llvm-project/llvm/lib/Analysis/Delinearization.cpp

141 lines
4.8 KiB
C++
Raw Normal View History

//===---- Delinearization.cpp - MultiDimensional Index Delinearization ----===//
//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
//
// This implements an analysis pass that tries to delinearize all GEP
// instructions in all loops using the SCEV analysis functionality. This pass is
// only used for testing purposes: if your pass needs delinearization, please
// use the on-demand SCEVAddRecExpr::delinearize() function.
//
//===----------------------------------------------------------------------===//
#include "llvm/Analysis/LoopInfo.h"
#include "llvm/Analysis/Passes.h"
#include "llvm/Analysis/ScalarEvolution.h"
#include "llvm/Analysis/ScalarEvolutionExpressions.h"
#include "llvm/IR/Constants.h"
#include "llvm/IR/DerivedTypes.h"
#include "llvm/IR/Function.h"
#include "llvm/IR/InstIterator.h"
#include "llvm/IR/Instructions.h"
#include "llvm/IR/LLVMContext.h"
#include "llvm/IR/Type.h"
#include "llvm/Pass.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/raw_ostream.h"
using namespace llvm;
#define DL_NAME "delinearize"
#define DEBUG_TYPE DL_NAME
namespace {
class Delinearization : public FunctionPass {
Delinearization(const Delinearization &); // do not implement
protected:
Function *F;
LoopInfo *LI;
ScalarEvolution *SE;
public:
static char ID; // Pass identification, replacement for typeid
Delinearization() : FunctionPass(ID) {
initializeDelinearizationPass(*PassRegistry::getPassRegistry());
}
bool runOnFunction(Function &F) override;
void getAnalysisUsage(AnalysisUsage &AU) const override;
void print(raw_ostream &O, const Module *M = nullptr) const override;
};
} // end anonymous namespace
void Delinearization::getAnalysisUsage(AnalysisUsage &AU) const {
AU.setPreservesAll();
AU.addRequired<LoopInfoWrapperPass>();
[PM] Port ScalarEvolution to the new pass manager. This change makes ScalarEvolution a stand-alone object and just produces one from a pass as needed. Making this work well requires making the object movable, using references instead of overwritten pointers in a number of places, and other refactorings. I've also wired it up to the new pass manager and added a RUN line to a test to exercise it under the new pass manager. This includes basic printing support much like with other analyses. But there is a big and somewhat scary change here. Prior to this patch ScalarEvolution was never *actually* invalidated!!! Re-running the pass just re-wired up the various other analyses and didn't remove any of the existing entries in the SCEV caches or clear out anything at all. This might seem OK as everything in SCEV that can uses ValueHandles to track updates to the values that serve as SCEV keys. However, this still means that as we ran SCEV over each function in the module, we kept accumulating more and more SCEVs into the cache. At the end, we would have a SCEV cache with every value that we ever needed a SCEV for in the entire module!!! Yowzers. The releaseMemory routine would dump all of this, but that isn't realy called during normal runs of the pipeline as far as I can see. To make matters worse, there *is* actually a key that we don't update with value handles -- there is a map keyed off of Loop*s. Because LoopInfo *does* release its memory from run to run, it is entirely possible to run SCEV over one function, then over another function, and then lookup a Loop* from the second function but find an entry inserted for the first function! Ouch. To make matters still worse, there are plenty of updates that *don't* trip a value handle. It seems incredibly unlikely that today GVN or another pass that invalidates SCEV can update values in *just* such a way that a subsequent run of SCEV will incorrectly find lookups in a cache, but it is theoretically possible and would be a nightmare to debug. With this refactoring, I've fixed all this by actually destroying and recreating the ScalarEvolution object from run to run. Technically, this could increase the amount of malloc traffic we see, but then again it is also technically correct. ;] I don't actually think we're suffering from tons of malloc traffic from SCEV because if we were, the fact that we never clear the memory would seem more likely to have come up as an actual problem before now. So, I've made the simple fix here. If in fact there are serious issues with too much allocation and deallocation, I can work on a clever fix that preserves the allocations (while clearing the data) between each run, but I'd prefer to do that kind of optimization with a test case / benchmark that shows why we need such cleverness (and that can test that we actually make it faster). It's possible that this will make some things faster by making the SCEV caches have higher locality (due to being significantly smaller) so until there is a clear benchmark, I think the simple change is best. Differential Revision: http://reviews.llvm.org/D12063 llvm-svn: 245193
