//===-- Passes.td - Transforms pass definition file --------*- tablegen -*-===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
//
// This file contains definitions for passes within the Optimizer/Transforms/
// directory.
//
//===----------------------------------------------------------------------===//
#ifndef FLANG_OPTIMIZER_TRANSFORMS_PASSES
#define FLANG_OPTIMIZER_TRANSFORMS_PASSES
include "mlir/Pass/PassBase.td"
def AbstractResultOpt
: Pass<"abstract-result"> {
let summary = "Convert fir.array, fir.box and fir.rec function result to "
"function argument";
let description = [{
This pass is required before code gen to the LLVM IR dialect,
including the pre-cg rewrite pass.
}];
let dependentDialects = [
"fir::FIROpsDialect", "mlir::func::FuncDialect"
];
let options = [
Option<"passResultAsBox", "abstract-result-as-box",
"bool", /*default=*/"false",
"Pass fir.array<T> result as fir.box<fir.array<T>> argument instead"
" of fir.ref<fir.array<T>>.">
];
}
def AffineDialectPromotion : Pass<"promote-to-affine", "::mlir::func::FuncOp"> {
let summary = "Promotes `fir.{do_loop,if}` to `affine.{for,if}`.";
let description = [{
Convert fir operations which satisfy affine constraints to the affine
dialect.
`fir.do_loop` will be converted to `affine.for` if the loops inside the body
can be converted and the indices for memory loads and stores satisfy
`affine.apply` criteria for symbols and dimensions.
`fir.if` will be converted to `affine.if` where possible. `affine.if`'s
condition uses an integer set (==, >=) and an analysis is done to determine
the fir condition's parent operations to construct the integer set.
`fir.load` (`fir.store`) will be converted to `affine.load` (`affine.store`)
where possible. This conversion includes adding a dummy `fir.convert` cast
to adapt values of type `!fir.ref<!fir.array>` to `memref`. This is done
because the affine dialect presently only understands the `memref` type.
}];
let constructor = "::fir::createPromoteToAffinePass()";
let dependentDialects = [
"fir::FIROpsDialect", "mlir::func::FuncDialect",
"mlir::affine::AffineDialect"
];
}
def AffineDialectDemotion : Pass<"demote-affine", "::mlir::func::FuncOp"> {
let summary = "Converts `affine.{load,store}` back to fir operations";
let description = [{
Affine dialect's default lowering for loads and stores is different from
fir as it uses the `memref` type. The `memref` type is not compatible with
the Fortran runtime. Therefore, conversion of memory operations back to
`fir.load` and `fir.store` with `!fir.ref<?>` types is required.
}];
let constructor = "::fir::createAffineDemotionPass()";
let dependentDialects = [
"fir::FIROpsDialect", "mlir::func::FuncDialect",
"mlir::affine::AffineDialect"
];
}
def AnnotateConstantOperands : Pass<"annotate-constant"> {
let summary = "Annotate constant operands to all FIR operations";
let description = [{
The MLIR canonicalizer makes a distinction between constants based on how
they are packaged in the IR. A constant value is wrapped in an Attr and that
Attr can be attached to an Op. There is a distinguished Op, ConstantOp, that
merely has one of these Attr attached.
The MLIR canonicalizer treats constants referenced by an Op and constants
referenced through a ConstantOp as having distinct semantics. This pass
eliminates that distinction, so hashconsing of Ops, basic blocks, etc.
behaves as one would expect.
}];
let constructor = "::fir::createAnnotateConstantOperandsPass()";
let dependentDialects = [ "fir::FIROpsDialect" ];
}
def ArrayValueCopy : Pass<"array-value-copy", "::mlir::func::FuncOp"> {
let summary = "Convert array value operations to memory operations.";
let description = [{
Transform the set of array value primitives to a memory-based array
representation.
The Ops `array_load`, `array_store`, `array_fetch`, and `array_update` are
used to manage abstract aggregate array values. A simple analysis is done
to determine if there are potential dependences between these operations.
If not, these array operations can be lowered to work directly on the memory
representation. If there is a potential conflict, a temporary is created
along with appropriate copy-in/copy-out operations. Here, a more refined
analysis might be deployed, such as using the affine framework.
This pass is required before code gen to the LLVM IR dialect.
