[linux-block.git] / kernel / pid_namespace.c

/*
 * Pid namespaces
 *
 * Authors:
 *    (C) 2007 Pavel Emelyanov <xemul@openvz.org>, OpenVZ, SWsoft Inc.
 *    (C) 2007 Sukadev Bhattiprolu <sukadev@us.ibm.com>, IBM
 *     Many thanks to Oleg Nesterov for comments and help
 *
 */

#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>
#include <linux/slab.h>
#include <linux/proc_fs.h>
#include <linux/reboot.h>
#include <linux/export.h>

#define BITS_PER_PAGE		(PAGE_SIZE*8)

struct pid_cache {
	int nr_ids;
	char name[16];
	struct kmem_cache *cachep;
	struct list_head list;
};

static LIST_HEAD(pid_caches_lh);
static DEFINE_MUTEX(pid_caches_mutex);
static struct kmem_cache *pid_ns_cachep;

/*
 * creates the kmem cache to allocate pids from.
 * @nr_ids: the number of numerical ids this pid will have to carry
 */

static struct kmem_cache *create_pid_cachep(int nr_ids)
{
	struct pid_cache *pcache;
	struct kmem_cache *cachep;

	mutex_lock(&pid_caches_mutex);
	list_for_each_entry(pcache, &pid_caches_lh, list)
		if (pcache->nr_ids == nr_ids)
			goto out;

	pcache = kmalloc(sizeof(struct pid_cache), GFP_KERNEL);
	if (pcache == NULL)
		goto err_alloc;

	snprintf(pcache->name, sizeof(pcache->name), "pid_%d", nr_ids);
	cachep = kmem_cache_create(pcache->name,
			sizeof(struct pid) + (nr_ids - 1) * sizeof(struct upid),
			0, SLAB_HWCACHE_ALIGN, NULL);
	if (cachep == NULL)
		goto err_cachep;

	pcache->nr_ids = nr_ids;
	pcache->cachep = cachep;
	list_add(&pcache->list, &pid_caches_lh);
out:
	mutex_unlock(&pid_caches_mutex);
	return pcache->cachep;

err_cachep:
	kfree(pcache);
err_alloc:
	mutex_unlock(&pid_caches_mutex);
	return NULL;
}

static void proc_cleanup_work(struct work_struct *work)
{
	struct pid_namespace *ns = container_of(work, struct pid_namespace, proc_work);
	pid_ns_release_proc(ns);
}

/* 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 user_namespace *user_ns,
	struct pid_namespace *parent_pid_ns)
{
	struct pid_namespace *ns;
	unsigned int level = parent_pid_ns->level + 1;
	int i;
	int err;

	if (level > MAX_PID_NS_LEVEL) {
		err = -EINVAL;
		goto out;
	}

	err = -ENOMEM;
	ns = kmem_cache_zalloc(pid_ns_cachep, GFP_KERNEL);
	if (ns == NULL)
		goto out;

	ns->pidmap[0].page = kzalloc(PAGE_SIZE, GFP_KERNEL);
	if (!ns->pidmap[0].page)
		goto out_free;

	ns->pid_cachep = create_pid_cachep(level + 1);
	if (ns->pid_cachep == NULL)
		goto out_free_map;

	kref_init(&ns->kref);
	ns->level = level;
	ns->parent = get_pid_ns(parent_pid_ns);
	ns->user_ns = get_user_ns(user_ns);
	INIT_WORK(&ns->proc_work, proc_cleanup_work);

	set_bit(0, ns->pidmap[0].page);
	atomic_set(&ns->pidmap[0].nr_free, BITS_PER_PAGE - 1);

	for (i = 1; i < PIDMAP_ENTRIES; i++)
		atomic_set(&ns->pidmap[i].nr_free, BITS_PER_PAGE);

	return ns;

out_free_map:
	kfree(ns->pidmap[0].page);
out_free:
	kmem_cache_free(pid_ns_cachep, ns);
out:
	return ERR_PTR(err);
}

static void destroy_pid_namespace(struct pid_namespace *ns)
{
	int i;

	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 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);
	if (task_active_pid_ns(current) != old_ns)
		return ERR_PTR(-EINVAL);
	return create_pid_namespace(user_ns, old_ns);
}

static void free_pid_ns(struct kref *kref)
{
	struct pid_namespace *ns;

	ns = container_of(kref, struct pid_namespace, kref);
	destroy_pid_namespace(ns);
}

void put_pid_ns(struct pid_namespace *ns)
{
	struct pid_namespace *parent;

	while (ns != &init_pid_ns) {
		parent = ns->parent;
		if (!kref_put(&ns->kref, free_pid_ns))
			break;
		ns = parent;
	}
}
EXPORT_SYMBOL_GPL(put_pid_ns);

void zap_pid_ns_processes(struct pid_namespace *pid_ns)
{
	int nr;
	int rc;
	struct task_struct *task, *me = current;

