feat: Support remote debug info from debuginfod

CMakePresets.json

@@ -48,6 +48,20 @@
         "POORPROF_DEPS": "vcpkg"
       }
     },
+    {
+      "name": "debug-vcpkg-local",
+      "displayName": "Debug vcpkg",
+      "description": "Debug build using deps from local instance of vcpkg",
+      "generator": "Ninja",
+      "binaryDir": "${sourceDir}/build/debug-vcpkg-local",
+      "cacheVariables": {
+        "CMAKE_BUILD_TYPE": "Debug",
+        "CMAKE_TOOLCHAIN_FILE": "/home/sergey/dev/vcpkg/scripts/buildsystems/vcpkg.cmake",
+        "CMAKE_EXPORT_COMPILE_COMMANDS": true,
+        "FORCE_COLORED_OUTPUT": true,
+        "POORPROF_DEPS": "vcpkg"
+      }
+    },
     {
       "name": "release-system",
       "displayName": "Release system",
@@ -105,6 +119,12 @@
       "configurePreset": "debug-system",
       "displayName": "Debug system",
       "description": "Debug build using deps from system"
+    },
+    {
+      "name": "debug-vcpkg-local",
+      "configurePreset": "debug-vcpkg-local",
+      "displayName": "Debug vcpkg with local installation",
+      "description": "Debug build using deps from vcpkg"
     }
   ]
 }

cmake/dependencies/system.cmake

@@ -6,6 +6,9 @@ list(APPEND POORPROF_PRIVATE_LIBRARIES fmt::fmt)
 find_package(ElfUtils REQUIRED)
 list(APPEND POORPROF_PRIVATE_LIBRARIES ${LIBDWARF_LIBRARIES} ${LIBDW_LIBRARIES} ${LIBELF_LIBRARIES})
 
+find_package(LibDebugInfoD REQUIRED)
+list(APPEND POORPROF_PRIVATE_LIBRARIES ${LIBDEBUGINFOD_LIBRARIES})
+
 find_package(absl REQUIRED)
 list(APPEND POORPROF_PRIVATE_LIBRARIES absl::flat_hash_map)
 
@@ -14,7 +17,3 @@ list(APPEND POORPROF_PRIVATE_LIBRARIES spdlog::spdlog)
 
 find_package(re2 REQUIRED)
 list(APPEND POORPROF_PRIVATE_LIBRARIES re2::re2)
-
-# set(Boost_USE_STATIC_LIBS ON)
-# find_package(Boost REQUIRED COMPONENTS iostreams)
-# list(APPEND POORPROF_PRIVATE_LIBRARIES Boost::iostreams)

cmake/dependencies/vcpkg.cmake

@@ -6,6 +6,9 @@ list(APPEND POORPROF_PRIVATE_LIBRARIES fmt::fmt)
 find_package(ElfUtils REQUIRED)
 list(APPEND POORPROF_PRIVATE_LIBRARIES ${LIBDWARF_LIBRARIES} ${LIBDW_LIBRARIES} ${LIBELF_LIBRARIES})
 
+find_package(LibDebugInfoD REQUIRED)
+list(APPEND POORPROF_PRIVATE_LIBRARIES ${LIBDEBUGINFOD_LIBRARIES})
+
 find_package(absl REQUIRED)
 list(APPEND POORPROF_PRIVATE_LIBRARIES absl::flat_hash_map)
 
@@ -14,7 +17,3 @@ list(APPEND POORPROF_PRIVATE_LIBRARIES spdlog::spdlog)
 
 find_package(re2 REQUIRED)
 list(APPEND POORPROF_PRIVATE_LIBRARIES re2::re2)
-
-# set(Boost_USE_STATIC_LIBS ON)
-# find_package(Boost REQUIRED COMPONENTS iostreams)
-# list(APPEND POORPROF_PRIVATE_LIBRARIES Boost::iostreams)

