linux-kernel - Re: [RFC PATCH] tracing: Introduce pipe_cpumask to avoid race on trace

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: <20230817214337.48de3ab8@gandalf.local.home>
Date:   Thu, 17 Aug 2023 21:43:37 -0400
From:   Steven Rostedt <rostedt@...dmis.org>
To:     Zheng Yejian <zhengyejian1@...wei.com>
Cc:     <laijs@...fujitsu.com>, <linux-kernel@...r.kernel.org>,
        <linux-trace-kernel@...r.kernel.org>, <mhiramat@...nel.org>
Subject: Re: [RFC PATCH] tracing: Introduce pipe_cpumask to avoid race on
 trace_pipes

On Fri, 18 Aug 2023 09:38:27 +0800
Zheng Yejian <zhengyejian1@...wei.com> wrote:

> >> from being opened by multiple people at the same time.  
> > 
> > I'm fine with the CPU mask.  
> 
> Hi, Steve, if a task open a trace_pipe file, then concurrently read it
> with multiple threads, then the read race problem may also happen, this
> patch will not prevent this case.

I was thinking this too and started to look at a fix for that when I
realized, it shouldn't be a problem. The open creates the iterator and it's
attached to the file descriptor. All the threads will be using the same
file descriptor, which means they will be using the same iterator. That
iter->ent from one thread will be still valid for the next thread.

-- Steve


> 
> Do we need to consider this case? Or just tell user not to do like this
> through some documents?