[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CACRpkdY+T8Rqg_irkLNvAC+o_QfwO2N+eB9X-y24t34_9Rg3ww@mail.gmail.com>
Date: Sun, 26 Nov 2023 00:46:03 +0100
From: Linus Walleij <linus.walleij@...aro.org>
To: Andrew Lunn <andrew@...n.ch>
Cc: Florian Fainelli <f.fainelli@...il.com>, Vladimir Oltean <olteanv@...il.com>,
"David S. Miller" <davem@...emloft.net>, Eric Dumazet <edumazet@...gle.com>,
Jakub Kicinski <kuba@...nel.org>, Paolo Abeni <pabeni@...hat.com>,
Christian Marangi <ansuelsmth@...il.com>, Tim Harvey <tharvey@...eworks.com>,
linux-kernel@...r.kernel.org, netdev@...r.kernel.org
Subject: Re: [PATCH RFC] net: dsa: mv88e6xxx: Support LED control
On Thu, Nov 23, 2023 at 5:13 PM Andrew Lunn <andrew@...n.ch> wrote:
> What i would really like to see happen is that the DSA core handles
> the registration of the LEDs, similar to how phylib does. The DT
> binding should be identical for all DSA devices, so there is no need
> for each driver to do its own parsing.
>
> There are some WIP patches at
>
> https://github.com/lunn/linux.git leds-offload-support-reduced-auto-netdev
>
> which implement this. Feel free to make use of them.
Oh it's quite a lot of patches, I really cannot drive that because there are
so many things about them that I don't understand the thinking behind...
But I like what I see!
While I defined the bits a bit differently, some of it looks similar enough.
> > +/* Entries are listed in selector order */
> > +static const struct mv88e6xxx_led_hwconfig mv88e6xxx_led_hwconfigs[] = {
>
> You need to be careful with naming. These are probably specific to the
> 6352. Different switches probably have different capabilities. So it
> would be good to have the names reflect the switch family they are
> valid for.
OK I'll try to name them like such.
> When we come to add support for other switch families, i wounder how
> tables like this scale. Is there some things which can be shared, if
> we break the table up? I need to check the data sheets.
We will see I guess. It falls back to the whole question of whether
supporting all variants in a single module is even scaling. So far
it does I guess? One day the MV88E6xxx may need to be broken
into submodules.
Yours,
Linus Walleij
Powered by blists - more mailing lists