[linux-2.6-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/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 struct pid_namespace *create_pid_namespace(struct pid_namespace *parent_pid_ns)
{
	struct pid_namespace *ns;
	unsigned int level = parent_pid_ns->level + 1;
	int i, 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);

	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);

	err = pid_ns_prepare_proc(ns);
	if (err)
		goto out_put_parent_pid_ns;

	return ns;

out_put_parent_pid_ns:
	put_pid_ns(parent_pid_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);
	kmem_cache_free(pid_ns_cachep, ns);
}

struct pid_namespace *copy_pid_ns(unsigned long flags, 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);
}

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 __unhash_process().
	 */
	for (;;) {
		bool need_wait = false;

		read_lock(&tasklist_lock);
		if (!list_empty(&current->children)) {
			__set_current_state(TASK_UNINTERRUPTIBLE);
			need_wait = true;
		}
		read_unlock(&tasklist_lock);

		if (!need_wait)
			break;
		schedule();
	}

	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 ctl_table tmp = *table;

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