Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754859AbaGCTAJ (ORCPT ); Thu, 3 Jul 2014 15:00:09 -0400 Received: from e23smtp07.au.ibm.com ([202.81.31.140]:50650 "EHLO e23smtp07.au.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751698AbaGCTAG (ORCPT ); Thu, 3 Jul 2014 15:00:06 -0400 Message-ID: <53B5A769.5030108@linux.vnet.ibm.com> Date: Fri, 04 Jul 2014 00:26:41 +0530 From: Raghavendra K T Organization: IBM User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:17.0) Gecko/20130625 Thunderbird/17.0.7 MIME-Version: 1.0 To: Linus Torvalds CC: Andrew Morton , Fengguang Wu , David Cohen , Al Viro , Damien Ramonda , Jan Kara , David Rientjes , Nishanth Aravamudan , linux-mm , Linux Kernel Mailing List Subject: Re: [PATCH] mm readahead: Fix sys_readahead breakage by reverting 2MB limit (bug 79111) References: <1404392547-11648-1-git-send-email-raghavendra.kt@linux.vnet.ibm.com> <53B59CB5.9060004@linux.vnet.ibm.com> <53B5A343.4090402@linux.vnet.ibm.com> In-Reply-To: Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit X-TM-AS-MML: disable X-Content-Scanned: Fidelis XPS MAILER x-cbid: 14070319-0260-0000-0000-000005487641 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 07/04/2014 12:23 AM, Linus Torvalds wrote: > On Thu, Jul 3, 2014 at 11:38 AM, Raghavendra K T > wrote: >> >> Okay, how about something like 256MB? I would be happy to send a patch >> for that change. > > I'd like to see some performance numbers. I know at least Fedora uses > "readahead()" in the startup scripts, do we have any performance > numbers for that? > > Also, I think 256MB is actually excessive. People still do have really > slow devices out there. USB-2 is still common, and drives that read at > 15MB/s are not unusual. Do we really want to do readahead() that can > take tens of seconds (and *will* take tens of seconds sycnhronously, > because the IO requests fill up). > > So I wouldn't go from 2 to 256. That seems like an excessive jump. I > was more thinking in the 4-8MB range. But even then, I think we should > always have technical reasons (ie preferably numbers) for the change, > not just randomly change it. Okay. I 'll take some time to do the analysis. I think we also should keep in mind of possible remote readahead that would cause unnecessary penalty. -- 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/