Blame view
fs/eventpoll.c
66.1 KB
2874c5fd2
|
1 |
// SPDX-License-Identifier: GPL-2.0-or-later |
1da177e4c
|
2 |
/* |
5071f97ec
|
3 4 |
* fs/eventpoll.c (Efficient event retrieval implementation) * Copyright (C) 2001,...,2009 Davide Libenzi |
1da177e4c
|
5 |
* |
1da177e4c
|
6 |
* Davide Libenzi <davidel@xmailserver.org> |
1da177e4c
|
7 |
*/ |
1da177e4c
|
8 9 |
#include <linux/init.h> #include <linux/kernel.h> |
174cd4b1e
|
10 |
#include <linux/sched/signal.h> |
1da177e4c
|
11 12 13 14 15 16 17 |
#include <linux/fs.h> #include <linux/file.h> #include <linux/signal.h> #include <linux/errno.h> #include <linux/mm.h> #include <linux/slab.h> #include <linux/poll.h> |
1da177e4c
|
18 19 20 21 22 |
#include <linux/string.h> #include <linux/list.h> #include <linux/hash.h> #include <linux/spinlock.h> #include <linux/syscalls.h> |
1da177e4c
|
23 24 25 26 27 |
#include <linux/rbtree.h> #include <linux/wait.h> #include <linux/eventpoll.h> #include <linux/mount.h> #include <linux/bitops.h> |
144efe3e3
|
28 |
#include <linux/mutex.h> |
da66f7cb0
|
29 |
#include <linux/anon_inodes.h> |
4d7e30d98
|
30 |
#include <linux/device.h> |
7c0f6ba68
|
31 |
#include <linux/uaccess.h> |
1da177e4c
|
32 33 |
#include <asm/io.h> #include <asm/mman.h> |
60063497a
|
34 |
#include <linux/atomic.h> |
138d22b58
|
35 36 |
#include <linux/proc_fs.h> #include <linux/seq_file.h> |
35280bd4a
|
37 |
#include <linux/compat.h> |
ae10b2b4e
|
38 |
#include <linux/rculist.h> |
bf3b9f637
|
39 |
#include <net/busy_poll.h> |
1da177e4c
|
40 |
|
1da177e4c
|
41 42 43 44 |
/* * LOCKING: * There are three level of locking required by epoll : * |
144efe3e3
|
45 |
* 1) epmutex (mutex) |
c7ea76302
|
46 |
* 2) ep->mtx (mutex) |
a218cc491
|
47 |
* 3) ep->lock (rwlock) |
1da177e4c
|
48 49 |
* * The acquire order is the one listed above, from 1 to 3. |
a218cc491
|
50 |
* We need a rwlock (ep->lock) because we manipulate objects |
1da177e4c
|
51 52 53 54 55 56 |
* from inside the poll callback, that might be triggered from * a wake_up() that in turn might be called from IRQ context. * So we can't sleep inside the poll callback and hence we need * a spinlock. During the event transfer loop (from kernel to * user space) we could end up sleeping due a copy_to_user(), so * we need a lock that will allow us to sleep. This lock is a |
d47de16c7
|
57 58 59 60 61 |
* mutex (ep->mtx). It is acquired during the event transfer loop, * during epoll_ctl(EPOLL_CTL_DEL) and during eventpoll_release_file(). * Then we also need a global mutex to serialize eventpoll_release_file() * and ep_free(). * This mutex is acquired by ep_free() during the epoll file |
1da177e4c
|
62 63 |
* cleanup path and it is also acquired by eventpoll_release_file() * if a file has been pushed inside an epoll set and it is then |
bf6a41db7
|
64 |
* close()d without a previous call to epoll_ctl(EPOLL_CTL_DEL). |
22bacca48
|
65 66 67 68 69 70 71 |
* It is also acquired when inserting an epoll fd onto another epoll * fd. We do this so that we walk the epoll tree and ensure that this * insertion does not create a cycle of epoll file descriptors, which * could lead to deadlock. We need a global mutex to prevent two * simultaneous inserts (A into B and B into A) from racing and * constructing a cycle without either insert observing that it is * going to. |
d8805e633
|
72 73 74 75 76 77 78 79 80 |
* It is necessary to acquire multiple "ep->mtx"es at once in the * case when one epoll fd is added to another. In this case, we * always acquire the locks in the order of nesting (i.e. after * epoll_ctl(e1, EPOLL_CTL_ADD, e2), e1->mtx will always be acquired * before e2->mtx). Since we disallow cycles of epoll file * descriptors, this ensures that the mutexes are well-ordered. In * order to communicate this nesting to lockdep, when walking a tree * of epoll file descriptors, we use the current recursion depth as * the lockdep subkey. |
d47de16c7
|
81 |
* It is possible to drop the "ep->mtx" and to use the global |
a218cc491
|
82 |
* mutex "epmutex" (together with "ep->lock") to have it working, |
d47de16c7
|
83 |
* but having "ep->mtx" will make the interface more scalable. |
144efe3e3
|
84 |
* Events that require holding "epmutex" are very rare, while for |
d47de16c7
|
85 86 |
* normal operations the epoll private "ep->mtx" will guarantee * a better scalability. |
1da177e4c
|
87 |
*/ |
1da177e4c
|
88 |
/* Epoll private bits inside the event mask */ |
df0108c5d
|
89 |
#define EP_PRIVATE_BITS (EPOLLWAKEUP | EPOLLONESHOT | EPOLLET | EPOLLEXCLUSIVE) |
1da177e4c
|
90 |
|
a9a08845e
|
91 |
#define EPOLLINOUT_BITS (EPOLLIN | EPOLLOUT) |
b6a515c8a
|
92 |
|
a9a08845e
|
93 |
#define EPOLLEXCLUSIVE_OK_BITS (EPOLLINOUT_BITS | EPOLLERR | EPOLLHUP | \ |
b6a515c8a
|
94 |
EPOLLWAKEUP | EPOLLET | EPOLLEXCLUSIVE) |
5071f97ec
|
95 96 |
/* Maximum number of nesting allowed inside epoll sets */ #define EP_MAX_NESTS 4 |
1da177e4c
|
97 |
|
b611967de
|
98 |
#define EP_MAX_EVENTS (INT_MAX / sizeof(struct epoll_event)) |
d47de16c7
|
99 |
#define EP_UNACTIVE_PTR ((void *) -1L) |
7ef9964e6
|
100 |
#define EP_ITEM_COST (sizeof(struct epitem) + sizeof(struct eppoll_entry)) |
1da177e4c
|
101 102 103 |
struct epoll_filefd { struct file *file; int fd; |
39732ca5a
|
104 |
} __packed; |
1da177e4c
|
105 106 |
/* |
5071f97ec
|
107 108 |
* Structure used to track possible nested calls, for too deep recursions * and loop cycles. |
1da177e4c
|
109 |
*/ |
5071f97ec
|
110 |
struct nested_call_node { |
1da177e4c
|
111 |
struct list_head llink; |
5071f97ec
|
112 |
void *cookie; |
3fe4a975d
|
113 |
void *ctx; |
1da177e4c
|
114 115 116 |
}; /* |
5071f97ec
|
117 118 |
* This structure is used as collector for nested calls, to check for * maximum recursion dept and loop cycles. |
1da177e4c
|
119 |
*/ |
5071f97ec
|
120 121 |
struct nested_calls { struct list_head tasks_call_list; |
1da177e4c
|
122 123 124 125 |
spinlock_t lock; }; /* |
d47de16c7
|
126 127 |
* Each file descriptor added to the eventpoll interface will * have an entry of this type linked to the "rbr" RB tree. |
39732ca5a
|
128 129 |
* Avoid increasing the size of this struct, there can be many thousands * of these on a server and we do not want this to take another cache line. |
d47de16c7
|
130 131 |
*/ struct epitem { |
ae10b2b4e
|
132 133 134 135 136 137 |
union { /* RB tree node links this structure to the eventpoll RB tree */ struct rb_node rbn; /* Used to free the struct epitem */ struct rcu_head rcu; }; |
d47de16c7
|
138 139 140 |
/* List header used to link this structure to the eventpoll ready list */ struct list_head rdllink; |
c7ea76302
|
141 142 143 144 145 |
/* * Works together "struct eventpoll"->ovflist in keeping the * single linked chain of items. */ struct epitem *next; |
d47de16c7
|
146 147 148 149 150 151 152 153 154 155 156 |
/* The file descriptor information this item refers to */ struct epoll_filefd ffd; /* Number of active wait queue attached to poll operations */ int nwait; /* List containing poll wait queues */ struct list_head pwqlist; /* The "container" of this item */ struct eventpoll *ep; |
d47de16c7
|
157 158 |
/* List header used to link this item to the "struct file" items list */ struct list_head fllink; |
4d7e30d98
|
159 |
/* wakeup_source used when EPOLLWAKEUP is set */ |
eea1d5859
|
160 |
struct wakeup_source __rcu *ws; |
4d7e30d98
|
161 |
|
c7ea76302
|
162 163 |
/* The structure that describe the interested events and the source fd */ struct epoll_event event; |
d47de16c7
|
164 165 166 |
}; /* |
1da177e4c
|
167 |
* This structure is stored inside the "private_data" member of the file |
bf6a41db7
|
168 |
* structure and represents the main data structure for the eventpoll |
1da177e4c
|
169 170 171 |
* interface. */ struct eventpoll { |
1da177e4c
|
172 |
/* |
d47de16c7
|
173 174 175 176 |
* This mutex is used to ensure that files are not removed * while epoll is using them. This is held during the event * collection loop, the file cleanup path, the epoll file exit * code and the ctl operations. |
1da177e4c
|
177 |
*/ |
d47de16c7
|
178 |
struct mutex mtx; |
1da177e4c
|
179 180 181 182 183 184 185 186 187 |
/* Wait queue used by sys_epoll_wait() */ wait_queue_head_t wq; /* Wait queue used by file->poll() */ wait_queue_head_t poll_wait; /* List of ready file descriptors */ struct list_head rdllist; |
a218cc491
|
188 189 |
/* Lock which protects rdllist and ovflist */ rwlock_t lock; |
67647d0fb
|
190 |
/* RB tree root used to store monitored fd structs */ |
b2ac2ea62
|
191 |
struct rb_root_cached rbr; |
d47de16c7
|
192 193 194 |
/* * This is a single linked list that chains all the "struct epitem" that |
25985edce
|
195 |
* happened while transferring ready events to userspace w/out |
a218cc491
|
196 |
* holding ->lock. |
d47de16c7
|
197 198 |
*/ struct epitem *ovflist; |
7ef9964e6
|
199 |
|
4d7e30d98
|
200 201 |
/* wakeup_source used when ep_scan_ready_list is running */ struct wakeup_source *ws; |
7ef9964e6
|
202 203 |
/* The user that created the eventpoll descriptor */ struct user_struct *user; |
28d82dc1c
|
204 205 206 207 |
struct file *file; /* used to optimize loop detection check */ |
18306c404
|
208 |
u64 gen; |
bf3b9f637
|
209 210 211 212 213 |
#ifdef CONFIG_NET_RX_BUSY_POLL /* used to track busy poll napi_id */ unsigned int napi_id; #endif |
efcdd350d
|
214 215 216 217 218 |
#ifdef CONFIG_DEBUG_LOCK_ALLOC /* tracks wakeup nests for lockdep validation */ u8 nests; #endif |
1da177e4c
|
219 220 221 222 223 224 225 226 |
}; /* Wait structure used by the poll hooks */ struct eppoll_entry { /* List header used to link this structure to the "struct epitem" */ struct list_head llink; /* The "base" pointer is set to the container "struct epitem" */ |
4f0989dbf
|
227 |
struct epitem *base; |
1da177e4c
|
