[Debuginfod] [Symbolizer] Break debuginfod out of libLLVM.

[mysterymath]

Summary

Debuginfod can pull in libcurl as a dependency, which isn't appropriate
for libLLVM. (See
https://gitlab.freedesktop.org/mesa/mesa/-/issues/5732).

This change breaks out debuginfod into a separate non-component library
that can be used directly in llvm-symbolizer. The tool can inject
debuginfod into the Symbolizer library via an abstract DebugInfoFetcher
interface, breaking the dependency of Symbolizer on debuinfod.

See https://github.com/llvm/llvm-project/issues/52731

[mysterymath]

Run git clang-format.

[phosek]

llvm/include/llvm/DebugInfo/Symbolize/DebugInfoFetcher.h

29 ↗

(On Diff #403839)

I'd probably go with a simple DIFetcher which better matches the derived types.

32–36 ↗

(On Diff #403839)

Do we really need the concept? Wouldn't a simple inheritance with virtual functions be sufficient? IIUC the reason for using concept is if you want to opt types that you do not own into the abstract interface, but that's not what we need here since all the types are part of LLVM.

[mysterymath]

DebugInfoFetcher -> DIFetcher

llvm/include/llvm/DebugInfo/Symbolize/DebugInfoFetcher.h

32–36 ↗

(On Diff #403839)

The other advantage is removing the requirement for any DIFetcher implementation to depend on Symbolizer. A concept allows the DIFetcher abstraction to properly belong to Symbolizer, i.e., point of use rather than point of implementation.

But, it's a fair amount of syntactic/semantic overhead for just that; my golang proclivities are probably showing a bit. I'm quite ambivalent about this; if this seems like overkill for the gain, let me know.

[mysterymath]

Simplify concept to interface.

[mysterymath]

I took a look at the existing concepts in LLVM, and given the overhead in code, they have to be born of deep frustrations. Given that those aren't present here, removed; it should be obvious when they are necessary.

[dblaikie]

In D118413#3292466, @mysterymath wrote:

I took a look at the existing concepts in LLVM, and given the overhead in code, they have to be born of deep frustrations. Given that those aren't present here, removed; it should be obvious when they are necessary.

Can't say I'm following this comment - could you describe it in more detail. What existing concepts? What gives the impression they were "Born of deep frustration"?

[mysterymath]

In D118413#3294741, @dblaikie wrote:

In D118413#3292466, @mysterymath wrote:

I took a look at the existing concepts in LLVM, and given the overhead in code, they have to be born of deep frustrations. Given that those aren't present here, removed; it should be obvious when they are necessary.

Can't say I'm following this comment - could you describe it in more detail. What existing concepts? What gives the impression they were "Born of deep frustration"?

Ah, this is in reference to the "concept/model" pattern from Sean Parent's "Inheritance is the base class of evil" talk: https://www.youtube.com/watch?v=bIhUE5uUFOA
It's a way of using templates to get golang-style structural typing interfaces.

It's used in a few places in LLVM; an example is AliasAnalysis.h and IIRC, TargetTransformInfo. You have to duplicate each method two or three times to implement the pattern, so it incurs a large constant overhead in code/maintenance, but the benefits are less clear. Since I don't fully understand why things were done that way for those specific places in LLVM where it's used, I can't effectively argue for the pattern here, so I was just cargo-culting by using it.

[dblaikie]

In D118413#3294847, @mysterymath wrote:

In D118413#3294741, @dblaikie wrote:

In D118413#3292466, @mysterymath wrote:

I took a look at the existing concepts in LLVM, and given the overhead in code, they have to be born of deep frustrations. Given that those aren't present here, removed; it should be obvious when they are necessary.

Can't say I'm following this comment - could you describe it in more detail. What existing concepts? What gives the impression they were "Born of deep frustration"?

Ah, this is in reference to the "concept/model" pattern from Sean Parent's "Inheritance is the base class of evil" talk: https://www.youtube.com/watch?v=bIhUE5uUFOA
It's a way of using templates to get golang-style structural typing interfaces.

