xref: /freebsd/contrib/llvm-project/llvm/lib/ProfileData/InstrProfReader.cpp (revision 85868e8a1daeaae7a0e48effb2ea2310ae3b02c6) 
1 //===- InstrProfReader.cpp - Instrumented profiling reader ----------------===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8 //
9 // This file contains support for reading profiling data for clang's
10 // instrumentation based PGO and coverage.
11 //
12 //===----------------------------------------------------------------------===//
13 
14 #include "llvm/ProfileData/InstrProfReader.h"
15 #include "llvm/ADT/ArrayRef.h"
16 #include "llvm/ADT/DenseMap.h"
17 #include "llvm/ADT/STLExtras.h"
18 #include "llvm/ADT/StringRef.h"
19 #include "llvm/IR/ProfileSummary.h"
20 #include "llvm/ProfileData/InstrProf.h"
21 #include "llvm/ProfileData/ProfileCommon.h"
22 #include "llvm/Support/Endian.h"
23 #include "llvm/Support/Error.h"
24 #include "llvm/Support/ErrorOr.h"
25 #include "llvm/Support/MemoryBuffer.h"
26 #include "llvm/Support/SymbolRemappingReader.h"
27 #include "llvm/Support/SwapByteOrder.h"
28 #include <algorithm>
29 #include <cctype>
30 #include <cstddef>
31 #include <cstdint>
32 #include <limits>
33 #include <memory>
34 #include <system_error>
35 #include <utility>
36 #include <vector>
37 
38 using namespace llvm;
39 
40 static Expected<std::unique_ptr<MemoryBuffer>>
41 setupMemoryBuffer(const Twine &Path) {
42   ErrorOr<std::unique_ptr<MemoryBuffer>> BufferOrErr =
43       MemoryBuffer::getFileOrSTDIN(Path);
44   if (std::error_code EC = BufferOrErr.getError())
45     return errorCodeToError(EC);
46   return std::move(BufferOrErr.get());
47 }
48 
49 static Error initializeReader(InstrProfReader &Reader) {
50   return Reader.readHeader();
51 }
52 
53 Expected<std::unique_ptr<InstrProfReader>>
54 InstrProfReader::create(const Twine &Path) {
55   // Set up the buffer to read.
56   auto BufferOrError = setupMemoryBuffer(Path);
57   if (Error E = BufferOrError.takeError())
58     return std::move(E);
59   return InstrProfReader::create(std::move(BufferOrError.get()));
60 }
61 
62 Expected<std::unique_ptr<InstrProfReader>>
63 InstrProfReader::create(std::unique_ptr<MemoryBuffer> Buffer) {
64   // Sanity check the buffer.
65   if (uint64_t(Buffer->getBufferSize()) > std::numeric_limits<uint64_t>::max())
66     return make_error<InstrProfError>(instrprof_error::too_large);
67 
68   if (Buffer->getBufferSize() == 0)
69     return make_error<InstrProfError>(instrprof_error::empty_raw_profile);
70 
71   std::unique_ptr<InstrProfReader> Result;
72   // Create the reader.
73   if (IndexedInstrProfReader::hasFormat(*Buffer))
74     Result.reset(new IndexedInstrProfReader(std::move(Buffer)));
75   else if (RawInstrProfReader64::hasFormat(*Buffer))
76     Result.reset(new RawInstrProfReader64(std::move(Buffer)));
77   else if (RawInstrProfReader32::hasFormat(*Buffer))
78     Result.reset(new RawInstrProfReader32(std::move(Buffer)));
79   else if (TextInstrProfReader::hasFormat(*Buffer))
80     Result.reset(new TextInstrProfReader(std::move(Buffer)));
81   else
82     return make_error<InstrProfError>(instrprof_error::unrecognized_format);
83 
84   // Initialize the reader and return the result.
85   if (Error E = initializeReader(*Result))
86     return std::move(E);
87 
88   return std::move(Result);
89 }
90 
91 Expected<std::unique_ptr<IndexedInstrProfReader>>
92 IndexedInstrProfReader::create(const Twine &Path, const Twine &RemappingPath) {
93   // Set up the buffer to read.
94   auto BufferOrError = setupMemoryBuffer(Path);
95   if (Error E = BufferOrError.takeError())
96     return std::move(E);
97 
98   // Set up the remapping buffer if requested.
99   std::unique_ptr<MemoryBuffer> RemappingBuffer;
100   std::string RemappingPathStr = RemappingPath.str();
101   if (!RemappingPathStr.empty()) {
102     auto RemappingBufferOrError = setupMemoryBuffer(RemappingPathStr);
103     if (Error E = RemappingBufferOrError.takeError())
104       return std::move(E);
105     RemappingBuffer = std::move(RemappingBufferOrError.get());
106   }
107 
108   return IndexedInstrProfReader::create(std::move(BufferOrError.get()),
109                                         std::move(RemappingBuffer));
110 }
111 
112 Expected<std::unique_ptr<IndexedInstrProfReader>>
113 IndexedInstrProfReader::create(std::unique_ptr<MemoryBuffer> Buffer,
114                                std::unique_ptr<MemoryBuffer> RemappingBuffer) {
115   // Sanity check the buffer.
