Implement --clean-strtab

Author: brenoguim

patchelf.1

@@ -131,6 +131,11 @@ old_name new_name
 
 Symbol names do not contain version specifier that are also shown in the output of the nm -D command from binutils. So instead of the name write@GLIBC_2.2.5 it is just write.
 
+.IP "--clean-strtab"
+Regenerates the ".dynstr" section removing all unused strings.
+
+Notice this may actually increase the size of ths section because it will undo string sharing between "vprintf" and "printf" some linkers create.
+
 .IP "--output FILE"
 Set the output file name.  If not specified, the input will be modified in place.
 

src/patchelf.cc

@@ -29,6 +29,7 @@
 #include <string_view>
 #include <unordered_map>
 #include <unordered_set>
+#include <variant>
 #include <vector>
 
 #include <cassert>
@@ -1903,6 +1904,88 @@ void ElfFile<ElfFileParamNames>::noDefaultLib()
     changed = true;
 }
 
+template<ElfFileParams>
+void ElfFile<ElfFileParamNames>::cleanStrTab()
+{
+    std::unordered_map<std::string, unsigned> requiredStrs2Idx {{"",0}};
+
+    // A collection of pairs, each containing:
+    //     - a pointer to the field that refer to str indices
+    //     - a pointer to the new index in `requiredStrs2Idx`
+    using StrIndexPtr = std::variant<Elf32_Word*, Elf64_Xword*>;
+    std::vector<std::pair<StrIndexPtr, unsigned*>> strRefs;
+
+    auto& strTabHdr = findSectionHeader(".dynstr");
+    auto strTab = getSectionSpan<char>(strTabHdr);
+
+    // Utility to collect a string index field from any table
+    auto collect = [&] (auto& idx) {
+        auto [it, _] = requiredStrs2Idx.emplace(&strTab[rdi(idx)], 0);
+        strRefs.emplace_back(&idx, &it->second);
+    };
+
+    // Iterate on tables known to store references to .dynstr
+    for (auto& sym : tryGetSectionSpan<Elf_Sym>(".dynsym"))
+        collect(sym.st_name);
+
+    for (auto& dyn : tryGetSectionSpan<Elf_Dyn>(".dynamic"))
+        switch (rdi(dyn.d_tag))
+        {
+            case DT_NEEDED:
+            case DT_SONAME:
+            case DT_RPATH:
+            case DT_RUNPATH: collect(dyn.d_un.d_val);
+            default:;
+        }
+
+    if (auto verdHdr = tryFindSectionHeader(".gnu.version_d"))
+    {
+        // Only collect fields if they use the strtab we are cleaning
+        if (&shdrs.at(rdi(verdHdr->get().sh_link)) == &strTabHdr)
+            forAll_ElfVer(getSectionSpan<Elf_Verdef>(*verdHdr),
+                [] (auto& /*vd*/) {},
+                [&] (auto& vda) { collect(vda.vda_name); }
+            );
+    }
+
+    if (auto vernHdr = tryFindSectionHeader(".gnu.version_r"))
+    {
+        // Only collect fields if they use the strtab we are cleaning
+        if (&shdrs.at(rdi(vernHdr->get().sh_link)) == &strTabHdr)
+            forAll_ElfVer(getSectionSpan<Elf_Verneed>(*vernHdr),
+                [&] (auto& vn) { collect(vn.vn_file); },
+                [&] (auto& vna) { collect(vna.vna_name); }
+            );
+    }
+
+    // Iterate on all required strings calculating the new position
+    size_t curIdx = 1;
+    for (auto& [str,idx] : requiredStrs2Idx)
+    {
+        idx = curIdx;
+        curIdx += str.size() + /*null terminator*/1;
+    }
+
+    // Add required strings to the new dynstr section
+    auto& newStrSec = replaceSection(".dynstr", curIdx);
+    for (auto& [str,idx] : requiredStrs2Idx)
+        std::copy(str.begin(), str.end()+1, &newStrSec[idx]);
+
+    // Iterate on all fields on all tables setting the new index value 
+    for (auto& [oldIndexPtr, newIdxPtr_] : strRefs)
+    {
+        auto newIdxPtr = newIdxPtr_; // Some compilers complain about 
+                                     // capturing structured bindings
+        std::visit(
+            [&] (auto* ptr) { wri(*ptr, *newIdxPtr); },
+            oldIndexPtr
+        );
+    }
+
+    changed = true;
+    this->rewriteSections();
+}
+
 template<ElfFileParams>
 void ElfFile<ElfFileParamNames>::addDebugTag()
 {
@@ -2271,6 +2354,7 @@ static bool removeRPath = false;
 static bool setRPath = false;
 static bool addRPath = false;
 static bool addDebugTag = false;
+static bool cleanStrTab = false;
 static bool renameDynamicSymbols = false;
 static bool printRPath = false;
 static std::string newRPath;
@@ -2342,6 +2426,9 @@ static void patchElf2(ElfFile && elfFile, const FileContents & fileContents, con
     if (renameDynamicSymbols)
         elfFile.renameDynamicSymbols(symbolsToRename);
 
