From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from userp1040.oracle.com ([156.151.31.81]:50999 "EHLO userp1040.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753907AbaFID0n (ORCPT ); Sun, 8 Jun 2014 23:26:43 -0400 Message-ID: <53952A02.30705@oracle.com> Date: Mon, 09 Jun 2014 11:29:06 +0800 From: Anand Jain MIME-Version: 1.0 To: Qu Wenruo , linux-btrfs@vger.kernel.org CC: dsterba@suse.cz, wangsl.fnst@cn.fujitsu.com, clm@fb.com, jbacik@fb.com Subject: Re: [PATCH V3] Btrfs: device_list_add() should not update list when mounted References: <1402025201-17140-1-git-send-email-anand.jain@oracle.com> <53951371.9010608@cn.fujitsu.com> <539519BC.7060109@oracle.com> <53951D70.6040303@cn.fujitsu.com> In-Reply-To: <53951D70.6040303@cn.fujitsu.com> Content-Type: text/plain; charset=UTF-8; format=flowed Sender: linux-btrfs-owner@vger.kernel.org List-ID: >> As mentioned in the other thread.. we expect user to check the devices >> before / after mount and wipefs the disks which should not belong to >> the fsid. my bad. I just realized that unmount and wipefs may not be an viable choice in case of btrfs as root fs. > For the case you mentioned, I think the behavior is OK, > always use the device with *current* largest generation is a acceptable > strategy, since the only things we can depend on the devices we > *current* see... Yeah. let me do that approach as well to take care of picking the correct disk when FS is _unmounted_. And, when FS is mounted we should NOT kick out device paths invariably. which means we still need the patch. Btrfs: device_list_add() should not update list when mounted Thanks, Anand