restore: correctly restore cgroup mounts inside a container

Before the nsroot= mount option, we were just getting lucky because the
cgroup superblocks "matched" when inspecting them from userspace, so we
were actually getting a bind mount from the host when migrating from within
cgroup namespaces.

Instead, let's actually do a new (i.e. not a bind mount) for cgroup
namespaces. For this, we need two things:

1. to prepare the cgroup namespace (and thus the cgroups) before the mount
   ns, so when the mount() occurrs it is relative to the right cgroup path.

2. not reject cgroup filesystems with no root. A cgroup ns mount looks
   like:

	 223 222 0:22 /lxc/unpriv /sys/fs/cgroup/systemd rw,nosuid,nodev,noexec,relatime - cgroup cgroup rw,xattr,release_agent=/lib/systemd/systemd-cgroups-agent,name=systemd,nsroot=/lxc/unpriv

   i.e. it has /lxc/unpriv as its root, and thus doesn't look rooted to CRIU.
   We use the fstype->parse hook to rewrite this root to /, since it
   is handled by the cgroup ns infrastructure.

v2: add new fstype->munge hook, allowing fstypes to munge their parsed
    mountinfo entries if they want to. this allows us to get rid of the
    ugly hacks with FSTYPE__CGROUP everywhere in teh patch.
v3: s/fstype->munge/fstype->parse for FSTYPE__CGROUP
Signed-off-by: Tycho Andersen <tycho.andersen@canonical.com>
Signed-off-by: Pavel Emelyanov <xemul@virtuozzo.com>

criu/cr-restore.c

@@ -1564,6 +1564,15 @@ static int restore_task_with_children(void *_arg)
 			goto err;
 	}
 
+	/*
+	 * Call this _before_ forking to optimize cgroups
+	 * restore -- if all tasks live in one set of cgroups
+	 * we will only move the root one there, others will
+	 * just have it inherited.
+	 */
+	if (prepare_task_cgroup(current) < 0)
+		goto err;
+
 	/* Restore root task */
 	if (current->parent == NULL) {
 		if (restore_finish_stage(CR_STATE_RESTORE_NS) < 0)
@@ -1596,15 +1605,6 @@ static int restore_task_with_children(void *_arg)
 	if (prepare_mappings())
 		goto err;
 
-	/*
-	 * Call this _before_ forking to optimize cgroups
-	 * restore -- if all tasks live in one set of cgroups
-	 * we will only move the root one there, others will
-	 * just have it inherited.
-	 */
-	if (prepare_task_cgroup(current) < 0)
-		goto err;
-
 	if (prepare_sigactions() < 0)
 		goto err;
 

criu/mount.c

@@ -1632,6 +1632,22 @@ out:
 	return ret;
 }
 
+static int cgroup_parse(struct mount_info *pm)
+{
+	if (!(root_ns_mask & CLONE_NEWCGROUP))
+		return 0;
+
+	/* cgroup namespaced mounts don't look rooted to CRIU, so let's fake it
+	 * here.
+	 */
+	xfree(pm->root);
+	pm->root = xstrdup("/");
+	if (!pm->root)
+		return -1;
+
+	return 0;
+}
+
 static int dump_empty_fs(struct mount_info *pm)
 {
 	int fd, ret = -1;
@@ -1717,6 +1733,7 @@ static struct fstype fstypes[32] = {
 	}, {
 		.name = "cgroup",
 		.code = FSTYPE__CGROUP,
+		.parse = cgroup_parse,
 	}, {
 		.name = "aufs",
 		.code = FSTYPE__AUFS,