Blame view
kernel/futex.c
49 KB
1da177e4c
|
1 2 3 4 5 6 7 8 9 10 |
/* * Fast Userspace Mutexes (which I call "Futexes!"). * (C) Rusty Russell, IBM 2002 * * Generalized futexes, futex requeueing, misc fixes by Ingo Molnar * (C) Copyright 2003 Red Hat Inc, All Rights Reserved * * Removed page pinning, fix privately mapped COW pages and other cleanups * (C) Copyright 2003, 2004 Jamie Lokier * |
0771dfefc
|
11 12 13 14 |
* Robust futex support started by Ingo Molnar * (C) Copyright 2006 Red Hat Inc, All Rights Reserved * Thanks to Thomas Gleixner for suggestions, analysis and fixes. * |
c87e2837b
|
15 16 17 18 |
* PI-futex support started by Ingo Molnar and Thomas Gleixner * Copyright (C) 2006 Red Hat, Inc., Ingo Molnar <mingo@redhat.com> * Copyright (C) 2006 Timesys Corp., Thomas Gleixner <tglx@timesys.com> * |
34f01cc1f
|
19 20 21 |
* PRIVATE futexes by Eric Dumazet * Copyright (C) 2007 Eric Dumazet <dada1@cosmosbay.com> * |
1da177e4c
|
22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 |
* Thanks to Ben LaHaise for yelling "hashed waitqueues" loudly * enough at me, Linus for the original (flawed) idea, Matthew * Kirkwood for proof-of-concept implementation. * * "The futexes are also cursed." * "But they come in a choice of three flavours!" * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation; either version 2 of the License, or * (at your option) any later version. * * This program is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License * along with this program; if not, write to the Free Software * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ #include <linux/slab.h> #include <linux/poll.h> #include <linux/fs.h> #include <linux/file.h> #include <linux/jhash.h> #include <linux/init.h> #include <linux/futex.h> #include <linux/mount.h> #include <linux/pagemap.h> #include <linux/syscalls.h> |
7ed20e1ad
|
53 |
#include <linux/signal.h> |
9adef58b1
|
54 |
#include <linux/module.h> |
fd5eea421
|
55 |
#include <linux/magic.h> |
b488893a3
|
56 57 |
#include <linux/pid.h> #include <linux/nsproxy.h> |
4732efbeb
|
58 |
#include <asm/futex.h> |
1da177e4c
|
59 |
|
c87e2837b
|
60 |
#include "rtmutex_common.h" |
a0c1e9073
|
61 |
int __read_mostly futex_cmpxchg_enabled; |
1da177e4c
|
62 63 64 |
#define FUTEX_HASHBITS (CONFIG_BASE_SMALL ? 4 : 8) /* |
c87e2837b
|
65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 |
* Priority Inheritance state: */ struct futex_pi_state { /* * list of 'owned' pi_state instances - these have to be * cleaned up in do_exit() if the task exits prematurely: */ struct list_head list; /* * The PI object: */ struct rt_mutex pi_mutex; struct task_struct *owner; atomic_t refcount; union futex_key key; }; /* |
1da177e4c
|
86 87 88 89 |
* We use this hashed waitqueue instead of a normal wait_queue_t, so * we can wake only the relevant ones (hashed queues may be shared). * * A futex_q has a woken state, just like tasks have TASK_RUNNING. |
ec92d0829
|
90 |
* It is considered woken when plist_node_empty(&q->list) || q->lock_ptr == 0. |
1da177e4c
|
91 |
* The order of wakup is always to make the first condition true, then |
73500ac54
|
92 |
* wake up q->waiter, then make the second condition true. |
1da177e4c
|
93 94 |
*/ struct futex_q { |
ec92d0829
|
95 |
struct plist_node list; |
73500ac54
|
96 97 |
/* There can only be a single waiter */ wait_queue_head_t waiter; |
1da177e4c
|
98 |
|
e2970f2fb
|
99 |
/* Which hash list lock to use: */ |
1da177e4c
|
100 |
spinlock_t *lock_ptr; |
e2970f2fb
|
101 |
/* Key which the futex is hashed on: */ |
1da177e4c
|
102 |
union futex_key key; |
c87e2837b
|
103 104 105 |
/* Optional priority inheritance state: */ struct futex_pi_state *pi_state; struct task_struct *task; |
cd689985c
|
106 107 108 |
/* Bitset for the optional bitmasked wakeup */ u32 bitset; |
1da177e4c
|
109 110 111 112 113 114 |
}; /* * Split the global futex_lock into every hash list lock. */ struct futex_hash_bucket { |
ec92d0829
|
115 116 |
spinlock_t lock; struct plist_head chain; |
1da177e4c
|
117 118 119 |
}; static struct futex_hash_bucket futex_queues[1<<FUTEX_HASHBITS]; |
1da177e4c
|
120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 |
/* * We hash on the keys returned from get_futex_key (see below). */ static struct futex_hash_bucket *hash_futex(union futex_key *key) { u32 hash = jhash2((u32*)&key->both.word, (sizeof(key->both.word)+sizeof(key->both.ptr))/4, key->both.offset); return &futex_queues[hash & ((1 << FUTEX_HASHBITS)-1)]; } /* * Return 1 if two futex_keys are equal, 0 otherwise. */ static inline int match_futex(union futex_key *key1, union futex_key *key2) { return (key1->both.word == key2->both.word && key1->both.ptr == key2->both.ptr && key1->both.offset == key2->both.offset); } |
38d47c1b7
|
140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 |
/* * Take a reference to the resource addressed by a key. * Can be called while holding spinlocks. * */ static void get_futex_key_refs(union futex_key *key) { if (!key->both.ptr) return; switch (key->both.offset & (FUT_OFF_INODE|FUT_OFF_MMSHARED)) { case FUT_OFF_INODE: atomic_inc(&key->shared.inode->i_count); break; case FUT_OFF_MMSHARED: atomic_inc(&key->private.mm->mm_count); break; } } /* * Drop a reference to the resource addressed by a key. * The hash bucket spinlock must not be held. */ static void drop_futex_key_refs(union futex_key *key) { |
90621c40c
|
166 167 168 |
if (!key->both.ptr) { /* If we're here then we tried to put a key we failed to get */ WARN_ON_ONCE(1); |
38d47c1b7
|
169 |
return; |
90621c40c
|
170 |
} |
38d47c1b7
|
171 172 173 174 175 176 177 178 179 180 |
switch (key->both.offset & (FUT_OFF_INODE|FUT_OFF_MMSHARED)) { case FUT_OFF_INODE: iput(key->shared.inode); break; case FUT_OFF_MMSHARED: mmdrop(key->private.mm); break; } } |
34f01cc1f
|
181 182 183 184 185 186 187 188 189 |
/** * get_futex_key - Get parameters which are the keys for a futex. * @uaddr: virtual address of the futex * @shared: NULL for a PROCESS_PRIVATE futex, * ¤t->mm->mmap_sem for a PROCESS_SHARED futex * @key: address where result is stored. * * Returns a negative error code or 0 * The key words are stored in *key on success. |
1da177e4c
|
190 |
* |
f3a43f3f6
|
191 |
* For shared mappings, it's (page->index, vma->vm_file->f_path.dentry->d_inode, |
1da177e4c
|
192 193 194 |
* offset_within_page). For private mappings, it's (uaddr, current->mm). * We can usually work out the index without swapping in the page. * |
34f01cc1f
|
195 196 197 |
* fshared is NULL for PROCESS_PRIVATE futexes * For other futexes, it points to ¤t->mm->mmap_sem and * caller must have taken the reader lock. but NOT any spinlocks. |
1da177e4c
|
198 |
*/ |
c2f9f2015
|
199 |
static int get_futex_key(u32 __user *uaddr, int fshared, union futex_key *key) |
1da177e4c
|
200 |
{ |
e2970f2fb
|
201 |
unsigned long address = (unsigned long)uaddr; |
1da177e4c
|
202 |
struct mm_struct *mm = current->mm; |
1da177e4c
|
203 204 205 206 207 208 |
struct page *page; int err; /* * The futex address must be "naturally" aligned. */ |
e2970f2fb
|
209 |
key->both.offset = address % PAGE_SIZE; |
34f01cc1f
|
210 |
if (unlikely((address % sizeof(u32)) != 0)) |
1da177e4c
|
211 |
return -EINVAL; |
e2970f2fb
|
212 |
address -= key->both.offset; |
1da177e4c
|
213 214 |
/* |
34f01cc1f
|
215 216 217 218 219 220 221 222 223 224 225 |
* PROCESS_PRIVATE futexes are fast. * As the mm cannot disappear under us and the 'key' only needs * virtual address, we dont even have to find the underlying vma. * Note : We do have to check 'uaddr' is a valid user address, * but access_ok() should be faster than find_vma() */ if (!fshared) { if (unlikely(!access_ok(VERIFY_WRITE, uaddr, sizeof(u32)))) return -EFAULT; key->private.mm = mm; key->private.address = address; |
42569c399
|
226 |
get_futex_key_refs(key); |
34f01cc1f
|
227 228 |
return 0; } |
1da177e4c
|
229 |
|
38d47c1b7
|
230 |
again: |
734b05b10
|
231 |
err = get_user_pages_fast(address, 1, 0, &page); |
38d47c1b7
|
232 233 234 235 236 237 238 239 240 |
if (err < 0) return err; lock_page(page); if (!page->mapping) { unlock_page(page); put_page(page); goto again; } |
1da177e4c
|
241 242 243 244 245 246 |
/* * Private mappings are handled in a simple way. * * NOTE: When userspace waits on a MAP_SHARED mapping, even if * it's a read-only handle, it's expected that futexes attach to |
38d47c1b7
|
247 |
* the object not the particular process. |
1da177e4c
|
248 |
*/ |
38d47c1b7
|
249 250 |
if (PageAnon(page)) { key->both.offset |= FUT_OFF_MMSHARED; /* ref taken on mm */ |
