gdb script for loading symbol files by correlating stack frames with /proc/$pid/maps

auto-add-symbol-files.py

# gdb script:
#
#   source /path/to/this/script
#   auto-add-symbol-files
#
# Tries to load symbol files by correlating stack frames with /proc/$pid/maps.
# Useful when debugging native code issues in programs running under Wine/Proton.
#
# Author: Nicolai Hähnle <nhaehnle@gmail.com>
# Use at your own risk, blah blah.

import gdb
import re
import subprocess

# RE for /proc/$pid/maps
RE_map = re.compile(r'([0-9a-f]+)-([0-9a-f]+) [r-][w-][x-]p ([0-9a-f]+) [0-9:]+ [0-9]+ +(.*)$')

# RE for output of readelf
RE_section = re.compile(r' *\[ *[0-9]+\] ([^ ]+) +(?:[^ ]+) +([0-9a-f]+) +(?:[0-9a-z]+)')

class AutoAddSymbolFiles(gdb.Command):
    def __init__(self):
        super(AutoAddSymbolFiles, self).__init__("auto-add-symbol-files", gdb.COMMAND_USER)

    def invoke(self, args, from_tty):
        print("Invoked blah", args)
        progspace = gdb.current_progspace()
        pid = gdb.selected_inferior().pid

        frame = gdb.newest_frame()
        depth = 0
        max_depth = None

        while frame is not None:
            if frame.function() is None:
                pc = frame.pc()
                if progspace.objfile_for_address(pc):
                    # We have no debug info even though an objfile has already been loaded
                    # for the address. Unwinding isn't going to work properly, so just bail.
                    print(f'Missing debug info at PC {pc:x}, stopping here.')
                    break

                if depth == max_depth:
                    print(f'Re-scanning the same frame, something must have gone wrong.')
                    break

                # See if we can find an objfile
                with open(f"/proc/{pid}/maps", "r") as maps:
                    for map_line in maps:
                        m = RE_map.match(map_line)
                        if m is None:
                            continue

                        begin = int(m.group(1), base=16)
                        end = int(m.group(2), base=16)
                        file_offset = int(m.group(3), base=16)
                        file_name = m.group(4)

                        if pc < begin or end <= pc:
                            continue

                        print(f'Found candidate object file {file_name} for PC {pc:x}')
                        break
                    else:
                        print(f'Could not find candidate object file for PC {pc:x}')
                        break

                # We found our candidate file. Get the section info by shelling out to readelf
                out = subprocess.check_output(['readelf', '-S', file_name]) \
                        .decode().split('\n')

                for out_line in out:
                    m = RE_section.match(out_line)
                    if m is None:
                        continue

                    section = m.group(1)
                    section_file_offset = int(m.group(2), base=16)
                    if section == '.text':
                        break
                else:
                    print(f'Did not find .text section')
                    break

                # Found the .text section
                text_address = begin - file_offset + section_file_offset
                print(f'Loading {file_name} with .text at {text_address:x}')
                gdb.execute(f'add-symbol-file {file_name} {text_address}')

                # Re-compute the frames
                max_depth = depth
                frame = gdb.newest_frame()
                depth = 0
                continue

            frame = frame.older()
            depth += 1

AutoAddSymbolFiles()