llvm/test/CodeGen/AMDGPU/amdgpu-cs-chain-fp-nosave.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
; RUN: llc -mtriple=amdgcn -mcpu=gfx1200 < %s 2>&1 | FileCheck -check-prefix=GFX12 %s
; RUN: llc -mtriple=amdgcn -mcpu=gfx942 < %s 2>&1 | FileCheck -check-prefix=GFX942 %s

; These situations are "special" in that they either have an alloca that is not
; in the entry block or that they have a dynamic alloca. Both situations affect
; prolog/epilog generation.

declare amdgpu_gfx void @foo()

define amdgpu_cs_chain void @test_alloca() {
; GFX12-LABEL: test_alloca:
; GFX12:       ; %bb.0: ; %.entry
; GFX12-NEXT:    s_wait_loadcnt_dscnt 0x0
; GFX12-NEXT:    s_wait_expcnt 0x0
; GFX12-NEXT:    s_wait_samplecnt 0x0
; GFX12-NEXT:    s_wait_bvhcnt 0x0
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    v_mov_b32_e32 v0, 0
; GFX12-NEXT:    s_mov_b32 s32, 16
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_mov_b32 s0, s32
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_add_co_i32 s32, s0, 0x200
; GFX12-NEXT:    scratch_store_b32 off, v0, s0
; GFX12-NEXT:    s_endpgm
;
; GFX942-LABEL: test_alloca:
; GFX942:       ; %bb.0: ; %.entry
; GFX942-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX942-NEXT:    s_mov_b32 s32, 16
; GFX942-NEXT:    s_mov_b32 s0, s32
; GFX942-NEXT:    v_mov_b32_e32 v0, 0
; GFX942-NEXT:    s_add_i32 s32, s0, 0x400
; GFX942-NEXT:    scratch_store_dword off, v0, s0
; GFX942-NEXT:    s_endpgm
.entry:
  br label %SW_C

SW_C:                                             ; preds = %.entry
  %v = alloca i32, i32 1, align 4, addrspace(5)
  store i32 0, ptr addrspace(5) %v, align 4
  ret void
}

define amdgpu_cs_chain void @test_alloca_var_uniform(i32 inreg %count) {
; GFX12-LABEL: test_alloca_var_uniform:
; GFX12:       ; %bb.0:
; GFX12-NEXT:    s_wait_loadcnt_dscnt 0x0
; GFX12-NEXT:    s_wait_expcnt 0x0
; GFX12-NEXT:    s_wait_samplecnt 0x0
; GFX12-NEXT:    s_wait_bvhcnt 0x0
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    s_lshl2_add_u32 s0, s0, 15
; GFX12-NEXT:    v_mov_b32_e32 v0, 0
; GFX12-NEXT:    s_mov_b32 s32, 16
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_and_b32 s0, s0, -16
; GFX12-NEXT:    s_mov_b32 s1, s32
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_lshl_b32 s0, s0, 5
; GFX12-NEXT:    scratch_store_b32 off, v0, s1
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_add_co_i32 s32, s1, s0
; GFX12-NEXT:    s_endpgm
;
; GFX942-LABEL: test_alloca_var_uniform:
; GFX942:       ; %bb.0:
; GFX942-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX942-NEXT:    s_lshl2_add_u32 s0, s0, 15
; GFX942-NEXT:    s_mov_b32 s32, 16
; GFX942-NEXT:    s_and_b32 s0, s0, -16
; GFX942-NEXT:    v_mov_b32_e32 v0, 0
; GFX942-NEXT:    s_lshl_b32 s0, s0, 6
; GFX942-NEXT:    s_mov_b32 s1, s32
; GFX942-NEXT:    s_add_i32 s32, s1, s0
; GFX942-NEXT:    scratch_store_dword off, v0, s1
; GFX942-NEXT:    s_endpgm
  %v = alloca i32, i32 %count, align 4, addrspace(5)
  store i32 0, ptr addrspace(5) %v, align 4
  ret void
}

