Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752039AbcDWCR3 (ORCPT ); Fri, 22 Apr 2016 22:17:29 -0400 Received: from mail-pf0-f194.google.com ([209.85.192.194]:33625 "EHLO mail-pf0-f194.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751718AbcDWCR1 (ORCPT ); Fri, 22 Apr 2016 22:17:27 -0400 From: Ratna Manoj Subject: Re: [PATCH] NBD: replace kill_bdev() with __invalidate_device() To: Markus Pargmann References: <56E711D1.9090708@gmail.com> <2358965.KUbuAa1sts@adelgunde> <56F34412.1050707@gmail.com> <11986900.ymBh9azTyz@adelgunde> Cc: pbonzini@redhat.com, jack@suse.cz, Gou Rao , Vinod Jayaraman , nbd-general@lists.sourceforge.net, linux-kernel@vger.kernel.org Message-ID: <571ADB31.2000609@gmail.com> Date: Sat, 23 Apr 2016 07:47:21 +0530 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101 Thunderbird/38.6.0 MIME-Version: 1.0 In-Reply-To: <11986900.ymBh9azTyz@adelgunde> Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 5025 Lines: 144 Thanks for the review. Atleast for ext4 this crash happens on a sys_umount() call, timing of which is not in control of block driver. Block driver cannot force the filesystems to be unmounted, and the file system does not expect buffers to get unmapped under it. Ext4 can be fixed with the this patch: http://www.spinics.net/lists/linux-ext4/msg51112.html It did not make to the kernel. It checks the state of the buffer head before committing. When we consider diskett/CD as user space thread that called NBD_DO_IT, this problem is analogous to changing disk with another or the same disk suddenly when the file system is still mounted. If we completely kill the block device we would loss some writes when same thread is reconnected. if we do not completely kill or if we only invalidate clean buffers, we will have inconsistency on re-attach with a different thread (analogous to replacing disk with different disk suddenly). Ratna. On Wed, Apr 20, 2016 at 4:36 PM, Markus Pargmann wrote: > Hi, > > On Thursday 24 March 2016 07:04:10 Ratna Manoj wrote: > > From: Ratna Manoj Bolla > > > > When a filesystem is mounted on a nbd device and on a disconnect, because > > of kill_bdev(), and resetting bdev size to zero, buffer_head mappings are > > getting destroyed under mounted filesystem. > > > > After a bdev size reset(i.e bdev->bd_inode->i_size = 0) on a disconnect, > > followed by a sys_umount(), > > generic_shutdown_super()->... > > ->__sync_blockdev()->... > > -> blkdev_writepages()->... > > ->do_invalidatepage()->... > > -> discard_buffer() is discarding superblock buffer_head > assumed > > to be in mapped state by ext4_commit_super(). > > > > > > > > Signed-off-by: Ratna Manoj Bolla > > --- > > This script reproduces both the kernel panic scenarios: > > > > $ qemu-img create -f qcow2 f.img 1G > > $ mkfs.ext4 f.img > > $ qemu-nbd -c /dev/nbd0 f.img > > $ mount /dev/nbd0 dir > > $ killall -KILL qemu-nbd > > $ sleep 1 > > $ ls dir > > $ umount dir > > > > Bug reports: > > http://www.kernelhub.org/?p=2&msg=361407 > > > https://www.mail-archive.com/nbd-general@lists.sourceforge.net/msg02388.html > > Thanks, please CC nbd-general@lists.sourceforge.net, > linux-kernel@vger.kernel.org as well. > > So this patch simply does not cleanup the blockdevice to avoid any > errors on the filesystem side. The userspace thread that called > NBD_DO_IT will exit immediately before the filesystem decided to release > the blockdevice. The nbd driver assumes that the shutdown was done and > accepts new clients setting up sockets and so on. Couldn't this lead to > a lot of problems? > > Currently NBD_DO_IT returns when it is save to use the NBD device again. > This patch changes this as the blockdevice may still be in use when > NBD_DO_IT returns. I think it would be better to delay NBD_DO_IT until > everything is cleaned up and all filesystems are closed. > > Best Regards, > > Markus > > > > > diff --git a/drivers/block/nbd.c b/drivers/block/nbd.c > > index f6b51d7..6e77b3a 100644 > > --- a/drivers/block/nbd.c > > +++ b/drivers/block/nbd.c > > @@ -119,7 +119,8 @@ static const char *nbdcmd_to_ascii(int cmd) > > > > static int nbd_size_clear(struct nbd_device *nbd, struct block_device > *bdev) > > { > > - bdev->bd_inode->i_size = 0; > > + if (bdev->bd_openers <= 1) > > + bdev->bd_inode->i_size = 0; > > set_capacity(nbd->disk, 0); > > kobject_uevent(&nbd_to_dev(nbd)->kobj, KOBJ_CHANGE); > > > > @@ -678,6 +679,9 @@ static void nbd_reset(struct nbd_device *nbd) > > > > static void nbd_bdev_reset(struct block_device *bdev) > > { > > + if (bdev->bd_openers > 1) > > + return; > > + > > set_device_ro(bdev, false); > > bdev->bd_inode->i_size = 0; > > if (max_part > 0) { > > @@ -735,7 +739,7 @@ static int __nbd_ioctl(struct block_device *bdev, > struct nbd_device *nbd, > > nbd_clear_que(nbd); > > BUG_ON(!list_empty(&nbd->queue_head)); > > BUG_ON(!list_empty(&nbd->waiting_queue)); > > - kill_bdev(bdev); > > + __invalidate_device(bdev, true); > > return 0; > > > > case NBD_SET_SOCK: { > > @@ -809,7 +813,7 @@ static int __nbd_ioctl(struct block_device *bdev, > struct nbd_device *nbd, > > > > sock_shutdown(nbd); > > nbd_clear_que(nbd); > > - kill_bdev(bdev); > > + __invalidate_device(bdev, true); > > nbd_bdev_reset(bdev); > > > > if (nbd->disconnect) /* user requested, ignore socket > errors */ > > > > > > > > -- > Pengutronix e.K. | | > Industrial Linux Solutions | http://www.pengutronix.de/ | > Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0 | > Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917-5555 | >