	/* Ignore SIGCHLD causing any terminated children to autoreap */
	spin_lock_irq(&me->sighand->siglock);
	me->sighand->action[SIGCHLD - 1].sa.sa_handler = SIG_IGN;
	spin_unlock_irq(&me->sighand->siglock);

	/*
	 * The last thread in the cgroup-init thread group is terminating.
	 * Find remaining pid_ts in the namespace, signal and wait for them
	 * to exit.
	 *
	 * Note:  This signals each threads in the namespace - even those that
	 * 	  belong to the same thread group, To avoid this, we would have
	 * 	  to walk the entire tasklist looking a processes in this
	 * 	  namespace, but that could be unnecessarily expensive if the
	 * 	  pid namespace has just a few processes. Or we need to
	 * 	  maintain a tasklist for each pid namespace.
	 *
	 */
	read_lock(&tasklist_lock);
	nr = next_pidmap(pid_ns, 1);
	while (nr > 0) {
		rcu_read_lock();

		task = pid_task(find_vpid(nr), PIDTYPE_PID);
		if (task && !__fatal_signal_pending(task))
			send_sig_info(SIGKILL, SEND_SIG_FORCED, task);

		rcu_read_unlock();

		nr = next_pidmap(pid_ns, nr);
	}
	read_unlock(&tasklist_lock);

	/* Firstly reap the EXIT_ZOMBIE children we may have. */
	do {
		clear_thread_flag(TIF_SIGPENDING);
		rc = sys_wait4(-1, NULL, __WALL, NULL);
	} while (rc != -ECHILD);

	/*
	 * sys_wait4() above can't reap the TASK_DEAD children.
	 * Make sure they all go away, see free_pid().
	 */
	for (;;) {
		set_current_state(TASK_UNINTERRUPTIBLE);
		if (pid_ns->nr_hashed == 1)
			break;
		schedule();
	}
	__set_current_state(TASK_RUNNING);

	if (pid_ns->reboot)
		current->signal->group_exit_code = pid_ns->reboot;

	acct_exit_ns(pid_ns);
	return;
}

#ifdef CONFIG_CHECKPOINT_RESTORE
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 && !ns_capable(pid_ns->user_ns, CAP_SYS_ADMIN))
		return -EPERM;

	/*
	 * Writing directly to ns' last_pid field is OK, since this field
	 * is volatile in a living namespace anyway and a code writing to
	 * it should synchronize its usage with external means.
	 */

	tmp.data = &pid_ns->last_pid;
	return proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
}

extern int pid_max;
static int zero = 0;
static struct ctl_table pid_ns_ctl_table[] = {
	{
		.procname = "ns_last_pid",
		.maxlen = sizeof(int),
		.mode = 0666, /* permissions are checked in the handler */
		.proc_handler = pid_ns_ctl_handler,
		.extra1 = &zero,
		.extra2 = &pid_max,
	},
	{ }
};
static struct ctl_path kern_path[] = { { .procname = "kernel", }, { } };
#endif	/* CONFIG_CHECKPOINT_RESTORE */

int reboot_pid_ns(struct pid_namespace *pid_ns, int cmd)
{
	if (pid_ns == &init_pid_ns)
		return 0;

	switch (cmd) {
	case LINUX_REBOOT_CMD_RESTART2:
	case LINUX_REBOOT_CMD_RESTART:
		pid_ns->reboot = SIGHUP;
		break;

	case LINUX_REBOOT_CMD_POWER_OFF:
	case LINUX_REBOOT_CMD_HALT:
		pid_ns->reboot = SIGINT;
		break;
	default:
		return -EINVAL;
	}

	read_lock(&tasklist_lock);
	force_sig(SIGKILL, pid_ns->child_reaper);
	read_unlock(&tasklist_lock);

	do_exit(0);

	/* Not reached */
	return 0;
}

static __init int pid_namespaces_init(void)
{
	pid_ns_cachep = KMEM_CACHE(pid_namespace, SLAB_PANIC);

#ifdef CONFIG_CHECKPOINT_RESTORE
	register_sysctl_paths(kern_path, pid_ns_ctl_table);
#endif
	return 0;
}

