From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S933195Ab2GBWqw (ORCPT ); Mon, 2 Jul 2012 18:46:52 -0400 Received: from mga01.intel.com ([192.55.52.88]:59276 "EHLO mga01.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932456Ab2GBWqt (ORCPT ); Mon, 2 Jul 2012 18:46:49 -0400 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="4.71,315,1320652800"; d="scan'208";a="187699399" Date: Mon, 2 Jul 2012 15:46:48 -0700 From: Andi Kleen To: Jiri Kosina Cc: linux-kernel@vger.kernel.org, torvalds@linux-foundation.org, shemminger@vyatta.com Subject: long boot delays caused by 070ad7e7 floppy change Message-ID: <20120702224648.GA20361@tassilo.jf.intel.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.5.21 (2010-09-15) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Jiri, One of my test systems has a new 20s+ boot delay in 3.5rc* It takes the floppy driver that long to determine I don't have a floppy [ 2.386322] calling floppy_init+0x0/0xe46 @ 1 [ 2.448977] Refined TSC clocksource calibration: 2594.106 MHz. [ 2.449021] Switching to clocksource tsc [ 22.371560] floppy0: no floppy controllers found [ 22.371676] initcall floppy_init+0x0/0xe46 returned -19 after 19553621 usecs In older kernels the delay is only about 3s. I tracked it down to commit 070ad7e793dc6ff753ee682ef7790b3373b471f6 Author: Jiri Kosina Date: Fri May 18 13:50:25 2012 +0200 floppy: convert to delayed work and single-thread wq The problem is not completely stable, a bisect actually doesn't find it and reverting the patch on current HEAD also doesn't help. However when setting the tree manually to one commit before yours the problem is gone and adding your patch adds it. I think the floppy init should run asynchronous, but it would be still good to make this faster again. Since block devices need to be synchronized at end of boot there is no way to hide a 20s delay completely. -Andi -- ak@linux.intel.com -- Speaking for myself only