[LLVMSymbolize] Move ModuleInfo into a separate class (SymbolizableModule).

Summary:
This is mostly NFC. It is a first step in cleaning up LLVMSymbolize
library. It removes "ModuleInfo" class which bundles together ObjectFile
and its debug info context in favor of:
  * abstract SymbolizableModule in public headers;
  * SymbolizableObjectFile subclass in implementation.

Additionally, SymbolizableObjectFile is now created via factory, so we
can properly detect object parsing error at this stage instead of keeping
the broken half-parsed object. As a next step, we would be able to
propagate the error all the way back to the library user.

Further improvements might include:
  * factoring out the logic of finding appropriate file with debug info
    for a given object file, and caching all parsed object files into a
    separate class [A].
  * factoring out DILineInfo rendering [B].
This would make what is now a heavyweight "LLVMSymbolizer" a relatively
straightforward class, that calls into [A] to turn filepath into a
SymbolizableModule, delegates actual symbolization to concrete SymbolizableModule
implementation, and lets [C] render the result.

Reviewers: dblaikie, echristo, rafael

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D14099

llvm-svn: 251662

1 files changed, 251 insertions, 0 deletions

diff --git a/llvm/lib/DebugInfo/Symbolize/SymbolizableObjectFile.cpp b/llvm/lib/DebugInfo/Symbolize/SymbolizableObjectFile.cpp
new file mode 100644
index 00000000000..65d4cce3dc1
--- /dev/null
+++ b/llvm/lib/DebugInfo/Symbolize/SymbolizableObjectFile.cpp
@@ -0,0 +1,251 @@
+//===-- SymbolizableObjectFile.cpp ----------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// Implementation of SymbolizableObjectFile class.
+//
+//===----------------------------------------------------------------------===//
+
+#include "SymbolizableObjectFile.h"
+#include "llvm/Object/SymbolSize.h"
+#include "llvm/Support/DataExtractor.h"
+
+namespace llvm {
+namespace symbolize {
+
+using namespace object;
+
+static DILineInfoSpecifier
+getDILineInfoSpecifier(FunctionNameKind FNKind) {
+  return DILineInfoSpecifier(
+      DILineInfoSpecifier::FileLineInfoKind::AbsoluteFilePath, FNKind);
+}
+
+ErrorOr<std::unique_ptr<SymbolizableObjectFile>>
+SymbolizableObjectFile::create(object::ObjectFile *Obj,
+                               std::unique_ptr<DIContext> DICtx) {
+  std::unique_ptr<SymbolizableObjectFile> res(
+      new SymbolizableObjectFile(Obj, std::move(DICtx)));
+  std::unique_ptr<DataExtractor> OpdExtractor;
+  uint64_t OpdAddress = 0;
+  // Find the .opd (function descriptor) section if any, for big-endian
+  // PowerPC64 ELF.
+  if (Obj->getArch() == Triple::ppc64) {
+    for (section_iterator Section : Obj->sections()) {
+      StringRef Name;
+      StringRef Data;
+      if (auto EC = Section->getName(Name))
+        return EC;
+      if (Name == ".opd") {
+        if (auto EC = Section->getContents(Data))
+          return EC;
+        OpdExtractor.reset(new DataExtractor(Data, Obj->isLittleEndian(),
+                                             Obj->getBytesInAddress()));
+        OpdAddress = Section->getAddress();
+        break;
+      }
+    }
+  }
+  std::vector<std::pair<SymbolRef, uint64_t>> Symbols =
+      computeSymbolSizes(*Obj);
+  for (auto &P : Symbols)
+    res->addSymbol(P.first, P.second, OpdExtractor.get(), OpdAddress);
+
+  // If this is a COFF object and we didn't find any symbols, try the export
+  // table.
+  if (Symbols.empty()) {
+    if (auto *CoffObj = dyn_cast<COFFObjectFile>(Obj))
+      if (auto EC = res->addCoffExportSymbols(CoffObj))
+        return EC;
+  }
+  return std::move(res);
+}
+
+SymbolizableObjectFile::SymbolizableObjectFile(ObjectFile *Obj,