1da177e4c
|
251 |
key->private.mm = mm; |
e2970f2fb
|
252 |
key->private.address = address; |
38d47c1b7
|
253 254 255 256 |
} else { key->both.offset |= FUT_OFF_INODE; /* inode-based key */ key->shared.inode = page->mapping->host; key->shared.pgoff = page->index; |
1da177e4c
|
257 |
} |
38d47c1b7
|
258 |
get_futex_key_refs(key); |
1da177e4c
|
259 |
|
38d47c1b7
|
260 261 262 |
unlock_page(page); put_page(page); return 0; |
1da177e4c
|
263 |
} |
38d47c1b7
|
264 |
static inline |
c2f9f2015
|
265 |
void put_futex_key(int fshared, union futex_key *key) |
1da177e4c
|
266 |
{ |
38d47c1b7
|
267 |
drop_futex_key_refs(key); |
1da177e4c
|
268 |
} |
36cf3b5c3
|
269 270 271 272 273 274 275 276 277 278 279 280 |
static u32 cmpxchg_futex_value_locked(u32 __user *uaddr, u32 uval, u32 newval) { u32 curval; pagefault_disable(); curval = futex_atomic_cmpxchg_inatomic(uaddr, uval, newval); pagefault_enable(); return curval; } static int get_futex_value_locked(u32 *dest, u32 __user *from) |
1da177e4c
|
281 282 |
{ int ret; |
a866374ae
|
283 |
pagefault_disable(); |
e2970f2fb
|
284 |
ret = __copy_from_user_inatomic(dest, from, sizeof(u32)); |
a866374ae
|
285 |
pagefault_enable(); |
1da177e4c
|
286 287 288 289 290 |
return ret ? -EFAULT : 0; } /* |
34f01cc1f
|
291 |
* Fault handling. |
c87e2837b
|
292 |
*/ |
c2f9f2015
|
293 |
static int futex_handle_fault(unsigned long address, int attempt) |
c87e2837b
|
294 295 296 |
{ struct vm_area_struct * vma; struct mm_struct *mm = current->mm; |
34f01cc1f
|
297 |
int ret = -EFAULT; |
c87e2837b
|
298 |
|
34f01cc1f
|
299 300 |
if (attempt > 2) return ret; |
c87e2837b
|
301 |
|
61270708e
|
302 |
down_read(&mm->mmap_sem); |
34f01cc1f
|
303 304 305 |
vma = find_vma(mm, address); if (vma && address >= vma->vm_start && (vma->vm_flags & VM_WRITE)) { |
83c54070e
|
306 307 308 309 310 311 312 313 314 |
int fault; fault = handle_mm_fault(mm, vma, address, 1); if (unlikely((fault & VM_FAULT_ERROR))) { #if 0 /* XXX: let's do this when we verify it is OK */ if (ret & VM_FAULT_OOM) ret = -ENOMEM; #endif } else { |
34f01cc1f
|
315 |
ret = 0; |
83c54070e
|
316 317 318 319 |
if (fault & VM_FAULT_MAJOR) current->maj_flt++; else current->min_flt++; |
34f01cc1f
|
320 |
} |
c87e2837b
|
321 |
} |
61270708e
|
322 |
up_read(&mm->mmap_sem); |
34f01cc1f
|
323 |
return ret; |
c87e2837b
|
324 325 326 327 328 329 330 331 332 333 334 |
} /* * PI code: */ static int refill_pi_state_cache(void) { struct futex_pi_state *pi_state; if (likely(current->pi_state_cache)) return 0; |
4668edc33
|
335 |
pi_state = kzalloc(sizeof(*pi_state), GFP_KERNEL); |
c87e2837b
|
336 337 338 |
if (!pi_state) return -ENOMEM; |
c87e2837b
|
339 340 341 342 |
INIT_LIST_HEAD(&pi_state->list); /* pi_mutex gets initialized later */ pi_state->owner = NULL; atomic_set(&pi_state->refcount, 1); |
38d47c1b7
|
343 |
pi_state->key = FUTEX_KEY_INIT; |
c87e2837b
|
344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 |
current->pi_state_cache = pi_state; return 0; } static struct futex_pi_state * alloc_pi_state(void) { struct futex_pi_state *pi_state = current->pi_state_cache; WARN_ON(!pi_state); current->pi_state_cache = NULL; return pi_state; } static void free_pi_state(struct futex_pi_state *pi_state) { if (!atomic_dec_and_test(&pi_state->refcount)) return; /* * If pi_state->owner is NULL, the owner is most probably dying * and has cleaned up the pi_state already */ if (pi_state->owner) { spin_lock_irq(&pi_state->owner->pi_lock); list_del_init(&pi_state->list); spin_unlock_irq(&pi_state->owner->pi_lock); rt_mutex_proxy_unlock(&pi_state->pi_mutex, pi_state->owner); } if (current->pi_state_cache) kfree(pi_state); else { /* * pi_state->list is already empty. * clear pi_state->owner. * refcount is at 0 - put it back to 1. */ pi_state->owner = NULL; atomic_set(&pi_state->refcount, 1); current->pi_state_cache = pi_state; } } /* * Look up the task based on what TID userspace gave us. * We dont trust it. */ static struct task_struct * futex_find_get_task(pid_t pid) { struct task_struct *p; |
c69e8d9c0
|
398 |
const struct cred *cred = current_cred(), *pcred; |
c87e2837b
|
399 |
|
d359b549b
|
400 |
rcu_read_lock(); |
228ebcbe6
|
401 |
p = find_task_by_vpid(pid); |
c69e8d9c0
|
402 |
if (!p) { |
a06381fec
|
403 |
p = ERR_PTR(-ESRCH); |
c69e8d9c0
|
404 405 406 407 408 409 410 411 |
} else { pcred = __task_cred(p); if (cred->euid != pcred->euid && cred->euid != pcred->uid) p = ERR_PTR(-ESRCH); else get_task_struct(p); } |
a06381fec
|
412 |
|
d359b549b
|
413 |
rcu_read_unlock(); |
c87e2837b
|
414 415 416 417 418 419 420 421 422 423 424 |
return p; } /* * This task is holding PI mutexes at exit time => bad. * Kernel cleans up PI-state, but userspace is likely hosed. * (Robust-futex cleanup is separate and might save the day for userspace.) */ void exit_pi_state_list(struct task_struct *curr) { |
c87e2837b
|
425 426 |
struct list_head *next, *head = &curr->pi_state_list; struct futex_pi_state *pi_state; |
627371d73
|
427 |
struct futex_hash_bucket *hb; |
38d47c1b7
|
428 |
union futex_key key = FUTEX_KEY_INIT; |
c87e2837b
|
429 |
|
a0c1e9073
|
430 431 |
if (!futex_cmpxchg_enabled) return; |
c87e2837b
|
432 433 434 |
/* * We are a ZOMBIE and nobody can enqueue itself on * pi_state_list anymore, but we have to be careful |
627371d73
|
435 |
* versus waiters unqueueing themselves: |
c87e2837b
|
436 437 438 439 440 441 442 |
*/ spin_lock_irq(&curr->pi_lock); while (!list_empty(head)) { next = head->next; pi_state = list_entry(next, struct futex_pi_state, list); key = pi_state->key; |
627371d73
|
443 |
hb = hash_futex(&key); |
c87e2837b
|
444 |
spin_unlock_irq(&curr->pi_lock); |
c87e2837b
|
445 446 447 |
spin_lock(&hb->lock); spin_lock_irq(&curr->pi_lock); |
627371d73
|
448 449 450 451 |
/* * We dropped the pi-lock, so re-check whether this * task still owns the PI-state: */ |
c87e2837b
|
452 453 454 455 |
if (head->next != next) { spin_unlock(&hb->lock); continue; } |
c87e2837b
|
456 |
WARN_ON(pi_state->owner != curr); |
627371d73
|
457 458 |
WARN_ON(list_empty(&pi_state->list)); list_del_init(&pi_state->list); |
c87e2837b
|
459 460 461 462 463 464 465 466 467 468 469 470 471 |
pi_state->owner = NULL; spin_unlock_irq(&curr->pi_lock); rt_mutex_unlock(&pi_state->pi_mutex); spin_unlock(&hb->lock); spin_lock_irq(&curr->pi_lock); } spin_unlock_irq(&curr->pi_lock); } static int |
d0aa7a70b
|
472 473 |
lookup_pi_state(u32 uval, struct futex_hash_bucket *hb, union futex_key *key, struct futex_pi_state **ps) |
c87e2837b
|
474 475 476 |
{ struct futex_pi_state *pi_state = NULL; struct futex_q *this, *next; |
ec92d0829
|
477 |
struct plist_head *head; |
c87e2837b
|
478 |
struct task_struct *p; |
778e9a9c3
|
479 |
pid_t pid = uval & FUTEX_TID_MASK; |
c87e2837b
|
480 481 |
head = &hb->chain; |
ec92d0829
|
482 |
plist_for_each_entry_safe(this, next, head, list) { |
d0aa7a70b
|
483 |
if (match_futex(&this->key, key)) { |
c87e2837b
|
484 485 486 487 488 |
/* * Another waiter already exists - bump up * the refcount and return its pi_state: */ pi_state = this->pi_state; |
06a9ec291
|
489 490 491 492 493 |
/* * Userspace might have messed up non PI and PI futexes */ if (unlikely(!pi_state)) return -EINVAL; |
627371d73
|
494 |
WARN_ON(!atomic_read(&pi_state->refcount)); |
778e9a9c3
|
495 496 |
WARN_ON(pid && pi_state->owner && pi_state->owner->pid != pid); |
627371d73
|
497 |
|
c87e2837b
|
498 |
atomic_inc(&pi_state->refcount); |
d0aa7a70b
|
499 |
*ps = pi_state; |
c87e2837b
|
500 501 502 503 504 505 |
return 0; } } /* |
e3f2ddeac
|
506 |
* We are the first waiter - try to look up the real owner and attach |
778e9a9c3
|
507 |
* the new pi_state to it, but bail out when TID = 0 |
c87e2837b
|
508 |
*/ |
778e9a9c3
|
509 |
if (!pid) |
e3f2ddeac
|
510 |
return -ESRCH; |
c87e2837b
|
511 |
p = futex_find_get_task(pid); |
778e9a9c3
|
512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 |
if (IS_ERR(p)) return PTR_ERR(p); /* * We need to look at the task state flags to figure out, * whether the task is exiting. To protect against the do_exit * change of the task flags, we do this protected by * p->pi_lock: */ spin_lock_irq(&p->pi_lock); if (unlikely(p->flags & PF_EXITING)) { /* * The task is on the way out. When PF_EXITPIDONE is * set, we know that the task has finished the * cleanup: */ int ret = (p->flags & PF_EXITPIDONE) ? -ESRCH : -EAGAIN; spin_unlock_irq(&p->pi_lock); put_task_struct(p); return ret; } |
c87e2837b
|
534 535 536 537 538 539 540 541 542 543 |