116   if (uint64_t(Buffer->getBufferSize()) > std::numeric_limits<uint64_t>::max())
117     return make_error<InstrProfError>(instrprof_error::too_large);
118 
119   // Create the reader.
120   if (!IndexedInstrProfReader::hasFormat(*Buffer))
121     return make_error<InstrProfError>(instrprof_error::bad_magic);
122   auto Result = std::make_unique<IndexedInstrProfReader>(
123       std::move(Buffer), std::move(RemappingBuffer));
124 
125   // Initialize the reader and return the result.
126   if (Error E = initializeReader(*Result))
127     return std::move(E);
128 
129   return std::move(Result);
130 }
131 
132 void InstrProfIterator::Increment() {
133   if (auto E = Reader->readNextRecord(Record)) {
134     // Handle errors in the reader.
135     InstrProfError::take(std::move(E));
136     *this = InstrProfIterator();
137   }
138 }
139 
140 bool TextInstrProfReader::hasFormat(const MemoryBuffer &Buffer) {
141   // Verify that this really looks like plain ASCII text by checking a
142   // 'reasonable' number of characters (up to profile magic size).
143   size_t count = std::min(Buffer.getBufferSize(), sizeof(uint64_t));
144   StringRef buffer = Buffer.getBufferStart();
145   return count == 0 ||
146          std::all_of(buffer.begin(), buffer.begin() + count,
147                      [](char c) { return isPrint(c) || ::isspace(c); });
148 }
149 
150 // Read the profile variant flag from the header: ":FE" means this is a FE
151 // generated profile. ":IR" means this is an IR level profile. Other strings
152 // with a leading ':' will be reported an error format.
153 Error TextInstrProfReader::readHeader() {
154   Symtab.reset(new InstrProfSymtab());
155   bool IsIRInstr = false;
156   if (!Line->startswith(":")) {
157     IsIRLevelProfile = false;
158     return success();
159   }
160   StringRef Str = (Line)->substr(1);
161   if (Str.equals_lower("ir"))
162     IsIRInstr = true;
163   else if (Str.equals_lower("fe"))
164     IsIRInstr = false;
165   else if (Str.equals_lower("csir")) {
166     IsIRInstr = true;
167     HasCSIRLevelProfile = true;
168   } else
169     return error(instrprof_error::bad_header);
170 
171   ++Line;
172   IsIRLevelProfile = IsIRInstr;
173   return success();
174 }
175 
176 Error
177 TextInstrProfReader::readValueProfileData(InstrProfRecord &Record) {
178 
179 #define CHECK_LINE_END(Line)                                                   \
180   if (Line.is_at_end())                                                        \
181     return error(instrprof_error::truncated);
182 #define READ_NUM(Str, Dst)                                                     \
183   if ((Str).getAsInteger(10, (Dst)))                                           \
184     return error(instrprof_error::malformed);
185 #define VP_READ_ADVANCE(Val)                                                   \
186   CHECK_LINE_END(Line);                                                        \
187   uint32_t Val;                                                                \
188   READ_NUM((*Line), (Val));                                                    \
189   Line++;
190 
191   if (Line.is_at_end())
192     return success();
193 
194   uint32_t NumValueKinds;
195   if (Line->getAsInteger(10, NumValueKinds)) {
196     // No value profile data
197     return success();
198   }
199   if (NumValueKinds == 0 || NumValueKinds > IPVK_Last + 1)
200     return error(instrprof_error::malformed);
201   Line++;
202 
203   for (uint32_t VK = 0; VK < NumValueKinds; VK++) {
204     VP_READ_ADVANCE(ValueKind);
205     if (ValueKind > IPVK_Last)
206       return error(instrprof_error::malformed);
207     VP_READ_ADVANCE(NumValueSites);
208     if (!NumValueSites)
209       continue;
210 
211     Record.reserveSites(VK, NumValueSites);
212     for (uint32_t S = 0; S < NumValueSites; S++) {
213       VP_READ_ADVANCE(NumValueData);
214 
215       std::vector<InstrProfValueData> CurrentValues;
216       for (uint32_t V = 0; V < NumValueData; V++) {
217         CHECK_LINE_END(Line);
218         std::pair<StringRef, StringRef> VD = Line->rsplit(':');
219         uint64_t TakenCount, Value;
220         if (ValueKind == IPVK_IndirectCallTarget) {
221           if (InstrProfSymtab::isExternalSymbol(VD.first)) {
222             Value = 0;
223           } else {
224             if (Error E = Symtab->addFuncName(VD.first))
225               return E;
226             Value = IndexedInstrProf::ComputeHash(VD.first);
227           }
228         } else {
229           READ_NUM(VD.first, Value);
230         }
231         READ_NUM(VD.second, TakenCount);
232         CurrentValues.push_back({Value, TakenCount});
233         Line++;
234       }
235       Record.addValueData(ValueKind, S, CurrentValues.data(), NumValueData,
236                           nullptr);
237     }
238   }
239   return success();
240 
241 #undef CHECK_LINE_END
242 #undef READ_NUM
243 #undef VP_READ_ADVANCE
244 }
245 
246 Error TextInstrProfReader::readNextRecord(NamedInstrProfRecord &Record) {
247   // Skip empty lines and comments.
