// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --include-generated-funcs --version 5 // REQUIRES: amdgpu-registered-target // RUN: %clang_cc1 -triple amdgcn-amd-amdhsa -target-cpu gfx1250 -emit-llvm -fcuda-is-device -o - %s | FileCheck %s #define __device__ __attribute__((device)) typedef int v4i __attribute__((ext_vector_type(4))); template __device__ v4i templated_amdgcn_flat_load_monitor_b128(v4i* inptr) { return __builtin_amdgcn_flat_load_monitor_b128(inptr, AO, Scope); } __device__ void test_amdgcn_flat_load_monitor_b128_from_template(v4i* inptr, v4i *out) { *out = templated_amdgcn_flat_load_monitor_b128<__ATOMIC_SEQ_CST, __MEMORY_SCOPE_SYSTEM>(inptr); } // CHECK-LABEL: define dso_local void @_Z48test_amdgcn_flat_load_monitor_b128_from_templatePDv4_iS0_( // CHECK-SAME: ptr noundef [[INPTR:%.*]], ptr noundef [[OUT:%.*]]) #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[INPTR_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) // CHECK-NEXT: [[OUT_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) // CHECK-NEXT: [[INPTR_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[INPTR_ADDR]] to ptr // CHECK-NEXT: [[OUT_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[OUT_ADDR]] to ptr // CHECK-NEXT: store ptr [[INPTR]], ptr [[INPTR_ADDR_ASCAST]], align 8 // CHECK-NEXT: store ptr [[OUT]], ptr [[OUT_ADDR_ASCAST]], align 8 // CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[INPTR_ADDR_ASCAST]], align 8 // CHECK-NEXT: [[CALL:%.*]] = call noundef <4 x i32> @_Z39templated_amdgcn_flat_load_monitor_b128ILj5ELj0EEDv4_iPS0_(ptr noundef [[TMP0]]) #[[ATTR2:[0-9]+]] // CHECK-NEXT: [[TMP1:%.*]] = load ptr, ptr [[OUT_ADDR_ASCAST]], align 8 // CHECK-NEXT: store <4 x i32> [[CALL]], ptr [[TMP1]], align 16 // CHECK-NEXT: ret void // // // CHECK-LABEL: define linkonce_odr noundef <4 x i32> @_Z39templated_amdgcn_flat_load_monitor_b128ILj5ELj0EEDv4_iPS0_( // CHECK-SAME: ptr noundef [[INPTR:%.*]]) #[[ATTR0]] comdat { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[INPTR_ADDR:%.*]] = alloca ptr, align 8, addrspace(5) // CHECK-NEXT: [[INPTR_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[INPTR_ADDR]] to ptr // CHECK-NEXT: store ptr [[INPTR]], ptr [[INPTR_ADDR_ASCAST]], align 8 // CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[INPTR_ADDR_ASCAST]], align 8 // CHECK-NEXT: [[TMP1:%.*]] = call <4 x i32> @llvm.amdgcn.flat.load.monitor.b128.v4i32(ptr [[TMP0]], i32 5, metadata [[META4:![0-9]+]]) // CHECK-NEXT: ret <4 x i32> [[TMP1]] // //. // CHECK: [[META4]] = !{!""} //.