It's used in a few places in LLVM; an example is AliasAnalysis.h and IIRC, TargetTransformInfo. You have to duplicate each method two or three times to implement the pattern, so it incurs a large constant overhead in code/maintenance, but the benefits are less clear. Since I don't fully understand why things were done that way for those specific places in LLVM where it's used, I can't effectively argue for the pattern here, so I was just cargo-culting by using it.

Ah, OK - thanks for the details.

[phosek]

LGTM

llvm/include/llvm/DebugInfo/Symbolize/DIFetcher.h

+//===-- llvm/DebugInfo/Symbolize/DIFetcher.h --------------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+///
+/// \file
+/// This file declares a DIFetcher abstraction for obtaining debug info from an
+/// arbitrary outside source.
+///
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_DEBUGINFO_SYMBOLIZE_DIFETCHER_H
+#define LLVM_DEBUGINFO_SYMBOLIZE_DIFETCHER_H
+
+#include <cstdint>
+#include <string>
+
+#include "llvm/ADT/ArrayRef.h"
+
+namespace llvm {
+namespace symbolize {
+
+/// The DIFetcher interface provides arbitrary mechanisms for obtaining debug
+/// info from an outside source.
+class DIFetcher {
+public:
+  virtual ~DIFetcher() = default;
+  virtual Optional<std::string>
+  fetchBuildID(ArrayRef<uint8_t> BuildID) const = 0;
+};
+
+/// LocalDIFetcher searches local cache directories for debug info.
+class LocalDIFetcher : public DIFetcher {
+public:
+  LocalDIFetcher(ArrayRef<std::string> DebugFileDirectory)
+      : DebugFileDirectory(DebugFileDirectory){};
+  virtual ~LocalDIFetcher() = default;
+
+  Optional<std::string> fetchBuildID(ArrayRef<uint8_t> BuildID) const override;
+
+private:
+  const ArrayRef<std::string> DebugFileDirectory;
+};
+
+} // end namespace symbolize
+} // end namespace llvm
+
+#endif // LLVM_DEBUGINFO_SYMBOLIZE_DIFETCHER_H

llvm/include/llvm/DebugInfo/Symbolize/Symbolize.h

 //===- Symbolize.h ----------------------------------------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // Header for LLVM symbolization library.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_DEBUGINFO_SYMBOLIZE_SYMBOLIZE_H
 #define LLVM_DEBUGINFO_SYMBOLIZE_SYMBOLIZE_H
 
+#include "llvm/DebugInfo/Symbolize/DIFetcher.h"
 #include "llvm/DebugInfo/Symbolize/SymbolizableModule.h"
 #include "llvm/Object/Binary.h"
 #include "llvm/Object/ELFObjectFile.h"
 #include "llvm/Object/ObjectFile.h"
 #include "llvm/Support/Error.h"
 #include <algorithm>
 #include <cstdint>
 #include <map>
public:
   symbolizeFrame(const std::string &ModuleName,
                  object::SectionedAddress ModuleOffset);
   void flush();
 
   static std::string
   DemangleName(const std::string &Name,
                const SymbolizableModule *DbiModuleDescriptor);
 
+  void addDIFetcher(std::unique_ptr<DIFetcher> Fetcher) {
+    DIFetchers.push_back(std::move(Fetcher));
+  }
+
 private:
   // Bundles together object file with code/data and object file with
   // corresponding debug info. These objects can be the same.
   using ObjectPair = std::pair<const ObjectFile *, const ObjectFile *>;
 
   template <typename T>
   Expected<DILineInfo>
   symbolizeCodeCommon(const T &ModuleSpecifier,
ObjectFile *lookUpDsymFile(const std::string &Path,
                              const std::string &ArchName);
   ObjectFile *lookUpDebuglinkObject(const std::string &Path,
                                     const ObjectFile *Obj,
                                     const std::string &ArchName);
   ObjectFile *lookUpBuildIDObject(const std::string &Path,
                                   const ELFObjectFileBase *Obj,
                                   const std::string &ArchName);
 