2015-08-17 10:08:17 +08:00
AU.addRequired<ScalarEvolutionWrapperPass>();
}
bool Delinearization::runOnFunction(Function &F) {
this->F = &F;
[PM] Port ScalarEvolution to the new pass manager. This change makes ScalarEvolution a stand-alone object and just produces one from a pass as needed. Making this work well requires making the object movable, using references instead of overwritten pointers in a number of places, and other refactorings. I've also wired it up to the new pass manager and added a RUN line to a test to exercise it under the new pass manager. This includes basic printing support much like with other analyses. But there is a big and somewhat scary change here. Prior to this patch ScalarEvolution was never *actually* invalidated!!! Re-running the pass just re-wired up the various other analyses and didn't remove any of the existing entries in the SCEV caches or clear out anything at all. This might seem OK as everything in SCEV that can uses ValueHandles to track updates to the values that serve as SCEV keys. However, this still means that as we ran SCEV over each function in the module, we kept accumulating more and more SCEVs into the cache. At the end, we would have a SCEV cache with every value that we ever needed a SCEV for in the entire module!!! Yowzers. The releaseMemory routine would dump all of this, but that isn't realy called during normal runs of the pipeline as far as I can see. To make matters worse, there *is* actually a key that we don't update with value handles -- there is a map keyed off of Loop*s. Because LoopInfo *does* release its memory from run to run, it is entirely possible to run SCEV over one function, then over another function, and then lookup a Loop* from the second function but find an entry inserted for the first function! Ouch. To make matters still worse, there are plenty of updates that *don't* trip a value handle. It seems incredibly unlikely that today GVN or another pass that invalidates SCEV can update values in *just* such a way that a subsequent run of SCEV will incorrectly find lookups in a cache, but it is theoretically possible and would be a nightmare to debug. With this refactoring, I've fixed all this by actually destroying and recreating the ScalarEvolution object from run to run. Technically, this could increase the amount of malloc traffic we see, but then again it is also technically correct. ;] I don't actually think we're suffering from tons of malloc traffic from SCEV because if we were, the fact that we never clear the memory would seem more likely to have come up as an actual problem before now. So, I've made the simple fix here. If in fact there are serious issues with too much allocation and deallocation, I can work on a clever fix that preserves the allocations (while clearing the data) between each run, but I'd prefer to do that kind of optimization with a test case / benchmark that shows why we need such cleverness (and that can test that we actually make it faster). It's possible that this will make some things faster by making the SCEV caches have higher locality (due to being significantly smaller) so until there is a clear benchmark, I think the simple change is best. Differential Revision: http://reviews.llvm.org/D12063 llvm-svn: 245193
2015-08-17 10:08:17 +08:00
SE = &getAnalysis<ScalarEvolutionWrapperPass>().getSE();
LI = &getAnalysis<LoopInfoWrapperPass>().getLoopInfo();
return false;
}
static Value *getPointerOperand(Instruction &Inst) {
if (LoadInst *Load = dyn_cast<LoadInst>(&Inst))
return Load->getPointerOperand();
else if (StoreInst *Store = dyn_cast<StoreInst>(&Inst))
return Store->getPointerOperand();
else if (GetElementPtrInst *Gep = dyn_cast<GetElementPtrInst>(&Inst))
return Gep->getPointerOperand();
return nullptr;
}
void Delinearization::print(raw_ostream &O, const Module *) const {
O << "Delinearization on function " << F->getName() << ":\n";
for (inst_iterator I = inst_begin(F), E = inst_end(F); I != E; ++I) {
Instruction *Inst = &(*I);
// Only analyze loads and stores.
if (!isa<StoreInst>(Inst) && !isa<LoadInst>(Inst) &&
!isa<GetElementPtrInst>(Inst))
continue;
const BasicBlock *BB = Inst->getParent();
// Delinearize the memory access as analyzed in all the surrounding loops.