228 229 230 231 232 |
/* * Wait queue item that will be linked to the target file wait * queue head. */ |
ac6424b98
|
233 |
wait_queue_entry_t wait; |
1da177e4c
|
234 235 236 237 |
/* The wait queue head that linked the "wait" wait queue item */ wait_queue_head_t *whead; }; |
1da177e4c
|
238 239 240 241 242 |
/* Wrapper struct used by poll queueing */ struct ep_pqueue { poll_table pt; struct epitem *epi; }; |
5071f97ec
|
243 244 245 246 |
/* Used by the ep_send_events() function as callback private data */ struct ep_send_events_data { int maxevents; struct epoll_event __user *events; |
d7ebbe46f
|
247 |
int res; |
5071f97ec
|
248 |
}; |
1da177e4c
|
249 |
/* |
7ef9964e6
|
250 251 |
* Configuration options available inside /proc/sys/fs/epoll/ */ |
7ef9964e6
|
252 |
/* Maximum number of epoll watched descriptors, per user */ |
52bd19f76
|
253 |
static long max_user_watches __read_mostly; |
7ef9964e6
|
254 255 |
/* |
d47de16c7
|
256 |
* This mutex is used to serialize ep_free() and eventpoll_release_file(). |
1da177e4c
|
257 |
*/ |
7ef9964e6
|
258 |
static DEFINE_MUTEX(epmutex); |
1da177e4c
|
259 |
|
18306c404
|
260 |
static u64 loop_check_gen = 0; |
22bacca48
|
261 262 |
/* Used to check for epoll file descriptor inclusion loops */ static struct nested_calls poll_loop_ncalls; |
1da177e4c
|
263 |
/* Slab cache used to allocate "struct epitem" */ |
e18b890bb
|
264 |
static struct kmem_cache *epi_cache __read_mostly; |
1da177e4c
|
265 266 |
/* Slab cache used to allocate "struct eppoll_entry" */ |
e18b890bb
|
267 |
static struct kmem_cache *pwq_cache __read_mostly; |
1da177e4c
|
268 |
|
28d82dc1c
|
269 270 271 272 273 |
/* * List of files with newly added links, where we may need to limit the number * of emanating paths. Protected by the epmutex. */ static LIST_HEAD(tfile_check_list); |
7ef9964e6
|
274 275 276 |
#ifdef CONFIG_SYSCTL #include <linux/sysctl.h> |
eec4844fa
|
277 |
static long long_zero; |
52bd19f76
|
278 |
static long long_max = LONG_MAX; |
7ef9964e6
|
279 |
|
1f7e0616c
|
280 |
struct ctl_table epoll_table[] = { |
7ef9964e6
|
281 |
{ |
7ef9964e6
|
282 283 |
.procname = "max_user_watches", .data = &max_user_watches, |
52bd19f76
|
284 |
.maxlen = sizeof(max_user_watches), |
7ef9964e6
|
285 |
.mode = 0644, |
52bd19f76
|
286 |
.proc_handler = proc_doulongvec_minmax, |
eec4844fa
|
287 |
.extra1 = &long_zero, |
52bd19f76
|
288 |
.extra2 = &long_max, |
7ef9964e6
|
289 |
}, |
ab09203e3
|
290 |
{ } |
7ef9964e6
|
291 292 |
}; #endif /* CONFIG_SYSCTL */ |
28d82dc1c
|
293 294 295 296 297 298 |
static const struct file_operations eventpoll_fops; static inline int is_file_epoll(struct file *f) { return f->f_op == &eventpoll_fops; } |
b030a4dd6
|
299 |
|
67647d0fb
|
300 |
/* Setup the structure that is used as key for the RB tree */ |
b030a4dd6
|
301 302 303 304 305 306 |
static inline void ep_set_ffd(struct epoll_filefd *ffd, struct file *file, int fd) { ffd->file = file; ffd->fd = fd; } |
67647d0fb
|
307 |
/* Compare RB tree keys */ |
b030a4dd6
|
308 309 310 311 312 313 |
static inline int ep_cmp_ffd(struct epoll_filefd *p1, struct epoll_filefd *p2) { return (p1->file > p2->file ? +1: (p1->file < p2->file ? -1 : p1->fd - p2->fd)); } |
b030a4dd6
|
314 |
/* Tells us if the item is currently linked */ |
992991c03
|
315 |
static inline int ep_is_linked(struct epitem *epi) |
b030a4dd6
|
316 |
{ |
992991c03
|
317 |
return !list_empty(&epi->rdllink); |
b030a4dd6
|
318 |
} |
ac6424b98
|
319 |
static inline struct eppoll_entry *ep_pwq_from_wait(wait_queue_entry_t *p) |
971316f05
|
320 321 322 |
{ return container_of(p, struct eppoll_entry, wait); } |
b030a4dd6
|
323 |
/* Get the "struct epitem" from a wait queue pointer */ |
ac6424b98
|
324 |
static inline struct epitem *ep_item_from_wait(wait_queue_entry_t *p) |
b030a4dd6
|
325 326 327 328 329 |
{ return container_of(p, struct eppoll_entry, wait)->base; } /* Get the "struct epitem" from an epoll queue wrapper */ |
cdac75e6f
|
330 |
static inline struct epitem *ep_item_from_epqueue(poll_table *p) |
b030a4dd6
|
331 332 333 |
{ return container_of(p, struct ep_pqueue, pt)->epi; } |
1da177e4c
|
334 |
/* Initialize the poll safe wake up structure */ |
5071f97ec
|
335 |
static void ep_nested_calls_init(struct nested_calls *ncalls) |
1da177e4c
|
336 |
{ |
5071f97ec
|
337 338 |
INIT_LIST_HEAD(&ncalls->tasks_call_list); spin_lock_init(&ncalls->lock); |
1da177e4c
|
339 |
} |
5071f97ec
|
340 |
/** |
3fb0e584a
|
341 342 343 344 345 346 347 348 349 |
* ep_events_available - Checks if ready events might be available. * * @ep: Pointer to the eventpoll context. * * Returns: Returns a value different than zero if ready events are available, * or zero otherwise. */ static inline int ep_events_available(struct eventpoll *ep) { |
c5a282e96
|
350 351 |
return !list_empty_careful(&ep->rdllist) || READ_ONCE(ep->ovflist) != EP_UNACTIVE_PTR; |
3fb0e584a
|
352 |
} |
bf3b9f637
|
353 354 355 356 357 358 359 |
#ifdef CONFIG_NET_RX_BUSY_POLL static bool ep_busy_loop_end(void *p, unsigned long start_time) { struct eventpoll *ep = p; return ep_events_available(ep) || busy_loop_timeout(start_time); } |
bf3b9f637
|
360 361 362 363 364 365 366 367 368 |
/* * Busy poll if globally on and supporting sockets found && no events, * busy loop will return if need_resched or ep_events_available. * * we must do our busy polling with irqs enabled */ static void ep_busy_loop(struct eventpoll *ep, int nonblock) { |
bf3b9f637
|
369 370 371 372 |
unsigned int napi_id = READ_ONCE(ep->napi_id); if ((napi_id >= MIN_NAPI_ID) && net_busy_loop_on()) napi_busy_loop(napi_id, nonblock ? NULL : ep_busy_loop_end, ep); |
bf3b9f637
|
373 374 375 376 |
} static inline void ep_reset_busy_poll_napi_id(struct eventpoll *ep) { |
bf3b9f637
|
377 378 |
if (ep->napi_id) ep->napi_id = 0; |
bf3b9f637
|
379 380 381 382 383 384 385 |
} /* * Set epoll busy poll NAPI ID from sk. */ static inline void ep_set_busy_poll_napi_id(struct epitem *epi) { |
bf3b9f637
|
386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 |
struct eventpoll *ep; unsigned int napi_id; struct socket *sock; struct sock *sk; int err; if (!net_busy_loop_on()) return; sock = sock_from_file(epi->ffd.file, &err); if (!sock) return; sk = sock->sk; if (!sk) return; napi_id = READ_ONCE(sk->sk_napi_id); ep = epi->ep; /* Non-NAPI IDs can be rejected * or * Nothing to do if we already have this ID */ if (napi_id < MIN_NAPI_ID || napi_id == ep->napi_id) return; /* record NAPI ID for use in next busy poll */ ep->napi_id = napi_id; |
bf3b9f637
|
415 |
} |
514056d50
|
416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 |
#else static inline void ep_busy_loop(struct eventpoll *ep, int nonblock) { } static inline void ep_reset_busy_poll_napi_id(struct eventpoll *ep) { } static inline void ep_set_busy_poll_napi_id(struct epitem *epi) { } #endif /* CONFIG_NET_RX_BUSY_POLL */ |
3fb0e584a
|
431 |
/** |
5071f97ec
|
432 433 434 435 436 437 |
* ep_call_nested - Perform a bound (possibly) nested call, by checking * that the recursion limit is not exceeded, and that * the same nested call (by the meaning of same cookie) is * no re-entered. * * @ncalls: Pointer to the nested_calls structure to be used for this call. |
5071f97ec
|
438 439 440 |
* @nproc: Nested call core function pointer. * @priv: Opaque data to be passed to the @nproc callback. * @cookie: Cookie to be used to identify this nested call. |
3fe4a975d
|
441 |
* @ctx: This instance context. |
5071f97ec
|
442 443 444 |
* * Returns: Returns the code returned by the @nproc callback, or -1 if * the maximum recursion limit has been exceeded. |
1da177e4c
|
445 |
*/ |
74bdc1298
|
446 |
static int ep_call_nested(struct nested_calls *ncalls, |
5071f97ec
|
447 |
int (*nproc)(void *, void *, int), void *priv, |
3fe4a975d
|
448 |
void *cookie, void *ctx) |
1da177e4c
|
449 |
{ |
5071f97ec
|
450 |
int error, call_nests = 0; |
1da177e4c
|
451 |
unsigned long flags; |
5071f97ec
|
452 453 454 |
struct list_head *lsthead = &ncalls->tasks_call_list; struct nested_call_node *tncur; struct nested_call_node tnode; |
1da177e4c
|
455 |
|
5071f97ec
|
456 |
spin_lock_irqsave(&ncalls->lock, flags); |
1da177e4c
|
457 |
|
5071f97ec
|
458 459 460 461 462 |
/* * Try to see if the current task is already inside this wakeup call. * We use a list here, since the population inside this set is always * very much limited. */ |
b70c39409
|
463 |
list_for_each_entry(tncur, lsthead, llink) { |
3fe4a975d
|
464 |
if (tncur->ctx == ctx && |
74bdc1298
|
465 |
(tncur->cookie == cookie || ++call_nests > EP_MAX_NESTS)) { |
1da177e4c
|
466 467 468 469 |
/* * Ops ... loop detected or maximum nest level reached. * We abort this wake by breaking the cycle itself. */ |
abff55cee
|
470 471 |
error = -1; goto out_unlock; |
1da177e4c
|
472 473 |
} } |
5071f97ec
|
474 |
/* Add the current task and cookie to the list */ |
3fe4a975d
|
475 |
tnode.ctx = ctx; |
5071f97ec
|
476 |
tnode.cookie = cookie; |
1da177e4c
|
477 |
list_add(&tnode.llink, lsthead); |
5071f97ec
|
478 |
spin_unlock_irqrestore(&ncalls->lock, flags); |
1da177e4c
|
479 |
|
5071f97ec
|
480 481 |
/* Call the nested function */ error = (*nproc)(priv, cookie, call_nests); |
1da177e4c
|
482 483 |
/* Remove the current task from the list */ |
5071f97ec
|
484 |
spin_lock_irqsave(&ncalls->lock, flags); |
1da177e4c
|
485 |
list_del(&tnode.llink); |
3fe4a975d
|
486 |
out_unlock: |
5071f97ec
|
487 488 489 490 |
spin_unlock_irqrestore(&ncalls->lock, flags); return error; } |
02edc6fc4
|
491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 |
