stm32 i2c slave

[jrmoulton]

This isn't quite finished. It only has the async slave implementation and the code needs to be cleaned up.

I could use some help debugging also.

In the listen method I enable the address match interrupt and register the waker in the poll closure.

The issue is that the runtime never polls the function again even when an address is matched.

If I write code that blocks instead of yielding and send and I2C communication the peripheral will still be enabled and will match the address.

I've tried also adding the wupen bit but that didn't make a difference

WUPEN:Wakeup from Stop mode enable 
0: Wakeup from Stop mode disable.
1: Wakeup from Stop mode enable.

Any idea why the runtime wouldn't poll the function again?

[bsodmike]

Hi Jared, interesting. Will respond soon if I figure anything out.

[jrmoulton]

I've figured out why the function wasn't being polled. Now just working on finishing up the implementation

[jrmoulton]

@Dirbaio the stm32 i2c supports multimaster for allowing both a master and a slave functionality. The slave mode is the default and the master mode is enabled by a start condition and ends with a stop condition.

Currently in this implementation I have split the slave from the master. This more closely matches the API in the embassy-rp crate. Would it be better to join them and merge the slave functionality into the regular i2c driver (in order to let you switch from slave to master without reconfiguring) or is it better to leave them split?

(edited)

[bsodmike]

I think being part of the regular driver works, if say you introduce a Option<Mode> type config option that is very obvious.

Thoughts?