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/DebugInfo/DIContext.h"
11 #include "llvm/DebugInfo/DWARF/DWARFContext.h"
12 #include "llvm/DebugInfo/DWARF/DWARFDie.h"
13 #include "llvm/DebugInfo/DWARF/DWARFExpression.h"
14 #include "llvm/DebugInfo/DWARF/DWARFFormValue.h"
15 #include "llvm/DebugInfo/DWARF/DWARFLocationExpression.h"
16 #include "llvm/DebugInfo/DWARF/DWARFUnit.h"
17 #include "llvm/Object/MachO.h"
18 #include "llvm/Support/Debug.h"
19 
20 #define DEBUG_TYPE "correlator"
21 
22 using namespace llvm;
23 
24 /// Get the __llvm_prf_cnts section.
25 Expected<object::SectionRef> getCountersSection(const object::ObjectFile &Obj) {
26   for (auto &Section : Obj.sections())
27     if (auto SectionName = Section.getName())
28       if (SectionName.get() == INSTR_PROF_CNTS_SECT_NAME)
29         return Section;
30   return make_error<InstrProfError>(
31       instrprof_error::unable_to_correlate_profile,
32       "could not find counter section (" INSTR_PROF_CNTS_SECT_NAME ")");
33 }
34 
35 const char *InstrProfCorrelator::FunctionNameAttributeName = "Function Name";
36 const char *InstrProfCorrelator::CFGHashAttributeName = "CFG Hash";
37 const char *InstrProfCorrelator::NumCountersAttributeName = "Num Counters";
38 
39 llvm::Expected<std::unique_ptr<InstrProfCorrelator::Context>>
40 InstrProfCorrelator::Context::get(std::unique_ptr<MemoryBuffer> Buffer,
41                                   const object::ObjectFile &Obj) {
42   auto CountersSection = getCountersSection(Obj);
43   if (auto Err = CountersSection.takeError())
44     return std::move(Err);
45   auto C = std::make_unique<Context>();
46   C->Buffer = std::move(Buffer);
47   C->CountersSectionStart = CountersSection->getAddress();
48   C->CountersSectionEnd = C->CountersSectionStart + CountersSection->getSize();
49   C->ShouldSwapBytes = Obj.isLittleEndian() != sys::IsLittleEndianHost;
50   return Expected<std::unique_ptr<Context>>(std::move(C));
51 }
52 
53 llvm::Expected<std::unique_ptr<InstrProfCorrelator>>
54 InstrProfCorrelator::get(StringRef DebugInfoFilename) {
55   auto DsymObjectsOrErr =
56       object::MachOObjectFile::findDsymObjectMembers(DebugInfoFilename);
57   if (auto Err = DsymObjectsOrErr.takeError())
58     return std::move(Err);
59   if (!DsymObjectsOrErr->empty()) {
60     // TODO: Enable profile correlation when there are multiple objects in a
61     // dSYM bundle.
62     if (DsymObjectsOrErr->size() > 1)
63       return make_error<InstrProfError>(
64           instrprof_error::unable_to_correlate_profile,
65           "using multiple objects is not yet supported");
66     DebugInfoFilename = *DsymObjectsOrErr->begin();
67   }
68   auto BufferOrErr =
69       errorOrToExpected(MemoryBuffer::getFile(DebugInfoFilename));
70   if (auto Err = BufferOrErr.takeError())
71     return std::move(Err);
72 
73   return get(std::move(*BufferOrErr));
74 }
75 
76 llvm::Expected<std::unique_ptr<InstrProfCorrelator>>
77 InstrProfCorrelator::get(std::unique_ptr<MemoryBuffer> Buffer) {
78   auto BinOrErr = object::createBinary(*Buffer);
79   if (auto Err = BinOrErr.takeError())
80     return std::move(Err);
81 
82   if (auto *Obj = dyn_cast<object::ObjectFile>(BinOrErr->get())) {
83     auto CtxOrErr = Context::get(std::move(Buffer), *Obj);
84     if (auto Err = CtxOrErr.takeError())
85       return std::move(Err);
86     auto T = Obj->makeTriple();
87     if (T.isArch64Bit())
88       return InstrProfCorrelatorImpl<uint64_t>::get(std::move(*CtxOrErr), *Obj);
89     if (T.isArch32Bit())
90       return InstrProfCorrelatorImpl<uint32_t>::get(std::move(*CtxOrErr), *Obj);