pi_state = alloc_pi_state(); /* * Initialize the pi_mutex in locked state and make 'p' * the owner of it: */ rt_mutex_init_proxy_locked(&pi_state->pi_mutex, p); /* Store the key for possible exit cleanups: */ |
d0aa7a70b
|
544 |
pi_state->key = *key; |
c87e2837b
|
545 |
|
627371d73
|
546 |
WARN_ON(!list_empty(&pi_state->list)); |
c87e2837b
|
547 548 549 550 551 |
list_add(&pi_state->list, &p->pi_state_list); pi_state->owner = p; spin_unlock_irq(&p->pi_lock); put_task_struct(p); |
d0aa7a70b
|
552 |
*ps = pi_state; |
c87e2837b
|
553 554 555 556 557 |
return 0; } /* |
1da177e4c
|
558 559 560 561 562 |
* The hash bucket lock must be held when this is called. * Afterwards, the futex_q must not be accessed. */ static void wake_futex(struct futex_q *q) { |
ec92d0829
|
563 |
plist_del(&q->list, &q->list.plist); |
1da177e4c
|
564 565 |
/* * The lock in wake_up_all() is a crucial memory barrier after the |
ec92d0829
|
566 |
* plist_del() and also before assigning to q->lock_ptr. |
1da177e4c
|
567 |
*/ |
73500ac54
|
568 |
wake_up(&q->waiter); |
1da177e4c
|
569 570 571 |
/* * The waiting task can free the futex_q as soon as this is written, * without taking any locks. This must come last. |
8e31108b9
|
572 573 574 575 576 |
* * A memory barrier is required here to prevent the following store * to lock_ptr from getting ahead of the wakeup. Clearing the lock * at the end of wake_up_all() does not prevent this store from * moving. |
1da177e4c
|
577 |
*/ |
ccdea2f88
|
578 |
smp_wmb(); |
1da177e4c
|
579 580 |
q->lock_ptr = NULL; } |
c87e2837b
|
581 582 583 584 585 586 587 588 |
static int wake_futex_pi(u32 __user *uaddr, u32 uval, struct futex_q *this) { struct task_struct *new_owner; struct futex_pi_state *pi_state = this->pi_state; u32 curval, newval; if (!pi_state) return -EINVAL; |
21778867b
|
589 |
spin_lock(&pi_state->pi_mutex.wait_lock); |
c87e2837b
|
590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 |
new_owner = rt_mutex_next_owner(&pi_state->pi_mutex); /* * This happens when we have stolen the lock and the original * pending owner did not enqueue itself back on the rt_mutex. * Thats not a tragedy. We know that way, that a lock waiter * is on the fly. We make the futex_q waiter the pending owner. */ if (!new_owner) new_owner = this->task; /* * We pass it to the next owner. (The WAITERS bit is always * kept enabled while there is PI state around. We must also * preserve the owner died bit.) */ |
e3f2ddeac
|
606 |
if (!(uval & FUTEX_OWNER_DIED)) { |
778e9a9c3
|
607 |
int ret = 0; |
b488893a3
|
608 |
newval = FUTEX_WAITERS | task_pid_vnr(new_owner); |
e3f2ddeac
|
609 |
|
36cf3b5c3
|
610 |
curval = cmpxchg_futex_value_locked(uaddr, uval, newval); |
778e9a9c3
|
611 |
|
e3f2ddeac
|
612 |
if (curval == -EFAULT) |
778e9a9c3
|
613 |
ret = -EFAULT; |
cde898fa8
|
614 |
else if (curval != uval) |
778e9a9c3
|
615 616 617 618 619 |
ret = -EINVAL; if (ret) { spin_unlock(&pi_state->pi_mutex.wait_lock); return ret; } |
e3f2ddeac
|
620 |
} |
c87e2837b
|
621 |
|
627371d73
|
622 623 624 625 626 627 628 |
spin_lock_irq(&pi_state->owner->pi_lock); WARN_ON(list_empty(&pi_state->list)); list_del_init(&pi_state->list); spin_unlock_irq(&pi_state->owner->pi_lock); spin_lock_irq(&new_owner->pi_lock); WARN_ON(!list_empty(&pi_state->list)); |
c87e2837b
|
629 630 |
list_add(&pi_state->list, &new_owner->pi_state_list); pi_state->owner = new_owner; |
627371d73
|
631 |
spin_unlock_irq(&new_owner->pi_lock); |
21778867b
|
632 |
spin_unlock(&pi_state->pi_mutex.wait_lock); |
c87e2837b
|
633 634 635 636 637 638 639 640 641 642 643 644 645 |
rt_mutex_unlock(&pi_state->pi_mutex); return 0; } static int unlock_futex_pi(u32 __user *uaddr, u32 uval) { u32 oldval; /* * There is no waiter, so we unlock the futex. The owner died * bit has not to be preserved here. We are the owner: */ |
36cf3b5c3
|
646 |
oldval = cmpxchg_futex_value_locked(uaddr, uval, 0); |
c87e2837b
|
647 648 649 650 651 652 653 654 |
if (oldval == -EFAULT) return oldval; if (oldval != uval) return -EAGAIN; return 0; } |
1da177e4c
|
655 |
/* |
8b8f319fc
|
656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 |
* Express the locking dependencies for lockdep: */ static inline void double_lock_hb(struct futex_hash_bucket *hb1, struct futex_hash_bucket *hb2) { if (hb1 <= hb2) { spin_lock(&hb1->lock); if (hb1 < hb2) spin_lock_nested(&hb2->lock, SINGLE_DEPTH_NESTING); } else { /* hb1 > hb2 */ spin_lock(&hb2->lock); spin_lock_nested(&hb1->lock, SINGLE_DEPTH_NESTING); } } /* |
1da177e4c
|
672 673 674 |
* Wake up all waiters hashed on the physical page that is mapped * to this virtual address: */ |
c2f9f2015
|
675 |
static int futex_wake(u32 __user *uaddr, int fshared, int nr_wake, u32 bitset) |
1da177e4c
|
676 |
{ |
e2970f2fb
|
677 |
struct futex_hash_bucket *hb; |
1da177e4c
|
678 |
struct futex_q *this, *next; |
ec92d0829
|
679 |
struct plist_head *head; |
38d47c1b7
|
680 |
union futex_key key = FUTEX_KEY_INIT; |
1da177e4c
|
681 |
int ret; |
cd689985c
|
682 683 |
if (!bitset) return -EINVAL; |
34f01cc1f
|
684 |
ret = get_futex_key(uaddr, fshared, &key); |
1da177e4c
|
685 686 |
if (unlikely(ret != 0)) goto out; |
e2970f2fb
|
687 688 689 |
hb = hash_futex(&key); spin_lock(&hb->lock); head = &hb->chain; |
1da177e4c
|
690 |
|
ec92d0829
|
691 |
plist_for_each_entry_safe(this, next, head, list) { |
1da177e4c
|
692 |
if (match_futex (&this->key, &key)) { |
ed6f7b10e
|
693 694 695 696 |
if (this->pi_state) { ret = -EINVAL; break; } |
cd689985c
|
697 698 699 700 |
/* Check if one of the bits is set in both bitsets */ if (!(this->bitset & bitset)) continue; |
1da177e4c
|
701 702 703 704 705 |
wake_futex(this); if (++ret >= nr_wake) break; } } |
e2970f2fb
|
706 |
spin_unlock(&hb->lock); |
38d47c1b7
|
707 |
put_futex_key(fshared, &key); |
42d35d48c
|
708 |
out: |
1da177e4c
|
709 710 711 712 |
return ret; } /* |
4732efbeb
|
713 714 715 |
* Wake up all waiters hashed on the physical page that is mapped * to this virtual address: */ |
e2970f2fb
|
716 |
static int |
c2f9f2015
|
717 |
futex_wake_op(u32 __user *uaddr1, int fshared, u32 __user *uaddr2, |
e2970f2fb
|
718 |
int nr_wake, int nr_wake2, int op) |
4732efbeb
|
719 |
{ |
38d47c1b7
|
720 |
union futex_key key1 = FUTEX_KEY_INIT, key2 = FUTEX_KEY_INIT; |
e2970f2fb
|
721 |
struct futex_hash_bucket *hb1, *hb2; |
ec92d0829
|
722 |
struct plist_head *head; |
4732efbeb
|
723 724 725 726 |
struct futex_q *this, *next; int ret, op_ret, attempt = 0; retryfull: |
34f01cc1f
|
727 |
ret = get_futex_key(uaddr1, fshared, &key1); |
4732efbeb
|
728 729 |
if (unlikely(ret != 0)) goto out; |
34f01cc1f
|
730 |
ret = get_futex_key(uaddr2, fshared, &key2); |
4732efbeb
|
731 |
if (unlikely(ret != 0)) |
42d35d48c
|
732 |
goto out_put_key1; |
4732efbeb
|
733 |
|
e2970f2fb
|
734 735 |
hb1 = hash_futex(&key1); hb2 = hash_futex(&key2); |
4732efbeb
|
736 737 |
retry: |
8b8f319fc
|
738 |
double_lock_hb(hb1, hb2); |
4732efbeb
|
739 |
|
e2970f2fb
|
740 |
op_ret = futex_atomic_op_inuser(op, uaddr2); |
4732efbeb
|
741 |
if (unlikely(op_ret < 0)) { |
e2970f2fb
|
742 |
u32 dummy; |
4732efbeb
|
743 |
|
e2970f2fb
|
744 745 746 |
spin_unlock(&hb1->lock); if (hb1 != hb2) spin_unlock(&hb2->lock); |
4732efbeb
|
747 |
|
7ee1dd3fe
|
748 |
#ifndef CONFIG_MMU |
e2970f2fb
|
749 750 751 752 |
/* * we don't get EFAULT from MMU faults if we don't have an MMU, * but we might get them from range checking */ |
7ee1dd3fe
|
753 |
ret = op_ret; |
42d35d48c
|
754 |
goto out_put_keys; |
7ee1dd3fe
|
755 |
#endif |
796f8d9b9
|
756 757 |
if (unlikely(op_ret != -EFAULT)) { ret = op_ret; |
42d35d48c
|
758 |
goto out_put_keys; |
796f8d9b9
|
759 |
} |
e2970f2fb
|
760 761 |
/* * futex_atomic_op_inuser needs to both read and write |
4732efbeb
|
762 763 764 |
* *(int __user *)uaddr2, but we can't modify it * non-atomically. Therefore, if get_user below is not * enough, we need to handle the fault ourselves, while |
e2970f2fb
|
765 766 |
* still holding the mmap_sem. */ |
4732efbeb
|
767 |
if (attempt++) { |
34f01cc1f
|
768 |
ret = futex_handle_fault((unsigned long)uaddr2, |
c2f9f2015
|
769 |
attempt); |
34f01cc1f
|
770 |
if (ret) |
42d35d48c
|
771 |
goto out_put_keys; |
4732efbeb
|
772 773 |
goto retry; } |
e2970f2fb
|
774 |
ret = get_user(dummy, uaddr2); |
4732efbeb
|
775 776 777 778 779 |
if (ret) return ret; goto retryfull; } |
e2970f2fb
|
780 |
head = &hb1->chain; |
4732efbeb
|
781 |
|
ec92d0829
|
782 |
plist_for_each_entry_safe(this, next, head, list) { |
4732efbeb
|
783 784 785 786 787 788 789 790 |
