netdev - Re: [PATCH] net: Revert tx queue length on partial failure in dev_qdisc_change_tx_queue

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]

Message-ID: <CANn89i+dif2qjKM6oO1o=BKutXoO6w9kWnnPfc50BDBJ7VpAeQ@mail.gmail.com>
Date: Tue, 22 Jul 2025 09:28:14 -0700
From: Eric Dumazet <edumazet@...gle.com>
To: Suchit K <suchitkarunakaran@...il.com>
Cc: davem@...emloft.net, kuba@...nel.org, pabeni@...hat.com, horms@...nel.org, 
	jhs@...atatu.com, xiyou.wangcong@...il.com, jiri@...nulli.us, sdf@...ichev.me, 
	kuniyu@...gle.com, aleksander.lobakin@...el.com, netdev@...r.kernel.org, 
	skhan@...uxfoundation.org, linux-kernel-mentees@...ts.linux.dev, 
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH] net: Revert tx queue length on partial failure in dev_qdisc_change_tx_queue_len()

On Tue, Jul 22, 2025 at 9:22 AM Suchit K <suchitkarunakaran@...il.com> wrote:
>
> >
> > WRITE_ONCE() is missing.
>
> Oops, I'm sorry about that.
>
> >
> > > +               while (i >= 0) {
> > > +                       qdisc_change_tx_queue_len(dev, &dev->_tx[i]);
> >
> > What happens if one of these calls fails ?
> >
> > I think a fix will be more complicated...
>
> I did consider that, but since I didn’t have a solution, I assumed it
> wouldn’t fail.

But this definitely could fail. Exactly the same way than the first time.

I also have a question. In the Qdisc_ops structure,
> there’s a function pointer for change_tx_queue_len, but I was only
> able to find a single implementation which is
> pfifo_fast_change_tx_queue_len. Is that the only one? Apologies if
> this isn’t the right place to ask such questions. I’d really
> appreciate any feedback. Thank you!

I think only pfifo_fast has to re-allocate its data structures.

Other qdiscs eventually dynamically read dev->tx_queue_len (thus the
WRITE_ONCE() I mentioned to you)