On Tue, Feb 15, 2022 at 11:00:54AM +0800, Jon Lin wrote:
> 在 2022/2/14 20:49, Mark Brown 写道:

> > As covered in the documentation setup() for one device may run while
> > another is active, therefore if multiple devices are configured in the
> > same register you should use a lock to ensure there can't be multiple
> > writes.  Note that the above appears to not just be setting the mode but
> > also the chip select so if you've got two SPI_CS_HIGH devices then
> > they'll both be going in and separately setting cr0.

> Is the io_mutex in function spi_setup is good enough?

It's not supposed to be for that but looking at the code quickly I
*think* setup() is never called with io_mutex held so it might well be
fine - you should double check though.  If not you'd need to add another
lock in your driver data.