/* * As described in commit 0ccf831cb lockdep: annotate epoll * the use of wait queues used by epoll is done in a very controlled * manner. Wake ups can nest inside each other, but are never done * with the same locking. For example: * * dfd = socket(...); * efd1 = epoll_create(); * efd2 = epoll_create(); * epoll_ctl(efd1, EPOLL_CTL_ADD, dfd, ...); * epoll_ctl(efd2, EPOLL_CTL_ADD, efd1, ...); * * When a packet arrives to the device underneath "dfd", the net code will * issue a wake_up() on its poll wake list. Epoll (efd1) has installed a * callback wakeup entry on that queue, and the wake_up() performed by the * "dfd" net code will end up in ep_poll_callback(). At this point epoll * (efd1) notices that it may have some event ready, so it needs to wake up * the waiters on its poll wait list (efd2). So it calls ep_poll_safewake() * that ends up in another wake_up(), after having checked about the * recursion constraints. That are, no more than EP_MAX_POLLWAKE_NESTS, to * avoid stack blasting. * * When CONFIG_DEBUG_LOCK_ALLOC is enabled, make sure lockdep can handle * this special case of epoll. */ |
2dfa4eeab
|
516 |
#ifdef CONFIG_DEBUG_LOCK_ALLOC |
57a173bdf
|
517 |
|
efcdd350d
|
518 |
static void ep_poll_safewake(struct eventpoll *ep, struct epitem *epi) |
5071f97ec
|
519 |
{ |
efcdd350d
|
520 |
struct eventpoll *ep_src; |
f6520c520
|
521 |
unsigned long flags; |
efcdd350d
|
522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 |
u8 nests = 0; /* * To set the subclass or nesting level for spin_lock_irqsave_nested() * it might be natural to create a per-cpu nest count. However, since * we can recurse on ep->poll_wait.lock, and a non-raw spinlock can * schedule() in the -rt kernel, the per-cpu variable are no longer * protected. Thus, we are introducing a per eventpoll nest field. * If we are not being call from ep_poll_callback(), epi is NULL and * we are at the first level of nesting, 0. Otherwise, we are being * called from ep_poll_callback() and if a previous wakeup source is * not an epoll file itself, we are at depth 1 since the wakeup source * is depth 0. If the wakeup source is a previous epoll file in the * wakeup chain then we use its nests value and record ours as * nests + 1. The previous epoll file nests value is stable since its * already holding its own poll_wait.lock. */ if (epi) { if ((is_file_epoll(epi->ffd.file))) { ep_src = epi->ffd.file->private_data; nests = ep_src->nests; } else { nests = 1; } } spin_lock_irqsave_nested(&ep->poll_wait.lock, flags, nests); ep->nests = nests + 1; wake_up_locked_poll(&ep->poll_wait, EPOLLIN); ep->nests = 0; spin_unlock_irqrestore(&ep->poll_wait.lock, flags); |
1da177e4c
|
552 |
} |
57a173bdf
|
553 |
#else |
efcdd350d
|
554 |
static void ep_poll_safewake(struct eventpoll *ep, struct epitem *epi) |
57a173bdf
|
555 |
{ |
efcdd350d
|
556 |
wake_up_poll(&ep->poll_wait, EPOLLIN); |
57a173bdf
|
557 558 559 |
} #endif |
971316f05
|
560 561 562 563 564 |
static void ep_remove_wait_queue(struct eppoll_entry *pwq) { wait_queue_head_t *whead; rcu_read_lock(); |
138e4ad67
|
565 566 567 568 569 570 571 |
/* * If it is cleared by POLLFREE, it should be rcu-safe. * If we read NULL we need a barrier paired with * smp_store_release() in ep_poll_callback(), otherwise * we rely on whead->lock. */ whead = smp_load_acquire(&pwq->whead); |
971316f05
|
572 573 574 575 |
if (whead) remove_wait_queue(whead, &pwq->wait); rcu_read_unlock(); } |
1da177e4c
|
576 |
/* |
d1bc90dd5
|
577 578 579 |
* This function unregisters poll callbacks from the associated file * descriptor. Must be called with "mtx" held (or "epmutex" if called from * ep_free). |
1da177e4c
|
580 |
*/ |
7699acd13
|
581 |
static void ep_unregister_pollwait(struct eventpoll *ep, struct epitem *epi) |
1da177e4c
|
582 |
{ |
7699acd13
|
583 584 |
struct list_head *lsthead = &epi->pwqlist; struct eppoll_entry *pwq; |
1da177e4c
|
585 |
|
d1bc90dd5
|
586 587 |
while (!list_empty(lsthead)) { pwq = list_first_entry(lsthead, struct eppoll_entry, llink); |
1da177e4c
|
588 |
|
d1bc90dd5
|
589 |
list_del(&pwq->llink); |
971316f05
|
590 |
ep_remove_wait_queue(pwq); |
d1bc90dd5
|
591 |
kmem_cache_free(pwq_cache, pwq); |
1da177e4c
|
592 |
} |
1da177e4c
|
593 |
} |
eea1d5859
|
594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 |
/* call only when ep->mtx is held */ static inline struct wakeup_source *ep_wakeup_source(struct epitem *epi) { return rcu_dereference_check(epi->ws, lockdep_is_held(&epi->ep->mtx)); } /* call only when ep->mtx is held */ static inline void ep_pm_stay_awake(struct epitem *epi) { struct wakeup_source *ws = ep_wakeup_source(epi); if (ws) __pm_stay_awake(ws); } static inline bool ep_has_wakeup_source(struct epitem *epi) { return rcu_access_pointer(epi->ws) ? true : false; } /* call when ep->mtx cannot be held (ep_poll_callback) */ static inline void ep_pm_stay_awake_rcu(struct epitem *epi) { struct wakeup_source *ws; rcu_read_lock(); ws = rcu_dereference(epi->ws); if (ws) __pm_stay_awake(ws); rcu_read_unlock(); } |
5071f97ec
|
625 626 627 628 629 630 631 632 |
/** * ep_scan_ready_list - Scans the ready list in a way that makes possible for * the scan code, to call f_op->poll(). Also allows for * O(NumReady) performance. * * @ep: Pointer to the epoll private data structure. * @sproc: Pointer to the scan callback. * @priv: Private opaque data passed to the @sproc callback. |
d8805e633
|
633 |
* @depth: The current depth of recursive f_op->poll calls. |
67347fe4e
|
634 |
* @ep_locked: caller already holds ep->mtx |
5071f97ec
|
635 636 637 |
* * Returns: The same integer error code returned by the @sproc callback. */ |
d85e2aa2e
|
638 639 |
static __poll_t ep_scan_ready_list(struct eventpoll *ep, __poll_t (*sproc)(struct eventpoll *, |
5071f97ec
|
640 |
struct list_head *, void *), |
67347fe4e
|
641 |
void *priv, int depth, bool ep_locked) |
5071f97ec
|
642 |
{ |
d85e2aa2e
|
643 |
__poll_t res; |
5071f97ec
|
644 |
struct epitem *epi, *nepi; |
296e236e9
|
645 |
LIST_HEAD(txlist); |
5071f97ec
|
646 |
|
92e641784
|
647 |
lockdep_assert_irqs_enabled(); |
5071f97ec
|
648 649 |
/* * We need to lock this because we could be hit by |
e057e15ff
|
650 |
* eventpoll_release_file() and epoll_ctl(). |
5071f97ec
|
651 |
*/ |
67347fe4e
|
652 653 654 |
if (!ep_locked) mutex_lock_nested(&ep->mtx, depth); |
5071f97ec
|
655 656 657 658 659 660 661 662 663 |
/* * Steal the ready list, and re-init the original one to the * empty list. Also, set ep->ovflist to NULL so that events * happening while looping w/out locks, are not lost. We cannot * have the poll callback to queue directly on ep->rdllist, * because we want the "sproc" callback to be able to do it * in a lockless way. */ |
a218cc491
|
664 |
write_lock_irq(&ep->lock); |
296e236e9
|
665 |
list_splice_init(&ep->rdllist, &txlist); |
c5a282e96
|
666 |
WRITE_ONCE(ep->ovflist, NULL); |
a218cc491
|
667 |
write_unlock_irq(&ep->lock); |
5071f97ec
|
668 669 670 671 |
/* * Now call the callback function. */ |
d85e2aa2e
|
672 |
res = (*sproc)(ep, &txlist, priv); |
5071f97ec
|
673 |
|
a218cc491
|
674 |
write_lock_irq(&ep->lock); |
5071f97ec
|
675 676 677 678 679 |
/* * During the time we spent inside the "sproc" callback, some * other events might have been queued by the poll callback. * We re-insert them inside the main ready-list here. */ |
c5a282e96
|
680 |
for (nepi = READ_ONCE(ep->ovflist); (epi = nepi) != NULL; |
5071f97ec
|
681 682 683 684 685 686 687 |
nepi = epi->next, epi->next = EP_UNACTIVE_PTR) { /* * We need to check if the item is already in the list. * During the "sproc" callback execution time, items are * queued into ->ovflist but the "txlist" might already * contain them, and the list_splice() below takes care of them. */ |
992991c03
|
688 |
if (!ep_is_linked(epi)) { |
c141175d0
|
689 690 691 692 693 |
/* * ->ovflist is LIFO, so we have to reverse it in order * to keep in FIFO. */ list_add(&epi->rdllink, &ep->rdllist); |
eea1d5859
|
694 |
ep_pm_stay_awake(epi); |
4d7e30d98
|
695 |
} |
5071f97ec
|
696 697 698 699 700 701 |
} /* * We need to set back ep->ovflist to EP_UNACTIVE_PTR, so that after * releasing the lock, events will be queued in the normal way inside * ep->rdllist. */ |
c5a282e96
|
702 |
WRITE_ONCE(ep->ovflist, EP_UNACTIVE_PTR); |
5071f97ec
|
703 704 705 706 707 |
/* * Quickly re-inject items left on "txlist". */ list_splice(&txlist, &ep->rdllist); |
4d7e30d98
|
708 |
__pm_relax(ep->ws); |
a218cc491
|
709 |
write_unlock_irq(&ep->lock); |
5071f97ec
|
710 |
|
67347fe4e
|
711 712 |
if (!ep_locked) mutex_unlock(&ep->mtx); |
5071f97ec
|
713 |
|
d85e2aa2e
|
714 |
return res; |
5071f97ec
|
715 |
} |
ae10b2b4e
|
716 717 718 719 720 |
static void epi_rcu_free(struct rcu_head *head) { struct epitem *epi = container_of(head, struct epitem, rcu); kmem_cache_free(epi_cache, epi); } |
1da177e4c
|
721 |
/* |
7699acd13
|
722 |
* Removes a "struct epitem" from the eventpoll RB tree and deallocates |
c7ea76302
|
723 |
* all the associated resources. Must be called with "mtx" held. |
7699acd13
|
724 725 726 |
*/ static int ep_remove(struct eventpoll *ep, struct epitem *epi) { |
7699acd13
|
727 |
struct file *file = epi->ffd.file; |
1da177e4c
|
728 |
|
92e641784
|
729 |
lockdep_assert_irqs_enabled(); |
1da177e4c
|
730 |
/* |
ee8ef0a4b
|
731 |
* Removes poll wait queue hooks. |
1da177e4c
|
732 |
*/ |
7699acd13
|
733 |
ep_unregister_pollwait(ep, epi); |
1da177e4c
|
734 |
|
7699acd13
|
735 |
/* Remove the current item from the list of epoll hooks */ |
684999149
|
736 |
spin_lock(&file->f_lock); |
ae10b2b4e
|
737 |
list_del_rcu(&epi->fllink); |
684999149
|
738 |
spin_unlock(&file->f_lock); |
1da177e4c
|
739 |
|
b2ac2ea62
|
740 |
rb_erase_cached(&epi->rbn, &ep->rbr); |
1da177e4c
|
741 |
|
a218cc491
|
742 |
write_lock_irq(&ep->lock); |
992991c03
|
743 |
if (ep_is_linked(epi)) |
c7ea76302
|
744 |
list_del_init(&epi->rdllink); |
a218cc491
|
745 |
write_unlock_irq(&ep->lock); |
1da177e4c
|
746 |
|
eea1d5859
|
747 |
wakeup_source_unregister(ep_wakeup_source(epi)); |
