netdev - Re: Unrecognized RTL8125 in r8169

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>] [day] [month] [year] [list]

Message-ID: <66af2954-b34d-73d6-d7eb-d697b709b5c7@gmail.com>
Date:   Wed, 3 Jun 2020 21:43:21 +0200
From:   Heiner Kallweit <hkallweit1@...il.com>
To:     Eivind Uggedal <eivind@...edal.com>
Cc:     "netdev@...r.kernel.org" <netdev@...r.kernel.org>
Subject: Re: Unrecognized RTL8125 in r8169

On 03.06.2020 20:57, Eivind Uggedal wrote:
> Hi,
> 
> I saw you added the initial RTL8125 support in r8169.
> 
> I have a recent x570 motherboard (MSI x570 Tomahawk) with onboard RTL8125. With r8169 from 5.4 and 5.7 I get:
> 
>> unknown chip XID 641
> 
> from the r8169 driver since XID==641 does not match any of the mask/val in rtl8169_get_mac_version().
> 
> I did not see any relevant unmerged changes in the netdev-next tree.
> 
> I have a hard time following these "magic" masks/values and could not correlate them to anything in the Realtek oot r8125 driver. I have some C experience, but zero driver/low level experience.
> 
Thanks for the report. XID 641 is CFG_METHOD_5 = RTL8125B in the r8125 vendor driver.
This new chip version isn't supported yet by the r8169 driver. Typically I wait for
reports like yours because:
- Not every chip version in the vendor driver makes it to the market.
  Adding support for it would be wasted effort.
- I don't have hardware for testing. I need users like you for it ..

> I could try to just fallback to RTL_GIGIA_MAC_VER_60 or RTL_GIGIA_MAC_VER_61 locally, but ideally I'd like a proper patch that can be mainlined.
> 
This may or may not work, partially also depending on the link partner. Typically each
chip version needs it own quirks, therefore they are treated differently by the vendor driver.

> At least one other poor soul has the same issue: 
> https://forums.unraid.net/topic/92998-dhcp-networking-issue-v-683/
> 
> Seeing as this board had a world wide release just days ago maybe this is a new revision of the RTL8125 chipset?
> 
> Appreciate any pointers and help to debug it further!
> 
There is not really something to debug. I have to port all chip quirks from
the vendor driver and ask Realtek for a firmware file.
I'll put it on my roadmap, maybe it makes it into 5.9 or 5.10. Until then
you would have to use vendor driver r8125.

> Regards,
> Eivind Uggedal

Heiner