+    if (cleanStrTab)
+        elfFile.cleanStrTab();
+
     if (elfFile.isChanged()){
         writeFile(fileName, elfFile.fileContents);
     } else if (alwaysWrite) {
@@ -2361,9 +2448,9 @@ 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, Elf32_Rel, Elf32_Rela, 32>(fileContents), fileContents, outputFileName2);
+            patchElf2(ElfFile<Elf32_Ehdr, Elf32_Phdr, Elf32_Shdr, Elf32_Addr, Elf32_Off, Elf32_Dyn, Elf32_Sym, Elf32_Versym, Elf32_Verdef, Elf32_Verdaux, Elf32_Verneed, Elf32_Vernaux, Elf32_Rel, Elf32_Rela, 32>(fileContents), fileContents, outputFileName2);
         else
-            patchElf2(ElfFile<Elf64_Ehdr, Elf64_Phdr, Elf64_Shdr, Elf64_Addr, Elf64_Off, Elf64_Dyn, Elf64_Sym, Elf64_Verneed, Elf64_Versym, Elf64_Rel, Elf64_Rela, 64>(fileContents), fileContents, outputFileName2);
+            patchElf2(ElfFile<Elf64_Ehdr, Elf64_Phdr, Elf64_Shdr, Elf64_Addr, Elf64_Off, Elf64_Dyn, Elf64_Sym, Elf64_Versym, Elf64_Verdef, Elf64_Verdaux, Elf64_Verneed, Elf64_Vernaux, Elf64_Rel, Elf64_Rela, 64>(fileContents), fileContents, outputFileName2);
     }
 }
 
@@ -2406,6 +2493,7 @@ static void showHelp(const std::string & progName)
   [--clear-execstack]\n\
   [--set-execstack]\n\
   [--rename-dynamic-symbols NAME_MAP_FILE]\tRenames dynamic symbols. The map file should contain two symbols (old_name new_name) per line\n\
+  [--clean-strtab]\n\
   [--output FILE]\n\
   [--debug]\n\
   [--version]\n\
@@ -2537,6 +2625,9 @@ static int mainWrapped(int argc, char * * argv)
         else if (arg == "--add-debug-tag") {
             addDebugTag = true;
         }
+        else if (arg == "--clean-strtab") {
+            cleanStrTab = true;
+        }
         else if (arg == "--rename-dynamic-symbols") {
             renameDynamicSymbols = true;
             if (++i == argc) error("missing argument");

src/patchelf.h

@@ -10,8 +10,8 @@
 
 using FileContents = std::shared_ptr<std::vector<unsigned char>>;
 