91   }
92   return make_error<InstrProfError>(
93       instrprof_error::unable_to_correlate_profile, "not an object file");
94 }
95 
96 Optional<size_t> InstrProfCorrelator::getDataSize() const {
97   if (auto *C = dyn_cast<InstrProfCorrelatorImpl<uint32_t>>(this)) {
98     return C->getDataSize();
99   } else if (auto *C = dyn_cast<InstrProfCorrelatorImpl<uint64_t>>(this)) {
100     return C->getDataSize();
101   }
102   return {};
103 }
104 
105 namespace llvm {
106 
107 template <>
108 InstrProfCorrelatorImpl<uint32_t>::InstrProfCorrelatorImpl(
109     std::unique_ptr<InstrProfCorrelator::Context> Ctx)
110     : InstrProfCorrelatorImpl(InstrProfCorrelatorKind::CK_32Bit,
111                               std::move(Ctx)) {}
112 template <>
113 InstrProfCorrelatorImpl<uint64_t>::InstrProfCorrelatorImpl(
114     std::unique_ptr<InstrProfCorrelator::Context> Ctx)
115     : InstrProfCorrelatorImpl(InstrProfCorrelatorKind::CK_64Bit,
116                               std::move(Ctx)) {}
117 template <>
118 bool InstrProfCorrelatorImpl<uint32_t>::classof(const InstrProfCorrelator *C) {
119   return C->getKind() == InstrProfCorrelatorKind::CK_32Bit;
120 }
121 template <>
122 bool InstrProfCorrelatorImpl<uint64_t>::classof(const InstrProfCorrelator *C) {
123   return C->getKind() == InstrProfCorrelatorKind::CK_64Bit;
124 }
125 
126 } // end namespace llvm
127 
128 template <class IntPtrT>
129 llvm::Expected<std::unique_ptr<InstrProfCorrelatorImpl<IntPtrT>>>
130 InstrProfCorrelatorImpl<IntPtrT>::get(
131     std::unique_ptr<InstrProfCorrelator::Context> Ctx,
132     const object::ObjectFile &Obj) {
133   if (Obj.isELF() || Obj.isMachO()) {
134     auto DICtx = DWARFContext::create(Obj);
135     return std::make_unique<DwarfInstrProfCorrelator<IntPtrT>>(std::move(DICtx),
136                                                                std::move(Ctx));
137   }
138   return make_error<InstrProfError>(
139       instrprof_error::unable_to_correlate_profile,
140       "unsupported debug info format (only DWARF is supported)");
141 }
142 
143 template <class IntPtrT>
144 Error InstrProfCorrelatorImpl<IntPtrT>::correlateProfileData() {
145   assert(Data.empty() && Names.empty() && NamesVec.empty());
146   correlateProfileDataImpl();
147   if (Data.empty() || NamesVec.empty())
148     return make_error<InstrProfError>(
149         instrprof_error::unable_to_correlate_profile,
150         "could not find any profile metadata in debug info");
151   auto Result =
152       collectPGOFuncNameStrings(NamesVec, /*doCompression=*/false, Names);
153   CounterOffsets.clear();
154   NamesVec.clear();
155   return Result;
156 }
157 
158 template <class IntPtrT>
159 void InstrProfCorrelatorImpl<IntPtrT>::addProbe(StringRef FunctionName,
160                                                 uint64_t CFGHash,
161                                                 IntPtrT CounterOffset,
162                                                 IntPtrT FunctionPtr,
163                                                 uint32_t NumCounters) {
164   // Check if a probe was already added for this counter offset.
165   if (!CounterOffsets.insert(CounterOffset).second)
166     return;
167   Data.push_back({
168       maybeSwap<uint64_t>(IndexedInstrProf::ComputeHash(FunctionName)),
169       maybeSwap<uint64_t>(CFGHash),
170       // In this mode, CounterPtr actually stores the section relative address
171       // of the counter.
172       maybeSwap<IntPtrT>(CounterOffset),
173       maybeSwap<IntPtrT>(FunctionPtr),
174       // TODO: Value profiling is not yet supported.