248   while (!Line.is_at_end() && (Line->empty() || Line->startswith("#")))
249     ++Line;
250   // If we hit EOF while looking for a name, we're done.
251   if (Line.is_at_end()) {
252     return error(instrprof_error::eof);
253   }
254 
255   // Read the function name.
256   Record.Name = *Line++;
257   if (Error E = Symtab->addFuncName(Record.Name))
258     return error(std::move(E));
259 
260   // Read the function hash.
261   if (Line.is_at_end())
262     return error(instrprof_error::truncated);
263   if ((Line++)->getAsInteger(0, Record.Hash))
264     return error(instrprof_error::malformed);
265 
266   // Read the number of counters.
267   uint64_t NumCounters;
268   if (Line.is_at_end())
269     return error(instrprof_error::truncated);
270   if ((Line++)->getAsInteger(10, NumCounters))
271     return error(instrprof_error::malformed);
272   if (NumCounters == 0)
273     return error(instrprof_error::malformed);
274 
275   // Read each counter and fill our internal storage with the values.
276   Record.Clear();
277   Record.Counts.reserve(NumCounters);
278   for (uint64_t I = 0; I < NumCounters; ++I) {
279     if (Line.is_at_end())
280       return error(instrprof_error::truncated);
281     uint64_t Count;
282     if ((Line++)->getAsInteger(10, Count))
283       return error(instrprof_error::malformed);
284     Record.Counts.push_back(Count);
285   }
286 
287   // Check if value profile data exists and read it if so.
288   if (Error E = readValueProfileData(Record))
289     return error(std::move(E));
290 
291   return success();
292 }
293 
294 template <class IntPtrT>
295 bool RawInstrProfReader<IntPtrT>::hasFormat(const MemoryBuffer &DataBuffer) {
296   if (DataBuffer.getBufferSize() < sizeof(uint64_t))
297     return false;
298   uint64_t Magic =
299     *reinterpret_cast<const uint64_t *>(DataBuffer.getBufferStart());
300   return RawInstrProf::getMagic<IntPtrT>() == Magic ||
301          sys::getSwappedBytes(RawInstrProf::getMagic<IntPtrT>()) == Magic;
302 }
303 
304 template <class IntPtrT>
305 Error RawInstrProfReader<IntPtrT>::readHeader() {
306   if (!hasFormat(*DataBuffer))
307     return error(instrprof_error::bad_magic);
308   if (DataBuffer->getBufferSize() < sizeof(RawInstrProf::Header))
309     return error(instrprof_error::bad_header);
310   auto *Header = reinterpret_cast<const RawInstrProf::Header *>(
311       DataBuffer->getBufferStart());
312   ShouldSwapBytes = Header->Magic != RawInstrProf::getMagic<IntPtrT>();
313   return readHeader(*Header);
314 }
315 
316 template <class IntPtrT>
317 Error RawInstrProfReader<IntPtrT>::readNextHeader(const char *CurrentPos) {
318   const char *End = DataBuffer->getBufferEnd();
319   // Skip zero padding between profiles.
320   while (CurrentPos != End && *CurrentPos == 0)
321     ++CurrentPos;
322   // If there's nothing left, we're done.
323   if (CurrentPos == End)
324     return make_error<InstrProfError>(instrprof_error::eof);
325   // If there isn't enough space for another header, this is probably just
326   // garbage at the end of the file.
327   if (CurrentPos + sizeof(RawInstrProf::Header) > End)
328     return make_error<InstrProfError>(instrprof_error::malformed);
329   // The writer ensures each profile is padded to start at an aligned address.
330   if (reinterpret_cast<size_t>(CurrentPos) % alignof(uint64_t))
331     return make_error<InstrProfError>(instrprof_error::malformed);
332   // The magic should have the same byte order as in the previous header.
333   uint64_t Magic = *reinterpret_cast<const uint64_t *>(CurrentPos);
334   if (Magic != swap(RawInstrProf::getMagic<IntPtrT>()))
335     return make_error<InstrProfError>(instrprof_error::bad_magic);
336 
337   // There's another profile to read, so we need to process the header.
