]>
Commit | Line | Data |
---|---|---|
457c8996 | 1 | // SPDX-License-Identifier: GPL-2.0-only |
1da177e4 LT |
2 | /* |
3 | * Generic pidhash and scalable, time-bounded PID allocator | |
4 | * | |
6d49e352 NYC |
5 | * (C) 2002-2003 Nadia Yvette Chambers, IBM |
6 | * (C) 2004 Nadia Yvette Chambers, Oracle | |
1da177e4 LT |
7 | * (C) 2002-2004 Ingo Molnar, Red Hat |
8 | * | |
9 | * pid-structures are backing objects for tasks sharing a given ID to chain | |
10 | * against. There is very little to them aside from hashing them and | |
11 | * parking tasks using given ID's on a list. | |
12 | * | |
13 | * The hash is always changed with the tasklist_lock write-acquired, | |
14 | * and the hash is only accessed with the tasklist_lock at least | |
15 | * read-acquired, so there's no additional SMP locking needed here. | |
16 | * | |
17 | * We have a list of bitmap pages, which bitmaps represent the PID space. | |
18 | * Allocating and freeing PIDs is completely lockless. The worst-case | |
19 | * allocation scenario when all but one out of 1 million PIDs possible are | |
20 | * allocated already: the scanning of 32 list entries and at most PAGE_SIZE | |
21 | * bytes. The typical fastpath is a single successful setbit. Freeing is O(1). | |
30e49c26 PE |
22 | * |
23 | * Pid namespaces: | |
24 | * (C) 2007 Pavel Emelyanov <xemul@openvz.org>, OpenVZ, SWsoft Inc. | |
25 | * (C) 2007 Sukadev Bhattiprolu <sukadev@us.ibm.com>, IBM | |
26 | * Many thanks to Oleg Nesterov for comments and help | |
27 | * | |
1da177e4 LT |
28 | */ |
29 | ||
30 | #include <linux/mm.h> | |
9984de1a | 31 | #include <linux/export.h> |
1da177e4 LT |
32 | #include <linux/slab.h> |
33 | #include <linux/init.h> | |
82524746 | 34 | #include <linux/rculist.h> |
57c8a661 | 35 | #include <linux/memblock.h> |
61a58c6c | 36 | #include <linux/pid_namespace.h> |
820e45db | 37 | #include <linux/init_task.h> |
3eb07c8c | 38 | #include <linux/syscalls.h> |
0bb80f24 | 39 | #include <linux/proc_ns.h> |
f57e515a | 40 | #include <linux/refcount.h> |
32fcb426 CB |
41 | #include <linux/anon_inodes.h> |
42 | #include <linux/sched/signal.h> | |
29930025 | 43 | #include <linux/sched/task.h> |
95846ecf | 44 | #include <linux/idr.h> |
1da177e4 | 45 | |
e1e871af | 46 | struct pid init_struct_pid = { |
f57e515a | 47 | .count = REFCOUNT_INIT(1), |
e1e871af DH |
48 | .tasks = { |
49 | { .first = NULL }, | |
50 | { .first = NULL }, | |
51 | { .first = NULL }, | |
52 | }, | |
53 | .level = 0, | |
54 | .numbers = { { | |
55 | .nr = 0, | |
56 | .ns = &init_pid_ns, | |
57 | }, } | |
58 | }; | |
1da177e4 LT |
59 | |
60 | int pid_max = PID_MAX_DEFAULT; | |
1da177e4 LT |
61 | |
62 | #define RESERVED_PIDS 300 | |
63 | ||
64 | int pid_max_min = RESERVED_PIDS + 1; | |
65 | int pid_max_max = PID_MAX_LIMIT; | |
66 | ||
1da177e4 LT |
67 | /* |
68 | * PID-map pages start out as NULL, they get allocated upon | |
69 | * first use and are never deallocated. This way a low pid_max | |
70 | * value does not cause lots of bitmaps to be allocated, but | |
71 | * the scheme scales to up to 4 million PIDs, runtime. | |
