netdev - Re: [PATCH v1] tcp: fix connection reset due to tw hashdance race.

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]

Date: Tue, 6 Jun 2023 10:57:50 +0200
From: Simon Horman <simon.horman@...igine.com>
To: Duan Muquan <duanmuquan@...du.com>
Cc: davem@...emloft.net, dsahern@...nel.org, edumazet@...gle.com,
	kuba@...nel.org, pabeni@...hat.com, netdev@...r.kernel.org,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH v1] tcp: fix connection reset due to tw hashdance race.

On Mon, Jun 05, 2023 at 11:51:40AM +0800, Duan Muquan wrote:

...

> Brief of the scenario:
> 
> 1. Server runs on CPU 0 and Client runs on CPU 1. Server closes
> connection actively and sends a FIN to client. The lookback's driver
> enqueues the FIN segment to backlog queue of CPU 0 via
> loopback_xmit()->netif_rx(), one of the conditions for non-delay ack
> meets in __tcp_ack_snd_check(), and the ACK is sent immediately.
> 
> 2. On loopback interface, the ACK is received and processed on CPU 0,
> the 'dance' from original sock to tw sock will perfrom, tw sock will

Hi Duan Muquan,

a minor nit from my side: perfrom -> perform

> be inserted to ehash table, then the original sock will be removed.

...