Code size report:

   bare-arm:    +0 +0.000% 
minimal x86:    +0 +0.000% 
   unix x64:    +0 +0.000% standard
      stm32:    +0 +0.000% PYBV10
     mimxrt:    +0 +0.000% TEENSY40
        rp2:    +0 +0.000% RPI_PICO_W
       samd:    +0 +0.000% ADAFRUIT_ITSYBITSY_M4_EXPRESS
  qemu rv32:    +0 +0.000% VIRT_RV32

@sebromero @ubidefeo

This is most likely a breaking change for users, because the filesystem partition size has shrunk by 4MiB. And I don't think that will be easily detected because existing filesystems will mount and work correctly until the ROMFS is used. And then when ROMFS is used and the flash for that is written to, the original filesystem will probably still mount correctly because the initial filesystem data is unchanged.

We will deploy "stock" romfs images along with 1.25.0, which will likely corrupt the filesystem as you've expected. However, romfs will be the main storage for all embedded assets moving forward, as we're already running out of flash on some boards (like the Nicla), so this change is unavoidable. When we deploy 1.25.0 we could display a warning and erase the filesystem as well.

We could indeed add a version check in our installer tool to at least warn the user and consequently perform an erase of the file system.

For Mpflash it should be possible to detect the step-ups to 1.25.0 and then, after confirmation:

• extract the filesystem, or files on the filesystem to the host
• erase , and flash
• Wait for reboot, and the MCU to mount and format the flash
• if needed: Extract the files from the downloaded image
• re-upload the files.

Other tools can do this as well, if they can check the fw version up-front

When we deploy 1.25.0 we could display a warning and erase the filesystem as well.

OK, sounds good, you can manage user expectations on your side for this change.

@iabdalkader I have now updated this after #16903, to name the partition #0. Please can you review it?

@iabdalkader I have now updated this after #16903, to name the partition #0. Please can you review it?

I can retest it again, later today.

Seems to be working fine. I tested build, deploy and read. Anything else to test? Otherwise it's good.

>>> import os
>>> os.listdir("/rom")
['test.txt']
>>> print(open("/rom/test.txt", "r").read())
HelloWorld!
``

Seems to be working fine. I tested build, deploy and read

Great, thanks, that should be enough.

I tested the address as well, using uctypes.addressof. The ROMFS should start at 0x00B00000 (11MiB) so this seems about right.

mpremote run ls_romfs.py 
addr: 0x00B00012  size: 12        alignment: NOT aligned name: test.txt

The address should be more like 0x90B00000, right?

The address should be more like 0x90B00000, right?

Yes, that's right. The CYW43 firmware for example is at 0x90F00000.