drivers: i2c: Introduce I2C timeout for SAM0

[kowalewskijan]

Adds configurable timeout for I2C transactions and provides bus recovery functionality.

Tested on SAME53J20A

[nandojve]

Hi @kowalewskijan ,

Interesting, a few points:

1- Could you explain on the commit messages what you are solving and why? It will be nice that you add information about a few use cases too. In this case, are you trying to solve some of the problems related on 36.6.2.4.2 Transmitting Address Packets?

2- I can imagine that it is not possible to have a HW test to avoid regressions for this, right?

3- Zephyr is moving to have all dependencies defined by each driver. This means that you need to add GPIO and PINCTRL dependencies at Kconfig.

[kowalewskijan]

@nandojve, what I mainly want to add is the timeout for I2C in sam0 driver. Right now, once the I2C is blocked due to malfunctioning I2C slave device or any other hardware issue that may cause the I2C to be not functional, the driver will hang the whole system during the device init phase. I believe introducing timeouts in such case will greatly improve usability for these SoCs. Regarding recovering the I2C it's just a typical way to recover it, but thinking about it, maybe it should handle the I2C errors more holistic based on STATUS register. If you want, we can split this into 1st having the timeout which is definitely beneficial for overall stability of the system for these SoCs and later add recovery features based on STATUS register.

[nandojve]

Hi @kowalewskijan ,

@nandojve, what I mainly want to add is the timeout for I2C in sam0 driver. Right now, once the I2C is blocked due to malfunctioning I2C slave device or any other hardware issue that may cause the I2C to be not functional, the driver will hang the whole system during the device init phase. I believe introducing timeouts in such case will greatly improve usability for these SoCs. Regarding recovering the I2C it's just a typical way to recover it, but thinking about it, maybe it should handle the I2C errors more holistic based on STATUS register. If you want, we can split this into 1st having the timeout which is definitely beneficial for overall stability of the system for these SoCs and later add recovery features based on STATUS register.

I like the idea about split into 2 parts. I'm fine about timeouts unless @teburd have a different opinion. What I really appreciate is to receive a solution that uses the STATUS registers to perform the bus recover.

[teburd]

Hi @kowalewskijan ,

@nandojve, what I mainly want to add is the timeout for I2C in sam0 driver. Right now, once the I2C is blocked due to malfunctioning I2C slave device or any other hardware issue that may cause the I2C to be not functional, the driver will hang the whole system during the device init phase. I believe introducing timeouts in such case will greatly improve usability for these SoCs. Regarding recovering the I2C it's just a typical way to recover it, but thinking about it, maybe it should handle the I2C errors more holistic based on STATUS register. If you want, we can split this into 1st having the timeout which is definitely beneficial for overall stability of the system for these SoCs and later add recovery features based on STATUS register.

I like the idea about split into 2 parts. I'm fine about timeouts unless @teburd have a different opinion. What I really appreciate is to receive a solution that uses the STATUS registers to perform the bus recover.

I don't have a strong opinion on this myself. I trust your judgment here @nandojve

[kowalewskijan]

@nandojve - I extracted only I2C timeouts for this PR