From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from aserp1040.oracle.com ([141.146.126.69]:34667 "EHLO aserp1040.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753479AbbE3Pcq (ORCPT ); Sat, 30 May 2015 11:32:46 -0400 From: Anand Jain To: linux-btrfs@vger.kernel.org Cc: dsterba@suse.cz Subject: [PATCH v5.2 24/42] Btrfs: free the stale device Date: Sat, 30 May 2015 23:32:25 +0800 Message-Id: <1432999945-4320-1-git-send-email-Anand.Jain@oracle.com> In-Reply-To: <1426845702-6298-25-git-send-email-anand.jain@oracle.com> References: <1426845702-6298-25-git-send-email-anand.jain@oracle.com> Sender: linux-btrfs-owner@vger.kernel.org List-ID: When the btrfs on a device is overwritten with a new btrfs (mkfs), the old btrfs instance in the kernel becomes stale. So with this patch if kernel finds device is overwritten then delete the stale fsid/uuid. Signed-off-by: Anand Jain --- v5.1->v5.2: accepts David's review comments with thanks. esp.ly the important rcu part v5->v5.1: since this deals with only devices in unmounted state, don't try to remove device link fs/btrfs/volumes.c | 62 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 62 insertions(+) diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index e50005f..0a4df7a 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -445,6 +445,62 @@ static void pending_bios_fn(struct btrfs_work *work) run_scheduled_bios(device); } + +void btrfs_free_stale_device(struct btrfs_device *cur_dev) +{ + struct btrfs_fs_devices *fs_devs; + struct btrfs_device *dev; + + if (!cur_dev->name) + return; + + list_for_each_entry(fs_devs, &fs_uuids, list) { + int del = 1; + + if (fs_devs->opened) + continue; + if (fs_devs->seeding) + continue; + + list_for_each_entry(dev, &fs_devs->devices, dev_list) { + + if (dev == cur_dev) + continue; + if (!dev->name) + continue; + + /* + * Todo: This won't be enough. What if the same device + * comes back (with new uuid and) with its mapper path? + * But for now, this does help as mostly an admin will + * either use mapper or non mapper path throughout. + */ + rcu_read_lock(); + del = strcmp(rcu_str_deref(dev->name), + rcu_str_deref(cur_dev->name)); + rcu_read_unlock(); + if (!del) + break; + } + + if (!del) { + /* delete the stale device */ + if (fs_devs->num_devices == 1) { + btrfs_sysfs_remove_fsid(fs_devs); + list_del(&fs_devs->list); + free_fs_devices(fs_devs); + } else { + fs_devs->num_devices--; + list_del(&dev->dev_list); + rcu_string_free(dev->name); + kfree(dev); + } + break; + } + } + return; +} + /* * Add new device to list of registered devices * @@ -560,6 +616,12 @@ static noinline int device_list_add(const char *path, if (!fs_devices->opened) device->generation = found_transid; + /* + * if there is new btrfs on an already registered device, + * then remove the stale device entry. + */ + btrfs_free_stale_device(device); + *fs_devices_ret = fs_devices; return ret; -- 2.4.1