[LICM] Cap the clobbering calls in LICM.

Summary:
Unlimitted number of calls to getClobberingAccess can lead to high
compile times in pathological cases.
Switching EnableLicmCap flag from bool to int, and enabling to default 100.
(tested to be appropriate for current bechmarks)
We can revisit this value when enabling MemorySSA.

Reviewers: sanjoy, chandlerc, george.burgess.iv

Subscribers: jlebar, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D57968

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353897 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/include/llvm/Transforms/Utils/LoopUtils.h b/include/llvm/Transforms/Utils/LoopUtils.h
index d1448874d4aa6c41210f5950fbb18444659ed502..a5883ac79bb6d875fdd6fe04ddf532909b0e3401 100644 (file)
--- a/include/llvm/Transforms/Utils/LoopUtils.h
+++ b/include/llvm/Transforms/Utils/LoopUtils.h
@@ -112,7 +112,7 @@ bool formLCSSARecursively(Loop &L, DominatorTree &DT, LoopInfo *LI,
 bool sinkRegion(DomTreeNode *, AliasAnalysis *, LoopInfo *, DominatorTree *,
                 TargetLibraryInfo *, TargetTransformInfo *, Loop *,
                 AliasSetTracker *, MemorySSAUpdater *, ICFLoopSafetyInfo *,
-                bool, OptimizationRemarkEmitter *);
+                bool, int &, OptimizationRemarkEmitter *);
 
 /// Walk the specified region of the CFG (defined by all blocks
 /// dominated by the specified block, and that are in the current loop) in depth
@@ -124,7 +124,7 @@ bool sinkRegion(DomTreeNode *, AliasAnalysis *, LoopInfo *, DominatorTree *,
 /// ORE. It returns changed status.
 bool hoistRegion(DomTreeNode *, AliasAnalysis *, LoopInfo *, DominatorTree *,
                  TargetLibraryInfo *, Loop *, AliasSetTracker *,
-                 MemorySSAUpdater *, ICFLoopSafetyInfo *, bool,
+                 MemorySSAUpdater *, ICFLoopSafetyInfo *, bool, int &,
                  OptimizationRemarkEmitter *);
 
 /// This function deletes dead loops. The caller of this function needs to
@@ -277,6 +277,7 @@ bool canSinkOrHoistInst(Instruction &I, AAResults *AA, DominatorTree *DT,
                         Loop *CurLoop, AliasSetTracker *CurAST,
                         MemorySSAUpdater *MSSAU, bool TargetExecutesOncePerLoop,
                         bool NoOfMemAccessesTooLarge,
+                        int *LicmMssaOptCap = nullptr,
                         OptimizationRemarkEmitter *ORE = nullptr);
 
 /// Returns a Min/Max operation corresponding to MinMaxRecurrenceKind.

diff --git a/lib/Transforms/Scalar/LICM.cpp b/lib/Transforms/Scalar/LICM.cpp
index c83889748d428b8c54854972acf7622b6b1f7c7e..7384ea0b0a35d7fcb6005d599d1ed9a9e5cbfeda 100644 (file)
--- a/lib/Transforms/Scalar/LICM.cpp
+++ b/lib/Transforms/Scalar/LICM.cpp
@@ -106,17 +106,19 @@ static cl::opt<int>
 LICMN2Theshold("licm-n2-threshold", cl::Hidden, cl::init(0),
                cl::desc("How many instruction to cross product using AA"));
 
-// Experimental option to allow imprecision in LICM (use MemorySSA cap) in
-// pathological cases, in exchange for faster compile. This is to be removed
-// if MemorySSA starts to address the same issue. This flag applies only when
-// LICM uses MemorySSA instead on AliasSetTracker. When the flag is disabled
-// (default), LICM calls MemorySSAWalker's getClobberingMemoryAccess, which
-// gets perfect accuracy. When flag is enabled, LICM will call into MemorySSA's
-// getDefiningAccess, which may not be precise, since optimizeUses is capped.
-static cl::opt<bool> EnableLicmCap(
-    "enable-licm-cap", cl::init(false), cl::Hidden,
-    cl::desc("Enable imprecision in LICM (uses MemorySSA cap) in "
-             "pathological cases, in exchange for faster compile"));
+// Experimental option to allow imprecision in LICM in pathological cases, in
+// exchange for faster compile. This is to be removed if MemorySSA starts to
+// address the same issue. This flag applies only when LICM uses MemorySSA
+// instead on AliasSetTracker. LICM calls MemorySSAWalker's
+// getClobberingMemoryAccess, up to the value of the Cap, getting perfect
+// accuracy. Afterwards, LICM will call into MemorySSA's getDefiningAccess,
+// which may not be precise, since optimizeUses is capped. The result is
+// correct, but we may not get as "far up" as possible to get which access is
+// clobbering the one queried.
+static cl::opt<int> LicmMssaOptCap(
+    "licm-mssa-optimization-cap", cl::init(100), cl::Hidden,
+    cl::desc("Enable imprecision in LICM in pathological cases, in exchange "
+             "for faster compile. Caps the MemorySSA clobbering calls."));
 
