From mboxrd@z Thu Jan 1 00:00:00 1970 From: Alan Cox Subject: Re: [GIT PULL] Core block IO bits for 2.6.39 Date: Sun, 27 Mar 2011 14:21:35 +0100 Message-ID: <20110327142135.376142dd@lxorguk.ukuu.org.uk> References: <4D8B4A89.80608@fusionio.com> <4D8D87B0.3080107@fusionio.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-14 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: Received: from earthlight.etchedpixels.co.uk ([81.2.110.250]:49523 "EHLO www.etchedpixels.co.uk" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1753324Ab1C0NVw convert rfc822-to-8bit (ORCPT ); Sun, 27 Mar 2011 09:21:52 -0400 In-Reply-To: Sender: linux-m68k-owner@vger.kernel.org List-Id: linux-m68k@vger.kernel.org To: Linus Torvalds Cc: Jens Axboe , Geert Uytterhoeven , "linux-kernel@vger.kernel.org" , Chris Mason , Linux/m68k On Sat, 26 Mar 2011 09:48:37 -0700 Linus Torvalds wrote: > On Fri, Mar 25, 2011 at 11:29 PM, Jens Axboe wr= ote: > > + > > + =A0 =A0 =A0 /* Use 3ms as that was the old plug delay */ > > + =A0 =A0 =A0 blk_delay_queue(q, msecs_to_jiffies(3)); >=20 > That's bogus. blk_delay_queue() already takes msecs, not jiffies. >=20 > Also, do we really need to delay every unplug like this? It seems sad= =2E > A 3ms delay is a long time these days - admittedly most people > hopefully use ATA these days if they have an SSD or something, but > still.. The ATA code contains several other bogus delays btw. So out of the box unless you are using AHCI you are continually running small pointless delays every command issued.