[llvm][ConstraintElimination]Insert ConditionFact into loop header in case of monotonic induction variables #112080
There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -186,9 +186,12 @@ struct State { | |
| LoopInfo &LI; | ||
| ScalarEvolution &SE; | ||
| SmallVector<FactOrCheck, 64> WorkList; | ||
| bool AddInductionInfoIntoHeader = false; | ||
|
|
||
| State(DominatorTree &DT, LoopInfo &LI, ScalarEvolution &SE, | ||
| bool AddInductionInfoIntoHeader = false) | ||
| : DT(DT), LI(LI), SE(SE), | ||
| AddInductionInfoIntoHeader(AddInductionInfoIntoHeader) {} | ||
|
|
||
| /// Process block \p BB and add known facts to work-list. | ||
| void addInfoFor(BasicBlock &BB); | ||
|
|
@@ -197,6 +200,8 @@ struct State { | |
| /// controlling the loop header. | ||
| void addInfoForInductions(BasicBlock &BB); | ||
|
|
||
| void addConditionFactsIntoLoopHeader(BasicBlock &BB); | ||
|
|
||
| /// Returns true if we can add a known condition from BB to its successor | ||
| /// block Succ. | ||
| bool canAddSuccessor(BasicBlock &BB, BasicBlock *Succ) const { | ||
|
|
@@ -900,7 +905,62 @@ static void dumpConstraint(ArrayRef<int64_t> C, | |
| } | ||
| #endif | ||
|
|
||
| // For monotonically decreasing/increasing variables in the loop, | ||
| // this will insert ConditionFact PN >= StartingValue or PN <= StartingValue | ||
| // associated with the loop header, where PN is the corresponding PHi node. | ||
| void State::addConditionFactsIntoLoopHeader(BasicBlock &BB) { | ||
| auto *L = LI.getLoopFor(&BB); | ||
| if (!L || L->getHeader() != &BB) | ||
| return; | ||
| DomTreeNode *DTN = DT.getNode(&BB); | ||
| for (PHINode &PN : L->getHeader()->phis()) { | ||
| if (PN.getNumIncomingValues() != 2 || PN.getParent() != &BB || | ||
| !SE.isSCEVable(PN.getType())) | ||
| continue; | ||
| auto *AR = dyn_cast_or_null<SCEVAddRecExpr>(SE.getSCEV(&PN)); | ||
| BasicBlock *LoopPred = L->getLoopPredecessor(); | ||
| if (!AR || AR->getLoop() != L || !LoopPred) | ||
| return; | ||
| const SCEV *StartSCEV = AR->getStart(); | ||
| Value *StartValue = nullptr; | ||
| if (auto *C = dyn_cast<SCEVConstant>(StartSCEV)) { | ||
| StartValue = C->getValue(); | ||
| } else { | ||
| StartValue = PN.getIncomingValueForBlock(LoopPred); | ||
| assert(SE.getSCEV(StartValue) == StartSCEV && "inconsistent start value"); | ||
| } | ||
| auto IncUnsigned = SE.getMonotonicPredicateType(AR, CmpInst::ICMP_UGT); | ||
| auto IncSigned = SE.getMonotonicPredicateType(AR, CmpInst::ICMP_SGT); | ||
|
|
||
| // Monotonically Increasing | ||
| bool MonotonicallyIncreasingUnsigned = | ||
| IncUnsigned && *IncUnsigned == ScalarEvolution::MonotonicallyIncreasing; | ||
| bool MonotonicallyIncreasingSigned = | ||
| IncSigned && *IncSigned == ScalarEvolution::MonotonicallyIncreasing; | ||
| if (MonotonicallyIncreasingUnsigned) | ||
| WorkList.push_back(FactOrCheck::getConditionFact(DTN, CmpInst::ICMP_UGE, | ||
| &PN, StartValue)); | ||
| if (MonotonicallyIncreasingSigned) | ||
| WorkList.push_back(FactOrCheck::getConditionFact(DTN, CmpInst::ICMP_SGE, | ||
| &PN, StartValue)); | ||
|
|
||
| // Monotonically Decreasing | ||
| bool MonotonicallyDecreasingUnsigned = | ||
| IncUnsigned && *IncUnsigned == ScalarEvolution::MonotonicallyDecreasing; | ||
| bool MonotonicallyDecreasingSigned = | ||
| IncSigned && *IncSigned == ScalarEvolution::MonotonicallyDecreasing; | ||
| if (MonotonicallyDecreasingUnsigned) | ||
| WorkList.push_back(FactOrCheck::getConditionFact(DTN, CmpInst::ICMP_ULE, | ||
| &PN, StartValue)); | ||
| if (MonotonicallyDecreasingSigned) | ||
| WorkList.push_back(FactOrCheck::getConditionFact(DTN, CmpInst::ICMP_SLE, | ||
| &PN, StartValue)); | ||
| } | ||
| } | ||
|
|
||
| void State::addInfoForInductions(BasicBlock &BB) { | ||
| if (AddInductionInfoIntoHeader) | ||
| addConditionFactsIntoLoopHeader(BB); | ||
| auto *L = LI.getLoopFor(&BB); | ||
| if (!L || L->getHeader() != &BB) | ||
| return; | ||
|
|
@@ -1358,7 +1418,7 @@ static std::optional<bool> checkCondition(CmpInst::Predicate Pred, Value *A, | |
| LLVM_DEBUG(dbgs() << "Checking " << *CheckInst << "\n"); | ||
|
|
||
| auto R = Info.getConstraintForSolving(Pred, A, B); | ||
| if (R.empty() || !R.isValid(Info)) { | ||
| LLVM_DEBUG(dbgs() << " failed to decompose condition\n"); | ||
| return std::nullopt; | ||
| } | ||
|
|
@@ -1671,6 +1731,16 @@ tryToSimplifyOverflowMath(IntrinsicInst *II, ConstraintInfo &Info, | |
