shylie/llvm-project
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[flang][cuda][NFC] Fix typo in CUFAllocDescriptor (#120797)

Browse Source 
Missing `r` in the function name.
This commit is contained in:
Valentin Clement (バレンタイン クレメン) 2024-12-20 13:57:47 -08:00 committed by GitHub
parent 81831ef3e7
commit e650ac1654
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
7 changed files with 19 additions and 19 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
flang/include/flang/Runtime/CUDA/descriptor.h
View File

@ -18,7 +18,7 @@ namespace Fortran::runtime::cuda {
extern "C" {
/// Allocate a descriptor in managed.
Descriptor *RTDECL(CUFAllocDesciptor)(
Descriptor *RTDECL(CUFAllocDescriptor)(
std::size_t, const char *sourceFile = nullptr, int sourceLine = 0);
/// Deallocate a descriptor allocated in managed or unified memory.

12
flang/lib/Optimizer/CodeGen/CodeGen.cpp
View File

@ -1210,12 +1210,12 @@ genCUFAllocDescriptor(mlir::Location loc,
llvmPointerType, {llvmIntPtrType, llvmPointerType, llvmInt32Type});
auto llvmFunc = mod.lookupSymbol<mlir::LLVM::LLVMFuncOp>(
RTNAME_STRING(CUFAllocDesciptor));
RTNAME_STRING(CUFAllocDescriptor));
auto funcFunc =
mod.lookupSymbol<mlir::func::FuncOp>(RTNAME_STRING(CUFAllocDesciptor));
mod.lookupSymbol<mlir::func::FuncOp>(RTNAME_STRING(CUFAllocDescriptor));
if (!llvmFunc && !funcFunc)
mlir::OpBuilder::atBlockEnd(mod.getBody())
.create<mlir::LLVM::LLVMFuncOp>(loc, RTNAME_STRING(CUFAllocDesciptor),
.create<mlir::LLVM::LLVMFuncOp>(loc, RTNAME_STRING(CUFAllocDescriptor),
fctTy);
mlir::Type structTy = typeConverter.convertBoxTypeAsStruct(boxTy);
@ -1224,7 +1224,7 @@ genCUFAllocDescriptor(mlir::Location loc,
genConstantIndex(loc, llvmIntPtrType, rewriter, boxSize);
llvm::SmallVector args = {sizeInBytes, sourceFile, sourceLine};
return rewriter
.create<mlir::LLVM::CallOp>(loc, fctTy, RTNAME_STRING(CUFAllocDesciptor),
.create<mlir::LLVM::CallOp>(loc, fctTy, RTNAME_STRING(CUFAllocDescriptor),
args)
.getResult();
}
@ -1735,7 +1735,7 @@ static bool isDeviceAllocation(mlir::Value val) {
(callOp.getCallee().value().getRootReference().getValue().starts_with(
RTNAME_STRING(CUFMemAlloc)) ||
callOp.getCallee().value().getRootReference().getValue().starts_with(
RTNAME_STRING(CUFAllocDesciptor))))
RTNAME_STRING(CUFAllocDescriptor))))
return true;
return false;
}
@ -3127,7 +3127,7 @@ struct LoadOpConversion : public fir::FIROpConversion<fir::LoadOp> {
inputBoxStorage.getDefiningOp())) {
if (callOp.getCallee() &&
(*callOp.getCallee())
.starts_with(RTNAME_STRING(CUFAllocDesciptor))) {
.starts_with(RTNAME_STRING(CUFAllocDescriptor))) {
// CUDA Fortran local descriptor are allocated in managed memory. So
// new storage must be allocated the same way.
auto mod = load->getParentOfType<mlir::ModuleOp>();

2
flang/lib/Optimizer/Transforms/CUFOpConversion.cpp
View File

@ -342,7 +342,7 @@ struct CUFAllocOpConversion : public mlir::OpRewritePattern<cuf::AllocOp> {
// Convert descriptor allocations to function call.
auto boxTy = mlir::dyn_cast_or_null<fir::BaseBoxType>(op.getInType());
mlir::func::FuncOp func =
fir::runtime::getRuntimeFunc<mkRTKey(CUFAllocDesciptor)>(loc, builder);
fir::runtime::getRuntimeFunc<mkRTKey(CUFAllocDescriptor)>(loc, builder);
auto fTy = func.getFunctionType();
mlir::Value sourceLine =
fir::factory::locationToLineNo(builder, loc, fTy.getInput(2));