cmake/modules/FindLibDebugInfoD.cmake

@@ -0,0 +1,60 @@
+# - Try to find libdebuginfod
+# Once done this will define
+#
+#  LIBDEBUGINFOD_FOUND - system has libdebuginfod
+#  LIBDEBUGINFOD_INCLUDE_DIRS - the libdebuginfod include directory
+#  LIBDEBUGINFOD_LIBRARIES - Link these to use libdebuginfod
+#  LIBDEBUGINFOD_DEFINITIONS - Compiler switches required for using libdebuginfod
+#
+
+find_package(PkgConfig QUIET)
+
+if(PKG_CONFIG_FOUND)
+    set(PKG_CONFIG_USE_CMAKE_PREFIX_PATH ON)
+    pkg_check_modules(PC_LIBDEBUGINFOD QUIET libdebuginfod)
+endif()
+
+find_path (LIBDEBUGINFOD_INCLUDE_DIR
+    NAMES
+        debuginfod.h
+    HINTS
+        ${PC_LIBDEBUGINFOD_INCLUDE_DIRS}
+    PATHS
+        /usr/include
+        /usr/include/elfutils
+        /usr/local/include
+        /usr/local/include/elfutils
+        /opt/local/include
+        /opt/local/include/elfutils
+        /sw/include
+        /sw/include/elfutils
+    ENV CPATH)
+
+find_library (LIBDEBUGINFOD_LIBRARY
+    NAMES
+        debuginfod
+    HINTS
+        ${PC_LIBDEBUGINFOD_LIBRARY_DIRS}
+    PATHS
+        /usr/lib
+        /usr/local/lib
+        /opt/local/lib
+        /sw/lib
+    ENV LIBRARY_PATH
+    ENV LD_LIBRARY_PATH)
+
+# Transitive deps
+find_package(CURL)
+
+include (FindPackageHandleStandardArgs)
+
+find_package_handle_standard_args(LibDebugInfoD DEFAULT_MSG
+    LIBDEBUGINFOD_LIBRARY
+    LIBDEBUGINFOD_INCLUDE_DIR
+    CURL_LIBRARIES)
+
+
+mark_as_advanced(LIBDEBUGINFOD_LIBRARY LIBDEBUGINFOD_INCLUDE_DIR)
+
+set(LIBDEBUGINFOD_LIBRARIES ${LIBDEBUGINFOD_LIBRARY} ${CURL_LIBRARIES})
+set(LIBDEBUGINFOD_INCLUDE_DIRS ${LIBDEBUGINFOD_INCLUDE_DIR} )

src/cli/main.cpp

@@ -1,7 +1,9 @@
+#include "dw/debuginfod.h"
 #include "dw/gdb_index.h"
 #include "util/ctrlc.h"
 #include "util/defer.h"
 #include "util/demangle.h"
+#include "util/error.h"
 #include "util/iterator.h"
 #include "util/literals.h"
 
@@ -76,11 +78,11 @@ class DwflError : public std::runtime_error {
 };
 
 #define CHECK_DWFL(...) \
-    if (int err = (__VA_ARGS__); err != 0) { \
+    if (auto err = (__VA_ARGS__); err != 0) { \
         if (err < 0) { \
             throw DwflError{}; \
         } \
-        throw std::system_error{err, std::system_category()}; \
+        throw std::system_error{static_cast<int>(err), std::system_category()}; \
     }
 
 class Unwinder {
@@ -125,7 +127,45 @@ class Unwinder {
 
     using SymbolList = absl::InlinedVector<Symbol, 2>;
 
-public:
+    static int FindDebugInfo(Dwfl_Module* mod, void** userdata,
+            const char* modname, Dwarf_Addr base,
+            const char* file_name, const char* debuglink_name,
+            GElf_Word debuglink_crc, char** debuginfo_filename)
+    {
+        spdlog::info("FindDebugInfo for module {} (file_name: {})", modname, file_name);
+
+        // Try to use standard search facility
+        int fd = dwfl_standard_find_debuginfo(mod, userdata, modname, base, file_name, debuglink_name, debuglink_crc, debuginfo_filename);
+        if (fd >= 0) {
+            return fd;
+        }
+
+        // Try to find debuginfo using debuginfod 
+        if (mod == nullptr) {
+            return -1;
+        }
+
+        GElf_Addr bias;
+        const unsigned char* buildIdBytes;
+        int len = dwfl_module_build_id(mod, &buildIdBytes, &bias);
+        if (len < 0) {
+            throw DwflError{};
+        }
+        std::string buildId(reinterpret_cast<const char*>(buildIdBytes), len);
+
+        if (!userdata) {
+            spdlog::error("Module {} is not registered in the unwinder", modname);
+            return -1;
+        }
+        Unwinder* that = static_cast<Unwinder*>(*userdata);
+        std::optional<int> res = that->RemoteDebugInfo_.FindDebugInfo(buildId);
+        if (res) {
+            return *res;
+        }
+
+        return -1;
+    }
+
 public:
     Unwinder(Options opts)
         : Options_{std::move(opts)}
@@ -135,7 +175,7 @@ class Unwinder {
         , DebugInfoPath_{DebugInfo_ ? DebugInfo_->data() : nullptr}
         , Callbacks_{
             .find_elf = dwfl_linux_proc_find_elf,
-            .find_debuginfo = dwfl_standard_find_debuginfo,
+            .find_debuginfo = &FindDebugInfo,
             .debuginfo_path = &DebugInfoPath_,
         }
     {
@@ -144,6 +184,7 @@ class Unwinder {
         }
         InitializeDwfl();
         FillDwflReport();
+        RegisterDwflModules();
         AttachToProcess();
         ValidateAttachedPid();
     }
@@ -667,9 +708,8 @@ class Unwinder {
         dwfl_report_begin(Dwfl_);
         if (CustomMaps_) {
             FILE* maps = fopen(CustomMaps_->data(), "r");
-            DEFER {
-                fclose(maps);
-            };
+            DEFER { fclose(maps); };
+
             int code = dwfl_linux_proc_maps_report(Dwfl_, maps);
             CHECK_DWFL(code);
         } else {
@@ -679,6 +719,20 @@ class Unwinder {
         CHECK_DWFL(dwfl_report_end(Dwfl_, nullptr, nullptr));
     }
 
