pid.c 10.1 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
/*
 * Generic pidhash and scalable, time-bounded PID allocator
 *
 * (C) 2002-2003 William Irwin, IBM
 * (C) 2004 William Irwin, Oracle
 * (C) 2002-2004 Ingo Molnar, Red Hat
 *
 * pid-structures are backing objects for tasks sharing a given ID to chain
 * against. There is very little to them aside from hashing them and
 * parking tasks using given ID's on a list.
 *
 * The hash is always changed with the tasklist_lock write-acquired,
 * and the hash is only accessed with the tasklist_lock at least
 * read-acquired, so there's no additional SMP locking needed here.
 *
 * We have a list of bitmap pages, which bitmaps represent the PID space.
 * Allocating and freeing PIDs is completely lockless. The worst-case
 * allocation scenario when all but one out of 1 million PIDs possible are
 * allocated already: the scanning of 32 list entries and at most PAGE_SIZE
 * bytes. The typical fastpath is a single successful setbit. Freeing is O(1).
 */

#include <linux/mm.h>
#include <linux/module.h>
#include <linux/slab.h>
#include <linux/init.h>
#include <linux/bootmem.h>
#include <linux/hash.h>
29
#include <linux/pid_namespace.h>
L
Linus Torvalds 已提交
30 31

#define pid_hashfn(nr) hash_long((unsigned long)nr, pidhash_shift)
32
static struct hlist_head *pid_hash;
L
Linus Torvalds 已提交
33
static int pidhash_shift;
34
static struct kmem_cache *pid_cachep;
L
Linus Torvalds 已提交
35 36 37 38 39 40 41 42 43 44

int pid_max = PID_MAX_DEFAULT;

#define RESERVED_PIDS		300

int pid_max_min = RESERVED_PIDS + 1;
int pid_max_max = PID_MAX_LIMIT;

#define BITS_PER_PAGE		(PAGE_SIZE*8)
#define BITS_PER_PAGE_MASK	(BITS_PER_PAGE-1)
S
Sukadev Bhattiprolu 已提交
45

46 47
static inline int mk_pid(struct pid_namespace *pid_ns,
		struct pidmap *map, int off)
S
Sukadev Bhattiprolu 已提交
48
{
49
	return (map - pid_ns->pidmap)*BITS_PER_PAGE + off;
S
Sukadev Bhattiprolu 已提交
50 51
}

L
Linus Torvalds 已提交
52 53 54 55 56 57 58 59 60
#define find_next_offset(map, off)					\
		find_next_zero_bit((map)->page, BITS_PER_PAGE, off)

/*
 * PID-map pages start out as NULL, they get allocated upon
 * first use and are never deallocated. This way a low pid_max
 * value does not cause lots of bitmaps to be allocated, but
 * the scheme scales to up to 4 million PIDs, runtime.
 */
61
struct pid_namespace init_pid_ns = {
C
Cedric Le Goater 已提交
62 63 64
	.kref = {
		.refcount       = ATOMIC_INIT(2),
	},
S
Sukadev Bhattiprolu 已提交
65 66 67
	.pidmap = {
		[ 0 ... PIDMAP_ENTRIES-1] = { ATOMIC_INIT(BITS_PER_PAGE), NULL }
	},
68 69
	.last_pid = 0,
	.child_reaper = &init_task
S
Sukadev Bhattiprolu 已提交
70
};
L
Linus Torvalds 已提交
71

72 73 74 75 76 77 78 79 80 81 82 83 84
/*
 * Note: disable interrupts while the pidmap_lock is held as an
 * interrupt might come in and do read_lock(&tasklist_lock).
 *
 * If we don't disable interrupts there is a nasty deadlock between
 * detach_pid()->free_pid() and another cpu that does
 * spin_lock(&pidmap_lock) followed by an interrupt routine that does
 * read_lock(&tasklist_lock);
 *
 * After we clean up the tasklist_lock and know there are no
 * irq handlers that take it we can leave the interrupts enabled.
 * For now it is easier to be safe than to prove it can't happen.
 */
S
Sukadev Bhattiprolu 已提交
85

L
Linus Torvalds 已提交
86 87
static  __cacheline_aligned_in_smp DEFINE_SPINLOCK(pidmap_lock);