}];
let constructor = "::fir::createArrayValueCopyPass()";
let dependentDialects = [ "fir::FIROpsDialect" ];
let options = [
Option<"optimizeConflicts", "optimize-conflicts", "bool",
/*default=*/"false",
"do more detailed conflict analysis to reduce the number "
"of temporaries">
];
}
def CharacterConversion : Pass<"character-conversion"> {
let summary = "Convert CHARACTER entities with different KINDs";
let description = [{
Translates entities of one CHARACTER KIND to another.
By default the translation is to naively zero-extend or truncate a code
point to fit the destination size.
}];
let dependentDialects = [ "fir::FIROpsDialect" ];
let options = [
Option<"useRuntimeCalls", "use-runtime-calls",
"std::string", /*default=*/"std::string{}",
"Generate runtime calls to a named set of conversion routines. "
"By default, the conversions may produce unexpected results.">
];
}
def CFGConversion : Pass<"cfg-conversion"> {
let summary = "Convert FIR structured control flow ops to CFG ops.";
let description = [{
Transform the `fir.do_loop`, `fir.if`, `fir.iterate_while` and
`fir.select_type` ops into plain old test and branch operations. Removing
the high-level control structures can enable other optimizations.
This pass is required before code gen to the LLVM IR dialect.
}];
let dependentDialects = [
"fir::FIROpsDialect", "mlir::func::FuncDialect"
];
let options = [
Option<"forceLoopToExecuteOnce", "always-execute-loop-body", "bool",
/*default=*/"false",
"force the body of a loop to execute at least once">,
Option<"setNSW", "set-nsw", "bool",
/*default=*/"false",
"set nsw on loop variable increment">
];
}
def ExternalNameConversion : Pass<"external-name-interop", "mlir::ModuleOp"> {
let summary = "Convert name for external interoperability";
let description = [{
Demangle FIR internal name and mangle them for external interoperability.
}];
let options = [
Option<"appendUnderscoreOpt", "append-underscore",
"bool", /*default=*/"true",
"Append trailing underscore to external names.">
];
}
def CompilerGeneratedNamesConversion : Pass<"compiler-generated-names",
"mlir::ModuleOp"> {
let summary = "Convert names of compiler generated globals";
let description = [{
Transforms names of compiler generated globals to avoid
characters that might be unsupported by some target toolchains.
All special symbols are replaced with a predefined 'X' character.
This is only done for uniqued names that are not externally facing.
The uniqued names always use '_Q' prefix, and the user entity names
are always lower cased, so using 'X' instead of the special symbols
will guarantee that the converted name will not conflict with the user
space. This pass does not affect the externally facing names,
because the expectation is that the compiler will not generate
externally facing names on its own, and these names cannot use
special symbols.
}];
}
def MemRefDataFlowOpt : Pass<"fir-memref-dataflow-opt", "::mlir::func::FuncOp"> {
let summary =
"Perform store/load forwarding and potentially removing dead stores.";
let description = [{
This pass performs store to load forwarding to eliminate memory accesses and
potentially the entire allocation if all the accesses are forwarded.
}];
let constructor = "::fir::createMemDataFlowOptPass()";
let dependentDialects = [
"fir::FIROpsDialect", "mlir::func::FuncDialect"
];
}
// This needs to be a "mlir::ModuleOp" pass, because we are creating debug for
// the module in this pass.
def AddDebugInfo : Pass<"add-debug-info", "mlir::ModuleOp"> {
let summary = "Add the debug info";
let description = [{
Emit debug info that can be understood by llvm.
}];
let constructor = "::fir::createAddDebugInfoPass()";
let dependentDialects = [
"fir::FIROpsDialect", "mlir::func::FuncDialect", "mlir::LLVM::LLVMDialect"
];
let options = [
Option<"debugLevel", "debug-level",
"mlir::LLVM::DIEmissionKind",
/*default=*/"mlir::LLVM::DIEmissionKind::Full",
"debug level",
[{::llvm::cl::values(
clEnumValN(mlir::LLVM::DIEmissionKind::Full, "Full", "Emit full debug info"),
clEnumValN(mlir::LLVM::DIEmissionKind::LineTablesOnly, "LineTablesOnly", "Emit line tables only"),
clEnumValN(mlir::LLVM::DIEmissionKind::None, "None", "Emit no debug information")
)}]
>,
Option<"isOptimized", "is-optimized",
"bool", /*default=*/"false",
"is optimized.">,
Option<"inputFilename", "file-name",
"std::string",
/*default=*/"std::string{}",
"name of the input source file">,
];
}
// This needs to be a "mlir::ModuleOp" pass, because it inserts simplified
// functions into the module, which is invalid if a finer grain mlir::Operation
// is used as the pass specification says to not touch things outside hte scope
// of the operation being processed.
def SimplifyIntrinsics : Pass<"simplify-intrinsics", "mlir::ModuleOp"> {
let summary = "Intrinsics simplification";
let description = [{
Qualifying intrinsics calls are replaced with calls to a specialized and
simplified function. The simplified function is added to the current module.
