llvm/test/CodeGen/AMDGPU/infinite-loop.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 6
; RUN: llc -mtriple=amdgpu6.00 -simplifycfg-require-and-preserve-domtree=1 < %s | FileCheck -check-prefix=SI %s
; RUN: opt -mtriple=amdgpu6.00 -S -amdgpu-unify-divergent-exit-nodes -verify -simplifycfg-require-and-preserve-domtree=1 %s | FileCheck -check-prefix=IR %s

define amdgpu_kernel void @infinite_loop(ptr addrspace(1) %out) {
; SI-LABEL: infinite_loop:
; SI:       ; %bb.0: ; %entry
; SI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x9
; SI-NEXT:    s_mov_b32 s3, 0xf000
; SI-NEXT:    s_mov_b32 s2, -1
; SI-NEXT:    v_mov_b32_e32 v0, 0x3e7
; SI-NEXT:    s_and_b64 vcc, exec, -1
; SI-NEXT:  .LBB0_1: ; %loop
; SI-NEXT:    ; =>This Inner Loop Header: Depth=1
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
; SI-NEXT:    s_waitcnt vmcnt(0)
; SI-NEXT:    s_mov_b64 vcc, vcc
; SI-NEXT:    s_cbranch_vccnz .LBB0_1
; SI-NEXT:  ; %bb.2: ; %DummyReturnBlock
; SI-NEXT:    s_endpgm
; IR-LABEL: define amdgpu_kernel void @infinite_loop(
; IR-SAME: ptr addrspace(1) [[OUT:%.*]]) {
; IR-NEXT:  [[ENTRY:.*:]]
; IR-NEXT:    br label %[[LOOP:.*]]
; IR:       [[LOOP]]:
; IR-NEXT:    store volatile i32 999, ptr addrspace(1) [[OUT]], align 4
; IR-NEXT:    br i1 true, label %[[LOOP]], label %[[DUMMYRETURNBLOCK:.*]]
; IR:       [[DUMMYRETURNBLOCK]]:
; IR-NEXT:    ret void
;
entry:
  br label %loop

loop:
  store volatile i32 999, ptr addrspace(1) %out, align 4
  br label %loop
}

define amdgpu_kernel void @infinite_loop_callbr(ptr addrspace(1) %out) {
; SI-LABEL: infinite_loop_callbr:
; SI:       ; %bb.0: ; %entry
; SI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x9
; SI-NEXT:    ;;#ASMSTART
; SI-NEXT:    ;;#ASMEND
; SI-NEXT:    s_mov_b32 s3, 0xf000
; SI-NEXT:    s_mov_b32 s2, -1
; SI-NEXT:    v_mov_b32_e32 v0, 0x3e7
; SI-NEXT:    s_and_b64 vcc, exec, 0
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
; SI-NEXT:    s_waitcnt vmcnt(0)
; SI-NEXT:    s_mov_b64 vcc, vcc
; SI-NEXT:    s_cbranch_vccnz .LBB1_2
; SI-NEXT:  .LBB1_1: ; %TransitionBlock
; SI-NEXT:    ; =>This Inner Loop Header: Depth=1
; SI-NEXT:    ;;#ASMSTART
; SI-NEXT:    ;;#ASMEND
; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
; SI-NEXT:    s_waitcnt vmcnt(0)
; SI-NEXT:    s_mov_b64 vcc, vcc
; SI-NEXT:    s_cbranch_vccz .LBB1_1
; SI-NEXT:  .LBB1_2: ; %DummyReturnBlock
; SI-NEXT:    s_endpgm
; IR-LABEL: define amdgpu_kernel void @infinite_loop_callbr(
; IR-SAME: ptr addrspace(1) [[OUT:%.*]]) {
; IR-NEXT:  [[ENTRY:.*:]]
; IR-NEXT:    callbr void asm "", ""()
; IR-NEXT:            to label %[[LOOP:.*]] []
; IR:       [[LOOP]]:
; IR-NEXT:    store volatile i32 999, ptr addrspace(1) [[OUT]], align 4
; IR-NEXT:    br i1 true, label %[[TRANSITIONBLOCK:.*]], label %[[DUMMYRETURNBLOCK:.*]]
; IR:       [[TRANSITIONBLOCK]]:
; IR-NEXT:    callbr void asm "", ""()
; IR-NEXT:            to label %[[LOOP]] []
; IR:       [[DUMMYRETURNBLOCK]]:
; IR-NEXT:    ret void
;
entry:
  callbr void asm "", ""() to label %loop []

