Non-CircuitPython firmware break the saves drive

[FoamyGuy]

CircuitPython version and board name

Adafruit CircuitPython 10.0.3 on 2025-10-17; Adafruit Fruit Jam with rp2350b

Code/REPL

# write to a file in saves
with open("/saves/testfile.txt", "w") as f:
    f.write("test content")


# read from a file in saves
# with open("/saves/testfile.txt", "r") as f:
#     print(f.read())

Behavior

After this sequence of events the saves drive ends up in a broken state:

• Start with CPy loaded and CPSAVES drive working as normal
• Write content to a file inside of /saves/, read back the content to verify that it was successfully written.
• Load a non CPy firmware UF2 such as apple 2e
• Load CPy firmware again

Now the CPSAVES drive does not seem to get mounted to my PC any more, only CIRCUITPY. Though I do hear 2 "device connected" jingles so it does seem to be trying. Possibly the 2 jingles is a not actually an indication of it trying to connect CPSAVES, I just noticed when CPSAVES is working I hear 3 jingles.

Trying to read the test file from /saves/ reports:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "saves_file_test.py", line 6, in <module>
OSError: [Errno 2] No such file/directory: /saves/testfile.txt

Trying to write a new file into /saves/ reports:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "saves_file_test.py", line 1, in <module>
OSError: [Errno 30] Read-only filesystem

os.listdir("/saves/") outputs only the placeholder.txt file.

Description

No response

Additional information

This isn't specific to the apple 2e emulator, I've had the same with the 286 emulator. Though I have not tried a more basic arduino program to see if it does the same. I could do that test if it is helpful to know.

Running storage.erase_filesystem() gets the CPSAVES drive back to a working state after this happens.

[dhalbert]

CircuitPython is only aware that there might be a CPSAVES partition if CIRCUITPY_SAVES_PARTITION_SIZE > 0. I'm not sure there's anything in flash that makes it clear it exists.

Do the non-CircuitPython firmware images know about the layout of flash? Do they use CIRCUITPY at all?

[dmcomm]

I think we're just talking about what happens if you overwrite the first part of flash with something smaller than the CircuitPython core. Like I noticed before that the contents of CIRCUITPY are preserved after using Arduino.