From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from userp2130.oracle.com ([156.151.31.86]:53594 "EHLO userp2130.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753979AbeEaHdA (ORCPT ); Thu, 31 May 2018 03:33:00 -0400 Received: from pps.filterd (userp2130.oracle.com [127.0.0.1]) by userp2130.oracle.com (8.16.0.22/8.16.0.22) with SMTP id w4V7W2Qh031678 for ; Thu, 31 May 2018 07:32:59 GMT Received: from userv0021.oracle.com (userv0021.oracle.com [156.151.31.71]) by userp2130.oracle.com with ESMTP id 2j9x4hamqn-1 (version=TLSv1.2 cipher=ECDHE-RSA-AES256-GCM-SHA384 bits=256 verify=OK) for ; Thu, 31 May 2018 07:32:59 +0000 Received: from userv0122.oracle.com (userv0122.oracle.com [156.151.31.75]) by userv0021.oracle.com (8.14.4/8.14.4) with ESMTP id w4V7WwmY019183 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-GCM-SHA384 bits=256 verify=OK) for ; Thu, 31 May 2018 07:32:59 GMT Received: from abhmp0016.oracle.com (abhmp0016.oracle.com [141.146.116.22]) by userv0122.oracle.com (8.14.4/8.14.4) with ESMTP id w4V7WwP6017390 for ; Thu, 31 May 2018 07:32:58 GMT From: Anand Jain To: linux-btrfs@vger.kernel.org Subject: [PATCH 2/2] btrfs: fix missing superblock update in the device delete commit transaction Date: Thu, 31 May 2018 15:35:29 +0800 Message-Id: <20180531073529.15309-2-anand.jain@oracle.com> In-Reply-To: <20180531073529.15309-1-anand.jain@oracle.com> References: <20180531073529.15309-1-anand.jain@oracle.com> Sender: linux-btrfs-owner@vger.kernel.org List-ID: When a device is deleted the btrfs_super_block::number_devices is reduced by 1, but since we are do that after the last commit transaction for the device delete transaction, the update to btrfs_super_block::number_devices waits for the next commit/fsync transaction to make it to the disk. This can be easily demonstrated using the following test case where I use the btrfs device ready cli to read the and report, as it reads the device directly (which is unnecessary for the mounted device) and checks for the fs_devices::total_devices. mkfs.btrfs -fq -dsingle -msingle $dev1 $dev2 mount $dev1 /btrfs btrfs dev del $dev2 /btrfs btrfs dev ready $dev1; echo RESULT=$? <-- 1 Without this patch RESULT returns 1, indicating not ready! Same thing using seed device: mkfs.btrfs -fq $dev1 btrfstune -S1 $dev1 mount $dev1 /btrfs btrfs dev add -f $dev2 /btrfs umount /btrfs mount $dev2 /btrfs btrfs dev del $dev1 /btrfs btrfs dev ready $dev2; echo RESULT=$? Signed-off-by: Anand Jain --- The seed device test case needs to fix the bug as in the patch, [PATCH 1/2] btrfs: fix parent in memory total_devices after seed delete As there is a bug in the btrfs_device_rm() code, fs/btrfs/volumes.c | 38 ++++++++++++++++++-------------------- 1 file changed, 18 insertions(+), 20 deletions(-) diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index 5296f26be2f6..4138f17d4e58 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -1843,46 +1843,32 @@ static void update_dev_time(const char *path_name) } static int btrfs_rm_dev_item(struct btrfs_fs_info *fs_info, - struct btrfs_device *device) + struct btrfs_device *device, + struct btrfs_trans_handle *trans) { struct btrfs_root *root = fs_info->chunk_root; int ret; struct btrfs_path *path; struct btrfs_key key; - struct btrfs_trans_handle *trans; path = btrfs_alloc_path(); if (!path) return -ENOMEM; - trans = btrfs_start_transaction(root, 0); - if (IS_ERR(trans)) { - btrfs_free_path(path); - return PTR_ERR(trans); - } key.objectid = BTRFS_DEV_ITEMS_OBJECTID; key.type = BTRFS_DEV_ITEM_KEY; key.offset = device->devid; ret = btrfs_search_slot(trans, root, &key, path, -1, 1); - if (ret) { - if (ret > 0) - ret = -ENOENT; - btrfs_abort_transaction(trans, ret); - btrfs_end_transaction(trans); + if (ret > 0) { + ret = -ENOENT; goto out; } ret = btrfs_del_item(trans, root, path); - if (ret) { - btrfs_abort_transaction(trans, ret); - btrfs_end_transaction(trans); - } out: btrfs_free_path(path); - if (!ret) - ret = btrfs_commit_transaction(trans); return ret; } @@ -1966,7 +1952,9 @@ int btrfs_rm_device(struct btrfs_fs_info *fs_info, const char *device_path, u64 devid) { struct btrfs_device *device; + struct btrfs_trans_handle *trans; struct btrfs_fs_devices *cur_devices; + struct btrfs_root *root = fs_info->dev_root; struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; u64 num_devices; int ret = 0; @@ -2014,14 +2002,23 @@ int btrfs_rm_device(struct btrfs_fs_info *fs_info, const char *device_path, if (ret) goto error_undo; + trans = btrfs_start_transaction(root, 0); + if (IS_ERR(trans)) { + ret = PTR_ERR(trans); + goto error_undo; + } + /* * TODO: the superblock still includes this device in its num_devices * counter although write_all_supers() is not locked out. This * could give a filesystem state which requires a degraded mount. */ - ret = btrfs_rm_dev_item(fs_info, device); - if (ret) + ret = btrfs_rm_dev_item(fs_info, device, trans); + if (ret) { + btrfs_abort_transaction(trans, ret); + btrfs_end_transaction(trans); goto error_undo; + } clear_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &device->dev_state); btrfs_scrub_cancel_dev(fs_info, device); @@ -2090,6 +2087,7 @@ int btrfs_rm_device(struct btrfs_fs_info *fs_info, const char *device_path, free_fs_devices(cur_devices); } + ret = btrfs_commit_transaction(trans); out: mutex_unlock(&uuid_mutex); return ret; -- 2.7.0