On Mon, Mar 20, 2023 at 05:36:17PM +0200, Andy Shevchenko wrote:
> On Mon, Mar 20, 2023 at 11:31:07AM -0400, William Breathitt Gray wrote:
> > On Mon, Mar 20, 2023 at 02:28:31PM +0200, Andy Shevchenko wrote:
> > > On Sat, Mar 18, 2023 at 10:59:51AM -0400, William Breathitt Gray wrote:
> > > > The Preset Register (PR), Flag Register (FLAG), and Filter Clock
> > > > Prescaler (PSC) have common usage patterns. Wrap up such usage into
> > > > dedicated functions to improve code clarity.
>
> ...
>
> > > > +static void quad8_preset_register_set(struct quad8 *const priv, const size_t id,
> > > > + const unsigned long preset)
> > > > +{
> > > > + struct channel_reg __iomem *const chan = priv->reg->channel + id;
> > > > + int i;
> > > > +
> > > > + /* Reset Byte Pointer */
> > > > + iowrite8(SELECT_RLD | RESET_BP, &chan->control);
> > > > +
> > > > + /* Set Preset Register */
> > > > + for (i = 0; i < 3; i++)
> > > > + iowrite8(preset >> (8 * i), &chan->data);
> > > > +}
> > >
> > > May we add generic __iowrite8_copy() / __ioread8_copy() instead?
> > >
> > > It seems that even current __ioread32_copy() and __iowrite32_copy() has to
> > > be amended to support IO.
>
> > Sure, I would use __iowrite8_copy() / __ioread8_copy() for these
> > situations if it were available.
>
> If needed, you may always introduce ones.
>
> > Is something equivalent available for the regmap API? I'm planning to
> > migrate this driver to the regmap API soon after this patch series is
> > merged, so the *_copy() calls would need to migrated as well.
>
> Yes. It's regmap bulk operations.
>
> --
> With Best Regards,
> Andy Shevchenko
After reading through the implementation for these functions I realized
they are actually doing something different than what's happening here.
The 104-QUAD-8 device exposes the 24-bit register by consecutive 8-bit
I/O operations on the same address; however, the iomap_copy and regmap
bulk functions operate on different addresses.
I'm not sure if there really is a way to make the 104-QUAD-8 operation
more generic for other drivers because it configures the current byte
pointer through a separate register from the data register (all of this
feel rather device specific), so I suspect keeping this function local
to 104-quad-8 is best for now.
William Breathitt Gray
Attachment:
signature.asc
Description: PGP signature
