I2C Slave Driver

[zephyrbot]

Reported by Richard Peters:

An API for I2C-Slaves is needed (Blocking and NonBlocking).
The supported plattforms should implement a slave driver.

(Imported from Jira ZEP-1486)

[zephyrbot]

by Kumar Gala:

We should try to make slave support consistent between SPI & I2C.

[dwagenk]

I'd like to work on a I2C slave driver for stm32 family. Has there been any discussion, rfc or similar on how the API should work? Or any started work on this issue? I didn't find anything in the mailinglist archives and on github.

I'm mostly interested in slave reciever, since it is a multi-master setup, so every component sends data as master and recieves as slave. Anyhow the API should be able to also handle slave transmitters. The problems that arise in this case are basically the same as with SPI, that @tbursztyka mentioned in #3782 (dumb slave that has prebuilt answer ready, or complicated slave, that needs to parse part of the message and than reply asap).
Any thoughts on this?

What I found was nrf5 spi slave (with legacy spi API) and esp32 i2c slave that look like they are developed against the existing (master centric) API and reinterpret it from a slave point of view. But I haven't found working samples or documentation on this anywhere.

[galak]

@mnkp might have some ideas on this. We need to workup a usage model of how we think slave mode will look from the API perspective first.

[tbursztyka]

@dwagenk I haven't had time to finalize SPI slave support, I mean: the proper way for complex slaves. I just rebased my working branch but there is still a big part missing: handling incoming master input's interruption which would call user's callback (i.e.: the one parsing the input asap and doing something about it). https://github.com/tbursztyka/zephyr/tree/spi_update_wip I hope I'll have time to finish this before 1.10.

[superna9999]

@dwagenk can you have a look at #5224 ?

[nashif]

this is now supported