+                                               std::unique_ptr<DIContext> DICtx)
+    : Module(Obj), DebugInfoContext(std::move(DICtx)) {}
+
+namespace {
+struct OffsetNamePair {
+  uint32_t Offset;
+  StringRef Name;
+  bool operator<(const OffsetNamePair &R) const {
+    return Offset < R.Offset;
+  }
+};
+}
+
+std::error_code SymbolizableObjectFile::addCoffExportSymbols(
+    const COFFObjectFile *CoffObj) {
+  // Get all export names and offsets.
+  std::vector<OffsetNamePair> ExportSyms;
+  for (const ExportDirectoryEntryRef &Ref : CoffObj->export_directories()) {
+    StringRef Name;
+    uint32_t Offset;
+    if (auto EC = Ref.getSymbolName(Name))
+      return EC;
+    if (auto EC = Ref.getExportRVA(Offset))
+      return EC;
+    ExportSyms.push_back(OffsetNamePair{Offset, Name});
+  }
+  if (ExportSyms.empty())
+    return std::error_code();
+
+  // Sort by ascending offset.
+  array_pod_sort(ExportSyms.begin(), ExportSyms.end());
+
+  // Approximate the symbol sizes by assuming they run to the next symbol.
+  // FIXME: This assumes all exports are functions.
+  uint64_t ImageBase = CoffObj->getImageBase();
+  for (auto I = ExportSyms.begin(), E = ExportSyms.end(); I != E; ++I) {
+    OffsetNamePair &Export = *I;
+    // FIXME: The last export has a one byte size now.
+    uint32_t NextOffset = I != E ? I->Offset : Export.Offset + 1;
+    uint64_t SymbolStart = ImageBase + Export.Offset;
+    uint64_t SymbolSize = NextOffset - Export.Offset;
+    SymbolDesc SD = {SymbolStart, SymbolSize};
+    Functions.insert(std::make_pair(SD, Export.Name));
+  }
+  return std::error_code();
+}
+
+std::error_code SymbolizableObjectFile::addSymbol(const SymbolRef &Symbol,
+                                                  uint64_t SymbolSize,
+                                                  DataExtractor *OpdExtractor,
+                                                  uint64_t OpdAddress) {
+  SymbolRef::Type SymbolType = Symbol.getType();
+  if (SymbolType != SymbolRef::ST_Function && SymbolType != SymbolRef::ST_Data)
+    return std::error_code();
+  ErrorOr<uint64_t> SymbolAddressOrErr = Symbol.getAddress();
+  if (auto EC = SymbolAddressOrErr.getError())
+    return EC;
+  uint64_t SymbolAddress = *SymbolAddressOrErr;
+  if (OpdExtractor) {
+    // For big-endian PowerPC64 ELF, symbols in the .opd section refer to
+    // function descriptors. The first word of the descriptor is a pointer to
+    // the function's code.
+    // For the purposes of symbolization, pretend the symbol's address is that
+    // of the function's code, not the descriptor.
+    uint64_t OpdOffset = SymbolAddress - OpdAddress;
+    uint32_t OpdOffset32 = OpdOffset;
+    if (OpdOffset == OpdOffset32 && 
+        OpdExtractor->isValidOffsetForAddress(OpdOffset32))
+      SymbolAddress = OpdExtractor->getAddress(&OpdOffset32);
+  }
+  ErrorOr<StringRef> SymbolNameOrErr = Symbol.getName();
+  if (auto EC = SymbolNameOrErr.getError())
+    return EC;
+  StringRef SymbolName = *SymbolNameOrErr;
+  // Mach-O symbol table names have leading underscore, skip it.
+  if (Module->isMachO() && SymbolName.size() > 0 && SymbolName[0] == '_')
+    SymbolName = SymbolName.drop_front();
+  // FIXME: If a function has alias, there are two entries in symbol table
+  // with same address size. Make sure we choose the correct one.
+  auto &M = SymbolType == SymbolRef::ST_Function ? Functions : Objects;
+  SymbolDesc SD = { SymbolAddress, SymbolSize };
+  M.insert(std::make_pair(SD, SymbolName));
+  return std::error_code();
+}
+
+// Return true if this is a 32-bit x86 PE COFF module.
+bool SymbolizableObjectFile::isWin32Module() const {
+  auto *CoffObject = dyn_cast<COFFObjectFile>(Module);
