1 //===-- InstrProfCorrelator.cpp -------------------------------------------===//
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 #include "llvm/ProfileData/InstrProfCorrelator.h"
10 #include "llvm/Object/MachO.h"
11 #include "llvm/Support/Debug.h"
12 #include "llvm/Support/FileSystem.h"
13 #include "llvm/Support/Path.h"
14 
15 #define DEBUG_TYPE "correlator"
16 
17 using namespace llvm;
18 
19 /// Get the __llvm_prf_cnts section.
20 Expected<object::SectionRef> getCountersSection(const object::ObjectFile &Obj) {
21   for (auto &Section : Obj.sections())
22     if (auto SectionName = Section.getName())
23       if (SectionName.get() == INSTR_PROF_CNTS_SECT_NAME)
24         return Section;
25   return make_error<InstrProfError>(
26       instrprof_error::unable_to_correlate_profile);
27 }
28 
29 const char *InstrProfCorrelator::FunctionNameAttributeName = "Function Name";
30 const char *InstrProfCorrelator::CFGHashAttributeName = "CFG Hash";
31 const char *InstrProfCorrelator::NumCountersAttributeName = "Num Counters";
32 
33 llvm::Expected<std::unique_ptr<InstrProfCorrelator::Context>>
34 InstrProfCorrelator::Context::get(std::unique_ptr<MemoryBuffer> Buffer,
35                                   const object::ObjectFile &Obj) {
36   auto CountersSection = getCountersSection(Obj);
37   if (auto Err = CountersSection.takeError())
38     return std::move(Err);
39   auto C = std::make_unique<Context>();
40   C->Buffer = std::move(Buffer);
41   C->CountersSectionStart = CountersSection->getAddress();
42   C->CountersSectionEnd = C->CountersSectionStart + CountersSection->getSize();
43   C->ShouldSwapBytes = Obj.isLittleEndian() != sys::IsLittleEndianHost;
44   return Expected<std::unique_ptr<Context>>(std::move(C));
45 }
46 
47 llvm::Expected<std::unique_ptr<InstrProfCorrelator>>
48 InstrProfCorrelator::get(StringRef DebugInfoFilename) {
49   auto DsymObjectsOrErr =
50       object::MachOObjectFile::findDsymObjectMembers(DebugInfoFilename);
51   if (auto Err = DsymObjectsOrErr.takeError())
52     return std::move(Err);
53   if (!DsymObjectsOrErr->empty()) {
54     // TODO: Enable profile correlation when there are multiple objects in a
55     // dSYM bundle.
56     if (DsymObjectsOrErr->size() > 1)
57       return createStringError(
58           std::error_code(),
59           "Profile correlation using multiple objects is not yet supported");
60     DebugInfoFilename = *DsymObjectsOrErr->begin();
61   }
62   auto BufferOrErr =
63       errorOrToExpected(MemoryBuffer::getFile(DebugInfoFilename));
64   if (auto Err = BufferOrErr.takeError())
65     return std::move(Err);
66 
67   return get(std::move(*BufferOrErr));
68 }
69 
70 llvm::Expected<std::unique_ptr<InstrProfCorrelator>>
71 InstrProfCorrelator::get(std::unique_ptr<MemoryBuffer> Buffer) {
72   auto BinOrErr = object::createBinary(*Buffer);
73   if (auto Err = BinOrErr.takeError())
74     return std::move(Err);
75 
76   if (auto *Obj = dyn_cast<object::ObjectFile>(BinOrErr->get())) {
77     auto CtxOrErr = Context::get(std::move(Buffer), *Obj);
78     if (auto Err = CtxOrErr.takeError())
79       return std::move(Err);
80     auto T = Obj->makeTriple();
81     if (T.isArch64Bit())
82       return InstrProfCorrelatorImpl<uint64_t>::get(std::move(*CtxOrErr), *Obj);
83     if (T.isArch32Bit())
84       return InstrProfCorrelatorImpl<uint32_t>::get(std::move(*CtxOrErr), *Obj);
85   }
86   return make_error<InstrProfError>(
87       instrprof_error::unable_to_correlate_profile);
88 }
89 
90 template <>
91 InstrProfCorrelatorImpl<uint32_t>::InstrProfCorrelatorImpl(
92     std::unique_ptr<InstrProfCorrelator::Context> Ctx)
93     : InstrProfCorrelatorImpl(InstrProfCorrelatorKind::CK_32Bit,