if (match_futex (&this->key, &key1)) { wake_futex(this); if (++ret >= nr_wake) break; } } if (op_ret > 0) { |
e2970f2fb
|
791 |
head = &hb2->chain; |
4732efbeb
|
792 793 |
op_ret = 0; |
ec92d0829
|
794 |
plist_for_each_entry_safe(this, next, head, list) { |
4732efbeb
|
795 796 797 798 799 800 801 802 |
if (match_futex (&this->key, &key2)) { wake_futex(this); if (++op_ret >= nr_wake2) break; } } ret += op_ret; } |
e2970f2fb
|
803 804 805 |
spin_unlock(&hb1->lock); if (hb1 != hb2) spin_unlock(&hb2->lock); |
42d35d48c
|
806 |
out_put_keys: |
38d47c1b7
|
807 |
put_futex_key(fshared, &key2); |
42d35d48c
|
808 |
out_put_key1: |
38d47c1b7
|
809 |
put_futex_key(fshared, &key1); |
42d35d48c
|
810 |
out: |
4732efbeb
|
811 812 813 814 |
return ret; } /* |
1da177e4c
|
815 816 817 |
* Requeue all waiters hashed on one physical page to another * physical page. */ |
c2f9f2015
|
818 |
static int futex_requeue(u32 __user *uaddr1, int fshared, u32 __user *uaddr2, |
e2970f2fb
|
819 |
int nr_wake, int nr_requeue, u32 *cmpval) |
1da177e4c
|
820 |
{ |
38d47c1b7
|
821 |
union futex_key key1 = FUTEX_KEY_INIT, key2 = FUTEX_KEY_INIT; |
e2970f2fb
|
822 |
struct futex_hash_bucket *hb1, *hb2; |
ec92d0829
|
823 |
struct plist_head *head1; |
1da177e4c
|
824 825 |
struct futex_q *this, *next; int ret, drop_count = 0; |
42d35d48c
|
826 |
retry: |
34f01cc1f
|
827 |
ret = get_futex_key(uaddr1, fshared, &key1); |
1da177e4c
|
828 829 |
if (unlikely(ret != 0)) goto out; |
34f01cc1f
|
830 |
ret = get_futex_key(uaddr2, fshared, &key2); |
1da177e4c
|
831 |
if (unlikely(ret != 0)) |
42d35d48c
|
832 |
goto out_put_key1; |
1da177e4c
|
833 |
|
e2970f2fb
|
834 835 |
hb1 = hash_futex(&key1); hb2 = hash_futex(&key2); |
1da177e4c
|
836 |
|
8b8f319fc
|
837 |
double_lock_hb(hb1, hb2); |
1da177e4c
|
838 |
|
e2970f2fb
|
839 840 |
if (likely(cmpval != NULL)) { u32 curval; |
1da177e4c
|
841 |
|
e2970f2fb
|
842 |
ret = get_futex_value_locked(&curval, uaddr1); |
1da177e4c
|
843 844 |
if (unlikely(ret)) { |
e2970f2fb
|
845 846 847 |
spin_unlock(&hb1->lock); if (hb1 != hb2) spin_unlock(&hb2->lock); |
1da177e4c
|
848 |
|
e2970f2fb
|
849 |
ret = get_user(curval, uaddr1); |
1da177e4c
|
850 851 852 |
if (!ret) goto retry; |
42d35d48c
|
853 |
goto out_put_keys; |
1da177e4c
|
854 |
} |
e2970f2fb
|
855 |
if (curval != *cmpval) { |
1da177e4c
|
856 857 858 859 |
ret = -EAGAIN; goto out_unlock; } } |
e2970f2fb
|
860 |
head1 = &hb1->chain; |
ec92d0829
|
861 |
plist_for_each_entry_safe(this, next, head1, list) { |
1da177e4c
|
862 863 864 865 866 |
if (!match_futex (&this->key, &key1)) continue; if (++ret <= nr_wake) { wake_futex(this); } else { |
59e0e0ace
|
867 868 869 870 871 |
/* * If key1 and key2 hash to the same bucket, no need to * requeue. */ if (likely(head1 != &hb2->chain)) { |
ec92d0829
|
872 873 |
plist_del(&this->list, &hb1->chain); plist_add(&this->list, &hb2->chain); |
59e0e0ace
|
874 |
this->lock_ptr = &hb2->lock; |
ec92d0829
|
875 876 877 |
#ifdef CONFIG_DEBUG_PI_LIST this->list.plist.lock = &hb2->lock; #endif |
778e9a9c3
|
878 |
} |
1da177e4c
|
879 |
this->key = key2; |
9adef58b1
|
880 |
get_futex_key_refs(&key2); |
1da177e4c
|
881 882 883 884 |
drop_count++; if (ret - nr_wake >= nr_requeue) break; |
1da177e4c
|
885 886 887 888 |
} } out_unlock: |
e2970f2fb
|
889 890 891 |
spin_unlock(&hb1->lock); if (hb1 != hb2) spin_unlock(&hb2->lock); |
1da177e4c
|
892 |
|
9adef58b1
|
893 |
/* drop_futex_key_refs() must be called outside the spinlocks. */ |
1da177e4c
|
894 |
while (--drop_count >= 0) |
9adef58b1
|
895 |
drop_futex_key_refs(&key1); |
1da177e4c
|
896 |
|
42d35d48c
|
897 |
out_put_keys: |
38d47c1b7
|
898 |
put_futex_key(fshared, &key2); |
42d35d48c
|
899 |
out_put_key1: |
38d47c1b7
|
900 |
put_futex_key(fshared, &key1); |
42d35d48c
|
901 |
out: |
1da177e4c
|
902 903 904 905 |
return ret; } /* The key must be already stored in q->key. */ |
82af7aca5
|
906 |
static inline struct futex_hash_bucket *queue_lock(struct futex_q *q) |
1da177e4c
|
907 |
{ |
e2970f2fb
|
908 |
struct futex_hash_bucket *hb; |
1da177e4c
|
909 |
|
73500ac54
|
910 |
init_waitqueue_head(&q->waiter); |
1da177e4c
|
911 |
|
9adef58b1
|
912 |
get_futex_key_refs(&q->key); |
e2970f2fb
|
913 914 |
hb = hash_futex(&q->key); q->lock_ptr = &hb->lock; |
1da177e4c
|
915 |
|
e2970f2fb
|
916 917 |
spin_lock(&hb->lock); return hb; |
1da177e4c
|
918 |
} |
82af7aca5
|
919 |
static inline void queue_me(struct futex_q *q, struct futex_hash_bucket *hb) |
1da177e4c
|
920 |
{ |
ec92d0829
|
921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 |
int prio; /* * The priority used to register this element is * - either the real thread-priority for the real-time threads * (i.e. threads with a priority lower than MAX_RT_PRIO) * - or MAX_RT_PRIO for non-RT threads. * Thus, all RT-threads are woken first in priority order, and * the others are woken last, in FIFO order. */ prio = min(current->normal_prio, MAX_RT_PRIO); plist_node_init(&q->list, prio); #ifdef CONFIG_DEBUG_PI_LIST q->list.plist.lock = &hb->lock; #endif plist_add(&q->list, &hb->chain); |
c87e2837b
|
938 |
q->task = current; |
e2970f2fb
|
939 |
spin_unlock(&hb->lock); |
1da177e4c
|
940 941 942 |
} static inline void |
e2970f2fb
|
943 |
queue_unlock(struct futex_q *q, struct futex_hash_bucket *hb) |
1da177e4c
|
944 |
{ |
e2970f2fb
|
945 |
spin_unlock(&hb->lock); |
9adef58b1
|
946 |
drop_futex_key_refs(&q->key); |
1da177e4c
|
947 948 949 950 951 952 |
} /* * queue_me and unqueue_me must be called as a pair, each * exactly once. They are called with the hashed spinlock held. */ |
1da177e4c
|
953 954 955 |
/* Return 1 if we were still queued (ie. 0 means we were woken) */ static int unqueue_me(struct futex_q *q) { |
1da177e4c
|
956 |
spinlock_t *lock_ptr; |
e2970f2fb
|
957 |
int ret = 0; |
1da177e4c
|
958 959 |
/* In the common case we don't take the spinlock, which is nice. */ |
42d35d48c
|
960 |
retry: |
1da177e4c
|
961 |
lock_ptr = q->lock_ptr; |
e91467ecd
|
962 |
barrier(); |
c80544dc0
|
963 |
if (lock_ptr != NULL) { |
1da177e4c
|
964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 |
spin_lock(lock_ptr); /* * q->lock_ptr can change between reading it and * spin_lock(), causing us to take the wrong lock. This * corrects the race condition. * * Reasoning goes like this: if we have the wrong lock, * q->lock_ptr must have changed (maybe several times) * between reading it and the spin_lock(). It can * change again after the spin_lock() but only if it was * already changed before the spin_lock(). It cannot, * however, change back to the original value. Therefore * we can detect whether we acquired the correct lock. */ if (unlikely(lock_ptr != q->lock_ptr)) { spin_unlock(lock_ptr); goto retry; } |
ec92d0829
|
982 983 |
WARN_ON(plist_node_empty(&q->list)); plist_del(&q->list, &q->list.plist); |
c87e2837b
|
984 985 |
BUG_ON(q->pi_state); |
1da177e4c
|
986 987 988 |
spin_unlock(lock_ptr); ret = 1; } |
9adef58b1
|
989 |
drop_futex_key_refs(&q->key); |
1da177e4c
|
990 991 |
return ret; } |
c87e2837b
|
992 993 |
/* * PI futexes can not be requeued and must remove themself from the |
d0aa7a70b
|
994 995 |
* hash bucket. The hash bucket lock (i.e. lock_ptr) is held on entry * and dropped here. |
c87e2837b
|
996 |
*/ |
d0aa7a70b
|
997 |
static void unqueue_me_pi(struct futex_q *q) |
c87e2837b
|
998 |
{ |
ec92d0829
|
999 1000 |
WARN_ON(plist_node_empty(&q->list)); plist_del(&q->list, &q->list.plist); |
c87e2837b
|
1001 1002 1003 1004 |
BUG_ON(!q->pi_state); free_pi_state(q->pi_state); q->pi_state = NULL; |
d0aa7a70b
|
1005 |
spin_unlock(q->lock_ptr); |
c87e2837b
|
1006 |
|
9adef58b1
|
1007 |
drop_futex_key_refs(&q->key); |
c87e2837b
|
1008 |
} |
d0aa7a70b
|
1009 |
/* |
cdf71a10c
|
1010 |
* Fixup the pi_state owner with the new owner. |
d0aa7a70b
|
1011 |
* |
778e9a9c3
|
1012 1013 |
* Must be called with hash bucket lock held and mm->sem held for non * private futexes. |
d0aa7a70b
|
1014 |
*/ |
778e9a9c3
|
1015 |
static int fixup_pi_state_owner(u32 __user *uaddr, struct futex_q *q, |
c2f9f2015
|
1016 |
struct task_struct *newowner, int fshared) |
d0aa7a70b
|
1017 |
{ |
cdf71a10c
|
1018 |
u32 newtid = task_pid_vnr(newowner) | FUTEX_WAITERS; |
d0aa7a70b
|
1019 |
struct futex_pi_state *pi_state = q->pi_state; |
1b7558e45
|
1020 |
struct task_struct *oldowner = pi_state->owner; |
d0aa7a70b
|
1021 |
u32 uval, curval, newval; |
1b7558e45
|
1022 |
int ret, attempt = 0; |
d0aa7a70b
|
1023 1024 |
/* Owner died? */ |
1b7558e45
|
1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 |
