linux-kernel - Re: [PATCH net v2 0/2] tcp_metrics: fix hanlding of route options

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 for Android: free password hash cracker in your pocket

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

Message-ID: <20250623093604.01b74726@mordecai.tesarici.cz>
Date: Mon, 23 Jun 2025 09:36:04 +0200
From: Petr Tesarik <ptesarik@...e.com>
To: Eric Dumazet <edumazet@...gle.com>
Cc: Paolo Abeni <pabeni@...hat.com>, "David S. Miller"
 <davem@...emloft.net>, Neal Cardwell <ncardwell@...gle.com>, Kuniyuki
 Iwashima <kuniyu@...gle.com>, "open list:NETWORKING [TCP]"
 <netdev@...r.kernel.org>, David Ahern <dsahern@...nel.org>, Jakub Kicinski
 <kuba@...nel.org>, open list <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH net v2 0/2] tcp_metrics: fix hanlding of route options

On Fri, 20 Jun 2025 06:24:23 -0700
Eric Dumazet <edumazet@...gle.com> wrote:

> On Fri, Jun 20, 2025 at 5:57 AM Petr Tesarik <ptesarik@...e.com> wrote:
> >
> > I ran into a couple of issues while trying to tweak TCP congestion
> > avoidance to analyze a potential performance regression. It turns out
> > that overriding the parameters with ip-route(8) does not work as
> > expected and appears to be buggy.  
> 
> Hi Petr
> 
> Could you add packetdrill tests as well ?

Glad to do that. But it will be my first time. ;-) Is there a tutorial?
I looked under Documentation/ and didn't see anything.

> Given this could accidentally break user setups, maybe net-next would be safer.

Yeah, you're right. Technically, it is a bugfix, but if it's been
broken for more than a decade without anyone complaining, it can't be
super-urgent.

> Some of us disable tcp_metrics, because metrics used one minute (or
> few seconds) in the past are not very helpful, and source of
> confusion.
> 
> (/proc/sys/net/ipv4/tcp_no_metrics_save set to 1)

Yes, I know about that one. FWIW it didn't help at all in my case,
because then the value from the routing table was ALWAYS ignored...

Petr T