 // Experimentally, memory promotion carries less importance than sinking and
 // hoisting. Limit when we do promotion when using MemorySSA, in order to save
@@ -149,7 +151,8 @@ static bool pointerInvalidatedByLoop(MemoryLocation MemLoc,
                                      AliasSetTracker *CurAST, Loop *CurLoop,
                                      AliasAnalysis *AA);
 static bool pointerInvalidatedByLoopWithMSSA(MemorySSA *MSSA, MemoryUse *MU,
-                                             Loop *CurLoop);
+                                             Loop *CurLoop,
+                                             int &LicmMssaOptCounter);
 static Instruction *CloneInstructionInExitBlock(
     Instruction &I, BasicBlock &ExitBlock, PHINode &PN, const LoopInfo *LI,
     const LoopSafetyInfo *SafetyInfo, MemorySSAUpdater *MSSAU);
@@ -307,6 +310,7 @@ bool LoopInvariantCodeMotion::runOnLoop(
   std::unique_ptr<AliasSetTracker> CurAST;
   std::unique_ptr<MemorySSAUpdater> MSSAU;
   bool NoOfMemAccTooLarge = false;
+  int LicmMssaOptCounter = 0;
 
   if (!MSSA) {
     LLVM_DEBUG(dbgs() << "LICM: Using Alias Set Tracker.\n");
@@ -352,11 +356,11 @@ bool LoopInvariantCodeMotion::runOnLoop(
   if (L->hasDedicatedExits())
     Changed |= sinkRegion(DT->getNode(L->getHeader()), AA, LI, DT, TLI, TTI, L,
                           CurAST.get(), MSSAU.get(), &SafetyInfo,
-                          NoOfMemAccTooLarge, ORE);
+                          NoOfMemAccTooLarge, LicmMssaOptCounter, ORE);
   if (Preheader)
     Changed |= hoistRegion(DT->getNode(L->getHeader()), AA, LI, DT, TLI, L,
                            CurAST.get(), MSSAU.get(), &SafetyInfo,
-                           NoOfMemAccTooLarge, ORE);
+                           NoOfMemAccTooLarge, LicmMssaOptCounter, ORE);
 
   // Now that all loop invariants have been removed from the loop, promote any
   // memory references to scalars that we can.
@@ -461,7 +465,7 @@ bool llvm::sinkRegion(DomTreeNode *N, AliasAnalysis *AA, LoopInfo *LI,
                       TargetTransformInfo *TTI, Loop *CurLoop,
                       AliasSetTracker *CurAST, MemorySSAUpdater *MSSAU,
                       ICFLoopSafetyInfo *SafetyInfo, bool NoOfMemAccTooLarge,
-                      OptimizationRemarkEmitter *ORE) {
+                      int &LicmMssaOptCounter, OptimizationRemarkEmitter *ORE) {
 
   // Verify inputs.
   assert(N != nullptr && AA != nullptr && LI != nullptr && DT != nullptr &&
@@ -505,7 +509,7 @@ bool llvm::sinkRegion(DomTreeNode *N, AliasAnalysis *AA, LoopInfo *LI,
       bool FreeInLoop = false;
       if (isNotUsedOrFreeInLoop(I, CurLoop, SafetyInfo, TTI, FreeInLoop) &&
           canSinkOrHoistInst(I, AA, DT, CurLoop, CurAST, MSSAU, true,
-                             NoOfMemAccTooLarge, ORE) &&
+                             NoOfMemAccTooLarge, &LicmMssaOptCounter, ORE) &&
           !I.mayHaveSideEffects()) {
         if (sink(I, LI, DT, CurLoop, SafetyInfo, MSSAU, ORE, FreeInLoop)) {
           if (!FreeInLoop) {
@@ -760,6 +764,7 @@ bool llvm::hoistRegion(DomTreeNode *N, AliasAnalysis *AA, LoopInfo *LI,
                        DominatorTree *DT, TargetLibraryInfo *TLI, Loop *CurLoop,
                        AliasSetTracker *CurAST, MemorySSAUpdater *MSSAU,
                        ICFLoopSafetyInfo *SafetyInfo, bool NoOfMemAccTooLarge,
+                       int &LicmMssaOptCounter,
                        OptimizationRemarkEmitter *ORE) {
   // Verify inputs.
   assert(N != nullptr && AA != nullptr && LI != nullptr && DT != nullptr &&
@@ -813,7 +818,7 @@ bool llvm::hoistRegion(DomTreeNode *N, AliasAnalysis *AA, LoopInfo *LI,
       // to that block.
       if (CurLoop->hasLoopInvariantOperands(&I) &&
           canSinkOrHoistInst(I, AA, DT, CurLoop, CurAST, MSSAU, true,
-                             NoOfMemAccTooLarge, ORE) &&
+                             NoOfMemAccTooLarge, &LicmMssaOptCounter, ORE) &&
           isSafeToExecuteUnconditionally(
               I, DT, CurLoop, SafetyInfo, ORE,
               CurLoop->getLoopPreheader()->getTerminator())) {
@@ -1040,13 +1045,15 @@ bool llvm::canSinkOrHoistInst(Instruction &I, AAResults *AA, DominatorTree *DT,
                               Loop *CurLoop, AliasSetTracker *CurAST,
                               MemorySSAUpdater *MSSAU,
                               bool TargetExecutesOncePerLoop,
-                              bool NoOfMemAccTooLarge,
+                              bool NoOfMemAccTooLarge, int *LicmMssaOptCounter,
                               OptimizationRemarkEmitter *ORE) {
   // If we don't understand the instruction, bail early.
   if (!isHoistableAndSinkableInst(I))
     return false;
 
