Reduce binary size while adding segments or relocating PHDR

Commit based on the work of @Miniman2718

Resolve: #1085

src/ELF/Binary.cpp

@@ -2648,6 +2648,16 @@ uint64_t Binary::relocate_phdr_table_v3() {
   // Reserve space for 10 user's segments
   static constexpr size_t USER_SEGMENTS = 10;
 
+  uint64_t last_off = 0;
+  for (const std::unique_ptr<Segment>& segment : segments_) {
+    if (segment != nullptr && segment->is_load()) {
+      last_off = std::max(last_off, segment->file_offset());
+      // Possible because segment are not overlapping
+      last_off += segment->physical_size();
+    }
+  }
+  last_off = align(last_off, static_cast<uint64_t>(get_pagesize(*this)));
+
   if (phdr_reloc_info_.new_offset > 0) {
     return phdr_reloc_info_.new_offset;
   }
@@ -2660,7 +2670,7 @@ uint64_t Binary::relocate_phdr_table_v3() {
     type() == Header::CLASS::ELF32 ? sizeof(details::ELF32::Elf_Phdr) :
                                      sizeof(details::ELF64::Elf_Phdr);
 
-  const uint64_t last_offset = virtual_size();
+  const uint64_t last_offset = last_off;
 
   LIEF_DEBUG("Moving segment table at the end of the binary (0x{:010x})",
              last_offset);
@@ -2688,7 +2698,7 @@ uint64_t Binary::relocate_phdr_table_v3() {
   phdr_load_segment->file_offset(phdr_reloc_info_.new_offset);
   phdr_load_segment->physical_size(new_segtbl_sz);
   phdr_load_segment->virtual_size(new_segtbl_sz);
-  phdr_load_segment->virtual_address(imagebase() + phdr_reloc_info_.new_offset);
+  phdr_load_segment->virtual_address(imagebase() + virtual_size());
   phdr_load_segment->physical_address(phdr_load_segment->virtual_address());
   phdr_load_segment->alignment(0x1000);
   phdr_load_segment->add(Segment::FLAGS::R);

src/ELF/Binary.tcc

@@ -364,12 +364,8 @@ Segment* Binary::add_segment<Header::FILE_TYPE::EXEC>(const Segment& segment, ui
     return nullptr;
   }
 
-  //const uint64_t phdr_size = type() == ELF_CLASS::ELFCLASS32 ?
-  //                                     sizeof(details::ELF32::Elf_Phdr) : sizeof(details::ELF64::Elf_Phdr);
-
   // Add the segment itself
   // ====================================================================
-  //datahandler_->make_hole(new_phdr_offset + phdr_size * header.numberof_segments(), phdr_size);
   header.numberof_segments(header.numberof_segments() + 1);
   span<const uint8_t> content_ref = segment.content();
   std::vector<uint8_t> content{content_ref.data(), std::end(content_ref)};
@@ -378,6 +374,7 @@ Segment* Binary::add_segment<Header::FILE_TYPE::EXEC>(const Segment& segment, ui
   uint64_t last_offset_sections = last_offset_section();
   uint64_t last_offset_segments = last_offset_segment();
 
+
   uint64_t last_offset = std::max<uint64_t>(last_offset_sections, last_offset_segments);
 
   const auto psize = static_cast<uint64_t>(get_pagesize(*this));
@@ -413,7 +410,7 @@ Segment* Binary::add_segment<Header::FILE_TYPE::EXEC>(const Segment& segment, ui
   new_segment->content(content);
 
   if (header.section_headers_offset() <= new_segment->file_offset() + new_segment->physical_size()) {
-    header.section_headers_offset(header.section_headers_offset() + new_segment->file_offset() + new_segment->physical_size());
+    header.section_headers_offset(new_segment->file_offset() + new_segment->physical_size());
   }
 
   const auto it_new_segment_place = std::find_if(segments_.rbegin(), segments_.rend(),

tests/elf/test_i872.py

@@ -1,14 +1,15 @@
 #!/usr/bin/env python
-
 import subprocess
-from subprocess import Popen
 import pytest
 import stat
 
 import lief
 import pathlib
 
-from utils import get_sample, is_linux, is_x86_64
+from subprocess import Popen
+from pathlib import Path
+
+from utils import get_sample, is_linux, is_x86_64, glibc_version
 
 def test_issue_872(tmp_path):
     tmp = pathlib.Path(tmp_path)
@@ -190,3 +191,42 @@ def test_docker_init(tmp_path, mode):
     with Popen([outpath.as_posix(), "--version"], **popen_args) as proc: # type: ignore
         stdout = proc.stdout.read()
         assert "tini version 0.19.0" in stdout, f"Error: {stdout}"
+
+@pytest.mark.skipif(not (is_linux() and is_x86_64()), reason="incompatible env")
+@pytest.mark.parametrize("mode", [
+    lief.ELF.Binary.PHDR_RELOC.SEGMENT_GAP,
+    lief.ELF.Binary.PHDR_RELOC.BSS_END
+])
+def test_python312d(tmp_path: Path, mode: lief.ELF.Binary.PHDR_RELOC):
+    sample = Path(get_sample("ELF/python3.12d"))
+
+    elf: lief.ELF.Binary = lief.ELF.parse(sample)
+    elf.relocate_phdr_table(mode)
+
+    segment = lief.ELF.Segment()
+    segment.type = lief.ELF.Segment.TYPE.LOAD
+    segment.content = [1] * 12
+
+    elf.add(segment)
+    outpath = tmp_path / "modified.elf"
+    elf.write(outpath.as_posix())
+
+    delta_size = outpath.stat().st_size - sample.stat().st_size
+
+    if mode == lief.ELF.Binary.PHDR_RELOC.SEGMENT_GAP:
+        assert delta_size < 0x2000
+
+    if mode == lief.ELF.Binary.PHDR_RELOC.BSS_END:
+        assert delta_size < 0x6000
+
+    if glibc_version() >= (2, 38):
+        outpath.chmod(outpath.stat().st_mode | stat.S_IEXEC)
+        popen_args = {
+            "stdout": subprocess.PIPE,
+            "stderr": subprocess.STDOUT,
+            "universal_newlines": True
+        }
+
+        with Popen([outpath.as_posix(), "--version"], **popen_args) as proc: # type: ignore
+            stdout = proc.stdout.read()
+            assert "Python 3.12.7" in stdout, f"Error: {stdout}"