2
flang/runtime/CUDA/descriptor.cpp
View File

@ -18,7 +18,7 @@ namespace Fortran::runtime::cuda {
extern "C" {
RT_EXT_API_GROUP_BEGIN
Descriptor *RTDEF(CUFAllocDesciptor)(
Descriptor *RTDEF(CUFAllocDescriptor)(
std::size_t sizeInBytes, const char *sourceFile, int sourceLine) {
return reinterpret_cast<Descriptor *>(CUFAllocManaged(sizeInBytes));
}

2
flang/test/Fir/CUDA/cuda-allocate.fir
View File

@ -15,7 +15,7 @@ func.func @_QPsub1() {
}
// CHECK-LABEL: func.func @_QPsub1()
// CHECK: %[[DESC_RT_CALL:.*]] = fir.call @_FortranACUFAllocDesciptor(%{{.*}}, %{{.*}}, %{{.*}}) : (i64, !fir.ref<i8>, i32) -> !fir.ref<!fir.box<none>>
// CHECK: %[[DESC_RT_CALL:.*]] = fir.call @_FortranACUFAllocDescriptor(%{{.*}}, %{{.*}}, %{{.*}}) : (i64, !fir.ref<i8>, i32) -> !fir.ref<!fir.box<none>>
// CHECK: %[[DESC:.*]] = fir.convert %[[DESC_RT_CALL]] : (!fir.ref<!fir.box<none>>) -> !fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>
// CHECK: %[[DECL_DESC:.*]]:2 = hlfir.declare %[[DESC]] {data_attr = #cuf.cuda<device>, fortran_attrs = #fir.var_attrs<allocatable>, uniq_name = "_QFsub1Ea"} : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>) -> (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>, !fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>)
// CHECK: %[[BOX_NONE:.*]] = fir.convert %[[DECL_DESC]]#1 : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>) -> !fir.ref<!fir.box<none>>

