linux-kernel - Re: [PATCH 2/2] tracing: Disable events in reverse order of their enable operation

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, 28 Nov 2023 16:06:03 +0100
From:   Petr Pavlu <petr.pavlu@...e.com>
To:     Steven Rostedt <rostedt@...dmis.org>
Cc:     mhiramat@...nel.org, mathieu.desnoyers@...icios.com,
        zhengyejian1@...wei.com, linux-trace-kernel@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH 2/2] tracing: Disable events in reverse order of their
 enable operation

On 11/27/23 18:58, Steven Rostedt wrote:
> On Mon, 27 Nov 2023 16:12:48 +0100
> Petr Pavlu <petr.pavlu@...e.com> wrote:
> 
>> Make the disable operation in __ftrace_event_enable_disable() use the
>> reverse order of the respective enable operation.
>>
>> This has two minor benefits:
>> * Disabling of buffered events via trace_buffered_event_disable() is
>>   done after unregistering the trace event. It closes a small window
>>   where an event would be still registered and could be hit, but would
>>   unnecessarily go directly to a ring buffer.
> 
> There's little benefit to the above. Going to the ring buffer and reverting
> it is just a bit more expensive, but should not be an issue with this small
> window.
> 
>> * The SOFT_DISABLED flag is now consistently set only when SOFT_MODE is
>>   also set.
> 
> This code is a bit fragile, and I rather not change the logic. There's a
> lot of corner cases.
> 
> I'm not saying that this is a bad change, I just don't want to add it and
> find out later it broke one of the corner cases. To add this would require
> an analysis that every input produces the same output with and without this
> change.
> 
> If you want to make a table showing all inputs between soft_disable and the
> flags, and show that the result produces the same updates, I'll then
> reconsidered applying this.

Ok, that is fair. It looked to me as a reasonable change but I don't
feel strongly about it and I understand your concern. I guess I'll drop
it in v2.

Thanks,
Petr