Re: [RFC v5 1/1] ns: add binfmt_misc to the user namespace

On 2018-10-09, Laurent Vivier [off-list ref] wrote:

Le 09/10/2018 à 17:16, Tycho Andersen a écrit :

quoted

On Tue, Oct 09, 2018 at 12:37:52PM +0200, Laurent Vivier wrote:

quoted

@@ -80,18 +74,32 @@ static int entry_count;
  */
 #define MAX_REGISTER_LENGTH 1920
 
+static struct binfmt_namespace *binfmt_ns(struct user_namespace *ns)
+{
+	struct binfmt_namespace *b_ns;
+
+	while (ns) {
+		b_ns = READ_ONCE(ns->binfmt_ns);
+		if (b_ns)
+			return b_ns;
+		ns = ns->parent;
+	}
+	WARN_ON_ONCE(1);

It looks like we warn here,

quoted

@@ -133,17 +141,18 @@ static int load_misc_binary(struct linux_binprm *bprm)
 	struct file *interp_file = NULL;
 	int retval;
 	int fd_binary = -1;
+	struct binfmt_namespace *ns = binfmt_ns(current_user_ns());
 
 	retval = -ENOEXEC;
-	if (!enabled)
+	if (!ns->enabled)

...but then in cases like this we immediately dereference the pointer
anyways and crash. Can we return some other error code here in the !ns
case so we don't crash?

My concern here is I don't want to add code to check an error case that
cannot happen. The first namespace binfmt_ns pointer is initialized with
&init_binfmt_ns, so the return value cannot be NULL.

I'd argue that BUG() is a better thing to do then -- if doing a dummy
error path makes no sense. Though IIRC BUG() is no longer a popular
thing to do.

-- 
Aleksa Sarai
Senior Software Engineer (Containers)
SUSE Linux GmbH
<https://www.cyphar.com/>