This comment regarding smtpmic seems to have been ignored.

jmc said the delay helped.  I can understand why it might.  It at least
tries to force the register write operation to *actually happen*.

The smtiic_wait_state() approach on the other hand, skips doing that
register write.

Doing the register operation when it is not ready is bad.  But skipping
it is probably also bad.

> I guess smtiic_wait_state() will spin for 1000 * delay(10),
> so it mostly wins.
> 
> When it fails, it returns ETIMEDOUT.  Converting that to EIO
> manually is a bit strange, but shrug it happens everywhere else
> in this driver.
> 
> Do any of the other smtiic functions also need this smtiic_wait_state?
> 
> Is smtiic_write_byte() doing this correctly?
> 

Reply via email to