netdev - Re: [PATCH net-next v2] net: pppoe: implement GRO support

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

Message-ID: <439bd3e3-85d1-44b9-9587-052bd9c35dec@nbd.name>
Date: Tue, 22 Jul 2025 16:56:26 +0200
From: Felix Fietkau <nbd@....name>
To: Paolo Abeni <pabeni@...hat.com>, Eric Dumazet <edumazet@...gle.com>
Cc: netdev@...r.kernel.org, Michal Ostrowski <mostrows@...thlink.net>,
 Andrew Lunn <andrew+netdev@...n.ch>, "David S. Miller"
 <davem@...emloft.net>, Jakub Kicinski <kuba@...nel.org>,
 linux-kernel@...r.kernel.org
Subject: Re: [PATCH net-next v2] net: pppoe: implement GRO support

On 22.07.25 16:47, Paolo Abeni wrote:
> On 7/22/25 4:04 PM, Eric Dumazet wrote:
>> On Wed, Jul 16, 2025 at 1:14 AM Felix Fietkau <nbd@....name> wrote:
>>>
>>> Only handles packets where the pppoe header length field matches the exact
>>> packet length. Significantly improves rx throughput.
>>>
>>> When running NAT traffic through a MediaTek MT7621 devices from a host
>>> behind PPPoE to a host directly connected via ethernet, the TCP throughput
>>> that the device is able to handle improves from ~130 Mbit/s to ~630 Mbit/s,
>>> using fraglist GRO.
>>>
>>> Signed-off-by: Felix Fietkau <nbd@....name>
>>> ---
>> 
>> Shouldn't we first add GSO support ?
> 
> I *think* the current __skb_gso_segment() should be able to segment a
> pppoe GSO packet, as the pppoe header is static/constant, skb->mac_len
> will include both eth/pppoe and skb->protocol should be the actual L3.

If the packet is received on a ppp device, __skb_gso_segment works 
afterwards.
However, for the bridge forwarding case, Eric is correct. In this case, 
skb->protocol will not be the L3 protocol.

I will add GSO support in v3.

Thanks,

- Felix