Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1030189AbWLOWT4 (ORCPT ); Fri, 15 Dec 2006 17:19:56 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1030187AbWLOWT4 (ORCPT ); Fri, 15 Dec 2006 17:19:56 -0500 Received: from srv5.dvmed.net ([207.36.208.214]:58573 "EHLO mail.dvmed.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1030186AbWLOWTz (ORCPT ); Fri, 15 Dec 2006 17:19:55 -0500 Message-ID: <45831F80.5060008@garzik.org> Date: Fri, 15 Dec 2006 17:19:44 -0500 From: Jeff Garzik User-Agent: Thunderbird 1.5.0.8 (X11/20061107) MIME-Version: 1.0 To: Alan CC: Andrew Morton , Neil Brown , Jurriaan , linux-kernel@vger.kernel.org, linux-raid@vger.kernel.org, Tejun Heo Subject: Re: sata badness in 2.6.20-rc1? [Was: Re: md patches in -mm] References: <20061204203410.6152efec.akpm@osdl.org> <17780.63770.228659.234534@cse.unsw.edu.au> <20061205061623.GA13749@amd64.of.nowhere> <20061205062142.GA14784@amd64.of.nowhere> <20061204224323.2e5d0494.akpm@osdl.org> <20061205105928.GA6482@amd64.of.nowhere> <17782.28505.303064.964551@cse.unsw.edu.au> <20061215192146.GA3616@amd64.of.nowhere> <17795.2681.523120.656367@cse.unsw.edu.au> <20061215130552.95860b72.akpm@osdl.org> <20061215133927.a8346372.akpm@osdl.org> <20061215220618.06f1873c@localhost.localdomain> In-Reply-To: <20061215220618.06f1873c@localhost.localdomain> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Spam-Score: -4.3 (----) X-Spam-Report: SpamAssassin version 3.1.7 on srv5.dvmed.net summary: Content analysis details: (-4.3 points, 5.0 required) Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1156 Lines: 32 Alan wrote: > On Fri, 15 Dec 2006 13:39:27 -0800 > Andrew Morton wrote: > >> On Fri, 15 Dec 2006 13:05:52 -0800 >> Andrew Morton wrote: >> >>> Jeff, I shall send all the sata patches which I have at you one single time >>> and I shall then drop the lot. So please don't flub them. >>> >>> I'll then do a rc1-mm2 without them. >> hm, this is looking like a lot of work for not much gain. Rafael, are >> you able to do a quick chop and tell us whether these: > > The md one and the long history of reports about parallel I/O causing > problems sounds a lot more like the kmap stuff you were worried about > Andrew. I'd be very intereste dto know if it happens on x86_32 built with > a standard memory split and no highmem.... 2.6.20-rc1 works, and 2.6.20-rc1 does not have the kmap_atomic() fix. Upstream does kmap_atomic(KM_USER0) and -mm does kmap_atomic(KM_IRQ0) Jeff - 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/