__initcall(pid_namespaces_init);
Commit	Line	Data
74bd59bb PE	1	/*
	2	* Pid namespaces
	3	*
	4	* Authors:
	5	* (C) 2007 Pavel Emelyanov <xemul@openvz.org>, OpenVZ, SWsoft Inc.
	6	* (C) 2007 Sukadev Bhattiprolu <sukadev@us.ibm.com>, IBM
	7	* Many thanks to Oleg Nesterov for comments and help
	8	*
	9	*/
	10
	11	#include <linux/pid.h>
	12	#include <linux/pid_namespace.h>
49f4d8b9	13	#include <linux/user_namespace.h>
74bd59bb PE	14	#include <linux/syscalls.h>
74bd59bb PE	15	#include <linux/err.h>
0b6b030f	16	#include <linux/acct.h>
5a0e3ad6	17	#include <linux/slab.h>
4308eebb	18	#include <linux/proc_fs.h>
cf3f8921	19	#include <linux/reboot.h>
523a6a94	20	#include <linux/export.h>
74bd59bb PE	21
	22	#define BITS_PER_PAGE (PAGE_SIZE*8)
	23
	24	struct pid_cache {
	25	int nr_ids;
	26	char name[16];
	27	struct kmem_cache *cachep;
	28	struct list_head list;
	29	};
	30
	31	static LIST_HEAD(pid_caches_lh);
	32	static DEFINE_MUTEX(pid_caches_mutex);
	33	static struct kmem_cache *pid_ns_cachep;
	34
	35	/*
	36	* creates the kmem cache to allocate pids from.
	37	* @nr_ids: the number of numerical ids this pid will have to carry
	38	*/
	39
	40	static struct kmem_cache *create_pid_cachep(int nr_ids)
	41	{
	42	struct pid_cache *pcache;
	43	struct kmem_cache *cachep;
	44
	45	mutex_lock(&pid_caches_mutex);
	46	list_for_each_entry(pcache, &pid_caches_lh, list)
	47	if (pcache->nr_ids == nr_ids)
	48	goto out;
	49
	50	pcache = kmalloc(sizeof(struct pid_cache), GFP_KERNEL);
	51	if (pcache == NULL)
	52	goto err_alloc;
	53
	54	snprintf(pcache->name, sizeof(pcache->name), "pid_%d", nr_ids);
	55	cachep = kmem_cache_create(pcache->name,
	56	sizeof(struct pid) + (nr_ids - 1) * sizeof(struct upid),
	57	0, SLAB_HWCACHE_ALIGN, NULL);
	58	if (cachep == NULL)
	59	goto err_cachep;
	60
	61	pcache->nr_ids = nr_ids;
	62	pcache->cachep = cachep;
	63	list_add(&pcache->list, &pid_caches_lh);
	64	out:
	65	mutex_unlock(&pid_caches_mutex);
	66	return pcache->cachep;
	67
	68	err_cachep:
	69	kfree(pcache);
	70	err_alloc:
	71	mutex_unlock(&pid_caches_mutex);
	72	return NULL;
	73	}
	74
0a01f2cc EB	75	static void proc_cleanup_work(struct work_struct *work)
	76	{
	77	struct pid_namespace *ns = container_of(work, struct pid_namespace, proc_work);
	78	pid_ns_release_proc(ns);
	79	}
	80
f2302505 AV	81	/* MAX_PID_NS_LEVEL is needed for limiting size of 'struct pid' */
	82	#define MAX_PID_NS_LEVEL 32
	83
49f4d8b9 EB	84	static struct pid_namespace create_pid_namespace(struct user_namespace user_ns,
49f4d8b9 EB	85	struct pid_namespace *parent_pid_ns)
74bd59bb PE	86	{
74bd59bb PE	87	struct pid_namespace *ns;
ed469a63	88	unsigned int level = parent_pid_ns->level + 1;
f2302505 AV	89	int i;
	90	int err;
	91
	92	if (level > MAX_PID_NS_LEVEL) {
	93	err = -EINVAL;
	94	goto out;
	95	}
74bd59bb	96
f2302505	97	err = -ENOMEM;
84406c15	98	ns = kmem_cache_zalloc(pid_ns_cachep, GFP_KERNEL);
74bd59bb PE	99	if (ns == NULL)
	100	goto out;
	101
	102	ns->pidmap[0].page = kzalloc(PAGE_SIZE, GFP_KERNEL);
	103	if (!ns->pidmap[0].page)
	104	goto out_free;
	105
	106	ns->pid_cachep = create_pid_cachep(level + 1);
	107	if (ns->pid_cachep == NULL)
	108	goto out_free_map;
	109
	110	kref_init(&ns->kref);
74bd59bb	111	ns->level = level;
ed469a63	112	ns->parent = get_pid_ns(parent_pid_ns);