loop:
  store volatile i32 999, ptr addrspace(1) %out, align 4
  callbr void asm "", ""() to label %loop []
}

define amdgpu_kernel void @infinite_loop_ret(ptr addrspace(1) %out) {
; SI-LABEL: infinite_loop_ret:
; SI:       ; %bb.0: ; %entry
; SI-NEXT:    v_cmp_eq_u32_e32 vcc, 1, v0
; SI-NEXT:    s_and_saveexec_b64 s[0:1], vcc
; SI-NEXT:    s_cbranch_execz .LBB2_3
; SI-NEXT:  ; %bb.1: ; %loop.preheader
; SI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x9
; SI-NEXT:    s_mov_b32 s3, 0xf000
; SI-NEXT:    s_mov_b32 s2, -1
; SI-NEXT:    v_mov_b32_e32 v0, 0x3e7
; SI-NEXT:    s_and_b64 vcc, exec, -1
; SI-NEXT:  .LBB2_2: ; %loop
; SI-NEXT:    ; =>This Inner Loop Header: Depth=1
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
; SI-NEXT:    s_waitcnt vmcnt(0)
; SI-NEXT:    s_mov_b64 vcc, vcc
; SI-NEXT:    s_cbranch_vccnz .LBB2_2
; SI-NEXT:  .LBB2_3: ; %UnifiedReturnBlock
; SI-NEXT:    s_endpgm
; IR-LABEL: define amdgpu_kernel void @infinite_loop_ret(
; IR-SAME: ptr addrspace(1) [[OUT:%.*]]) {
; IR-NEXT:  [[ENTRY:.*:]]
; IR-NEXT:    [[TMP:%.*]] = tail call i32 @llvm.amdgcn.workitem.id.x()
; IR-NEXT:    [[COND:%.*]] = icmp eq i32 [[TMP]], 1
; IR-NEXT:    br i1 [[COND]], label %[[LOOP:.*]], label %[[UNIFIEDRETURNBLOCK:.*]]
; IR:       [[LOOP]]:
; IR-NEXT:    store volatile i32 999, ptr addrspace(1) [[OUT]], align 4
; IR-NEXT:    br i1 true, label %[[LOOP]], label %[[UNIFIEDRETURNBLOCK]]
; IR:       [[UNIFIEDRETURNBLOCK]]:
; IR-NEXT:    ret void
;
entry:
  %tmp = tail call i32 @llvm.amdgcn.workitem.id.x()
  %cond = icmp eq i32 %tmp, 1
  br i1 %cond, label %loop, label %return

