netdev - Re: bpf redirect will loop the packets

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: <2ff91572-443c-dea6-ebed-10bc99d080bc@iogearbox.net>
Date:   Mon, 1 Nov 2021 15:41:24 +0100
From:   Daniel Borkmann <daniel@...earbox.net>
To:     Tonghao Zhang <xiangxia.m.yue@...il.com>,
        Cong Wang <xiyou.wangcong@...il.com>, ast@...nel.org,
        andrii@...nel.org, songliubraving@...com, yhs@...com
Cc:     Linux Kernel Network Developers <netdev@...r.kernel.org>
Subject: Re: bpf redirect will loop the packets

On 11/1/21 7:32 AM, Tonghao Zhang wrote:
> Hi
> I found the issue,
> if we use the bpf_redirect to the same netdevice in the ingress path.
> the packet loop.
> and now linux doesn't check that. In tx path,
> softnet_data.xmit.recursion will limit the recursion.
> I think we should check that in ingress. any thoughts?

Nope, since this is xmit, not receive. This goes through a rescheduling
point and could end up on a different CPU potentially. This is not any
different in than how other infra handles this case. Point here is that
it's not freezing the box, so admin can react to misconfig and then undo
it again.