49f4d8b9	113	ns->user_ns = get_user_ns(user_ns);
0a01f2cc	114	INIT_WORK(&ns->proc_work, proc_cleanup_work);
74bd59bb PE	115
	116	set_bit(0, ns->pidmap[0].page);
	117	atomic_set(&ns->pidmap[0].nr_free, BITS_PER_PAGE - 1);
	118
84406c15	119	for (i = 1; i < PIDMAP_ENTRIES; i++)
74bd59bb	120	atomic_set(&ns->pidmap[i].nr_free, BITS_PER_PAGE);
74bd59bb PE	121
	122	return ns;
	123
	124	out_free_map:
	125	kfree(ns->pidmap[0].page);
	126	out_free:
	127	kmem_cache_free(pid_ns_cachep, ns);
	128	out:
4308eebb	129	return ERR_PTR(err);
74bd59bb PE	130	}
	131
	132	static void destroy_pid_namespace(struct pid_namespace *ns)
	133	{
	134	int i;
	135
	136	for (i = 0; i < PIDMAP_ENTRIES; i++)
	137	kfree(ns->pidmap[i].page);
49f4d8b9	138	put_user_ns(ns->user_ns);
74bd59bb PE	139	kmem_cache_free(pid_ns_cachep, ns);
	140	}
	141
49f4d8b9 EB	142	struct pid_namespace *copy_pid_ns(unsigned long flags,
49f4d8b9 EB	143	struct user_namespace user_ns, struct pid_namespace old_ns)
74bd59bb	144	{
74bd59bb	145	if (!(flags & CLONE_NEWPID))
dca4a979	146	return get_pid_ns(old_ns);
e5a47386	147	if (flags & (CLONE_THREAD\|CLONE_PARENT))
dca4a979	148	return ERR_PTR(-EINVAL);
225778d6 EB	149	if (task_active_pid_ns(current) != old_ns)
225778d6 EB	150	return ERR_PTR(-EINVAL);
49f4d8b9	151	return create_pid_namespace(user_ns, old_ns);
74bd59bb PE	152	}
74bd59bb PE	153
bbc2e3ef	154	static void free_pid_ns(struct kref *kref)
74bd59bb	155	{
bbc2e3ef	156	struct pid_namespace *ns;
74bd59bb PE	157
74bd59bb PE	158	ns = container_of(kref, struct pid_namespace, kref);
74bd59bb	159	destroy_pid_namespace(ns);
bbc2e3ef	160	}
74bd59bb	161
bbc2e3ef CG	162	void put_pid_ns(struct pid_namespace *ns)
	163	{
	164	struct pid_namespace *parent;
	165
	166	while (ns != &init_pid_ns) {
	167	parent = ns->parent;
	168	if (!kref_put(&ns->kref, free_pid_ns))
	169	break;
	170	ns = parent;
	171	}
74bd59bb	172	}
bbc2e3ef	173	EXPORT_SYMBOL_GPL(put_pid_ns);
74bd59bb PE	174
	175	void zap_pid_ns_processes(struct pid_namespace *pid_ns)
	176	{
	177	int nr;
	178	int rc;
00c10bc1 EB	179	struct task_struct task, me = current;
	180
	181	/* Ignore SIGCHLD causing any terminated children to autoreap */
	182	spin_lock_irq(&me->sighand->siglock);
	183	me->sighand->action[SIGCHLD - 1].sa.sa_handler = SIG_IGN;
	184	spin_unlock_irq(&me->sighand->siglock);
74bd59bb PE	185
	186	/*
	187	* The last thread in the cgroup-init thread group is terminating.
	188	* Find remaining pid_ts in the namespace, signal and wait for them
	189	* to exit.
	190	*
	191	* Note: This signals each threads in the namespace - even those that
	192	* belong to the same thread group, To avoid this, we would have
	193	* to walk the entire tasklist looking a processes in this
	194	* namespace, but that could be unnecessarily expensive if the
	195	* pid namespace has just a few processes. Or we need to
	196	* maintain a tasklist for each pid namespace.
	197	*
	198	*/
	199	read_lock(&tasklist_lock);
	200	nr = next_pidmap(pid_ns, 1);
	201	while (nr > 0) {
e4da026f SB	202	rcu_read_lock();
e4da026f SB	203
e4da026f	204	task = pid_task(find_vpid(nr), PIDTYPE_PID);
a02d6fd6 ON	205	if (task && !__fatal_signal_pending(task))
a02d6fd6 ON	206	send_sig_info(SIGKILL, SEND_SIG_FORCED, task);
e4da026f SB	207