338   auto *Header = reinterpret_cast<const RawInstrProf::Header *>(CurrentPos);
339   return readHeader(*Header);
340 }
341 
342 template <class IntPtrT>
343 Error RawInstrProfReader<IntPtrT>::createSymtab(InstrProfSymtab &Symtab) {
344   if (Error E = Symtab.create(StringRef(NamesStart, NamesSize)))
345     return error(std::move(E));
346   for (const RawInstrProf::ProfileData<IntPtrT> *I = Data; I != DataEnd; ++I) {
347     const IntPtrT FPtr = swap(I->FunctionPointer);
348     if (!FPtr)
349       continue;
350     Symtab.mapAddress(FPtr, I->NameRef);
351   }
352   return success();
353 }
354 
355 template <class IntPtrT>
356 Error RawInstrProfReader<IntPtrT>::readHeader(
357     const RawInstrProf::Header &Header) {
358   Version = swap(Header.Version);
359   if (GET_VERSION(Version) != RawInstrProf::Version)
360     return error(instrprof_error::unsupported_version);
361 
362   CountersDelta = swap(Header.CountersDelta);
363   NamesDelta = swap(Header.NamesDelta);
364   auto DataSize = swap(Header.DataSize);
365   auto CountersSize = swap(Header.CountersSize);
366   NamesSize = swap(Header.NamesSize);
367   ValueKindLast = swap(Header.ValueKindLast);
368 
369   auto DataSizeInBytes = DataSize * sizeof(RawInstrProf::ProfileData<IntPtrT>);
370   auto PaddingSize = getNumPaddingBytes(NamesSize);
371 
372   ptrdiff_t DataOffset = sizeof(RawInstrProf::Header);
373   ptrdiff_t CountersOffset = DataOffset + DataSizeInBytes;
374   ptrdiff_t NamesOffset = CountersOffset + sizeof(uint64_t) * CountersSize;
375   ptrdiff_t ValueDataOffset = NamesOffset + NamesSize + PaddingSize;
376 
377   auto *Start = reinterpret_cast<const char *>(&Header);
378   if (Start + ValueDataOffset > DataBuffer->getBufferEnd())
379     return error(instrprof_error::bad_header);
380 
381   Data = reinterpret_cast<const RawInstrProf::ProfileData<IntPtrT> *>(
382       Start + DataOffset);
383   DataEnd = Data + DataSize;
384   CountersStart = reinterpret_cast<const uint64_t *>(Start + CountersOffset);
385   NamesStart = Start + NamesOffset;
386   ValueDataStart = reinterpret_cast<const uint8_t *>(Start + ValueDataOffset);
387 
388   std::unique_ptr<InstrProfSymtab> NewSymtab = std::make_unique<InstrProfSymtab>();
389   if (Error E = createSymtab(*NewSymtab.get()))
390     return E;
391 
392   Symtab = std::move(NewSymtab);
393   return success();
394 }
395 
396 template <class IntPtrT>
397 Error RawInstrProfReader<IntPtrT>::readName(NamedInstrProfRecord &Record) {
398   Record.Name = getName(Data->NameRef);
399   return success();
400 }
401 
402 template <class IntPtrT>
403 Error RawInstrProfReader<IntPtrT>::readFuncHash(NamedInstrProfRecord &Record) {
404   Record.Hash = swap(Data->FuncHash);
405   return success();
406 }
407 
408 template <class IntPtrT>
409 Error RawInstrProfReader<IntPtrT>::readRawCounts(
410     InstrProfRecord &Record) {
411   uint32_t NumCounters = swap(Data->NumCounters);
412   IntPtrT CounterPtr = Data->CounterPtr;
413   if (NumCounters == 0)
414     return error(instrprof_error::malformed);
415 
416   auto *NamesStartAsCounter = reinterpret_cast<const uint64_t *>(NamesStart);
417   ptrdiff_t MaxNumCounters = NamesStartAsCounter - CountersStart;
418 
419   // Check bounds. Note that the counter pointer embedded in the data record
420   // may itself be corrupt.
421   if (NumCounters > MaxNumCounters)
422     return error(instrprof_error::malformed);
423   ptrdiff_t CounterOffset = getCounterOffset(CounterPtr);
424   if (CounterOffset < 0 || CounterOffset > MaxNumCounters ||
425       (CounterOffset + NumCounters) > MaxNumCounters)
426     return error(instrprof_error::malformed);
427 
428   auto RawCounts = makeArrayRef(getCounter(CounterOffset), NumCounters);
429 
430   if (ShouldSwapBytes) {
431     Record.Counts.clear();
432     Record.Counts.reserve(RawCounts.size());
433     for (uint64_t Count : RawCounts)
434       Record.Counts.push_back(swap(Count));
435   } else
436     Record.Counts = RawCounts;
437 
438   return success();
439 }
440 
441 template <class IntPtrT>
442 Error RawInstrProfReader<IntPtrT>::readValueProfilingData(
443     InstrProfRecord &Record) {
444   Record.clearValueData();
445   CurValueDataSize = 0;
446   // Need to match the logic in value profile dumper code in compiler-rt:
447   uint32_t NumValueKinds = 0;
448   for (uint32_t I = 0; I < IPVK_Last + 1; I++)
449     NumValueKinds += (Data->NumValueSites[I] != 0);
450 
451   if (!NumValueKinds)
452     return success();
453 
454   Expected<std::unique_ptr<ValueProfData>> VDataPtrOrErr =
455       ValueProfData::getValueProfData(
456           ValueDataStart, (const unsigned char *)DataBuffer->getBufferEnd(),
457           getDataEndianness());
458 
459   if (Error E = VDataPtrOrErr.takeError())
460     return E;
461 
462   // Note that besides deserialization, this also performs the conversion for
463   // indirect call targets.  The function pointers from the raw profile are
464   // remapped into function name hashes.