This function can be inlined by a general purpose inlining pass.
}];
let options = [
Option<"enableExperimental", "enable-experimental", "bool",
/*default=*/"false",
"Enable experimental code that may not always work correctly">
];
}
def MemoryAllocationOpt : Pass<"memory-allocation-opt", "mlir::func::FuncOp"> {
let summary = "Convert stack to heap allocations and vice versa.";
let description = [{
Convert stack allocations to heap allocations and vice versa based on
estimated size, lifetime, usage patterns, the call tree, etc.
}];
let dependentDialects = [ "fir::FIROpsDialect" ];
let options = [
Option<"dynamicArrayOnHeap", "dynamic-array-on-heap",
"bool", /*default=*/"false",
"Allocate all arrays with runtime determined size on heap.">,
Option<"maxStackArraySize", "maximum-array-alloc-size",
"std::size_t", /*default=*/"~static_cast<std::size_t>(0)",
"Set maximum number of elements of an array allocated on the stack.">
];
}
// This needs to be a "mlir::ModuleOp" pass, because it inserts global constants
def ConstantArgumentGlobalisationOpt : Pass<"constant-argument-globalisation-opt", "mlir::ModuleOp"> {
let summary = "Convert constant function arguments to global constants.";
let description = [{
Convert scalar literals of function arguments to global constants.
}];
let dependentDialects = [ "fir::FIROpsDialect" ];
}
def StackArrays : Pass<"stack-arrays", "mlir::func::FuncOp"> {
let summary = "Move local array allocations from heap memory into stack memory";
let description = [{
Convert heap allocations for arrays, even those of unknown size, into stack
allocations.
}];
let dependentDialects = [ "fir::FIROpsDialect" ];
}
def StackReclaim : Pass<"stack-reclaim"> {
let summary = "Insert stacksave/stackrestore in region with allocas";
let description = [{
Insert stacksave/stackrestore in loop region to reclaim alloca done in its
scope.
}];
let dependentDialects = [ "mlir::LLVM::LLVMDialect" ];
}
def AddAliasTags : Pass<"fir-add-alias-tags", "mlir::ModuleOp"> {
let summary = "Add tbaa tags to operations that implement FirAliasAnalysisOpInterface";
let description = [{
TBAA (type based alias analysis) is one method to pass pointer alias information
from language frontends to LLVM. This pass uses fir::AliasAnalysis to add this
information to fir.load and fir.store operations.
Additional tags are added during codegen. See fir::TBAABuilder.
This needs to be a separate pass so that it happens before structured control
flow operations are lowered to branches and basic blocks (this makes tracing
the source of values much eaiser). The other TBAA tags need to be applied to
box loads and stores which are implicit in FIR and so cannot be annotated
until codegen.
TODO: this is currently a pass on mlir::ModuleOp to avoid parallelism. In
theory, each operation could be considered in prallel, so long as there
aren't races adding new tags to the mlir context.
}];
let dependentDialects = [ "fir::FIROpsDialect" ];
}
def SimplifyRegionLite : Pass<"simplify-region-lite", "mlir::ModuleOp"> {
let summary = "Region simplification";
let description = [{
Run region DCE and erase unreachable blocks in regions.
}];
}
def AlgebraicSimplification : Pass<"flang-algebraic-simplification"> {
let summary = "";
let description = [{
Run algebraic simplifications for Math/Complex/etc. dialect operations.
This is a flang specific pass, because we may want to "tune"
the rewrite patterns specifically for Fortran (e.g. increase
the limit for constant exponent value that defines the cases
when pow(x, constant) is transformed into a set of multiplications, etc.).