16
flang/test/Fir/CUDA/cuda-code-gen.mlir
View File

@ -8,7 +8,7 @@ module attributes {dlti.dl_spec = #dlti.dl_spec<#dlti.dl_entry<f80, dense<128> :
%c48 = arith.constant 48 : index
%1 = fir.convert %c48 : (index) -> i64
%2 = fir.convert %0 : (!fir.ref<!fir.char<1,8>>) -> !fir.ref<i8>
%3 = fir.call @_FortranACUFAllocDesciptor(%1, %2, %c4_i32) : (i64, !fir.ref<i8>, i32) -> !fir.ref<!fir.box<none>>
%3 = fir.call @_FortranACUFAllocDescriptor(%1, %2, %c4_i32) : (i64, !fir.ref<i8>, i32) -> !fir.ref<!fir.box<none>>
%4 = fir.convert %3 : (!fir.ref<!fir.box<none>>) -> !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>
%5 = fir.zero_bits !fir.heap<!fir.array<?xi32>>
%6 = fircg.ext_embox %5(%c0) {allocator_idx = 2 : i32} : (!fir.heap<!fir.array<?xi32>>, index) -> !fir.box<!fir.heap<!fir.array<?xi32>>>
@ -18,13 +18,13 @@ module attributes {dlti.dl_spec = #dlti.dl_spec<#dlti.dl_entry<f80, dense<128> :
}
// CHECK-LABEL: llvm.func @_QQmain()
// CHECK-COUNT-2: llvm.call @_FortranACUFAllocDesciptor
// CHECK-COUNT-2: llvm.call @_FortranACUFAllocDescriptor
fir.global linkonce @_QQclX3C737464696E3E00 constant : !fir.char<1,8> {
%0 = fir.string_lit "<stdin>\00"(8) : !fir.char<1,8>
fir.has_value %0 : !fir.char<1,8>
}
func.func private @_FortranACUFAllocDesciptor(i64, !fir.ref<i8>, i32) -> !fir.ref<!fir.box<none>> attributes {fir.runtime}
func.func private @_FortranACUFAllocDescriptor(i64, !fir.ref<i8>, i32) -> !fir.ref<!fir.box<none>> attributes {fir.runtime}
}
// -----
@ -55,7 +55,7 @@ module attributes {dlti.dl_spec = #dlti.dl_spec<f80 = dense<128> : vector<2xi64>
// CHECK-LABEL: llvm.func @_QQmain()
// CHECK: llvm.call @_FortranACUFMemAlloc
// CHECK: llvm.call @_FortranACUFAllocDesciptor
// CHECK: llvm.call @_FortranACUFAllocDescriptor
// -----
@ -73,7 +73,7 @@ module attributes {dlti.dl_spec = #dlti.dl_spec<f80 = dense<128> : vector<2xi64>
%c72 = arith.constant 72 : index
%3 = fir.convert %c72 : (index) -> i64
%4 = fir.convert %2 : (!fir.ref<!fir.char<1,11>>) -> !fir.ref<i8>
%5 = fir.call @_FortranACUFAllocDesciptor(%3, %4, %c14_i32) : (i64, !fir.ref<i8>, i32) -> !fir.ref<!fir.box<none>>
%5 = fir.call @_FortranACUFAllocDescriptor(%3, %4, %c14_i32) : (i64, !fir.ref<i8>, i32) -> !fir.ref<!fir.box<none>>
%6 = fir.convert %5 : (!fir.ref<!fir.box<none>>) -> !fir.ref<!fir.box<!fir.heap<!fir.array<?x?xf32>>>>
%7 = fir.zero_bits !fir.heap<!fir.array<?x?xf32>>
%8 = fircg.ext_embox %7(%c0, %c0) {allocator_idx = 2 : i32} : (!fir.heap<!fir.array<?x?xf32>>, index, index) -> !fir.box<!fir.heap<!fir.array<?x?xf32>>>
@ -83,7 +83,7 @@ module attributes {dlti.dl_spec = #dlti.dl_spec<f80 = dense<128> : vector<2xi64>
%c48 = arith.constant 48 : index
%10 = fir.convert %c48 : (index) -> i64
%11 = fir.convert %9 : (!fir.ref<!fir.char<1,11>>) -> !fir.ref<i8>
%12 = fir.call @_FortranACUFAllocDesciptor(%10, %11, %c20_i32) : (i64, !fir.ref<i8>, i32) -> !fir.ref<!fir.box<none>>
%12 = fir.call @_FortranACUFAllocDescriptor(%10, %11, %c20_i32) : (i64, !fir.ref<i8>, i32) -> !fir.ref<!fir.box<none>>
%13 = fir.convert %12 : (!fir.ref<!fir.box<none>>) -> !fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>
%14 = fir.zero_bits !fir.heap<!fir.array<?xf32>>
%15 = fircg.ext_embox %14(%c0) {allocator_idx = 2 : i32} : (!fir.heap<!fir.array<?xf32>>, index) -> !fir.box<!fir.heap<!fir.array<?xf32>>>
@ -120,9 +120,9 @@ module attributes {dlti.dl_spec = #dlti.dl_spec<f80 = dense<128> : vector<2xi64>
%0 = fir.string_lit "dummy.mlir\00"(11) : !fir.char<1,11>
fir.has_value %0 : !fir.char<1,11>
}
func.func private @_FortranACUFAllocDesciptor(i64, !fir.ref<i8>, i32) -> !fir.ref<!fir.box<none>> attributes {fir.runtime}
func.func private @_FortranACUFAllocDescriptor(i64, !fir.ref<i8>, i32) -> !fir.ref<!fir.box<none>> attributes {fir.runtime}
func.func private @_FortranACUFAllocatableAllocate(!fir.ref<!fir.box<none>>, i64, i1, !fir.box<none>, !fir.ref<i8>, i32) -> i32 attributes {fir.runtime}
}
// CHECK-LABEL: llvm.func @_QQmain()
// CHECK-COUNT-4: llvm.call @_FortranACUFAllocDesciptor
// CHECK-COUNT-4: llvm.call @_FortranACUFAllocDescriptor

2
flang/unittests/Runtime/CUDA/AllocatorCUF.cpp
View File

@ -66,7 +66,7 @@ TEST(AllocatableCUFTest, DescriptorAllocationTest) {
// REAL(4), DEVICE, ALLOCATABLE :: a(:)
auto a{createAllocatable(TypeCategory::Real, 4)};
Descriptor *desc = nullptr;
desc = RTNAME(CUFAllocDesciptor)(a->SizeInBytes());
desc = RTNAME(CUFAllocDescriptor)(a->SizeInBytes());
EXPECT_TRUE(desc != nullptr);
RTNAME(CUFFreeDesciptor)(desc);
}