Improve the logic of the AUTO symbol counting (idea from: @adamjseitz)

lief-project · Jun 3, 2023 · 528435f · 528435f
1 parent 0d7c39d
commit 528435f
Show file tree

Hide file tree

Showing 3 changed files with 46 additions and 22 deletions.
diff --git a/src/ELF/Parser.tcc b/src/ELF/Parser.tcc
@@ -437,39 +437,50 @@ result<uint32_t> Parser::get_numberof_dynamic_symbols(DYNSYM_COUNT_METHODS mtd)
     case DYNSYM_COUNT_METHODS::COUNT_AUTO:
     default:
       {
-        uint32_t nb_dynsym     = 0;
-        uint32_t nb_dynsym_tmp = 0;
+        uint32_t nb_reloc   = 0;
+        uint32_t nb_section = 0;
+        uint32_t nb_hash    = 0;
 
-        auto res = get_numberof_dynamic_symbols<ELF_T>(DYNSYM_COUNT_METHODS::COUNT_RELOCATIONS);
-        if (res) {
-          nb_dynsym = res.value();
+        if (auto res = get_numberof_dynamic_symbols<ELF_T>(DYNSYM_COUNT_METHODS::COUNT_RELOCATIONS)) {
+          nb_reloc = *res;
         }
-        res = get_numberof_dynamic_symbols<ELF_T>(DYNSYM_COUNT_METHODS::COUNT_SECTION);
-        if (res) {
-          nb_dynsym_tmp = res.value();
+
+        if (auto res = get_numberof_dynamic_symbols<ELF_T>(DYNSYM_COUNT_METHODS::COUNT_SECTION)) {
+          nb_section = *res;
+        }
+
+        if (auto res = get_numberof_dynamic_symbols<ELF_T>(DYNSYM_COUNT_METHODS::COUNT_HASH)) {
+          nb_hash = *res;
         }
 
-        if (nb_dynsym_tmp < Parser::NB_MAX_SYMBOLS &&
-            nb_dynsym_tmp > nb_dynsym              &&
-            (nb_dynsym_tmp - nb_dynsym) < Parser::DELTA_NB_SYMBOLS)
+        LIEF_DEBUG("#dynsym.reloc: {}",   nb_reloc);
+        LIEF_DEBUG("#dynsym.section: {}", nb_section);
+        LIEF_DEBUG("#dynsym.hash: {}",    nb_hash);
+
+        if (nb_hash > 0 && nb_section == nb_hash) {
+          return nb_hash;
+        }
+
+        uint32_t candidate = nb_reloc;
+        if (nb_section < Parser::NB_MAX_SYMBOLS &&
+            nb_section > nb_reloc               &&
+            (nb_section - nb_reloc) < Parser::DELTA_NB_SYMBOLS)
         {
-          nb_dynsym = nb_dynsym_tmp;
+          candidate = nb_section;
         }
 
-        res = get_numberof_dynamic_symbols<ELF_T>(DYNSYM_COUNT_METHODS::COUNT_HASH);
-        if (!res) {
-          // Fail to get number of symbols from the hash table
-          return nb_dynsym;
+        if (nb_hash == 0) {
+          return candidate;
         }
 
-        nb_dynsym_tmp = res.value();
-        if (nb_dynsym_tmp < Parser::NB_MAX_SYMBOLS &&
-            nb_dynsym_tmp > nb_dynsym              &&
-            (nb_dynsym_tmp - nb_dynsym) < Parser::DELTA_NB_SYMBOLS)
+        if (nb_hash < Parser::NB_MAX_SYMBOLS &&
+            nb_hash > candidate              &&
+            (nb_hash - candidate) < Parser::DELTA_NB_SYMBOLS)
         {
-          nb_dynsym = nb_dynsym_tmp;
+          candidate = nb_hash;
         }
-        return nb_dynsym;
+
+        return candidate;
       }
   }
 }

diff --git a/tests/elf/test_parser.py b/tests/elf/test_parser.py
@@ -10,6 +10,14 @@ def test_symbol_count():
     assert len(gcc2.symbols) == 158
     assert len(gcc3.symbols) == 158
 
+def test_issue_922():
+    libcrypto_path = get_sample('ELF/libcrypto.so')
+    auto = lief.ELF.parse(libcrypto_path)
+    assert len(auto.symbols) == 14757
+
+    section = lief.ELF.parse(libcrypto_path, lief.ELF.DYNSYM_COUNT_METHODS.SECTION)
+    assert len(section.symbols) == 14757
+
 def test_tiny():
     tiny = lief.parse(get_sample('ELF/ELF32_x86_binary_tiny01.bin'))
     assert len(tiny.segments) == 1

diff --git a/tests/elf/test_section_frame.py b/tests/elf/test_section_frame.py
@@ -41,6 +41,9 @@ def test_frame(tmp_path):
 
 def test_sectionless(tmp_path):
     elf: lief.ELF.Binary = lief.parse(get_sample("ELF/mbedtls_selftest.elf64"))
+
+    assert len(elf.dynamic_symbols) == 40
+
     header: lief.ELF.Header = elf.header
 
     header.numberof_sections     = 0
@@ -52,6 +55,8 @@ def test_sectionless(tmp_path):
     sectionless = lief.parse(out.as_posix())
     out = pathlib.Path(tmp_path) / "mbedtls_selftest.sectionless.built"
 
+    assert len(sectionless.dynamic_symbols) == 40
+
     sectionless.add_library(sectionless.libraries[0])
     sectionless.write(out.as_posix())