| return Changed; | ||
| } | ||
|
|
||
| static unsigned int getNumConditionalBranches(Function &F) { | ||
| unsigned int NumCondBranches = 0; | ||
| for (BasicBlock &BB : F) { | ||
| BranchInst *BranchInstr = dyn_cast_or_null<BranchInst>(BB.getTerminator()); | ||
| if (BranchInstr && BranchInstr->isConditional()) | ||
| NumCondBranches++; | ||
| } | ||
| return NumCondBranches; | ||
| } | ||
|
|
||
| static bool eliminateConstraints(Function &F, DominatorTree &DT, LoopInfo &LI, | ||
| ScalarEvolution &SE, | ||
| OptimizationRemarkEmitter &ORE) { | ||
|
|
@@ -1680,7 +1750,9 @@ static bool eliminateConstraints(Function &F, DominatorTree &DT, LoopInfo &LI, | |
| for (Value &Arg : F.args()) | ||
| FunctionArgs.push_back(&Arg); | ||
| ConstraintInfo Info(F.getDataLayout(), FunctionArgs); | ||
| unsigned int NumCondBranches = getNumConditionalBranches(F); | ||
| State S(DT, LI, SE, | ||
| /* AddInductionInfoIntoHeader= */ NumCondBranches < MaxRows / 5); | ||
|
There was a problem hiding this comment. I was not aware of the existing Doing a whole walk over the function to count conditional branches feels somewhat heavy and we potentially get an amount of facts in the number of There was a problem hiding this comment. The reason I added this condition is to avoid regression in some edge cases (see analysis by @dtcxzyw). If we add extra constrains in loop headers and reach MaxRows limit, some constraints are dropped that are not otherwise. I am happy to remove it if we are OK with those regressions. There was a problem hiding this comment. It just feels like a lot of trouble (or rather a lot of memory read) to go through the whole function just to count the number of conditional branches... So I am wondering if there is ways to limit this without having an extra pass over the whole function... |
||
| std::unique_ptr<Module> ReproducerModule( | ||
| DumpReproducers ? new Module(F.getName(), F.getContext()) : nullptr); | ||
|
|
||
|
|
||
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,47 @@ | ||
| ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5 | ||
| ; RUN: opt < %s -passes=constraint-elimination -S | FileCheck %s | ||
|
|
||
| define i32 @foo() { | ||
| ; CHECK-LABEL: define i32 @foo() { | ||
| ; CHECK-NEXT: init: | ||
| ; CHECK-NEXT: br label %[[OUTER_LOOP_CONTROL:.*]] | ||
| ; CHECK: outer.loop.control: | ||
| ; CHECK-NEXT: [[X_0:%.*]] = phi i32 [ 0, [[INIT:%.*]] ], [ [[X_OUTER:%.*]], %[[OUTER_LOOP_INC:.*]] ] | ||
| ; CHECK-NEXT: [[TMP0:%.*]] = icmp slt i32 [[X_0]], 10 | ||
| ; CHECK-NEXT: br i1 [[TMP0]], label %[[INNER_LOOP_CONTROL:.*]], label %[[EXIT:.*]] | ||
| ; CHECK: inner.loop.control: | ||
| ; CHECK-NEXT: [[X_1:%.*]] = phi i32 [ [[X_0]], %[[OUTER_LOOP_CONTROL]] ], [ [[X_INNER:%.*]], %[[INNER_LOOP_BODY:.*]] ] | ||
| ; CHECK-NEXT: br i1 false, label %[[INNER_LOOP_BODY]], label %[[OUTER_LOOP_INC]] | ||
| ; CHECK: inner.loop.body: | ||
| ; CHECK-NEXT: [[X_INNER]] = add nsw i32 [[X_1]], -1 | ||
| ; CHECK-NEXT: br label %[[INNER_LOOP_CONTROL]] | ||
| ; CHECK: outer.loop.inc: | ||
| ; CHECK-NEXT: [[X_OUTER]] = add nsw i32 [[X_1]], 2 | ||
| ; CHECK-NEXT: br label %[[OUTER_LOOP_CONTROL]] | ||
| ; CHECK: exit: | ||
| ; CHECK-NEXT: ret i32 [[X_0]] | ||
| ; | ||
| init: | ||
| br label %outer.loop.control | ||
|
|
||
| outer.loop.control: ; preds = %init, %outer.loop.inc | ||
| %x.0 = phi i32 [ 0, %init ], [ %x.outer, %outer.loop.inc ] | ||
| %0 = icmp slt i32 %x.0, 10 | ||
| br i1 %0, label %inner.loop.control, label %exit | ||
|
|
||
| inner.loop.control: ; preds = %outer.loop.control, %inner.loop.body | ||
| %x.1 = phi i32 [ %x.0, %outer.loop.control ], [ %x.inner, %inner.loop.body ] | ||
| %1 = icmp sgt i32 %x.1, 20 | ||
| br i1 %1, label %inner.loop.body, label %outer.loop.inc | ||
|
|
||
| inner.loop.body: ; preds = %inner.loop.control | ||
| %x.inner = add nsw i32 %x.1, -1 | ||
| br label %inner.loop.control | ||
|
|
||
| outer.loop.inc: ; preds = %inner.loop.control | ||
| %x.outer = add nsw i32 %x.1, 2 | ||
| br label %outer.loop.control | ||
|
|
||
| exit: ; preds = %1 | ||
| ret i32 %x.0 | ||
| } |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Oops, something went wrong.
Oops, something went wrong.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Uh oh!
There was an error while loading. Please reload this page.