Allow users to delete the snapshots/subvols they own. Instead of CAP_SYS_ADMIN, require that - the user has write and exec permission on the parent directory - security_inode_rmdir() doesn''t object - the user has write and exec permission on the subvol directory - the user owns the subvol directory - the directory and subvol append flags are not set This is a bit more strict than the requirements for ''rm -f subvol/*'', which is allowed with just ''wx'' on non-owned non-sticky dirs. It is less strict than ''rmdir subvol'', which has additional requirements if the parent directory is sticky. It is less strict than ''rm -rf subvol'', which would require scanning the entire subvol to ensure all content is deletable. Signed-off-by: Sage Weil <sage@newdream.net> --- fs/btrfs/ioctl.c | 25 ++++++++++++++++++++++--- 1 files changed, 22 insertions(+), 3 deletions(-) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 9254b3d..b31283a 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -1227,9 +1227,6 @@ static noinline int btrfs_ioctl_snap_destroy(struct file *file, int ret; int err = 0; - if (!capable(CAP_SYS_ADMIN)) - return -EPERM; - vol_args = memdup_user(arg, sizeof(*vol_args)); if (IS_ERR(vol_args)) return PTR_ERR(vol_args); @@ -1264,6 +1261,28 @@ static noinline int btrfs_ioctl_snap_destroy(struct file *file, goto out_dput; } + /* + * Allow subvol deletion if we own the subvol and we would + * (approximately) be allowed to rmdir it. Strictly speaking, + * we could possibly delete everything with fewer permissions, + * or might require more permissions to remove all files + * contained by the subvol, but we aren''t trying to mimic + * directory semantics perfectly. + */ + err = inode_permission(dir, MAY_WRITE | MAY_EXEC); + if (err) + goto out_dput; + err = security_inode_rmdir(dir, dentry); + if (err) + goto out_dput; + err = inode_permission(inode, MAY_WRITE | MAY_EXEC); + if (err) + goto out_dput; + err = -EPERM; + if (inode->i_uid != current_fsuid() || IS_APPEND(dir) || + IS_APPEND(inode)) + goto out_dput; + dest = BTRFS_I(inode)->root; mutex_lock(&inode->i_mutex); -- 1.7.0 -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html