+  bool findDebugBinary(const std::string &OrigPath,
+                       const std::string &DebuglinkName, uint32_t CRCHash,
+                       std::string &Result);
+
+  bool findDebugBinary(const ArrayRef<uint8_t> BuildID, std::string &Result);
+
   /// Returns pair of pointers to object and debug object.
   Expected<ObjectPair> getOrCreateObjectPair(const std::string &Path,
                                              const std::string &ArchName);
 
   /// Return a pointer to object file at specified path, for a specified
   /// architecture (e.g. if path refers to a Mach-O universal binary, only one
   /// object file from it will be returned).
   Expected<ObjectFile *> getOrCreateObject(const std::string &Path,
private:
   std::map<std::string, OwningBinary<Binary>> BinaryForPath;
 
   /// Parsed object file for path/architecture pair, where "path" refers
   /// to Mach-O universal binary.
   std::map<std::pair<std::string, std::string>, std::unique_ptr<ObjectFile>>
       ObjectForUBPathAndArch;
 
   Options Opts;
+
+  SmallVector<std::unique_ptr<DIFetcher>> DIFetchers;
 };
 
 } // end namespace symbolize
 } // end namespace llvm
 
 #endif // LLVM_DEBUGINFO_SYMBOLIZE_SYMBOLIZE_H

llvm/include/llvm/Debuginfod/DIFetcher.h

+//===- llvm/DebugInfod/DIFetcher.h - Debug info fetcher----------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+///
+/// \file
+/// This file declares a DIFetcher implementation for obtaining debug info from
+/// debuginfod.
+///
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_DEBUGINFOD_DIFETCHER_H
+#define LLVM_DEBUGINFOD_DIFETCHER_H
+
+#include "llvm/ADT/ArrayRef.h"
+#include "llvm/DebugInfo/Symbolize/DIFetcher.h"
+
+namespace llvm {
+
+class DebuginfodDIFetcher : public symbolize::DIFetcher {
+public:
+  virtual ~DebuginfodDIFetcher() = default;
+
+  /// Fetches the given Build ID using debuginfod and returns a local path to
+  /// the resulting debug binary.
+  Optional<std::string> fetchBuildID(ArrayRef<uint8_t> BuildID) const override;
+};
+
+} // namespace llvm
+
+#endif // LLVM_DEBUGINFOD_DIFETCHER_H

llvm/lib/DebugInfo/Symbolize/CMakeLists.txt

 add_llvm_component_library(LLVMSymbolize
+  DIFetcher.cpp
   DIPrinter.cpp
   SymbolizableObjectFile.cpp
   Symbolize.cpp
 
   ADDITIONAL_HEADER_DIRS
   ${LLVM_MAIN_INCLUDE_DIR}/llvm/DebugInfo/Symbolize
 
   LINK_COMPONENTS
   DebugInfoDWARF
   DebugInfoPDB
-  Debuginfod
   Object
   Support
   Demangle
   )

