Add support for symbol name remapping

Author: brenoguim

src/patchelf.cc

@@ -17,16 +17,18 @@
  */
 
 #include <algorithm>
+#include <fstream>
 #include <limits>
 #include <map>
 #include <memory>
+#include <optional>
 #include <set>
 #include <sstream>
 #include <stdexcept>
 #include <string>
 #include <unordered_map>
+#include <unordered_set>
 #include <vector>
-#include <optional>
 
 #include <cassert>
 #include <cerrno>
@@ -553,6 +555,27 @@ std::optional<std::reference_wrapper<Elf_Shdr>> ElfFile<ElfFileParamNames>::tryF
     return {};
 }
 
+template<ElfFileParams>
+template<class T>
+span<T> ElfFile<ElfFileParamNames>::getSectionSpan(const Elf_Shdr & shdr) const
+{
+    return span((T*)(fileContents->data() + rdi(shdr.sh_offset)), rdi(shdr.sh_size)/sizeof(T));
+}
+
+template<ElfFileParams>
+template<class T>
+span<T> ElfFile<ElfFileParamNames>::getSectionSpan(const SectionName & sectionName)
+{
+    return getSectionSpan<T>(findSectionHeader(sectionName));
+}
+
+template<ElfFileParams>
+template<class T>
+span<T> ElfFile<ElfFileParamNames>::tryGetSectionSpan(const SectionName & sectionName)
+{
+    auto shdrOpt = tryFindSectionHeader(sectionName);
+    return shdrOpt ? getSectionSpan<T>(*shdrOpt) : span<T>();
+}
 
 template<ElfFileParams>
 unsigned int ElfFile<ElfFileParamNames>::getSectionIndex(const SectionName & sectionName)
@@ -1861,6 +1884,222 @@ void ElfFile<ElfFileParamNames>::addDebugTag()
     changed = true;
 }
 
