This is a simple patch to allow reflinks to be made crossing subvolume boundaries. The only complication I found in implementing this is that btrfs_ioctl_clone reuses a btrfs_path. This is a slight problem because once we allow the source and destination inodes to come from different btrfs_roots, we need to know what root to pass into the various calls to btrfs_release_path, and keeping track of this would further complicate the control flow of btrfs_ioctl_clone. Fortunately btrfs_release_path does not use it's first argument, patch number 1 removes it. Patch number 2 is then just a simple matter of distinguishing between the source and destination btrfs_roots in btrfs_ioctl_clone. If removing the first argument from btrfs_release_path is undesirable, then btrfs_ioctl_clone could simply pass NULL. There are a few other places in btrfs that already do this. I have tested this by cp -a --reflink=always all the files found in the UML slackware root_fs, and also by cloning a test file with hundreds of random extents and holes. Larry D'Anna (2): btrfs: remove unused argument 'root' from btrfs_release_path btrfs: allow cross-subvolume BTRFS_IOC_CLONE fs/btrfs/ctree.c | 28 ++++++------ fs/btrfs/ctree.h | 2 +- fs/btrfs/dir-item.c | 2 +- fs/btrfs/extent-tree.c | 74 ++++++++++++++++---------------- fs/btrfs/file-item.c | 12 +++--- fs/btrfs/file.c | 12 +++--- fs/btrfs/free-space-cache.c | 14 +++--- fs/btrfs/inode.c | 42 +++++++++--------- fs/btrfs/ioctl.c | 22 +++++---- fs/btrfs/relocation.c | 30 +++++++------- fs/btrfs/root-tree.c | 10 ++-- fs/btrfs/tree-defrag.c | 2 +- fs/btrfs/tree-log.c | 98 +++++++++++++++++++++--------------------- fs/btrfs/volumes.c | 16 ++++---- fs/btrfs/xattr.c | 4 +- 15 files changed, 185 insertions(+), 183 deletions(-) -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