llvm/lib/DebugInfo/Symbolize/DIFetcher.cpp

+//===-- lib/DebugInfo/Symbolize/DIFetcher.cpp -----------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+///
+/// \file
+/// This file defines the implementation of the local debug info fetcher, which
+/// searches cache directories.
+///
+//===----------------------------------------------------------------------===//
+
+#include "llvm/DebugInfo/Symbolize/DIFetcher.h"
+
+#include "llvm/Debuginfod/Debuginfod.h"
+#include "llvm/Support/FileSystem.h"
+#include "llvm/Support/Path.h"
+
+namespace llvm {
+namespace symbolize {
+
+Optional<std::string>
+LocalDIFetcher::fetchBuildID(ArrayRef<uint8_t> BuildID) const {
+  auto GetDebugPath = [&](StringRef Directory) {
+    SmallString<128> Path{Directory};
+    sys::path::append(Path, ".build-id",
+                      llvm::toHex(BuildID[0], /*LowerCase=*/true),
+                      llvm::toHex(BuildID.slice(1), /*LowerCase=*/true));
+    Path += ".debug";
+    return Path;
+  };
+  if (DebugFileDirectory.empty()) {
+    SmallString<128> Path = GetDebugPath(
+#if defined(__NetBSD__)
+        // Try /usr/libdata/debug/.build-id/../...
+        "/usr/libdata/debug"
+#else
+        // Try /usr/lib/debug/.build-id/../...
+        "/usr/lib/debug"
+#endif
+    );
+    if (llvm::sys::fs::exists(Path))
+      return std::string(Path);
+  } else {
+    for (const auto &Directory : DebugFileDirectory) {
+      // Try <debug-file-directory>/.build-id/../...
+      SmallString<128> Path = GetDebugPath(Directory);
+      if (llvm::sys::fs::exists(Path))
+        return std::string(Path);
+    }
+  }
+  return None;
+}
+
+} // namespace symbolize
+} // namespace llvm

llvm/lib/DebugInfo/Symbolize/Symbolize.cpp

 #include "SymbolizableObjectFile.h"
 
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/BinaryFormat/COFF.h"
 #include "llvm/Config/config.h"
 #include "llvm/DebugInfo/DWARF/DWARFContext.h"
 #include "llvm/DebugInfo/PDB/PDB.h"
 #include "llvm/DebugInfo/PDB/PDBContext.h"
-#include "llvm/Debuginfod/Debuginfod.h"
+#include "llvm/DebugInfo/Symbolize/DIFetcher.h"
 #include "llvm/Demangle/Demangle.h"
 #include "llvm/Object/COFF.h"
 #include "llvm/Object/MachO.h"
 #include "llvm/Object/MachOUniversal.h"
 #include "llvm/Support/CRC.h"
 #include "llvm/Support/Casting.h"
 #include "llvm/Support/Compression.h"
 #include "llvm/Support/DataExtractor.h"

 bool checkFileCRC(StringRef Path, uint32_t CRCHash) {
   ErrorOr<std::unique_ptr<MemoryBuffer>> MB =
       MemoryBuffer::getFileOrSTDIN(Path);
   if (!MB)
     return false;
   return CRCHash == llvm::crc32(arrayRefFromStringRef(MB.get()->getBuffer()));
 }
 
-bool findDebugBinary(const std::string &OrigPath,
-                     const std::string &DebuglinkName, uint32_t CRCHash,
-                     const std::string &FallbackDebugPath,
-                     std::string &Result) {
-  SmallString<16> OrigDir(OrigPath);
-  llvm::sys::path::remove_filename(OrigDir);
-  SmallString<16> DebugPath = OrigDir;
-  // Try relative/path/to/original_binary/debuglink_name
-  llvm::sys::path::append(DebugPath, DebuglinkName);
-  if (checkFileCRC(DebugPath, CRCHash)) {
-    Result = std::string(DebugPath.str());
-    return true;
-  }
-  // Try relative/path/to/original_binary/.debug/debuglink_name
-  DebugPath = OrigDir;
-  llvm::sys::path::append(DebugPath, ".debug", DebuglinkName);
-  if (checkFileCRC(DebugPath, CRCHash)) {
-    Result = std::string(DebugPath.str());
-    return true;
-  }
-  // Make the path absolute so that lookups will go to
-  // "/usr/lib/debug/full/path/to/debug", not
-  // "/usr/lib/debug/to/debug"
-  llvm::sys::fs::make_absolute(OrigDir);
-  if (!FallbackDebugPath.empty()) {
-    // Try <FallbackDebugPath>/absolute/path/to/original_binary/debuglink_name
-    DebugPath = FallbackDebugPath;
-  } else {
-#if defined(__NetBSD__)
-    // Try /usr/libdata/debug/absolute/path/to/original_binary/debuglink_name
-    DebugPath = "/usr/libdata/debug";
-#else
-    // Try /usr/lib/debug/absolute/path/to/original_binary/debuglink_name
-    DebugPath = "/usr/lib/debug";
-#endif
-  }
-  llvm::sys::path::append(DebugPath, llvm::sys::path::relative_path(OrigDir),
-                          DebuglinkName);
-  if (checkFileCRC(DebugPath, CRCHash)) {
-    Result = std::string(DebugPath.str());
-    return true;
-  }
-  return false;
-}
-
 bool getGNUDebuglinkContents(const ObjectFile *Obj, std::string &DebugName,
                              uint32_t &CRCHash) {
   if (!Obj)
     return false;
   for (const SectionRef &Section : Obj->sections()) {
     StringRef Name;
     consumeError(Section.getName().moveInto(Name));
 
