Re: SPI regression with today's build

[Andy Shevchenko <andy.shevchenko@xxxxxxxxx>]

On Wed, May 8, 2019 at 9:27 PM Noralf Trønnes <noralf@xxxxxxxxxxx> wrote:
> [cc:Martin]
> Den 08.05.2019 19.07, skrev Nicolas Saenz Julienne:

> Unless Martin Sperl, who wrote spi_split_transfers_maxsize(), has other
> suggestions, I think we should just revert this patch.

I gave a suggestion during LPC 2016 about this all stuff. We have to
provide a DMA channel capability to tell user what is the maximum
length is supported by the DMA engine. In such case the user may
prepare the transfers which are satisfying limitations and no
recreation is necessary, moreover, it will be performance wise to do
so.

-- 
With Best Regards,
Andy Shevchenko