88
static fastcall void free_pidmap(struct pid_namespace *pid_ns, int pid)
L
Linus Torvalds 已提交
89
{
90
	struct pidmap *map = pid_ns->pidmap + pid / BITS_PER_PAGE;
L
Linus Torvalds 已提交
91 92 93 94 95 96
	int offset = pid & BITS_PER_PAGE_MASK;

	clear_bit(offset, map->page);
	atomic_inc(&map->nr_free);
}

97
static int alloc_pidmap(struct pid_namespace *pid_ns)
L
Linus Torvalds 已提交
98
{
99
	int i, offset, max_scan, pid, last = pid_ns->last_pid;
100
	struct pidmap *map;
L
Linus Torvalds 已提交
101 102 103 104 105

	pid = last + 1;
	if (pid >= pid_max)
		pid = RESERVED_PIDS;
	offset = pid & BITS_PER_PAGE_MASK;
106
	map = &pid_ns->pidmap[pid/BITS_PER_PAGE];
L
Linus Torvalds 已提交
107 108 109
	max_scan = (pid_max + BITS_PER_PAGE - 1)/BITS_PER_PAGE - !offset;
	for (i = 0; i <= max_scan; ++i) {
		if (unlikely(!map->page)) {
S
Sukadev Bhattiprolu 已提交
110
			void *page = kzalloc(PAGE_SIZE, GFP_KERNEL);
L
Linus Torvalds 已提交
111 112 113 114
			/*
			 * Free the page if someone raced with us
			 * installing it:
			 */
115
			spin_lock_irq(&pidmap_lock);
L
Linus Torvalds 已提交
116
			if (map->page)
S
Sukadev Bhattiprolu 已提交
117
				kfree(page);
L
Linus Torvalds 已提交
118
			else
S
Sukadev Bhattiprolu 已提交
119
				map->page = page;
120
			spin_unlock_irq(&pidmap_lock);
L
Linus Torvalds 已提交
121 122 123 124 125 126 127
			if (unlikely(!map->page))
				break;
		}
		if (likely(atomic_read(&map->nr_free))) {
			do {
				if (!test_and_set_bit(offset, map->page)) {
					atomic_dec(&map->nr_free);
128
					pid_ns->last_pid = pid;
L
Linus Torvalds 已提交
129 130 131
					return pid;
				}
				offset = find_next_offset(map, offset);
132
				pid = mk_pid(pid_ns, map, offset);
L
Linus Torvalds 已提交
133 134 135 136 137 138 139 140 141 142
			/*
			 * find_next_offset() found a bit, the pid from it
			 * is in-bounds, and if we fell back to the last
			 * bitmap block and the final block was the same
			 * as the starting point, pid is before last_pid.
			 */
			} while (offset < BITS_PER_PAGE && pid < pid_max &&
					(i != max_scan || pid < last ||
					    !((last+1) & BITS_PER_PAGE_MASK)));
		}
143
		if (map < &pid_ns->pidmap[(pid_max-1)/BITS_PER_PAGE]) {
L
Linus Torvalds 已提交
144 145 146
			++map;
			offset = 0;
		} else {
147
			map = &pid_ns->pidmap[0];
L
Linus Torvalds 已提交
148 149 150 151
			offset = RESERVED_PIDS;
			if (unlikely(last == offset))
				break;
		}
152
		pid = mk_pid(pid_ns, map, offset);
L
Linus Torvalds 已提交
153 154 155 156
	}
	return -1;
}

157
static int next_pidmap(struct pid_namespace *pid_ns, int last)
158 159
{
	int offset;
160
	struct pidmap *map, *end;
161 162

	offset = (last + 1) & BITS_PER_PAGE_MASK;
163 164
	map = &pid_ns->pidmap[(last + 1)/BITS_PER_PAGE];
	end = &pid_ns->pidmap[PIDMAP_ENTRIES];
165
	for (; map < end; map++, offset = 0) {
166 167 168 169
		if (unlikely(!map->page))
			continue;
		offset = find_next_bit((map)->page, BITS_PER_PAGE, offset);
		if (offset < BITS_PER_PAGE)
170
			return mk_pid(pid_ns, map, offset);
171 172 173 174
	}
	return -1;
}