loop:
  store volatile i32 999, ptr addrspace(1) %out, align 4
  br label %loop

return:
  ret void
}

define amdgpu_kernel void @infinite_loop_ret_callbr(ptr addrspace(1) %out) {
; SI-LABEL: infinite_loop_ret_callbr:
; SI:       ; %bb.0: ; %entry
; SI-NEXT:    v_cmp_eq_u32_e32 vcc, 1, v0
; SI-NEXT:    v_cndmask_b32_e64 v0, 0, 1, vcc
; SI-NEXT:    ;;#ASMSTART
; SI-NEXT:    ;;#ASMEND
; SI-NEXT:  ; %bb.1: ; %loop.preheader
; SI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x9
; SI-NEXT:    s_mov_b32 s3, 0xf000
; SI-NEXT:    s_mov_b32 s2, -1
; SI-NEXT:    v_mov_b32_e32 v0, 0x3e7
; SI-NEXT:    s_and_b64 vcc, exec, 0
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
; SI-NEXT:    s_waitcnt vmcnt(0)
; SI-NEXT:    s_mov_b64 vcc, vcc
; SI-NEXT:    s_cbranch_vccnz .LBB3_3
; SI-NEXT:  .LBB3_2: ; %TransitionBlock
; SI-NEXT:    ; =>This Inner Loop Header: Depth=1
; SI-NEXT:    ;;#ASMSTART
; SI-NEXT:    ;;#ASMEND
; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
; SI-NEXT:    s_waitcnt vmcnt(0)
; SI-NEXT:    s_mov_b64 vcc, vcc
; SI-NEXT:    s_cbranch_vccz .LBB3_2
; SI-NEXT:  .LBB3_3: ; Inline asm indirect target
; SI-NEXT:    ; %return
; SI-NEXT:    ; Label of block must be emitted
; SI-NEXT:    s_endpgm
; IR-LABEL: define amdgpu_kernel void @infinite_loop_ret_callbr(
; IR-SAME: ptr addrspace(1) [[OUT:%.*]]) {
; IR-NEXT:  [[ENTRY:.*:]]
; IR-NEXT:    [[TMP:%.*]] = tail call i32 @llvm.amdgcn.workitem.id.x()
; IR-NEXT:    [[COND:%.*]] = icmp eq i32 [[TMP]], 1
; IR-NEXT:    [[COND32:%.*]] = zext i1 [[COND]] to i32
; IR-NEXT:    callbr void asm "", "r,!i"(i32 [[COND32]])
; IR-NEXT:            to label %[[LOOP:.*]] [label %[[RETURN:.*]]]
; IR:       [[LOOP]]:
; IR-NEXT:    store volatile i32 999, ptr addrspace(1) [[OUT]], align 4
; IR-NEXT:    br i1 true, label %[[TRANSITIONBLOCK:.*]], label %[[DUMMYRETURNBLOCK:.*]]
; IR:       [[TRANSITIONBLOCK]]:
; IR-NEXT:    callbr void asm "", ""()
; IR-NEXT:            to label %[[LOOP]] []
; IR:       [[RETURN]]:
; IR-NEXT:    ret void
; IR:       [[DUMMYRETURNBLOCK]]:
; IR-NEXT:    ret void
;
entry:
  %tmp = tail call i32 @llvm.amdgcn.workitem.id.x()
  %cond = icmp eq i32 %tmp, 1
  %cond32 = zext i1 %cond to i32
  callbr void asm "", "r,!i"(i32 %cond32) to label %loop [label %return]

loop:
  store volatile i32 999, ptr addrspace(1) %out, align 4
  callbr void asm "", ""() to label %loop []

return:
  ret void
}

define amdgpu_kernel void @infinite_loops(ptr addrspace(1) %out) {
; SI-LABEL: infinite_loops:
; SI:       ; %bb.0: ; %entry
; SI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x9
; SI-NEXT:    s_mov_b32 s3, 0xf000
; SI-NEXT:    s_mov_b32 s2, -1
; SI-NEXT:    s_cbranch_scc0 .LBB4_3
; SI-NEXT:  ; %bb.1: ; %loop1.preheader
; SI-NEXT:    v_mov_b32_e32 v0, 0x3e7
; SI-NEXT:    s_and_b64 vcc, exec, -1
; SI-NEXT:  .LBB4_2: ; %loop1
; SI-NEXT:    ; =>This Inner Loop Header: Depth=1
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
; SI-NEXT:    s_waitcnt vmcnt(0)
; SI-NEXT:    s_mov_b64 vcc, vcc
; SI-NEXT:    s_cbranch_vccnz .LBB4_2
; SI-NEXT:    s_branch .LBB4_5
; SI-NEXT:  .LBB4_3:
; SI-NEXT:    v_mov_b32_e32 v0, 0x378
; SI-NEXT:    s_and_b64 vcc, exec, -1
; SI-NEXT:  .LBB4_4: ; %loop2
; SI-NEXT:    ; =>This Inner Loop Header: Depth=1
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
; SI-NEXT:    s_waitcnt vmcnt(0)
; SI-NEXT:    s_mov_b64 vcc, vcc
; SI-NEXT:    s_cbranch_vccnz .LBB4_4
; SI-NEXT:  .LBB4_5: ; %DummyReturnBlock
; SI-NEXT:    s_endpgm
; IR-LABEL: define amdgpu_kernel void @infinite_loops(
; IR-SAME: ptr addrspace(1) [[OUT:%.*]]) {
; IR-NEXT:  [[ENTRY:.*:]]
; IR-NEXT:    br i1 poison, label %[[LOOP1:.*]], label %[[LOOP2:.*]]
; IR:       [[LOOP1]]:
; IR-NEXT:    store volatile i32 999, ptr addrspace(1) [[OUT]], align 4
; IR-NEXT:    br i1 true, label %[[LOOP1]], label %[[DUMMYRETURNBLOCK:.*]]
; IR:       [[LOOP2]]:
; IR-NEXT:    store volatile i32 888, ptr addrspace(1) [[OUT]], align 4
; IR-NEXT:    br i1 true, label %[[LOOP2]], label %[[DUMMYRETURNBLOCK]]
; IR:       [[DUMMYRETURNBLOCK]]:
; IR-NEXT:    ret void
;
entry:
  br i1 poison, label %loop1, label %loop2