else if (auto *O = dyn_cast<ELFObjectFile<ELF64LE>>(Obj))
     BuildID = getBuildID(O->getELFFile());
   else if (auto *O = dyn_cast<ELFObjectFile<ELF64BE>>(Obj))
     BuildID = getBuildID(O->getELFFile());
   else
     llvm_unreachable("unsupported file format");
   return BuildID;
 }
 
-bool findDebugBinary(const std::vector<std::string> &DebugFileDirectory,
-                     const ArrayRef<uint8_t> BuildID, std::string &Result) {
-  auto getDebugPath = [&](StringRef Directory) {
-    SmallString<128> Path{Directory};
-    sys::path::append(Path, ".build-id",
-                      llvm::toHex(BuildID[0], /*LowerCase=*/true),
-                      llvm::toHex(BuildID.slice(1), /*LowerCase=*/true));
-    Path += ".debug";
-    return Path;
-  };
-  if (DebugFileDirectory.empty()) {
-    SmallString<128> Path = getDebugPath(
-#if defined(__NetBSD__)
-        // Try /usr/libdata/debug/.build-id/../...
-        "/usr/libdata/debug"
-#else
-        // Try /usr/lib/debug/.build-id/../...
-        "/usr/lib/debug"
-#endif
-    );
-    if (llvm::sys::fs::exists(Path)) {
-      Result = std::string(Path.str());
-      return true;
-    }
-  } else {
-    for (const auto &Directory : DebugFileDirectory) {
-      // Try <debug-file-directory>/.build-id/../...
-      SmallString<128> Path = getDebugPath(Directory);
-      if (llvm::sys::fs::exists(Path)) {
-        Result = std::string(Path.str());
-        return true;
-      }
-    }
-  }
-  // Try debuginfod client cache and known servers.
-  Expected<std::string> PathOrErr = getCachedOrDownloadDebuginfo(BuildID);
-  if (!PathOrErr) {
-    consumeError(PathOrErr.takeError());
-    return false;
-  }
-  Result = *PathOrErr;
-  return true;
-}
-
 } // end anonymous namespace
 
 ObjectFile *LLVMSymbolizer::lookUpDsymFile(const std::string &ExePath,
                                            const MachOObjectFile *MachExeObj,
                                            const std::string &ArchName) {
   // On Darwin we may find DWARF in separate object file in
   // resource directory.
   std::vector<std::string> DsymPaths;

 ObjectFile *LLVMSymbolizer::lookUpDebuglinkObject(const std::string &Path,
                                                   const ObjectFile *Obj,
                                                   const std::string &ArchName) {
   std::string DebuglinkName;
   uint32_t CRCHash;
   std::string DebugBinaryPath;
   if (!getGNUDebuglinkContents(Obj, DebuglinkName, CRCHash))
     return nullptr;
-  if (!findDebugBinary(Path, DebuglinkName, CRCHash, Opts.FallbackDebugPath,
-                       DebugBinaryPath))
+  if (!findDebugBinary(Path, DebuglinkName, CRCHash, DebugBinaryPath))
     return nullptr;
   auto DbgObjOrErr = getOrCreateObject(DebugBinaryPath, ArchName);
   if (!DbgObjOrErr) {
     // Ignore errors, the file might not exist.
     consumeError(DbgObjOrErr.takeError());
     return nullptr;
   }
   return DbgObjOrErr.get();
 }
 
