On Mon, May 18, 2020 at 02:04:53PM +0300, Serge Semin wrote:
> On Mon, May 18, 2020 at 11:51:30AM +0100, Mark Brown wrote:

> > > - semantically the xfer argument isn't optional and we can't fetch it that easy
> > >   in the dmaengine completion callbacks.

> > Not sure I follow this.

> I mean is it Ok to call the spi_delay_exec like this: spi_delay_exec(delay, NULL),
> with null passed instead of xfer pointer? Semantically the pointer is required only
> if we'd need to calculate the SPI_DELAY_UNIT_SCK delay, but here we'll need
> USECS and NSECS delays. So at the first glace there is no problem with passed
> NULL instead of xfer. But doing so we'd rely on the semantic peculiarity, which
> may seem a bit hackish.

Yes, that should be fine if you don't specify a SCK delay.  There's no
reason to be looking at that if the delay isn't specified in SCKs.