1 ; RUN: llc < %s -mtriple=thumbv7-apple-darwin -mcpu=cortex-a8
2 target datalayout = "e-p:32:32:32-i1:8:32-i8:8:32-i16:16:32-i32:32:32-i64:32:32-f32:32:32-f64:32:32-v64:64:64-v128:128:128-a0:0:32-n32"
3 target triple = "thumbv7-apple-darwin10"
5 ; This function would crash LiveIntervalAnalysis by creating a chain of 4 INSERT_SUBREGs of the same register.
6 define arm_apcscc void @NEON_vst4q_u32(i32* nocapture %sp0, i32* nocapture %sp1, i32* nocapture %sp2, i32* nocapture %sp3, i32* %dp) nounwind {
8 %0 = bitcast i32* %sp0 to <4 x i32>* ; <<4 x i32>*> [#uses=1]
9 %1 = load <4 x i32>* %0, align 16 ; <<4 x i32>> [#uses=1]
10 %2 = bitcast i32* %sp1 to <4 x i32>* ; <<4 x i32>*> [#uses=1]
11 %3 = load <4 x i32>* %2, align 16 ; <<4 x i32>> [#uses=1]
12 %4 = bitcast i32* %sp2 to <4 x i32>* ; <<4 x i32>*> [#uses=1]
13 %5 = load <4 x i32>* %4, align 16 ; <<4 x i32>> [#uses=1]
14 %6 = bitcast i32* %sp3 to <4 x i32>* ; <<4 x i32>*> [#uses=1]
15 %7 = load <4 x i32>* %6, align 16 ; <<4 x i32>> [#uses=1]
16 %8 = bitcast i32* %dp to i8* ; <i8*> [#uses=1]
17 tail call void @llvm.arm.neon.vst4.v4i32(i8* %8, <4 x i32> %1, <4 x i32> %3, <4 x i32> %5, <4 x i32> %7)
21 declare void @llvm.arm.neon.vst4.v4i32(i8*, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>) nounwind
23 @sbuf = common global [16 x i32] zeroinitializer, align 16 ; <[16 x i32]*> [#uses=5]
24 @dbuf = common global [16 x i32] zeroinitializer ; <[16 x i32]*> [#uses=2]
26 ; This function creates 4 chained INSERT_SUBREGS and then invokes the register scavenger.
27 ; The first INSERT_SUBREG needs an <undef> use operand for that to work.
28 define arm_apcscc i32 @main() nounwind {
32 bb: ; preds = %bb, %bb.nph
33 %0 = phi i32 [ 0, %bb.nph ], [ %1, %bb ] ; <i32> [#uses=4]
34 %scevgep = getelementptr [16 x i32]* @sbuf, i32 0, i32 %0 ; <i32*> [#uses=1]
35 %scevgep5 = getelementptr [16 x i32]* @dbuf, i32 0, i32 %0 ; <i32*> [#uses=1]
36 store i32 %0, i32* %scevgep, align 4
37 store i32 -1, i32* %scevgep5, align 4
38 %1 = add nsw i32 %0, 1 ; <i32> [#uses=2]
39 %exitcond = icmp eq i32 %1, 16 ; <i1> [#uses=1]
40 br i1 %exitcond, label %bb2, label %bb
43 %2 = load <4 x i32>* bitcast ([16 x i32]* @sbuf to <4 x i32>*), align 16 ; <<4 x i32>> [#uses=1]
44 %3 = load <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32]* @sbuf, i32 0, i32 4) to <4 x i32>*), align 16 ; <<4 x i32>> [#uses=1]
45 %4 = load <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32]* @sbuf, i32 0, i32 8) to <4 x i32>*), align 16 ; <<4 x i32>> [#uses=1]
46 %5 = load <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32]* @sbuf, i32 0, i32 12) to <4 x i32>*), align 16 ; <<4 x i32>> [#uses=1]
47 tail call void @llvm.arm.neon.vst4.v4i32(i8* bitcast ([16 x i32]* @dbuf to i8*), <4 x i32> %2, <4 x i32> %3, <4 x i32> %4, <4 x i32> %5) nounwind