On Tue, Jun 06, 2023 at 01:04:51PM +0100, Mark Brown wrote: > On Tue, Jun 06, 2023 at 01:01:01PM +0100, Conor Dooley wrote: > > On Tue, Jun 06, 2023 at 12:55:58PM +0100, Mark Brown wrote: > > > > It is obviously going to corrupt the transfer if we deassert chip select > > > without being asked, it is only valid to change it during a message if > > > cs_change is set. > > > So, reading between the lines, I shouldn't have doubted myself and this > > is an issue in the SPI controller or its driver? > > Yes. Note that it may be that the hardware is limited and can't do > whatever it's being asked to do. Yah, looks like they were using the ecspi on an imx8mp & there are apparently known issues there with the generation of CS on it. GPIO CS seems to do the job for them though, so one problem less for me to worry about :)
Attachment:
signature.asc
Description: PGP signature
