From owner-freebsd-stable@FreeBSD.ORG Fri Aug 20 20:50:42 2004 Return-Path: Delivered-To: freebsd-stable@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id B3B0316A4CE for ; Fri, 20 Aug 2004 20:50:42 +0000 (GMT) Received: from mail021.syd.optusnet.com.au (mail021.syd.optusnet.com.au [211.29.132.132]) by mx1.FreeBSD.org (Postfix) with ESMTP id 9945143D46 for ; Fri, 20 Aug 2004 20:50:41 +0000 (GMT) (envelope-from PeterJeremy@optushome.com.au) Received: from cirb503493.alcatel.com.au (c211-30-75-229.belrs2.nsw.optusnet.com.au [211.30.75.229]) i7KKodU06639 for ; Sat, 21 Aug 2004 06:50:40 +1000 Received: from cirb503493.alcatel.com.au (localhost.alcatel.com.au [127.0.0.1])i7KKodxP040089 for ; Sat, 21 Aug 2004 06:50:39 +1000 (EST) (envelope-from pjeremy@cirb503493.alcatel.com.au) Received: (from pjeremy@localhost) by cirb503493.alcatel.com.au (8.12.10/8.12.9/Submit) id i7KKodwm040088 for freebsd-stable@freebsd.org; Sat, 21 Aug 2004 06:50:39 +1000 (EST) (envelope-from pjeremy) Date: Sat, 21 Aug 2004 06:50:39 +1000 From: Peter Jeremy To: freebsd-stable@freebsd.org Message-ID: <20040820205038.GN423@cirb503493.alcatel.com.au> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Disposition: inline Content-Transfer-Encoding: 8bit User-Agent: Mutt/1.4.2i Subject: Excessive fdisk(8) delays X-BeenThere: freebsd-stable@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Production branch of FreeBSD source code List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 20 Aug 2004 20:50:42 -0000 I haven't noticed this before and don't recall it being mentioned before... I just did an 'fdisk ad1' whilst ad0 was being hammered and the fdisk spat out all the expected information then hung (in "atalck") for about 4½ minutes before exiting. This was repeatable and very disconcerting the first time. Background: The system is an AMD Athlon-XP running -STABLE from the beginning of August with two 100GB WD drives running at UDMA100 attached to channel 0 of a VIA VT8233A. The "hammering" was a dd of the raw disk, a "find /" and a "cvs update" (systat -v report 100% utilisation of ad0). A quick check of the code shows "atalck" is used within ATA_SLEEPLOCK_CH to wait for the channel to be come idle and this macro is used fairly extensively within the ata code. Having I/O requests queued for over 4 minutes seems excessive - how difficult would it be for the ata code to better round-robin requests? (Possibly via a wakeup at the end of ata_start()). -- Peter Jeremy