if (!pi_state->owner) newtid |= FUTEX_OWNER_DIED; /* * We are here either because we stole the rtmutex from the * pending owner or we are the pending owner which failed to * get the rtmutex. We have to replace the pending owner TID * in the user space variable. This must be atomic as we have * to preserve the owner died bit here. * * Note: We write the user space value _before_ changing the * pi_state because we can fault here. Imagine swapped out * pages or a fork, which was running right before we acquired * mmap_sem, that marked all the anonymous memory readonly for * cow. * * Modifying pi_state _before_ the user space value would * leave the pi_state in an inconsistent state when we fault * here, because we need to drop the hash bucket lock to * handle the fault. This might be observed in the PID check * in lookup_pi_state. */ retry: if (get_futex_value_locked(&uval, uaddr)) goto handle_fault; while (1) { newval = (uval & FUTEX_OWNER_DIED) | newtid; curval = cmpxchg_futex_value_locked(uaddr, uval, newval); if (curval == -EFAULT) goto handle_fault; if (curval == uval) break; uval = curval; } /* * We fixed up user space. Now we need to fix the pi_state * itself. */ |
d0aa7a70b
|
1067 1068 1069 1070 1071 |
if (pi_state->owner != NULL) { spin_lock_irq(&pi_state->owner->pi_lock); WARN_ON(list_empty(&pi_state->list)); list_del_init(&pi_state->list); spin_unlock_irq(&pi_state->owner->pi_lock); |
1b7558e45
|
1072 |
} |
d0aa7a70b
|
1073 |
|
cdf71a10c
|
1074 |
pi_state->owner = newowner; |
d0aa7a70b
|
1075 |
|
cdf71a10c
|
1076 |
spin_lock_irq(&newowner->pi_lock); |
d0aa7a70b
|
1077 |
WARN_ON(!list_empty(&pi_state->list)); |
cdf71a10c
|
1078 1079 |
list_add(&pi_state->list, &newowner->pi_state_list); spin_unlock_irq(&newowner->pi_lock); |
1b7558e45
|
1080 |
return 0; |
d0aa7a70b
|
1081 |
|
d0aa7a70b
|
1082 |
/* |
1b7558e45
|
1083 1084 1085 1086 1087 1088 1089 1090 |
* To handle the page fault we need to drop the hash bucket * lock here. That gives the other task (either the pending * owner itself or the task which stole the rtmutex) the * chance to try the fixup of the pi_state. So once we are * back from handling the fault we need to check the pi_state * after reacquiring the hash bucket lock and before trying to * do another fixup. When the fixup has been done already we * simply return. |
d0aa7a70b
|
1091 |
*/ |
1b7558e45
|
1092 1093 |
handle_fault: spin_unlock(q->lock_ptr); |
778e9a9c3
|
1094 |
|
c2f9f2015
|
1095 |
ret = futex_handle_fault((unsigned long)uaddr, attempt++); |
778e9a9c3
|
1096 |
|
1b7558e45
|
1097 |
spin_lock(q->lock_ptr); |
778e9a9c3
|
1098 |
|
1b7558e45
|
1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 |
/* * Check if someone else fixed it for us: */ if (pi_state->owner != oldowner) return 0; if (ret) return ret; goto retry; |
d0aa7a70b
|
1109 |
} |
34f01cc1f
|
1110 1111 |
/* * In case we must use restart_block to restart a futex_wait, |
ce6bd420f
|
1112 |
* we encode in the 'flags' shared capability |
34f01cc1f
|
1113 |
*/ |
1acdac104
|
1114 1115 |
#define FLAGS_SHARED 0x01 #define FLAGS_CLOCKRT 0x02 |
34f01cc1f
|
1116 |
|
72c1bbf30
|
1117 |
static long futex_wait_restart(struct restart_block *restart); |
36cf3b5c3
|
1118 |
|
c2f9f2015
|
1119 |
static int futex_wait(u32 __user *uaddr, int fshared, |
1acdac104
|
1120 |
u32 val, ktime_t *abs_time, u32 bitset, int clockrt) |
1da177e4c
|
1121 |
{ |
c87e2837b
|
1122 1123 |
struct task_struct *curr = current; DECLARE_WAITQUEUE(wait, curr); |
e2970f2fb
|
1124 |
struct futex_hash_bucket *hb; |
1da177e4c
|
1125 |
struct futex_q q; |
e2970f2fb
|
1126 1127 |
u32 uval; int ret; |
bd197234b
|
1128 |
struct hrtimer_sleeper t; |
c19384b5b
|
1129 |
int rem = 0; |
1da177e4c
|
1130 |
|
cd689985c
|
1131 1132 |
if (!bitset) return -EINVAL; |
c87e2837b
|
1133 |
q.pi_state = NULL; |
cd689985c
|
1134 |
q.bitset = bitset; |
42d35d48c
|
1135 |
retry: |
38d47c1b7
|
1136 |
q.key = FUTEX_KEY_INIT; |
34f01cc1f
|
1137 |
ret = get_futex_key(uaddr, fshared, &q.key); |
1da177e4c
|
1138 |
if (unlikely(ret != 0)) |
42d35d48c
|
1139 |
goto out; |
1da177e4c
|
1140 |
|
82af7aca5
|
1141 |
hb = queue_lock(&q); |
1da177e4c
|
1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 |
/* * Access the page AFTER the futex is queued. * Order is important: * * Userspace waiter: val = var; if (cond(val)) futex_wait(&var, val); * Userspace waker: if (cond(var)) { var = new; futex_wake(&var); } * * The basic logical guarantee of a futex is that it blocks ONLY * if cond(var) is known to be true at the time of blocking, for * any cond. If we queued after testing *uaddr, that would open * a race condition where we could block indefinitely with * cond(var) false, which would violate the guarantee. * * A consequence is that futex_wait() can return zero and absorb * a wakeup when *uaddr != val on entry to the syscall. This is * rare, but normal. * |
34f01cc1f
|
1160 1161 |
* for shared futexes, we hold the mmap semaphore, so the mapping * cannot have changed since we looked it up in get_futex_key. |
1da177e4c
|
1162 |
*/ |
e2970f2fb
|
1163 |
ret = get_futex_value_locked(&uval, uaddr); |
1da177e4c
|
1164 1165 |
if (unlikely(ret)) { |
e2970f2fb
|
1166 |
queue_unlock(&q, hb); |
42d35d48c
|
1167 |
put_futex_key(fshared, &q.key); |
1da177e4c
|
1168 |
|
e2970f2fb
|
1169 |
ret = get_user(uval, uaddr); |
1da177e4c
|
1170 1171 1172 1173 1174 |
if (!ret) goto retry; return ret; } |
c87e2837b
|
1175 1176 |
ret = -EWOULDBLOCK; if (uval != val) |
42d35d48c
|
1177 |
goto out_unlock_put_key; |
1da177e4c
|
1178 1179 |
/* Only actually queue if *uaddr contained val. */ |
82af7aca5
|
1180 |
queue_me(&q, hb); |
1da177e4c
|
1181 1182 |
/* |
1da177e4c
|
1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 |
* There might have been scheduling since the queue_me(), as we * cannot hold a spinlock across the get_user() in case it * faults, and we cannot just set TASK_INTERRUPTIBLE state when * queueing ourselves into the futex hash. This code thus has to * rely on the futex_wake() code removing us from hash when it * wakes us up. */ /* add_wait_queue is the barrier after __set_current_state. */ __set_current_state(TASK_INTERRUPTIBLE); |
73500ac54
|
1193 |
add_wait_queue(&q.waiter, &wait); |
1da177e4c
|
1194 |
/* |
ec92d0829
|
1195 |
* !plist_node_empty() is safe here without any lock. |
1da177e4c
|
1196 1197 |
* q.lock_ptr != 0 is not safe, because of ordering against wakeup. */ |
ec92d0829
|
1198 |
if (likely(!plist_node_empty(&q.list))) { |
c19384b5b
|
1199 1200 1201 |
if (!abs_time) schedule(); else { |
ae4b748e8
|
1202 1203 1204 1205 |
unsigned long slack; slack = current->timer_slack_ns; if (rt_task(current)) slack = 0; |
1acdac104
|
1206 1207 1208 1209 |
hrtimer_init_on_stack(&t.timer, clockrt ? CLOCK_REALTIME : CLOCK_MONOTONIC, HRTIMER_MODE_ABS); |
c19384b5b
|
1210 |
hrtimer_init_sleeper(&t, current); |
ae4b748e8
|
1211 |
hrtimer_set_expires_range_ns(&t.timer, *abs_time, slack); |
c19384b5b
|
1212 |
|
cc584b213
|
1213 |
hrtimer_start_expires(&t.timer, HRTIMER_MODE_ABS); |
3588a085c
|
1214 1215 |
if (!hrtimer_active(&t.timer)) t.task = NULL; |
c19384b5b
|
1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 |
/* * the timer could have already expired, in which * case current would be flagged for rescheduling. * Don't bother calling schedule. */ if (likely(t.task)) schedule(); hrtimer_cancel(&t.timer); |
72c1bbf30
|
1226 |
|
c19384b5b
|
1227 1228 |
/* Flag if a timeout occured */ rem = (t.task == NULL); |
237fc6e7a
|
1229 1230 |
destroy_hrtimer_on_stack(&t.timer); |
c19384b5b
|
1231 |
} |
72c1bbf30
|
1232 |
} |
1da177e4c
|
1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 |
__set_current_state(TASK_RUNNING); /* * NOTE: we don't remove ourselves from the waitqueue because * we are the only user of it. */ /* If we were woken (and unqueued), we succeeded, whatever. */ if (!unqueue_me(&q)) return 0; |
c19384b5b
|
1243 |
if (rem) |
1da177e4c
|
1244 |
return -ETIMEDOUT; |
72c1bbf30
|
1245 |
|
e2970f2fb
|
1246 1247 1248 1249 |
/* * We expect signal_pending(current), but another thread may * have handled it for us already. */ |
c19384b5b
|
1250 |
if (!abs_time) |
72c1bbf30
|
1251 1252 1253 1254 1255 |
return -ERESTARTSYS; else { struct restart_block *restart; restart = ¤t_thread_info()->restart_block; restart->fn = futex_wait_restart; |
ce6bd420f
|
1256 1257 1258 |
restart->futex.uaddr = (u32 *)uaddr; restart->futex.val = val; restart->futex.time = abs_time->tv64; |
cd689985c
|
1259 |
restart->futex.bitset = bitset; |