ae10b2b4e
|
748 749 750 751 752 753 754 755 |
/* * At this point it is safe to free the eventpoll item. Use the union * field epi->rcu, since we are trying to minimize the size of * 'struct epitem'. The 'rbn' field is no longer in use. Protected by * ep->mtx. The rcu read side, reverse_path_check_proc(), does not make * use of the rbn field. */ call_rcu(&epi->rcu, epi_rcu_free); |
1da177e4c
|
756 |
|
52bd19f76
|
757 |
atomic_long_dec(&ep->user->epoll_watches); |
7ef9964e6
|
758 |
|
c7ea76302
|
759 |
return 0; |
1da177e4c
|
760 |
} |
7699acd13
|
761 |
static void ep_free(struct eventpoll *ep) |
1da177e4c
|
762 |
{ |
7699acd13
|
763 764 |
struct rb_node *rbp; struct epitem *epi; |
1da177e4c
|
765 |
|
7699acd13
|
766 767 |
/* We need to release all tasks waiting for these file */ if (waitqueue_active(&ep->poll_wait)) |
efcdd350d
|
768 |
ep_poll_safewake(ep, NULL); |
1da177e4c
|
769 |
|
7699acd13
|
770 771 772 |
/* * We need to lock this because we could be hit by * eventpoll_release_file() while we're freeing the "struct eventpoll". |
d47de16c7
|
773 |
* We do not need to hold "ep->mtx" here because the epoll file |
7699acd13
|
774 775 |
* is on the way to be removed and no one has references to it * anymore. The only hit might come from eventpoll_release_file() but |
25985edce
|
776 |
* holding "epmutex" is sufficient here. |
7699acd13
|
777 778 |
*/ mutex_lock(&epmutex); |
1da177e4c
|
779 780 |
/* |
7699acd13
|
781 |
* Walks through the whole tree by unregistering poll callbacks. |
1da177e4c
|
782 |
*/ |
b2ac2ea62
|
783 |
for (rbp = rb_first_cached(&ep->rbr); rbp; rbp = rb_next(rbp)) { |
7699acd13
|
784 785 786 |
epi = rb_entry(rbp, struct epitem, rbn); ep_unregister_pollwait(ep, epi); |
91cf5ab60
|
787 |
cond_resched(); |
7699acd13
|
788 |
} |
1da177e4c
|
789 790 |
/* |
7699acd13
|
791 792 |
* Walks through the whole tree by freeing each "struct epitem". At this * point we are sure no poll callbacks will be lingering around, and also by |
d47de16c7
|
793 |
* holding "epmutex" we can be sure that no file cleanup code will hit |
a218cc491
|
794 |
* us during this operation. So we can avoid the lock on "ep->lock". |
ddf676c38
|
795 796 |
* We do not need to lock ep->mtx, either, we only do it to prevent * a lockdep warning. |
1da177e4c
|
797 |
*/ |
ddf676c38
|
798 |
mutex_lock(&ep->mtx); |
b2ac2ea62
|
799 |
while ((rbp = rb_first_cached(&ep->rbr)) != NULL) { |
7699acd13
|
800 801 |
epi = rb_entry(rbp, struct epitem, rbn); ep_remove(ep, epi); |
91cf5ab60
|
802 |
cond_resched(); |
7699acd13
|
803 |
} |
ddf676c38
|
804 |
mutex_unlock(&ep->mtx); |
1da177e4c
|
805 |
|
7699acd13
|
806 |
mutex_unlock(&epmutex); |
d47de16c7
|
807 |
mutex_destroy(&ep->mtx); |
7ef9964e6
|
808 |
free_uid(ep->user); |
4d7e30d98
|
809 |
wakeup_source_unregister(ep->ws); |
f0ee9aabb
|
810 |
kfree(ep); |
7699acd13
|
811 |
} |
1da177e4c
|
812 |
|
7699acd13
|
813 814 815 |
static int ep_eventpoll_release(struct inode *inode, struct file *file) { struct eventpoll *ep = file->private_data; |
1da177e4c
|
816 |
|
f0ee9aabb
|
817 |
if (ep) |
7699acd13
|
818 |
ep_free(ep); |
7699acd13
|
819 |
|
7699acd13
|
820 |
return 0; |
1da177e4c
|
821 |
} |
d85e2aa2e
|
822 |
static __poll_t ep_read_events_proc(struct eventpoll *ep, struct list_head *head, |
37b5e5212
|
823 824 825 826 827 828 829 830 831 |
void *priv); static void ep_ptable_queue_proc(struct file *file, wait_queue_head_t *whead, poll_table *pt); /* * Differs from ep_eventpoll_poll() in that internal callers already have * the ep->mtx so we need to start from depth=1, such that mutex_lock_nested() * is correctly annotated. */ |
d85e2aa2e
|
832 |
static __poll_t ep_item_poll(const struct epitem *epi, poll_table *pt, |
bec1a502d
|
833 |
int depth) |
450d89ec0
|
834 |
{ |
37b5e5212
|
835 836 |
struct eventpoll *ep; bool locked; |
450d89ec0
|
837 |
pt->_key = epi->event.events; |
37b5e5212
|
838 |
if (!is_file_epoll(epi->ffd.file)) |
9965ed174
|
839 |
return vfs_poll(epi->ffd.file, pt) & epi->event.events; |
450d89ec0
|
840 |
|
37b5e5212
|
841 842 843 |
ep = epi->ffd.file->private_data; poll_wait(epi->ffd.file, &ep->poll_wait, pt); locked = pt && (pt->_qproc == ep_ptable_queue_proc); |
450d89ec0
|
844 |
|
37b5e5212
|
845 846 847 |
return ep_scan_ready_list(epi->ffd.file->private_data, ep_read_events_proc, &depth, depth, locked) & epi->event.events; |
450d89ec0
|
848 |
} |
d85e2aa2e
|
849 |
static __poll_t ep_read_events_proc(struct eventpoll *ep, struct list_head *head, |
296e236e9
|
850 |
void *priv) |
5071f97ec
|
851 852 |
{ struct epitem *epi, *tmp; |
626cf2366
|
853 |
poll_table pt; |
37b5e5212
|
854 |
int depth = *(int *)priv; |
5071f97ec
|
855 |
|
626cf2366
|
856 |
init_poll_funcptr(&pt, NULL); |
37b5e5212
|
857 |
depth++; |
450d89ec0
|
858 |
|
5071f97ec
|
859 |
list_for_each_entry_safe(epi, tmp, head, rdllink) { |
37b5e5212
|
860 |
if (ep_item_poll(epi, &pt, depth)) { |
a9a08845e
|
861 |
return EPOLLIN | EPOLLRDNORM; |
37b5e5212
|
862 |
} else { |
5071f97ec
|
863 864 865 866 867 |
/* * Item has been dropped into the ready list by the poll * callback, but it's not actually ready, as far as * caller requested events goes. We can remove it here. */ |
eea1d5859
|
868 |
__pm_relax(ep_wakeup_source(epi)); |
5071f97ec
|
869 |
list_del_init(&epi->rdllink); |
296e236e9
|
870 |
} |
5071f97ec
|
871 872 873 874 |
} return 0; } |
a11e1d432
|
875 |
static __poll_t ep_eventpoll_poll(struct file *file, poll_table *wait) |
11c5ad0ec
|
876 877 878 |
{ struct eventpoll *ep = file->private_data; int depth = 0; |
7699acd13
|
879 |
|
a11e1d432
|
880 881 |
/* Insert inside our poll wait queue */ poll_wait(file, &ep->poll_wait, wait); |
5071f97ec
|
882 883 |
/* * Proceed to find out if wanted events are really available inside |
37b5e5212
|
884 |
* the ready list. |
5071f97ec
|
885 |
*/ |
37b5e5212
|
886 887 |
return ep_scan_ready_list(ep, ep_read_events_proc, &depth, depth, false); |
7699acd13
|
888 |
} |
138d22b58
|
889 |
#ifdef CONFIG_PROC_FS |
a3816ab0e
|
890 |
static void ep_show_fdinfo(struct seq_file *m, struct file *f) |
138d22b58
|
891 892 893 |
{ struct eventpoll *ep = f->private_data; struct rb_node *rbp; |
138d22b58
|
894 895 |
mutex_lock(&ep->mtx); |
b2ac2ea62
|
896 |
for (rbp = rb_first_cached(&ep->rbr); rbp; rbp = rb_next(rbp)) { |
138d22b58
|
897 |
struct epitem *epi = rb_entry(rbp, struct epitem, rbn); |
77493f04b
|
898 |
struct inode *inode = file_inode(epi->ffd.file); |
138d22b58
|
899 |
|
77493f04b
|
900 901 902 |
seq_printf(m, "tfd: %8d events: %8x data: %16llx " " pos:%lli ino:%lx sdev:%x ", |
a3816ab0e
|
903 |
epi->ffd.fd, epi->event.events, |
77493f04b
|
904 905 906 |
(long long)epi->event.data, (long long)epi->ffd.file->f_pos, inode->i_ino, inode->i_sb->s_dev); |
a3816ab0e
|
907 |
if (seq_has_overflowed(m)) |
138d22b58
|
908 909 910 |
break; } mutex_unlock(&ep->mtx); |
138d22b58
|
911 912 |
} #endif |
7699acd13
|
913 914 |
/* File callbacks that implement the eventpoll file behaviour */ static const struct file_operations eventpoll_fops = { |
138d22b58
|
915 916 917 |
#ifdef CONFIG_PROC_FS .show_fdinfo = ep_show_fdinfo, #endif |
7699acd13
|
918 |
.release = ep_eventpoll_release, |
a11e1d432
|
919 |
.poll = ep_eventpoll_poll, |
6038f373a
|
920 |
.llseek = noop_llseek, |
7699acd13
|
921 |
}; |
b611967de
|
922 |
/* |
7699acd13
|
923 924 925 |
* This is called from eventpoll_release() to unlink files from the eventpoll * interface. We need to have this facility to cleanup correctly files that are * closed without being removed from the eventpoll interface. |
b611967de
|
926 |
*/ |
7699acd13
|
927 |
void eventpoll_release_file(struct file *file) |
b611967de
|
928 |
{ |
7699acd13
|
929 |
struct eventpoll *ep; |
ebe06187b
|
930 |
struct epitem *epi, *next; |
b611967de
|
931 932 |
/* |
684999149
|
933 |
* We don't want to get "file->f_lock" because it is not |
7699acd13
|
934 |
* necessary. It is not necessary because we're in the "struct file" |
25985edce
|
935 |
* cleanup path, and this means that no one is using this file anymore. |
5071f97ec
|
936 |
* So, for example, epoll_ctl() cannot hit here since if we reach this |
67647d0fb
|
937 |
* point, the file counter already went to zero and fget() would fail. |
d47de16c7
|
938 |
* The only hit might come from ep_free() but by holding the mutex |
7699acd13
|
939 |
* will correctly serialize the operation. We do need to acquire |
d47de16c7
|
940 |
* "ep->mtx" after "epmutex" because ep_remove() requires it when called |
7699acd13
|
941 |
* from anywhere but ep_free(). |
684999149
|
942 943 |
* * Besides, ep_remove() acquires the lock, so we can't hold it here. |
b611967de
|
944 |
*/ |
7699acd13
|
945 |
mutex_lock(&epmutex); |
ebe06187b
|
946 |
list_for_each_entry_safe(epi, next, &file->f_ep_links, fllink) { |
7699acd13
|
947 |
ep = epi->ep; |
d8805e633
|
948 |
mutex_lock_nested(&ep->mtx, 0); |
7699acd13
|
949 |
ep_remove(ep, epi); |
d47de16c7
|
950 |
mutex_unlock(&ep->mtx); |
b611967de
|
951 |
} |
7699acd13
|
952 |
mutex_unlock(&epmutex); |
b611967de
|
953 |
} |
53d2be79d
|
954 |
static int ep_alloc(struct eventpoll **pep) |
1da177e4c
|
955 |
{ |
7ef9964e6
|
956 957 958 |
int error; struct user_struct *user; struct eventpoll *ep; |
1da177e4c
|
959 |
|
7ef9964e6
|
960 |
user = get_current_user(); |
7ef9964e6
|
961 962 963 964 |
error = -ENOMEM; ep = kzalloc(sizeof(*ep), GFP_KERNEL); if (unlikely(!ep)) goto free_uid; |
1da177e4c
|
965 |
|
d47de16c7
|
966 |
mutex_init(&ep->mtx); |
a218cc491
|
967 |
rwlock_init(&ep->lock); |
1da177e4c
|
968 969 970 |
init_waitqueue_head(&ep->wq); init_waitqueue_head(&ep->poll_wait); INIT_LIST_HEAD(&ep->rdllist); |
b2ac2ea62
|
971 |
ep->rbr = RB_ROOT_CACHED; |
d47de16c7
|
972 |
ep->ovflist = EP_UNACTIVE_PTR; |
7ef9964e6
|
973 |
ep->user = user; |
1da177e4c
