| Message ID | 20220330110210.3374165-1-michael@walle.cc (mailing list archive) |
|---|---|
| State | RFC |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | [RFC,net-next] net: lan966x: make PHY reset support optional | expand |
The 03/30/2022 13:02, Michael Walle wrote: Hi Michael, > > The PHY subsystem as well as the MIIM mdio driver (in case of the > integrated PHYs) will already take care of the resets of any external > and internal PHY. There is no need for this reset anymore, so mark it > optionally to be backwards compatible. > > Signed-off-by: Michael Walle <michael@walle.cc> > --- > > Horatiu, what do you think, should it be removed altogether? I think it is OK to remove it altogether. If you get both [1] and [2] in. > There is > no user for that in mainline and I don't know about downstream but the > reset driver doesn't really work (as it also resets the GPIO/SGPIO) Yes because I didn't manage to send yet those patches. But if your get yours in that is fine for me. My problem was, if after the probe of the MDIO controller it was probed the SGPIO then the PHYs will be in reset because the SGPIO was resetting the swich. But you put the reset of the swich on the pinctrl which will be probed before the MDIO, so that should be fine. > and conceptionally the property is on the wrong DT node. All of the > drawbacks should have been addressed by my patches for the miim [1] > and the pinctrl driver [2]. > > [1] https://lore.kernel.org/netdev/20220318201324.1647416-1-michael@walle.cc/ > [2] https://lore.kernel.org/linux-gpio/20220313154640.63813-1-michael@walle.cc/ > > drivers/net/ethernet/microchip/lan966x/lan966x_main.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/drivers/net/ethernet/microchip/lan966x/lan966x_main.c b/drivers/net/ethernet/microchip/lan966x/lan966x_main.c > index 1f8c67f0261b..0765064d2845 100644 > --- a/drivers/net/ethernet/microchip/lan966x/lan966x_main.c > +++ b/drivers/net/ethernet/microchip/lan966x/lan966x_main.c > @@ -916,7 +916,7 @@ static int lan966x_reset_switch(struct lan966x *lan966x) > return dev_err_probe(lan966x->dev, PTR_ERR(switch_reset), > "Could not obtain switch reset"); > > - phy_reset = devm_reset_control_get_shared(lan966x->dev, "phy"); > + phy_reset = devm_reset_control_get_optional_shared(lan966x->dev, "phy"); > if (IS_ERR(phy_reset)) > return dev_err_probe(lan966x->dev, PTR_ERR(phy_reset), > "Could not obtain phy reset\n"); > -- > 2.30.2 >
diff --git a/drivers/net/ethernet/microchip/lan966x/lan966x_main.c b/drivers/net/ethernet/microchip/lan966x/lan966x_main.c index 1f8c67f0261b..0765064d2845 100644 --- a/drivers/net/ethernet/microchip/lan966x/lan966x_main.c +++ b/drivers/net/ethernet/microchip/lan966x/lan966x_main.c @@ -916,7 +916,7 @@ static int lan966x_reset_switch(struct lan966x *lan966x) return dev_err_probe(lan966x->dev, PTR_ERR(switch_reset), "Could not obtain switch reset"); - phy_reset = devm_reset_control_get_shared(lan966x->dev, "phy"); + phy_reset = devm_reset_control_get_optional_shared(lan966x->dev, "phy"); if (IS_ERR(phy_reset)) return dev_err_probe(lan966x->dev, PTR_ERR(phy_reset), "Could not obtain phy reset\n");
The PHY subsystem as well as the MIIM mdio driver (in case of the integrated PHYs) will already take care of the resets of any external and internal PHY. There is no need for this reset anymore, so mark it optionally to be backwards compatible. Signed-off-by: Michael Walle <michael@walle.cc> --- Horatiu, what do you think, should it be removed altogether? There is no user for that in mainline and I don't know about downstream but the reset driver doesn't really work (as it also resets the GPIO/SGPIO) and conceptionally the property is on the wrong DT node. All of the drawbacks should have been addressed by my patches for the miim [1] and the pinctrl driver [2]. [1] https://lore.kernel.org/netdev/20220318201324.1647416-1-michael@walle.cc/ [2] https://lore.kernel.org/linux-gpio/20220313154640.63813-1-michael@walle.cc/ drivers/net/ethernet/microchip/lan966x/lan966x_main.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)