   MemorySSA *MSSA = MSSAU ? MSSAU->getMemorySSA() : nullptr;
+  if (MSSA)
+    assert(LicmMssaOptCounter != nullptr && "Counter cannot be null.");
 
   // Loads have extra constraints we have to verify before we can hoist them.
   if (LoadInst *LI = dyn_cast<LoadInst>(&I)) {
@@ -1073,7 +1080,8 @@ bool llvm::canSinkOrHoistInst(Instruction &I, AAResults *AA, DominatorTree *DT,
                                              CurLoop, AA);
     else
       Invalidated = pointerInvalidatedByLoopWithMSSA(
-          MSSA, cast<MemoryUse>(MSSA->getMemoryAccess(LI)), CurLoop);
+          MSSA, cast<MemoryUse>(MSSA->getMemoryAccess(LI)), CurLoop,
+          *LicmMssaOptCounter);
     // Check loop-invariant address because this may also be a sinkable load
     // whose address is not necessarily loop-invariant.
     if (ORE && Invalidated && CurLoop->isLoopInvariant(LI->getPointerOperand()))
@@ -1118,7 +1126,8 @@ bool llvm::canSinkOrHoistInst(Instruction &I, AAResults *AA, DominatorTree *DT,
                   CurAST, CurLoop, AA);
             else
               Invalidated = pointerInvalidatedByLoopWithMSSA(
-                  MSSA, cast<MemoryUse>(MSSA->getMemoryAccess(CI)), CurLoop);
+                  MSSA, cast<MemoryUse>(MSSA->getMemoryAccess(CI)), CurLoop,
+                  *LicmMssaOptCounter);
             if (Invalidated)
               return false;
           }
@@ -1177,8 +1186,9 @@ bool llvm::canSinkOrHoistInst(Instruction &I, AAResults *AA, DominatorTree *DT,
     } else { // MSSAU
       if (isOnlyMemoryAccess(SI, CurLoop, MSSAU))
         return true;
-      if (!EnableLicmCap) {
+      if (*LicmMssaOptCounter < LicmMssaOptCap) {
         auto *Source = MSSA->getSkipSelfWalker()->getClobberingMemoryAccess(SI);
+        (*LicmMssaOptCounter)++;
         // If there are no clobbering Defs in the loop, we still need to check
         // for interfering Uses. If there are more accesses than the Promotion
         // cap, give up, we're not walking a list that long. Otherwise, walk the
@@ -2195,13 +2205,16 @@ static bool pointerInvalidatedByLoop(MemoryLocation MemLoc,
 }
 
 static bool pointerInvalidatedByLoopWithMSSA(MemorySSA *MSSA, MemoryUse *MU,
-                                             Loop *CurLoop) {
+                                             Loop *CurLoop,
+                                             int &LicmMssaOptCounter) {
   MemoryAccess *Source;
-  // See declaration of EnableLicmCap for usage details.
-  if (EnableLicmCap)
+  // See declaration of LicmMssaOptCap for usage details.
+  if (LicmMssaOptCounter >= LicmMssaOptCap)
     Source = MU->getDefiningAccess();
-  else
+  else {
     Source = MSSA->getSkipSelfWalker()->getClobberingMemoryAccess(MU);
+    LicmMssaOptCounter++;
+  }
   return !MSSA->isLiveOnEntryDef(Source) &&
          CurLoop->contains(Source->getBlock());
 }