|
974 |
|
53d2be79d
|
975 |
*pep = ep; |
1da177e4c
|
976 |
|
1da177e4c
|
977 |
return 0; |
7ef9964e6
|
978 979 980 981 |
free_uid: free_uid(user); return error; |
1da177e4c
|
982 |
} |
1da177e4c
|
983 |
/* |
c7ea76302
|
984 985 986 |
* Search the file inside the eventpoll tree. The RB tree operations * are protected by the "mtx" mutex, and ep_find() must be called with * "mtx" held. |
1da177e4c
|
987 988 989 990 |
*/ static struct epitem *ep_find(struct eventpoll *ep, struct file *file, int fd) { int kcmp; |
1da177e4c
|
991 992 993 |
struct rb_node *rbp; struct epitem *epi, *epir = NULL; struct epoll_filefd ffd; |
b030a4dd6
|
994 |
ep_set_ffd(&ffd, file, fd); |
b2ac2ea62
|
995 |
for (rbp = ep->rbr.rb_root.rb_node; rbp; ) { |
1da177e4c
|
996 |
epi = rb_entry(rbp, struct epitem, rbn); |
b030a4dd6
|
997 |
kcmp = ep_cmp_ffd(&ffd, &epi->ffd); |
1da177e4c
|
998 999 1000 1001 1002 |
if (kcmp > 0) rbp = rbp->rb_right; else if (kcmp < 0) rbp = rbp->rb_left; else { |
1da177e4c
|
1003 1004 1005 1006 |
epir = epi; break; } } |
1da177e4c
|
1007 |
|
1da177e4c
|
1008 1009 |
return epir; } |
92ef6da3d
|
1010 |
#ifdef CONFIG_CHECKPOINT_RESTORE |
0791e3644
|
1011 1012 1013 1014 |
static struct epitem *ep_find_tfd(struct eventpoll *ep, int tfd, unsigned long toff) { struct rb_node *rbp; struct epitem *epi; |
b2ac2ea62
|
1015 |
for (rbp = rb_first_cached(&ep->rbr); rbp; rbp = rb_next(rbp)) { |
0791e3644
|
1016 1017 1018 1019 1020 1021 1022 1023 1024 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 |
epi = rb_entry(rbp, struct epitem, rbn); if (epi->ffd.fd == tfd) { if (toff == 0) return epi; else toff--; } cond_resched(); } return NULL; } struct file *get_epoll_tfile_raw_ptr(struct file *file, int tfd, unsigned long toff) { struct file *file_raw; struct eventpoll *ep; struct epitem *epi; if (!is_file_epoll(file)) return ERR_PTR(-EINVAL); ep = file->private_data; mutex_lock(&ep->mtx); epi = ep_find_tfd(ep, tfd, toff); if (epi) file_raw = epi->ffd.file; else file_raw = ERR_PTR(-ENOENT); mutex_unlock(&ep->mtx); return file_raw; } |
92ef6da3d
|
1051 |
#endif /* CONFIG_CHECKPOINT_RESTORE */ |
0791e3644
|
1052 |
|
a218cc491
|
1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 |
/** * Adds a new entry to the tail of the list in a lockless way, i.e. * multiple CPUs are allowed to call this function concurrently. * * Beware: it is necessary to prevent any other modifications of the * existing list until all changes are completed, in other words * concurrent list_add_tail_lockless() calls should be protected * with a read lock, where write lock acts as a barrier which * makes sure all list_add_tail_lockless() calls are fully * completed. * * Also an element can be locklessly added to the list only in one * direction i.e. either to the tail either to the head, otherwise * concurrent access will corrupt the list. * * Returns %false if element has been already added to the list, %true * otherwise. */ static inline bool list_add_tail_lockless(struct list_head *new, struct list_head *head) { struct list_head *prev; /* * This is simple 'new->next = head' operation, but cmpxchg() * is used in order to detect that same element has been just * added to the list from another CPU: the winner observes * new->next == new. */ if (cmpxchg(&new->next, new, head) != new) return false; /* * Initially ->next of a new element must be updated with the head * (we are inserting to the tail) and only then pointers are atomically * exchanged. XCHG guarantees memory ordering, thus ->next should be * updated before pointers are actually swapped and pointers are * swapped before prev->next is updated. */ prev = xchg(&head->prev, new); /* * It is safe to modify prev->next and new->prev, because a new element * is added only to the tail and new->next is updated before XCHG. */ prev->next = new; new->prev = prev; return true; } /** * Chains a new epi entry to the tail of the ep->ovflist in a lockless way, * i.e. multiple CPUs are allowed to call this function concurrently. * * Returns %false if epi element has been already chained, %true otherwise. */ static inline bool chain_epi_lockless(struct epitem *epi) { struct eventpoll *ep = epi->ep; |
0c54a6a44
|
1115 1116 1117 |
/* Fast preliminary check */ if (epi->next != EP_UNACTIVE_PTR) return false; |
a218cc491
|
1118 1119 1120 1121 1122 1123 1124 1125 1126 |
/* Check that the same epi has not been just chained from another CPU */ if (cmpxchg(&epi->next, EP_UNACTIVE_PTR, NULL) != EP_UNACTIVE_PTR) return false; /* Atomically exchange tail */ epi->next = xchg(&ep->ovflist, epi); return true; } |
1da177e4c
|
1127 |
/* |
7699acd13
|
1128 |
* This is the callback that is passed to the wait queue wakeup |
bf6a41db7
|
1129 |
* mechanism. It is called by the stored file descriptors when they |
7699acd13
|
1130 |
* have events to report. |
a218cc491
|
1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 |
* * This callback takes a read lock in order not to content with concurrent * events from another file descriptors, thus all modifications to ->rdllist * or ->ovflist are lockless. Read lock is paired with the write lock from * ep_scan_ready_list(), which stops all list modifications and guarantees * that lists state is seen correctly. * * Another thing worth to mention is that ep_poll_callback() can be called * concurrently for the same @epi from different CPUs if poll table was inited * with several wait queues entries. Plural wakeup from different CPUs of a * single wait queue is serialized by wq.lock, but the case when multiple wait * queues are used should be detected accordingly. This is detected using * cmpxchg() operation. |
1da177e4c
|
1144 |
*/ |
ac6424b98
|
1145 |
static int ep_poll_callback(wait_queue_entry_t *wait, unsigned mode, int sync, void *key) |
1da177e4c
|
1146 |
{ |
7699acd13
|
1147 |
int pwake = 0; |
7699acd13
|
1148 1149 |
struct epitem *epi = ep_item_from_wait(wait); struct eventpoll *ep = epi->ep; |
3ad6f93e9
|
1150 |
__poll_t pollflags = key_to_poll(key); |
a218cc491
|
1151 |
unsigned long flags; |
df0108c5d
|
1152 |
int ewake = 0; |
1da177e4c
|
1153 |
|
a218cc491
|
1154 |
read_lock_irqsave(&ep->lock, flags); |
1da177e4c
|
1155 |
|
bf3b9f637
|
1156 |
ep_set_busy_poll_napi_id(epi); |
7699acd13
|
1157 1158 1159 1160 1161 1162 1163 |
/* * If the event mask does not contain any poll(2) event, we consider the * descriptor to be disabled. This condition is likely the effect of the * EPOLLONESHOT bit that disables the descriptor when an event is received, * until the next EPOLL_CTL_MOD will be issued. */ if (!(epi->event.events & ~EP_PRIVATE_BITS)) |
d47de16c7
|
1164 1165 1166 |
goto out_unlock; /* |
2dfa4eeab
|
1167 1168 1169 1170 1171 |
* Check the events coming with the callback. At this stage, not * every device reports the events in the "key" parameter of the * callback. We need to be able to handle both cases here, hence the * test for "key" != NULL before the event match test. */ |
3ad6f93e9
|
1172 |
if (pollflags && !(pollflags & epi->event.events)) |
2dfa4eeab
|
1173 1174 1175 |
goto out_unlock; /* |
bf6a41db7
|
1176 |
* If we are transferring events to userspace, we can hold no locks |
d47de16c7
|
1177 |
* (because we're accessing user memory, and because of linux f_op->poll() |
bf6a41db7
|
1178 |
* semantics). All the events that happen during that period of time are |
d47de16c7
|
1179 1180 |
* chained in ep->ovflist and requeued later on. */ |
c5a282e96
|
1181 |
if (READ_ONCE(ep->ovflist) != EP_UNACTIVE_PTR) { |
0c54a6a44
|
1182 1183 1184 1185 1186 |
if (chain_epi_lockless(epi)) ep_pm_stay_awake_rcu(epi); } else if (!ep_is_linked(epi)) { /* In the usual case, add event to ready list. */ if (list_add_tail_lockless(&epi->rdllink, &ep->rdllist)) |
c3e320b61
|
1187 |
ep_pm_stay_awake_rcu(epi); |
4d7e30d98
|
1188 |
} |
7699acd13
|
1189 |
|
7699acd13
|
1190 1191 1192 1193 |
/* * Wake up ( if active ) both the eventpoll wait list and the ->poll() * wait list. */ |
df0108c5d
|
1194 |
if (waitqueue_active(&ep->wq)) { |
b6a515c8a
|
1195 |
if ((epi->event.events & EPOLLEXCLUSIVE) && |
3ad6f93e9
|
1196 1197 |
!(pollflags & POLLFREE)) { switch (pollflags & EPOLLINOUT_BITS) { |
a9a08845e
|
1198 1199 |
case EPOLLIN: if (epi->event.events & EPOLLIN) |
b6a515c8a
|
1200 1201 |
ewake = 1; break; |
a9a08845e
|
1202 1203 |
case EPOLLOUT: if (epi->event.events & EPOLLOUT) |
b6a515c8a
|
1204 1205 1206 1207 1208 1209 1210 |
ewake = 1; break; case 0: ewake = 1; break; } } |
a218cc491
|
1211 |
wake_up(&ep->wq); |
df0108c5d
|
1212 |
} |
7699acd13
|
1213 1214 |
if (waitqueue_active(&ep->poll_wait)) pwake++; |
d47de16c7
|
1215 |
out_unlock: |
a218cc491
|
1216 |
read_unlock_irqrestore(&ep->lock, flags); |
1da177e4c
|
1217 |
|
7699acd13
|
1218 1219 |
/* We have to call this outside the lock */ if (pwake) |
efcdd350d
|
1220 |
ep_poll_safewake(ep, epi); |
7699acd13
|
1221 |
|
138e4ad67
|
1222 1223 |
if (!(epi->event.events & EPOLLEXCLUSIVE)) ewake = 1; |
3ad6f93e9
|
1224 |
if (pollflags & POLLFREE) { |
138e4ad67
|
1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 |
/* * If we race with ep_remove_wait_queue() it can miss * ->whead = NULL and do another remove_wait_queue() after * us, so we can't use __remove_wait_queue(). */ list_del_init(&wait->entry); /* * ->whead != NULL protects us from the race with ep_free() * or ep_remove(), ep_remove_wait_queue() takes whead->lock * held by the caller. Once we nullify it, nothing protects * ep/epi or even wait. */ smp_store_release(&ep_pwq_from_wait(wait)->whead, NULL); } |
df0108c5d
|
1239 |
|
138e4ad67
|
1240 |
return ewake; |
7699acd13
|
1241 |
} |
1da177e4c
|
1242 1243 1244 1245 1246 1247 1248 1249 |
/* * This is the callback that is used to add our wait queue to the * target file wakeup lists. */ static void ep_ptable_queue_proc(struct file *file, wait_queue_head_t *whead, poll_table *pt) { |
