| Message ID | E1nNdJV-00AsoS-Qi@rmk-PC.armlinux.org.uk (mailing list archive) |
|---|---|
| State | Accepted |
| Commit | 79fda660bdbba839f434cf8674ff4a20d88fbeb2 |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | net: dsa: ocelot: phylink updates | expand |
On Fri, Feb 25, 2022 at 04:19:25PM +0000, Russell King (Oracle) wrote: > Populate the supported interfaces bitmap for the Ocelot DSA switches. > > Since all sub-drivers only support a single interface mode, defined by > ocelot_port->phy_mode, we can handle this in the main driver code > without reference to the sub-driver. > > Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk> > --- Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Tested-by: Vladimir Oltean <vladimir.oltean@nxp.com>
On Fri, Feb 25, 2022 at 04:25:30PM +0000, Vladimir Oltean wrote: > On Fri, Feb 25, 2022 at 04:19:25PM +0000, Russell King (Oracle) wrote: > > Populate the supported interfaces bitmap for the Ocelot DSA switches. > > > > Since all sub-drivers only support a single interface mode, defined by > > ocelot_port->phy_mode, we can handle this in the main driver code > > without reference to the sub-driver. > > > > Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk> > > --- > > Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> > Tested-by: Vladimir Oltean <vladimir.oltean@nxp.com> Brilliant, thanks. This is the final driver in net-next that was making use of phylink_set_pcs(), so once this series is merged, that function will only be used by phylink internally. The next patch I have in the queue is to remove that function. Marek Behún will be very happy to see phylink_set_pcs() gone.
On Fri, 25 Feb 2022 16:31:52 +0000 "Russell King (Oracle)" <linux@armlinux.org.uk> wrote: > On Fri, Feb 25, 2022 at 04:25:30PM +0000, Vladimir Oltean wrote: > > On Fri, Feb 25, 2022 at 04:19:25PM +0000, Russell King (Oracle) wrote: > > > Populate the supported interfaces bitmap for the Ocelot DSA switches. > > > > > > Since all sub-drivers only support a single interface mode, defined by > > > ocelot_port->phy_mode, we can handle this in the main driver code > > > without reference to the sub-driver. > > > > > > Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk> > > > --- > > > > Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> > > Tested-by: Vladimir Oltean <vladimir.oltean@nxp.com> > > Brilliant, thanks. > > This is the final driver in net-next that was making use of > phylink_set_pcs(), so once this series is merged, that function will > only be used by phylink internally. The next patch I have in the queue > is to remove that function. > > Marek Behún will be very happy to see phylink_set_pcs() gone. Yes, finally we can convert mv88e6xxx fully :)
On Fri, Feb 25, 2022 at 06:16:53PM +0100, Marek Behún wrote: > On Fri, 25 Feb 2022 16:31:52 +0000 > "Russell King (Oracle)" <linux@armlinux.org.uk> wrote: > > > On Fri, Feb 25, 2022 at 04:25:30PM +0000, Vladimir Oltean wrote: > > > On Fri, Feb 25, 2022 at 04:19:25PM +0000, Russell King (Oracle) wrote: > > > > Populate the supported interfaces bitmap for the Ocelot DSA switches. > > > > > > > > Since all sub-drivers only support a single interface mode, defined by > > > > ocelot_port->phy_mode, we can handle this in the main driver code > > > > without reference to the sub-driver. > > > > > > > > Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk> > > > > --- > > > > > > Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> > > > Tested-by: Vladimir Oltean <vladimir.oltean@nxp.com> > > > > Brilliant, thanks. > > > > This is the final driver in net-next that was making use of > > phylink_set_pcs(), so once this series is merged, that function will > > only be used by phylink internally. The next patch I have in the queue > > is to remove that function. > > > > Marek Behún will be very happy to see phylink_set_pcs() gone. > > Yes, finally we can convert mv88e6xxx fully :) ... changing the subject line to show we've drifted off topic ... Yes, once we've worked out what the PCS interface should look like in order to deal with the 88E6393 errata workaround that needs to be run each time the interface changes or whenever we "power up" the PCS. I think that needs to be discussed, because I can see no clean and clear solution that doesn't have some kind of down-side. The existing pcs_enable/pcs_disable I have in my tree fits well with the idea of changing a PCS, but not for being called every time we do a major config when the PCS isn't being changed. To see what I mean, would someone who didn't have the 6393 issue be happy with this sequence on every major configuration, even when the PCS isn't being changed: mac_prepare() pcs_disable() mac_config() pcs_enable() pcs_config() pcs_an_restart() mac_finish() I thought about calling pcs_disable() pcs_prepare() but that then throws out pcs_enable(), unless we do that after pcs_config() - but what if pcs_enable() is used to clear the PDOWN bit of the PCS, or other (possibly external) PCS power control that prevents its registers being accessed. I'm also thinking, having had another recent look at mv88e6xxx_mac_config(), we would need to move this: if (mode != MLO_AN_PHY || !mv88e6xxx_phy_is_internal(ds, port)) { /* In inband mode, the link may come up at any time while the * link is not forced down. Force the link down while we * reconfigure the interface mode. */ if (mode == MLO_AN_INBAND && p->interface != state->interface && chip->info->ops->port_set_link) chip->info->ops->port_set_link(chip, port, LINK_FORCED_DOWN); into a mac_prepare() callback so it still happens prior to the PCS being called - which will need to happen between mac_prepare() and mac_config() for the errata. That means extending the mac_prepare() and mac_finish() methods into DSA as well. I do have a patch to add those additional callbacks to DSA, but I currently have no code that makes use of them (so haven't sent it yet.) See "net: dsa: add support for new phylink calls". I think I would prefer at this point - to get the mt7530 changes settled, which will then allow the phylink_helper_basex_speed() helper to be removed, do a few further phylink/sfp code cleanups (using %pe consistently in that code to print errors) and then wait until the next kernel cycle before tackling mv88e6xxx.
