From mboxrd@z Thu Jan  1 00:00:00 1970
From: Andrew Morton <akpm@linux-foundation.org>
Subject: Re: AIM7 40% regression with 2.6.26-rc1
Date: Wed, 7 May 2008 10:16:24 -0700
Message-ID: <20080507101624.5524d3c1.akpm@linux-foundation.org>
References: <1210052904.3453.30.camel@ymzhang>
	<20080506114449.GC32591@elte.hu>
	<20080506120934.GH19219@parisc-linux.org>
	<20080506162332.GI19219@parisc-linux.org>
	<20080506102153.5484c6ac.akpm@linux-foundation.org>
	<20080507163811.GY19219@parisc-linux.org>
	<alpine.LFD.1.10.0805070948250.3024@woody.linux-foundation.org>
	<alpine.LFD.1.10.0805070956060.3024@woody.linux-foundation.org>
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Cc: Matthew Wilcox <matthew@wil.cx>, Ingo Molnar <mingo@elte.hu>,
	"J. Bruce Fields" <bfields@citi.umich.edu>,
	"Zhang, Yanmin" <yanmin_zhang@linux.intel.com>,
	LKML <linux-kernel@vger.kernel.org>,
	Alexander Viro <viro@ftp.linux.org.uk>,
	linux-fsdevel@vger.kernel.org
To: Linus Torvalds <torvalds@linux-foundation.org>
Return-path: <linux-fsdevel-owner@vger.kernel.org>
Received: from smtp1.linux-foundation.org ([140.211.169.13]:52582 "EHLO
	smtp1.linux-foundation.org" rhost-flags-OK-OK-OK-OK)
	by vger.kernel.org with ESMTP id S1752483AbYEGRQl (ORCPT
	<rfc822;linux-fsdevel@vger.kernel.org>);
	Wed, 7 May 2008 13:16:41 -0400
In-Reply-To: <alpine.LFD.1.10.0805070956060.3024@woody.linux-foundation.org>
Sender: linux-fsdevel-owner@vger.kernel.org
List-ID: <linux-fsdevel.vger.kernel.org>

On Wed, 7 May 2008 10:08:18 -0700 (PDT) Linus Torvalds <torvalds@linux-foundation.org> wrote:

> Which is why I'm 100% convinced it's not even worth saving the old code. 
> It needs to use mutexes, or spinlocks. I bet it has *nothing* to do with 
> "slow path" other than the fact that it gets to that slow path much more 
> these days.

Stupid question: why doesn't lock_kernel() use a mutex?

(stupid answer: it'll trigger might_sleep() checks when we do it early in
boot with irqs disabled, but we can fix that)

(And __might_sleep()'s system_state check might even save us from that)

Of course, we shouldn't change anything until we've worked out why the new
semaphores got slower.