loop1:
  store volatile i32 999, ptr addrspace(1) %out, align 4
  br label %loop1

loop2:
  store volatile i32 888, ptr addrspace(1) %out, align 4
  br label %loop2
}

define amdgpu_kernel void @infinite_loops_callbr(ptr addrspace(1) %out) {
; SI-LABEL: infinite_loops_callbr:
; SI:       ; %bb.0: ; %entry
; SI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x9
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    ;;#ASMSTART
; SI-NEXT:    ;;#ASMEND
; SI-NEXT:  ; %bb.1: ; %loop1.preheader
; SI-NEXT:    s_mov_b32 s3, 0xf000
; SI-NEXT:    s_mov_b32 s2, -1
; SI-NEXT:    v_mov_b32_e32 v0, 0x3e7
; SI-NEXT:    s_mov_b64 vcc, 0
; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
; SI-NEXT:    s_waitcnt vmcnt(0)
; SI-NEXT:  .LBB5_2: ; %TransitionBlock1
; SI-NEXT:    ; =>This Inner Loop Header: Depth=1
; SI-NEXT:    ;;#ASMSTART
; SI-NEXT:    ;;#ASMEND
; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
; SI-NEXT:    s_waitcnt vmcnt(0)
; SI-NEXT:    s_mov_b64 vcc, vcc
; SI-NEXT:    s_cbranch_vccz .LBB5_2
; SI-NEXT:  .LBB5_3: ; %DummyReturnBlock
; SI-NEXT:    s_endpgm
; SI-NEXT:  .LBB5_4: ; Inline asm indirect target
; SI-NEXT:    ; %loop2.preheader
; SI-NEXT:    ; Label of block must be emitted
; SI-NEXT:    s_mov_b32 s3, 0xf000
; SI-NEXT:    s_mov_b32 s2, -1
; SI-NEXT:    v_mov_b32_e32 v0, 0x378
; SI-NEXT:    s_mov_b64 vcc, 0
; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
; SI-NEXT:    s_waitcnt vmcnt(0)
; SI-NEXT:  .LBB5_5: ; %TransitionBlock
; SI-NEXT:    ; =>This Inner Loop Header: Depth=1
; SI-NEXT:    ;;#ASMSTART
; SI-NEXT:    ;;#ASMEND
; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
; SI-NEXT:    s_waitcnt vmcnt(0)
; SI-NEXT:    s_mov_b64 vcc, vcc
; SI-NEXT:    s_cbranch_vccz .LBB5_5
; SI-NEXT:    s_branch .LBB5_3
; IR-LABEL: define amdgpu_kernel void @infinite_loops_callbr(
; IR-SAME: ptr addrspace(1) [[OUT:%.*]]) {
; IR-NEXT:  [[ENTRY:.*:]]
; IR-NEXT:    callbr void asm "", "r,!i"(i32 poison)
; IR-NEXT:            to label %[[LOOP1:.*]] [label %[[LOOP2:.*]]]
; IR:       [[LOOP1]]:
; IR-NEXT:    store volatile i32 999, ptr addrspace(1) [[OUT]], align 4
; IR-NEXT:    br i1 true, label %[[TRANSITIONBLOCK:.*]], label %[[DUMMYRETURNBLOCK:.*]]
; IR:       [[TRANSITIONBLOCK]]:
; IR-NEXT:    callbr void asm "", ""()
; IR-NEXT:            to label %[[LOOP1]] []
; IR:       [[LOOP2]]:
; IR-NEXT:    store volatile i32 888, ptr addrspace(1) [[OUT]], align 4
; IR-NEXT:    br i1 true, label %[[TRANSITIONBLOCK1:.*]], label %[[DUMMYRETURNBLOCK]]
; IR:       [[TRANSITIONBLOCK1]]:
; IR-NEXT:    callbr void asm "", ""()
; IR-NEXT:            to label %[[LOOP2]] []
; IR:       [[DUMMYRETURNBLOCK]]:
; IR-NEXT:    ret void
;
entry:
  callbr void asm "", "r,!i"(i32 poison) to label %loop1 [label %loop2]

