Radio img extractor - Android Q&A, Help & Troubleshooting

Hello ... so i have an Radio.img and i know inside there are this files
(bootloader) Validating 'radio.default.xml'
(bootloader) Committing 'radio.default.xml'
(bootloader) - flashing 'NON-HLOS.bin' to 'modem'
(bootloader) - flashing 'fsg.mbn' to 'fsg'
(bootloader) - erasing 'modemst1'
(bootloader) - erasing 'modemst2'.
How can i extract NON-HLOS and fsg ? thanks in advance ...

---

I know this is an ancient thread, but it's still the first search result, so I figured a solution could help anyone else that stumbles upon this..
I made a quick and dirty extractor that works at least for motorola edge 2021 xt2141 radio images. These files seem to start with magic "SINGLE_N_LONELY" and end with "LONELY_N_SINGLE". Filenames are provided, followed by the length of the contents (in little endian), then the contents.
This script will try to open radio.img in the current dir if a filename is not provided. Dumped files will go right in the working dir, so be careful. File content reading isn't done in chunks here, so be mindful of memory usage. Likely not an issue, but you can code in some chunking if needed.
Code:
#!/usr/bin/env python
import io
import sys
# supply filename as argument or default to 'radio.img'
try:
filename = sys.argv[1]
except IndexError:
filename = 'radio.img'
with open(filename, 'rb') as f:
magic = f.read(0x100).strip(b'\0').decode()
print(magic)
assert magic == 'SINGLE_N_LONELY'
while True:
# filename
fn = f.read(0xF0).strip(b'\0').decode()
print(fn)
if fn == 'LONELY_N_SINGLE':
break
# size of file in little endian
f.seek(0x08, io.SEEK_CUR)
l = int.from_bytes(f.read(0x08), 'little')
print(l)
# warning: not reading in chunks...
# warning: outputs to working dir
with open(fn, 'wb') as o:
o.write(f.read(l))
# seek remainder
rem = 0x10 - (l % 0x10)
if rem < 0x10:
f.seek(rem, io.SEEK_CUR)
# seek until next filename
while not f.read(0x10).strip(b'\0'):
continue
# rewind back to start of filename
f.seek(-0x10, io.SEEK_CUR)
Note the resulting images will likely be in sparse format. You'll need simg2img to convert to raw images if you're trying to mount or otherwise manhandle the images.
If interested in dumping carrier profiles (from inside the fsg image), EfsTools has an extractMbn function. Not sure how to reassemble though. https://github.com/JohnBel/EfsTools

---

ziddey said:
I know this is an ancient thread, but it's still the first search result, so I figured a solution could help anyone else that stumbles upon this..
I made a quick and dirty extractor that works at least for motorola edge 2021 xt2141 radio images. These files seem to start with magic "SINGLE_N_LONELY" and end with "LONELY_N_SINGLE". Filenames are provided, followed by the length of the contents (in little endian), then the contents.
This script will try to open radio.img in the current dir if a filename is not provided. Dumped files will go right in the working dir, so be careful. File content reading isn't done in chunks here, so be mindful of memory usage. Likely not an issue, but you can code in some chunking if needed.
Code:
#!/usr/bin/env python
import io
import sys
# supply filename as argument or default to 'radio.img'
try:
filename = sys.argv[1]
except IndexError:
filename = 'radio.img'
with open(filename, 'rb') as f:
magic = f.read(0x100).strip(b'\0').decode()
print(magic)
assert magic == 'SINGLE_N_LONELY'
while True:
# filename
fn = f.read(0xF0).strip(b'\0').decode()
print(fn)
if fn == 'LONELY_N_SINGLE':
break
# size of file in little endian
f.seek(0x08, io.SEEK_CUR)
l = int.from_bytes(f.read(0x08), 'little')
print(l)
# warning: not reading in chunks...
# warning: outputs to working dir
with open(fn, 'wb') as o:
o.write(f.read(l))
# seek remainder
rem = 0x10 - (l % 0x10)
if rem < 0x10:
f.seek(rem, io.SEEK_CUR)
# seek until next filename
while not f.read(0x10).strip(b'\0'):
continue
# rewind back to start of filename
f.seek(-0x10, io.SEEK_CUR)
Note the resulting images will likely be in sparse format. You'll need simg2img to convert to raw images if you're trying to mount or otherwise manhandle the images.
If interested in dumping carrier profiles (from inside the fsg image), EfsTools has an extractMbn function. Not sure how to reassemble though. https://github.com/JohnBel/EfsTools
Click to expand...
Click to collapse
Thanks for making python script to unpack these SINGLE_N_LONELY header files(bootloader.img, radio.img, singleimage.bin, gpt.bin) from Moto Stock ROM zips.
But why reading filename only 240 bytes and skipping 8 bytes instead of reading whole 248 bytes?
This guy wrote to read 248 bytes instead https://forum.xda-developers.com/t/...t-of-the-moto-g-5g-plus.4371213/post-87807175
I also made quick and dirty unpacked using Lua 5.3 at https://forum.xda-developers.com/t/...t-of-the-moto-g-5g-plus.4371213/post-87931915
I guess one of us has to post this to github, since I can't find any Open Source tool to unpack this simple format image files.
Currently, only star tool that we can find from some of blankflash files(eg. this) and imjtool can unpack these SINGLE_N_LONELY header files as far as I know. But I guess these are not Open Source.
Thanks

---

HemanthJabalpuri said:
But why reading filename only 240 bytes and skipping 8 bytes instead of reading whole 248 bytes?
This guy wrote to read 248 bytes instead https://forum.xda-developers.com/t/...t-of-the-moto-g-5g-plus.4371213/post-87807175
Click to expand...
Click to collapse
Ah neat. I only used xt2141 radio images as reference for approximating the file format. It's been a while, but I think based on the actual positioning of the filenames in the images I was testing, I wasn't sure if the final 8 bytes were part of the filename or padding.
Likewise, I wasn't sure of how padding works after the file data, so I just did a dumb seek and rewind.