-#define ElfFileParams class Elf_Ehdr, class Elf_Phdr, class Elf_Shdr, class Elf_Addr, class Elf_Off, class Elf_Dyn, class Elf_Sym, class Elf_Verneed, class Elf_Versym, class Elf_Rel, class Elf_Rela, unsigned ElfClass
-#define ElfFileParamNames Elf_Ehdr, Elf_Phdr, Elf_Shdr, Elf_Addr, Elf_Off, Elf_Dyn, Elf_Sym, Elf_Verneed, Elf_Versym, Elf_Rel, Elf_Rela, ElfClass
+#define ElfFileParams class Elf_Ehdr, class Elf_Phdr, class Elf_Shdr, class Elf_Addr, class Elf_Off, class Elf_Dyn, class Elf_Sym, class Elf_Versym, class Elf_Verdef, class Elf_Verdaux, class Elf_Verneed, class Elf_Vernaux, class Elf_Rel, class Elf_Rela, unsigned ElfClass
+#define ElfFileParamNames Elf_Ehdr, Elf_Phdr, Elf_Shdr, Elf_Addr, Elf_Off, Elf_Dyn, Elf_Sym, Elf_Versym, Elf_Verdef, Elf_Verdaux, Elf_Verneed, Elf_Vernaux, Elf_Rel, Elf_Rela, ElfClass
 
 template<class T>
 struct span
@@ -175,6 +175,8 @@ class ElfFile
 
     void modifyExecstack(ExecstackMode op);
 
+    void cleanStrTab();
+
 private:
     struct GnuHashTable {
         using BloomWord = Elf_Addr;
@@ -226,7 +228,6 @@ class ElfFile
     void changeRelocTableSymIds(const Elf_Shdr& shdr, RemapFn&& old2newSymId)
     {
         static_assert(std::is_same_v<ElfRelType, Elf_Rel> || std::is_same_v<ElfRelType, Elf_Rela>);
-
         for (auto& r : getSectionSpan<ElfRelType>(shdr))
         {
             auto info = rdi(r.r_info);
@@ -237,6 +238,37 @@ class ElfFile
         }
     }
 
+    template<class T, class U>
+    auto follow(U* ptr, size_t offset) -> T* {
+        return offset ? (T*)(((char*)ptr)+offset) : nullptr;
+    };
+
+    template<class VdFn, class VaFn>
+    void forAll_ElfVer(span<Elf_Verdef> vdspan, VdFn&& vdfn, VaFn&& vafn)
+    {
+        auto* vd = vdspan.begin();
+        for (; vd; vd = follow<Elf_Verdef>(vd, rdi(vd->vd_next)))
+        {
+            vdfn(*vd);
+            auto va = follow<Elf_Verdaux>(vd, rdi(vd->vd_aux));
+            for (; va; va = follow<Elf_Verdaux>(va, rdi(va->vda_next)))
+                vafn(*va);
+        }
+    }
+
+    template<class VnFn, class VaFn>
+    void forAll_ElfVer(span<Elf_Verneed> vnspan, VnFn&& vnfn, VaFn&& vafn)
+    {
+        auto* vn = vnspan.begin();
+        for (; vn; vn = follow<Elf_Verneed>(vn, rdi(vn->vn_next)))
+        {
+            vnfn(*vn);
+            auto va = follow<Elf_Vernaux>(vn, rdi(vn->vn_aux));
+            for (; va; va = follow<Elf_Vernaux>(va, rdi(va->vna_next)))
+                vafn(*va);
+        }
+    }
+
     /* Convert an integer in big or little endian representation (as
        specified by the ELF header) to this platform's integer
        representation. */

tests/clean-strtab.sh

@@ -0,0 +1,32 @@
+#! /bin/sh -e
+SCRATCH=scratch/$(basename "$0" .sh)
+PATCHELF=$(readlink -f "../src/patchelf")
+
+rm -rf "${SCRATCH}"
+mkdir -p "${SCRATCH}"
+
+cp libfoo.so "${SCRATCH}/"
+
+cd "${SCRATCH}"
+
+the_string=VERY_SPECIFIC_STRING
+check_count() {
+    count="$(strings libfoo.so | grep -c $the_string || true)"
+    expected=$1
+    echo "####### Checking count. Expected: $expected"
+    [ "$count" = "$expected" ] || exit 1
+}
+
+check_count 0
+
+${PATCHELF} --clean-strtab libfoo.so
+check_count 0
+
+${PATCHELF} --add-needed $the_string libfoo.so
+check_count 1
+
+${PATCHELF} --remove-needed $the_string libfoo.so
+check_count 1
+
+${PATCHELF} --clean-strtab libfoo.so
+check_count 0