define amdgpu_cs_chain void @test_alloca_var(i32 %count) {
; GFX12-LABEL: test_alloca_var:
; GFX12:       ; %bb.0:
; GFX12-NEXT:    s_wait_loadcnt_dscnt 0x0
; GFX12-NEXT:    s_wait_expcnt 0x0
; GFX12-NEXT:    s_wait_samplecnt 0x0
; GFX12-NEXT:    s_wait_bvhcnt 0x0
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    v_lshl_add_u32 v0, v8, 2, 15
; GFX12-NEXT:    s_mov_b32 s1, exec_lo
; GFX12-NEXT:    s_mov_b32 s0, 0
; GFX12-NEXT:    s_mov_b32 s32, 16
; GFX12-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX12-NEXT:    v_dual_mov_b32 v0, 0 :: v_dual_and_b32 v1, -16, v0
; GFX12-NEXT:  .LBB2_1: ; =>This Inner Loop Header: Depth=1
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_ctz_i32_b32 s2, s1
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX12-NEXT:    v_readlane_b32 s3, v1, s2
; GFX12-NEXT:    s_bitset0_b32 s1, s2
; GFX12-NEXT:    s_max_u32 s0, s0, s3
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_cmp_lg_u32 s1, 0
; GFX12-NEXT:    s_cbranch_scc1 .LBB2_1
; GFX12-NEXT:  ; %bb.2:
; GFX12-NEXT:    s_mov_b32 s1, s32
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    v_lshl_add_u32 v1, s0, 5, s1
; GFX12-NEXT:    scratch_store_b32 off, v0, s1
; GFX12-NEXT:    v_readfirstlane_b32 s32, v1
; GFX12-NEXT:    s_endpgm
;
; GFX942-LABEL: test_alloca_var:
; GFX942:       ; %bb.0:
; GFX942-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX942-NEXT:    v_lshl_add_u32 v0, v8, 2, 15
; GFX942-NEXT:    v_and_b32_e32 v1, -16, v0
; GFX942-NEXT:    v_mov_b32_e32 v0, 0
; GFX942-NEXT:    s_mov_b64 s[0:1], exec
; GFX942-NEXT:    s_mov_b32 s2, 0
; GFX942-NEXT:    s_mov_b32 s32, 16
; GFX942-NEXT:  .LBB2_1: ; =>This Inner Loop Header: Depth=1
; GFX942-NEXT:    s_ff1_i32_b64 s3, s[0:1]
; GFX942-NEXT:    v_readlane_b32 s4, v1, s3
; GFX942-NEXT:    s_bitset0_b64 s[0:1], s3
; GFX942-NEXT:    s_max_u32 s2, s2, s4
; GFX942-NEXT:    s_cmp_lg_u64 s[0:1], 0
; GFX942-NEXT:    s_cbranch_scc1 .LBB2_1
; GFX942-NEXT:  ; %bb.2:
; GFX942-NEXT:    s_mov_b32 s0, s32
; GFX942-NEXT:    v_mov_b32_e32 v1, s0
; GFX942-NEXT:    v_lshl_add_u32 v1, s2, 6, v1
; GFX942-NEXT:    scratch_store_dword off, v0, s0
; GFX942-NEXT:    v_readfirstlane_b32 s32, v1
; GFX942-NEXT:    s_endpgm
  %v = alloca i32, i32 %count, align 4, addrspace(5)
  store i32 0, ptr addrspace(5) %v, align 4
  ret void
}