}];
let dependentDialects = [ "mlir::math::MathDialect" ];
let constructor = "::fir::createAlgebraicSimplificationPass()";
}
def PolymorphicOpConversion : Pass<"fir-polymorphic-op", "mlir::ModuleOp"> {
let summary =
"Simplify operations on polymorphic types";
let description = [{
This pass breaks up the lowering of operations on polymorphic types by
introducing an intermediate FIR level that simplifies code geneation.
}];
let dependentDialects = [
"fir::FIROpsDialect", "mlir::func::FuncDialect"
];
}
def LoopVersioning : Pass<"loop-versioning", "mlir::func::FuncOp"> {
let summary = "Loop Versioning";
let description = [{
Loop Versioning pass adds a check and two variants of a loop when the input
array is an assumed shape array, to optimize for the (often common) case where
an array has element sized stride. The element sizes stride allows some
loops to be vectorized as well as other loop optimizations.
}];
let dependentDialects = [ "fir::FIROpsDialect" ];
}
def VScaleAttr : Pass<"vscale-attr", "mlir::func::FuncOp"> {
let summary = "Add vscale_range attribute to functions";
let description = [{
Set an attribute for the vscale range on functions, to allow scalable
vector operations to be used on processors with variable vector length.
}];
let options = [
Option<"vscaleRange", "vscale-range",
"std::pair<unsigned, unsigned>", /*default=*/"std::pair<unsigned, unsigned>{}",
"vector scale range">,
];
}
def FunctionAttr : Pass<"function-attr", "mlir::func::FuncOp"> {
let summary = "Pass that adds function attributes expected at LLVM IR level";
let description = [{ This feature introduces a general attribute aimed at
customizing function characteristics.
Options include:
Add "frame-pointer" attribute to functions: Set an attribute for the frame
pointer on functions, to avoid saving the frame pointer in a register in
functions where it is unnecessary. This eliminates the need for
instructions to save, establish, and restore frame pointers, while also
freeing up an additional register in numerous functions. However, this
approach can make debugging unfeasible on certain machines.
}];
let options = [
Option<"framePointerKind", "frame-pointer",
"mlir::LLVM::framePointerKind::FramePointerKind",
/*default=*/"mlir::LLVM::framePointerKind::FramePointerKind{}",
"frame pointer">,
Option<"noInfsFPMath", "no-infs-fp-math",
"bool", /*default=*/"false",
"Set the no-infs-fp-math attribute on functions in the module.">,
Option<"noNaNsFPMath", "no-nans-fp-math",
"bool", /*default=*/"false",
"Set the no-nans-fp-math attribute on functions in the module.">,
Option<"approxFuncFPMath", "approx-func-fp-math",
"bool", /*default=*/"false",
"Set the approx-func-fp-math attribute on functions in the module.">,
Option<"noSignedZerosFPMath", "no-signed-zeros-fp-math",
"bool", /*default=*/"false",
"Set the no-signed-zeros-fp-math attribute on functions in the module.">,
Option<"unsafeFPMath", "unsafe-fp-math",
"bool", /*default=*/"false",
"Set the unsafe-fp-math attribute on functions in the module.">,
Option<"tuneCPU", "tune-cpu",
"llvm::StringRef", /*default=*/"llvm::StringRef{}",
"Set the tune-cpu attribute on functions in the module.">,
];
}
def AssumedRankOpConversion : Pass<"fir-assumed-rank-op", "mlir::ModuleOp"> {
let summary =
"Simplify operations on assumed-rank types";
let description = [{
This pass breaks up the lowering of operations on assumed-rank types by
introducing an intermediate FIR level that simplifies code generation.
}];
let dependentDialects = [
"fir::FIROpsDialect", "mlir::func::FuncDialect"
];
}
def CufOpConversion : Pass<"cuf-convert", "mlir::ModuleOp"> {
let summary = "Convert some CUF operations to runtime calls";
let dependentDialects = [
"fir::FIROpsDialect"
];
}
def CufImplicitDeviceGlobal :
Pass<"cuf-implicit-device-global", "mlir::ModuleOp"> {
let summary = "Flag globals used in device function with data attribute";
let dependentDialects = [
"cuf::CUFDialect"
];
}
def CUFAddConstructor : Pass<"cuf-add-constructor", "mlir::ModuleOp"> {
let summary = "Add constructor to register CUDA Fortran allocators";
let dependentDialects = [
"mlir::func::FuncDialect"
];
}
#endif // FLANG_OPTIMIZER_TRANSFORMS_PASSES
Codebase Browser
llvm