linux-kernel - Re: Reorganize perf kernel side

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: <20151206095406.GB4022@gmail.com>
Date:	Sun, 6 Dec 2015 10:54:06 +0100
From:	Ingo Molnar <mingo@...nel.org>
To:	Borislav Petkov <bp@...en8.de>
Cc:	Peter Zijlstra <peterz@...radead.org>,
	lkml <linux-kernel@...r.kernel.org>,
	Thomas Gleixner <tglx@...utronix.de>,
	Arnaldo Carvalho de Melo <acme@...radead.org>,
	Jiri Olsa <jolsa@...hat.com>,
	Namhyung Kim <namhyung@...nel.org>
Subject: Re: Reorganize perf kernel side

* Borislav Petkov <bp@...en8.de> wrote:

> Hi guys,
> 
> so I've had my eyes on this for a long time now and it has managed to
> get on my nerves just enough to do something about it :-)
> 
> So how about moving perf stuff to arch/x86/perf/ and get rid of the
> prefixes in the filenames. This also flattens our folder structure which
> is a good thing and which we've been talking about in the past.
> 
> In order to diminish churn, I can do the whole thing in 4-5 patches'
> sets, after having run enough *config smoke tests and 0day bot too.
> Anyway, something like that.
> 
> perf_event_<vendor>_<type>.c
> 
> can then move to arch/x86/perf/<vendor>/type.c
> 
> and have much saner structure.
> 
> Thoughts?

Yeah, it would be lovely if you could do that - but could we please name it 
'events' instead of 'perf', to follow the existing namespace pattern we are using 
for the core bits, where we have kernel/events/ for the core bits, not 
kernel/perf/?

Also, how about naming the core x86 bits like this:

  arch/x86/events/core.c

which would give us a clear path to split-out core functionality eventually, such 
as:

  arch/x86/events/sched.c
  arch/x86/events/constraints.c
  ...etc...
  ...

Just like we've already split out functionality from kernel/events/core.c into 
kernel/events/ring_buffer.c.

Btw., kernel/sched/ is using a similar approach, there's core.c, and various 
split-out sub-modules. We are slowly migrating from the original humunguous 
kernel/sched.c to a more finegrained kernel/sched/subsys.c structure.

... and as you suggested, the x86 vendor dependent bits would be in their own, but 
easily accessible directory close to the core, as you suggested:

  arch/x86/events/intel/
  arch/x86/events/amd/
  ...

as a lot of work is happening in that space, so promoting it up in the namespace 
helps.

So, as an example, we'd have renames like this:

     arch/x86/kernel/cpu/perf_event_intel_uncore_nhmex.c
  => arch/x86/events/intel/uncore_nhmex.c

which will be 30% easier to type! Once our muscle memory has re-trained that is. ;-)

Thanks,

	Ingo
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/