loop1:
  store volatile i32 999, ptr addrspace(1) %out, align 4
  callbr void asm "", ""() to label %loop1 []

loop2:
  store volatile i32 888, ptr addrspace(1) %out, align 4
  callbr void asm "", ""() to label %loop2 []
}

define amdgpu_kernel void @infinite_loop_nest_ret(ptr addrspace(1) %out) {
; SI-LABEL: infinite_loop_nest_ret:
; SI:       ; %bb.0: ; %entry
; SI-NEXT:    v_cmp_ne_u32_e32 vcc, 1, v0
; SI-NEXT:    s_and_saveexec_b64 s[0:1], vcc
; SI-NEXT:    s_cbranch_execz .LBB6_5
; SI-NEXT:  ; %bb.1: ; %outer_loop.preheader
; SI-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0x9
; SI-NEXT:    v_cmp_ne_u32_e64 s[0:1], 3, v0
; SI-NEXT:    s_mov_b32 s7, 0xf000
; SI-NEXT:    s_mov_b32 s6, -1
; SI-NEXT:    v_mov_b32_e32 v0, 0x3e7
; SI-NEXT:  .LBB6_2: ; %outer_loop
; SI-NEXT:    ; =>This Loop Header: Depth=1
; SI-NEXT:    ; Child Loop BB6_3 Depth 2
; SI-NEXT:    s_mov_b64 s[2:3], 0
; SI-NEXT:  .LBB6_3: ; %inner_loop
; SI-NEXT:    ; Parent Loop BB6_2 Depth=1
; SI-NEXT:    ; => This Inner Loop Header: Depth=2
; SI-NEXT:    s_and_b64 s[8:9], exec, s[0:1]
; SI-NEXT:    s_or_b64 s[2:3], s[8:9], s[2:3]
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
; SI-NEXT:    s_waitcnt vmcnt(0)
; SI-NEXT:    s_andn2_b64 exec, exec, s[2:3]
; SI-NEXT:    s_cbranch_execnz .LBB6_3
; SI-NEXT:  ; %bb.4: ; %loop.exit.guard
; SI-NEXT:    ; in Loop: Header=BB6_2 Depth=1
; SI-NEXT:    s_or_b64 exec, exec, s[2:3]
; SI-NEXT:    s_branch .LBB6_2
; SI-NEXT:  .LBB6_5: ; %UnifiedReturnBlock
; SI-NEXT:    s_endpgm
; IR-LABEL: define amdgpu_kernel void @infinite_loop_nest_ret(
; IR-SAME: ptr addrspace(1) [[OUT:%.*]]) {
; IR-NEXT:  [[ENTRY:.*:]]
; IR-NEXT:    [[TMP:%.*]] = tail call i32 @llvm.amdgcn.workitem.id.x()
; IR-NEXT:    [[COND1:%.*]] = icmp ne i32 [[TMP]], 1
; IR-NEXT:    br i1 [[COND1]], label %[[OUTER_LOOP:.*]], label %[[UNIFIEDRETURNBLOCK:.*]]
; IR:       [[OUTER_LOOP]]:
; IR-NEXT:    br label %[[INNER_LOOP:.*]]
; IR:       [[INNER_LOOP]]:
; IR-NEXT:    store volatile i32 999, ptr addrspace(1) [[OUT]], align 4
; IR-NEXT:    [[COND3:%.*]] = icmp eq i32 [[TMP]], 3
; IR-NEXT:    br i1 true, label %[[TRANSITIONBLOCK:.*]], label %[[UNIFIEDRETURNBLOCK]]
; IR:       [[TRANSITIONBLOCK]]:
; IR-NEXT:    br i1 [[COND3]], label %[[INNER_LOOP]], label %[[OUTER_LOOP]]
; IR:       [[UNIFIEDRETURNBLOCK]]:
; IR-NEXT:    ret void
;
entry:
  %tmp = tail call i32 @llvm.amdgcn.workitem.id.x()
  %cond1 = icmp ne i32 %tmp, 1  ; avoid following BB optimizing away through the domination
  br i1 %cond1, label %outer_loop, label %return