 ObjectFile *LLVMSymbolizer::lookUpBuildIDObject(const std::string &Path,
                                                 const ELFObjectFileBase *Obj,
                                                 const std::string &ArchName) {
   auto BuildID = getBuildID(Obj);
   if (!BuildID)
     return nullptr;
   if (BuildID->size() < 2)
     return nullptr;
   std::string DebugBinaryPath;
-  if (!findDebugBinary(Opts.DebugFileDirectory, *BuildID, DebugBinaryPath))
+  if (!findDebugBinary(*BuildID, DebugBinaryPath))
     return nullptr;
   auto DbgObjOrErr = getOrCreateObject(DebugBinaryPath, ArchName);
   if (!DbgObjOrErr) {
     consumeError(DbgObjOrErr.takeError());
     return nullptr;
   }
   return DbgObjOrErr.get();
 }
 
+bool LLVMSymbolizer::findDebugBinary(const std::string &OrigPath,
+                                     const std::string &DebuglinkName,
+                                     uint32_t CRCHash, std::string &Result) {
+  SmallString<16> OrigDir(OrigPath);
+  llvm::sys::path::remove_filename(OrigDir);
+  SmallString<16> DebugPath = OrigDir;
+  // Try relative/path/to/original_binary/debuglink_name
+  llvm::sys::path::append(DebugPath, DebuglinkName);
+  if (checkFileCRC(DebugPath, CRCHash)) {
+    Result = std::string(DebugPath.str());
+    return true;
+  }
+  // Try relative/path/to/original_binary/.debug/debuglink_name
+  DebugPath = OrigDir;
+  llvm::sys::path::append(DebugPath, ".debug", DebuglinkName);
+  if (checkFileCRC(DebugPath, CRCHash)) {
+    Result = std::string(DebugPath.str());
+    return true;
+  }
+  // Make the path absolute so that lookups will go to
+  // "/usr/lib/debug/full/path/to/debug", not
+  // "/usr/lib/debug/to/debug"
+  llvm::sys::fs::make_absolute(OrigDir);
+  if (!Opts.FallbackDebugPath.empty()) {
+    // Try <FallbackDebugPath>/absolute/path/to/original_binary/debuglink_name
+    DebugPath = Opts.FallbackDebugPath;
+  } else {
+#if defined(__NetBSD__)
+    // Try /usr/libdata/debug/absolute/path/to/original_binary/debuglink_name
+    DebugPath = "/usr/libdata/debug";
+#else
+    // Try /usr/lib/debug/absolute/path/to/original_binary/debuglink_name
+    DebugPath = "/usr/lib/debug";
+#endif
+  }
+  llvm::sys::path::append(DebugPath, llvm::sys::path::relative_path(OrigDir),
+                          DebuglinkName);
+  if (checkFileCRC(DebugPath, CRCHash)) {
+    Result = std::string(DebugPath.str());
+    return true;
+  }
+  return false;
+}
+
+bool LLVMSymbolizer::findDebugBinary(const ArrayRef<uint8_t> BuildID,
+                                     std::string &Result) {
+  Optional<std::string> Path;
+  Path = LocalDIFetcher(Opts.DebugFileDirectory).fetchBuildID(BuildID);
+  if (Path) {
+    Result = std::move(*Path);
+    return true;
+  }
+
+  // Try caller-provided debug info fetchers.
+  for (const std::unique_ptr<DIFetcher> &Fetcher : DIFetchers) {
+    Path = Fetcher->fetchBuildID(BuildID);
+    if (Path) {
+      Result = std::move(*Path);
+      return true;
+    }
+  }
+
+  return false;
+}
+
 Expected<LLVMSymbolizer::ObjectPair>
 LLVMSymbolizer::getOrCreateObjectPair(const std::string &Path,
                                       const std::string &ArchName) {
   auto I = ObjectPairForPathArch.find(std::make_pair(Path, ArchName));
   if (I != ObjectPairForPathArch.end())
     return I->second;
 
