linux-kernel - Re: [PATCH 2/2] trace-cmd: Use tracecmd_peek_next_data() in fgraph_ent

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: <20160712083212.04b9622e@gandalf.local.home>
Date:	Tue, 12 Jul 2016 08:32:12 -0400
From:	Steven Rostedt <rostedt@...dmis.org>
To:	Namhyung Kim <namhyung@...nel.org>
Cc:	LKML <linux-kernel@...r.kernel.org>, Joonsoo Kim <js1304@...il.com>
Subject: Re: [PATCH 2/2] trace-cmd: Use tracecmd_peek_next_data() in
 fgraph_ent_handler

On Tue, 12 Jul 2016 09:17:25 +0900
Namhyung Kim <namhyung@...nel.org> wrote:

> > Hmm, but what happens if the next data is just some random event on
> > another CPU. Do we want to break it up just because there's data on
> > another cpu?  
> 
> Yes, I think we should break.  Isn't it natural to show an event in
> the middle of a function if it occurred before returning from the
> function?  It would be more acccurate output IMHO.  I guess most leaf
> functions are small so the end result would almost same.

OK, that sounds fine then.

> 
> 
> > 
> > I wonder if we should grab a record from the same cpu and if it isn't
> > the return, then try another cpu?  
> 
> But in this case, it's a problem even if it's the return of the same
> function.  The task can be migrated to another cpu during the
> function, and then can be migrated back to the original cpu while
> calling same function again.  The entry of the first invocation would
> match to the exit of the Nth invocation..

I'll keep your patch as is then.

Thanks,

-- Steve