Branch Folding: Accept explicit threshold for tail merge size.

This is prep work for allowing the threshold to be different during layout,
and to enforce a single threshold between merging and duplicating during
layout. No observable change intended.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@279117 91177308-0d34-0410-b5e6-96231b3b80d8

lib/CodeGen/BranchFolding.cpp

@@ -110,9 +110,12 @@ bool BranchFolderPass::runOnMachineFunction(MachineFunction &MF) {
 
 BranchFolder::BranchFolder(bool defaultEnableTailMerge, bool CommonHoist,
                            MBFIWrapper &FreqInfo,
-                           const MachineBranchProbabilityInfo &ProbInfo)
-    : EnableHoistCommonCode(CommonHoist), MBBFreqInfo(FreqInfo),
-      MBPI(ProbInfo) {
+                           const MachineBranchProbabilityInfo &ProbInfo,
+                           unsigned MinTailLength)
+    : EnableHoistCommonCode(CommonHoist), MinCommonTailLength(MinTailLength),
+      MBBFreqInfo(FreqInfo), MBPI(ProbInfo) {
+  if (MinCommonTailLength == 0)
+    MinCommonTailLength = TailMergeSize;
   switch (FlagEnableTailMerge) {
   case cl::BOU_UNSET: EnableTailMerge = defaultEnableTailMerge; break;
   case cl::BOU_TRUE: EnableTailMerge = true; break;
@@ -591,9 +594,21 @@ static unsigned CountTerminators(MachineBasicBlock *MBB,
 /// and decide if it would be profitable to merge those tails.  Return the
 /// length of the common tail and iterators to the first common instruction
 /// in each block.
+/// MBB1, MBB2      The blocks to check
+/// MinCommonTailLength  Minimum size of tail block to be merged.
+/// CommonTailLen   Out parameter to record the size of the shared tail between
+///                 MBB1 and MBB2
+/// I1, I2          Iterator references that will be changed to point to the first
+///                 instruction in the common tail shared by MBB1,MBB2
+/// SuccBB          A common successor of MBB1, MBB2 which are in a canonical form
+///                 relative to SuccBB
+/// PredBB          The layout predecessor of SuccBB, if any.
+/// FuncletMembership  map from block to funclet #.
+/// AfterPlacement  True if we are merging blocks after layout. Stricter
+///                 thresholds apply to prevent undoing tail-duplication.
 static bool
 ProfitableToMerge(MachineBasicBlock *MBB1, MachineBasicBlock *MBB2,
-                  unsigned minCommonTailLength, unsigned &CommonTailLen,
+                  unsigned MinCommonTailLength, unsigned &CommonTailLen,
                   MachineBasicBlock::iterator &I1,
                   MachineBasicBlock::iterator &I2, MachineBasicBlock *SuccBB,
                   MachineBasicBlock *PredBB,
@@ -651,7 +666,7 @@ ProfitableToMerge(MachineBasicBlock *MBB1, MachineBasicBlock *MBB2,
     ++EffectiveTailLen;
 
   // Check if the common tail is long enough to be worthwhile.
-  if (EffectiveTailLen >= minCommonTailLength)
+  if (EffectiveTailLen >= MinCommonTailLength)
     return true;
 
   // If we are optimizing for code size, 2 instructions in common is enough if
@@ -674,7 +689,7 @@ ProfitableToMerge(MachineBasicBlock *MBB1, MachineBasicBlock *MBB2,
 /// those blocks appear in MergePotentials (where they are not necessarily
 /// consecutive).
 unsigned BranchFolder::ComputeSameTails(unsigned CurHash,
-                                        unsigned minCommonTailLength,
+                                        unsigned MinCommonTailLength,
                                         MachineBasicBlock *SuccBB,
                                         MachineBasicBlock *PredBB) {
   unsigned maxCommonTailLength = 0U;
@@ -687,7 +702,7 @@ unsigned BranchFolder::ComputeSameTails(unsigned CurHash,
     for (MPIterator I = std::prev(CurMPIter); I->getHash() == CurHash; --I) {
       unsigned CommonTailLen;
       if (ProfitableToMerge(CurMPIter->getBlock(), I->getBlock(),
-                            minCommonTailLength,
+                            MinCommonTailLength,
                             CommonTailLen, TrialBBI1, TrialBBI2,
                             SuccBB, PredBB,
                             FuncletMembership,
@@ -832,14 +847,13 @@ mergeMMOsFromMemoryOperations(MachineBasicBlock::iterator MBBIStartPos,
 // branch to Succ added (but the predecessor/successor lists need no
 // adjustment). The lone predecessor of Succ that falls through into Succ,
 // if any, is given in PredBB.
+// MinCommonTailLength - Except for the special cases below, tail-merge if
+// there are at least this many instructions in common.
 bool BranchFolder::TryTailMergeBlocks(MachineBasicBlock *SuccBB,
-                                      MachineBasicBlock *PredBB) {
+                                      MachineBasicBlock *PredBB,
+                                      unsigned MinCommonTailLength) {
   bool MadeChange = false;
 