   auto ObjOrErr = getOrCreateObject(Path, ArchName);

llvm/lib/Debuginfod/CMakeLists.txt

 # Link LibCURL if the user wants it
 if (LLVM_ENABLE_CURL)
   set(imported_libs CURL::libcurl)
 endif()
 
-add_llvm_component_library(LLVMDebuginfod
+# Note: This isn't a component, since that could potentially add a libcurl
+# dependency to libLLVM.
+add_llvm_library(LLVMDebuginfod
   Debuginfod.cpp
+  DIFetcher.cpp
   HTTPClient.cpp
 
   ADDITIONAL_HEADER_DIRS
   ${LLVM_MAIN_INCLUDE_DIR}/llvm/Debuginfod
 
   LINK_LIBS
   ${imported_libs}
 
   LINK_COMPONENTS
   Support
   )
-
-# This block is only needed for llvm-config. When we deprecate llvm-config and
-# move to using CMake export, this block can be removed.
-if(LLVM_ENABLE_CURL)
-  # CMAKE_BUILD_TYPE is only meaningful to single-configuration generators.
-  if(CMAKE_BUILD_TYPE)
-    string(TOUPPER ${CMAKE_BUILD_TYPE} build_type)
-    get_property(curl_library TARGET CURL::libcurl PROPERTY LOCATION_${build_type})
-  endif()
-  if(NOT curl_library)
-    get_property(curl_library TARGET CURL::libcurl PROPERTY LOCATION)
-  endif()
-  get_library_name(${curl_library} curl_library)
-  set(llvm_system_libs ${llvm_system_libs} "${curl_library}")
-endif()

llvm/lib/Debuginfod/DIFetcher.cpp

+//===- llvm/DebugInfod/DIFetcher.cpp - Debug info fetcher -----------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+///
+/// \file
+/// This file defines a DIFetcher implementation for obtaining debug info
+/// from debuginfod.
+///
+//===----------------------------------------------------------------------===//
+
+#include "llvm/Debuginfod/DIFetcher.h"
+
+#include "llvm/Debuginfod/Debuginfod.h"
+
+using namespace llvm;
+
+Optional<std::string>
+DebuginfodDIFetcher::fetchBuildID(ArrayRef<uint8_t> BuildID) const {
+  Expected<std::string> PathOrErr = getCachedOrDownloadDebuginfo(BuildID);
+  if (PathOrErr)
+    return *PathOrErr;
+  consumeError(PathOrErr.takeError());
+  return None;
+}

llvm/tools/llvm-debuginfod-find/CMakeLists.txt

 set(LLVM_LINK_COMPONENTS
-  Debuginfod
   Support
   )
 add_llvm_tool(llvm-debuginfod-find
   llvm-debuginfod-find.cpp
   )
+target_link_libraries(llvm-debuginfod-find PRIVATE LLVMDebuginfod)
 if(LLVM_INSTALL_BINUTILS_SYMLINKS)
   add_llvm_tool_symlink(debuginfod-find llvm-debuginfod-find)
 endif()

llvm/tools/llvm-symbolizer/CMakeLists.txt

 # FIXME: As we plan to execute llvm-symbolizer binary from compiler-rt
 # libraries, it has to be compiled for all supported targets (x86_64, i386 etc).
 # This means that we need LLVM libraries to be compiled for these
 # targets as well. Currently, there is no support for such a build strategy.
 