94                               std::move(Ctx)) {}
95 template <>
96 InstrProfCorrelatorImpl<uint64_t>::InstrProfCorrelatorImpl(
97     std::unique_ptr<InstrProfCorrelator::Context> Ctx)
98     : InstrProfCorrelatorImpl(InstrProfCorrelatorKind::CK_64Bit,
99                               std::move(Ctx)) {}
100 template <>
101 bool InstrProfCorrelatorImpl<uint32_t>::classof(const InstrProfCorrelator *C) {
102   return C->getKind() == InstrProfCorrelatorKind::CK_32Bit;
103 }
104 template <>
105 bool InstrProfCorrelatorImpl<uint64_t>::classof(const InstrProfCorrelator *C) {
106   return C->getKind() == InstrProfCorrelatorKind::CK_64Bit;
107 }
108 
109 template <class IntPtrT>
110 llvm::Expected<std::unique_ptr<InstrProfCorrelatorImpl<IntPtrT>>>
111 InstrProfCorrelatorImpl<IntPtrT>::get(
112     std::unique_ptr<InstrProfCorrelator::Context> Ctx,
113     const object::ObjectFile &Obj) {
114   if (Obj.isELF() || Obj.isMachO()) {
115     auto DICtx = DWARFContext::create(Obj);
116     return std::make_unique<DwarfInstrProfCorrelator<IntPtrT>>(std::move(DICtx),
117                                                                std::move(Ctx));
118   }
119   return make_error<InstrProfError>(instrprof_error::unsupported_debug_format);
120 }
121 
122 template <class IntPtrT>
123 Error InstrProfCorrelatorImpl<IntPtrT>::correlateProfileData() {
124   assert(Data.empty() && CompressedNames.empty() && Names.empty());
125   correlateProfileDataImpl();
126   auto Result =
127       collectPGOFuncNameStrings(Names, /*doCompression=*/true, CompressedNames);
128   Names.clear();
129   return Result;
130 }
131 
132 template <class IntPtrT>
133 void InstrProfCorrelatorImpl<IntPtrT>::addProbe(StringRef FunctionName,
134                                                 uint64_t CFGHash,
135                                                 IntPtrT CounterOffset,
136                                                 IntPtrT FunctionPtr,
137                                                 uint32_t NumCounters) {
138   Data.push_back({
139       maybeSwap<uint64_t>(IndexedInstrProf::ComputeHash(FunctionName)),
140       maybeSwap<uint64_t>(CFGHash),
141       // In this mode, CounterPtr actually stores the section relative address
142       // of the counter.
143       maybeSwap<IntPtrT>(CounterOffset),
144       maybeSwap<IntPtrT>(FunctionPtr),
145       // TODO: Value profiling is not yet supported.
146       /*ValuesPtr=*/maybeSwap<IntPtrT>(0),
147       maybeSwap<uint32_t>(NumCounters),
148       /*NumValueSites=*/{maybeSwap<uint16_t>(0), maybeSwap<uint16_t>(0)},
149   });
150   Names.push_back(FunctionName.str());
151 }
152 
153 template <class IntPtrT>
154 llvm::Optional<uint64_t>
155 DwarfInstrProfCorrelator<IntPtrT>::getLocation(const DWARFDie &Die) const {
156   auto Locations = Die.getLocations(dwarf::DW_AT_location);
157   if (!Locations) {
158     consumeError(Locations.takeError());
159     return {};
160   }
161   auto &DU = *Die.getDwarfUnit();
162   for (auto &Location : *Locations) {
163     auto AddressSize = DU.getAddressByteSize();
164     DataExtractor Data(Location.Expr, DICtx->isLittleEndian(), AddressSize);
165     DWARFExpression Expr(Data, AddressSize);
166     for (auto &Op : Expr)
167       if (Op.getCode() == dwarf::DW_OP_addr)
168         return Op.getRawOperand(0);
169   }
170   return {};
171 }
172 
173 template <class IntPtrT>
174 bool DwarfInstrProfCorrelator<IntPtrT>::isDIEOfProbe(const DWARFDie &Die) {
175   const auto &ParentDie = Die.getParent();
176   if (!Die.isValid() || !ParentDie.isValid() || Die.isNULL())
177     return false;
178   if (Die.getTag() != dwarf::DW_TAG_variable)
179     return false;
180   if (!ParentDie.isSubprogramDIE())