-  // Except for the special cases below, tail-merge if there are at least
-  // this many instructions in common.
-  unsigned minCommonTailLength = TailMergeSize;
-
   DEBUG(dbgs() << "\nTryTailMergeBlocks: ";
         for (unsigned i = 0, e = MergePotentials.size(); i != e; ++i)
           dbgs() << "BB#" << MergePotentials[i].getBlock()->getNumber()
@@ -852,8 +866,8 @@ bool BranchFolder::TryTailMergeBlocks(MachineBasicBlock *SuccBB,
                    << PredBB->getNumber() << "\n";
         }
         dbgs() << "Looking for common tails of at least "
-               << minCommonTailLength << " instruction"
-               << (minCommonTailLength == 1 ? "" : "s") << '\n';
+               << MinCommonTailLength << " instruction"
+               << (MinCommonTailLength == 1 ? "" : "s") << '\n';
        );
 
   // Sort by hash value so that blocks with identical end sequences sort
@@ -867,10 +881,10 @@ bool BranchFolder::TryTailMergeBlocks(MachineBasicBlock *SuccBB,
     // Build SameTails, identifying the set of blocks with this hash code
     // and with the maximum number of instructions in common.
     unsigned maxCommonTailLength = ComputeSameTails(CurHash,
-                                                    minCommonTailLength,
+                                                    MinCommonTailLength,
                                                     SuccBB, PredBB);
 
-    // If we didn't find any pair that has at least minCommonTailLength
+    // If we didn't find any pair that has at least MinCommonTailLength
     // instructions in common, remove all blocks with this hash code and retry.
     if (SameTails.empty()) {
       RemoveBlocksWithHash(CurHash, SuccBB, PredBB);
@@ -976,7 +990,7 @@ bool BranchFolder::TailMergeBlocks(MachineFunction &MF) {
 
     // See if we can do any tail merging on those.
     if (MergePotentials.size() >= 2)
-      MadeChange |= TryTailMergeBlocks(nullptr, nullptr);
+      MadeChange |= TryTailMergeBlocks(nullptr, nullptr, MinCommonTailLength);
   }
 
   // Look at blocks (IBB) with multiple predecessors (PBB).
@@ -1110,7 +1124,7 @@ bool BranchFolder::TailMergeBlocks(MachineFunction &MF) {
         TriedMerging.insert(MergePotentials[i].getBlock());
 
     if (MergePotentials.size() >= 2)
-      MadeChange |= TryTailMergeBlocks(IBB, PredBB);
+      MadeChange |= TryTailMergeBlocks(IBB, PredBB, MinCommonTailLength);
 
     // Reinsert an unconditional branch if needed. The 1 below can occur as a
     // result of removing blocks in TryTailMergeBlocks.

lib/CodeGen/BranchFolding.h

@@ -29,9 +29,13 @@ namespace llvm {
   public:
     class MBFIWrapper;
 
-    explicit BranchFolder(bool defaultEnableTailMerge, bool CommonHoist,
+    explicit BranchFolder(bool defaultEnableTailMerge,
+                          bool CommonHoist,
                           MBFIWrapper &MBFI,
-                          const MachineBranchProbabilityInfo &MBPI);
+                          const MachineBranchProbabilityInfo &MBPI,
+                          // Min tail length to merge. Defaults to commandline
+                          // flag. Ignored for optsize.
+                          unsigned MinCommonTailLength = 0);
 
     bool OptimizeFunction(MachineFunction &MF, const TargetInstrInfo *tii,
                           const TargetRegisterInfo *tri, MachineModuleInfo *mmi,
@@ -99,6 +103,7 @@ namespace llvm {
     bool EnableTailMerge;
     bool EnableHoistCommonCode;
     bool UpdateLiveIns;
+    unsigned MinCommonTailLength;
     const TargetInstrInfo *TII;
     const TargetRegisterInfo *TRI;
     MachineModuleInfo *MMI;
@@ -129,7 +134,8 @@ namespace llvm {
 
     bool TailMergeBlocks(MachineFunction &MF);
     bool TryTailMergeBlocks(MachineBasicBlock* SuccBB,
-                       MachineBasicBlock* PredBB);
+                       MachineBasicBlock* PredBB,
+                       unsigned MinCommonTailLength);
     void setCommonTailEdgeWeights(MachineBasicBlock &TailMBB);
     void computeLiveIns(MachineBasicBlock &MBB);
     void ReplaceTailWithBranchTo(MachineBasicBlock::iterator OldInst,

lib/CodeGen/MachineBlockPlacement.cpp

@@ -1747,8 +1747,10 @@ bool MachineBlockPlacement::runOnMachineFunction(MachineFunction &MF) {
                          BranchFoldPlacement;
   // No tail merging opportunities if the block number is less than four.
   if (MF.size() > 3 && EnableTailMerge) {
+    // Default to the standard tail-merge-size option.
+    unsigned TailMergeSize = 0;
     BranchFolder BF(/*EnableTailMerge=*/true, /*CommonHoist=*/false, *MBFI,
-                    *MBPI);
+                    *MBPI, TailMergeSize);
 
     if (BF.OptimizeFunction(MF, TII, MF.getSubtarget().getRegisterInfo(),
                             getAnalysisIfAvailable<MachineModuleInfo>(), MLI,