netdev - Re: [Bug 209423] WARN_ON_ONCE() at rtl8169_tso_csum

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: <CANn89iK1pFsa76PCB2Bu0ZtJNuAG9mwhVtQUKeQByCuURFac_w@mail.gmail.com>
Date:   Tue, 19 Jan 2021 14:58:29 +0100
From:   Eric Dumazet <edumazet@...gle.com>
To:     Heiner Kallweit <hkallweit1@...il.com>
Cc:     Juerg Haefliger <juerg.haefliger@...onical.com>,
        Eric Dumazet <eric.dumazet@...il.com>,
        "netdev@...r.kernel.org" <netdev@...r.kernel.org>,
        Microchip Linux Driver Support <UNGLinuxDriver@...rochip.com>,
        Woojung Huh <woojung.huh@...rochip.com>
Subject: Re: [Bug 209423] WARN_ON_ONCE() at rtl8169_tso_csum_v2()

On Tue, Jan 19, 2021 at 2:47 PM Heiner Kallweit <hkallweit1@...il.com> wrote:
>
>
> In tcp_add_backlog() we have the following that looks like it could
> be related to the problem. gso_type doesn't get set, not sure however
> whether this is a bug or intentional (because we expect gso_type
> to be set already or because it's supposed to be set somewhere else).
> Meybe Eric can comment on this.
>
>
>         if (!shinfo->gso_size)
>                 shinfo->gso_size = skb->len - hdrlen;
>
>         if (!shinfo->gso_segs)
>                 shinfo->gso_segs = 1;
>

Yes, at this point TCP is supposed to own the skb, which is partially true.

Check for skb_cloned() in places like skb_try_coalesce()

I think that calling skb_unclone() would be terribly expensive for all
these USB drivers having fake skb
(all clones from a giant one), and thus very big headroom that would
be copied from generic expand head.