From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <linux-kernel-owner@vger.kernel.org>
Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand
	id S933897AbaJDAgc (ORCPT <rfc822;w@1wt.eu>);
	Fri, 3 Oct 2014 20:36:32 -0400
Received: from mx1.redhat.com ([209.132.183.28]:8909 "EHLO mx1.redhat.com"
	rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP
	id S1755097AbaJDAga (ORCPT <rfc822;linux-kernel@vger.kernel.org>);
	Fri, 3 Oct 2014 20:36:30 -0400
Date: Sat, 4 Oct 2014 02:33:00 +0200
From: Oleg Nesterov <oleg@redhat.com>
To: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Sasha Levin <sasha.levin@oracle.com>,
        Frederic Weisbecker <fweisbec@gmail.com>,
        Ingo Molnar <mingo@kernel.org>, Peter Anvin <hpa@zytor.com>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        Peter Zijlstra <peterz@infradead.org>,
        Andy Lutomirski <luto@amacapital.net>,
        Denys Vlasenko <dvlasenk@redhat.com>,
        Thomas Gleixner <tglx@linutronix.de>,
        Chuck Ebbert <cebbert.lkml@gmail.com>
Subject: Re: [tip:x86/asm] x86: Speed up ___preempt_schedule*() by using
	THUNK helpers
Message-ID: <20141004003300.GA6297@redhat.com>
References: <20140921184153.GA23727@redhat.com> <tip-0ad6e3c5199be12c9745da8f8b9e3c9f8066c235@git.kernel.org> <542E2B05.5080607@oracle.com> <20141003232631.GA3439@redhat.com> <CA+55aFwT3KFkGMgWz9Gag+xMHD=5X3RNz-iMswGD88_wnUT-Eg@mail.gmail.com> <CA+55aFxBAFMxAsSvgOcVuWy4FWH77E4NT8M3SYvQzxGBJ_qkHw@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <CA+55aFxBAFMxAsSvgOcVuWy4FWH77E4NT8M3SYvQzxGBJ_qkHw@mail.gmail.com>
User-Agent: Mutt/1.5.18 (2008-05-17)
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

On 10/03, Linus Torvalds wrote:
>
> On Fri, Oct 3, 2014 at 5:01 PM, Linus Torvalds
> <torvalds@linux-foundation.org> wrote:
> >
> > The real fix would appear to be to use
> > "preempt_enable_no_resched_notrace()", which your patch did, but
> > without the loop.
>
> Actually, the real fix would be to not be stupid, and just make the
> code do something like
>
> >        if (likely(!preemptible()))
> >                return;
> >
> >         __preempt_count_add(PREEMPT_ACTIVE);
> >         prev_ctx = exception_enter();
> >
> >         __schedule();
> >
> >         exception_exit(prev_ctx);
> >         __preempt_count_sub(PREEMPT_ACTIVE);
>
> and *not* enable preemption around the scheduling at all. The whole
> enable and then re-disable seems entirely broken, and comes from the
> code using "preempt_schedule()" which doesn't work while preemption is
> disabled. So don't do that then.

Again, it is too late for me... Most probably I am wrong, but somehow
it seems to me that the real fix should try to kill preempt_schedule_context()
altogether and teach preempt_schedule() to play well with CONTEXT_TRACKING.

Oleg.