80 lines · plain
1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py2; RUN: llc -mtriple=amdgcn -mcpu=gfx90a -amdgpu-atomic-optimizer-strategy=None < %s | FileCheck %s3 4define protected amdgpu_kernel void @IllegalGEPConst(i32 %a, ptr addrspace(1) %b, double %c) {5; CHECK-LABEL: IllegalGEPConst:6; CHECK: ; %bb.0: ; %entry7; CHECK-NEXT: s_mov_b32 s12, SCRATCH_RSRC_DWORD08; CHECK-NEXT: s_mov_b32 s13, SCRATCH_RSRC_DWORD19; CHECK-NEXT: s_load_dword s8, s[4:5], 0x2410; CHECK-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x2c11; CHECK-NEXT: s_mov_b32 s14, -112; CHECK-NEXT: s_mov_b32 s15, 0xe0000013; CHECK-NEXT: s_add_u32 s12, s12, s1114; CHECK-NEXT: s_addc_u32 s13, s13, 015; CHECK-NEXT: s_waitcnt lgkmcnt(0)16; CHECK-NEXT: s_ashr_i32 s9, s8, 3117; CHECK-NEXT: s_lshl_b64 s[4:5], s[8:9], 318; CHECK-NEXT: s_add_u32 s0, s0, s419; CHECK-NEXT: s_addc_u32 s1, s1, s520; CHECK-NEXT: s_add_u32 s0, s0, -821; CHECK-NEXT: s_mov_b64 s[6:7], src_shared_base22; CHECK-NEXT: s_addc_u32 s1, s1, -123; CHECK-NEXT: s_cmp_eq_u32 s1, s724; CHECK-NEXT: s_cselect_b64 s[4:5], -1, 025; CHECK-NEXT: s_andn2_b64 vcc, exec, s[4:5]26; CHECK-NEXT: s_mov_b64 s[4:5], -127; CHECK-NEXT: s_cbranch_vccnz .LBB0_328; CHECK-NEXT: ; %bb.1: ; %Flow629; CHECK-NEXT: s_andn2_b64 vcc, exec, s[4:5]30; CHECK-NEXT: s_cbranch_vccz .LBB0_831; CHECK-NEXT: .LBB0_2: ; %atomicrmw.phi32; CHECK-NEXT: s_endpgm33; CHECK-NEXT: .LBB0_3: ; %atomicrmw.check.private34; CHECK-NEXT: s_mov_b64 s[4:5], src_private_base35; CHECK-NEXT: s_cmp_eq_u32 s1, s536; CHECK-NEXT: s_cselect_b64 s[4:5], -1, 037; CHECK-NEXT: s_andn2_b64 vcc, exec, s[4:5]38; CHECK-NEXT: s_mov_b64 s[4:5], -139; CHECK-NEXT: s_cbranch_vccz .LBB0_540; CHECK-NEXT: ; %bb.4: ; %atomicrmw.global41; CHECK-NEXT: v_mov_b32_e32 v0, 042; CHECK-NEXT: v_pk_mov_b32 v[2:3], s[2:3], s[2:3] op_sel:[0,1]43; CHECK-NEXT: global_atomic_add_f64 v0, v[2:3], s[0:1]44; CHECK-NEXT: s_waitcnt vmcnt(0)45; CHECK-NEXT: buffer_wbinvl1_vol46; CHECK-NEXT: s_mov_b64 s[4:5], 047; CHECK-NEXT: .LBB0_5: ; %Flow48; CHECK-NEXT: s_andn2_b64 vcc, exec, s[4:5]49; CHECK-NEXT: s_cbranch_vccnz .LBB0_750; CHECK-NEXT: ; %bb.6: ; %atomicrmw.private51; CHECK-NEXT: s_cmp_lg_u64 s[0:1], 052; CHECK-NEXT: s_cselect_b32 s4, s0, -153; CHECK-NEXT: v_mov_b32_e32 v2, s454; CHECK-NEXT: buffer_load_dword v0, v2, s[12:15], 0 offen55; CHECK-NEXT: buffer_load_dword v1, v2, s[12:15], 0 offen offset:456; CHECK-NEXT: s_waitcnt vmcnt(0)57; CHECK-NEXT: v_add_f64 v[0:1], v[0:1], s[2:3]58; CHECK-NEXT: buffer_store_dword v0, v2, s[12:15], 0 offen59; CHECK-NEXT: buffer_store_dword v1, v2, s[12:15], 0 offen offset:460; CHECK-NEXT: .LBB0_7: ; %Flow561; CHECK-NEXT: s_cbranch_execnz .LBB0_262; CHECK-NEXT: .LBB0_8: ; %atomicrmw.shared63; CHECK-NEXT: s_cmp_lg_u64 s[0:1], 064; CHECK-NEXT: s_cselect_b32 s0, s0, -165; CHECK-NEXT: v_mov_b32_e32 v0, s066; CHECK-NEXT: v_pk_mov_b32 v[2:3], s[2:3], s[2:3] op_sel:[0,1]67; CHECK-NEXT: ds_add_f64 v0, v[2:3]68; CHECK-NEXT: s_waitcnt lgkmcnt(0)69; CHECK-NEXT: s_endpgm70entry:71 %i = add nsw i32 %a, -172 %i.2 = sext i32 %i to i6473 %i.3 = getelementptr inbounds double, ptr addrspace(1) %b, i64 %i.274 %i.4 = addrspacecast ptr addrspace(1) %i.3 to ptr75 %i.5 = atomicrmw fadd ptr %i.4, double %c syncscope("agent") seq_cst, align 8, !amdgpu.no.fine.grained.memory !076 ret void77}78 79!0 = !{}80