[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <707fbd5a-5e1b-4034-a1ca-337ec7eca1bd@lunn.ch>
Date: Thu, 4 Dec 2025 22:07:34 +0100
From: Andrew Lunn <andrew@...n.ch>
To: John Madieu <john.madieu.xa@...renesas.com>
Cc: prabhakar.mahadev-lad.rj@...renesas.com, andrew+netdev@...n.ch,
davem@...emloft.net, edumazet@...gle.com, kuba@...nel.org,
pabeni@...hat.com, robh@...nel.org, krzk+dt@...nel.org,
conor+dt@...nel.org, geert+renesas@...der.be,
biju.das.jz@...renesas.com, claudiu.beznea@...on.dev,
linux@...linux.org.uk, magnus.damm@...il.com,
mcoquelin.stm32@...il.com, alexandre.torgue@...s.st.com,
netdev@...r.kernel.org, linux-renesas-soc@...r.kernel.org,
devicetree@...r.kernel.org
Subject: Re: [PATCH net-next 0/3] net: stmmac: add physical port
identification support
On Thu, Dec 04, 2025 at 04:37:26PM +0000, John Madieu wrote:
> This series adds physical port identification support to the stmmac driver,
> enabling userspace to query hardware-stable identifiers for network interfaces
> via ndo_get_phys_port_id() and ndo_get_phys_port_name().
>
> On systems with multiple ethernet controllers sharing the same driver,
> physical port identification provides stable identifiers that persist
> across reboots and are independent of interface enumeration order.
> This is particularly useful for predictable network interface naming
> and for correlating interfaces with physical connectors.
>
> The implementation follows a two-tier approach:
>
> 1. Generic stmmac support: Default implementations use the permanent MAC
> address as port ID and bus_id for port naming. This provides immediate
> benefit for all stmmac-based platforms.
This information is already available to user space via
sysfs. udev/systemd can use this. How does systemd currently name
these interfaces using its rules?
https://www.freedesktop.org/software/systemd/man/latest/systemd.net-naming-scheme.html
> 2. Glue driver override: Platform drivers can provide custom callbacks
> for hardware-specific identification schemes. The Renesas GBETH driver
> implements this to support device tree-based port identification,
> addressing cases where hardware lacks unique identification registers.
Why is the MAC address not sufficient? What makes Renesas GBETH
special so it needs something different?
Andrew
Powered by blists - more mailing lists