465   VDataPtrOrErr.get()->deserializeTo(Record, Symtab.get());
466   CurValueDataSize = VDataPtrOrErr.get()->getSize();
467   return success();
468 }
469 
470 template <class IntPtrT>
471 Error RawInstrProfReader<IntPtrT>::readNextRecord(NamedInstrProfRecord &Record) {
472   if (atEnd())
473     // At this point, ValueDataStart field points to the next header.
474     if (Error E = readNextHeader(getNextHeaderPos()))
475       return error(std::move(E));
476 
477   // Read name ad set it in Record.
478   if (Error E = readName(Record))
479     return error(std::move(E));
480 
481   // Read FuncHash and set it in Record.
482   if (Error E = readFuncHash(Record))
483     return error(std::move(E));
484 
485   // Read raw counts and set Record.
486   if (Error E = readRawCounts(Record))
487     return error(std::move(E));
488 
489   // Read value data and set Record.
490   if (Error E = readValueProfilingData(Record))
491     return error(std::move(E));
492 
493   // Iterate.
494   advanceData();
495   return success();
496 }
497 
498 namespace llvm {
499 
500 template class RawInstrProfReader<uint32_t>;
501 template class RawInstrProfReader<uint64_t>;
502 
503 } // end namespace llvm
504 
505 InstrProfLookupTrait::hash_value_type
506 InstrProfLookupTrait::ComputeHash(StringRef K) {
507   return IndexedInstrProf::ComputeHash(HashType, K);
508 }
509 
510 using data_type = InstrProfLookupTrait::data_type;
511 using offset_type = InstrProfLookupTrait::offset_type;
512 
513 bool InstrProfLookupTrait::readValueProfilingData(
514     const unsigned char *&D, const unsigned char *const End) {
515   Expected<std::unique_ptr<ValueProfData>> VDataPtrOrErr =
516       ValueProfData::getValueProfData(D, End, ValueProfDataEndianness);
517 
518   if (VDataPtrOrErr.takeError())
519     return false;
520 
521   VDataPtrOrErr.get()->deserializeTo(DataBuffer.back(), nullptr);
522   D += VDataPtrOrErr.get()->TotalSize;
523 
524   return true;
525 }
526 
527 data_type InstrProfLookupTrait::ReadData(StringRef K, const unsigned char *D,
528                                          offset_type N) {
529   using namespace support;
530 
531   // Check if the data is corrupt. If so, don't try to read it.
532   if (N % sizeof(uint64_t))
533     return data_type();
534 
535   DataBuffer.clear();
536   std::vector<uint64_t> CounterBuffer;
537 
538   const unsigned char *End = D + N;
539   while (D < End) {
540     // Read hash.
541     if (D + sizeof(uint64_t) >= End)
542       return data_type();
543     uint64_t Hash = endian::readNext<uint64_t, little, unaligned>(D);
544 
545     // Initialize number of counters for GET_VERSION(FormatVersion) == 1.
546     uint64_t CountsSize = N / sizeof(uint64_t) - 1;
547     // If format version is different then read the number of counters.
548     if (GET_VERSION(FormatVersion) != IndexedInstrProf::ProfVersion::Version1) {
549       if (D + sizeof(uint64_t) > End)
550         return data_type();
551       CountsSize = endian::readNext<uint64_t, little, unaligned>(D);
552     }
553     // Read counter values.
554     if (D + CountsSize * sizeof(uint64_t) > End)
555       return data_type();
556 
557     CounterBuffer.clear();
558     CounterBuffer.reserve(CountsSize);
559     for (uint64_t J = 0; J < CountsSize; ++J)
560       CounterBuffer.push_back(endian::readNext<uint64_t, little, unaligned>(D));
561 
562     DataBuffer.emplace_back(K, Hash, std::move(CounterBuffer));
563 
564     // Read value profiling data.