+static uint32_t gnuHash(std::string_view name) {
+    uint32_t h = 5381;
+    for (uint8_t c : name)
+        h = ((h << 5) + h) + c;
+    return h;
+}
+
+template<ElfFileParams>
+auto ElfFile<ElfFileParamNames>::GnuHashTable::parse(span<char> sectionData) -> GnuHashTable
+{
+    auto hdr = (Header*)sectionData.begin();
+    auto bloomFilters = span((BloomWord*)(hdr+1), hdr->maskwords);
+    auto buckets = span((uint32_t*)bloomFilters.end(), hdr->numBuckets);
+    auto table = span(buckets.end(), ((uint32_t*)sectionData.end()) - buckets.end());
+    return GnuHashTable{*hdr, bloomFilters, buckets, table};
+}
+
+template<ElfFileParams>
+void ElfFile<ElfFileParamNames>::rebuildGnuHashTable(const char* strTab, span<Elf_Sym> dynsyms)
+{
+    auto sectionData = tryGetSectionSpan<char>(".gnu.hash");
+    if (!sectionData)
+        return;
+
+    auto ght = GnuHashTable::parse(sectionData);
+
+    // Only work with the last "m_table.size()" symbols from dynsyms which are the
+    // symbols that belong to the hash table
+    auto firstSymIdx = dynsyms.size() - ght.m_table.size();
+    auto symsToInsert = span(dynsyms.begin() + firstSymIdx, dynsyms.end());
+
+    // Only use the range of symbol versions that will be changed
+    auto versyms = tryGetSectionSpan<Elf_Versym>(".gnu.version");
+    if (versyms)
+        versyms = span(versyms.begin() + firstSymIdx, versyms.end());
+
+    struct Entry
+    {
+        Elf_Sym sym;
+        uint32_t hash, bucketIdx;
+        Elf_Versym vsym {};
+        uint32_t originalPos;
+    };
+
+    std::vector<Entry> entries;
+    entries.reserve(symsToInsert.size());
+
+    uint32_t pos = 0;
+    for (auto& sym : symsToInsert)
+    {
+        Entry e;
+        e.sym = sym;
+        e.hash = gnuHash(strTab + rdi(e.sym.st_name));
+        e.originalPos = pos++;
+        e.bucketIdx = e.hash % ght.m_hdr.numBuckets;
+        if (versyms)
+            e.vsym = versyms[e.originalPos];
+
+        entries.push_back(e);
+    }
+
+    // Sort the entries based on the buckets. This is a requirement for gnu hash table to work
+    std::sort(entries.begin(), entries.end(), [&] (auto& l, auto& r) {
+        return l.bucketIdx < r.bucketIdx;
+    });
+
+    // Update the symbol table with the new order and
+    // all tables that refer to symbols through indexes in the symbol table
+    std::vector<uint32_t> old2new(entries.size());
+    for (size_t i = 0; i < entries.size(); ++i)
+    {
+        auto& entry = entries[i];
+        old2new[entry.originalPos] = i + firstSymIdx;
+        symsToInsert[i] = entry.sym;
+        if (versyms)
+            versyms[i] = entry.vsym;
+    }
+
+    auto fixRelocationTable = [&old2new, firstSymIdx, this] <class ER> (auto& hdr)
+    {
+        auto rela = getSectionSpan<ER>(hdr);
+        for (auto& r : rela)
+        {
+            auto info = rdi(r.r_info);
+            auto oldSymIdx = rel_getSymId(info);
+            if (oldSymIdx >= firstSymIdx)
+            {
+                auto newSymIdx = old2new[oldSymIdx - firstSymIdx];
+                if (newSymIdx != oldSymIdx) {
+                    wri(r.r_info, rel_setSymId(info, newSymIdx));
+                }
+            }
+        }
+    };
+
+    for (unsigned int i = 1; i < rdi(hdr()->e_shnum); ++i)
+    {
+        auto& shdr = shdrs.at(i);
+        auto shtype = rdi(shdr.sh_type);
+        if (shtype == SHT_REL)
+            fixRelocationTable.template operator()<Elf_Rel>(shdr);
+        else if (shtype == SHT_RELA)
+            fixRelocationTable.template operator()<Elf_Rela>(shdr);
+    }
+
+    // Update bloom filters
+    std::fill(ght.m_bloomFilters.begin(), ght.m_bloomFilters.end(), 0); 
+    for (size_t i = 0; i < entries.size(); ++i)
+    {
+        auto h = entries[i].hash;
+        size_t idx = (h / ElfWordSize) % ght.m_bloomFilters.size();
+        auto val = rdi(ght.m_bloomFilters[idx]);
+        val |= uint64_t(1) << (h % ElfWordSize);
+        val |= uint64_t(1) << ((h >> ght.m_hdr.shift2) % ElfWordSize);
+        wri(ght.m_bloomFilters[idx], val);
+    }
+
+    // Fill buckets
+    std::fill(ght.m_buckets.begin(), ght.m_buckets.end(), 0); 
+    for (size_t i = 0; i < entries.size(); ++i)
+    {
+        auto symBucketIdx = entries[i].bucketIdx;
+        if (!ght.m_buckets[symBucketIdx])
+            ght.m_buckets[symBucketIdx] = i + ght.m_hdr.symndx;
+    }
+
+    // Fill hash table
+    for (size_t i = 0; i < entries.size(); ++i)
+    {
+        auto& n = entries[i];
+        bool isLast = (i == entries.size() - 1) || (n.bucketIdx != entries[i+1].bucketIdx);
+        // Add hash with first bit indicating end of chain
+        ght.m_table[i] = isLast ? (n.hash | 1) : (n.hash & ~1);
+    }
+}
+
+static uint32_t sysvHash(std::string_view name) {
+    uint32_t h = 0;
+    for (uint8_t c : name)
+    {
+        h = (h << 4) + c;
+        uint32_t g = h & 0xf0000000;
+        if (g != 0)
+            h ^= g >> 24;
+        h &= ~g;
+    }
+    return h;
+}
+
+template<ElfFileParams>
+auto ElfFile<ElfFileParamNames>::HashTable::parse(span<char> sectionData) -> HashTable
+{
+    auto hdr = (Header*)sectionData.begin();
+    auto buckets = span((uint32_t*)(hdr+1), hdr->numBuckets);
+    auto table = span(buckets.end(), ((uint32_t*)sectionData.end()) - buckets.end());
+    return HashTable{*hdr, buckets, table};
+}
+
+template<ElfFileParams>
+void ElfFile<ElfFileParamNames>::rebuildHashTable(const char* strTab, span<Elf_Sym> dynsyms)
+{
+    auto sectionData = tryGetSectionSpan<char>(".hash");
+    if (!sectionData)
+        return;
+
+    auto ht = HashTable::parse(sectionData);
+
+    std::fill(ht.m_buckets.begin(), ht.m_buckets.end(), 0);
+    std::fill(ht.m_chain.begin(), ht.m_chain.end(), 0);
+
+    auto symsToInsert = span(dynsyms.end() - ht.m_chain.size(), dynsyms.end());
+
+    for (auto& sym : symsToInsert)
+    {
+        auto name = strTab + rdi(sym.st_name);
+        uint32_t i = &sym - dynsyms.begin();
+        uint32_t hash = sysvHash(name) % ht.m_buckets.size();
+        ht.m_chain[i] = ht.m_buckets[hash];
+        ht.m_buckets[hash] = i;
+    }
+}
+
+template<ElfFileParams>
+void ElfFile<ElfFileParamNames>::renameDynamicSymbols(const std::unordered_map<std::string_view, std::string>& remap)
+{
+    auto dynsyms = getSectionSpan<Elf_Sym>(".dynsym");
+    auto strTab = getSectionSpan<char>(".dynstr");
+
+    std::vector<char> extraStrings;
+    extraStrings.reserve(remap.size() * 30); // Just an estimate
+    for (size_t i = 0; i < dynsyms.size(); i++)
+    {
+        auto& dynsym = dynsyms[i];
+        std::string_view name = &strTab[rdi(dynsym.st_name)];
+        auto it = remap.find(name);
+        if (it != remap.end())
+        {
+            wri(dynsym.st_name, strTab.size() + extraStrings.size());
+            auto& newName = it->second;
+            extraStrings.insert(extraStrings.end(), newName.data(), newName.data() + newName.size() + 1);
+            changed = true;
+        }
+    }
+
+    if (changed)
+    {
+        auto& newSec = replaceSection(".dynstr", strTab.size() + extraStrings.size());
+        std::copy(extraStrings.begin(), extraStrings.end(), newSec.begin() + strTab.size());
+
+        rebuildGnuHashTable(newSec.data(), dynsyms);
+        rebuildHashTable(newSec.data(), dynsyms);
+    }
+
+    this->rewriteSections();
+}
+
 template<ElfFileParams>
 void ElfFile<ElfFileParamNames>::clearSymbolVersions(const std::set<std::string> & syms)
 {
@@ -1904,12 +2143,15 @@ static bool removeRPath = false;
 static bool setRPath = false;
 static bool addRPath = false;
 static bool addDebugTag = false;
+static bool renameDynamicSymbols = false;
 static bool printRPath = false;
 static std::string newRPath;
 static std::set<std::string> neededLibsToRemove;
 static std::map<std::string, std::string> neededLibsToReplace;
 static std::set<std::string> neededLibsToAdd;
 static std::set<std::string> symbolsToClearVersion;
+static std::unordered_map<std::string_view, std::string> symbolsToRename;
+static std::unordered_set<std::string> symbolsToRenameKeys;
 static bool printNeeded = false;
 static bool noDefaultLib = false;
 