outer_loop:
  ; %cond2 = icmp eq i32 %tmp, 2
  ; br i1 %cond2, label %outer_loop, label %inner_loop
  br label %inner_loop

inner_loop:                                     ; preds = %LeafBlock, %LeafBlock1
  store volatile i32 999, ptr addrspace(1) %out, align 4
  %cond3 = icmp eq i32 %tmp, 3
  br i1 %cond3, label %inner_loop, label %outer_loop

return:
  ret void
}

define amdgpu_kernel void @infinite_loop_nest_ret_callbr(ptr addrspace(1) %out) {
; SI-LABEL: infinite_loop_nest_ret_callbr:
; SI:       ; %bb.0: ; %entry
; SI-NEXT:    v_cmp_ne_u32_e32 vcc, 1, v0
; SI-NEXT:    v_cndmask_b32_e64 v1, 0, 1, vcc
; SI-NEXT:    ;;#ASMSTART
; SI-NEXT:    ;;#ASMEND
; SI-NEXT:  ; %bb.1: ; %outer_loop.preheader
; SI-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0x9
; SI-NEXT:    v_cmp_eq_u32_e32 vcc, 3, v0
; SI-NEXT:    v_cndmask_b32_e64 v0, 0, 1, vcc
; SI-NEXT:    s_mov_b32 s7, 0xf000
; SI-NEXT:    s_mov_b32 s6, -1
; SI-NEXT:    v_mov_b32_e32 v1, 0x3e7
; SI-NEXT:    s_and_b64 s[0:1], exec, 0
; SI-NEXT:  .LBB7_2: ; %outer_loop
; SI-NEXT:    ; =>This Loop Header: Depth=1
; SI-NEXT:    ; Child Loop BB7_3 Depth 2
; SI-NEXT:    ;;#ASMSTART
; SI-NEXT:    ;;#ASMEND
; SI-NEXT:  .LBB7_3: ; %inner_loop
; SI-NEXT:    ; Parent Loop BB7_2 Depth=1
; SI-NEXT:    ; => This Inner Loop Header: Depth=2
; SI-NEXT:    s_mov_b64 vcc, s[0:1]
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    buffer_store_dword v1, off, s[4:7], 0
; SI-NEXT:    s_waitcnt vmcnt(0)
; SI-NEXT:    s_mov_b64 vcc, vcc
; SI-NEXT:    s_cbranch_vccnz .LBB7_6
; SI-NEXT:  ; %bb.4: ; %TransitionBlock
; SI-NEXT:    ; in Loop: Header=BB7_3 Depth=2
; SI-NEXT:    ;;#ASMSTART
; SI-NEXT:    ;;#ASMEND
; SI-NEXT:    s_branch .LBB7_3
; SI-NEXT:  .LBB7_5: ; Inline asm indirect target
; SI-NEXT:    ; %TransitionBlock.target.outer_loop
; SI-NEXT:    ; in Loop: Header=BB7_2 Depth=1
; SI-NEXT:    ; Label of block must be emitted
; SI-NEXT:    s_mov_b64 s[2:3], 0
; SI-NEXT:    s_branch .LBB7_7
; SI-NEXT:  .LBB7_6: ; in Loop: Header=BB7_2 Depth=1
; SI-NEXT:    s_mov_b64 s[2:3], -1
; SI-NEXT:  .LBB7_7: ; %loop.exit.guard
; SI-NEXT:    ; in Loop: Header=BB7_2 Depth=1
