pidns: Capture the user namespace and filter ns_last_pid

- Capture the the user namespace that creates the pid namespace
- Use that user namespace to test if it is ok to write to
  /proc/sys/kernel/ns_last_pid.

Zhao Hongjiang <zhaohongjiang@huawei.com> noticed I was missing a put_user_ns
in when destroying a pid_ns.  I have foloded his patch into this one
so that bisects will work properly.

Acked-by: Serge Hallyn <serge.hallyn@canonical.com>
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>

3 files changed, 14 insertions, 6 deletions

diff --git a/kernel/nsproxy.c b/kernel/nsproxy.c
index 7e1c3de1ce45..ca27d2c5264d 100644
--- a/kernel/nsproxy.c
+++ b/kernel/nsproxy.c
@@ -84,7 +84,7 @@ static struct nsproxy *create_new_namespaces(unsigned long flags,
 		goto out_ipc;
 	}
 
-	new_nsp->pid_ns = copy_pid_ns(flags, task_active_pid_ns(tsk));
+	new_nsp->pid_ns = copy_pid_ns(flags, task_cred_xxx(tsk, user_ns), task_active_pid_ns(tsk));
 	if (IS_ERR(new_nsp->pid_ns)) {
 		err = PTR_ERR(new_nsp->pid_ns);
 		goto out_pid;
diff --git a/kernel/pid.c b/kernel/pid.c
index aebd4f5aaf41..2a624f1486e1 100644
--- a/kernel/pid.c
+++ b/kernel/pid.c
@@ -78,6 +78,7 @@ struct pid_namespace init_pid_ns = {
 	.last_pid = 0,
 	.level = 0,
 	.child_reaper = &init_task,
+	.user_ns = &init_user_ns,
 };
 EXPORT_SYMBOL_GPL(init_pid_ns);
 
diff --git a/kernel/pid_namespace.c b/kernel/pid_namespace.c
index 7b07cc0dfb75..b2604950aa50 100644
--- a/kernel/pid_namespace.c
+++ b/kernel/pid_namespace.c
@@ -10,6 +10,7 @@
 
 #include <linux/pid.h>
 #include <linux/pid_namespace.h>
+#include <linux/user_namespace.h>
 #include <linux/syscalls.h>
 #include <linux/err.h>
 #include <linux/acct.h>
@@ -74,7 +75,8 @@ err_alloc:
 /* MAX_PID_NS_LEVEL is needed for limiting size of 'struct pid' */
 #define MAX_PID_NS_LEVEL 32
 
-static struct pid_namespace *create_pid_namespace(struct pid_namespace *parent_pid_ns)
+static struct pid_namespace *create_pid_namespace(struct user_namespace *user_ns,
+	struct pid_namespace *parent_pid_ns)
 {
 	struct pid_namespace *ns;
 	unsigned int level = parent_pid_ns->level + 1;
@@ -102,6 +104,7 @@ static struct pid_namespace *create_pid_namespace(struct pid_namespace *parent_p
 	kref_init(&ns->kref);
 	ns->level = level;
 	ns->parent = get_pid_ns(parent_pid_ns);
+	ns->user_ns = get_user_ns(user_ns);
 
 	set_bit(0, ns->pidmap[0].page);
 	atomic_set(&ns->pidmap[0].nr_free, BITS_PER_PAGE - 1);
@@ -117,6 +120,7 @@ static struct pid_namespace *create_pid_namespace(struct pid_namespace *parent_p
 
 out_put_parent_pid_ns:
 	put_pid_ns(parent_pid_ns);
+	put_user_ns(user_ns);
 out_free_map:
 	kfree(ns->pidmap[0].page);
 out_free:
@@ -131,16 +135,18 @@ static void destroy_pid_namespace(struct pid_namespace *ns)
 
 	for (i = 0; i < PIDMAP_ENTRIES; i++)
 		kfree(ns->pidmap[i].page);
+	put_user_ns(ns->user_ns);
 	kmem_cache_free(pid_ns_cachep, ns);
 }
 
-struct pid_namespace *copy_pid_ns(unsigned long flags, struct pid_namespace *old_ns)
+struct pid_namespace *copy_pid_ns(unsigned long flags,
+	struct user_namespace *user_ns, struct pid_namespace *old_ns)
 {
 	if (!(flags & CLONE_NEWPID))
 		return get_pid_ns(old_ns);
 	if (flags & (CLONE_THREAD|CLONE_PARENT))
 		return ERR_PTR(-EINVAL);
-	return create_pid_namespace(old_ns);
+	return create_pid_namespace(user_ns, old_ns);
 }
 
 static void free_pid_ns(struct kref *kref)
@@ -239,9 +245,10 @@ void zap_pid_ns_processes(struct pid_namespace *pid_ns)
 static int pid_ns_ctl_handler(struct ctl_table *table, int write,
 		void __user *buffer, size_t *lenp, loff_t *ppos)
 {
+	struct pid_namespace *pid_ns = task_active_pid_ns(current);
 	struct ctl_table tmp = *table;
 
-	if (write && !capable(CAP_SYS_ADMIN))
+	if (write && !ns_capable(pid_ns->user_ns, CAP_SYS_ADMIN))
 		return -EPERM;
 
 	/*
@@ -250,7 +257,7 @@ static int pid_ns_ctl_handler(struct ctl_table *table, int write,
 	 * it should synchronize its usage with external means.
 	 */
 
-	tmp.data = &current->nsproxy->pid_ns->last_pid;
+	tmp.data = &pid_ns->last_pid;
 	return proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
 }