@@ -1959,6 +2201,9 @@ static void patchElf2(ElfFile && elfFile, const FileContents & fileContents, con
     if (addDebugTag)
         elfFile.addDebugTag();
 
+    if (renameDynamicSymbols)
+        elfFile.renameDynamicSymbols(symbolsToRename);
+
     if (elfFile.isChanged()){
         writeFile(fileName, elfFile.fileContents);
     } else if (alwaysWrite) {
@@ -1978,9 +2223,9 @@ static void patchElf()
         const std::string & outputFileName2 = outputFileName.empty() ? fileName : outputFileName;
 
         if (getElfType(fileContents).is32Bit)
-            patchElf2(ElfFile<Elf32_Ehdr, Elf32_Phdr, Elf32_Shdr, Elf32_Addr, Elf32_Off, Elf32_Dyn, Elf32_Sym, Elf32_Verneed, Elf32_Versym>(fileContents), fileContents, outputFileName2);
+            patchElf2(ElfFile<Elf32_Ehdr, Elf32_Phdr, Elf32_Shdr, Elf32_Addr, Elf32_Off, Elf32_Dyn, Elf32_Sym, Elf32_Verneed, Elf32_Versym, Elf32_Rel, Elf32_Rela>(fileContents), fileContents, outputFileName2);
         else
-            patchElf2(ElfFile<Elf64_Ehdr, Elf64_Phdr, Elf64_Shdr, Elf64_Addr, Elf64_Off, Elf64_Dyn, Elf64_Sym, Elf64_Verneed, Elf64_Versym>(fileContents), fileContents, outputFileName2);
+            patchElf2(ElfFile<Elf64_Ehdr, Elf64_Phdr, Elf64_Shdr, Elf64_Addr, Elf64_Off, Elf64_Dyn, Elf64_Sym, Elf64_Verneed, Elf64_Versym, Elf64_Rel, Elf64_Rela>(fileContents), fileContents, outputFileName2);
     }
 }
 
@@ -2019,6 +2264,7 @@ void showHelp(const std::string & progName)
   [--no-sort]\t\tDo not sort program+section headers; useful for debugging patchelf.\n\
   [--clear-symbol-version SYMBOL]\n\
   [--add-debug-tag]\n\
+  [--rename-dynamic-symbols NAME_MAP_FILE]\tRenames dynamic symbols. The name map file should contain two symbols (old_name new_name) per line\n\
   [--output FILE]\n\
   [--debug]\n\
   [--version]\n\
@@ -2141,6 +2387,25 @@ int mainWrapped(int argc, char * * argv)
         else if (arg == "--add-debug-tag") {
             addDebugTag = true;
         }
+        else if (arg == "--rename-dynamic-symbols") {
+            renameDynamicSymbols = true;
+            if (++i == argc) error("missing argument");
+
+            std::ifstream infile(argv[i]);
+            if (!infile) error(fmt("Cannot open map file ", argv[i]));
+
+            std::string from, to;
+            while (true)
+            {
+                if (!(infile >> from))
+                    break;
+                if (!(infile >> to))
+                    error("Odd number of symbols in map file");
+                if (symbolsToRenameKeys.count(from))
+                    error(fmt("Symbol appears twice in the map file: ", from.c_str()));
+                symbolsToRename[*symbolsToRenameKeys.insert(from).first] = to;
+            }
+        }
         else if (arg == "--help" || arg == "-h" ) {
             showHelp(argv[0]);
             return 0;