[llvm] [StaticDataLayout][PGO] Add profile format for static data layout, and the classes to operate on the profiles. (PR #138170)

[Mingming Liu via llvm-commits]

================
@@ -0,0 +1,246 @@
+#include "llvm/ProfileData/DataAccessProf.h"
+#include "llvm/ADT/DenseMapInfoVariant.h"
+#include "llvm/ADT/STLExtras.h"
+#include "llvm/ProfileData/InstrProf.h"
+#include "llvm/Support/Compression.h"
+#include "llvm/Support/Endian.h"
+#include "llvm/Support/Errc.h"
+#include "llvm/Support/Error.h"
+#include "llvm/Support/StringSaver.h"
+#include "llvm/Support/raw_ostream.h"
+#include <sys/types.h>
+
+namespace llvm {
+namespace data_access_prof {
+
+// If `Map` has an entry keyed by `Str`, returns the entry iterator. Otherwise,
+// creates an owned copy of `Str`, adds a map entry for it and returns the
+// iterator.
+static MapVector<StringRef, uint64_t>::iterator
+saveStringToMap(MapVector<StringRef, uint64_t> &Map,
+                llvm::UniqueStringSaver &saver, StringRef Str) {
+  auto [Iter, Inserted] = Map.try_emplace(saver.save(Str), Map.size());
+  return Iter;
+}
+
+const DataAccessProfRecord *
+DataAccessProfData::getProfileRecord(const SymbolID SymbolID) const {
+  auto Key = SymbolID;
+  if (std::holds_alternative<StringRef>(SymbolID))
+    Key = InstrProfSymtab::getCanonicalName(std::get<StringRef>(SymbolID));
+
+  auto It = Records.find(Key);
+  if (It != Records.end())
+    return &It->second;
+
+  return nullptr;
+}
+
+bool DataAccessProfData::isKnownColdSymbol(const SymbolID SymID) const {
+  if (std::holds_alternative<uint64_t>(SymID))
+    return KnownColdHashes.count(std::get<uint64_t>(SymID));
+  return KnownColdSymbols.count(std::get<StringRef>(SymID));
+}
+
+Error DataAccessProfData::addSymbolizedDataAccessProfile(SymbolID Symbol,
+                                                         uint64_t AccessCount) {
+  uint64_t RecordID = -1;
+  bool IsStringLiteral = false;
+  SymbolID Key;
+  if (std::holds_alternative<uint64_t>(Symbol)) {
+    RecordID = std::get<uint64_t>(Symbol);
+    Key = RecordID;
+    IsStringLiteral = true;
+  } else {
+    StringRef SymbolName = std::get<StringRef>(Symbol);
+    if (SymbolName.empty())
+      return make_error<StringError>("Empty symbol name",
+                                     llvm::errc::invalid_argument);
+
+    StringRef CanonicalName = InstrProfSymtab::getCanonicalName(SymbolName);
+    Key = CanonicalName;
+    RecordID = saveStringToMap(StrToIndexMap, saver, CanonicalName)->second;
+    IsStringLiteral = false;
+  }
+
+  auto [Iter, Inserted] = Records.try_emplace(
+      Key, DataAccessProfRecord{RecordID, AccessCount, IsStringLiteral});
+  if (!Inserted)
+    return make_error<StringError>("Duplicate symbol or string literal added. "
+                                   "User of DataAccessProfData should "
+                                   "aggregate count for the same symbol. ",
+                                   llvm::errc::invalid_argument);
+
+  return Error::success();
+}
+
+Error DataAccessProfData::addSymbolizedDataAccessProfile(
+    SymbolID SymbolID, uint64_t AccessCount,
+    const llvm::SmallVector<DataLocation> &Locations) {
+  if (Error E = addSymbolizedDataAccessProfile(SymbolID, AccessCount))
+    return E;
+
+  auto &Record = Records.back().second;
+  for (const auto &Location : Locations)
+    Record.Locations.push_back(
+        {saveStringToMap(StrToIndexMap, saver, Location.FileName)->first,
+         Location.Line});
+
+  return Error::success();
+}
+
+Error DataAccessProfData::addKnownSymbolWithoutSamples(SymbolID SymbolID) {
+  if (std::holds_alternative<uint64_t>(SymbolID)) {
+    KnownColdHashes.insert(std::get<uint64_t>(SymbolID));
+    return Error::success();
+  }
+  StringRef SymbolName = std::get<StringRef>(SymbolID);
+  if (SymbolName.empty())
+    return make_error<StringError>("Empty symbol name",
+                                   llvm::errc::invalid_argument);
+  StringRef CanonicalSymName = InstrProfSymtab::getCanonicalName(SymbolName);
+  KnownColdSymbols.insert(CanonicalSymName);
+  return Error::success();
+}
+
+Error DataAccessProfData::deserialize(const unsigned char *&Ptr) {
+  uint64_t NumSampledSymbols =
+      support::endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
+  uint64_t NumColdKnownSymbols =
+      support::endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
+  if (Error E = deserializeStrings(Ptr, NumSampledSymbols, NumColdKnownSymbols))
+    return E;
+
+  uint64_t Num =
+      support::endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
+  for (uint64_t I = 0; I < Num; ++I)
+    KnownColdHashes.insert(
+        support::endian::readNext<uint64_t, llvm::endianness::little>(Ptr));
+
+  return deserializeRecords(Ptr);
+}
+
+Error DataAccessProfData::serializeStrings(ProfOStream &OS) const {
----------------
mingmingl-llvm wrote:

sounds good to make the names more targeted. Renamed to `{serialize,deserialize}SymbolsAndFilenames`. 

https://github.com/llvm/llvm-project/pull/138170