netdev - Re: [PATCH v2 net] tcp: allow again tcp_disconnect() when threads are waiting

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: Fri, 13 Oct 2023 11:25:11 +0200
From: Paolo Abeni <pabeni@...hat.com>
To: Xin Guo <guoxin0309@...il.com>, Eric Dumazet <edumazet@...gle.com>
Cc: netdev@...r.kernel.org, Ayush Sawal <ayush.sawal@...lsio.com>, "David S.
 Miller" <davem@...emloft.net>, Jakub Kicinski <kuba@...nel.org>, David
 Ahern <dsahern@...nel.org>,  mptcp@...ts.linux.dev, Boris Pismenny
 <borisp@...dia.com>, Tom Deseyn <tdeseyn@...hat.com>
Subject: Re: [PATCH v2 net] tcp: allow again tcp_disconnect() when threads
 are waiting

Hi,

On Fri, 2023-10-13 at 12:42 +0800, Xin Guo wrote:
> In my view, this patch is NOT so good, and it seems that trying to fix
> a problem temporarily without knowing its root cause,

First thing first, please avoid top posting when replying to the ML.

I don't follow the above statement. The root case of the problem
addressed here is stated in the commit message: the blamed commit
explicitly disables a functionality used by the user-space. We must
avoid breaking the user-space.

> because sk_wait_event function should know nothing about the other
> functions were called or not,
> but now this patch added a logic to let sk_wait_event know the
> specific tcp_dissconnect function was called by other threads or NOT,
> honestly speaking, it is NOT a good designation,

Why?

> so what is root cause about the problem which [0] commit want to fix?

The mentioned commit changelog is quite descriptive about the problem,
please read it.

> can we have a way to fix it directly instead of denying
> tcp_disconnect() when threads are waiting?

Yes, this patch.

Cheers,

Paolo