Re: [PATCH net-next 01/10] net: pcs: xpcs: move PCS reset to .pcs_pre_config()

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

On Tue, Oct 01, 2024 at 11:34:42PM +0300, Serge Semin wrote:
> Hi Russell
> 
> On Tue, Oct 01, 2024 at 05:04:10PM GMT, Russell King (Oracle) wrote:
> > Move the PCS reset to .pcs_pre_config() rather than at creation time,
> > which means we call the reset function with the interface that we're
> > actually going to be using to talk to the downstream device.
> > 
> > Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com>
> > Tested-by: Vladimir Oltean <vladimir.oltean@nxp.com> # sja1105
> > Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
> 
> Continuing the RFC discussion. As I mentioned here:
> https://lore.kernel.org/netdev/mykeabksgikgk6otbub2i3ksfettbozuhqy3gt5vyezmemvttg@cpjn5bcfiwei/
> The PCS-reset procedure actually can be converted to being independent
> from the PHY-interface. Thus you won't need to move the PCS resetting
> to the pre_config() method, and get rid from the pointer to
> dw_xpcs_compat utilization each time the reset is required.
> 
> So why not to merge in my patch to your series as a pre-requisite
> change and then this patch can be converted to just dropping the
> xpcs_find_compat() method call from the xpcs_init_iface() function?
> Alternatively the dropping can be just incorporated into my patch.

I'm wondering why we seem to be having a communication issue here.

I'm not sure which part of "keeping the functional changes to a
minimum for a cleanup series" you're not understanding. This is
one of the basics for kernel development... and given that you're
effectively maintaining stmmac, it's something you _should_ know.

So no, I'm going to outright refuse to merge your patch in to this
series, because as I see it, it would be wrong to do so. This is
a _cleanup_ series, not a functional change series, and what you're
proposing _changes_ the _way_ reset happens in this driver beyond
the minimum that is required for this cleanup. It's introducing a
completely _new_ way of writing to the devices registers to do
the reset that's different.

The more differences there are, the more chances there are of
regressions.

So, again, no..

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