 set(LLVM_TARGET_DEFINITIONS Opts.td)
 tablegen(LLVM Opts.inc -gen-opt-parser-defs)
 add_public_tablegen_target(SymbolizerOptsTableGen)
 
 set(LLVM_LINK_COMPONENTS
   DebugInfoDWARF
   DebugInfoPDB
-  Debuginfod
   Demangle
   Object
   Option
   Support
   Symbolize
   )
 
 add_llvm_tool(llvm-symbolizer
   llvm-symbolizer.cpp
+
   DEPENDS
   SymbolizerOptsTableGen
   )
 
+target_link_libraries(llvm-symbolizer PRIVATE LLVMDebuginfod)
+
 add_llvm_tool_symlink(llvm-addr2line llvm-symbolizer)
 
 if(LLVM_INSTALL_BINUTILS_SYMLINKS)
   add_llvm_tool_symlink(addr2line llvm-symbolizer)
 endif()

llvm/tools/llvm-symbolizer/llvm-symbolizer.cpp

 //
 //===----------------------------------------------------------------------===//
 
 #include "Opts.inc"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Config/config.h"
 #include "llvm/DebugInfo/Symbolize/DIPrinter.h"
 #include "llvm/DebugInfo/Symbolize/Symbolize.h"
+#include "llvm/Debuginfod/DIFetcher.h"
 #include "llvm/Debuginfod/HTTPClient.h"
 #include "llvm/Option/Arg.h"
 #include "llvm/Option/ArgList.h"
 #include "llvm/Option/Option.h"
 #include "llvm/Support/COM.h"
 #include "llvm/Support/CommandLine.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/FileSystem.h"
return StringSwitch<FunctionNameKind>(A->getValue())
         .Case("none", FunctionNameKind::None)
         .Case("short", FunctionNameKind::ShortName)
         .Default(FunctionNameKind::LinkageName);
   return IsAddr2Line ? FunctionNameKind::None : FunctionNameKind::LinkageName;
 }
 
 int main(int argc, char **argv) {
   InitLLVM X(argc, argv);
-  // The HTTPClient must be initialized for use by the debuginfod client.
-  HTTPClient::initialize();
   sys::InitializeCOMRAII COM(sys::COMThreadingMode::MultiThreaded);
 
   bool IsAddr2Line = sys::path::stem(argv[0]).contains("addr2line");
   BumpPtrAllocator A;
   StringSaver Saver(A);
   SymbolizerOptTable Tbl;
   opt::InputArgList Args = parseOptions(argc, argv, IsAddr2Line, Saver, Tbl);
 
if (strcmp(A->getValue(), "GNU") == 0)
       Style = OutputStyle::GNU;
     else if (strcmp(A->getValue(), "JSON") == 0)
       Style = OutputStyle::JSON;
     else
       Style = OutputStyle::LLVM;
   }
 
   LLVMSymbolizer Symbolizer(Opts);
+
+  // Look up symbols using the debuginfod client.
+  Symbolizer.addDIFetcher(std::make_unique<DebuginfodDIFetcher>());
+  // The HTTPClient must be initialized for use by the debuginfod client.
+  HTTPClient::initialize();
+
   std::unique_ptr<DIPrinter> Printer;
   if (Style == OutputStyle::GNU)
     Printer = std::make_unique<GNUPrinter>(outs(), errs(), Config);
   else if (Style == OutputStyle::JSON)
     Printer = std::make_unique<JSONPrinter>(outs(), Config);
   else
     Printer = std::make_unique<LLVMPrinter>(outs(), errs(), Config);
 

llvm/unittests/Debuginfod/CMakeLists.txt

-set(LLVM_LINK_COMPONENTS
-  Debuginfod
-  )
-
 add_llvm_unittest(DebuginfodTests
   HTTPClientTests.cpp
   DebuginfodTests.cpp
   )
 
-target_link_libraries(DebuginfodTests PRIVATE LLVMTestingSupport)
+target_link_libraries(DebuginfodTests PRIVATE
+  LLVMDebuginfod
+  LLVMTestingSupport
+  )