netdev - Re: [PATCH net-next v2 01/11] tls: rx: allow only one reader at a time

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: <CANn89iKcmSfWgvZjzNGbsrndmCch2HC_EPZ7qmGboDNaWoviNQ@mail.gmail.com>
Date:   Wed, 20 Jul 2022 19:09:59 +0200
From:   Eric Dumazet <edumazet@...gle.com>
To:     Jakub Kicinski <kuba@...nel.org>
Cc:     David Miller <davem@...emloft.net>,
        netdev <netdev@...r.kernel.org>, Paolo Abeni <pabeni@...hat.com>,
        Boris Pismenny <borisp@...dia.com>,
        John Fastabend <john.fastabend@...il.com>,
        Maxim Mikityanskiy <maximmi@...dia.com>,
        Tariq Toukan <tariqt@...dia.com>, vfedorenko@...ek.ru
Subject: Re: [PATCH net-next v2 01/11] tls: rx: allow only one reader at a time

On Wed, Jul 20, 2022 at 6:59 PM Jakub Kicinski <kuba@...nel.org> wrote:
>
> On Wed, 20 Jul 2022 10:37:02 +0200 Eric Dumazet wrote:
> > > +               if (!timeo)
> > > +                       return -EAGAIN;
> >
> > We return with socket lock held, and callers seem to not release the lock.
> >
> > > +               if (signal_pending(current))
> > > +                       return sock_intr_errno(timeo);
> >
> > same here.
>
> Thanks a lot for catching these.
>
> > Let's wait for syzbot to catch up :)
>
> I'll send the fixes later today. This is just a passing comment, right?
> There isn't a report you know is coming? Otherwise I can wait to give
> syzbot credit, too.

I now have a full syzbot report, with a repro and bisection, I am
releasing it now.

>
> I have two additional questions while I have you :)
>
> Is the timeo supposed to be for the entire operation? Right now TLS
> seems to use a fresh timeo every time it goes to wait so the cumulative
> wait can be much longer, as long as some data keeps coming in :/

Good question. I am not sure how this timeout is used in applications, but
I would think it serves as a way to make sure a stall is detected.
So restarting the timeout every time there is progress would make sense.

Application needing a different behavior can still use regular timer,
independent of networking, ( alarm() being the most simple one)

>
> Last one - I posted a bit of a disemboweling patch for TCP, LMK if it's
> no bueno:
>
> https://lore.kernel.org/all/20220719231129.1870776-6-kuba@kernel.org/