b030a4dd6
|
1250 |
struct epitem *epi = ep_item_from_epqueue(pt); |
1da177e4c
|
1251 |
struct eppoll_entry *pwq; |
e94b17660
|
1252 |
if (epi->nwait >= 0 && (pwq = kmem_cache_alloc(pwq_cache, GFP_KERNEL))) { |
1da177e4c
|
1253 1254 1255 |
init_waitqueue_func_entry(&pwq->wait, ep_poll_callback); pwq->whead = whead; pwq->base = epi; |
df0108c5d
|
1256 1257 1258 1259 |
if (epi->event.events & EPOLLEXCLUSIVE) add_wait_queue_exclusive(whead, &pwq->wait); else add_wait_queue(whead, &pwq->wait); |
1da177e4c
|
1260 1261 |
list_add_tail(&pwq->llink, &epi->pwqlist); epi->nwait++; |
296e236e9
|
1262 |
} else { |
1da177e4c
|
1263 1264 |
/* We have to signal that an error occurred */ epi->nwait = -1; |
296e236e9
|
1265 |
} |
1da177e4c
|
1266 |
} |
1da177e4c
|
1267 1268 1269 |
static void ep_rbtree_insert(struct eventpoll *ep, struct epitem *epi) { int kcmp; |
b2ac2ea62
|
1270 |
struct rb_node **p = &ep->rbr.rb_root.rb_node, *parent = NULL; |
1da177e4c
|
1271 |
struct epitem *epic; |
b2ac2ea62
|
1272 |
bool leftmost = true; |
1da177e4c
|
1273 1274 1275 1276 |
while (*p) { parent = *p; epic = rb_entry(parent, struct epitem, rbn); |
b030a4dd6
|
1277 |
kcmp = ep_cmp_ffd(&epi->ffd, &epic->ffd); |
b2ac2ea62
|
1278 |
if (kcmp > 0) { |
1da177e4c
|
1279 |
p = &parent->rb_right; |
b2ac2ea62
|
1280 1281 |
leftmost = false; } else |
1da177e4c
|
1282 1283 1284 |
p = &parent->rb_left; } rb_link_node(&epi->rbn, parent, p); |
b2ac2ea62
|
1285 |
rb_insert_color_cached(&epi->rbn, &ep->rbr, leftmost); |
1da177e4c
|
1286 |
} |
a80a6b85b
|
1287 |
|
28d82dc1c
|
1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 |
#define PATH_ARR_SIZE 5 /* * These are the number paths of length 1 to 5, that we are allowing to emanate * from a single file of interest. For example, we allow 1000 paths of length * 1, to emanate from each file of interest. This essentially represents the * potential wakeup paths, which need to be limited in order to avoid massive * uncontrolled wakeup storms. The common use case should be a single ep which * is connected to n file sources. In this case each file source has 1 path * of length 1. Thus, the numbers below should be more than sufficient. These * path limits are enforced during an EPOLL_CTL_ADD operation, since a modify * and delete can't add additional paths. Protected by the epmutex. */ static const int path_limits[PATH_ARR_SIZE] = { 1000, 500, 100, 50, 10 }; static int path_count[PATH_ARR_SIZE]; static int path_count_inc(int nests) { |
93dc6107a
|
1305 1306 1307 |
/* Allow an arbitrary number of depth 1 paths */ if (nests == 0) return 0; |
28d82dc1c
|
1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 |
if (++path_count[nests] > path_limits[nests]) return -1; return 0; } static void path_count_init(void) { int i; for (i = 0; i < PATH_ARR_SIZE; i++) path_count[i] = 0; } static int reverse_path_check_proc(void *priv, void *cookie, int call_nests) { int error = 0; struct file *file = priv; struct file *child_file; struct epitem *epi; |
ae10b2b4e
|
1327 1328 1329 |
/* CTL_DEL can remove links here, but that can't increase our count */ rcu_read_lock(); list_for_each_entry_rcu(epi, &file->f_ep_links, fllink) { |
28d82dc1c
|
1330 1331 1332 1333 1334 1335 1336 1337 1338 |
child_file = epi->ep->file; if (is_file_epoll(child_file)) { if (list_empty(&child_file->f_ep_links)) { if (path_count_inc(call_nests)) { error = -1; break; } } else { error = ep_call_nested(&poll_loop_ncalls, |
28d82dc1c
|
1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 |
reverse_path_check_proc, child_file, child_file, current); } if (error != 0) break; } else { printk(KERN_ERR "reverse_path_check_proc: " "file is not an ep! "); } } |
ae10b2b4e
|
1351 |
rcu_read_unlock(); |
28d82dc1c
|
1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 |
return error; } /** * reverse_path_check - The tfile_check_list is list of file *, which have * links that are proposed to be newly added. We need to * make sure that those added links don't add too many * paths such that we will spend all our time waking up * eventpoll objects. * * Returns: Returns zero if the proposed links don't create too many paths, * -1 otherwise. */ static int reverse_path_check(void) { |
28d82dc1c
|
1367 1368 1369 1370 1371 |
int error = 0; struct file *current_file; /* let's call this for all tfiles */ list_for_each_entry(current_file, &tfile_check_list, f_tfile_llink) { |
28d82dc1c
|
1372 |
path_count_init(); |
74bdc1298
|
1373 |
error = ep_call_nested(&poll_loop_ncalls, |
28d82dc1c
|
1374 1375 1376 1377 1378 1379 1380 |
reverse_path_check_proc, current_file, current_file, current); if (error) break; } return error; } |
4d7e30d98
|
1381 1382 |
static int ep_create_wakeup_source(struct epitem *epi) { |
3701cb59d
|
1383 |
struct name_snapshot n; |
eea1d5859
|
1384 |
struct wakeup_source *ws; |
4d7e30d98
|
1385 1386 |
if (!epi->ep->ws) { |
c8377adfa
|
1387 |
epi->ep->ws = wakeup_source_register(NULL, "eventpoll"); |
4d7e30d98
|
1388 1389 1390 |
if (!epi->ep->ws) return -ENOMEM; } |
3701cb59d
|
1391 1392 1393 |
take_dentry_name_snapshot(&n, epi->ffd.file->f_path.dentry); ws = wakeup_source_register(NULL, n.name.name); release_dentry_name_snapshot(&n); |
eea1d5859
|
1394 1395 |
if (!ws) |
4d7e30d98
|
1396 |
return -ENOMEM; |
eea1d5859
|
1397 |
rcu_assign_pointer(epi->ws, ws); |
4d7e30d98
|
1398 1399 1400 |
return 0; } |
eea1d5859
|
1401 1402 |
/* rare code path, only used when EPOLL_CTL_MOD removes a wakeup source */ static noinline void ep_destroy_wakeup_source(struct epitem *epi) |
4d7e30d98
|
1403 |
{ |
eea1d5859
|
1404 |
struct wakeup_source *ws = ep_wakeup_source(epi); |
d6d67e723
|
1405 |
RCU_INIT_POINTER(epi->ws, NULL); |
eea1d5859
|
1406 1407 1408 1409 1410 1411 1412 1413 |
/* * wait for ep_pm_stay_awake_rcu to finish, synchronize_rcu is * used internally by wakeup_source_remove, too (called by * wakeup_source_unregister), so we cannot use call_rcu */ synchronize_rcu(); wakeup_source_unregister(ws); |
4d7e30d98
|
1414 |
} |
c7ea76302
|
1415 1416 1417 |
/* * Must be called with "mtx" held. */ |
bec1a502d
|
1418 |
static int ep_insert(struct eventpoll *ep, const struct epoll_event *event, |
67347fe4e
|
1419 |
struct file *tfile, int fd, int full_check) |
1da177e4c
|
1420 |
{ |
d85e2aa2e
|
1421 1422 |
int error, pwake = 0; __poll_t revents; |
52bd19f76
|
1423 |
long user_watches; |
1da177e4c
|
1424 1425 |
struct epitem *epi; struct ep_pqueue epq; |
92e641784
|
1426 |
lockdep_assert_irqs_enabled(); |
52bd19f76
|
1427 1428 |
user_watches = atomic_long_read(&ep->user->epoll_watches); if (unlikely(user_watches >= max_user_watches)) |
7ef9964e6
|
1429 |
return -ENOSPC; |
e94b17660
|
1430 |
if (!(epi = kmem_cache_alloc(epi_cache, GFP_KERNEL))) |
7ef9964e6
|
1431 |
return -ENOMEM; |
1da177e4c
|
1432 1433 |
/* Item initialization follow here ... */ |
1da177e4c
|
1434 1435 |
INIT_LIST_HEAD(&epi->rdllink); INIT_LIST_HEAD(&epi->fllink); |
1da177e4c
|
1436 1437 |
INIT_LIST_HEAD(&epi->pwqlist); epi->ep = ep; |
b030a4dd6
|
1438 |
ep_set_ffd(&epi->ffd, tfile, fd); |
1da177e4c
|
1439 |
epi->event = *event; |
1da177e4c
|
1440 |
epi->nwait = 0; |
d47de16c7
|
1441 |
epi->next = EP_UNACTIVE_PTR; |
4d7e30d98
|
1442 1443 1444 1445 1446 |
if (epi->event.events & EPOLLWAKEUP) { error = ep_create_wakeup_source(epi); if (error) goto error_create_wakeup_source; } else { |
eea1d5859
|
1447 |
RCU_INIT_POINTER(epi->ws, NULL); |
4d7e30d98
|
1448 |
} |
1da177e4c
|
1449 |
|
f8d4f44df
|
1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 |
/* Add the current item to the list of active epoll hook for this file */ spin_lock(&tfile->f_lock); list_add_tail_rcu(&epi->fllink, &tfile->f_ep_links); spin_unlock(&tfile->f_lock); /* * Add the current item to the RB tree. All RB tree operations are * protected by "mtx", and ep_insert() is called with "mtx" held. */ ep_rbtree_insert(ep, epi); /* now check if we've created too many backpaths */ error = -EINVAL; if (full_check && reverse_path_check()) goto error_remove_epi; |
1da177e4c
|
1465 1466 1467 1468 1469 1470 1471 |
/* Initialize the poll table using the queue callback */ epq.epi = epi; init_poll_funcptr(&epq.pt, ep_ptable_queue_proc); /* * Attach the item to the poll hooks and get current event bits. * We can safely use the file* here because its usage count has |
c7ea76302
|
1472 1473 1474 |
* been increased by the caller of this function. Note that after * this operation completes, the poll callback can start hitting * the new item. |
1da177e4c
|
1475 |
*/ |
37b5e5212
|
1476 |
revents = ep_item_poll(epi, &epq.pt, 1); |
1da177e4c
|
1477 1478 1479 1480 1481 1482 |
/* * We have to check if something went wrong during the poll wait queue * install process. Namely an allocation for a wait queue failed due * high memory pressure. */ |
7ef9964e6
|
1483 |
error = -ENOMEM; |
1da177e4c
|
1484 |
if (epi->nwait < 0) |
7699acd13
|
1485 |
goto error_unregister; |
1da177e4c
|
1486 |
|
c7ea76302
|
1487 |
/* We have to drop the new item inside our item list to keep track of it */ |
a218cc491
|
1488 |
write_lock_irq(&ep->lock); |
c7ea76302
|
1489 |
|
bf3b9f637
|
1490 1491 |
/* record NAPI ID of new item if present */ ep_set_busy_poll_napi_id(epi); |
1da177e4c
|
1492 |
/* If the file is already "ready" we drop it inside the ready list */ |
992991c03
|
1493 |
if (revents && !ep_is_linked(epi)) { |
1da177e4c
|
1494 |
list_add_tail(&epi->rdllink, &ep->rdllist); |
eea1d5859
|
1495 |
ep_pm_stay_awake(epi); |
1da177e4c
|
1496 1497 1498 |
/* Notify waiting tasks that events are available */ if (waitqueue_active(&ep->wq)) |
a218cc491
|
1499 |
wake_up(&ep->wq); |
1da177e4c
|
1500 1501 1502 |
if (waitqueue_active(&ep->poll_wait)) pwake++; } |
a218cc491
|
1503 |
write_unlock_irq(&ep->lock); |
1da177e4c
|
1504 |
|
52bd19f76
|
1505 |
atomic_long_inc(&ep->user->epoll_watches); |
7ef9964e6
|
1506 |
|
1da177e4c
|
1507 1508 |
/* We have to call this outside the lock */ if (pwake) |
efcdd350d
|
1509 |
ep_poll_safewake(ep, NULL); |
1da177e4c
|
1510 |
|
1da177e4c
|
1511 |
return 0; |
f8d4f44df
|
1512 1513 |
error_unregister: ep_unregister_pollwait(ep, epi); |
28d82dc1c
|
1514 1515 |
error_remove_epi: spin_lock(&tfile->f_lock); |
ae10b2b4e
|
1516 |
list_del_rcu(&epi->fllink); |
28d82dc1c
|
1517 |
spin_unlock(&tfile->f_lock); |
b2ac2ea62
|
1518 |
rb_erase_cached(&epi->rbn, &ep->rbr); |
28d82dc1c
|
1519 |
|
1da177e4c
|
1520 1521 |
/* * We need to do this because an event could have been arrived on some |
67647d0fb
|
1522 1523 1524 |
* allocated wait queue. Note that we don't care about the ep->ovflist * list, since that is used/cleaned only inside a section bound by "mtx". * And ep_insert() is called with "mtx" held. |
1da177e4c
|
1525 |
*/ |
a218cc491
|
1526 |
write_lock_irq(&ep->lock); |
992991c03
|
1527 |
if (ep_is_linked(epi)) |
6192bd536
|
1528 |
list_del_init(&epi->rdllink); |
a218cc491
|
1529 |
write_unlock_irq(&ep->lock); |
1da177e4c
|
1530 |
|
eea1d5859
|
1531 |
wakeup_source_unregister(ep_wakeup_source(epi)); |
4d7e30d98
|
1532 1533 |
error_create_wakeup_source: |
b030a4dd6
|
1534 |
kmem_cache_free(epi_cache, epi); |
7ef9964e6
|
1535 |
|
1da177e4c
|
1536 1537 |
return error; } |
1da177e4c
|
1538 1539 |
/* * Modify the interest event mask by dropping an event if the new mask |
c7ea76302
|
1540 |
* has a match in the current file status. Must be called with "mtx" held. |
1da177e4c
|
1541 |
*/ |
bec1a502d
|
1542 1543 |
static int ep_modify(struct eventpoll *ep, struct epitem *epi, const struct epoll_event *event) |
1da177e4c
|
1544 1545 |
{ int pwake = 0; |
626cf2366
|
1546 |
poll_table pt; |
92e641784
|
1547 |
lockdep_assert_irqs_enabled(); |
626cf2366
|
1548 |
init_poll_funcptr(&pt, NULL); |
1da177e4c
|
1549 1550 |
/* |
e057e15ff
|
1551 1552 1553 |
* Set the new event interest mask before calling f_op->poll(); * otherwise we might miss an event that happens between the * f_op->poll() call and the new event set registering. |
1da177e4c
|
1554 |
*/ |
128dd1759
|
1555 |
epi->event.events = event->events; /* need barrier below */ |
e057e15ff
|
1556 |
epi->event.data = event->data; /* protected by mtx */ |
4d7e30d98
|
1557 |
if (epi->event.events & EPOLLWAKEUP) { |
eea1d5859
|
1558 |
if (!ep_has_wakeup_source(epi)) |
4d7e30d98
|
1559 |
ep_create_wakeup_source(epi); |
eea1d5859
|
1560 |
} else if (ep_has_wakeup_source(epi)) { |
4d7e30d98
|
1561 1562 |
ep_destroy_wakeup_source(epi); } |
1da177e4c
|
1563 1564 |
/* |
128dd1759
|
1565 1566 1567 1568 1569 |
* The following barrier has two effects: * * 1) Flush epi changes above to other CPUs. This ensures * we do not miss events from ep_poll_callback if an * event occurs immediately after we call f_op->poll(). |
a218cc491
|
1570 |
* We need this because we did not take ep->lock while |
128dd1759
|
1571 |
* changing epi above (but ep_poll_callback does take |
a218cc491
|
1572 |
* ep->lock). |
128dd1759
|
1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 |
* * 2) We also need to ensure we do not miss _past_ events * when calling f_op->poll(). This barrier also * pairs with the barrier in wq_has_sleeper (see * comments for wq_has_sleeper). * * This barrier will now guarantee ep_poll_callback or f_op->poll * (or both) will notice the readiness of an item. */ smp_mb(); /* |
1da177e4c
|
1585 1586 |
* Get current event bits. We can safely use the file* here because * its usage count has been increased by the caller of this function. |
c7ea76302
|
1587 |
* If the item is "hot" and it is not registered inside the ready |
67647d0fb
|
1588 |
* list, push it inside. |
1da177e4c
|
1589 |
*/ |
69112736e
|
1590 |
if (ep_item_poll(epi, &pt, 1)) { |
a218cc491
|
1591 |
write_lock_irq(&ep->lock); |
992991c03
|
1592 |
if (!ep_is_linked(epi)) { |
c7ea76302
|
1593 |
list_add_tail(&epi->rdllink, &ep->rdllist); |
eea1d5859
|
1594 |
ep_pm_stay_awake(epi); |
c7ea76302
|
1595 1596 1597 |
/* Notify waiting tasks that events are available */ if (waitqueue_active(&ep->wq)) |
a218cc491
|
1598 |
wake_up(&ep->wq); |
c7ea76302
|
1599 1600 |
if (waitqueue_active(&ep->poll_wait)) pwake++; |
7699acd13
|
1601 |
} |
a218cc491
|
1602 |
write_unlock_irq(&ep->lock); |
7699acd13
|
1603 |
} |
1da177e4c
|
1604 |
|
7699acd13
|
1605 1606 |
/* We have to call this outside the lock */ if (pwake) |
efcdd350d
|
1607 |
ep_poll_safewake(ep, NULL); |
1da177e4c
|
1608 |
|
7699acd13
|
1609 |
return 0; |
1da177e4c
|
1610 |
} |
d85e2aa2e
|
1611 |
static __poll_t ep_send_events_proc(struct eventpoll *ep, struct list_head *head, |
296e236e9
|
1612 |
void *priv) |
1da177e4c
|
1613 |
{ |
5071f97ec
|
1614 |
struct ep_send_events_data *esed = priv; |
d85e2aa2e
|
1615 |
__poll_t revents; |
4e0982a00
|
1616 1617 |
struct epitem *epi, *tmp; struct epoll_event __user *uevent = esed->events; |
eea1d5859
|
1618 |
struct wakeup_source *ws; |
626cf2366
|
1619 1620 1621 |
poll_table pt; init_poll_funcptr(&pt, NULL); |
4e0982a00
|
1622 |
esed->res = 0; |
1da177e4c
|
1623 |
|
296e236e9
|
1624 |
/* |
5071f97ec
|
1625 1626 1627 |
* We can loop without lock because we are passed a task private list. * Items cannot vanish during the loop because ep_scan_ready_list() is * holding "mtx" during this call. |
296e236e9
|
1628 |
*/ |
21877e1a5
|
1629 |
lockdep_assert_held(&ep->mtx); |
4e0982a00
|
1630 1631 1632 |
list_for_each_entry_safe(epi, tmp, head, rdllink) { if (esed->res >= esed->maxevents) break; |
d47de16c7
|
1633 |
|
4d7e30d98
|
1634 1635 1636 1637 1638 1639 1640 1641 1642 |
/* * Activate ep->ws before deactivating epi->ws to prevent * triggering auto-suspend here (in case we reactive epi->ws * below). * * This could be rearranged to delay the deactivation of epi->ws * instead, but then epi->ws would temporarily be out of sync * with ep_is_linked(). */ |
eea1d5859
|
1643 1644 1645 1646 1647 1648 |
ws = ep_wakeup_source(epi); if (ws) { if (ws->active) __pm_stay_awake(ep->ws); __pm_relax(ws); } |
d47de16c7
|
1649 |
list_del_init(&epi->rdllink); |
1da177e4c
|
1650 |
|
296e236e9
|
1651 |
/* |
5071f97ec
|
1652 1653 |
* If the event mask intersect the caller-requested one, * deliver the event to userspace. Again, ep_scan_ready_list() |
4e0982a00
|
1654 |
* is holding ep->mtx, so no operations coming from userspace |
5071f97ec
|
1655 |
* can change the item. |
296e236e9
|
1656 |
*/ |
4e0982a00
|
1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 |
revents = ep_item_poll(epi, &pt, 1); if (!revents) continue; if (__put_user(revents, &uevent->events) || __put_user(epi->event.data, &uevent->data)) { list_add(&epi->rdllink, head); ep_pm_stay_awake(epi); if (!esed->res) esed->res = -EFAULT; return 0; } esed->res++; uevent++; if (epi->event.events & EPOLLONESHOT) epi->event.events &= EP_PRIVATE_BITS; else if (!(epi->event.events & EPOLLET)) { /* * If this file has been added with Level * Trigger mode, we need to insert back inside * the ready list, so that the next call to * epoll_wait() will check again the events * availability. At this point, no one can insert * into ep->rdllist besides us. The epoll_ctl() * callers are locked out by * ep_scan_ready_list() holding "mtx" and the * poll callback will queue them in ep->ovflist. */ list_add_tail(&epi->rdllink, &ep->rdllist); ep_pm_stay_awake(epi); |
296e236e9
|
1687 1688 |
} } |
5071f97ec
|
1689 |
|
d7ebbe46f
|
1690 |
return 0; |
5071f97ec
|
1691 |
} |
d47de16c7
|
1692 |
|
296e236e9
|
1693 1694 |
static int ep_send_events(struct eventpoll *ep, struct epoll_event __user *events, int maxevents) |
5071f97ec
|
1695 1696 |
{ struct ep_send_events_data esed; |
1da177e4c
|
1697 |
|
5071f97ec
|
1698 1699 |
esed.maxevents = maxevents; esed.events = events; |
6192bd536
|
1700 |
|
d7ebbe46f
|
1701 1702 |
ep_scan_ready_list(ep, ep_send_events_proc, &esed, 0, false); return esed.res; |
1da177e4c
|
1703 |
} |
766b9f928
|
1704 |
static inline struct timespec64 ep_set_mstimeout(long ms) |
0781b909b
|
1705 |
{ |
766b9f928
|
1706 |
struct timespec64 now, ts = { |
0781b909b
|
1707 1708 1709 |
.tv_sec = ms / MSEC_PER_SEC, .tv_nsec = NSEC_PER_MSEC * (ms % MSEC_PER_SEC), }; |
766b9f928
|
1710 1711 |
ktime_get_ts64(&now); return timespec64_add_safe(now, ts); |
0781b909b
|
1712 |
} |
f4d93ad74
|
1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 |
/** * ep_poll - Retrieves ready events, and delivers them to the caller supplied * event buffer. * * @ep: Pointer to the eventpoll context. * @events: Pointer to the userspace buffer where the ready events should be * stored. * @maxevents: Size (in terms of number of events) of the caller event buffer. * @timeout: Maximum timeout for the ready events fetch operation, in * milliseconds. If the @timeout is zero, the function will not block, * while if the @timeout is less than zero, the function will block * until at least one event has been retrieved (or an error * occurred). * * Returns: Returns the number of ready events which have been fetched, or an * error code, in case of error. */ |
1da177e4c
|
1730 1731 1732 |
static int ep_poll(struct eventpoll *ep, struct epoll_event __user *events, int maxevents, long timeout) { |
f4d93ad74
|
1733 |
int res = 0, eavail, timed_out = 0; |
da8b44d5a
|
1734 |
u64 slack = 0; |
ac6424b98
|
1735 |
wait_queue_entry_t wait; |
95aac7b1c
|
1736 |
ktime_t expires, *to = NULL; |
679abf381
|
1737 |
lockdep_assert_irqs_enabled(); |
95aac7b1c
|
1738 |
if (timeout > 0) { |
766b9f928
|
1739 |
struct timespec64 end_time = ep_set_mstimeout(timeout); |
0781b909b
|
1740 |
|
95aac7b1c
|
1741 1742 |
