Re: [PATCH v10] net: mdio: Add RTL9300 MDIO driver
From: Andrew Lunn
Date: Thu Mar 13 2025 - 16:40:53 EST
On Thu, Mar 13, 2025 at 08:37:18PM +0000, Chris Packham wrote:
> On 14/03/2025 09:35, Andrew Lunn wrote:
> > On Thu, Mar 13, 2025 at 07:54:39PM +0000, Chris Packham wrote:
> >> +cc netdev, lkml
> >>
> >> On 14/03/2025 01:34, Andrew Lunn wrote:
> >>>> + /* Put the interfaces into C45 mode if required */
> >>>> + glb_ctrl_mask = GENMASK(19, 16);
> >>>> + for (i = 0; i < MAX_SMI_BUSSES; i++)
> >>>> + if (priv->smi_bus_is_c45[i])
> >>>> + glb_ctrl_val |= GLB_CTRL_INTF_SEL(i);
> >>>> +
> >>>> + fwnode_for_each_child_node(node, child)
> >>>> + if (fwnode_device_is_compatible(child, "ethernet-phy-ieee802.3-c45"))
> >>>> + priv->smi_bus_is_c45[mdio_bus] = true;
> >>>> +
> >>> This needs more explanation. Some PHYs mix C22 and C45, e.g. the > 1G
> >>> speed support registers are in the C45 address space, but <= 1G is in
> >>> the C22 space. And 1G PHYs which support EEE need access to C45 space
> >>> for the EEE registers.
> >> Ah good point. The MDIO interfaces are either in GPHY (i.e. clause 22)
> >> or 10GPHY mode (i.e. clause 45). This does mean we can't support support
> >> both c45 and c22 on the same MDIO bus (whether that's one PHY that
> >> supports both or two different PHYs). I'll add a comment to that effect
> >> and I should probably only provide bus->read/write or
> >> bus->read_c45/write_c45 depending on the mode.
> > Is there more to it than this? Because why not just set the mode per
> > bus transaction?
>
> It's a bus level setting at init time. You can't dynamically switch modes.
Why not? The bus is only every doing one transaction at a time, so why
not switch it per transaction?
Andrew