netdev - Re: PHY firmware update method

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Thu, 29 Sep 2022 09:04:07 +0200
From:   Jiri Pirko <jiri@...nulli.us>
To:     Andrew Lunn <andrew@...n.ch>
Cc:     Michael Walle <michael@...le.cc>,
        Heiner Kallweit <hkallweit1@...il.com>,
        Russell King <linux@...linux.org.uk>, netdev@...r.kernel.org,
        kuba@...nel.org
Subject: Re: PHY firmware update method

Wed, Sep 28, 2022 at 02:28:27PM CEST, andrew@...n.ch wrote:
>On Wed, Sep 28, 2022 at 01:27:13PM +0200, Michael Walle wrote:
>> Hi,
>> 
>> There are PHYs whose firmware can be updated. Usually, they have
>> an internal ROM and you can add patches on top of that, or there
>> might be an external flash device which can have a more recent
>> firmware version installed which can be programmed in-place
>> through the PHY.
>> 
>> The firmware update for a PHY is usually quite simple, but there
>> seems to be no infrastructure in the kernel for that. There is the
>> ETHTOOL_FLASHDEV ioctl for upgrading the firmware of a NIC it seems.
>> Other than that I haven't found anything. And before going in a wrong
>> directions I'd like to hear your thoughts on how to do it. I.e. how
>> should the interface to the userspace look like.
>> 
>> Also I think the PHY should be taken offline, similar to the cable
>> test.
>
>I've seen a few different ways of doing this.
>
>One is to load the firmware from disk every boot using
>request_firmware(). Then parse the header, determine if it is newer
>than what the PHY is already using, and if so, upgrade the PHY. If you
>do this during probe, it should be transparent, no user interaction
>required.
>
>I've also seen the FLASH made available as just another mtd
>device. User space can then write to it, and then do a {cold} boot.
>
>devlink has become the standard way for upgrading firmware on complex
>network devices, like NICs and TOR switches. That is probably a good
>solution here. The problem is, what devlink instance to use. Only a
>few MAC drivers are using devlink, so it is unlikely the MAC driver
>the PHY is attached to has a devlink instance. Do we create a devlink
>instance for the PHY?

Ccing Jakub. I don't think it is good idea to create a devlink instance
per-PHY. However, on the other hand, we have a devlink instance per
devlink linecard now. The devlink linecard however has devlink
representation, which PHY does not have.

Perhaps now is the time to dust-off my devlink components implementation
and use it for PHYs? IDF. Jakub, WDYT.


>
>You might want to talk to Jiri about this.
>
>The other issue is actually getting the firmware. Many manufactures
>seem reluctant to allow redistribution as required by linux-firmware.
>There is no point adding firmware upgrade if you cannot redistribute
>the firmware.
>
>    Andrew