[memprof] Speed up caller-callee pair extraction (llvm#116184)

We know that the MemProf profile has a lot of duplicate call stacks.
Extracting caller-callee pairs from a call stack we've seen before is
a wasteful effort.

This patch makes the extraction more efficient by first coming up with
a work list of linear call stack IDs -- the set of starting positions
in the radix tree array -- and then extract caller-callee pairs from
each call stack in the work list.

We implement the work list as a bit vector because we expect the work
list to be dense in the range [0, RadixTreeSize).  Also, we want the
set insertion to be cheap.

Without this patch, it takes 25 seconds to extract caller-callee pairs
from a large MemProf profile.  This patch shortenes that down to 4
seconds.

Author: kazutakahirata

llvm/include/llvm/ProfileData/InstrProfReader.h

@@ -683,6 +683,8 @@ class IndexedMemProfReader {
   const unsigned char *FrameBase = nullptr;
   /// The starting address of the call stack array.
   const unsigned char *CallStackBase = nullptr;
+  // The number of elements in the radix tree array.
+  unsigned RadixTreeSize = 0;
 
   Error deserializeV012(const unsigned char *Start, const unsigned char *Ptr,
                         uint64_t FirstWord);

llvm/lib/ProfileData/InstrProfReader.cpp

@@ -1303,6 +1303,12 @@ Error IndexedMemProfReader::deserializeV3(const unsigned char *Start,
   FrameBase = Ptr;
   CallStackBase = Start + CallStackPayloadOffset;
 
+  // Compute the number of elements in the radix tree array.  Since we use this
+  // to reserve enough bits in a BitVector, it's totally OK if we overestimate
+  // this number a little bit because of padding just before the next section.
+  RadixTreeSize = (RecordPayloadOffset - CallStackPayloadOffset) /
+                  sizeof(memprof::LinearFrameId);
+
   // Now initialize the table reader with a pointer into data buffer.
   MemProfRecordTable.reset(MemProfRecordHashTable::Create(
       /*Buckets=*/Start + RecordTableOffset,
@@ -1674,11 +1680,22 @@ IndexedMemProfReader::getMemProfCallerCalleePairs() const {
   memprof::LinearFrameIdConverter FrameIdConv(FrameBase);
   memprof::CallerCalleePairExtractor Extractor(CallStackBase, FrameIdConv);
 
+  // The set of linear call stack IDs that we need to traverse from.  We expect
+  // the set to be dense, so we use a BitVector.
+  BitVector Worklist(RadixTreeSize);
+
+  // Collect the set of linear call stack IDs.  Since we expect a lot of
+  // duplicates, we first collect them in the form of a bit vector before
+  // processing them.
   for (const memprof::IndexedMemProfRecord &IndexedRecord :
        MemProfRecordTable->data())
     for (const memprof::IndexedAllocationInfo &IndexedAI :
          IndexedRecord.AllocSites)
-      Extractor(IndexedAI.CSId);
+      Worklist.set(IndexedAI.CSId);
+
+  // Collect caller-callee pairs for each linear call stack ID in Worklist.
+  for (unsigned CS : Worklist.set_bits())
+    Extractor(CS);
 
   DenseMap<uint64_t, SmallVector<memprof::CallEdgeTy, 0>> Pairs =
       std::move(Extractor.CallerCalleePairs);

llvm/lib/ProfileData/InstrProfWriter.cpp

@@ -601,7 +601,8 @@ writeMemProfCallStackArray(
         &MemProfCallStackData,
     llvm::DenseMap<memprof::FrameId, memprof::LinearFrameId>
         &MemProfFrameIndexes,
-    llvm::DenseMap<memprof::FrameId, memprof::FrameStat> &FrameHistogram) {
+    llvm::DenseMap<memprof::FrameId, memprof::FrameStat> &FrameHistogram,
+    unsigned &NumElements) {
   llvm::DenseMap<memprof::CallStackId, memprof::LinearCallStackId>
       MemProfCallStackIndexes;
 
@@ -610,6 +611,7 @@ writeMemProfCallStackArray(
                 FrameHistogram);
   for (auto I : Builder.getRadixArray())
     OS.write32(I);
+  NumElements = Builder.getRadixArray().size();
   MemProfCallStackIndexes = Builder.takeCallStackPos();
 
   // Release the memory of this vector as it is no longer needed.
@@ -771,15 +773,26 @@ static Error writeMemProfV3(ProfOStream &OS,
       writeMemProfFrameArray(OS, MemProfData.Frames, FrameHistogram);
 
   uint64_t CallStackPayloadOffset = OS.tell();
+  // The number of elements in the call stack array.
+  unsigned NumElements = 0;
   llvm::DenseMap<memprof::CallStackId, memprof::LinearCallStackId>
-      MemProfCallStackIndexes = writeMemProfCallStackArray(
-          OS, MemProfData.CallStacks, MemProfFrameIndexes, FrameHistogram);
+      MemProfCallStackIndexes =
+          writeMemProfCallStackArray(OS, MemProfData.CallStacks,
+                                     MemProfFrameIndexes, FrameHistogram,
+                                     NumElements);
 
   uint64_t RecordPayloadOffset = OS.tell();
   uint64_t RecordTableOffset =
       writeMemProfRecords(OS, MemProfData.Records, &Schema, memprof::Version3,
                           &MemProfCallStackIndexes);
 
+  // IndexedMemProfReader::deserializeV3 computes the number of elements in the
+  // call stack array from the difference between CallStackPayloadOffset and
+  // RecordPayloadOffset.  Verify that the computation works.
+  assert(CallStackPayloadOffset +
+             NumElements * sizeof(memprof::LinearFrameId) ==
+         RecordPayloadOffset);
+
   uint64_t Header[] = {
       CallStackPayloadOffset,
       RecordPayloadOffset,