72 | */ | |
61a58c6c | 73 | struct pid_namespace init_pid_ns = { |
1e24edca | 74 | .kref = KREF_INIT(2), |
f6bb2a2c | 75 | .idr = IDR_INIT(init_pid_ns.idr), |
e8cfbc24 | 76 | .pid_allocated = PIDNS_ADDING, |
faacbfd3 PE |
77 | .level = 0, |
78 | .child_reaper = &init_task, | |
49f4d8b9 | 79 | .user_ns = &init_user_ns, |
435d5f4b | 80 | .ns.inum = PROC_PID_INIT_INO, |
33c42940 AV |
81 | #ifdef CONFIG_PID_NS |
82 | .ns.ops = &pidns_operations, | |
83 | #endif | |
3fbc9648 | 84 | }; |
198fe21b | 85 | EXPORT_SYMBOL_GPL(init_pid_ns); |
1da177e4 | 86 | |
92476d7f EB |
87 | /* |
88 | * Note: disable interrupts while the pidmap_lock is held as an | |
89 | * interrupt might come in and do read_lock(&tasklist_lock). | |
90 | * | |
91 | * If we don't disable interrupts there is a nasty deadlock between | |
92 | * detach_pid()->free_pid() and another cpu that does | |
93 | * spin_lock(&pidmap_lock) followed by an interrupt routine that does | |
94 | * read_lock(&tasklist_lock); | |
95 | * | |
96 | * After we clean up the tasklist_lock and know there are no | |
97 | * irq handlers that take it we can leave the interrupts enabled. | |
98 | * For now it is easier to be safe than to prove it can't happen. | |
99 | */ | |
3fbc9648 | 100 | |
1da177e4 LT |
101 | static __cacheline_aligned_in_smp DEFINE_SPINLOCK(pidmap_lock); |
102 | ||
7ad5b3a5 | 103 | void put_pid(struct pid *pid) |
92476d7f | 104 | { |
baf8f0f8 PE |
105 | struct pid_namespace *ns; |
106 | ||
92476d7f EB |
107 | if (!pid) |
108 | return; | |
baf8f0f8 | 109 | |
8ef047aa | 110 | ns = pid->numbers[pid->level].ns; |
f57e515a | 111 | if (refcount_dec_and_test(&pid->count)) { |
baf8f0f8 | 112 | kmem_cache_free(ns->pid_cachep, pid); |
b461cc03 | 113 | put_pid_ns(ns); |
8ef047aa | 114 | } |
92476d7f | 115 | } |
bbf73147 | 116 | EXPORT_SYMBOL_GPL(put_pid); |
92476d7f EB |
117 | |
118 | static void delayed_put_pid(struct rcu_head *rhp) | |
119 | { | |
120 | struct pid *pid = container_of(rhp, struct pid, rcu); | |
121 | put_pid(pid); | |
122 | } | |
123 | ||
7ad5b3a5 | 124 | void free_pid(struct pid *pid) |
92476d7f EB |
125 | { |
126 | /* We can be called with write_lock_irq(&tasklist_lock) held */ | |
8ef047aa | 127 | int i; |
92476d7f EB |
128 | unsigned long flags; |
129 | ||
130 | spin_lock_irqsave(&pidmap_lock, flags); | |
0a01f2cc EB |
131 | for (i = 0; i <= pid->level; i++) { |
132 | struct upid *upid = pid->numbers + i; | |
af4b8a83 | 133 | struct pid_namespace *ns = upid->ns; |
e8cfbc24 | 134 | switch (--ns->pid_allocated) { |
a6064885 | 135 | case 2: |
af4b8a83 EB |
136 | case 1: |
137 | /* When all that is left in the pid namespace | |
138 | * is the reaper wake up the reaper. The reaper | |
139 | * may be sleeping in zap_pid_ns_processes(). | |
140 | */ | |
141 | wake_up_process(ns->child_reaper); | |
142 | break; | |
e8cfbc24 | 143 | case PIDNS_ADDING: |
314a8ad0 ON |
144 | /* Handle a fork failure of the first process */ |
145 | WARN_ON(ns->child_reaper); | |
e8cfbc24 | 146 | ns->pid_allocated = 0; |
314a8ad0 | 147 | /* fall through */ |
