Implement --clean-strtab

NixOS · Feb 20, 2023 · 6bfd015 · 6bfd015
1 parent 6ba561f
commit 6bfd015
Show file tree

Hide file tree

Showing 4 changed files with 147 additions and 4 deletions.
diff --git a/src/patchelf.cc b/src/patchelf.cc
@@ -28,6 +28,7 @@
 #include <string>
 #include <unordered_map>
 #include <unordered_set>
+#include <variant>
 #include <vector>
 
 #include <cassert>
@@ -1855,7 +1856,78 @@ void ElfFile<ElfFileParamNames>::noDefaultLib()
 template<ElfFileParams>
 void ElfFile<ElfFileParamNames>::cleanStrTab()
 {
+    std::unordered_map<std::string, unsigned> requiredStrs2Idx {{"",0}};
 
+    // A collection of pointers to the fields that refer to str indices
+    // and a pointer to the new index value to be calculated
+    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()+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.begin()+idx);
+
+    // Iterate on all fields on all tables setting the new index value 
+    for (auto& [oldIndexPtr, newIdxPtr] : strRefs)
+        std::visit(
+            [&] (auto* ptr) { wri(*ptr, *newIdxPtr); },
+            oldIndexPtr
+        );
+
+    changed = true;
+    this->rewriteSections();
 }
 
 template<ElfFileParams>
@@ -2230,9 +2302,9 @@ 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);
     }
 }
 

diff --git a/src/patchelf.h b/src/patchelf.h
@@ -1,7 +1,7 @@
 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
@@ -206,6 +206,44 @@ class ElfFile
         return info;
     }
 
+    template<class T, class U>
+    auto follow(U* ptr, size_t offset) -> T* {
+        return offset ? (T*)(((char*)ptr)+offset) : nullptr;
+    };
+
+    template<class Fn1, class Fn2>
+    void forAll_ElfVer(span<Elf_Verdef> vdspan, Fn1&& fn1, Fn2&& fn2)
+    {
+        auto* vd = vdspan.begin();
+        while (vd)
+        {
+            fn1(*vd);
+            auto va = follow<Elf_Verdaux>(vd, rdi(vd->vd_aux));
+            while (va)
+            {
+                fn2(*va);
+                va = follow<Elf_Verdaux>(va, rdi(va->vda_next));
+            }
+            vd = follow<Elf_Verdef>(vd, rdi(vd->vd_next));
+        }
+    }
+
+    template<class Fn1, class Fn2>
+    void forAll_ElfVer(span<Elf_Verneed> vnspan, Fn1&& fn1, Fn2&& fn2)
+    {
+        auto* vn = vnspan.begin();
+        while (vn)
+        {
+            fn1(*vn);
+            auto va = follow<Elf_Vernaux>(vn, rdi(vn->vn_aux));
+            while (va)
+            {
+                fn2(*va);
+                va = follow<Elf_Vernaux>(va, rdi(va->vna_next));
+            }
+            vn = follow<Elf_Verneed>(vn, rdi(vn->vn_next));
+        }
+    }
 
     void clearSymbolVersions(const std::set<std::string> & syms);
 

diff --git a/tests/Makefile.am b/tests/Makefile.am
@@ -44,6 +44,7 @@ src_TESTS = \
   replace-add-needed.sh \
   add-debug-tag.sh \
   rename-dynamic-symbols.sh \
+  clean-strtab.sh \
   empty-note.sh
 
 build_TESTS = \

diff --git a/tests/clean-strtab.sh b/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 ${SCRATCH}
+
+the_string=VERY_SPECIFIC_STRING
+function check_count {
+    count="$(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