// Do not analyze memory accesses outside loops.
for (Loop *L = LI->getLoopFor(BB); L != nullptr; L = L->getParentLoop()) {
const SCEV *AccessFn = SE->getSCEVAtScope(getPointerOperand(*Inst), L);
const SCEVUnknown *BasePointer =
dyn_cast<SCEVUnknown>(SE->getPointerBase(AccessFn));
// Do not delinearize if we cannot find the base pointer.
if (!BasePointer)
break;
AccessFn = SE->getMinusSCEV(AccessFn, BasePointer);
O << "\n";
O << "Inst:" << *Inst << "\n";
O << "In Loop with Header: " << L->getHeader()->getName() << "\n";
O << "AccessFunction: " << *AccessFn << "\n";
SmallVector<const SCEV *, 3> Subscripts, Sizes;
SE->delinearize(AccessFn, Subscripts, Sizes, SE->getElementSize(Inst));
if (Subscripts.size() == 0 || Sizes.size() == 0 ||
split delinearization pass in 3 steps To compute the dimensions of the array in a unique way, we split the delinearization analysis in three steps: - find parametric terms in all memory access functions - compute the array dimensions from the set of terms - compute the delinearized access functions for each dimension The first step is executed on all the memory access functions such that we gather all the patterns in which an array is accessed. The second step reduces all this information in a unique description of the sizes of the array. The third step is delinearizing each memory access function following the common description of the shape of the array computed in step 2. This rewrite of the delinearization pass also solves a problem we had with the previous implementation: because the previous algorithm was by induction on the structure of the SCEV, it would not correctly recognize the shape of the array when the memory access was not following the nesting of the loops: for example, see polly/test/ScopInfo/multidim_only_ivs_3d_reverse.ll ; void foo(long n, long m, long o, double A[n][m][o]) { ; ; for (long i = 0; i < n; i++) ; for (long j = 0; j < m; j++) ; for (long k = 0; k < o; k++) ; A[i][k][j] = 1.0; Starting with this patch we no longer delinearize access functions that do not contain parameters, for example in test/Analysis/DependenceAnalysis/GCD.ll ;; for (long int i = 0; i < 100; i++) ;; for (long int j = 0; j < 100; j++) { ;; A[2*i - 4*j] = i; ;; *B++ = A[6*i + 8*j]; these accesses will not be delinearized as the upper bound of the loops are constants, and their access functions do not contain SCEVUnknown parameters. llvm-svn: 208232
2014-05-08 02:01:20 +08:00
Subscripts.size() != Sizes.size()) {
O << "failed to delinearize\n";
continue;
}
O << "Base offset: " << *BasePointer << "\n";
O << "ArrayDecl[UnknownSize]";
split delinearization pass in 3 steps To compute the dimensions of the array in a unique way, we split the delinearization analysis in three steps: - find parametric terms in all memory access functions - compute the array dimensions from the set of terms - compute the delinearized access functions for each dimension The first step is executed on all the memory access functions such that we gather all the patterns in which an array is accessed. The second step reduces all this information in a unique description of the sizes of the array. The third step is delinearizing each memory access function following the common description of the shape of the array computed in step 2. This rewrite of the delinearization pass also solves a problem we had with the previous implementation: because the previous algorithm was by induction on the structure of the SCEV, it would not correctly recognize the shape of the array when the memory access was not following the nesting of the loops: for example, see polly/test/ScopInfo/multidim_only_ivs_3d_reverse.ll ; void foo(long n, long m, long o, double A[n][m][o]) { ; ; for (long i = 0; i < n; i++) ; for (long j = 0; j < m; j++) ; for (long k = 0; k < o; k++) ; A[i][k][j] = 1.0; Starting with this patch we no longer delinearize access functions that do not contain parameters, for example in test/Analysis/DependenceAnalysis/GCD.ll ;; for (long int i = 0; i < 100; i++) ;; for (long int j = 0; j < 100; j++) { ;; A[2*i - 4*j] = i; ;; *B++ = A[6*i + 8*j]; these accesses will not be delinearized as the upper bound of the loops are constants, and their access functions do not contain SCEVUnknown parameters. llvm-svn: 208232
2014-05-08 02:01:20 +08:00
int Size = Subscripts.size();
for (int i = 0; i < Size - 1; i++)
O << "[" << *Sizes[i] << "]";
O << " with elements of " << *Sizes[Size - 1] << " bytes.\n";
O << "ArrayRef";
for (int i = 0; i < Size; i++)
O << "[" << *Subscripts[i] << "]";
O << "\n";
}
}
}
char Delinearization::ID = 0;
static const char delinearization_name[] = "Delinearization";
INITIALIZE_PASS_BEGIN(Delinearization, DL_NAME, delinearization_name, true,
true)
INITIALIZE_PASS_DEPENDENCY(LoopInfoWrapperPass)
INITIALIZE_PASS_END(Delinearization, DL_NAME, delinearization_name, true, true)
FunctionPass *llvm::createDelinearizationPass() { return new Delinearization; }