af4b8a83 | 148 | case 0: |
af4b8a83 EB |
149 | schedule_work(&ns->proc_work); |
150 | break; | |
5e1182de | 151 | } |
95846ecf GS |
152 | |
153 | idr_remove(&ns->idr, upid->nr); | |
0a01f2cc | 154 | } |
92476d7f EB |
155 | spin_unlock_irqrestore(&pidmap_lock, flags); |
156 | ||
92476d7f EB |
157 | call_rcu(&pid->rcu, delayed_put_pid); |
158 | } | |
159 | ||
49cb2fc4 AR |
160 | struct pid *alloc_pid(struct pid_namespace *ns, pid_t *set_tid, |
161 | size_t set_tid_size) | |
92476d7f EB |
162 | { |
163 | struct pid *pid; | |
164 | enum pid_type type; | |
8ef047aa PE |
165 | int i, nr; |
166 | struct pid_namespace *tmp; | |
198fe21b | 167 | struct upid *upid; |
35f71bc0 | 168 | int retval = -ENOMEM; |
92476d7f | 169 | |
49cb2fc4 AR |
170 | /* |
171 | * set_tid_size contains the size of the set_tid array. Starting at | |
172 | * the most nested currently active PID namespace it tells alloc_pid() | |
173 | * which PID to set for a process in that most nested PID namespace | |
174 | * up to set_tid_size PID namespaces. It does not have to set the PID | |
175 | * for a process in all nested PID namespaces but set_tid_size must | |
176 | * never be greater than the current ns->level + 1. | |
177 | */ | |
178 | if (set_tid_size > ns->level + 1) | |
179 | return ERR_PTR(-EINVAL); | |
180 | ||
baf8f0f8 | 181 | pid = kmem_cache_alloc(ns->pid_cachep, GFP_KERNEL); |
92476d7f | 182 | if (!pid) |
35f71bc0 | 183 | return ERR_PTR(retval); |
92476d7f | 184 | |
8ef047aa | 185 | tmp = ns; |
0a01f2cc | 186 | pid->level = ns->level; |
95846ecf | 187 | |
8ef047aa | 188 | for (i = ns->level; i >= 0; i--) { |
49cb2fc4 AR |
189 | int tid = 0; |
190 | ||
191 | if (set_tid_size) { | |
192 | tid = set_tid[ns->level - i]; | |
193 | ||
194 | retval = -EINVAL; | |
195 | if (tid < 1 || tid >= pid_max) | |
196 | goto out_free; | |
197 | /* | |
198 | * Also fail if a PID != 1 is requested and | |
199 | * no PID 1 exists. | |
200 | */ | |
201 | if (tid != 1 && !tmp->child_reaper) | |
202 | goto out_free; | |
203 | retval = -EPERM; | |
204 | if (!ns_capable(tmp->user_ns, CAP_SYS_ADMIN)) | |
205 | goto out_free; | |
206 | set_tid_size--; | |
207 | } | |
95846ecf GS |
208 | |
209 | idr_preload(GFP_KERNEL); | |
210 | spin_lock_irq(&pidmap_lock); | |
211 | ||
49cb2fc4 AR |
212 | if (tid) { |
213 | nr = idr_alloc(&tmp->idr, NULL, tid, | |
214 | tid + 1, GFP_ATOMIC); | |
215 | /* | |
216 | * If ENOSPC is returned it means that the PID is | |
217 | * alreay in use. Return EEXIST in that case. | |
218 | */ | |
219 | if (nr == -ENOSPC) | |
220 | nr = -EEXIST; | |
221 | } else { | |
222 | int pid_min = 1; | |
223 | /* | |
224 | * init really needs pid 1, but after reaching the | |
225 | * maximum wrap back to RESERVED_PIDS | |
226 | */ | |
227 | if (idr_get_cursor(&tmp->idr) > RESERVED_PIDS) | |
228 | pid_min = RESERVED_PIDS; | |
229 | ||
230 | /* | |
231 | * Store a null pointer so find_pid_ns does not find | |
232 | * a partially initialized PID (see below). | |
233 | */ | |
234 | nr = idr_alloc_cyclic(&tmp->idr, NULL, pid_min, | |
235 | pid_max, GFP_ATOMIC); | |