175       /*ValuesPtr=*/maybeSwap<IntPtrT>(0),
176       maybeSwap<uint32_t>(NumCounters),
177       /*NumValueSites=*/{maybeSwap<uint16_t>(0), maybeSwap<uint16_t>(0)},
178   });
179   NamesVec.push_back(FunctionName.str());
180 }
181 
182 template <class IntPtrT>
183 llvm::Optional<uint64_t>
184 DwarfInstrProfCorrelator<IntPtrT>::getLocation(const DWARFDie &Die) const {
185   auto Locations = Die.getLocations(dwarf::DW_AT_location);
186   if (!Locations) {
187     consumeError(Locations.takeError());
188     return {};
189   }
190   auto &DU = *Die.getDwarfUnit();
191   auto AddressSize = DU.getAddressByteSize();
192   for (auto &Location : *Locations) {
193     DataExtractor Data(Location.Expr, DICtx->isLittleEndian(), AddressSize);
194     DWARFExpression Expr(Data, AddressSize);
195     for (auto &Op : Expr) {
196       if (Op.getCode() == dwarf::DW_OP_addr) {
197         return Op.getRawOperand(0);
198       } else if (Op.getCode() == dwarf::DW_OP_addrx) {
199         uint64_t Index = Op.getRawOperand(0);
200         if (auto SA = DU.getAddrOffsetSectionItem(Index))
201           return SA->Address;
202       }
203     }
204   }
205   return {};
206 }
207 
208 template <class IntPtrT>
209 bool DwarfInstrProfCorrelator<IntPtrT>::isDIEOfProbe(const DWARFDie &Die) {
210   const auto &ParentDie = Die.getParent();
211   if (!Die.isValid() || !ParentDie.isValid() || Die.isNULL())
212     return false;
213   if (Die.getTag() != dwarf::DW_TAG_variable)
214     return false;
215   if (!ParentDie.isSubprogramDIE())
216     return false;
217   if (!Die.hasChildren())
218     return false;
219   if (const char *Name = Die.getName(DINameKind::ShortName))
220     return StringRef(Name).startswith(getInstrProfCountersVarPrefix());
221   return false;
222 }
223 
224 template <class IntPtrT>
225 void DwarfInstrProfCorrelator<IntPtrT>::correlateProfileDataImpl() {
226   auto maybeAddProbe = [&](DWARFDie Die) {
227     if (!isDIEOfProbe(Die))
228       return;
229     Optional<const char *> FunctionName;
230     Optional<uint64_t> CFGHash;
231     Optional<uint64_t> CounterPtr = getLocation(Die);
232     auto FunctionPtr =
233         dwarf::toAddress(Die.getParent().find(dwarf::DW_AT_low_pc));
234     Optional<uint64_t> NumCounters;
235     for (const DWARFDie &Child : Die.children()) {
236       if (Child.getTag() != dwarf::DW_TAG_LLVM_annotation)
237         continue;
238       auto AnnotationFormName = Child.find(dwarf::DW_AT_name);
239       auto AnnotationFormValue = Child.find(dwarf::DW_AT_const_value);
240       if (!AnnotationFormName || !AnnotationFormValue)
241         continue;
242       auto AnnotationNameOrErr = AnnotationFormName->getAsCString();
243       if (auto Err = AnnotationNameOrErr.takeError()) {
244         consumeError(std::move(Err));
245         continue;
246       }
247       StringRef AnnotationName = *AnnotationNameOrErr;
248       if (AnnotationName.compare(
249               InstrProfCorrelator::FunctionNameAttributeName) == 0) {
250         if (auto EC =
251                 AnnotationFormValue->getAsCString().moveInto(FunctionName))
252           consumeError(std::move(EC));
253       } else if (AnnotationName.compare(
254                      InstrProfCorrelator::CFGHashAttributeName) == 0) {
255         CFGHash = AnnotationFormValue->getAsUnsignedConstant();
256       } else if (AnnotationName.compare(
257                      InstrProfCorrelator::NumCountersAttributeName) == 0) {
258         NumCounters = AnnotationFormValue->getAsUnsignedConstant();
259       }
260     }
261     if (!FunctionName || !CFGHash || !CounterPtr || !NumCounters) {
262       LLVM_DEBUG(dbgs() << "Incomplete DIE for probe\n\tFunctionName: "
263                         << FunctionName << "\n\tCFGHash: " << CFGHash
264                         << "\n\tCounterPtr: " << CounterPtr
265                         << "\n\tNumCounters: " << NumCounters);
266       LLVM_DEBUG(Die.dump(dbgs()));
267       return;
268     }
269     uint64_t CountersStart = this->Ctx->CountersSectionStart;
270     uint64_t CountersEnd = this->Ctx->CountersSectionEnd;
271     if (*CounterPtr < CountersStart || *CounterPtr >= CountersEnd) {
272       LLVM_DEBUG(
273           dbgs() << "CounterPtr out of range for probe\n\tFunction Name: "
274                  << FunctionName << "\n\tExpected: [0x"
275                  << Twine::utohexstr(CountersStart) << ", 0x"
276                  << Twine::utohexstr(CountersEnd) << ")\n\tActual: 0x"
277                  << Twine::utohexstr(*CounterPtr));
278       LLVM_DEBUG(Die.dump(dbgs()));
279       return;
280     }
281     if (!FunctionPtr) {
282       LLVM_DEBUG(dbgs() << "Could not find address of " << *FunctionName
283                         << "\n");
284       LLVM_DEBUG(Die.dump(dbgs()));
285     }
286     this->addProbe(*FunctionName, *CFGHash, *CounterPtr - CountersStart,
287                    FunctionPtr.value_or(0), *NumCounters);
288   };
289   for (auto &CU : DICtx->normal_units())
290     for (const auto &Entry : CU->dies())
291       maybeAddProbe(DWARFDie(CU.get(), &Entry));
292   for (auto &CU : DICtx->dwo_units())
293     for (const auto &Entry : CU->dies())
294       maybeAddProbe(DWARFDie(CU.get(), &Entry));
295 }
296