81 lines · plain
1// RUN: fir-opt --split-input-file --cuf-add-constructor %s | FileCheck %s2 3module attributes {dlti.dl_spec = #dlti.dl_spec<#dlti.dl_entry<!llvm.ptr, dense<64> : vector<4xi64>>, #dlti.dl_entry<!llvm.ptr<271>, dense<32> : vector<4xi64>>, #dlti.dl_entry<!llvm.ptr<270>, dense<32> : vector<4xi64>>, #dlti.dl_entry<f128, dense<128> : vector<2xi64>>, #dlti.dl_entry<f64, dense<64> : vector<2xi64>>, #dlti.dl_entry<f80, dense<128> : vector<2xi64>>, #dlti.dl_entry<f16, dense<16> : vector<2xi64>>, #dlti.dl_entry<i32, dense<32> : vector<2xi64>>, #dlti.dl_entry<i16, dense<16> : vector<2xi64>>, #dlti.dl_entry<i128, dense<128> : vector<2xi64>>, #dlti.dl_entry<i8, dense<8> : vector<2xi64>>, #dlti.dl_entry<!llvm.ptr<272>, dense<64> : vector<4xi64>>, #dlti.dl_entry<i64, dense<64> : vector<2xi64>>, #dlti.dl_entry<i1, dense<8> : vector<2xi64>>, #dlti.dl_entry<"dlti.endianness", "little">, #dlti.dl_entry<"dlti.stack_alignment", 128 : i64>>, fir.defaultkind = "a1c4d8i4l4r4", fir.kindmap = "", gpu.container_module, llvm.data_layout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-i128:128-f80:128-n8:16:32:64-S128", llvm.ident = "flang version 20.0.0 (https://github.com/llvm/llvm-project.git cae351f3453a0a26ec8eb2ddaf773c24a29d929e)", llvm.target_triple = "x86_64-unknown-linux-gnu"} {4 5 fir.global @_QMmtestsEn(dense<[3, 4, 5, 6, 7]> : tensor<5xi32>) {data_attr = #cuf.cuda<device>} : !fir.array<5xi32>6 fir.global @_QMmtestsEndev {data_attr = #cuf.cuda<device>} : !fir.box<!fir.heap<!fir.array<?xi32>>> {7 %c0 = arith.constant 0 : index8 %0 = fir.zero_bits !fir.heap<!fir.array<?xi32>>9 %1 = fircg.ext_embox %0(%c0) {allocator_idx = 2 : i32} : (!fir.heap<!fir.array<?xi32>>, index) -> !fir.box<!fir.heap<!fir.array<?xi32>>>10 fir.has_value %1 : !fir.box<!fir.heap<!fir.array<?xi32>>>11 }12 13 gpu.module @cuda_device_mod {14 }15}16 17// CHECK: gpu.module @cuda_device_mod18 19// CHECK: llvm.func internal @__cudaFortranConstructor() {20// CHECK-DAG: %[[MODULE:.*]] = cuf.register_module @cuda_device_mod -> !llvm.ptr21// CHECK-DAG: %[[VAR_NAME:.*]] = fir.address_of(@_QQ{{.*}}) : !fir.ref<!fir.char<1,12>>22// CHECK-DAG: %[[VAR_ADDR:.*]] = fir.address_of(@_QMmtestsEn) : !fir.ref<!fir.array<5xi32>>23// CHECK-DAG: %[[MODULE2:.*]] = fir.convert %[[MODULE]] : (!llvm.ptr) -> !fir.ref<!fir.llvm_ptr<i8>>24// CHECK-DAG: %[[VAR_ADDR2:.*]] = fir.convert %[[VAR_ADDR]] : (!fir.ref<!fir.array<5xi32>>) -> !fir.ref<i8>25// CHECK-DAG: %[[VAR_NAME2:.*]] = fir.convert %[[VAR_NAME]] : (!fir.ref<!fir.char<1,12>>) -> !fir.ref<i8>26// CHECK-DAG: %[[CST:.*]] = arith.constant 20 : index27// CHECK-DAG: %[[CST2:.*]] = fir.convert %[[CST]] : (index) -> i6428// CHECK-DAG: fir.call @_FortranACUFRegisterVariable(%[[MODULE2]], %[[VAR_ADDR2]], %[[VAR_NAME2]], %[[CST2]]) : (!fir.ref<!fir.llvm_ptr<i8>>, !fir.ref<i8>, !fir.ref<i8>, i64) -> ()29// CHECK-DAG: %[[BOX:.*]] = fir.address_of(@_QMmtestsEndev) : !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>30// CHECK-DAG: %[[BOXREF:.*]] = fir.convert %[[BOX]] : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>) -> !fir.ref<i8>31// CHECK-DAG: fir.call @_FortranACUFRegisterVariable(%[[MODULE:.*]], %[[BOXREF]], %{{.*}}, %{{.