On Wed, Jun 12, 2024 at 09:48:16PM +0100, Conor Dooley wrote:
> On Wed, Jun 12, 2024 at 05:40:39PM +0100, Mark Brown wrote:
> > On Wed, Jun 12, 2024 at 04:48:32PM +0100, Conor Dooley wrote:
> >
> > > + //TODO: questionable robustness if both cs_change and cs_off toggle
> > > + list_for_each_entry(t, &m->transfers, transfer_list) {
> > > + //cs_change being set means we need to re-enable
> >
> > Is it not possible to implement prepare_message() and transfer_one()
> > rather than open coding all this?
>
> If I can, I will. I already found one issue with the cs toggling in the
> code Cyril gave me and I need to figure out why there's a udelay(750)
> required later on in the function anyway!
Actually wasn't too bad, glad to be rid of the open coded CS handling.
I still need the as of yet not understood udelay(), but it looks even
worse now:
+static int mchp_coreqspi_unprepare_message(struct spi_controller *ctlr, struct spi_message *m)
+{
+ udelay(750);
+
+ return 0;
+}
Attachment:
signature.asc
Description: PGP signature