565     if (GET_VERSION(FormatVersion) > IndexedInstrProf::ProfVersion::Version2 &&
566         !readValueProfilingData(D, End)) {
567       DataBuffer.clear();
568       return data_type();
569     }
570   }
571   return DataBuffer;
572 }
573 
574 template <typename HashTableImpl>
575 Error InstrProfReaderIndex<HashTableImpl>::getRecords(
576     StringRef FuncName, ArrayRef<NamedInstrProfRecord> &Data) {
577   auto Iter = HashTable->find(FuncName);
578   if (Iter == HashTable->end())
579     return make_error<InstrProfError>(instrprof_error::unknown_function);
580 
581   Data = (*Iter);
582   if (Data.empty())
583     return make_error<InstrProfError>(instrprof_error::malformed);
584 
585   return Error::success();
586 }
587 
588 template <typename HashTableImpl>
589 Error InstrProfReaderIndex<HashTableImpl>::getRecords(
590     ArrayRef<NamedInstrProfRecord> &Data) {
591   if (atEnd())
592     return make_error<InstrProfError>(instrprof_error::eof);
593 
594   Data = *RecordIterator;
595 
596   if (Data.empty())
597     return make_error<InstrProfError>(instrprof_error::malformed);
598 
599   return Error::success();
600 }
601 
602 template <typename HashTableImpl>
603 InstrProfReaderIndex<HashTableImpl>::InstrProfReaderIndex(
604     const unsigned char *Buckets, const unsigned char *const Payload,
605     const unsigned char *const Base, IndexedInstrProf::HashT HashType,
606     uint64_t Version) {
607   FormatVersion = Version;
608   HashTable.reset(HashTableImpl::Create(
609       Buckets, Payload, Base,
610       typename HashTableImpl::InfoType(HashType, Version)));
611   RecordIterator = HashTable->data_begin();
612 }
613 
614 namespace {
615 /// A remapper that does not apply any remappings.
616 class InstrProfReaderNullRemapper : public InstrProfReaderRemapper {
617   InstrProfReaderIndexBase &Underlying;
618 
619 public:
620   InstrProfReaderNullRemapper(InstrProfReaderIndexBase &Underlying)
621       : Underlying(Underlying) {}
622 
623   Error getRecords(StringRef FuncName,
624                    ArrayRef<NamedInstrProfRecord> &Data) override {
625     return Underlying.getRecords(FuncName, Data);
626   }
627 };
628 }
629 
630 /// A remapper that applies remappings based on a symbol remapping file.
631 template <typename HashTableImpl>
632 class llvm::InstrProfReaderItaniumRemapper
633     : public InstrProfReaderRemapper {
634 public:
635   InstrProfReaderItaniumRemapper(
636       std::unique_ptr<MemoryBuffer> RemapBuffer,
637       InstrProfReaderIndex<HashTableImpl> &Underlying)
638       : RemapBuffer(std::move(RemapBuffer)), Underlying(Underlying) {
639   }
640 
641   /// Extract the original function name from a PGO function name.
642   static StringRef extractName(StringRef Name) {
643     // We can have multiple :-separated pieces; there can be pieces both
644     // before and after the mangled name. Find the first part that starts
645     // with '_Z'; we'll assume that's the mangled name we want.
646     std::pair<StringRef, StringRef> Parts = {StringRef(), Name};
647     while (true) {
648       Parts = Parts.second.split(':');
649       if (Parts.first.startswith("_Z"))
650         return Parts.first;
651       if (Parts.second.empty())
652         return Name;
653     }
654   }
655 
656   /// Given a mangled name extracted from a PGO function name, and a new
657   /// form for that mangled name, reconstitute the name.
658   static void reconstituteName(StringRef OrigName, StringRef ExtractedName,
659                                StringRef Replacement,
660                                SmallVectorImpl<char> &Out) {
661     Out.reserve(OrigName.size() + Replacement.size() - ExtractedName.size());
662     Out.insert(Out.end(), OrigName.begin(), ExtractedName.begin());
663     Out.insert(Out.end(), Replacement.begin(), Replacement.end());
664     Out.insert(Out.end(), ExtractedName.end(), OrigName.end());
665   }
666 
667   Error populateRemappings() override {
668     if (Error E = Remappings.read(*RemapBuffer))
669       return E;
670     for (StringRef Name : Underlying.HashTable->keys()) {
671       StringRef RealName = extractName(Name);
672       if (auto Key = Remappings.insert(RealName)) {
673         // FIXME: We could theoretically map the same equivalence class to
674         // multiple names in the profile data. If that happens, we should
675         // return NamedInstrProfRecords from all of them.
676         MappedNames.insert({Key, RealName});
677       }
678     }
679     return Error::success();
680   }
681 
682   Error getRecords(StringRef FuncName,
683                    ArrayRef<NamedInstrProfRecord> &Data) override {
684     StringRef RealName = extractName(FuncName);
685     if (auto Key = Remappings.lookup(RealName)) {
686       StringRef Remapped = MappedNames.lookup(Key);
687       if (!Remapped.empty()) {
688         if (RealName.begin() == FuncName.begin() &&
689             RealName.end() == FuncName.end())
690           FuncName = Remapped;
691         else {
692           // Try rebuilding the name from the given remapping.
693           SmallString<256> Reconstituted;
694           reconstituteName(FuncName, RealName, Remapped, Reconstituted);
695           Error E = Underlying.getRecords(Reconstituted, Data);
696           if (!E)
697             return E;
698 
699           // If we failed because the name doesn't exist, fall back to asking
700           // about the original name.
