btrfs-progs: do not add stale device into fs_devices
authorLiu Bo <bo.li.liu@oracle.com>
Wed, 11 Oct 2017 17:57:16 +0000 (11:57 -0600)
committerDavid Sterba <dsterba@suse.com>
Tue, 14 Nov 2017 14:59:00 +0000 (15:59 +0100)
If one of btrfs' devices was pulled out and we've replaced it with a
new one, then they have the same uuid.

If that device gets reconnected, 'btrfs filesystem show' will show the
stale one instead of the new one, but on the kernel side btrfs has a fix
not to include the stale one, this could confuse users as people may
monitor btrfs by running that command.

This does the similar thing to what kernel side has done.

Signed-off-by: Liu Bo <bo.li.liu@oracle.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
[ format string adjustments ]
Signed-off-by: David Sterba <dsterba@suse.com>
volumes.c

index 2209e5a..2fde476 100644 (file)
--- a/volumes.c
+++ b/volumes.c
@@ -138,7 +138,21 @@ static int device_list_add(const char *path,
                list_add(&device->dev_list, &fs_devices->devices);
                device->fs_devices = fs_devices;
        } else if (!device->name || strcmp(device->name, path)) {
-               char *name = strdup(path);
+               char *name;
+
+               /*
+                * The existing device has newer generation, so this one could
+                * be a stale one, don't add it.
+                */
+               if (found_transid < device->generation) {
+                       warning(
+       "adding device %s gen %llu but found an existing device %s gen %llu",
+                               path, found_transid, device->name,
+                               device->generation);
+                       return -EEXIST;
+               }
+
+               name = strdup(path);
                 if (!name)
                         return -ENOMEM;
                 kfree(device->name);