Re: [RFC PATCH v1 2/9] net: dsa: microchip: ksz8795: add phylink support

[Oleksij Rempel]

Hi Vladimir,

On Thu, May 06, 2021 at 03:13:21PM +0300, Vladimir Oltean wrote:
> Hi Oleksij,
> 
> On Wed, May 05, 2021 at 11:20:18AM +0200, Oleksij Rempel wrote:
> > From: Michael Grzeschik <m.grzeschik@xxxxxxxxxxxxxx>
> > 
> > This patch adds the phylink support to the ksz8795 driver, since
> > phylib is obsolete for dsa drivers.
> > 
> > Signed-off-by: Michael Grzeschik <m.grzeschik@xxxxxxxxxxxxxx>
> > Signed-off-by: Oleksij Rempel <o.rempel@xxxxxxxxxxxxxx>
> > ---
> >  drivers/net/dsa/microchip/ksz8795.c | 73 +++++++++++++++++++++++++++++
> >  1 file changed, 73 insertions(+)
> > 
> > diff --git a/drivers/net/dsa/microchip/ksz8795.c b/drivers/net/dsa/microchip/ksz8795.c
> > index 4ca352fbe81c..0ddaf2547f18 100644
> > --- a/drivers/net/dsa/microchip/ksz8795.c
> > +++ b/drivers/net/dsa/microchip/ksz8795.c
> > @@ -18,6 +18,7 @@
> >  #include <linux/micrel_phy.h>
> >  #include <net/dsa.h>
> >  #include <net/switchdev.h>
> > +#include <linux/phylink.h>
> >  
> >  #include "ksz_common.h"
> >  #include "ksz8795_reg.h"
> > @@ -1420,11 +1421,83 @@ static int ksz8_setup(struct dsa_switch *ds)
> >  	return 0;
> >  }
> >  
> > +static int ksz_get_state(struct dsa_switch *ds, int port,
> > +					  struct phylink_link_state *state)
> > +{
> > +	struct ksz_device *dev = ds->priv;
> > +	struct ksz8 *ksz8 = dev->priv;
> > +	const u8 *regs = ksz8->regs;
> > +	u8 speed, link;
> > +
> > +	ksz_pread8(dev, port, regs[P_LINK_STATUS], &link);
> > +	ksz_pread8(dev, port, regs[P_SPEED_STATUS], &speed);
> > +
> > +	state->link = !!(link & PORT_STAT_LINK_GOOD);
> > +	if (state->link) {
> > +		state->speed =
> > +			(speed & PORT_STAT_SPEED_100MBIT) ? SPEED_100 : SPEED_10;
> > +		state->duplex =
> > +			(speed & PORT_STAT_FULL_DUPLEX) ? DUPLEX_FULL : DUPLEX_HALF;
> > +	}
> > +
> > +	return 0;
> > +}
> > +
> > +static void ksz_validate(struct dsa_switch *ds, int port,
> > +			       unsigned long *supported,
> > +			       struct phylink_link_state *state)
> > +{
> > +	__ETHTOOL_DECLARE_LINK_MODE_MASK(mask) = { 0, };
> > +	struct ksz_device *dev = ds->priv;
> > +
> > +	if (port == dev->cpu_port) {
> > +		if ((state->interface != PHY_INTERFACE_MODE_RMII) &&
> > +		   (state->interface != PHY_INTERFACE_MODE_MII))
> > +			goto unsupported;
> > +	} else if (port > dev->port_cnt) {
> > +		bitmap_zero(supported, __ETHTOOL_LINK_MODE_MASK_NBITS);
> > +		dev_err(ds->dev, "Unsupported port: %i\n", port);
> > +		return;
> > +	} else {
> > +		if (state->interface != PHY_INTERFACE_MODE_INTERNAL)
> > +			goto unsupported;
> > +	}
> > +
> > +	/* Allow all the expected bits */
> > +	phylink_set_port_modes(mask);
> > +	phylink_set(mask, Autoneg);
> > +
> > +	phylink_set(mask, Pause);
> > +	/* Silicon Errata Sheet (DS80000830A): Asym_Pause limit to port 2 */
> > +	if (port || !ksz_is_ksz88x3(dev))
> > +		phylink_set(mask, Asym_Pause);
> > +
> > +	/* 10M and 100M are only supported */
> > +	phylink_set(mask, 10baseT_Half);
> > +	phylink_set(mask, 10baseT_Full);
> > +	phylink_set(mask, 100baseT_Half);
> > +	phylink_set(mask, 100baseT_Full);
> > +
> > +	bitmap_and(supported, supported, mask,
> > +		   __ETHTOOL_LINK_MODE_MASK_NBITS);
> > +	bitmap_and(state->advertising, state->advertising, mask,
> > +		   __ETHTOOL_LINK_MODE_MASK_NBITS);
> > +
> > +	return;
> > +
> > +unsupported:
> > +	bitmap_zero(supported, __ETHTOOL_LINK_MODE_MASK_NBITS);
> > +	dev_err(ds->dev, "Unsupported interface: %d, port: %d\n",
> > +		state->interface, port);
> > +}
> > +
> >  static const struct dsa_switch_ops ksz8_switch_ops = {
> >  	.get_tag_protocol	= ksz8_get_tag_protocol,
> >  	.setup			= ksz8_setup,
> >  	.phy_read		= ksz_phy_read16,
> >  	.phy_write		= ksz_phy_write16,
> > +	.phylink_validate	= ksz_validate,
> > +	.phylink_mac_link_state	= ksz_get_state,
> >  	.phylink_mac_link_down	= ksz_mac_link_down,
> >  	.port_enable		= ksz_enable_port,
> >  	.get_strings		= ksz8_get_strings,
> > -- 
> > 2.29.2
> > 
> 
> I've asked Prasanna about this too, but for one reason or another I am
> still not edified. Is this change a compliance thing, or do you actually
> gain anything at all from phylink?
> https://patchwork.kernel.org/project/netdevbpf/patch/20210422094257.1641396-6-prasanna.vengateshan@xxxxxxxxxxxxx/
> What made you submit the patch?

This switch has proper flow control only on one port, see 
Module 4: Port 1 does not respond to received flow control PAUSE frames:
http://ww1.microchip.com/downloads/en/DeviceDoc/KSZ8873-Errata-DS80000830A.pdf

With phylink we can disable Pause here, instead of bypassing it to the PHY
driver.

Regards,
Oleksij
-- 
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |