apparmor: grab ns lock and refresh when looking up changehat child profiles

[ Upstream commit 32e9276 ]

There was a race condition involving change_hat and profile replacement in
which replacement of the parent profile during a changehat operation could
result in the list of children becoming empty and the changehat operation
failing. To prevent this:
 - grab the namespace lock until we've built the hat transition, and
 - use aa_get_newest_profile to avoid using stale profile objects.

Link: https://bugs.launchpad.net/bugs/2139664
Fixes: 89dbf19 ("apparmor: move change_hat mediation to using labels")
Reviewed-by: Georgia Garcia <georgia.garcia@canonical.com>
Signed-off-by: Ryan Lee <ryan.lee@canonical.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

Author: canonical-rlee287

security/apparmor/domain.c

@@ -13,6 +13,7 @@
 #include <linux/fs.h>
 #include <linux/file.h>
 #include <linux/mount.h>
+#include <linux/mutex.h>
 #include <linux/syscalls.h>
 #include <linux/personality.h>
 #include <linux/xattr.h>
@@ -1061,6 +1062,7 @@ static struct aa_label *change_hat(const struct cred *subj_cred,
 				   int count, int flags)
 {
 	struct aa_profile *profile, *root, *hat = NULL;
+	struct aa_ns *ns, *new_ns;
 	struct aa_label *new;
 	struct label_it it;
 	bool sibling = false;
@@ -1071,6 +1073,32 @@ static struct aa_label *change_hat(const struct cred *subj_cred,
 	AA_BUG(!hats);
 	AA_BUG(count < 1);
 
+	/*
+	 * Acquire the newest label and then hold the lock until we choose a
+	 * hat, so that profile replacement doesn't atomically truncate the
+	 * list of potential hats. Because we are getting the namespaces from
+	 * the profiles and label, we can rely on the namespaces being live
+	 * and avoid incrementing their refcounts while grabbing the lock.
+	 */
+	label = aa_get_label(label);
+	ns = labels_ns(label);
+
+retry:
+	mutex_lock_nested(&ns->lock, ns->level);
+	if (label_is_stale(label)) {
+		new = aa_get_newest_label(label);
+		new_ns = labels_ns(new);
+		if (new_ns != ns) {
+			aa_put_label(new);
+			mutex_unlock(&ns->lock);
+			ns = new_ns;
+			label = new;
+			goto retry;
+		}
+		aa_put_label(label);
+		label = new;
+	}
+
 	if (PROFILE_IS_HAT(labels_profile(label)))
 		sibling = true;
 
@@ -1079,7 +1107,7 @@ static struct aa_label *change_hat(const struct cred *subj_cred,
 		name = hats[i];
 		label_for_each_in_ns(it, labels_ns(label), label, profile) {
 			if (sibling && PROFILE_IS_HAT(profile)) {
-				root = aa_get_profile_rcu(&profile->parent);
+				root = aa_get_profile(profile->parent);
 			} else if (!sibling && !PROFILE_IS_HAT(profile)) {
 				root = aa_get_profile(profile);
 			} else {	/* conflicting change type */
@@ -1139,6 +1167,7 @@ static struct aa_label *change_hat(const struct cred *subj_cred,
 				      GLOBAL_ROOT_UID, info, error);
 		}
 	}
+	mutex_unlock(&ns->lock);
 	return ERR_PTR(error);
 
 build:
@@ -1151,7 +1180,7 @@ static struct aa_label *change_hat(const struct cred *subj_cred,
 		error = -ENOMEM;
 		goto fail;
 	} /* else if (IS_ERR) build_change_hat has logged error so return new */
-
+	mutex_unlock(&ns->lock);
 	return new;
 }