From mboxrd@z Thu Jan 1 00:00:00 1970 From: Russell King Subject: Re: [BUG 2.6.17-git] kmem_cache_create: duplicate cache scsi_cmd_cache Date: Fri, 12 May 2006 21:37:50 +0100 Message-ID: <20060512203749.GB17120@flint.arm.linux.org.uk> References: <20060511151456.GD3755@harddisk-recovery.com> <15ddcffd0605112153q57f139a1k7068e204a3eeaf1f@mail.gmail.com> <20060512171632.GA29077@harddisk-recovery.com> <1147456038.3769.39.camel@mulgrave.il.steeleye.com> <1147460325.3769.46.camel@mulgrave.il.steeleye.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from caramon.arm.linux.org.uk ([212.18.232.186]:10509 "EHLO caramon.arm.linux.org.uk") by vger.kernel.org with ESMTP id S1750762AbWELUh4 (ORCPT ); Fri, 12 May 2006 16:37:56 -0400 Content-Disposition: inline In-Reply-To: <1147460325.3769.46.camel@mulgrave.il.steeleye.com> Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: James Bottomley Cc: Linus Torvalds , Erik Mouw , Or Gerlitz , linux-scsi@vger.kernel.org, axboe@suse.de, Linux Kernel Mailing List On Fri, May 12, 2006 at 01:58:45PM -0500, James Bottomley wrote: > I suggest simply reversing this patch at the moment. If Russell and > Jens can tell me what they're trying to do I'll see if there's another > way to do it. When a MMC card is pulled, we remove the MMC device structure (which is what the driverfs_dev points at.) At this point, the MMC layer *totally* forgets about the MMC device and deletes it. Unfortunately, an uncounted reference is kept while the partition is mounted by the gendisk layer, which when the partition is unmounted via hotplug causes another hotplug event to be generated with respect to this freed MMC device structure, and hence you get an oops. Since the MMC layer has lost all knowledge of the device, the only possible solution is as given in that patch. -- Russell King Linux kernel 2.6 ARM Linux - http://www.arm.linux.org.uk/ maintainer of: 2.6 Serial core