236 | } | |
95846ecf GS |
237 | spin_unlock_irq(&pidmap_lock); |
238 | idr_preload_end(); | |
239 | ||
287980e4 | 240 | if (nr < 0) { |
f83606f5 | 241 | retval = (nr == -ENOSPC) ? -EAGAIN : nr; |
8ef047aa | 242 | goto out_free; |
35f71bc0 | 243 | } |
92476d7f | 244 | |
8ef047aa PE |
245 | pid->numbers[i].nr = nr; |
246 | pid->numbers[i].ns = tmp; | |
247 | tmp = tmp->parent; | |
248 | } | |
249 | ||
0a01f2cc | 250 | if (unlikely(is_child_reaper(pid))) { |
c0ee5549 | 251 | if (pid_ns_prepare_proc(ns)) |
0a01f2cc EB |
252 | goto out_free; |
253 | } | |
254 | ||
b461cc03 | 255 | get_pid_ns(ns); |
f57e515a | 256 | refcount_set(&pid->count, 1); |
92476d7f EB |
257 | for (type = 0; type < PIDTYPE_MAX; ++type) |
258 | INIT_HLIST_HEAD(&pid->tasks[type]); | |
259 | ||
b53b0b9d JFG |
260 | init_waitqueue_head(&pid->wait_pidfd); |
261 | ||
417e3152 | 262 | upid = pid->numbers + ns->level; |
92476d7f | 263 | spin_lock_irq(&pidmap_lock); |
e8cfbc24 | 264 | if (!(ns->pid_allocated & PIDNS_ADDING)) |
5e1182de | 265 | goto out_unlock; |
0a01f2cc | 266 | for ( ; upid >= pid->numbers; --upid) { |
95846ecf GS |
267 | /* Make the PID visible to find_pid_ns. */ |
268 | idr_replace(&upid->ns->idr, pid, upid->nr); | |
e8cfbc24 | 269 | upid->ns->pid_allocated++; |
0a01f2cc | 270 | } |
92476d7f EB |
271 | spin_unlock_irq(&pidmap_lock); |
272 | ||
92476d7f EB |
273 | return pid; |
274 | ||
5e1182de | 275 | out_unlock: |
6e666884 | 276 | spin_unlock_irq(&pidmap_lock); |
24c037eb ON |
277 | put_pid_ns(ns); |
278 | ||
92476d7f | 279 | out_free: |
95846ecf | 280 | spin_lock_irq(&pidmap_lock); |
1a80dade MW |
281 | while (++i <= ns->level) { |
282 | upid = pid->numbers + i; | |
283 | idr_remove(&upid->ns->idr, upid->nr); | |
284 | } | |
95846ecf | 285 | |
c0ee5549 EB |
286 | /* On failure to allocate the first pid, reset the state */ |
287 | if (ns->pid_allocated == PIDNS_ADDING) | |
288 | idr_set_cursor(&ns->idr, 0); | |
289 | ||
95846ecf | 290 | spin_unlock_irq(&pidmap_lock); |
8ef047aa | 291 | |
baf8f0f8 | 292 | kmem_cache_free(ns->pid_cachep, pid); |
35f71bc0 | 293 | return ERR_PTR(retval); |
92476d7f EB |
294 | } |
295 | ||
c876ad76 EB |
296 | void disable_pid_allocation(struct pid_namespace *ns) |
297 | { | |
298 | spin_lock_irq(&pidmap_lock); | |
e8cfbc24 | 299 | ns->pid_allocated &= ~PIDNS_ADDING; |
c876ad76 EB |
300 | spin_unlock_irq(&pidmap_lock); |
301 | } | |
302 | ||
7ad5b3a5 | 303 | struct pid *find_pid_ns(int nr, struct pid_namespace *ns) |
1da177e4 | 304 | { |
e8cfbc24 | 305 | return idr_find(&ns->idr, nr); |
1da177e4 | 306 | } |
198fe21b | 307 | EXPORT_SYMBOL_GPL(find_pid_ns); |
1da177e4 | 308 | |
8990571e PE |
309 | struct pid *find_vpid(int nr) |
310 | { | |
17cf22c3 | 311 | return find_pid_ns(nr, task_active_pid_ns(current)); |
8990571e PE |
312 | } |
313 | EXPORT_SYMBOL_GPL(find_vpid); | |
314 | ||
2c470475 EB |
315 | static struct pid **task_pid_ptr(struct task_struct *task, enum pid_type type) |
316 | { | |
317 | return (type == PIDTYPE_PID) ? | |
318 | &task->thread_pid : | |
2c470475 EB |
319 | &task->signal->pids[type]; |
