Dump ext2 filesystem via U-Boot mmc menu

mmc_uboot_dump.py

import os.path
import re
import sys
import struct
import serial
from tqdm import tqdm
import re


DIR_R = re.compile(r'^<DIR>[ \t]+(\d+)[ \t]+(.+)$')
FILE_R = re.compile(r'^[ \t]+(\d+)[ \t]+(.+)$')
SYM_R = re.compile(r'^<SYM>[ \t]+(\d+)[ \t]+(.+)$')
FILE_GOOD_R = re.compile(r'(\d+) bytes read in .+$')
IGNORE = ['.qmlc', '.jsc', '.wav', '.caf', '.mkv', '.png', '.mo', '.css', '.pdf']
IGNORE_NAMES = tuple(['libQt5'])
IGNORE_DIRS = ['.//usr/share/zoneinfo', './/usr/lib/gstreamer-1.0', './/usr/lib/locale']
RAM_ADDR = 0x10000000
BLOCK_SIZE = 0x1000


def exec_cmd(s, cmd):
    # print('executing %s' % cmd.decode())
    s.write(b'%s\n' % cmd)
    lines = []

    line = ''
    while True:
        b = s.read()

        if b in [b'\n', b'\r']:
            line = line.rstrip('\r\n')
            if len(line):
                lines.append(line)
            line = ''
            continue
        elif b == b'' and line.startswith('=> '):
            break

        line += b.decode(errors='replace')

    return lines


def parse_hex_lines(lines, max_size):
    res = bytearray()
    for line in lines:
        data, _ = line.split("    ", maxsplit=1)
        _, strdata = data.split(maxsplit=1)
        data = bytes.fromhex(strdata)
        res.extend(data)

    return bytes(res)[:max_size]



def dump_file(s, path, size):
    w_path = '/'.join(['./', path])
    fn, fe = os.path.splitext(path)
    fn = os.path.basename(fn)

    if size == 0 or fe in IGNORE or fn.startswith(IGNORE_NAMES):
        with open(w_path, 'wb') as w:
            pass
        return

    if os.path.exists(w_path) and os.path.getsize(w_path) == size:
        return

    res = exec_cmd(s, b'ext2load mmc 1:2 %x %s' % (RAM_ADDR, path.encode()))

    m = FILE_GOOD_R.match(res[1])
    real_sz = -1

    if m is not None:
        real_sz = m.group(1)

    if real_sz == -1:
        with open(w_path, 'wb') as w:
            pass
        return

    off = 0
    bs = min(size, BLOCK_SIZE)

    with open(w_path, 'wb') as w:
        with tqdm(total=size) as pb:
            lines = exec_cmd(s, b'md.b %x %x' % (RAM_ADDR + off, bs))[1:]

            buf = parse_hex_lines(lines, bs)
            w.write(buf)
            pb.update(bs)
            size -= bs
            off += bs

            while size > 0:
                lines = exec_cmd(s, b'\n')

                blk = min(size, bs)
                buf = parse_hex_lines(lines, blk)
                w.write(buf)
                pb.update(blk)
                size -= blk
                off += blk


def dump_mmc_1_2(s, root='./', level=0):
    os.makedirs(root, exist_ok=True)

    if root in IGNORE_DIRS:
        return

    res = exec_cmd(s, b'ext2ls mmc 1:2 %s' % root.encode())
    res = res[1:]
    count = len(res) - 2

    dirs = []
    for i, line in enumerate(res):
        m = DIR_R.match(line)

        if not m:
            m = SYM_R.match(line)

            if not m:
                m = FILE_R.match(line)

                if not m:
                    continue

                # do file
                size = int(m.group(1))
                name = m.group(2)
                print('%s[%04d/%04d] %s %d' % (' ' * level, i + 1 - 2, count, name, size))
                dump_file(s, '/'.join([root, name]), size)
                continue


            # do sym
            size = int(m.group(1))
            name = m.group(2)
            print('%s[%04d/%04d] <%s> %d' % (' ' * level, i + 1 - 2, count, name, size))
            dump_file(s, '/'.join([root, name]), 0)
            continue

        # do dir
        name = m.group(2)

        if name not in ['.', '..']:
            print('%s[%04d/%04d] [%s]' % (' ' * level, i + 1 - 2, count, name))
            dump_mmc_1_2(s, '/'.join([root, name]), level+1)
        # return sub_dirs


def main():
    # s = serial.Serial('COM15', 115200, timeout=0.001)
    s = serial.Serial('/dev/ttyUSB0', 115200, timeout=0.001)
    dump_mmc_1_2(s)
    s.close()


if __name__ == '__main__':
    main()