; SI-NEXT:    s_and_b64 vcc, exec, s[2:3]
; SI-NEXT:    s_cbranch_vccz .LBB7_2
; SI-NEXT:  .LBB7_8: ; Inline asm indirect target
; SI-NEXT:    ; %return
; SI-NEXT:    ; Label of block must be emitted
; SI-NEXT:    s_endpgm
; IR-LABEL: define amdgpu_kernel void @infinite_loop_nest_ret_callbr(
; IR-SAME: ptr addrspace(1) [[OUT:%.*]]) {
; IR-NEXT:  [[ENTRY:.*:]]
; IR-NEXT:    [[TMP:%.*]] = tail call i32 @llvm.amdgcn.workitem.id.x()
; IR-NEXT:    [[COND1:%.*]] = icmp ne i32 [[TMP]], 1
; IR-NEXT:    [[COND1_32:%.*]] = zext i1 [[COND1]] to i32
; IR-NEXT:    callbr void asm "", "r,!i"(i32 [[COND1_32]])
; IR-NEXT:            to label %[[OUTER_LOOP:.*]] [label %[[RETURN:.*]]]
; IR:       [[OUTER_LOOP]]:
; IR-NEXT:    callbr void asm "", ""()
; IR-NEXT:            to label %[[INNER_LOOP:.*]] []
; IR:       [[INNER_LOOP]]:
; IR-NEXT:    store volatile i32 999, ptr addrspace(1) [[OUT]], align 4
; IR-NEXT:    [[COND3:%.*]] = icmp eq i32 [[TMP]], 3
; IR-NEXT:    [[COND3_32:%.*]] = zext i1 [[COND3]] to i32
; IR-NEXT:    br i1 true, label %[[TRANSITIONBLOCK:.*]], label %[[DUMMYRETURNBLOCK:.*]]
; IR:       [[TRANSITIONBLOCK]]:
; IR-NEXT:    callbr void asm "", "r,!i"(i32 [[COND3_32]])
; IR-NEXT:            to label %[[INNER_LOOP]] [label %[[OUTER_LOOP]]]
; IR:       [[RETURN]]:
; IR-NEXT:    ret void
; IR:       [[DUMMYRETURNBLOCK]]:
; IR-NEXT:    ret void
;
entry:
  %tmp = tail call i32 @llvm.amdgcn.workitem.id.x()
  %cond1 = icmp ne i32 %tmp, 1  ; avoid following BB optimizing away through the domination
  %cond1_32 = zext i1 %cond1 to i32
  callbr void asm "", "r,!i"(i32 %cond1_32) to label %outer_loop [label %return]

outer_loop:
  ; %cond2 = icmp eq i32 %tmp, 2
  ; br i1 %cond2, label %outer_loop, label %inner_loop
  callbr void asm "", ""() to label %inner_loop []

inner_loop:                                     ; preds = %LeafBlock, %LeafBlock1
  store volatile i32 999, ptr addrspace(1) %out, align 4
  %cond3 = icmp eq i32 %tmp, 3
  %cond3_32 = zext i1 %cond3 to i32
  callbr void asm "", "r,!i"(i32 %cond3_32) to label %inner_loop [label %outer_loop]

return:
  ret void
}

declare i32 @llvm.amdgcn.workitem.id.x()