On Tue, Oct 19, 2021 at 04:30:37PM +0100, Daniel Thompson wrote:
> On Tue, Oct 19, 2021 at 02:09:27PM +0200, Uwe Kleine-König wrote:
> > Up to now ili9320_remove() returns zero unconditionally. Make it return
> > void instead which makes it easier to see in the callers that there is
> > no error to handle.
> 
> Ultimately the actions here boil down to an SPI write which could fail
> so we could make an attempt to correct the broken error propagation
> inside ili9320.c instead...
> 
> > Also the return value of spi remove callbacks is ignored anyway.
> 
> ... which does inform the above question.
> 
> Is this patch a local fix or part of a larger plan to eventually change
> the return value of the spi_remove to void?

This is the motivation yes, so there isn't any sane way to propagate an
error. The only thing you could sensibly do is to print an error
message. But the device will go away no matter what you do. (Well, you
could do an endless-loop to "delay" removing, but ...)

Best regards
Uwe

-- 
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | https://www.pengutronix.de/ |