summaryrefslogtreecommitdiffstats
path: root/llvm/test/CodeGen/AMDGPU/build-vector-insert-elt-infloop.ll
blob: fd81c0438d6f88aa9016f369596530abcde25027 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s

; There was an infinite loop in DAGCombiner from a target build_vector
; combine and a generic insert_vector_elt combine.

; GCN-LABEL: {{^}}combine_loop:
; GCN: flat_load_ushort
; GCN: flat_store_short
; GCN: v_lshlrev_b32_e32 v{{[0-9]+}}, 16,
define amdgpu_kernel void @combine_loop(i16* %arg) #0 {
bb:
  br label %bb1

bb1:
  %tmp = phi <2 x i16> [ <i16 15360, i16 15360>, %bb ], [ %tmp5, %bb1 ]
  %tmp2 = phi half [ 0xH0000, %bb ], [ %tmp8, %bb1 ]
  %tmp3 = load volatile half, half* null, align 536870912
  %tmp4 = bitcast half %tmp3 to i16
  %tmp5 = insertelement <2 x i16> <i16 0, i16 undef>, i16 %tmp4, i32 1
  %tmp6 = bitcast i16* %arg to half*
  store volatile half %tmp2, half* %tmp6, align 2
  %tmp7 = bitcast <2 x i16> %tmp to <2 x half>
  %tmp8 = extractelement <2 x half> %tmp7, i32 0
  br label %bb1
}

attributes #0 = { nounwind }
OpenPOWER on IntegriCloud