701           if (Error Unhandled = handleErrors(
702                   std::move(E), [](std::unique_ptr<InstrProfError> Err) {
703                     return Err->get() == instrprof_error::unknown_function
704                                ? Error::success()
705                                : Error(std::move(Err));
706                   }))
707             return Unhandled;
708         }
709       }
710     }
711     return Underlying.getRecords(FuncName, Data);
712   }
713 
714 private:
715   /// The memory buffer containing the remapping configuration. Remappings
716   /// holds pointers into this buffer.
717   std::unique_ptr<MemoryBuffer> RemapBuffer;
718 
719   /// The mangling remapper.
720   SymbolRemappingReader Remappings;
721 
722   /// Mapping from mangled name keys to the name used for the key in the
723   /// profile data.
724   /// FIXME: Can we store a location within the on-disk hash table instead of
725   /// redoing lookup?
726   DenseMap<SymbolRemappingReader::Key, StringRef> MappedNames;
727 
728   /// The real profile data reader.
729   InstrProfReaderIndex<HashTableImpl> &Underlying;
730 };
731 
732 bool IndexedInstrProfReader::hasFormat(const MemoryBuffer &DataBuffer) {
733   using namespace support;
734 
735   if (DataBuffer.getBufferSize() < 8)
736     return false;
737   uint64_t Magic =
738       endian::read<uint64_t, little, aligned>(DataBuffer.getBufferStart());
739   // Verify that it's magical.
740   return Magic == IndexedInstrProf::Magic;
741 }
742 
743 const unsigned char *
744 IndexedInstrProfReader::readSummary(IndexedInstrProf::ProfVersion Version,
745                                     const unsigned char *Cur, bool UseCS) {
746   using namespace IndexedInstrProf;
747   using namespace support;
748 
749   if (Version >= IndexedInstrProf::Version4) {
750     const IndexedInstrProf::Summary *SummaryInLE =
751         reinterpret_cast<const IndexedInstrProf::Summary *>(Cur);
752     uint64_t NFields =
753         endian::byte_swap<uint64_t, little>(SummaryInLE->NumSummaryFields);
754     uint64_t NEntries =
755         endian::byte_swap<uint64_t, little>(SummaryInLE->NumCutoffEntries);
756     uint32_t SummarySize =
757         IndexedInstrProf::Summary::getSize(NFields, NEntries);
758     std::unique_ptr<IndexedInstrProf::Summary> SummaryData =
759         IndexedInstrProf::allocSummary(SummarySize);
760 
761     const uint64_t *Src = reinterpret_cast<const uint64_t *>(SummaryInLE);
762     uint64_t *Dst = reinterpret_cast<uint64_t *>(SummaryData.get());
763     for (unsigned I = 0; I < SummarySize / sizeof(uint64_t); I++)
764       Dst[I] = endian::byte_swap<uint64_t, little>(Src[I]);
765 
766     SummaryEntryVector DetailedSummary;
767     for (unsigned I = 0; I < SummaryData->NumCutoffEntries; I++) {
768       const IndexedInstrProf::Summary::Entry &Ent = SummaryData->getEntry(I);
769       DetailedSummary.emplace_back((uint32_t)Ent.Cutoff, Ent.MinBlockCount,
770                                    Ent.NumBlocks);
771     }
772     std::unique_ptr<llvm::ProfileSummary> &Summary =
773         UseCS ? this->CS_Summary : this->Summary;
774 
775     // initialize InstrProfSummary using the SummaryData from disk.
776     Summary = std::make_unique<ProfileSummary>(
777         UseCS ? ProfileSummary::PSK_CSInstr : ProfileSummary::PSK_Instr,
778         DetailedSummary, SummaryData->get(Summary::TotalBlockCount),
779         SummaryData->get(Summary::MaxBlockCount),
780         SummaryData->get(Summary::MaxInternalBlockCount),
781         SummaryData->get(Summary::MaxFunctionCount),
782         SummaryData->get(Summary::TotalNumBlocks),
783         SummaryData->get(Summary::TotalNumFunctions));
784     return Cur + SummarySize;
785   } else {
786     // The older versions do not support a profile summary. This just computes
787     // an empty summary, which will not result in accurate hot/cold detection.
788     // We would need to call addRecord for all NamedInstrProfRecords to get the
789     // correct summary. However, this version is old (prior to early 2016) and
790     // has not been supporting an accurate summary for several years.
791     InstrProfSummaryBuilder Builder(ProfileSummaryBuilder::DefaultCutoffs);
792     Summary = Builder.getSummary();
793     return Cur;
794   }
795 }
796 
797 Error IndexedInstrProfReader::readHeader() {
798   using namespace support;
799 
800   const unsigned char *Start =
801       (const unsigned char *)DataBuffer->getBufferStart();
802   const unsigned char *Cur = Start;
803   if ((const unsigned char *)DataBuffer->getBufferEnd() - Cur < 24)
804     return error(instrprof_error::truncated);
805 
806   auto *Header = reinterpret_cast<const IndexedInstrProf::Header *>(Cur);
807   Cur += sizeof(IndexedInstrProf::Header);
808 
809   // Check the magic number.