320 | } | |
321 | ||
e713d0da SB |
322 | /* |
323 | * attach_pid() must be called with the tasklist_lock write-held. | |
324 | */ | |
81907739 | 325 | void attach_pid(struct task_struct *task, enum pid_type type) |
1da177e4 | 326 | { |
2c470475 EB |
327 | struct pid *pid = *task_pid_ptr(task, type); |
328 | hlist_add_head_rcu(&task->pid_links[type], &pid->tasks[type]); | |
1da177e4 LT |
329 | } |
330 | ||
24336eae ON |
331 | static void __change_pid(struct task_struct *task, enum pid_type type, |
332 | struct pid *new) | |
1da177e4 | 333 | { |
2c470475 | 334 | struct pid **pid_ptr = task_pid_ptr(task, type); |
92476d7f EB |
335 | struct pid *pid; |
336 | int tmp; | |
1da177e4 | 337 | |
2c470475 | 338 | pid = *pid_ptr; |
1da177e4 | 339 | |
2c470475 EB |
340 | hlist_del_rcu(&task->pid_links[type]); |
341 | *pid_ptr = new; | |
1da177e4 | 342 | |
92476d7f | 343 | for (tmp = PIDTYPE_MAX; --tmp >= 0; ) |
1d416a11 | 344 | if (pid_has_task(pid, tmp)) |
92476d7f | 345 | return; |
1da177e4 | 346 | |
92476d7f | 347 | free_pid(pid); |
1da177e4 LT |
348 | } |
349 | ||
24336eae ON |
350 | void detach_pid(struct task_struct *task, enum pid_type type) |
351 | { | |
352 | __change_pid(task, type, NULL); | |
353 | } | |
354 | ||
355 | void change_pid(struct task_struct *task, enum pid_type type, | |
356 | struct pid *pid) | |
357 | { | |
358 | __change_pid(task, type, pid); | |
81907739 | 359 | attach_pid(task, type); |
24336eae ON |
360 | } |
361 | ||
c18258c6 | 362 | /* transfer_pid is an optimization of attach_pid(new), detach_pid(old) */ |
7ad5b3a5 | 363 | void transfer_pid(struct task_struct *old, struct task_struct *new, |
c18258c6 EB |
364 | enum pid_type type) |
365 | { | |
2c470475 EB |
366 | if (type == PIDTYPE_PID) |
367 | new->thread_pid = old->thread_pid; | |
368 | hlist_replace_rcu(&old->pid_links[type], &new->pid_links[type]); | |
c18258c6 EB |
369 | } |
370 | ||
7ad5b3a5 | 371 | struct task_struct *pid_task(struct pid *pid, enum pid_type type) |
1da177e4 | 372 | { |
92476d7f EB |
373 | struct task_struct *result = NULL; |
374 | if (pid) { | |
375 | struct hlist_node *first; | |
67bdbffd | 376 | first = rcu_dereference_check(hlist_first_rcu(&pid->tasks[type]), |
db1466b3 | 377 | lockdep_tasklist_lock_is_held()); |
92476d7f | 378 | if (first) |
2c470475 | 379 | result = hlist_entry(first, struct task_struct, pid_links[(type)]); |
92476d7f EB |
380 | } |
381 | return result; | |
382 | } | |
eccba068 | 383 | EXPORT_SYMBOL(pid_task); |
1da177e4 | 384 | |
92476d7f | 385 | /* |
9728e5d6 | 386 | * Must be called under rcu_read_lock(). |
92476d7f | 387 | */ |
17f98dcf | 388 | struct task_struct *find_task_by_pid_ns(pid_t nr, struct pid_namespace *ns) |
92476d7f | 389 | { |
f78f5b90 PM |
390 | RCU_LOCKDEP_WARN(!rcu_read_lock_held(), |
391 | "find_task_by_pid_ns() needs rcu_read_lock() protection"); | |
17f98dcf | 392 | return pid_task(find_pid_ns(nr, ns), PIDTYPE_PID); |
92476d7f | 393 | } |
1da177e4 | 394 | |
228ebcbe PE |
395 | struct task_struct *find_task_by_vpid(pid_t vnr) |
396 | { | |
