Re: [PATCH 2/2] spi: Add driver for IMG SPFI controller

[Mark Brown <broonie-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>]

[-- Attachment #1: Type: text/plain, Size: 2273 bytes --]

On Wed, Nov 12, 2014 at 02:54:57PM -0800, Andrew Bresticker wrote:
> On Wed, Nov 12, 2014 at 2:07 PM, Mark Brown <broonie-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org> wrote:

> >>  drivers/spi/spi-img.c | 703 ++++++++++++++++++++++++++++++++++++++++++++++++++

> > How about spi-img-spfi?  That way if someone makes another SPI
> > controller (say a more generic one, this one seems flash specialized)
> > there won't be a collision.

> Despite the name, I believe this controller is used for generic SPI
> stuff as well.  I'm not sure if there is a separate one which is more
> generic (James?).

It would still be better to use a name less impressively generic - this
is an entire company, not even a product line.

> >> +             cpu_relax();

> > Seems random - we already relax in the data transfer?

> We don't relax in the transfers - would that be a better place to put
> it?  I thought it was better here since we reach this point once the
> TX FIFO has filled or RX FIFO has emptied.

Oh, that was the FIFO drain I was thinking of.  I guess here is fine.

> >> +     if (tx_buf)
> >> +             spfi_flush_tx_fifo(spfi);
> >> +     spfi_disable(spfi);

> > What does the enable and disable actually do?  Should this be runtime
> > PM?

> It starts/stops the transfer.  The control registers (bit clock,
> transfer mode, etc.) must be programmed before the enable bit is set
> and the bit does not clear itself upon completion of the transfer.  I
> don't think it makes sense to have this be part of runtime PM.

Perhaps these functions need to be called start() and stop() then - the
names sound like they gate the IP?

> > This will unconditionally claim to have handled an interrupt even if it
> > didn't get any interrupt status it has ever heard of.  At the very least
> > it should log unknown interrupts, ideally return IRQ_NONE though since
> > it seems to be a clear on read interrupt that's a bit misleading.

> There's a clear register actually (see the writel() above), but yes,
> an error and returning IRQ_NONE sound appropriate in the event of an
> unexpected interrupt.

Don't add the error print - the interrupt core has diagnostics already
and it won't be nice if the interrupt ends up shared.  My recommendation
was intended as an either/or.

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 473 bytes --]