> ... changing the subject line to show we've drifted off topic ... > > Yes, once we've worked out what the PCS interface should look like in > order to deal with the 88E6393 errata workaround that needs to be run > each time the interface changes or whenever we "power up" the PCS. Hi Russell The erratas are not limited to 6393. For the 6390 there is an errata where you need to "power up" the PCS before you change cmode, otherwise TX works, but RX just drops frames rather than pass them to the MAC. I've not looked at the details of your proposal, and maybe it is a none issue, i just wanted to make sure you are aware of this. Andrew
On Fri, Feb 25, 2022 at 06:57:33PM +0100, Andrew Lunn wrote: > > ... changing the subject line to show we've drifted off topic ... > > > > Yes, once we've worked out what the PCS interface should look like in > > order to deal with the 88E6393 errata workaround that needs to be run > > each time the interface changes or whenever we "power up" the PCS. > > Hi Russell > > The erratas are not limited to 6393. For the 6390 there is an errata > where you need to "power up" the PCS before you change cmode, > otherwise TX works, but RX just drops frames rather than pass them to > the MAC. > > I've not looked at the details of your proposal, and maybe it is a > none issue, i just wanted to make sure you are aware of this. Thanks - Marek has been keeping me on the straight and narrow with these issues, but it's good to know that it's not necessary for the other Marvell variants. That's been at the back of my mind a bit as we're getting closer to the point that we need to sort this out. I should also mention - once the mt7530 and mv88e6xxx drivers are sorted, I believe we will then be in a position to kill off all the "legacy_pre_march2020" stuff in DSA since DSA will no longer need the legacy phylink behaviour. Sadly, though, that doesn't mean "legacy_pre_march2020" can be removed, we still have mtk_eth_soc reliant on the old behaviour.
diff --git a/drivers/net/dsa/ocelot/felix.c b/drivers/net/dsa/ocelot/felix.c index 9959407fede8..a1a6ace39aab 100644 --- a/drivers/net/dsa/ocelot/felix.c +++ b/drivers/net/dsa/ocelot/felix.c @@ -778,6 +778,15 @@ static int felix_vlan_del(struct dsa_switch *ds, int port, return ocelot_vlan_del(ocelot, port, vlan->vid); } +static void felix_phylink_get_caps(struct dsa_switch *ds, int port, + struct phylink_config *config) +{ + struct ocelot *ocelot = ds->priv; + + __set_bit(ocelot->ports[port]->phy_mode, + config->supported_interfaces); +} + static void felix_phylink_validate(struct dsa_switch *ds, int port, unsigned long *supported, struct phylink_link_state *state) @@ -1587,6 +1596,7 @@ const struct dsa_switch_ops felix_switch_ops = { .get_ethtool_stats = felix_get_ethtool_stats, .get_sset_count = felix_get_sset_count, .get_ts_info = felix_get_ts_info, + .phylink_get_caps = felix_phylink_get_caps, .phylink_validate = felix_phylink_validate, .phylink_mac_config = felix_phylink_mac_config, .phylink_mac_link_down = felix_phylink_mac_link_down,
Populate the supported interfaces bitmap for the Ocelot DSA switches. Since all sub-drivers only support a single interface mode, defined by ocelot_port->phy_mode, we can handle this in the main driver code without reference to the sub-driver. Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk> --- drivers/net/dsa/ocelot/felix.c | 10 ++++++++++ 1 file changed, 10 insertions(+)