+  return CoffObject && CoffObject->getMachine() == COFF::IMAGE_FILE_MACHINE_I386;
+}
+
+uint64_t SymbolizableObjectFile::getModulePreferredBase() const {
+  if (auto *CoffObject = dyn_cast<COFFObjectFile>(Module))
+    return CoffObject->getImageBase();
+  return 0;
+}
+
+bool SymbolizableObjectFile::getNameFromSymbolTable(SymbolRef::Type Type,
+                                                    uint64_t Address,
+                                                    std::string &Name,
+                                                    uint64_t &Addr,
+                                                    uint64_t &Size) const {
+  const auto &SymbolMap = Type == SymbolRef::ST_Function ? Functions : Objects;
+  if (SymbolMap.empty())
+    return false;
+  SymbolDesc SD = { Address, Address };
+  auto SymbolIterator = SymbolMap.upper_bound(SD);
+  if (SymbolIterator == SymbolMap.begin())
+    return false;
+  --SymbolIterator;
+  if (SymbolIterator->first.Size != 0 &&
+      SymbolIterator->first.Addr + SymbolIterator->first.Size <= Address)
+    return false;
+  Name = SymbolIterator->second.str();
+  Addr = SymbolIterator->first.Addr;
+  Size = SymbolIterator->first.Size;
+  return true;
+}
+
+DILineInfo SymbolizableObjectFile::symbolizeCode(uint64_t ModuleOffset,
+                                                 FunctionNameKind FNKind,
+                                                 bool UseSymbolTable) const {
+  DILineInfo LineInfo;
+  if (DebugInfoContext) {
+    LineInfo = DebugInfoContext->getLineInfoForAddress(
+        ModuleOffset, getDILineInfoSpecifier(FNKind));
+  }
+  // Override function name from symbol table if necessary.
+  if (FNKind == FunctionNameKind::LinkageName && UseSymbolTable) {
+    std::string FunctionName;
+    uint64_t Start, Size;
+    if (getNameFromSymbolTable(SymbolRef::ST_Function, ModuleOffset,
+                               FunctionName, Start, Size)) {
+      LineInfo.FunctionName = FunctionName;
+    }
+  }
+  return LineInfo;
+}
+
+DIInliningInfo SymbolizableObjectFile::symbolizeInlinedCode(
+    uint64_t ModuleOffset, FunctionNameKind FNKind, bool UseSymbolTable) const {
+  DIInliningInfo InlinedContext;
+
+  if (DebugInfoContext)
+    InlinedContext = DebugInfoContext->getInliningInfoForAddress(
+        ModuleOffset, getDILineInfoSpecifier(FNKind));
+  // Make sure there is at least one frame in context.
+  if (InlinedContext.getNumberOfFrames() == 0)
+    InlinedContext.addFrame(DILineInfo());
+
+  if (FNKind != FunctionNameKind::LinkageName || !UseSymbolTable)
+    return InlinedContext;
+
+  // Override the function name in lower frame with name from symbol table.
+  // We can't directly change the last element of DIInliningInfo, so copy
+  // all frames into new context, replacing function name in the last one.
+  DIInliningInfo PatchedInlinedContext;
+  for (uint32_t i = 0, n = InlinedContext.getNumberOfFrames(); i < n; i++) {
+    DILineInfo LineInfo = InlinedContext.getFrame(i);
+    if (i == n - 1) {
+      std::string FunctionName;
+      uint64_t Start, Size;
+      if (getNameFromSymbolTable(SymbolRef::ST_Function, ModuleOffset,
+                                 FunctionName, Start, Size)) {
+        LineInfo.FunctionName = FunctionName;
+      }
+    }
+    PatchedInlinedContext.addFrame(LineInfo);
+  }
+  return PatchedInlinedContext;
+}
+
+bool SymbolizableObjectFile::symbolizeData(uint64_t ModuleOffset,
+                                           std::string &Name, uint64_t &Start,
+                                           uint64_t &Size) const {
+  return getNameFromSymbolTable(SymbolRef::ST_Data, ModuleOffset, Name, Start,
+                                Size);
+}
+
+}  // namespace symbolize
+}  // namespace llvm
+