define amdgpu_cs_chain void @test_alloca_and_call() {
; GFX12-LABEL: test_alloca_and_call:
; GFX12:       ; %bb.0: ; %.entry
; GFX12-NEXT:    s_wait_loadcnt_dscnt 0x0
; GFX12-NEXT:    s_wait_expcnt 0x0
; GFX12-NEXT:    s_wait_samplecnt 0x0
; GFX12-NEXT:    s_wait_bvhcnt 0x0
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    s_getpc_b64 s[0:1]
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_sext_i32_i16 s1, s1
; GFX12-NEXT:    s_add_co_u32 s0, s0, foo@gotpcrel32@lo+12
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_add_co_ci_u32 s1, s1, foo@gotpcrel32@hi+24
; GFX12-NEXT:    v_mov_b32_e32 v0, 0
; GFX12-NEXT:    s_load_b64 s[0:1], s[0:1], 0x0
; GFX12-NEXT:    s_mov_b32 s32, 16
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_mov_b32 s2, s32
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_add_co_i32 s32, s2, 0x200
; GFX12-NEXT:    scratch_store_b32 off, v0, s2
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_swappc_b64 s[30:31], s[0:1]
; GFX12-NEXT:    s_endpgm
;
; GFX942-LABEL: test_alloca_and_call:
; GFX942:       ; %bb.0: ; %.entry
; GFX942-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX942-NEXT:    s_getpc_b64 s[0:1]
; GFX942-NEXT:    s_add_u32 s0, s0, foo@gotpcrel32@lo+4
; GFX942-NEXT:    s_addc_u32 s1, s1, foo@gotpcrel32@hi+12
; GFX942-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x0
; GFX942-NEXT:    s_mov_b32 s32, 16
; GFX942-NEXT:    s_mov_b32 s2, s32
; GFX942-NEXT:    v_mov_b32_e32 v0, 0
; GFX942-NEXT:    s_add_i32 s32, s2, 0x400
; GFX942-NEXT:    scratch_store_dword off, v0, s2
; GFX942-NEXT:    s_waitcnt lgkmcnt(0)
; GFX942-NEXT:    s_swappc_b64 s[30:31], s[0:1]
; GFX942-NEXT:    s_endpgm
.entry:
  br label %SW_C

SW_C:                                             ; preds = %.entry
  %v = alloca i32, i32 1, align 4, addrspace(5)
  store i32 0, ptr addrspace(5) %v, align 4
  call amdgpu_gfx void @foo()
  ret void
}

define amdgpu_cs_chain void @test_alloca_and_call_var_uniform(i32 inreg %count) {
; GFX12-LABEL: test_alloca_and_call_var_uniform:
; GFX12:       ; %bb.0:
; GFX12-NEXT:    s_wait_loadcnt_dscnt 0x0
; GFX12-NEXT:    s_wait_expcnt 0x0
; GFX12-NEXT:    s_wait_samplecnt 0x0
; GFX12-NEXT:    s_wait_bvhcnt 0x0
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    s_getpc_b64 s[2:3]
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_sext_i32_i16 s3, s3
; GFX12-NEXT:    s_add_co_u32 s2, s2, foo@gotpcrel32@lo+12
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_add_co_ci_u32 s3, s3, foo@gotpcrel32@hi+24
; GFX12-NEXT:    s_lshl2_add_u32 s0, s0, 15
; GFX12-NEXT:    s_load_b64 s[2:3], s[2:3], 0x0
; GFX12-NEXT:    v_mov_b32_e32 v0, 0
; GFX12-NEXT:    s_mov_b32 s32, 16
; GFX12-NEXT:    s_and_b32 s0, s0, -16
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_mov_b32 s1, s32
; GFX12-NEXT:    s_lshl_b32 s0, s0, 5
; GFX12-NEXT:    scratch_store_b32 off, v0, s1
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_add_co_i32 s32, s1, s0
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_swappc_b64 s[30:31], s[2:3]
; GFX12-NEXT:    s_endpgm
;
; GFX942-LABEL: test_alloca_and_call_var_uniform:
; GFX942:       ; %bb.0:
; GFX942-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX942-NEXT:    s_lshl2_add_u32 s0, s0, 15
; GFX942-NEXT:    s_and_b32 s0, s0, -16
; GFX942-NEXT:    s_lshl_b32 s2, s0, 6
; GFX942-NEXT:    s_getpc_b64 s[0:1]
; GFX942-NEXT:    s_add_u32 s0, s0, foo@gotpcrel32@lo+4
; GFX942-NEXT:    s_addc_u32 s1, s1, foo@gotpcrel32@hi+12
; GFX942-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x0
; GFX942-NEXT:    s_mov_b32 s32, 16
; GFX942-NEXT:    v_mov_b32_e32 v0, 0
; GFX942-NEXT:    s_mov_b32 s3, s32
; GFX942-NEXT:    s_add_i32 s32, s3, s2
; GFX942-NEXT:    scratch_store_dword off, v0, s3
; GFX942-NEXT:    s_waitcnt lgkmcnt(0)
; GFX942-NEXT:    s_swappc_b64 s[30:31], s[0:1]
; GFX942-NEXT:    s_endpgm
  %v = alloca i32, i32 %count, align 4, addrspace(5)
  store i32 0, ptr addrspace(5) %v, align 4
  call amdgpu_gfx void @foo()
  ret void
}