ce6bd420f
|
1260 |
restart->futex.flags = 0; |
34f01cc1f
|
1261 |
if (fshared) |
ce6bd420f
|
1262 |
restart->futex.flags |= FLAGS_SHARED; |
1acdac104
|
1263 1264 |
if (clockrt) restart->futex.flags |= FLAGS_CLOCKRT; |
72c1bbf30
|
1265 1266 |
return -ERESTART_RESTARTBLOCK; } |
1da177e4c
|
1267 |
|
42d35d48c
|
1268 |
out_unlock_put_key: |
c87e2837b
|
1269 |
queue_unlock(&q, hb); |
38d47c1b7
|
1270 |
put_futex_key(fshared, &q.key); |
42d35d48c
|
1271 1272 |
out: |
c87e2837b
|
1273 1274 |
return ret; } |
72c1bbf30
|
1275 1276 1277 |
static long futex_wait_restart(struct restart_block *restart) { |
ce6bd420f
|
1278 |
u32 __user *uaddr = (u32 __user *)restart->futex.uaddr; |
c2f9f2015
|
1279 |
int fshared = 0; |
ce6bd420f
|
1280 |
ktime_t t; |
72c1bbf30
|
1281 |
|
ce6bd420f
|
1282 |
t.tv64 = restart->futex.time; |
72c1bbf30
|
1283 |
restart->fn = do_no_restart_syscall; |
ce6bd420f
|
1284 |
if (restart->futex.flags & FLAGS_SHARED) |
c2f9f2015
|
1285 |
fshared = 1; |
cd689985c
|
1286 |
return (long)futex_wait(uaddr, fshared, restart->futex.val, &t, |
1acdac104
|
1287 1288 |
restart->futex.bitset, restart->futex.flags & FLAGS_CLOCKRT); |
72c1bbf30
|
1289 |
} |
c87e2837b
|
1290 1291 1292 1293 1294 1295 |
/* * Userspace tried a 0 -> TID atomic transition of the futex value * and failed. The kernel side here does the whole locking operation: * if there are waiters then it will block, it does PI, etc. (Due to * races the kernel might see a 0 value of the futex too.) */ |
c2f9f2015
|
1296 |
static int futex_lock_pi(u32 __user *uaddr, int fshared, |
34f01cc1f
|
1297 |
int detect, ktime_t *time, int trylock) |
c87e2837b
|
1298 |
{ |
c5780e976
|
1299 |
struct hrtimer_sleeper timeout, *to = NULL; |
c87e2837b
|
1300 1301 1302 1303 |
struct task_struct *curr = current; struct futex_hash_bucket *hb; u32 uval, newval, curval; struct futex_q q; |
778e9a9c3
|
1304 |
int ret, lock_taken, ownerdied = 0, attempt = 0; |
c87e2837b
|
1305 1306 1307 |
if (refill_pi_state_cache()) return -ENOMEM; |
c19384b5b
|
1308 |
if (time) { |
c5780e976
|
1309 |
to = &timeout; |
237fc6e7a
|
1310 1311 |
hrtimer_init_on_stack(&to->timer, CLOCK_REALTIME, HRTIMER_MODE_ABS); |
c5780e976
|
1312 |
hrtimer_init_sleeper(to, current); |
cc584b213
|
1313 |
hrtimer_set_expires(&to->timer, *time); |
c5780e976
|
1314 |
} |
c87e2837b
|
1315 |
q.pi_state = NULL; |
42d35d48c
|
1316 |
retry: |
38d47c1b7
|
1317 |
q.key = FUTEX_KEY_INIT; |
34f01cc1f
|
1318 |
ret = get_futex_key(uaddr, fshared, &q.key); |
c87e2837b
|
1319 |
if (unlikely(ret != 0)) |
42d35d48c
|
1320 |
goto out; |
c87e2837b
|
1321 |
|
42d35d48c
|
1322 |
retry_unlocked: |
82af7aca5
|
1323 |
hb = queue_lock(&q); |
c87e2837b
|
1324 |
|
42d35d48c
|
1325 |
retry_locked: |
778e9a9c3
|
1326 |
ret = lock_taken = 0; |
d0aa7a70b
|
1327 |
|
c87e2837b
|
1328 1329 1330 1331 1332 |
/* * To avoid races, we attempt to take the lock here again * (by doing a 0 -> TID atomic cmpxchg), while holding all * the locks. It will most likely not succeed. */ |
b488893a3
|
1333 |
newval = task_pid_vnr(current); |
c87e2837b
|
1334 |
|
36cf3b5c3
|
1335 |
curval = cmpxchg_futex_value_locked(uaddr, 0, newval); |
c87e2837b
|
1336 1337 1338 |
if (unlikely(curval == -EFAULT)) goto uaddr_faulted; |
778e9a9c3
|
1339 1340 1341 1342 |
/* * Detect deadlocks. In case of REQUEUE_PI this is a valid * situation and we return success to user space. */ |
b488893a3
|
1343 |
if (unlikely((curval & FUTEX_TID_MASK) == task_pid_vnr(current))) { |
bd197234b
|
1344 |
ret = -EDEADLK; |
42d35d48c
|
1345 |
goto out_unlock_put_key; |
c87e2837b
|
1346 1347 1348 |
} /* |
778e9a9c3
|
1349 |
* Surprise - we got the lock. Just return to userspace: |
c87e2837b
|
1350 1351 |
*/ if (unlikely(!curval)) |
42d35d48c
|
1352 |
goto out_unlock_put_key; |
c87e2837b
|
1353 1354 |
uval = curval; |
778e9a9c3
|
1355 |
|
d0aa7a70b
|
1356 |
/* |
778e9a9c3
|
1357 1358 |
* Set the WAITERS flag, so the owner will know it has someone * to wake at next unlock |
d0aa7a70b
|
1359 |
*/ |
778e9a9c3
|
1360 1361 1362 1363 |
newval = curval | FUTEX_WAITERS; /* * There are two cases, where a futex might have no owner (the |
bd197234b
|
1364 1365 1366 |
* owner TID is 0): OWNER_DIED. We take over the futex in this * case. We also do an unconditional take over, when the owner * of the futex died. |
778e9a9c3
|
1367 1368 1369 1370 |
* * This is safe as we are protected by the hash bucket lock ! */ if (unlikely(ownerdied || !(curval & FUTEX_TID_MASK))) { |
bd197234b
|
1371 |
/* Keep the OWNER_DIED bit */ |
b488893a3
|
1372 |
newval = (curval & ~FUTEX_TID_MASK) | task_pid_vnr(current); |
778e9a9c3
|
1373 1374 1375 |
ownerdied = 0; lock_taken = 1; } |
c87e2837b
|
1376 |
|
36cf3b5c3
|
1377 |
curval = cmpxchg_futex_value_locked(uaddr, uval, newval); |
c87e2837b
|
1378 1379 1380 1381 1382 |
if (unlikely(curval == -EFAULT)) goto uaddr_faulted; if (unlikely(curval != uval)) goto retry_locked; |
778e9a9c3
|
1383 |
/* |
bd197234b
|
1384 |
* We took the lock due to owner died take over. |
778e9a9c3
|
1385 |
*/ |
bd197234b
|
1386 |
if (unlikely(lock_taken)) |
42d35d48c
|
1387 |
goto out_unlock_put_key; |
d0aa7a70b
|
1388 |
|
c87e2837b
|
1389 1390 1391 1392 |
/* * We dont have the lock. Look up the PI state (or create it if * we are the first waiter): */ |
d0aa7a70b
|
1393 |
ret = lookup_pi_state(uval, hb, &q.key, &q.pi_state); |
c87e2837b
|
1394 1395 |
if (unlikely(ret)) { |
778e9a9c3
|
1396 |
switch (ret) { |
c87e2837b
|
1397 |
|
778e9a9c3
|
1398 1399 1400 1401 1402 1403 |
case -EAGAIN: /* * Task is exiting and we just wait for the * exit to complete. */ queue_unlock(&q, hb); |
778e9a9c3
|
1404 1405 |
cond_resched(); goto retry; |
c87e2837b
|
1406 |
|
778e9a9c3
|
1407 1408 1409 1410 1411 1412 1413 |
case -ESRCH: /* * No owner found for this futex. Check if the * OWNER_DIED bit is set to figure out whether * this is a robust futex or not. */ if (get_futex_value_locked(&curval, uaddr)) |
c87e2837b
|
1414 |
goto uaddr_faulted; |
778e9a9c3
|
1415 1416 1417 1418 1419 1420 1421 1422 |
/* * We simply start over in case of a robust * futex. The code above will take the futex * and return happy. */ if (curval & FUTEX_OWNER_DIED) { ownerdied = 1; |
c87e2837b
|
1423 |
goto retry_locked; |
778e9a9c3
|
1424 1425 |
} default: |
42d35d48c
|
1426 |
goto out_unlock_put_key; |
c87e2837b
|
1427 |
} |
c87e2837b
|
1428 1429 1430 1431 1432 |
} /* * Only actually queue now that the atomic ops are done: */ |
82af7aca5
|
1433 |
queue_me(&q, hb); |
c87e2837b
|
1434 |
|
c87e2837b
|
1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 |
WARN_ON(!q.pi_state); /* * Block on the PI mutex: */ if (!trylock) ret = rt_mutex_timed_lock(&q.pi_state->pi_mutex, to, 1); else { ret = rt_mutex_trylock(&q.pi_state->pi_mutex); /* Fixup the trylock return value: */ ret = ret ? 0 : -EWOULDBLOCK; } |
a99e4e413
|
1446 |
spin_lock(q.lock_ptr); |
c87e2837b
|
1447 |
|
778e9a9c3
|
1448 1449 1450 1451 1452 1453 1454 |
if (!ret) { /* * Got the lock. We might not be the anticipated owner * if we did a lock-steal - fix up the PI-state in * that case: */ if (q.pi_state->owner != curr) |
1b7558e45
|
1455 |
ret = fixup_pi_state_owner(uaddr, &q, curr, fshared); |
778e9a9c3
|
1456 |
} else { |
c87e2837b
|
1457 1458 |
/* * Catch the rare case, where the lock was released |
778e9a9c3
|
1459 1460 |
* when we were on the way back before we locked the * hash bucket. |
c87e2837b
|
1461 |
*/ |
cdf71a10c
|
1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 |
if (q.pi_state->owner == curr) { /* * Try to get the rt_mutex now. This might * fail as some other task acquired the * rt_mutex after we removed ourself from the * rt_mutex waiters list. */ if (rt_mutex_trylock(&q.pi_state->pi_mutex)) ret = 0; else { /* * pi_state is incorrect, some other * task did a lock steal and we * returned due to timeout or signal * without taking the rt_mutex. Too * late. We can access the * rt_mutex_owner without locking, as * the other task is now blocked on * the hash bucket lock. Fix the state * up. */ struct task_struct *owner; int res; owner = rt_mutex_owner(&q.pi_state->pi_mutex); |
1b7558e45
|
1487 1488 |
res = fixup_pi_state_owner(uaddr, &q, owner, fshared); |
cdf71a10c
|
1489 |
|
cdf71a10c
|
1490 1491 1492 1493 |
/* propagate -EFAULT, if the fixup failed */ if (res) ret = res; } |
778e9a9c3
|
1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 |
