Re: [PATCH net-next v3 3/6] net: hisilicon: add support for hisi_femac core on Hi3798MV200

[Andrew Lunn]

On Tue, Feb 20, 2024 at 04:14:36AM +0800, Yang Xiwen wrote:
> On 2/20/2024 4:03 AM, Andrew Lunn wrote:
> > > Note it's unable to put the MDIO bus node outside of MAC controller
> > > (i.e. at the same level in the parent bus node). Because we need to
> > > control all clocks and resets in FEMAC driver due to the phy reset
> > > procedure. So the clocks can't be assigned to MDIO bus device, which is
> > > an essential resource for the MDIO bus to work.
> > What PHY driver is being used? If there a specific PHY driver for this
> > hardware? Does it implement soft reset?
> 
> I'm using generic PHY driver.
> 
> It implements IEEE C22 standard. So there is a soft reset in BMCR register.
> 
> > 
> > I'm wondering if you can skip hardware reset of the PHY and only do a
> > software reset.
> 
> There must be someone to deassert the hardware reset control signal for the
> PHY. We can't rely on the boot loader to do that. And here even we choose to
> skip the hardware reset procedure, the sequence of deasserting the reset
> signals is also very important. (i.e. first PHY, then MAC and MACIF).
> Opposite to the normal sequence. (we normally first register MAC driver, and
> then PHY).

There are a few MACs which require the PHY to provide a clock to the
MAC before they can use their DMA engine. The PHY provides typically a
25MHz clock, which is used to driver the DMA. So long as you don't
touch the DMA, you can access other parts of the MAC before the PHY is
generating the clock.

So it might be possible to take the MAC and MACIF out of reset, then
create the MDIO bus, probe the PHY, take it out of reset so its
generating the clock, and then complete the rest of the MAC setup.

	Andrew