+    // Fill userdata in each dwfl module.
+    void RegisterDwflModules() {
+        ptrdiff_t offset = 0;
+        do {
+            offset = dwfl_getmodules(Dwfl_, +[](Dwfl_Module* mod, void**, const char* , Dwarf_Addr, void* arg) -> int {
+                void** userdata;
+                dwfl_module_info(mod, &userdata, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr);
+                *userdata = arg;
+                return DWARF_CB_OK;
+            }, this, offset);
+        } while (offset > 0);
+        CHECK_DWFL(offset);
+    }
+
     void AttachToProcess() {
         CHECK_DWFL(dwfl_linux_proc_attach(Dwfl_, Pid_, /*assume_ptrace_stopped=*/false));
     }
@@ -704,6 +758,8 @@ class Unwinder {
     absl::flat_hash_map<pid_t, std::string> ThreadNameCache_;
     absl::flat_hash_map<size_t, TraceInfo> Traces_;
     absl::flat_hash_map<Dwfl_Module*, std::unique_ptr<ObjectFile>> Modules_;
+
+    RemoteDebugInfo RemoteDebugInfo_;
 };
 
 #undef CHECK_DWFL

src/dw/CMakeLists.txt

@@ -1,10 +1,12 @@
 set(
     SOURCES
+    debuginfod.cpp
     dw.cpp
 )
 
 set(
     HEADERS
+    debuginfod.h
     dw.h
 )
 
@@ -16,7 +18,7 @@ set_target_properties(poorprof_dw PROPERTIES
 )
 
 target_link_libraries(poorprof_dw
-    PUBLIC ${LIBDWARF_LIBRARIES}
+    PUBLIC ${LIBDWARF_LIBRARIES} ${LIBDEBUGINFOD_LIBRARIES}
     PUBLIC ${CMAKE_DL_LIBS}
 )
 

src/dw/debuginfod.cpp

@@ -0,0 +1,59 @@
+#include "debuginfod.h"
+
+#include "util/assert.h"
+#include "util/defer.h"
+
+#include <spdlog/spdlog.h>
+
+#include <elfutils/debuginfod.h>
+
+#ifdef __linux__
+#include <dlfcn.h>
+#endif
+
+
+namespace poorprof::dw {
+
+RemoteDebugInfo::RemoteDebugInfo()
+{
+    Client_ = debuginfod_begin();
+    if (Client_) {
+        debuginfod_set_progressfn(Client_, +[](debuginfod_client*, long a, long b) {
+            spdlog::info("Loading {}/{}", a, b);
+            return 0;
+        });
+        spdlog::info("Successfully initialized debuginfod client");
+    } else {
+        spdlog::info("Failed to initialize debuginfod client");
+    }
+}
+
+RemoteDebugInfo::~RemoteDebugInfo() {
+    if (Client_) {
+        debuginfod_end(std::exchange(Client_, nullptr));
+    }
+}
+
+std::optional<FileDescriptor> RemoteDebugInfo::FindDebugInfo(std::string buildId) {
+    if (!Client_) {
+        return std::nullopt;
+    }
+
+    char* path = nullptr;
+    DEFER {
+        ::free(path);
+    };
+
+    spdlog::info("Loading remote debug info");
+    const unsigned char* ptr = reinterpret_cast<const unsigned char*>(buildId.data());
+    int fd = debuginfod_find_debuginfo(Client_, ptr, buildId.size(), &path);
+    if (fd <= 0) {
+        spdlog::warn("Failed to find remote debug info: {}", strerror(-fd));
+        return std::nullopt;
+    }
+
+    spdlog::info("Successfully fetched remote debug info");
+    return fd;
+}
+
+} // namespace poorprof::dw

src/dw/debuginfod.h

@@ -0,0 +1,30 @@
+#pragma once
+
+#include "util/fs.h"
+#include "util/noncopyable.h"
+
+#include <absl/container/flat_hash_map.h>
+
+#include <optional>
+
+
+struct debuginfod_client;
+
+namespace poorprof::dw {
+
+using FileDescriptor = int;
+
+class RemoteDebugInfo : util::NonCopyable {
+    struct DebugInfoLib;
+
+public:
+    RemoteDebugInfo();
+    ~RemoteDebugInfo();
+
+    std::optional<FileDescriptor> FindDebugInfo(std::string buildId);
+
+private:
+    debuginfod_client* Client_ = nullptr;
+};
+
+} // namespace poorprof::dw

src/util/fs.h

@@ -0,0 +1,10 @@
+#pragma once
+
+#include <filesystem>
+
+
+namespace poorprof {
+
+namespace fs = std::filesystem;
+
+} // namespace poorprof