SDCC 4.5.0 conflicts with ITE code signature, puts `HOME` segment at wrong address

[clark-ja]

On the current master (commit d15bc38), building the ec.rom for the darp10-b results in a wrong size, where it is 262080 bytes instead of 262144 bytes.

Steps to reproduce

1. Clone the repository and initialize submodules
2. Build with make BOARD=system76/darp10-b
3. Check the size of the ec.rom with, e.g., wc -c build/ec.rom

Disclaimer: I currently do not have another platform where I can try it out so I am not 100% sure if there is something wrong on my machine causing this issue.

[clark-ja]

Sorry, this clearly appears to be a bug on my system as objcopy -I ihex -O binary --gap-fill=0xFF --pad-to=262144 build/ec.ihx build/ec.rom results in a file build/ec.rom of size 262080 bytes...

[clark-ja]

I found the problem: This bug happens with sdcc=4.5.0 but not with sdcc=4.4.0. I leave this issue here for reference. You may close it.

[crawfxrd]

Reverting to SDCC's makebin might work.

If so, would also remove the need for installing the host GCC toolchain (gcc-avr would still be needed for the Arduino tools). Well, not entirely: The tool needs cc available.

[python-destroyer]

I bricked my darp10 because of this issue. I'm on Debian 13 which has gcc 14.2.0. When trying to build and flash EC I was getting the size error, so I just padded the firmware with objcopy -I binary -O binary --gap-fill=0xFF --pad-to=262144 build/ec.rom and flashed it.

After flashing the padded image laptop has no signs of life, pressing the power button does nothing, even LEDs don't light up when connecting the charger.

I switched to #521 and image builds with proper size, flashed it externally with Arduino Mega, but the laptop still has no signs of life. Any suggestions?

[crawfxrd]

Don't use SDCC 4.5.0.

Build the firmware using a container image that has sdcc <4.5.0.
e.g., Ubuntu 24.04 (4.2.0) or Fedora 43 (4.4.0)

[python-destroyer]

Thanks @crawfxrd, I built and flashed the EC firmware with Ubuntu 24.04 in VM and that revived my PC.