//===-- TBAABuilder.cpp -- TBAA builder definitions -----------------------===//
//
// 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
//
//===----------------------------------------------------------------------===//
//
// Coding style: https://mlir.llvm.org/getting_started/DeveloperGuide/
//
//===----------------------------------------------------------------------===//
#include "flang/Optimizer/CodeGen/TBAABuilder.h"
#include "flang/Optimizer/Dialect/FIRType.h"
#include "llvm/ADT/TypeSwitch.h"
#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
#include <mlir/Dialect/LLVMIR/LLVMAttrs.h>
#include <mlir/Dialect/LLVMIR/LLVMDialect.h>
#include <mlir/Dialect/LLVMIR/LLVMTypes.h>
#define DEBUG_TYPE "flang-tbaa-builder"
using namespace mlir;
using namespace mlir::LLVM;
static llvm::cl::opt<bool> disableTBAA(
"disable-tbaa",
llvm::cl::desc("disable attaching TBAA tags to memory accessing operations "
"to override default Flang behavior"),
llvm::cl::init(false));
// disabling this will play badly with the FIR TBAA pass, leading to worse
// performance
static llvm::cl::opt<bool> perFunctionTBAATrees(
"per-function-tbaa-trees",
llvm::cl::desc("Give each function an independent TBAA tree (default)"),
llvm::cl::init(true), llvm::cl::Hidden);
// tagAttachmentLimit is a debugging option that allows limiting
// the number of TBAA access tag attributes attached to operations.
// It is set to kTagAttachmentUnlimited by default denoting "no limit".
static constexpr unsigned kTagAttachmentUnlimited =
std::numeric_limits<unsigned>::max();
static llvm::cl::opt<unsigned>
tagAttachmentLimit("tbaa-attach-tag-max", llvm::cl::desc(""),
llvm::cl::init(kTagAttachmentUnlimited));
namespace fir {
TBAABuilder::TBAABuilder(MLIRContext *context, bool applyTBAA,
bool forceUnifiedTree)
: enableTBAA(applyTBAA && !disableTBAA),
trees(/*separatePerFunction=*/perFunctionTBAATrees && !forceUnifiedTree) {
// TODO: the TBAA tags created here are rooted in the root scope
// of the enclosing function. This does not work best with MLIR inlining.
// A better approach is to root them according to the scopes they belong to
// and that were used by AddAliasTagsPass to create TBAA tags before
// the CodeGen. For example:
// subroutine caller(a, b, ptr)
// real, target :: a(:), b(:)
// integer, pointer :: ptr(:)
// call callee(a, b, ptr)
// end
// subroutine callee(a, b, ptr)
// real :: a(:), b(:)
// integer, pointer :: ptr(:)
// do i=...
// a(ptr(i)) = b(ptr(i))
// end do
// end
//
// When callee is inlined, the dummy arguments 'a' and 'b' will
// be rooted in TBAA tree corresponding to the `call callee` call site,
// saying that the references to 'a' and 'b' cannot alias each other.
// These tags will be created by AddAliasTagsPass, but it will not be able
// to create any tags for 'ptr' references.
// During the CodeGen, we create 'any data access' tags for the
// 'ptr' acceses. If they are rooted within the root scope of `caller`,
// they end up in a different TBAA tree with the 'a' and 'b' access
// tags, so 'ptr', 'a' and 'b' references MayAlias. Moreover,
// the box access of 'ptr' will also be in a different TBAA tree
// with 'a' and 'b' tags, meaning they can also alias.
// This will prevent LLVM vectorization even with memory conflict checks.
// It seems that we'd better move all TBAA tags assignment to
// AddAliasTagsPass, which can at least rely on the dummy arguments scopes.
if (!enableTBAA)
return;
}
TBAATagAttr TBAABuilder::getAccessTag(TBAATypeDescriptorAttr baseTypeDesc,
TBAATypeDescriptorAttr accessTypeDesc,
int64_t offset) {
TBAATagAttr &tag = tagsMap[{baseTypeDesc, accessTypeDesc, offset}];
if (tag)
return tag;
// Initialize new tag.
tag = TBAATagAttr::get(baseTypeDesc, accessTypeDesc, offset);
return tag;
}
TBAATagAttr TBAABuilder::getAnyBoxAccessTag(mlir::LLVM::LLVMFuncOp func) {
TBAATypeDescriptorAttr boxMemberTypeDesc = trees[func].boxMemberTypeDesc;
return getAccessTag(boxMemberTypeDesc, boxMemberTypeDesc, /*offset=*/0);
}
TBAATagAttr TBAABuilder::getBoxAccessTag(Type baseFIRType, Type accessFIRType,
GEPOp gep,
mlir::LLVM::LLVMFuncOp func) {
return getAnyBoxAccessTag(func);
}
TBAATagAttr TBAABuilder::getAnyDataAccessTag(mlir::LLVM::LLVMFuncOp func) {
TBAATypeDescriptorAttr anyDataAccessTypeDesc = trees[func].anyDataTypeDesc;
return getAccessTag(anyDataAccessTypeDesc, anyDataAccessTypeDesc,
/*offset=*/0);
}
TBAATagAttr TBAABuilder::getDataAccessTag(Type baseFIRType, Type accessFIRType,
GEPOp gep,
mlir::LLVM::LLVMFuncOp func) {
return getAnyDataAccessTag(func);
}
TBAATagAttr TBAABuilder::getAnyAccessTag(mlir::LLVM::LLVMFuncOp func) {
TBAATypeDescriptorAttr anyAccessTypeDesc = trees[func].anyAccessDesc;
return getAccessTag(anyAccessTypeDesc, anyAccessTypeDesc, /*offset=*/0);
}
void TBAABuilder::attachTBAATag(AliasAnalysisOpInterface op, Type baseFIRType,
Type accessFIRType, GEPOp gep) {
if (!enableTBAA)
return;
mlir::LLVM::LLVMFuncOp func = op->getParentOfType<mlir::LLVM::LLVMFuncOp>();
if (!func)
return;
++tagAttachmentCounter;
if (tagAttachmentLimit != kTagAttachmentUnlimited &&
tagAttachmentCounter > tagAttachmentLimit)
return;
LLVM_DEBUG(llvm::dbgs() << "Attaching TBAA tag #" << tagAttachmentCounter
<< "\n");
TBAATagAttr tbaaTagSym;
if (fir::isRecordWithDescriptorMember(baseFIRType)) {
// A memory access that addresses an aggregate that contains
// a mix of data members and descriptor members may alias
// with both data and descriptor accesses.
// Conservatively set any-access tag if there is any descriptor member.
tbaaTagSym = getAnyAccessTag(func);
} else if (mlir::isa<fir::BaseBoxType>(baseFIRType)) {
tbaaTagSym = getBoxAccessTag(baseFIRType, accessFIRType, gep, func);
} else {
tbaaTagSym = getDataAccessTag(baseFIRType, accessFIRType, gep, func);
}
if (!tbaaTagSym)
return;
op.setTBAATags(ArrayAttr::get(op->getContext(), tbaaTagSym));
}
} // namespace fir