*}})32//33 34// -----35 36// Checking that constant global variables are not registered37 38// CHECK: @_FortranACUFRegisterAllocator39// CHECK-NOT: fir.call @_FortranACUFRegisterVariable40 41module attributes {dlti.dl_spec = #dlti.dl_spec<i8 = dense<8> : vector<2xi64>, i16 = dense<16> : vector<2xi64>, i1 = dense<8> : vector<2xi64>, !llvm.ptr = dense<64> : vector<4xi64>, f80 = dense<128> : vector<2xi64>, i128 = dense<128> : vector<2xi64>, i64 = dense<64> : vector<2xi64>, !llvm.ptr<271> = dense<32> : vector<4xi64>, !llvm.ptr<272> = dense<64> : vector<4xi64>, f128 = dense<128> : vector<2xi64>, !llvm.ptr<270> = dense<32> : vector<4xi64>, f16 = dense<16> : vector<2xi64>, f64 = dense<64> : vector<2xi64>, i32 = dense<32> : vector<2xi64>, "dlti.stack_alignment" = 128 : i64, "dlti.endianness" = "little">, fir.defaultkind = "a1c4d8i4l4r4", fir.kindmap = "", gpu.container_module, llvm.data_layout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-i128:128-f80:128-n8:16:32:64-S128", llvm.ident = "flang version 20.0.0 (https://github.com/llvm/llvm-project.git 3372303188df0f7f8ac26e7ab610cf8b0f716d42)", llvm.target_triple = "x86_64-unknown-linux-gnu"} {42 fir.global @_QMiso_c_bindingECc_int constant : i3243 44 45 fir.type_info @_QM__fortran_builtinsT__builtin_c_ptr noinit nodestroy nofinal : !fir.type<_QM__fortran_builtinsT__builtin_c_ptr{__address:i64}>46 gpu.module @cuda_device_mod {47 fir.global @_QMiso_c_bindingECc_int {data_attr = #cuf.cuda<constant>} constant : i3248 gpu.func @_QMdevmodPdevsub(%arg0: !fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>) -> i32 {49 %c0 = arith.constant 0 : index50 %c4_i32 = arith.constant 4 : i3251 %0 = fir.alloca i32 {bindc_name = "devsub", uniq_name = "_QMdevmodFdevsubEdevsub"}52 %1 = fir.alloca i32 {bindc_name = "__builtin_warpsize", uniq_name = "_QM__fortran_builtinsEC__builtin_warpsize"}53 %2 = fir.load %arg0 : !fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>54 %3:3 = fir.box_dims %2, %c0 : (!fir.box<!fir.heap<!fir.array<?xf32>>>, index) -> (index, index, index)55 %4 = fir.convert %3#1 : (index) -> i3256 %5 = arith.muli %4, %c4_i32 : i3257 fir.store %5 to %0 : !fir.ref<i32>58 %6 = fir.load %0 : !fir.ref<i32>59 gpu.return %6 : i3260 }61 }62}63 64// -----65 66module attributes {dlti.dl_spec = #dlti.dl_spec<i8 = dense<8> : vector<2xi64>, i16 = dense<16> : vector<2xi64>, i1 = dense<8> : vector<2xi64>, !llvm.ptr = dense<64> : vector<4xi64>, f80 = dense<128> : vector<2xi64>, i128 = dense<128> : vector<2xi64>, i64 = dense<64> : vector<2xi64>, !llvm.ptr<271> = dense<32> : vector<4xi64>, !llvm.ptr<272> = dense<64> : vector<4xi64>, f128 = dense<128> : vector<2xi64>, !llvm.ptr<270> = dense<32> : vector<4xi64>, f16 = dense<16> : vector<2xi64>, f64 = dense<64> : vector<2xi64>, i32 = dense<32> : vector<2xi64>, "dlti.stack_alignment" = 128 : i64, "dlti.endianness" = "little">, fir.defaultkind = "a1c4d8i4l4r4", fir.kindmap = "", gpu.container_module, llvm.data_layout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-i128:128-f80:128-n8:16:32:64-S128", llvm.ident = "flang version 20.0.0 (https://github.com/llvm/llvm-project.git 3372303188df0f7f8ac26e7ab610cf8b0f716d42)", llvm.target_triple = "x86_64-unknown-linux-gnu"} {67 fir.global @_QMmEa00 {data_attr = #cuf.cuda<managed>} : !fir.box<!fir.heap<!fir.array<?x?x?x?x?xf64>>> {68 %c0 = arith.constant 0 : index69 %0 = fir.zero_bits !fir.heap<!fir.array<?x?x?x?x?xf64>>70 %1 = fir.shape %c0, %c0, %c0, %c0, %c0 : (index, index, index, index, index) -> !fir.shape<5>71 %2 = fir.embox %0(%1) {allocator_idx = 3 : i32} : (!fir.heap<!fir.array<?x?x?x?x?xf64>>, !fir.shape<5>) -> !fir.box<!fir.heap<!fir.array<?x?x?x?x?xf64>>>72 fir.has_value %2 : !fir.box<!fir.heap<!fir.array<?x?x?x?x?xf64>>>73 }74 gpu.module @cuda_device_mod {75 }76}77 78// CHECK: llvm.func internal @__cudaFortranConstructor()79// CHECK: fir.address_of(@_QMmEa00)80// CHECK: fir.call @_FortranACUFRegisterVariable81