linux-kernel - Re: [PATCH] smp: generic ipi

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] [day] [month] [year] [list]

Message-ID: <767faf00-0dc0-698a-910f-100608d10d81@gmail.com>
Date:   Thu, 21 May 2020 21:11:36 +0100
From:   Wojciech Kudla <wk.kernel@...il.com>
To:     Nadav Amit <namit@...are.com>
Cc:     LKML <linux-kernel@...r.kernel.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Peter Zijlstra <peterz@...radead.org>
Subject: Re: [PATCH] smp: generic ipi_raise tracepoint

On 21/05/2020 20:00, Nadav Amit wrote:

>> -	if (llist_add(&csd->llist, &per_cpu(call_single_queue, cpu)))
>> +	if (llist_add(&csd->llist, &per_cpu(call_single_queue, cpu))) {
>> +		if (trace_ipi_raise_enabled())
> 
> Why do you need this check? trace_ipi_raise() will do the same check before
> actual tracing:
> 
> 	if (static_key_false(&__tracepoint_##name.key)
> 

Yes, my motivation for conditional logic was performance-driven.
Thanks for pointing out the implicit check.

> 
> In general, I think there are too many trace-points. They look benign(i.e.,
> free), but can cause worse code to be generated as they behave as a memory
> clobber. Many times the same result can be achieved with a probe.
> 

Thank you for the review, I agree this may not be optimal. Let's just stop here.
There's a different patch I submitted today that follows Peter's suggestions
about smp function calls being much more sensible target for new tracepoints.

http://lkml.kernel.org/r/e6141d56-1da1-6c00-198f-cbe4385327ff@gmail.com

Thanks,
W.