17cf22c3 | 397 | return find_task_by_pid_ns(vnr, task_active_pid_ns(current)); |
228ebcbe | 398 | } |
228ebcbe | 399 | |
2ee08260 MR |
400 | struct task_struct *find_get_task_by_vpid(pid_t nr) |
401 | { | |
402 | struct task_struct *task; | |
403 | ||
404 | rcu_read_lock(); | |
405 | task = find_task_by_vpid(nr); | |
406 | if (task) | |
407 | get_task_struct(task); | |
408 | rcu_read_unlock(); | |
409 | ||
410 | return task; | |
411 | } | |
412 | ||
1a657f78 ON |
413 | struct pid *get_task_pid(struct task_struct *task, enum pid_type type) |
414 | { | |
415 | struct pid *pid; | |
416 | rcu_read_lock(); | |
2c470475 | 417 | pid = get_pid(rcu_dereference(*task_pid_ptr(task, type))); |
1a657f78 ON |
418 | rcu_read_unlock(); |
419 | return pid; | |
420 | } | |
77c100c8 | 421 | EXPORT_SYMBOL_GPL(get_task_pid); |
1a657f78 | 422 | |
7ad5b3a5 | 423 | struct task_struct *get_pid_task(struct pid *pid, enum pid_type type) |
92476d7f EB |
424 | { |
425 | struct task_struct *result; | |
426 | rcu_read_lock(); | |
427 | result = pid_task(pid, type); | |
428 | if (result) | |
429 | get_task_struct(result); | |
430 | rcu_read_unlock(); | |
431 | return result; | |
1da177e4 | 432 | } |
77c100c8 | 433 | EXPORT_SYMBOL_GPL(get_pid_task); |
1da177e4 | 434 | |
92476d7f | 435 | struct pid *find_get_pid(pid_t nr) |
1da177e4 LT |
436 | { |
437 | struct pid *pid; | |
438 | ||
92476d7f | 439 | rcu_read_lock(); |
198fe21b | 440 | pid = get_pid(find_vpid(nr)); |
92476d7f | 441 | rcu_read_unlock(); |
1da177e4 | 442 | |
92476d7f | 443 | return pid; |
1da177e4 | 444 | } |
339caf2a | 445 | EXPORT_SYMBOL_GPL(find_get_pid); |
1da177e4 | 446 | |
7af57294 PE |
447 | pid_t pid_nr_ns(struct pid *pid, struct pid_namespace *ns) |
448 | { | |
449 | struct upid *upid; | |
450 | pid_t nr = 0; | |
451 | ||
452 | if (pid && ns->level <= pid->level) { | |
453 | upid = &pid->numbers[ns->level]; | |
454 | if (upid->ns == ns) | |
455 | nr = upid->nr; | |
456 | } | |
457 | return nr; | |
458 | } | |
4f82f457 | 459 | EXPORT_SYMBOL_GPL(pid_nr_ns); |
7af57294 | 460 | |
44c4e1b2 EB |
461 | pid_t pid_vnr(struct pid *pid) |
462 | { | |
17cf22c3 | 463 | return pid_nr_ns(pid, task_active_pid_ns(current)); |
44c4e1b2 EB |
464 | } |
465 | EXPORT_SYMBOL_GPL(pid_vnr); | |
466 | ||
52ee2dfd ON |
467 | pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type, |
468 | struct pid_namespace *ns) | |
2f2a3a46 | 469 | { |
52ee2dfd ON |
470 | pid_t nr = 0; |
471 | ||
472 | rcu_read_lock(); | |
473 | if (!ns) | |
17cf22c3 | 474 | ns = task_active_pid_ns(current); |
2c470475 EB |
475 | if (likely(pid_alive(task))) |
476 | nr = pid_nr_ns(rcu_dereference(*task_pid_ptr(task, type)), ns); | |
52ee2dfd ON |
477 | rcu_read_unlock(); |
478 | ||
479 | return nr; | |
2f2a3a46 | 480 | } |
52ee2dfd | 481 | EXPORT_SYMBOL(__task_pid_nr_ns); |
2f2a3a46 | 482 | |
61bce0f1 EB |
483 | struct pid_namespace *task_active_pid_ns(struct task_struct *tsk) |
484 | { | |
485 | return ns_of_pid(task_pid(tsk)); | |
486 | } | |
487 | EXPORT_SYMBOL_GPL(task_active_pid_ns); | |
488 | ||
0804ef4b | 489 | /* |
