diff options
author | David Sterba <dsterba@suse.cz> | 2013-04-26 15:20:23 +0000 |
---|---|---|
committer | Josef Bacik <jbacik@fusionio.com> | 2013-05-06 15:55:25 -0400 |
commit | 55793c0d0381176e727389325d9a47f7f0b5387f (patch) | |
tree | 580c9b7f5562e314bf13a8cdb92b342bc17db75a /fs | |
parent | f7a52a40cabea38b99b5053bc4f7cf45f4997603 (diff) | |
download | op-kernel-dev-55793c0d0381176e727389325d9a47f7f0b5387f.zip op-kernel-dev-55793c0d0381176e727389325d9a47f7f0b5387f.tar.gz |
btrfs: read entire device info under lock
There's a theoretical possibility of reading stale (or even more
theoretically, freed) data from DEV_INFO ioctl when the device would
disappear between an early mutex unlock and data being copied from the
device structure.
Signed-off-by: David Sterba <dsterba@suse.cz>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
Diffstat (limited to 'fs')
-rw-r--r-- | fs/btrfs/ioctl.c | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 3aa3750..0de4a2f 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -2432,7 +2432,6 @@ static long btrfs_ioctl_dev_info(struct btrfs_root *root, void __user *arg) mutex_lock(&fs_devices->device_list_mutex); dev = btrfs_find_device(root->fs_info, di_args->devid, s_uuid, NULL); - mutex_unlock(&fs_devices->device_list_mutex); if (!dev) { ret = -ENODEV; @@ -2456,6 +2455,7 @@ static long btrfs_ioctl_dev_info(struct btrfs_root *root, void __user *arg) } out: + mutex_unlock(&fs_devices->device_list_mutex); if (ret == 0 && copy_to_user(arg, di_args, sizeof(*di_args))) ret = -EFAULT; |