Add IDA scripts to strip debug info

scripting/ida-extraction-scripts/elf-replace-dynstr.py

+#!/usr/bin/env python3
+
+import argparse
+import sys
+
+from elftools.elf.elffile import ELFFile
+from elftools.elf.dynamic import DynamicSegment
+
+def log_error(msg):
+    sys.stderr.write("[ERROR] {}\n".format(msg))
+
+def log(msg):
+    sys.stderr.write(msg + "\n")
+
+def unique_or_none(list):
+    if not list:
+        return None
+    assert len(list) == 1
+    return list[0]
+
+def main():
+    parser = argparse.ArgumentParser(description="Rewrite portions of .dynstr.")
+    parser.add_argument("elf_path", metavar="ELF", help="path to the ELF file.")
+    args = parser.parse_args()
+
+    with open(args.elf_path, "rb+") as elf_file:
+        elf = ELFFile(elf_file)
+
+        # Blank dynstr section
+        dynamic = unique_or_none([segment
+                                  for segment
+                                  in elf.iter_segments()
+                                  if type(segment) is DynamicSegment])
+
+        if dynamic is None:
+            log("Not a dynamic executable")
+            return 0
+
+        address = unique_or_none([tag.entry.d_val
+                                  for tag
+                                  in dynamic.iter_tags()
+                                  if tag.entry.d_tag == "DT_STRTAB"])
+
+        offset = None
+        if address:
+            offset = unique_or_none(list(elf.address_offsets(address)))
+
+        size = unique_or_none([tag.entry.d_val
+                               for tag
+                               in dynamic.iter_tags()
+                               if tag.entry.d_tag == "DT_STRSZ"])
+
+        if offset is None or size is None:
+            log("DT_STRTAB not found")
+            return 0
+
+        elf_file.seek(offset)
+        original = elf_file.read(size)
+
+        new = "\x00" * size
+        new = new.encode("ascii")
+        elf_file.seek(offset)
+        elf_file.write(new)
+        log("patching dynstr with zeroes")
+
+        # Blank strtab section
+        strtab = unique_or_none([section for section in elf.iter_sections() if section.name == ".strtab"])
+        offset = strtab.header["sh_offset"]
+        size = strtab.header["sh_size"]
+
+        elf_file.seek(offset)
+        original = elf_file.read(size)
+
+        new = "\x00" * size
+        new = new.encode("ascii")
+        elf_file.seek(offset)
+        elf_file.write(new)
+        log("patching strtab with zeroes")
+
+    return 0
+
+if __name__ == "__main__":
+    sys.exit(main())

scripting/ida-extraction-scripts/strip-all.sh

+#!/bin/bash
+# With this step, we remove some additional debug infos from binaries, that, if
+# present, some decompilers take advantage to perform optimizations orthogonal
+# to our evaluation purposes.
+
+if [ $# -eq 0 ]; then
+  echo "No arguments supplied"
+  exit 1
+fi
+
+arch=$1
+basepath=$(pwd)
+workdir=$basepath/workdir-$arch
+stripped_dir=$workdir/stripped
+
+if [ ! -d $stripped_dir ]; then
+  mkdir $stripped_dir;
+fi
+
+cd $arch
+
+# Use the `objcopy` tool to remove the `.dynstr` section from the ELF
+for filename in *
+do
+  echo "Removing the .dynstr section from the ELF"
+  if [[ "$arch" == "x86-64"* ]]; then
+    cp $filename $stripped_dir/$filename.stripped
+    $basepath/scripting/ida-extraction-scripts/elf-replace-dynstr.py $stripped_dir/$filename.stripped
+  else
+    echo "Unsupported architecture"
+    exit 1
+  fi
+done;
+
+# Use the `strip` tool to remove dbg info
+for filename in *
+do
+  echo "Stripping input: $filename"
+  if [[ "$arch" == "x86-64"* ]]; then
+    strip --strip-debug $stripped_dir/$filename.stripped
+  elif [[ "$arch" == "arm"* ]]; then
+    armv7a-hardfloat-linux-uclibceabi-strip --strip-debug $stripped_dir/$filename.stripped
+  elif [[ "$arch" == "mips"* ]]; then
+    mips-unknown-linux-musl-strip --strip-debug $stripped_dir/$filename.stripped
+  else
+    echo "Unsupported architecture"
+    exit 1
+  fi
+done;