Merge pull request #11657 from ClickHouse/dont-use-elf-if-foreign

Don't use debug info from ELF file if it doesn't correspond to the running binary.

src/Common/Elf.cpp

@@ -54,6 +54,18 @@ Elf::Elf(const std::string & path)
         throw Exception("The ELF is truncated (section names string table points after end of file)", ErrorCodes::CANNOT_PARSE_ELF);
 
     section_names = reinterpret_cast<const char *>(mapped + section_names_offset);
+
+    /// Get program headers
+
+    ElfOff program_header_offset = header->e_phoff;
+    uint16_t program_header_num_entries = header->e_phnum;
+
+    if (!program_header_offset
+        || !program_header_num_entries
+        || program_header_offset + program_header_num_entries * sizeof(ElfPhdr) > elf_size)
+        throw Exception("The ELF is truncated (program header points after end of file)", ErrorCodes::CANNOT_PARSE_ELF);
+
+    program_headers = reinterpret_cast<const ElfPhdr *>(mapped + program_header_offset);
 }
 
 
@@ -104,6 +116,40 @@ std::optional<Elf::Section> Elf::findSectionByName(const char * name) const
 }
 
 
+String Elf::getBuildID() const
+{
+    for (size_t idx = 0; idx < header->e_phnum; ++idx)
+    {
+        const ElfPhdr & phdr = program_headers[idx];
+
+        if (phdr.p_type == PT_NOTE)
+            return getBuildID(mapped + phdr.p_offset, phdr.p_filesz);
+    }
+    return {};
+}
+
+
+String Elf::getBuildID(const char * nhdr_pos, size_t size)
+{
+    const char * nhdr_end = nhdr_pos + size;
+
+    while (nhdr_pos < nhdr_end)
+    {
+        const ElfNhdr & nhdr = *reinterpret_cast<const ElfNhdr *>(nhdr_pos);
+
+        nhdr_pos += sizeof(ElfNhdr) + nhdr.n_namesz;
+        if (nhdr.n_type == NT_GNU_BUILD_ID)
+        {
+            const char * build_id = nhdr_pos;
+            return {build_id, nhdr.n_descsz};
+        }
+        nhdr_pos += nhdr.n_descsz;
+    }
+
+    return {};
+}
+
+
 const char * Elf::Section::name() const
 {
     if (!elf.section_names)

src/Common/Elf.h

@@ -17,6 +17,7 @@ using ElfEhdr = ElfW(Ehdr);
 using ElfOff = ElfW(Off);
 using ElfPhdr = ElfW(Phdr);
 using ElfShdr = ElfW(Shdr);
+using ElfNhdr = ElfW(Nhdr);
 using ElfSym = ElfW(Sym);
 
 
@@ -53,12 +54,18 @@ public:
     const char * end() const { return mapped + elf_size; }
     size_t size() const { return elf_size; }
 
+    /// Obtain build id from PT_NOTES section of program headers. Return empty string if does not exist.
+    /// The string is returned in binary. Note that "readelf -n ./clickhouse-server" prints it in hex.
+    String getBuildID() const;
+    static String getBuildID(const char * nhdr_pos, size_t size);
+
 private:
     MMapReadBufferFromFile in;
     size_t elf_size;
     const char * mapped;
     const ElfEhdr * header;
     const ElfShdr * section_headers;
+    const ElfPhdr * program_headers;
     const char * section_names = nullptr;
 };
 

src/Common/SymbolIndex.cpp

@@ -196,6 +196,20 @@ void collectSymbolsFromProgramHeaders(dl_phdr_info * info,
 }
 
 
+String getBuildIDFromProgramHeaders(dl_phdr_info * info)
+{
+    for (size_t header_index = 0; header_index < info->dlpi_phnum; ++header_index)
+    {
+        const ElfPhdr & phdr = info->dlpi_phdr[header_index];
+        if (phdr.p_type != PT_NOTE)
+            continue;
+
+        return Elf::getBuildID(reinterpret_cast<const char *>(info->dlpi_addr + phdr.p_vaddr), phdr.p_memsz);
+    }
+    return {};
+}
+
+
 void collectSymbolsFromELFSymbolTable(
     dl_phdr_info * info,
     const Elf & elf,
@@ -283,8 +297,31 @@ void collectSymbolsFromELF(dl_phdr_info * info,
 
     object_name = std::filesystem::exists(debug_info_path) ? debug_info_path : canonical_path;
 
+    /// But we have to compare Build ID to check that debug info corresponds to the same executable.
+    String our_build_id = getBuildIDFromProgramHeaders(info);
+
     SymbolIndex::Object object;
     object.elf = std::make_unique<Elf>(object_name);
+
+    String file_build_id = object.elf->getBuildID();
+
+    if (our_build_id != file_build_id)
+    {
+        /// If debug info doesn't correspond to our binary, fallback to the info in our binary.
+        if (object_name != canonical_path)
+        {
+            object_name = canonical_path;
+            object.elf = std::make_unique<Elf>(object_name);
+
+            /// But it can still be outdated, for example, if executable file was deleted from filesystem and replaced by another file.
+            file_build_id = object.elf->getBuildID();
+            if (our_build_id != file_build_id)
+                return;
+        }
+        else
+            return;
+    }
+
     object.address_begin = reinterpret_cast<const void *>(info->dlpi_addr);
     object.address_end = reinterpret_cast<const void *>(info->dlpi_addr + object.elf->size());
     object.name = object_name;