define amdgpu_cs_chain void @test_alloca_and_call_var(i32 %count) {
; GFX12-LABEL: test_alloca_and_call_var:
; GFX12:       ; %bb.0:
; GFX12-NEXT:    s_wait_loadcnt_dscnt 0x0
; GFX12-NEXT:    s_wait_expcnt 0x0
; GFX12-NEXT:    s_wait_samplecnt 0x0
; GFX12-NEXT:    s_wait_bvhcnt 0x0
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    v_lshl_add_u32 v0, v8, 2, 15
; GFX12-NEXT:    s_mov_b32 s1, exec_lo
; GFX12-NEXT:    s_mov_b32 s0, 0
; GFX12-NEXT:    s_mov_b32 s32, 16
; GFX12-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX12-NEXT:    v_dual_mov_b32 v0, 0 :: v_dual_and_b32 v1, -16, v0
; GFX12-NEXT:  .LBB5_1: ; =>This Inner Loop Header: Depth=1
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_ctz_i32_b32 s2, s1
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX12-NEXT:    v_readlane_b32 s3, v1, s2
; GFX12-NEXT:    s_bitset0_b32 s1, s2
; GFX12-NEXT:    s_max_u32 s0, s0, s3
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_cmp_lg_u32 s1, 0
; GFX12-NEXT:    s_cbranch_scc1 .LBB5_1
; GFX12-NEXT:  ; %bb.2:
; GFX12-NEXT:    s_getpc_b64 s[2:3]
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_sext_i32_i16 s3, s3
; GFX12-NEXT:    s_add_co_u32 s2, s2, foo@gotpcrel32@lo+12
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_add_co_ci_u32 s3, s3, foo@gotpcrel32@hi+24
; GFX12-NEXT:    s_mov_b32 s1, s32
; GFX12-NEXT:    s_load_b64 s[2:3], s[2:3], 0x0
; GFX12-NEXT:    v_lshl_add_u32 v1, s0, 5, s1
; GFX12-NEXT:    scratch_store_b32 off, v0, s1
; GFX12-NEXT:    v_readfirstlane_b32 s32, v1
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    s_wait_alu depctr_va_sdst(0)
; GFX12-NEXT:    s_swappc_b64 s[30:31], s[2:3]
; GFX12-NEXT:    s_endpgm
;
; GFX942-LABEL: test_alloca_and_call_var:
; GFX942:       ; %bb.0:
; GFX942-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX942-NEXT:    v_lshl_add_u32 v0, v8, 2, 15
; GFX942-NEXT:    v_and_b32_e32 v1, -16, v0
; GFX942-NEXT:    v_mov_b32_e32 v0, 0
; GFX942-NEXT:    s_mov_b64 s[0:1], exec
; GFX942-NEXT:    s_mov_b32 s2, 0
; GFX942-NEXT:    s_mov_b32 s32, 16
; GFX942-NEXT:  .LBB5_1: ; =>This Inner Loop Header: Depth=1
; GFX942-NEXT:    s_ff1_i32_b64 s3, s[0:1]
; GFX942-NEXT:    v_readlane_b32 s4, v1, s3
; GFX942-NEXT:    s_bitset0_b64 s[0:1], s3
; GFX942-NEXT:    s_max_u32 s2, s2, s4
; GFX942-NEXT:    s_cmp_lg_u64 s[0:1], 0
; GFX942-NEXT:    s_cbranch_scc1 .LBB5_1
; GFX942-NEXT:  ; %bb.2:
; GFX942-NEXT:    s_getpc_b64 s[0:1]
; GFX942-NEXT:    s_add_u32 s0, s0, foo@gotpcrel32@lo+4
; GFX942-NEXT:    s_addc_u32 s1, s1, foo@gotpcrel32@hi+12
; GFX942-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x0
; GFX942-NEXT:    s_mov_b32 s3, s32
; GFX942-NEXT:    v_mov_b32_e32 v1, s3
; GFX942-NEXT:    v_lshl_add_u32 v1, s2, 6, v1
; GFX942-NEXT:    scratch_store_dword off, v0, s3
; GFX942-NEXT:    v_readfirstlane_b32 s32, v1
; GFX942-NEXT:    s_waitcnt lgkmcnt(0)
; GFX942-NEXT:    s_swappc_b64 s[30:31], s[0:1]
; GFX942-NEXT:    s_endpgm
  %v = alloca i32, i32 %count, align 4, addrspace(5)
  store i32 0, ptr addrspace(5) %v, align 4
  call amdgpu_gfx void @foo()
  ret void
}

