ %cmp = icmp ult i64 %inc, 200
+ br i1 %cmp, label %for.body, label %for.end
-for.end: ; preds = %for.cond
+for.end: ; preds = %for.body
ret void
}
@@ -116,26 +108,24 @@ define void @add_ints_as_0_1_2(i32* %a, i32 addrspace(1)* %b, i32 addrspace(2)*
; CHECK-LABEL: @add_ints_as_0_1_2(
; CHECK-NOT: <4 x i32>
; CHECK: ret
-entry:
- br label %for.cond
-for.cond: ; preds = %for.body, %entry
- %i.0 = phi i64 [ 0, %entry ], [ %inc, %for.body ]
- %cmp = icmp ult i64 %i.0, 200
- br i1 %cmp, label %for.body, label %for.end
+entry:
+ br label %for.body
-for.body: ; preds = %for.cond
- %arrayidx = getelementptr inbounds i32 addrspace(1)* %b, i64 %i.0
+for.body: ; preds = %entry, %for.body
+ %i.01 = phi i64 [ 0, %entry ], [ %inc, %for.body ]
+ %arrayidx = getelementptr inbounds i32 addrspace(1)* %b, i64 %i.01
%0 = load i32 addrspace(1)* %arrayidx, align 4
- %arrayidx1 = getelementptr inbounds i32 addrspace(2)* %c, i64 %i.0
+ %arrayidx1 = getelementptr inbounds i32 addrspace(2)* %c, i64 %i.01
%1 = load i32 addrspace(2)* %arrayidx1, align 4
%add = add nsw i32 %0, %1
- %arrayidx2 = getelementptr inbounds i32* %a, i64 %i.0
+ %arrayidx2 = getelementptr inbounds i32* %a, i64 %i.01
store i32 %add, i32* %arrayidx2, align 4
- %inc = add i64 %i.0, 1
- br label %for.cond
+ %inc = add i64 %i.01, 1
+ %cmp = icmp ult i64 %inc, 200
+ br i1 %cmp, label %for.body, label %for.end
-for.end: ; preds = %for.cond
+for.end: ; preds = %for.body
ret void
}
diff --git a/test/Transforms/LoopVectorize/unsized-pointee-crash.ll b/test/Transforms/LoopVectorize/unsized-pointee-crash.ll
new file mode 100644
index 000000000000..5cc98378b952
--- /dev/null
+++ b/test/Transforms/LoopVectorize/unsized-pointee-crash.ll
@@ -0,0 +1,24 @@
+; RUN: opt -S -loop-vectorize < %s | FileCheck %s
+target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
+target triple = "x86_64-unknown-linux-gnu"
+
+; CHECK-LABEL: @fn1
+define void @fn1() {
+entry:
+ br label %for.body
+
+for.body:
+ %b.05 = phi i32 (...)* [ undef, %entry ], [ %1, %for.body ]
+ %a.04 = phi i32 [ 0, %entry ], [ %inc, %for.body ]
+ %0 = bitcast i32 (...)* %b.05 to i8*
+ %add.ptr = getelementptr i8* %0, i64 1
+ %1 = bitcast i8* %add.ptr to i32 (...)*
+; CHECK: %[[cst:.*]] = bitcast i32 (...)* {{.*}} to i8*
+; CHECK-NEXT: %[[gep:.*]] = getelementptr i8* %[[cst]], i64 1
+ %inc = add nsw i32 %a.04, 1
+ %exitcond = icmp eq i32 %a.04, 63
+ br i1 %exitcond, label %for.end, label %for.body
+
+for.end:
+ ret void
+}
diff --git a/test/Transforms/LoopVectorize/vect.stats.ll b/test/Transforms/LoopVectorize/vect.stats.ll
index 92ec24f726ee..9adf2bb6cf14 100644
--- a/test/Transforms/LoopVectorize/vect.stats.ll
+++ b/test/Transforms/LoopVectorize/vect.stats.ll
@@ -13,53 +13,47 @@ target triple = "x86_64-unknown-linux-gnu"
define void @vectorized(float* nocapture %a, i64 %size) {
entry:
- %cmp1 = icmp sgt i64 %size, 0
- br i1 %cmp1, label %for.header, label %for.end
+ %cmp1 = icmp sle i64 %size, 0
+ %cmp21 = icmp sgt i64 0, %size
+ %or.cond = or i1 %cmp1, %cmp21
+ br i1 %or.cond, label %for.end, label %for.body
-for.header:
- %indvars.iv = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %entry ]
- %cmp2 = icmp sgt i64 %indvars.iv, %size
- br i1 %cmp2, label %for.end, label %for.body
-
-for.body:
-
- %arrayidx = getelementptr inbounds float* %a, i64 %indvars.iv
+for.body: ; preds = %entry, %for.body
+ %indvars.iv2 = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds float* %a, i64 %indvars.iv2
%0 = load float* %arrayidx, align 4
%mul = fmul float %0, %0
store float %mul, float* %arrayidx, align 4
+ %indvars.iv.next = add nuw nsw i64 %indvars.iv2, 1
+ %cmp2 = icmp sgt i64 %indvars.iv.next, %size
+ br i1 %cmp2, label %for.end, label %for.body
- %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
- br label %for.header
-
-for.end:
+for.end: ; preds = %entry, %for.body
ret void
}
define void @not_vectorized(float* nocapture %a, i64 %size) {
entry:
- %cmp1 = icmp sgt i64 %size, 0
- br i1 %cmp1, label %for.header, label %for.end
+ %cmp1 = icmp sle i64 %size, 0
+ %cmp21 = icmp sgt i64 0, %size
+ %or.cond = or i1 %cmp1, %cmp21
+ br i1 %or.cond, label %for.end, label %for.body
-for.header:
- %indvars.iv = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %entry ]
- %cmp2 = icmp sgt i64 %indvars.iv, %size
- br i1 %cmp2, label %for.end, label %for.body
-
-for.body:
-
- %0 = add nsw i64 %indvars.iv, -5
+for.body: ; preds = %entry, %for.body
+ %indvars.iv2 = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %entry ]
+ %0 = add nsw i64 %indvars.iv2, -5
%arrayidx = getelementptr inbounds float* %a, i64 %0
%1 = load float* %arrayidx, align 4
- %2 = add nsw i64 %indvars.iv, 2
+ %2 = add nsw i64 %indvars.iv2, 2
%arrayidx2 = getelementptr inbounds float* %a, i64 %2
%3 = load float* %arrayidx2, align 4
%mul = fmul float %1, %3
- %arrayidx4 = getelementptr inbounds float* %a, i64 %indvars.iv
+ %arrayidx4 = getelementptr inbounds float* %a, i64 %indvars.iv2
store float %mul, float* %arrayidx4, align 4
+ %indvars.iv.next = add nuw nsw i64 %indvars.iv2, 1
+ %cmp2 = icmp sgt i64 %indvars.iv.next, %size
+ br i1 %cmp2, label %for.end, label %for.body
- %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
- br label %for.header
-
-for.end:
+for.end: ; preds = %entry, %for.body
ret void
} \ No newline at end of file