Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755369Ab0KXRqa (ORCPT ); Wed, 24 Nov 2010 12:46:30 -0500 Received: from mail-yw0-f46.google.com ([209.85.213.46]:47507 "EHLO mail-yw0-f46.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753051Ab0KXRq3 convert rfc822-to-8bit (ORCPT ); Wed, 24 Nov 2010 12:46:29 -0500 DomainKey-Signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:sender:in-reply-to:references:date :x-google-sender-auth:message-id:subject:from:to:cc:content-type :content-transfer-encoding; b=a+wfqbhG8+32Nm96vFf6BFGEnZBd3ZnsEBNpDicKyNtRFTMY73sunprNSzpgsBfnN9 +Ik3hgabPRFn1NE8/Pd8pK8qXzL6JOhap4yTN+kTsvez5Gx+hNmYHuBBlYd1A6RhpBnn Hf2x/qvCtOnVV0zAYSIcKd8SqG0ypwyc3n8dA= MIME-Version: 1.0 In-Reply-To: References: <20101122161158.02699d10.akpm@linux-foundation.org> <1290501502.2390.7029.camel@nimitz> <1290529171.2390.7994.camel@nimitz> Date: Wed, 24 Nov 2010 19:46:28 +0200 X-Google-Sender-Auth: 9294wgu7DDnWi3fyhwryRtLpQXg Message-ID: Subject: Re: Sudden and massive page cache eviction From: Pekka Enberg To: =?ISO-8859-1?Q?Peter_Sch=FCller?= Cc: Dave Hansen , Andrew Morton , linux-kernel@vger.kernel.org, Mattias de Zalenski , linux-mm@kvack.org Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8BIT Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1436 Lines: 37 Hi Peter, 2010/11/24 Peter Sch?ller : >>> I forgot to address the second part of this question: How would I best >>> inspect whether the kernel is doing that? >> >> You can, for example, record >> >> ?cat /proc/meminfo | grep Huge >> >> for large page allocations. > > Those show zero a per my other post. However I got the impression Dave > was asking about regular but larger-than-one-page allocations internal > to the kernel, while the Huge* lines in /proc/meminfo refers to > allocations specifically done by userland applications doing huge page > allocation on a system with huge pages enabled - or am I confused? He was asking about both (large page allocations and higher order allocations). >> The "pagesperslab" column of /proc/slabinfo tells you how many pages >> slab allocates from the page allocator. > > Seems to be what vmstat -m reports. No, "vmstat -m" reports _total number_ of pages allocated. We're interested in how many pages slab allocator whenever it needs to allocate memory for a new slab. That's represented by the "pagesperslab" column of /proc/slabinfo from which you can deduce the page allocation order. Pekka -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/