	208	rcu_read_unlock();
	209
74bd59bb PE	210	nr = next_pidmap(pid_ns, nr);
	211	}
	212	read_unlock(&tasklist_lock);
	213
6347e900	214	/* Firstly reap the EXIT_ZOMBIE children we may have. */
74bd59bb PE	215	do {
	216	clear_thread_flag(TIF_SIGPENDING);
	217	rc = sys_wait4(-1, NULL, __WALL, NULL);
	218	} while (rc != -ECHILD);
	219
6347e900 EB	220	/*
6347e900 EB	221	* sys_wait4() above can't reap the TASK_DEAD children.
af4b8a83	222	* Make sure they all go away, see free_pid().
6347e900 EB	223	*/
6347e900 EB	224	for (;;) {
af4b8a83 EB	225	set_current_state(TASK_UNINTERRUPTIBLE);
af4b8a83 EB	226	if (pid_ns->nr_hashed == 1)
6347e900 EB	227	break;
	228	schedule();
	229	}
af4b8a83	230	__set_current_state(TASK_RUNNING);
6347e900	231
cf3f8921 DL	232	if (pid_ns->reboot)
	233	current->signal->group_exit_code = pid_ns->reboot;
	234
0b6b030f	235	acct_exit_ns(pid_ns);
74bd59bb PE	236	return;
	237	}
	238
98ed57ee	239	#ifdef CONFIG_CHECKPOINT_RESTORE
b8f566b0 PE	240	static int pid_ns_ctl_handler(struct ctl_table *table, int write,
	241	void __user buffer, size_t lenp, loff_t *ppos)
	242	{
49f4d8b9	243	struct pid_namespace *pid_ns = task_active_pid_ns(current);
b8f566b0 PE	244	struct ctl_table tmp = *table;
b8f566b0 PE	245
49f4d8b9	246	if (write && !ns_capable(pid_ns->user_ns, CAP_SYS_ADMIN))
b8f566b0 PE	247	return -EPERM;
	248
	249	/*
	250	* Writing directly to ns' last_pid field is OK, since this field
	251	* is volatile in a living namespace anyway and a code writing to
	252	* it should synchronize its usage with external means.
	253	*/
	254
49f4d8b9	255	tmp.data = &pid_ns->last_pid;
579035dc	256	return proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
b8f566b0 PE	257	}
b8f566b0 PE	258
579035dc AV	259	extern int pid_max;
579035dc AV	260	static int zero = 0;
b8f566b0 PE	261	static struct ctl_table pid_ns_ctl_table[] = {
	262	{
	263	.procname = "ns_last_pid",
	264	.maxlen = sizeof(int),
	265	.mode = 0666, /* permissions are checked in the handler */
	266	.proc_handler = pid_ns_ctl_handler,
579035dc AV	267	.extra1 = &zero,
579035dc AV	268	.extra2 = &pid_max,
b8f566b0 PE	269	},
	270	{ }
	271	};
b8f566b0	272	static struct ctl_path kern_path[] = { { .procname = "kernel", }, { } };
98ed57ee	273	#endif /* CONFIG_CHECKPOINT_RESTORE */
b8f566b0	274
cf3f8921 DL	275	int reboot_pid_ns(struct pid_namespace *pid_ns, int cmd)
	276	{
	277	if (pid_ns == &init_pid_ns)
	278	return 0;
	279
	280	switch (cmd) {
	281	case LINUX_REBOOT_CMD_RESTART2:
	282	case LINUX_REBOOT_CMD_RESTART:
	283	pid_ns->reboot = SIGHUP;
	284	break;
	285
	286	case LINUX_REBOOT_CMD_POWER_OFF:
	287	case LINUX_REBOOT_CMD_HALT:
	288	pid_ns->reboot = SIGINT;
	289	break;
	290	default:
	291	return -EINVAL;
	292	}
	293
	294	read_lock(&tasklist_lock);
	295	force_sig(SIGKILL, pid_ns->child_reaper);
	296	read_unlock(&tasklist_lock);
	297
	298	do_exit(0);
	299
	300	/* Not reached */
	301	return 0;
	302	}
	303
74bd59bb PE	304	static __init int pid_namespaces_init(void)
	305	{
	306	pid_ns_cachep = KMEM_CACHE(pid_namespace, SLAB_PANIC);
98ed57ee CG	307
98ed57ee CG	308	#ifdef CONFIG_CHECKPOINT_RESTORE
b8f566b0	309	register_sysctl_paths(kern_path, pid_ns_ctl_table);
98ed57ee	310	#endif
74bd59bb PE	311	return 0;
	312	}
	313
	314	__initcall(pid_namespaces_init);