Panics in release mode

[jordens]

The code should work fine in release mode.

[HarryMakes]

Hi, I was testing with --release to see which part of main.rs (commit fa742bd) in the release profile would cause the panic. My observation:

• Initialisation of the channels and fans hardware seem to cause panic; I2C SCL and SDA, if initialised separately, won't cause the panic.
• LEDs can be initialised successfully. However, they don't seem to work (turning the lights on), e.g. by calling their update functions in a "button press" handler.

The firmware by cargo-building in the debug profile doesn't panic upon initialisation. But the ELF generated and *.DFU converted between the debug and release profile differ quite a lot:

• release: ELF is 3033064 B, converted DFU is 173345 B
• debug: ELF is 7081552 B, converted DFU is 514889 B

[ryan-summers]

I have probed the I2C lines and looks like this has to do with resetting the ADS7924. Specifically, after issuing a software reset to the device, we currently wait 100uS for the device to come back online. A probe of the I2C lines indicates this may not be enough time for the device to come back up:

I think an appropriate fix here is to instead poll continuously after the reset is sent to wait for the device to come back online.

[ryan-summers]

Additional testing indicates we cannot continuously poll the device either, as it will continuously NAK the request. I would recommend we just increase the delay period to 500uS and see if we have any further issues.