} else { /* * Paranoia check. If we did not take the lock * in the trylock above, then we should not be * the owner of the rtmutex, neither the real * nor the pending one: */ if (rt_mutex_owner(&q.pi_state->pi_mutex) == curr) printk(KERN_ERR "futex_lock_pi: ret = %d " "pi-mutex: %p pi-state %p ", ret, q.pi_state->pi_mutex.owner, q.pi_state->owner); |
c87e2837b
|
1507 |
} |
c87e2837b
|
1508 |
} |
778e9a9c3
|
1509 1510 |
/* Unqueue and drop the lock */ unqueue_me_pi(&q); |
c87e2837b
|
1511 |
|
237fc6e7a
|
1512 1513 |
if (to) destroy_hrtimer_on_stack(&to->timer); |
c5780e976
|
1514 |
return ret != -EINTR ? ret : -ERESTARTNOINTR; |
c87e2837b
|
1515 |
|
42d35d48c
|
1516 |
out_unlock_put_key: |
c87e2837b
|
1517 |
queue_unlock(&q, hb); |
42d35d48c
|
1518 |
out_put_key: |
38d47c1b7
|
1519 |
put_futex_key(fshared, &q.key); |
42d35d48c
|
1520 |
out: |
237fc6e7a
|
1521 1522 |
if (to) destroy_hrtimer_on_stack(&to->timer); |
c87e2837b
|
1523 |
return ret; |
42d35d48c
|
1524 |
uaddr_faulted: |
c87e2837b
|
1525 |
/* |
b56863630
|
1526 1527 1528 1529 1530 |
* We have to r/w *(int __user *)uaddr, and we have to modify it * atomically. Therefore, if we continue to fault after get_user() * below, we need to handle the fault ourselves, while still holding * the mmap_sem. This can occur if the uaddr is under contention as * we have to drop the mmap_sem in order to call get_user(). |
c87e2837b
|
1531 |
*/ |
778e9a9c3
|
1532 |
queue_unlock(&q, hb); |
c87e2837b
|
1533 |
if (attempt++) { |
c2f9f2015
|
1534 |
ret = futex_handle_fault((unsigned long)uaddr, attempt); |
34f01cc1f
|
1535 |
if (ret) |
42d35d48c
|
1536 |
goto out_put_key; |
778e9a9c3
|
1537 |
goto retry_unlocked; |
c87e2837b
|
1538 |
} |
c87e2837b
|
1539 |
ret = get_user(uval, uaddr); |
b56863630
|
1540 |
if (!ret) |
c87e2837b
|
1541 |
goto retry; |
237fc6e7a
|
1542 1543 |
if (to) destroy_hrtimer_on_stack(&to->timer); |
c87e2837b
|
1544 1545 1546 1547 |
return ret; } /* |
c87e2837b
|
1548 1549 1550 1551 |
* Userspace attempted a TID -> 0 atomic transition, and failed. * This is the in-kernel slowpath: we look up the PI state (if any), * and do the rt-mutex unlock. */ |
c2f9f2015
|
1552 |
static int futex_unlock_pi(u32 __user *uaddr, int fshared) |
c87e2837b
|
1553 1554 1555 1556 |
{ struct futex_hash_bucket *hb; struct futex_q *this, *next; u32 uval; |
ec92d0829
|
1557 |
struct plist_head *head; |
38d47c1b7
|
1558 |
union futex_key key = FUTEX_KEY_INIT; |
c87e2837b
|
1559 1560 1561 1562 1563 1564 1565 1566 |
int ret, attempt = 0; retry: if (get_user(uval, uaddr)) return -EFAULT; /* * We release only a lock we actually own: */ |
b488893a3
|
1567 |
if ((uval & FUTEX_TID_MASK) != task_pid_vnr(current)) |
c87e2837b
|
1568 |
return -EPERM; |
c87e2837b
|
1569 |
|
34f01cc1f
|
1570 |
ret = get_futex_key(uaddr, fshared, &key); |
c87e2837b
|
1571 1572 1573 1574 |
if (unlikely(ret != 0)) goto out; hb = hash_futex(&key); |
778e9a9c3
|
1575 |
retry_unlocked: |
c87e2837b
|
1576 |
spin_lock(&hb->lock); |
c87e2837b
|
1577 1578 1579 1580 1581 |
/* * To avoid races, try to do the TID -> 0 atomic transition * again. If it succeeds then we can return without waking * anyone else up: */ |
36cf3b5c3
|
1582 |
if (!(uval & FUTEX_OWNER_DIED)) |
b488893a3
|
1583 |
uval = cmpxchg_futex_value_locked(uaddr, task_pid_vnr(current), 0); |
36cf3b5c3
|
1584 |
|
c87e2837b
|
1585 1586 1587 1588 1589 1590 1591 |
if (unlikely(uval == -EFAULT)) goto pi_faulted; /* * Rare case: we managed to release the lock atomically, * no need to wake anyone else up: */ |
b488893a3
|
1592 |
if (unlikely(uval == task_pid_vnr(current))) |
c87e2837b
|
1593 1594 1595 1596 1597 1598 1599 |
goto out_unlock; /* * Ok, other tasks may need to be woken up - check waiters * and do the wakeup if necessary: */ head = &hb->chain; |
ec92d0829
|
1600 |
plist_for_each_entry_safe(this, next, head, list) { |
c87e2837b
|
1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 |
if (!match_futex (&this->key, &key)) continue; ret = wake_futex_pi(uaddr, uval, this); /* * The atomic access to the futex value * generated a pagefault, so retry the * user-access and the wakeup: */ if (ret == -EFAULT) goto pi_faulted; goto out_unlock; } /* * No waiters - kernel unlocks the futex: */ |
e3f2ddeac
|
1616 1617 1618 1619 1620 |
if (!(uval & FUTEX_OWNER_DIED)) { ret = unlock_futex_pi(uaddr, uval); if (ret == -EFAULT) goto pi_faulted; } |
c87e2837b
|
1621 1622 1623 |
out_unlock: spin_unlock(&hb->lock); |
38d47c1b7
|
1624 |
put_futex_key(fshared, &key); |
c87e2837b
|
1625 |
|
42d35d48c
|
1626 |
out: |
c87e2837b
|
1627 1628 1629 1630 |
return ret; pi_faulted: /* |
b56863630
|
1631 1632 1633 1634 1635 |
* We have to r/w *(int __user *)uaddr, and we have to modify it * atomically. Therefore, if we continue to fault after get_user() * below, we need to handle the fault ourselves, while still holding * the mmap_sem. This can occur if the uaddr is under contention as * we have to drop the mmap_sem in order to call get_user(). |
c87e2837b
|
1636 |
*/ |
778e9a9c3
|
1637 |
spin_unlock(&hb->lock); |
c87e2837b
|
1638 |
if (attempt++) { |
c2f9f2015
|
1639 |
ret = futex_handle_fault((unsigned long)uaddr, attempt); |
34f01cc1f
|
1640 |
if (ret) |
778e9a9c3
|
1641 |
goto out; |
187226f57
|
1642 |
uval = 0; |
778e9a9c3
|
1643 |
goto retry_unlocked; |
c87e2837b
|
1644 |
} |
c87e2837b
|
1645 |
ret = get_user(uval, uaddr); |
b56863630
|
1646 |
if (!ret) |
c87e2837b
|
1647 |
goto retry; |
1da177e4c
|
1648 1649 |
return ret; } |
0771dfefc
|
1650 1651 1652 1653 1654 1655 1656 |
/* * Support for robust futexes: the kernel cleans up held futexes at * thread exit time. * * Implementation: user-space maintains a per-thread list of locks it * is holding. Upon do_exit(), the kernel carefully walks this list, * and marks all locks that are owned by this thread with the |
c87e2837b
|
1657 |
* FUTEX_OWNER_DIED bit, and wakes up a waiter (if any). The list is |
0771dfefc
|
1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 |
* always manipulated with the lock held, so the list is private and * per-thread. Userspace also maintains a per-thread 'list_op_pending' * field, to allow the kernel to clean up if the thread dies after * acquiring the lock, but just before it could have added itself to * the list. There can only be one such pending lock. */ /** * sys_set_robust_list - set the robust-futex list head of a task * @head: pointer to the list-head * @len: length of the list-head, as userspace expects */ asmlinkage long sys_set_robust_list(struct robust_list_head __user *head, size_t len) { |
a0c1e9073
|
1674 1675 |
if (!futex_cmpxchg_enabled) return -ENOSYS; |
0771dfefc
|
1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 |
/* * The kernel knows only one size for now: */ if (unlikely(len != sizeof(*head))) return -EINVAL; current->robust_list = head; return 0; } /** * sys_get_robust_list - get the robust-futex list head of a task * @pid: pid of the process [zero for current task] * @head_ptr: pointer to a list-head pointer, the kernel fills it in * @len_ptr: pointer to a length field, the kernel fills in the header size */ asmlinkage long |
ba46df984
|
1694 |
sys_get_robust_list(int pid, struct robust_list_head __user * __user *head_ptr, |
0771dfefc
|
1695 1696 |
size_t __user *len_ptr) { |
ba46df984
|
1697 |
struct robust_list_head __user *head; |
0771dfefc
|
1698 |
unsigned long ret; |
c69e8d9c0
|
1699 |
const struct cred *cred = current_cred(), *pcred; |
0771dfefc
|
1700 |
|
a0c1e9073
|
1701 1702 |
if (!futex_cmpxchg_enabled) return -ENOSYS; |
0771dfefc
|
1703 1704 1705 1706 1707 1708 |
if (!pid) head = current->robust_list; else { struct task_struct *p; ret = -ESRCH; |
aaa2a97eb
|
1709 |
rcu_read_lock(); |
228ebcbe6
|
1710 |
p = find_task_by_vpid(pid); |
0771dfefc
|
1711 1712 1713 |
if (!p) goto err_unlock; ret = -EPERM; |
c69e8d9c0
|
1714 1715 1716 |
pcred = __task_cred(p); if (cred->euid != pcred->euid && cred->euid != pcred->uid && |
76aac0e9a
|
1717 |
!capable(CAP_SYS_PTRACE)) |
0771dfefc
|
1718 1719 |
goto err_unlock; head = p->robust_list; |
aaa2a97eb
|
1720 |
rcu_read_unlock(); |
0771dfefc
|
1721 1722 1723 1724 1725 1726 1727 |
} if (put_user(sizeof(*head), len_ptr)) return -EFAULT; return put_user(head, head_ptr); err_unlock: |
aaa2a97eb
|
1728 |
rcu_read_unlock(); |
0771dfefc
|
1729 1730 1731 1732 1733 1734 1735 1736 |
return ret; } /* * Process a futex-list entry, check whether it's owned by the * dying task, and do notification if so: */ |
e3f2ddeac
|
1737 |
int handle_futex_death(u32 __user *uaddr, struct task_struct *curr, int pi) |
0771dfefc
|
1738 |
{ |
e3f2ddeac
|
1739 |
u32 uval, nval, mval; |
