Btrfs: check total number of devices when removing missing
authorJosef Bacik <josef@redhat.com>
Wed, 27 Jan 2010 02:09:38 +0000 (02:09 +0000)
committerChris Mason <chris.mason@oracle.com>
Thu, 28 Jan 2010 21:20:39 +0000 (16:20 -0500)
If you have a disk failure in RAID1 and then add a new disk to the
array, and then try to remove the missing volume, it will fail.  The
reason is the sanity check only looks at the total number of rw devices,
which is just 2 because we have 2 good disks and 1 bad one.  Instead
check the total number of devices in the array to make sure we can
actually remove the device.  Tested this with a failed disk setup and
with this test we can now run

btrfs-vol -r missing /mount/point

and it works fine.

Signed-off-by: Josef Bacik <josef@redhat.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
fs/btrfs/volumes.c

index 5eb7459..41ecbb2 100644 (file)
@@ -1135,7 +1135,7 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path)
                root->fs_info->avail_metadata_alloc_bits;
 
        if ((all_avail & BTRFS_BLOCK_GROUP_RAID10) &&
-           root->fs_info->fs_devices->rw_devices <= 4) {
+           root->fs_info->fs_devices->num_devices <= 4) {
                printk(KERN_ERR "btrfs: unable to go below four devices "
                       "on raid10\n");
                ret = -EINVAL;
@@ -1143,7 +1143,7 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path)
        }
 
        if ((all_avail & BTRFS_BLOCK_GROUP_RAID1) &&
-           root->fs_info->fs_devices->rw_devices <= 2) {
+           root->fs_info->fs_devices->num_devices <= 2) {
                printk(KERN_ERR "btrfs: unable to go below two "
                       "devices on raid1\n");
                ret = -EINVAL;