ZEN: Add sysctl and CONFIG to disallow unprivileged CLONE_NEWUSER

Our default behavior continues to match the vanilla kernel.

include/linux/user_namespace.h

@@ -137,6 +137,8 @@ static inline void set_rlimit_ucount_max(struct user_namespace *ns,
 
 #ifdef CONFIG_USER_NS
 
+extern int unprivileged_userns_clone;
+
 static inline struct user_namespace *get_user_ns(struct user_namespace *ns)
 {
 	if (ns)
@@ -170,6 +172,8 @@ extern bool current_in_userns(const struct user_namespace *target_ns);
 struct ns_common *ns_get_owner(struct ns_common *ns);
 #else
 
+#define unprivileged_userns_clone 0
+
 static inline struct user_namespace *get_user_ns(struct user_namespace *ns)
 {
 	return &init_user_ns;

init/Kconfig

@@ -1200,6 +1200,22 @@ config USER_NS
 
 	  If unsure, say N.
 
+config USER_NS_UNPRIVILEGED
+	bool "Allow unprivileged users to create namespaces"
+	default y
+	depends on USER_NS
+	help
+	  When disabled, unprivileged users will not be able to create
+	  new namespaces. Allowing users to create their own namespaces
+	  has been part of several recent local privilege escalation
+	  exploits, so if you need user namespaces but are
+	  paranoid^Wsecurity-conscious you want to disable this.
+
+	  This setting can be overridden at runtime via the
+	  kernel.unprivileged_userns_clone sysctl.
+
+	  If unsure, say Y.
+
 config PID_NS
 	bool "PID Namespaces"
 	default y

kernel/fork.c

@@ -99,6 +99,10 @@
 #include <linux/io_uring.h>
 #include <linux/bpf.h>
 
+#ifdef CONFIG_USER_NS
+#include <linux/user_namespace.h>
+#endif
+
 #include <asm/pgalloc.h>
 #include <linux/uaccess.h>
 #include <asm/mmu_context.h>
@@ -1899,6 +1903,10 @@ static __latent_entropy struct task_struct *copy_process(
 	if ((clone_flags & (CLONE_NEWUSER|CLONE_FS)) == (CLONE_NEWUSER|CLONE_FS))
 		return ERR_PTR(-EINVAL);
 
+	if ((clone_flags & CLONE_NEWUSER) && !unprivileged_userns_clone)
+		if (!capable(CAP_SYS_ADMIN))
+			return ERR_PTR(-EPERM);
+
 	/*
 	 * Thread groups must share signals as well, and detached threads
 	 * can only be started up within the thread group.
@@ -3003,6 +3011,12 @@ int ksys_unshare(unsigned long unshare_flags)
 	if (unshare_flags & CLONE_NEWNS)
 		unshare_flags |= CLONE_FS;
 
+	if ((unshare_flags & CLONE_NEWUSER) && !unprivileged_userns_clone) {
+		err = -EPERM;
+		if (!capable(CAP_SYS_ADMIN))
+			goto bad_unshare_out;
+	}
+
 	err = check_unshare_flags(unshare_flags);
 	if (err)
 		goto bad_unshare_out;

kernel/sysctl.c

@@ -105,6 +105,9 @@
 #ifdef CONFIG_LOCKUP_DETECTOR
 #include <linux/nmi.h>
 #endif
+#ifdef CONFIG_USER_NS
+#include <linux/user_namespace.h>
+#endif
 
 #if defined(CONFIG_SYSCTL)
 
@@ -1908,6 +1911,15 @@ static struct ctl_table kern_table[] = {
 		.proc_handler	= proc_dointvec,
 	},
 #endif
+#ifdef CONFIG_USER_NS
+	{
+		.procname	= "unprivileged_userns_clone",
+		.data		= &unprivileged_userns_clone,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec,
+	},
+#endif
 #ifdef CONFIG_PROC_SYSCTL
 	{
 		.procname	= "tainted",

kernel/user_namespace.c

@@ -21,6 +21,13 @@
 #include <linux/bsearch.h>
 #include <linux/sort.h>
 
+/* sysctl */
+#ifdef CONFIG_USER_NS_UNPRIVILEGED
+int unprivileged_userns_clone = 1;
+#else
+int unprivileged_userns_clone;
+#endif
+
 static struct kmem_cache *user_ns_cachep __read_mostly;
 static DEFINE_MUTEX(userns_state_mutex);