define amdgpu_cs_chain void @test_call_and_alloca() {
; GFX12-LABEL: test_call_and_alloca:
; GFX12:       ; %bb.0: ; %.entry
; GFX12-NEXT:    s_wait_loadcnt_dscnt 0x0
; GFX12-NEXT:    s_wait_expcnt 0x0
; GFX12-NEXT:    s_wait_samplecnt 0x0
; GFX12-NEXT:    s_wait_bvhcnt 0x0
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    s_getpc_b64 s[0:1]
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_sext_i32_i16 s1, s1
; GFX12-NEXT:    s_add_co_u32 s0, s0, foo@gotpcrel32@lo+12
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_add_co_ci_u32 s1, s1, foo@gotpcrel32@hi+24
; GFX12-NEXT:    s_mov_b32 s32, 16
; GFX12-NEXT:    s_load_b64 s[0:1], s[0:1], 0x0
; GFX12-NEXT:    s_mov_b32 s4, s32
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_add_co_i32 s32, s4, 0x200
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_swappc_b64 s[30:31], s[0:1]
; GFX12-NEXT:    v_mov_b32_e32 v0, 0
; GFX12-NEXT:    scratch_store_b32 off, v0, s4
; GFX12-NEXT:    s_endpgm
;
; GFX942-LABEL: test_call_and_alloca:
; GFX942:       ; %bb.0: ; %.entry
; GFX942-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX942-NEXT:    s_getpc_b64 s[0:1]
; GFX942-NEXT:    s_add_u32 s0, s0, foo@gotpcrel32@lo+4
; GFX942-NEXT:    s_addc_u32 s1, s1, foo@gotpcrel32@hi+12
; GFX942-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x0
; GFX942-NEXT:    s_mov_b32 s32, 16
; GFX942-NEXT:    s_mov_b32 s4, s32
; GFX942-NEXT:    s_add_i32 s32, s4, 0x400
; GFX942-NEXT:    s_waitcnt lgkmcnt(0)
; GFX942-NEXT:    s_swappc_b64 s[30:31], s[0:1]
; GFX942-NEXT:    v_mov_b32_e32 v0, 0
; GFX942-NEXT:    scratch_store_dword off, v0, s4
; GFX942-NEXT:    s_endpgm
.entry:
  br label %SW_C