025dfdaf | 490 | * Used by proc to find the first pid that is greater than or equal to nr. |
0804ef4b | 491 | * |
e49859e7 | 492 | * If there is a pid at nr this function is exactly the same as find_pid_ns. |
0804ef4b | 493 | */ |
198fe21b | 494 | struct pid *find_ge_pid(int nr, struct pid_namespace *ns) |
0804ef4b | 495 | { |
95846ecf | 496 | return idr_get_next(&ns->idr, &nr); |
0804ef4b EB |
497 | } |
498 | ||
32fcb426 CB |
499 | /** |
500 | * pidfd_create() - Create a new pid file descriptor. | |
501 | * | |
502 | * @pid: struct pid that the pidfd will reference | |
503 | * | |
504 | * This creates a new pid file descriptor with the O_CLOEXEC flag set. | |
505 | * | |
506 | * Note, that this function can only be called after the fd table has | |
507 | * been unshared to avoid leaking the pidfd to the new process. | |
508 | * | |
509 | * Return: On success, a cloexec pidfd is returned. | |
510 | * On error, a negative errno number will be returned. | |
511 | */ | |
512 | static int pidfd_create(struct pid *pid) | |
513 | { | |
514 | int fd; | |
515 | ||
516 | fd = anon_inode_getfd("[pidfd]", &pidfd_fops, get_pid(pid), | |
517 | O_RDWR | O_CLOEXEC); | |
518 | if (fd < 0) | |
519 | put_pid(pid); | |
520 | ||
521 | return fd; | |
522 | } | |
523 | ||
524 | /** | |
525 | * pidfd_open() - Open new pid file descriptor. | |
526 | * | |
527 | * @pid: pid for which to retrieve a pidfd | |
528 | * @flags: flags to pass | |
529 | * | |
530 | * This creates a new pid file descriptor with the O_CLOEXEC flag set for | |
531 | * the process identified by @pid. Currently, the process identified by | |
532 | * @pid must be a thread-group leader. This restriction currently exists | |
533 | * for all aspects of pidfds including pidfd creation (CLONE_PIDFD cannot | |
534 | * be used with CLONE_THREAD) and pidfd polling (only supports thread group | |
535 | * leaders). | |
536 | * | |
537 | * Return: On success, a cloexec pidfd is returned. | |
538 | * On error, a negative errno number will be returned. | |
539 | */ | |
540 | SYSCALL_DEFINE2(pidfd_open, pid_t, pid, unsigned int, flags) | |
541 | { | |
1e1d0f0b | 542 | int fd; |
32fcb426 CB |
543 | struct pid *p; |
544 | ||
545 | if (flags) | |
546 | return -EINVAL; | |
547 | ||
548 | if (pid <= 0) | |
549 | return -EINVAL; | |
550 | ||
551 | p = find_get_pid(pid); | |
552 | if (!p) | |
553 | return -ESRCH; | |
554 | ||
1e1d0f0b CB |
555 | if (pid_has_task(p, PIDTYPE_TGID)) |
556 | fd = pidfd_create(p); | |
557 | else | |
558 | fd = -EINVAL; | |
32fcb426 | 559 | |
32fcb426 CB |
560 | put_pid(p); |
561 | return fd; | |
562 | } | |
563 | ||
95846ecf | 564 | void __init pid_idr_init(void) |
1da177e4 | 565 | { |
840d6fe7 | 566 | /* Verify no one has done anything silly: */ |
e8cfbc24 | 567 | BUILD_BUG_ON(PID_MAX_LIMIT >= PIDNS_ADDING); |
c876ad76 | 568 | |
72680a19 HB |
569 | /* bump default and minimum pid_max based on number of cpus */ |
570 | pid_max = min(pid_max_max, max_t(int, pid_max, | |
571 | PIDS_PER_CPU_DEFAULT * num_possible_cpus())); | |
572 | pid_max_min = max_t(int, pid_max_min, | |