175 176 177 178 179 180 181 182
fastcall void put_pid(struct pid *pid)
{
	if (!pid)
		return;
	if ((atomic_read(&pid->count) == 1) ||
	     atomic_dec_and_test(&pid->count))
		kmem_cache_free(pid_cachep, pid);
}
183
EXPORT_SYMBOL_GPL(put_pid);
184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199

static void delayed_put_pid(struct rcu_head *rhp)
{
	struct pid *pid = container_of(rhp, struct pid, rcu);
	put_pid(pid);
}

fastcall void free_pid(struct pid *pid)
{
	/* We can be called with write_lock_irq(&tasklist_lock) held */
	unsigned long flags;

	spin_lock_irqsave(&pidmap_lock, flags);
	hlist_del_rcu(&pid->pid_chain);
	spin_unlock_irqrestore(&pidmap_lock, flags);

200
	free_pidmap(&init_pid_ns, pid->nr);
201 202 203 204 205 206 207 208 209 210 211 212 213
	call_rcu(&pid->rcu, delayed_put_pid);
}

struct pid *alloc_pid(void)
{
	struct pid *pid;
	enum pid_type type;
	int nr = -1;

	pid = kmem_cache_alloc(pid_cachep, GFP_KERNEL);
	if (!pid)
		goto out;

214
	nr = alloc_pidmap(current->nsproxy->pid_ns);
215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236
	if (nr < 0)
		goto out_free;

	atomic_set(&pid->count, 1);
	pid->nr = nr;
	for (type = 0; type < PIDTYPE_MAX; ++type)
		INIT_HLIST_HEAD(&pid->tasks[type]);

	spin_lock_irq(&pidmap_lock);
	hlist_add_head_rcu(&pid->pid_chain, &pid_hash[pid_hashfn(pid->nr)]);
	spin_unlock_irq(&pidmap_lock);

out:
	return pid;

out_free:
	kmem_cache_free(pid_cachep, pid);
	pid = NULL;
	goto out;
}

struct pid * fastcall find_pid(int nr)
L
Linus Torvalds 已提交
237 238 239 240
{
	struct hlist_node *elem;
	struct pid *pid;

I
Ingo Molnar 已提交
241
	hlist_for_each_entry_rcu(pid, elem,
242
			&pid_hash[pid_hashfn(nr)], pid_chain) {
L
Linus Torvalds 已提交
243 244 245 246 247
		if (pid->nr == nr)
			return pid;
	}
	return NULL;
}
248
EXPORT_SYMBOL_GPL(find_pid);
L
Linus Torvalds 已提交
249

250
int fastcall attach_pid(struct task_struct *task, enum pid_type type, int nr)
L
Linus Torvalds 已提交
251
{
252 253 254 255 256 257
	struct pid_link *link;
	struct pid *pid;

	link = &task->pids[type];
	link->pid = pid = find_pid(nr);
	hlist_add_head_rcu(&link->node, &pid->tasks[type]);
L
Linus Torvalds 已提交
258 259 260 261

	return 0;
}

262
void fastcall detach_pid(struct task_struct *task, enum pid_type type)
L
Linus Torvalds 已提交
263
{
264 265 266
	struct pid_link *link;
	struct pid *pid;
	int tmp;
L
Linus Torvalds 已提交
267

268 269
	link = &task->pids[type];
	pid = link->pid;
L
Linus Torvalds 已提交
270

271 272
	hlist_del_rcu(&link->node);
	link->pid = NULL;
L
Linus Torvalds 已提交
273

274 275 276
	for (tmp = PIDTYPE_MAX; --tmp >= 0; )
		if (!hlist_empty(&pid->tasks[tmp]))
			return;
L
Linus Torvalds 已提交
277

278
	free_pid(pid);
L
Linus Torvalds 已提交
279 280
}

281 282 283 284 285 286 287 288 289
/* transfer_pid is an optimization of attach_pid(new), detach_pid(old) */
void fastcall transfer_pid(struct task_struct *old, struct task_struct *new,
			   enum pid_type type)
{
	new->pids[type].pid = old->pids[type].pid;
	hlist_replace_rcu(&old->pids[type].node, &new->pids[type].node);
	old->pids[type].pid = NULL;
}

290
struct task_struct * fastcall pid_task(struct pid *pid, enum pid_type type)
L
Linus Torvalds 已提交
291
{
292 293 294 295 296 297 298 299 300
	struct task_struct *result = NULL;
	if (pid) {
		struct hlist_node *first;
		first = rcu_dereference(pid->tasks[type].first);
		if (first)
			result = hlist_entry(first, struct task_struct, pids[(type)].node);
	}
	return result;
}
L
Linus Torvalds 已提交
301