810   uint64_t Magic = endian::byte_swap<uint64_t, little>(Header->Magic);
811   if (Magic != IndexedInstrProf::Magic)
812     return error(instrprof_error::bad_magic);
813 
814   // Read the version.
815   uint64_t FormatVersion = endian::byte_swap<uint64_t, little>(Header->Version);
816   if (GET_VERSION(FormatVersion) >
817       IndexedInstrProf::ProfVersion::CurrentVersion)
818     return error(instrprof_error::unsupported_version);
819 
820   Cur = readSummary((IndexedInstrProf::ProfVersion)FormatVersion, Cur,
821                     /* UseCS */ false);
822   if (FormatVersion & VARIANT_MASK_CSIR_PROF)
823     Cur = readSummary((IndexedInstrProf::ProfVersion)FormatVersion, Cur,
824                       /* UseCS */ true);
825 
826   // Read the hash type and start offset.
827   IndexedInstrProf::HashT HashType = static_cast<IndexedInstrProf::HashT>(
828       endian::byte_swap<uint64_t, little>(Header->HashType));
829   if (HashType > IndexedInstrProf::HashT::Last)
830     return error(instrprof_error::unsupported_hash_type);
831 
832   uint64_t HashOffset = endian::byte_swap<uint64_t, little>(Header->HashOffset);
833 
834   // The rest of the file is an on disk hash table.
835   auto IndexPtr =
836       std::make_unique<InstrProfReaderIndex<OnDiskHashTableImplV3>>(
837           Start + HashOffset, Cur, Start, HashType, FormatVersion);
838 
839   // Load the remapping table now if requested.
840   if (RemappingBuffer) {
841     Remapper = std::make_unique<
842         InstrProfReaderItaniumRemapper<OnDiskHashTableImplV3>>(
843         std::move(RemappingBuffer), *IndexPtr);
844     if (Error E = Remapper->populateRemappings())
845       return E;
846   } else {
847     Remapper = std::make_unique<InstrProfReaderNullRemapper>(*IndexPtr);
848   }
849   Index = std::move(IndexPtr);
850 
851   return success();
852 }
853 
854 InstrProfSymtab &IndexedInstrProfReader::getSymtab() {
855   if (Symtab.get())
856     return *Symtab.get();
857 
858   std::unique_ptr<InstrProfSymtab> NewSymtab = std::make_unique<InstrProfSymtab>();
859   if (Error E = Index->populateSymtab(*NewSymtab.get())) {
860     consumeError(error(InstrProfError::take(std::move(E))));
861   }
862 
863   Symtab = std::move(NewSymtab);
864   return *Symtab.get();
865 }
866 
867 Expected<InstrProfRecord>
868 IndexedInstrProfReader::getInstrProfRecord(StringRef FuncName,
869                                            uint64_t FuncHash) {
870   ArrayRef<NamedInstrProfRecord> Data;
871   Error Err = Remapper->getRecords(FuncName, Data);
872   if (Err)
873     return std::move(Err);
874   // Found it. Look for counters with the right hash.
875   for (unsigned I = 0, E = Data.size(); I < E; ++I) {
876     // Check for a match and fill the vector if there is one.
877     if (Data[I].Hash == FuncHash) {
878       return std::move(Data[I]);
879     }
880   }
881   return error(instrprof_error::hash_mismatch);
882 }
883 
884 Error IndexedInstrProfReader::getFunctionCounts(StringRef FuncName,
885                                                 uint64_t FuncHash,
886                                                 std::vector<uint64_t> &Counts) {
887   Expected<InstrProfRecord> Record = getInstrProfRecord(FuncName, FuncHash);
888   if (Error E = Record.takeError())
889     return error(std::move(E));
890 
891   Counts = Record.get().Counts;
892   return success();
893 }
894 
895 Error IndexedInstrProfReader::readNextRecord(NamedInstrProfRecord &Record) {
896   ArrayRef<NamedInstrProfRecord> Data;
897 
898   Error E = Index->getRecords(Data);
899   if (E)
900     return error(std::move(E));
901 
902   Record = Data[RecordIndex++];
903   if (RecordIndex >= Data.size()) {
904     Index->advanceToNextKey();
905     RecordIndex = 0;
906   }
907   return success();
908 }
909 
910 void InstrProfReader::accumulateCounts(CountSumOrPercent &Sum, bool IsCS) {
911   uint64_t NumFuncs = 0;
912   for (const auto &Func : *this) {
913     if (isIRLevelProfile()) {
914       bool FuncIsCS = NamedInstrProfRecord::hasCSFlagInHash(Func.Hash);
915       if (FuncIsCS != IsCS)
916         continue;
917     }
918     Func.accumulateCounts(Sum);
919     ++NumFuncs;
920   }
921   Sum.NumEntries = NumFuncs;
922 }
923