0771dfefc
|
1740 |
|
8f17d3a50
|
1741 1742 |
retry: if (get_user(uval, uaddr)) |
0771dfefc
|
1743 |
return -1; |
b488893a3
|
1744 |
if ((uval & FUTEX_TID_MASK) == task_pid_vnr(curr)) { |
0771dfefc
|
1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 |
/* * Ok, this dying thread is truly holding a futex * of interest. Set the OWNER_DIED bit atomically * via cmpxchg, and if the value had FUTEX_WAITERS * set, wake up a waiter (if any). (We have to do a * futex_wake() even if OWNER_DIED is already set - * to handle the rare but possible case of recursive * thread-death.) The rest of the cleanup is done in * userspace. */ |
e3f2ddeac
|
1755 1756 |
mval = (uval & FUTEX_WAITERS) | FUTEX_OWNER_DIED; nval = futex_atomic_cmpxchg_inatomic(uaddr, uval, mval); |
c87e2837b
|
1757 1758 1759 1760 |
if (nval == -EFAULT) return -1; if (nval != uval) |
8f17d3a50
|
1761 |
goto retry; |
0771dfefc
|
1762 |
|
e3f2ddeac
|
1763 1764 1765 1766 |
/* * Wake robust non-PI futexes here. The wakeup of * PI futexes happens in exit_pi_state(): */ |
36cf3b5c3
|
1767 |
if (!pi && (uval & FUTEX_WAITERS)) |
c2f9f2015
|
1768 |
futex_wake(uaddr, 1, 1, FUTEX_BITSET_MATCH_ANY); |
0771dfefc
|
1769 1770 1771 1772 1773 |
} return 0; } /* |
e3f2ddeac
|
1774 1775 1776 |
* Fetch a robust-list pointer. Bit 0 signals PI futexes: */ static inline int fetch_robust_entry(struct robust_list __user **entry, |
ba46df984
|
1777 1778 |
struct robust_list __user * __user *head, int *pi) |
e3f2ddeac
|
1779 1780 |
{ unsigned long uentry; |
ba46df984
|
1781 |
if (get_user(uentry, (unsigned long __user *)head)) |
e3f2ddeac
|
1782 |
return -EFAULT; |
ba46df984
|
1783 |
*entry = (void __user *)(uentry & ~1UL); |
e3f2ddeac
|
1784 1785 1786 1787 1788 1789 |
*pi = uentry & 1; return 0; } /* |
0771dfefc
|
1790 1791 1792 1793 1794 1795 1796 1797 |
* Walk curr->robust_list (very carefully, it's a userspace list!) * and mark any locks found there dead, and notify any waiters. * * We silently return on any sign of list-walking problem. */ void exit_robust_list(struct task_struct *curr) { struct robust_list_head __user *head = curr->robust_list; |
9f96cb1e8
|
1798 1799 |
struct robust_list __user *entry, *next_entry, *pending; unsigned int limit = ROBUST_LIST_LIMIT, pi, next_pi, pip; |
0771dfefc
|
1800 |
unsigned long futex_offset; |
9f96cb1e8
|
1801 |
int rc; |
0771dfefc
|
1802 |
|
a0c1e9073
|
1803 1804 |
if (!futex_cmpxchg_enabled) return; |
0771dfefc
|
1805 1806 1807 1808 |
/* * Fetch the list head (which was registered earlier, via * sys_set_robust_list()): */ |
e3f2ddeac
|
1809 |
if (fetch_robust_entry(&entry, &head->list.next, &pi)) |
0771dfefc
|
1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 |
return; /* * Fetch the relative futex offset: */ if (get_user(futex_offset, &head->futex_offset)) return; /* * Fetch any possibly pending lock-add first, and handle it * if it exists: */ |
e3f2ddeac
|
1820 |
if (fetch_robust_entry(&pending, &head->list_op_pending, &pip)) |
0771dfefc
|
1821 |
return; |
e3f2ddeac
|
1822 |
|
9f96cb1e8
|
1823 |
next_entry = NULL; /* avoid warning with gcc */ |
0771dfefc
|
1824 1825 |
while (entry != &head->list) { /* |
9f96cb1e8
|
1826 1827 1828 1829 1830 |
* Fetch the next entry in the list before calling * handle_futex_death: */ rc = fetch_robust_entry(&next_entry, &entry->next, &next_pi); /* |
0771dfefc
|
1831 |
* A pending lock might already be on the list, so |
c87e2837b
|
1832 |
* don't process it twice: |
0771dfefc
|
1833 1834 |
*/ if (entry != pending) |
ba46df984
|
1835 |
if (handle_futex_death((void __user *)entry + futex_offset, |
e3f2ddeac
|
1836 |
curr, pi)) |
0771dfefc
|
1837 |
return; |
9f96cb1e8
|
1838 |
if (rc) |
0771dfefc
|
1839 |
return; |
9f96cb1e8
|
1840 1841 |
entry = next_entry; pi = next_pi; |
0771dfefc
|
1842 1843 1844 1845 1846 1847 1848 1849 |
/* * Avoid excessively long or circular lists: */ if (!--limit) break; cond_resched(); } |
9f96cb1e8
|
1850 1851 1852 1853 |
if (pending) handle_futex_death((void __user *)pending + futex_offset, curr, pip); |
0771dfefc
|
1854 |
} |
c19384b5b
|
1855 |
long do_futex(u32 __user *uaddr, int op, u32 val, ktime_t *timeout, |
e2970f2fb
|
1856 |
u32 __user *uaddr2, u32 val2, u32 val3) |
1da177e4c
|
1857 |
{ |
1acdac104
|
1858 |
int clockrt, ret = -ENOSYS; |
34f01cc1f
|
1859 |
int cmd = op & FUTEX_CMD_MASK; |
c2f9f2015
|
1860 |
int fshared = 0; |
34f01cc1f
|
1861 1862 |
if (!(op & FUTEX_PRIVATE_FLAG)) |
c2f9f2015
|
1863 |
fshared = 1; |
1da177e4c
|
1864 |
|
1acdac104
|
1865 1866 1867 |
clockrt = op & FUTEX_CLOCK_REALTIME; if (clockrt && cmd != FUTEX_WAIT_BITSET) return -ENOSYS; |
1da177e4c
|
1868 |
|
34f01cc1f
|
1869 |
switch (cmd) { |
1da177e4c
|
1870 |
case FUTEX_WAIT: |
cd689985c
|
1871 1872 |
val3 = FUTEX_BITSET_MATCH_ANY; case FUTEX_WAIT_BITSET: |
1acdac104
|
1873 |
ret = futex_wait(uaddr, fshared, val, timeout, val3, clockrt); |
1da177e4c
|
1874 1875 |
break; case FUTEX_WAKE: |
cd689985c
|
1876 1877 1878 |
val3 = FUTEX_BITSET_MATCH_ANY; case FUTEX_WAKE_BITSET: ret = futex_wake(uaddr, fshared, val, val3); |
1da177e4c
|
1879 |
break; |
1da177e4c
|
1880 |
case FUTEX_REQUEUE: |
34f01cc1f
|
1881 |
ret = futex_requeue(uaddr, fshared, uaddr2, val, val2, NULL); |
1da177e4c
|
1882 1883 |
break; case FUTEX_CMP_REQUEUE: |
34f01cc1f
|
1884 |
ret = futex_requeue(uaddr, fshared, uaddr2, val, val2, &val3); |
1da177e4c
|
1885 |
break; |
4732efbeb
|
1886 |
case FUTEX_WAKE_OP: |
34f01cc1f
|
1887 |
ret = futex_wake_op(uaddr, fshared, uaddr2, val, val2, val3); |
4732efbeb
|
1888 |
break; |
c87e2837b
|
1889 |
case FUTEX_LOCK_PI: |
a0c1e9073
|
1890 1891 |
if (futex_cmpxchg_enabled) ret = futex_lock_pi(uaddr, fshared, val, timeout, 0); |
c87e2837b
|
1892 1893 |
break; case FUTEX_UNLOCK_PI: |
a0c1e9073
|
1894 1895 |
if (futex_cmpxchg_enabled) ret = futex_unlock_pi(uaddr, fshared); |
c87e2837b
|
1896 1897 |
break; case FUTEX_TRYLOCK_PI: |
a0c1e9073
|
1898 1899 |
if (futex_cmpxchg_enabled) ret = futex_lock_pi(uaddr, fshared, 0, timeout, 1); |
c87e2837b
|
1900 |
break; |
1da177e4c
|
1901 1902 1903 1904 1905 |
default: ret = -ENOSYS; } return ret; } |
e2970f2fb
|
1906 |
asmlinkage long sys_futex(u32 __user *uaddr, int op, u32 val, |
1da177e4c
|
1907 |
struct timespec __user *utime, u32 __user *uaddr2, |
e2970f2fb
|
1908 |
u32 val3) |
1da177e4c
|
1909 |
{ |
c19384b5b
|
1910 1911 |
struct timespec ts; ktime_t t, *tp = NULL; |
e2970f2fb
|
1912 |
u32 val2 = 0; |
34f01cc1f
|
1913 |
int cmd = op & FUTEX_CMD_MASK; |
1da177e4c
|
1914 |
|
cd689985c
|
1915 1916 |
if (utime && (cmd == FUTEX_WAIT || cmd == FUTEX_LOCK_PI || cmd == FUTEX_WAIT_BITSET)) { |
c19384b5b
|
1917 |
if (copy_from_user(&ts, utime, sizeof(ts)) != 0) |
1da177e4c
|
1918 |
return -EFAULT; |
c19384b5b
|
1919 |
if (!timespec_valid(&ts)) |
9741ef964
|
1920 |
return -EINVAL; |
c19384b5b
|
1921 1922 |
t = timespec_to_ktime(ts); |
34f01cc1f
|
1923 |
if (cmd == FUTEX_WAIT) |
5a7780e72
|
1924 |
t = ktime_add_safe(ktime_get(), t); |
c19384b5b
|
1925 |
tp = &t; |
1da177e4c
|
1926 1927 |
} /* |
34f01cc1f
|
1928 |
* requeue parameter in 'utime' if cmd == FUTEX_REQUEUE. |
f54f09861
|
1929 |
* number of waiters to wake in 'utime' if cmd == FUTEX_WAKE_OP. |
1da177e4c
|
1930 |
*/ |
f54f09861
|
1931 1932 |
if (cmd == FUTEX_REQUEUE || cmd == FUTEX_CMP_REQUEUE || cmd == FUTEX_WAKE_OP) |
e2970f2fb
|
1933 |
val2 = (u32) (unsigned long) utime; |
1da177e4c
|
1934 |
|
c19384b5b
|
1935 |
return do_futex(uaddr, op, val, tp, uaddr2, val2, val3); |
1da177e4c
|
1936 |
} |
f6d107fb1
|
1937 |
static int __init futex_init(void) |
1da177e4c
|
1938 |
{ |
a0c1e9073
|
1939 |
u32 curval; |
3e4ab747e
|
1940 |
int i; |
95362fa90
|
1941 |
|
a0c1e9073
|
1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 |
/* * This will fail and we want it. Some arch implementations do * runtime detection of the futex_atomic_cmpxchg_inatomic() * functionality. We want to know that before we call in any * of the complex code paths. Also we want to prevent * registration of robust lists in that case. NULL is * guaranteed to fault and we get -EFAULT on functional * implementation, the non functional ones will return * -ENOSYS. */ curval = cmpxchg_futex_value_locked(NULL, 0, 0); if (curval == -EFAULT) futex_cmpxchg_enabled = 1; |
3e4ab747e
|
1955 1956 1957 1958 |
for (i = 0; i < ARRAY_SIZE(futex_queues); i++) { plist_head_init(&futex_queues[i].chain, &futex_queues[i].lock); spin_lock_init(&futex_queues[i].lock); } |
1da177e4c
|
1959 1960 |
return 0; } |
f6d107fb1
|
1961 |
__initcall(futex_init); |