302 303 304
/*
 * Must be called under rcu_read_lock() or with tasklist_lock read-held.
 */
305
struct task_struct *find_task_by_pid_type(int type, int nr)
306 307 308
{
	return pid_task(find_pid(nr), type);
}
L
Linus Torvalds 已提交
309

310
EXPORT_SYMBOL(find_task_by_pid_type);
L
Linus Torvalds 已提交
311

312 313 314 315 316 317 318 319 320
struct pid *get_task_pid(struct task_struct *task, enum pid_type type)
{
	struct pid *pid;
	rcu_read_lock();
	pid = get_pid(task->pids[type].pid);
	rcu_read_unlock();
	return pid;
}

321 322 323 324 325 326 327 328 329
struct task_struct *fastcall get_pid_task(struct pid *pid, enum pid_type type)
{
	struct task_struct *result;
	rcu_read_lock();
	result = pid_task(pid, type);
	if (result)
		get_task_struct(result);
	rcu_read_unlock();
	return result;
L
Linus Torvalds 已提交
330 331
}

332
struct pid *find_get_pid(pid_t nr)
L
Linus Torvalds 已提交
333 334 335
{
	struct pid *pid;

336 337 338
	rcu_read_lock();
	pid = get_pid(find_pid(nr));
	rcu_read_unlock();
L
Linus Torvalds 已提交
339

340
	return pid;
L
Linus Torvalds 已提交
341 342
}

343 344 345 346 347 348 349 350 351 352 353 354 355
/*
 * Used by proc to find the first pid that is greater then or equal to nr.
 *
 * If there is a pid at nr this function is exactly the same as find_pid.
 */
struct pid *find_ge_pid(int nr)
{
	struct pid *pid;

	do {
		pid = find_pid(nr);
		if (pid)
			break;
356
		nr = next_pidmap(current->nsproxy->pid_ns, nr);
357 358 359 360
	} while (nr > 0);

	return pid;
}
361
EXPORT_SYMBOL_GPL(find_get_pid);
362

C
Cedric Le Goater 已提交
363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382
int copy_pid_ns(int flags, struct task_struct *tsk)
{
	struct pid_namespace *old_ns = tsk->nsproxy->pid_ns;
	int err = 0;

	if (!old_ns)
		return 0;

	get_pid_ns(old_ns);
	return err;
}

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

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

L
Linus Torvalds 已提交
383 384 385 386 387 388 389
/*
 * The pid hash table is scaled according to the amount of memory in the
 * machine.  From a minimum of 16 slots up to 4096 slots at one gigabyte or
 * more.
 */
void __init pidhash_init(void)
{
390
	int i, pidhash_size;
L
Linus Torvalds 已提交
391 392 393 394 395 396 397 398
	unsigned long megabytes = nr_kernel_pages >> (20 - PAGE_SHIFT);

	pidhash_shift = max(4, fls(megabytes * 4));
	pidhash_shift = min(12, pidhash_shift);
	pidhash_size = 1 << pidhash_shift;

	printk("PID hash table entries: %d (order: %d, %Zd bytes)\n",
		pidhash_size, pidhash_shift,
399 400 401 402 403 404 405
		pidhash_size * sizeof(struct hlist_head));

	pid_hash = alloc_bootmem(pidhash_size *	sizeof(*(pid_hash)));
	if (!pid_hash)
		panic("Could not alloc pidhash!\n");
	for (i = 0; i < pidhash_size; i++)
		INIT_HLIST_HEAD(&pid_hash[i]);
L
Linus Torvalds 已提交
406 407 408 409
}

void __init pidmap_init(void)
{
410
	init_pid_ns.pidmap[0].page = kzalloc(PAGE_SIZE, GFP_KERNEL);
411
	/* Reserve PID 0. We never call free_pidmap(0) */
412 413
	set_bit(0, init_pid_ns.pidmap[0].page);
	atomic_dec(&init_pid_ns.pidmap[0].nr_free);
414

415
	pid_cachep = KMEM_CACHE(pid, SLAB_PANIC);
L
Linus Torvalds 已提交
416
}