Home | History | Annotate | Download | only in LoopVectorize
      1 ; RUN: opt -S -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine < %s | FileCheck %s
      2 
      3 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"
      4 target triple = "x86_64-apple-macosx10.8.0"
      5 
      6 @uf = common global [100 x i32] zeroinitializer, align 16
      7 @xi = common global [100 x i32] zeroinitializer, align 16
      8 @q = common global [100 x i32] zeroinitializer, align 16
      9 
     10 ; PR16455
     11 
     12 
     13 ; Due to a bug in the way we handled reverse induction stores we would generate
     14 ; a shuffle too many.
     15 
     16 define void @t()  {
     17 entry:
     18   br label %for.body
     19 
     20 ; CHECK-LABEL: @t(
     21 ; CHECK: vector.body:
     22 ; CHECK: load <4 x i32>
     23 ; CHECK: [[VAR1:%[a-zA-Z0-9]+]] = shufflevector
     24 ; CHECK: load <4 x i32>
     25 ; CHECK: [[VAR2:%[a-zA-Z0-9]+]] = shufflevector
     26 ; CHECK: [[VAR3:%[a-zA-Z0-9]+]] = add nsw <4 x i32> [[VAR2]], [[VAR1]]
     27 ; CHECK: [[VAR4:%[a-zA-Z0-9]+]] = shufflevector <4 x i32> [[VAR3]]
     28 ; CHECK: store <4 x i32> [[VAR4]]
     29 ; CHECK: load <4 x i32>
     30 ; CHECK: [[VAR5:%[a-zA-Z0-9]+]] = shufflevector
     31 ; CHECK-NOT: add nsw <4 x i32> [[VAR4]], [[VAR5]]
     32 ; CHECK-NOT: add nsw <4 x i32> [[VAR5]], [[VAR4]]
     33 ; CHECK: add nsw <4 x i32> [[VAR3]], [[VAR5]]
     34 
     35 for.body:
     36   %indvars.iv = phi i64 [ 93, %entry ], [ %indvars.iv.next, %for.body ]
     37   %0 = add i64 %indvars.iv, 1
     38   %arrayidx = getelementptr inbounds [100 x i32]* @uf, i64 0, i64 %0
     39   %arrayidx3 = getelementptr inbounds [100 x i32]* @xi, i64 0, i64 %0
     40   %1 = load i32* %arrayidx3, align 4
     41   %2 = load i32* %arrayidx, align 4
     42   %add4 = add nsw i32 %2, %1
     43   store i32 %add4, i32* %arrayidx, align 4
     44   %arrayidx7 = getelementptr inbounds [100 x i32]* @q, i64 0, i64 %0
     45   %3 = load i32* %arrayidx7, align 4
     46   %add8 = add nsw i32 %add4, %3
     47   store i32 %add8, i32* %arrayidx, align 4
     48   %indvars.iv.next = add i64 %indvars.iv, -1
     49   %4 = trunc i64 %indvars.iv.next to i32
     50   %cmp = icmp ugt i32 %4, 2
     51   br i1 %cmp, label %for.body, label %for.end
     52 
     53 for.end:
     54   ret void
     55 }
     56