SW_C:                                             ; preds = %.entry
  %v = alloca i32, i32 1, align 4, addrspace(5)
  call amdgpu_gfx void @foo()
  store i32 0, ptr addrspace(5) %v, align 4
  ret void
}

define amdgpu_cs_chain void @test_call_and_alloca_var_uniform(i32 inreg %count) {
; GFX12-LABEL: test_call_and_alloca_var_uniform:
; GFX12:       ; %bb.0:
; GFX12-NEXT:    s_wait_loadcnt_dscnt 0x0
; GFX12-NEXT:    s_wait_expcnt 0x0
; GFX12-NEXT:    s_wait_samplecnt 0x0
; GFX12-NEXT:    s_wait_bvhcnt 0x0
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    s_getpc_b64 s[2:3]
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_sext_i32_i16 s3, s3
; GFX12-NEXT:    s_add_co_u32 s2, s2, foo@gotpcrel32@lo+12
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_add_co_ci_u32 s3, s3, foo@gotpcrel32@hi+24
; GFX12-NEXT:    s_lshl2_add_u32 s0, s0, 15
; GFX12-NEXT:    s_load_b64 s[2:3], s[2:3], 0x0
; GFX12-NEXT:    s_mov_b32 s32, 16
; GFX12-NEXT:    s_and_b32 s0, s0, -16
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_mov_b32 s4, s32
; GFX12-NEXT:    s_lshl_b32 s0, s0, 5
; GFX12-NEXT:    v_mov_b32_e32 v40, 0
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_add_co_i32 s32, s4, s0
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_swappc_b64 s[30:31], s[2:3]
; GFX12-NEXT:    scratch_store_b32 off, v40, s4
; GFX12-NEXT:    s_endpgm
;
; GFX942-LABEL: test_call_and_alloca_var_uniform:
; GFX942:       ; %bb.0:
; GFX942-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX942-NEXT:    s_lshl2_add_u32 s0, s0, 15
; GFX942-NEXT:    s_and_b32 s0, s0, -16
; GFX942-NEXT:    s_lshl_b32 s2, s0, 6
; GFX942-NEXT:    s_getpc_b64 s[0:1]
; GFX942-NEXT:    s_add_u32 s0, s0, foo@gotpcrel32@lo+4
; GFX942-NEXT:    s_addc_u32 s1, s1, foo@gotpcrel32@hi+12
; GFX942-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x0
; GFX942-NEXT:    s_mov_b32 s32, 16
; GFX942-NEXT:    s_mov_b32 s4, s32
; GFX942-NEXT:    v_mov_b32_e32 v40, 0
; GFX942-NEXT:    s_add_i32 s32, s4, s2
; GFX942-NEXT:    s_waitcnt lgkmcnt(0)
; GFX942-NEXT:    s_swappc_b64 s[30:31], s[0:1]
; GFX942-NEXT:    scratch_store_dword off, v40, s4
; GFX942-NEXT:    s_endpgm
  %v = alloca i32, i32 %count, align 4, addrspace(5)
  call amdgpu_gfx void @foo()
  store i32 0, ptr addrspace(5) %v, align 4
  ret void
}

