1; RUN: llc < %s -O0 -mcpu=cortex-a8 | FileCheck %s
2target 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-v64:32:64-v128:32:128-a0:0:32-n32"
3target triple = "thumbv7-apple-darwin10"
4
5; This tests the fast register allocator's handling of partial redefines:
6;
7;      %reg1028:dsub_0, %reg1028:dsub_1 = VLD1q64 %reg1025...
8;      %reg1030:dsub_1 = COPY killed %reg1028:dsub_0
9;
10; %reg1028 gets allocated %Q0, and if %reg1030 is reloaded for the partial
11; redef, it cannot also get %Q0.
12
13; CHECK: vld1.64 {d16, d17}, [r{{.}}]
14; CHECK-NOT: vld1.64 {d16, d17}
15; CHECK: vmov.f64
16
17define i32 @test(i8* %arg) nounwind {
18entry:
19 %0 = call <2 x i64> @llvm.arm.neon.vld1.v2i64.p0i8(i8* %arg, i32 1)
20 %1 = shufflevector <2 x i64> undef, <2 x i64> %0, <2 x i32> <i32 1, i32 2>
21 store <2 x i64> %1, <2 x i64>* undef, align 16
22 ret i32 undef
23}
24
25declare <2 x i64> @llvm.arm.neon.vld1.v2i64.p0i8(i8*, i32) nounwind readonly
26