573 | PIDS_PER_CPU_MIN * num_possible_cpus()); | |
574 | pr_info("pid_max: default: %u minimum: %u\n", pid_max, pid_max_min); | |
575 | ||
95846ecf | 576 | idr_init(&init_pid_ns.idr); |
92476d7f | 577 | |
74bd59bb | 578 | init_pid_ns.pid_cachep = KMEM_CACHE(pid, |
5d097056 | 579 | SLAB_HWCACHE_ALIGN | SLAB_PANIC | SLAB_ACCOUNT); |
1da177e4 | 580 | } |
8649c322 SD |
581 | |
582 | static struct file *__pidfd_fget(struct task_struct *task, int fd) | |
583 | { | |
584 | struct file *file; | |
585 | int ret; | |
586 | ||
587 | ret = mutex_lock_killable(&task->signal->cred_guard_mutex); | |
588 | if (ret) | |
589 | return ERR_PTR(ret); | |
590 | ||
591 | if (ptrace_may_access(task, PTRACE_MODE_ATTACH_REALCREDS)) | |
592 | file = fget_task(task, fd); | |
593 | else | |
594 | file = ERR_PTR(-EPERM); | |
595 | ||
596 | mutex_unlock(&task->signal->cred_guard_mutex); | |
597 | ||
598 | return file ?: ERR_PTR(-EBADF); | |
599 | } | |
600 | ||
601 | static int pidfd_getfd(struct pid *pid, int fd) | |
602 | { | |
603 | struct task_struct *task; | |
604 | struct file *file; | |
605 | int ret; | |
606 | ||
607 | task = get_pid_task(pid, PIDTYPE_PID); | |
608 | if (!task) | |
609 | return -ESRCH; | |
610 | ||
611 | file = __pidfd_fget(task, fd); | |
612 | put_task_struct(task); | |
613 | if (IS_ERR(file)) | |
614 | return PTR_ERR(file); | |
615 | ||
616 | ret = security_file_receive(file); | |
617 | if (ret) { | |
618 | fput(file); | |
619 | return ret; | |
620 | } | |
621 | ||
622 | ret = get_unused_fd_flags(O_CLOEXEC); | |
623 | if (ret < 0) | |
624 | fput(file); | |
625 | else | |
626 | fd_install(ret, file); | |
627 | ||
628 | return ret; | |
629 | } | |
630 | ||
631 | /** | |
632 | * sys_pidfd_getfd() - Get a file descriptor from another process | |
633 | * | |
634 | * @pidfd: the pidfd file descriptor of the process | |
635 | * @fd: the file descriptor number to get | |
636 | * @flags: flags on how to get the fd (reserved) | |
637 | * | |
638 | * This syscall gets a copy of a file descriptor from another process | |
639 | * based on the pidfd, and file descriptor number. It requires that | |
640 | * the calling process has the ability to ptrace the process represented | |
641 | * by the pidfd. The process which is having its file descriptor copied | |
642 | * is otherwise unaffected. | |
643 | * | |
644 | * Return: On success, a cloexec file descriptor is returned. | |
645 | * On error, a negative errno number will be returned. | |
646 | */ | |
647 | SYSCALL_DEFINE3(pidfd_getfd, int, pidfd, int, fd, | |
648 | unsigned int, flags) | |
649 | { | |
650 | struct pid *pid; | |
651 | struct fd f; | |
652 | int ret; | |
653 | ||
654 | /* flags is currently unused - make sure it's unset */ | |
655 | if (flags) | |
656 | return -EINVAL; | |
657 | ||
658 | f = fdget(pidfd); | |
659 | if (!f.file) | |
660 | return -EBADF; | |
661 | ||
662 | pid = pidfd_pid(f.file); | |
663 | if (IS_ERR(pid)) | |
664 | ret = PTR_ERR(pid); | |
665 | else | |
666 | ret = pidfd_getfd(pid, fd); | |
667 | ||
668 | fdput(f); | |
669 | return ret; | |
670 | } |