181     return false;
182   if (!Die.hasChildren())
183     return false;
184   if (const char *Name = Die.getName(DINameKind::ShortName))
185     return StringRef(Name).startswith(getInstrProfCountersVarPrefix());
186   return false;
187 }
188 
189 template <class IntPtrT>
190 void DwarfInstrProfCorrelator<IntPtrT>::correlateProfileDataImpl() {
191   auto maybeAddProbe = [&](DWARFDie Die) {
192     if (!isDIEOfProbe(Die))
193       return;
194     Optional<const char *> FunctionName;
195     Optional<uint64_t> CFGHash;
196     Optional<uint64_t> CounterPtr = getLocation(Die);
197     auto FunctionPtr =
198         dwarf::toAddress(Die.getParent().find(dwarf::DW_AT_low_pc));
199     Optional<uint64_t> NumCounters;
200     for (const DWARFDie &Child : Die.children()) {
201       if (Child.getTag() != dwarf::DW_TAG_LLVM_annotation)
202         continue;
203       auto AnnotationFormName = Child.find(dwarf::DW_AT_name);
204       auto AnnotationFormValue = Child.find(dwarf::DW_AT_const_value);
205       if (!AnnotationFormName || !AnnotationFormValue)
206         continue;
207       auto AnnotationNameOrErr = AnnotationFormName->getAsCString();
208       if (auto Err = AnnotationNameOrErr.takeError()) {
209         consumeError(std::move(Err));
210         continue;
211       }
212       StringRef AnnotationName = *AnnotationNameOrErr;
213       if (AnnotationName.compare(
214               InstrProfCorrelator::FunctionNameAttributeName) == 0) {
215         if (auto EC =
216                 AnnotationFormValue->getAsCString().moveInto(FunctionName))
217           consumeError(std::move(EC));
218       } else if (AnnotationName.compare(
219                      InstrProfCorrelator::CFGHashAttributeName) == 0) {
220         CFGHash = AnnotationFormValue->getAsUnsignedConstant();
221       } else if (AnnotationName.compare(
222                      InstrProfCorrelator::NumCountersAttributeName) == 0) {
223         NumCounters = AnnotationFormValue->getAsUnsignedConstant();
224       }
225     }
226     if (!FunctionName || !CFGHash || !CounterPtr || !NumCounters) {
227       LLVM_DEBUG(dbgs() << "Incomplete DIE for probe\n\tFunctionName: "
228                         << FunctionName << "\n\tCFGHash: " << CFGHash
229                         << "\n\tCounterPtr: " << CounterPtr
230                         << "\n\tNumCounters: " << NumCounters);
231       LLVM_DEBUG(Die.dump(dbgs()));
232       return;
233     }
234     uint64_t CountersStart = this->Ctx->CountersSectionStart;
235     uint64_t CountersEnd = this->Ctx->CountersSectionEnd;
236     if (*CounterPtr < CountersStart || *CounterPtr >= CountersEnd) {
237       LLVM_DEBUG(
238           dbgs() << "CounterPtr out of range for probe\n\tFunction Name: "
239                  << FunctionName << "\n\tExpected: [0x"
240                  << Twine::utohexstr(CountersStart) << ", 0x"
241                  << Twine::utohexstr(CountersEnd) << ")\n\tActual: 0x"
242                  << Twine::utohexstr(*CounterPtr));
243       LLVM_DEBUG(Die.dump(dbgs()));
244       return;
245     }
246     if (!FunctionPtr) {
247       LLVM_DEBUG(dbgs() << "Could not find address of " << *FunctionName
248                         << "\n");
249       LLVM_DEBUG(Die.dump(dbgs()));
250     }
251     this->addProbe(*FunctionName, *CFGHash, *CounterPtr - CountersStart,
252                    FunctionPtr.getValueOr(0), *NumCounters);
253   };
254   for (auto &CU : DICtx->normal_units())
255     for (const auto &Entry : CU->dies())
256       maybeAddProbe(DWARFDie(CU.get(), &Entry));
257   for (auto &CU : DICtx->dwo_units())
258     for (const auto &Entry : CU->dies())
259       maybeAddProbe(DWARFDie(CU.get(), &Entry));
260 }
261