Re: [PATCH net-next] net: dsa: bcm_sf2: provide own phylink MAC operations

["Russell King (Oracle)" <linux@armlinux.org.uk>]

On Thu, Apr 18, 2024 at 09:52:02AM -0700, Florian Fainelli wrote:
> On 4/16/24 20:13, Florian Fainelli wrote:
> > 
> > 
> > On 4/16/2024 11:16 AM, Russell King (Oracle) wrote:
> > > On Tue, Apr 16, 2024 at 10:44:38AM -0700, Florian Fainelli wrote:
> > > > On 4/16/24 03:19, Russell King (Oracle) wrote:
> > > > > Convert bcm_sf2 to provide its own phylink MAC operations, thus
> > > > > avoiding the shim layer in DSA's port.c
> > > > > 
> > > > > Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
> > > > 
> > > > Acked-by: Florian Fainelli <florian.fainelli@broadcom.com>
> > > > Tested-by: Florian Fainelli <florian.fainelli@broadcom.com>
> > > 
> > > Great, thanks for testing.
> > > 
> > > (Unrelated to this patch... so please don't delay applying based on
> > > ongoing discussion!)
> > > 
> > > The other Broadcom driver, b53, isn't going to be as simple - I believe
> > > it uses a mixture of the .adjust_link method for shared ports, and
> > > .phylink_mac_* for user ports. That makes it very awkward now, given
> > > the check that was added (and suggested by Vladimir) to check for the
> > > legacy methods if dsa_switch's .phylink_mac_ops is populated.
> > > 
> > > Is there any scope for converting b53 to use only phylink methods for
> > > everything, thus eliminating the .adjust_link callback?
> > 
> > It is on the TODO list for sure, and there might be a window later this
> > week to actually work on removing the adjust_link callback once and for
> > all.
> 
> This is what I came up with so far:
> 
> https://github.com/ffainelli/linux/commits/b53-phylink
> 
> Will test later today on the various devices I have.

A few comments...

In "net: dsa: b53: Configure RGMII for 531x5 and MII for 5325", the
commit description says about adding to b53_mac_config(), but the
code change modifies b53_phylink_mac_link_up(). I would think that
the former (as mentioned in the commit message) was the better
place for this rather than in b53_phylink_mac_link_up().

However, I see you move this to b53_phylink_mac_prepare() in
"net: dsa: b53: Move MII/RGMII configuration to mac_prepare" but
I'm not sure why you've chosen mac_prepare() over mac_config().

FYI, the order in which methods at the MAC and PCS are called on a
reconfiguration is:

	mac_prepare()
	if changing pcs && oldpcs:
		oldpcs::pcs_disable()
	if newpcs:
		newpcs::pre_config()
	mac_config()
	if newpcs:
		newpcs::post_config()
		if changing pcs:
			newpcs::pcs_enable()
		newpcs::pcs_config()
		newpcs::pcs_an_restart()
	mac_finish()

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!