define amdgpu_cs_chain void @test_call_and_alloca_var(i32 %count) {
; GFX12-LABEL: test_call_and_alloca_var:
; GFX12:       ; %bb.0:
; GFX12-NEXT:    s_wait_loadcnt_dscnt 0x0
; GFX12-NEXT:    s_wait_expcnt 0x0
; GFX12-NEXT:    s_wait_samplecnt 0x0
; GFX12-NEXT:    s_wait_bvhcnt 0x0
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    v_lshl_add_u32 v0, v8, 2, 15
; GFX12-NEXT:    v_mov_b32_e32 v40, 0
; GFX12-NEXT:    s_mov_b32 s1, exec_lo
; GFX12-NEXT:    s_mov_b32 s0, 0
; GFX12-NEXT:    s_mov_b32 s32, 16
; GFX12-NEXT:    v_and_b32_e32 v0, -16, v0
; GFX12-NEXT:  .LBB8_1: ; =>This Inner Loop Header: Depth=1
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_ctz_i32_b32 s2, s1
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX12-NEXT:    v_readlane_b32 s3, v0, s2
; GFX12-NEXT:    s_bitset0_b32 s1, s2
; GFX12-NEXT:    s_max_u32 s0, s0, s3
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_cmp_lg_u32 s1, 0
; GFX12-NEXT:    s_cbranch_scc1 .LBB8_1
; GFX12-NEXT:  ; %bb.2:
; GFX12-NEXT:    s_getpc_b64 s[2:3]
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_sext_i32_i16 s3, s3
; GFX12-NEXT:    s_add_co_u32 s2, s2, foo@gotpcrel32@lo+12
; GFX12-NEXT:    s_wait_alu depctr_sa_sdst(0)
; GFX12-NEXT:    s_add_co_ci_u32 s3, s3, foo@gotpcrel32@hi+24
; GFX12-NEXT:    s_mov_b32 s4, s32
; GFX12-NEXT:    s_load_b64 s[2:3], s[2:3], 0x0
; GFX12-NEXT:    v_lshl_add_u32 v0, s0, 5, s4
; GFX12-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX12-NEXT:    v_readfirstlane_b32 s32, v0
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    s_wait_alu depctr_va_sdst(0)
; GFX12-NEXT:    s_swappc_b64 s[30:31], s[2:3]
; GFX12-NEXT:    scratch_store_b32 off, v40, s4
; GFX12-NEXT:    s_endpgm
;
; GFX942-LABEL: test_call_and_alloca_var:
; GFX942:       ; %bb.0:
; GFX942-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX942-NEXT:    v_lshl_add_u32 v0, v8, 2, 15
; GFX942-NEXT:    v_and_b32_e32 v0, -16, v0
; GFX942-NEXT:    v_mov_b32_e32 v40, 0
; GFX942-NEXT:    s_mov_b64 s[0:1], exec
; GFX942-NEXT:    s_mov_b32 s2, 0
; GFX942-NEXT:    s_mov_b32 s32, 16
; GFX942-NEXT:  .LBB8_1: ; =>This Inner Loop Header: Depth=1
; GFX942-NEXT:    s_ff1_i32_b64 s3, s[0:1]
; GFX942-NEXT:    v_readlane_b32 s4, v0, s3
; GFX942-NEXT:    s_bitset0_b64 s[0:1], s3
; GFX942-NEXT:    s_max_u32 s2, s2, s4
; GFX942-NEXT:    s_cmp_lg_u64 s[0:1], 0
; GFX942-NEXT:    s_cbranch_scc1 .LBB8_1
; GFX942-NEXT:  ; %bb.2:
; GFX942-NEXT:    s_getpc_b64 s[0:1]
; GFX942-NEXT:    s_add_u32 s0, s0, foo@gotpcrel32@lo+4
; GFX942-NEXT:    s_addc_u32 s1, s1, foo@gotpcrel32@hi+12
; GFX942-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x0
; GFX942-NEXT:    s_mov_b32 s4, s32
; GFX942-NEXT:    v_mov_b32_e32 v0, s4
; GFX942-NEXT:    v_lshl_add_u32 v0, s2, 6, v0
; GFX942-NEXT:    s_nop 0
; GFX942-NEXT:    v_readfirstlane_b32 s32, v0
; GFX942-NEXT:    s_waitcnt lgkmcnt(0)
; GFX942-NEXT:    s_swappc_b64 s[30:31], s[0:1]
; GFX942-NEXT:    scratch_store_dword off, v40, s4
; GFX942-NEXT:    s_endpgm
  %v = alloca i32, i32 %count, align 4, addrspace(5)
  call amdgpu_gfx void @foo()
  store i32 0, ptr addrspace(5) %v, align 4
  ret void
}