slack = select_estimate_accuracy(&end_time); to = &expires; |
766b9f928
|
1743 |
*to = timespec64_to_ktime(end_time); |
95aac7b1c
|
1744 |
} else if (timeout == 0) { |
f4d93ad74
|
1745 1746 |
/* * Avoid the unnecessary trip to the wait queue loop, if the |
c5a282e96
|
1747 1748 1749 1750 |
* caller specified a non blocking operation. We still need * lock because we could race and not see an epi being added * to the ready list while in irq callback. Thus incorrectly * returning 0 back to userspace. |
f4d93ad74
|
1751 |
*/ |
95aac7b1c
|
1752 |
timed_out = 1; |
c5a282e96
|
1753 |
|
a218cc491
|
1754 |
write_lock_irq(&ep->lock); |
c5a282e96
|
1755 |
eavail = ep_events_available(ep); |
a218cc491
|
1756 |
write_unlock_irq(&ep->lock); |
c5a282e96
|
1757 |
|
35cff1a6e
|
1758 |
goto send_events; |
95aac7b1c
|
1759 |
} |
1da177e4c
|
1760 |
|
f4d93ad74
|
1761 |
fetch_events: |
bf3b9f637
|
1762 1763 1764 |
if (!ep_events_available(ep)) ep_busy_loop(ep, timed_out); |
c5a282e96
|
1765 1766 |
eavail = ep_events_available(ep); if (eavail) |
35cff1a6e
|
1767 |
goto send_events; |
1da177e4c
|
1768 |
|
c5a282e96
|
1769 1770 1771 1772 1773 1774 |
/* * Busy poll timed out. Drop NAPI ID for now, we can add * it back in when we have moved a socket with a valid NAPI * ID onto the ready list. */ ep_reset_busy_poll_napi_id(ep); |
bf3b9f637
|
1775 |
|
412895f03
|
1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 |
do { /* * Internally init_wait() uses autoremove_wake_function(), * thus wait entry is removed from the wait queue on each * wakeup. Why it is important? In case of several waiters * each new wakeup will hit the next waiter, giving it the * chance to harvest new event. Otherwise wakeup can be * lost. This is also good performance-wise, because on * normal wakeup path no need to call __remove_wait_queue() * explicitly, thus ep->lock is not taken, which halts the * event delivery. */ init_wait(&wait); |
1da177e4c
|
1789 |
|
65759097d
|
1790 |
write_lock_irq(&ep->lock); |
1da177e4c
|
1791 |
/* |
65759097d
|
1792 1793 1794 |
* Barrierless variant, waitqueue_active() is called under * the same lock on wakeup ep_poll_callback() side, so it * is safe to avoid an explicit barrier. |
bf3b9f637
|
1795 |
*/ |
65759097d
|
1796 |
__set_current_state(TASK_INTERRUPTIBLE); |
bf3b9f637
|
1797 |
/* |
65759097d
|
1798 1799 1800 1801 1802 |
* Do the final check under the lock. ep_scan_ready_list() * plays with two lists (->rdllist and ->ovflist) and there * is always a race when both lists are empty for short * period of time although events are pending, so lock is * important. |
1da177e4c
|
1803 |
*/ |
65759097d
|
1804 1805 1806 1807 1808 1809 |
eavail = ep_events_available(ep); if (!eavail) { if (signal_pending(current)) res = -EINTR; else __add_wait_queue_exclusive(&ep->wq, &wait); |
c5a282e96
|
1810 |
} |
65759097d
|
1811 |
write_unlock_irq(&ep->lock); |
95aac7b1c
|
1812 |
|
1afb979cd
|
1813 1814 1815 |
if (!eavail && !res) timed_out = !schedule_hrtimeout_range(to, slack, HRTIMER_MODE_ABS); |
1da177e4c
|
1816 |
|
1afb979cd
|
1817 1818 1819 1820 1821 |
/* * We were woken up, thus go and try to harvest some events. * If timed out and still on the wait queue, recheck eavail * carefully under lock, below. */ |
412895f03
|
1822 |
eavail = 1; |
412895f03
|
1823 |
} while (0); |
1da177e4c
|
1824 |
|
c5a282e96
|
1825 |
__set_current_state(TASK_RUNNING); |
1da177e4c
|
1826 |
|
412895f03
|
1827 1828 |
if (!list_empty_careful(&wait.entry)) { write_lock_irq(&ep->lock); |
1afb979cd
|
1829 1830 1831 1832 1833 1834 1835 1836 |
/* * If the thread timed out and is not on the wait queue, it * means that the thread was woken up after its timeout expired * before it could reacquire the lock. Thus, when wait.entry is * empty, it needs to harvest events. */ if (timed_out) eavail = list_empty(&wait.entry); |
412895f03
|
1837 1838 1839 |
__remove_wait_queue(&ep->wq, &wait); write_unlock_irq(&ep->lock); } |
35cff1a6e
|
1840 |
send_events: |
65759097d
|
1841 1842 1843 1844 1845 1846 1847 1848 1849 |
if (fatal_signal_pending(current)) { /* * Always short-circuit for fatal signals to allow * threads to make a timely exit without the chance of * finding more events available and fetching * repeatedly. */ res = -EINTR; } |
1da177e4c
|
1850 1851 1852 1853 1854 1855 |
/* * Try to transfer events to user space. In case we get 0 events and * there's still timeout left over, we go trying again in search of * more luck. */ if (!res && eavail && |
95aac7b1c
|
1856 |
!(res = ep_send_events(ep, events, maxevents)) && !timed_out) |
f4d93ad74
|
1857 |
goto fetch_events; |
1da177e4c
|
1858 1859 1860 |
return res; } |
22bacca48
|
1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 |
/** * ep_loop_check_proc - Callback function to be passed to the @ep_call_nested() * API, to verify that adding an epoll file inside another * epoll structure, does not violate the constraints, in * terms of closed loops, or too deep chains (which can * result in excessive stack usage). * * @priv: Pointer to the epoll file to be currently checked. * @cookie: Original cookie for this call. This is the top-of-the-chain epoll * data structure pointer. * @call_nests: Current dept of the @ep_call_nested() call stack. * * Returns: Returns zero if adding the epoll @file inside current epoll * structure @ep does not violate the constraints, or -1 otherwise. */ static int ep_loop_check_proc(void *priv, void *cookie, int call_nests) { int error = 0; struct file *file = priv; struct eventpoll *ep = file->private_data; |
28d82dc1c
|
1881 |
struct eventpoll *ep_tovisit; |
22bacca48
|
1882 1883 |
struct rb_node *rbp; struct epitem *epi; |
d8805e633
|
1884 |
mutex_lock_nested(&ep->mtx, call_nests + 1); |
18306c404
|
1885 |
ep->gen = loop_check_gen; |
b2ac2ea62
|
1886 |
for (rbp = rb_first_cached(&ep->rbr); rbp; rbp = rb_next(rbp)) { |
22bacca48
|
1887 1888 |
epi = rb_entry(rbp, struct epitem, rbn); if (unlikely(is_file_epoll(epi->ffd.file))) { |
28d82dc1c
|
1889 |
ep_tovisit = epi->ffd.file->private_data; |
18306c404
|
1890 |
if (ep_tovisit->gen == loop_check_gen) |
28d82dc1c
|
1891 |
continue; |
74bdc1298
|
1892 |
error = ep_call_nested(&poll_loop_ncalls, |
28d82dc1c
|
1893 1894 |
ep_loop_check_proc, epi->ffd.file, ep_tovisit, current); |
22bacca48
|
1895 1896 |
if (error != 0) break; |
28d82dc1c
|
1897 1898 1899 1900 1901 1902 1903 1904 1905 |
} else { /* * If we've reached a file that is not associated with * an ep, then we need to check if the newly added * links are going to add too many wakeup paths. We do * this by adding it to the tfile_check_list, if it's * not already there, and calling reverse_path_check() * during ep_insert(). */ |
a9ed4a656
|
1906 |
if (list_empty(&epi->ffd.file->f_tfile_llink)) { |
77f4689de
|
1907 1908 1909 |
if (get_file_rcu(epi->ffd.file)) list_add(&epi->ffd.file->f_tfile_llink, &tfile_check_list); |
a9ed4a656
|
1910 |
} |
22bacca48
|
1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 |
} } mutex_unlock(&ep->mtx); return error; } /** * ep_loop_check - Performs a check to verify that adding an epoll file (@file) * another epoll file (represented by @ep) does not create * closed loops or too deep chains. * * @ep: Pointer to the epoll private data structure. * @file: Pointer to the epoll file to be checked. * * Returns: Returns zero if adding the epoll @file inside current epoll * structure @ep does not violate the constraints, or -1 otherwise. */ static int ep_loop_check(struct eventpoll *ep, struct file *file) { |
18306c404
|
1931 |
return ep_call_nested(&poll_loop_ncalls, |
22bacca48
|
1932 |
ep_loop_check_proc, file, ep, current); |
28d82dc1c
|
1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 |
} static void clear_tfile_check_list(void) { struct file *file; /* first clear the tfile_check_list */ while (!list_empty(&tfile_check_list)) { file = list_first_entry(&tfile_check_list, struct file, f_tfile_llink); list_del_init(&file->f_tfile_llink); |
a9ed4a656
|
1944 |
fput(file); |
28d82dc1c
|
1945 1946 |
} INIT_LIST_HEAD(&tfile_check_list); |
22bacca48
|
1947 |
} |
7699acd13
|
1948 |
/* |
523723bb5
|
1949 |
* Open an eventpoll file descriptor. |
7699acd13
|
1950 |
*/ |
791eb22ee
|
1951 |
static int do_epoll_create(int flags) |
7699acd13
|
1952 |
{ |
28d82dc1c
|
1953 |
int error, fd; |
bb57c3edc
|
1954 |
struct eventpoll *ep = NULL; |
28d82dc1c
|
1955 |
struct file *file; |
7699acd13
|
1956 |
|
e38b36f32
|
1957 1958 |
/* Check the EPOLL_* constant for consistency. */ BUILD_BUG_ON(EPOLL_CLOEXEC != O_CLOEXEC); |
296e236e9
|
1959 1960 |
if (flags & ~EPOLL_CLOEXEC) return -EINVAL; |
7699acd13
|
1961 |
/* |
bb57c3edc
|
1962 |
* Create the internal data structure ("struct eventpoll"). |
7699acd13
|
1963 |
*/ |
9fe5ad9c8
|
1964 |
error = ep_alloc(&ep); |
bb57c3edc
|
1965 1966 |
if (error < 0) return error; |
7699acd13
|
1967 1968 |
/* * Creates all the items needed to setup an eventpoll file. That is, |
2030a42ce
|
1969 |
* a file structure and a free file descriptor. |
7699acd13
|
1970 |
*/ |
28d82dc1c
|
1971 1972 1973 1974 1975 1976 |
fd = get_unused_fd_flags(O_RDWR | (flags & O_CLOEXEC)); if (fd < 0) { error = fd; goto out_free_ep; } file = anon_inode_getfile("[eventpoll]", &eventpoll_fops, ep, |
628ff7c1d
|
1977 |
O_RDWR | (flags & O_CLOEXEC)); |
28d82dc1c
|
1978 1979 1980 1981 |
if (IS_ERR(file)) { error = PTR_ERR(file); goto out_free_fd; } |
28d82dc1c
|
1982 |
ep->file = file; |
98022748f
|
1983 |
fd_install(fd, file); |
28d82dc1c
|
1984 1985 1986 1987 1988 1989 |
return fd; out_free_fd: put_unused_fd(fd); out_free_ep: ep_free(ep); |
bb57c3edc
|
1990 |
return error; |
7699acd13
|
1991 |
} |
791eb22ee
|
1992 1993 1994 1995 |
SYSCALL_DEFINE1(epoll_create1, int, flags) { return do_epoll_create(flags); } |
5a8a82b1d
|
1996 |
SYSCALL_DEFINE1(epoll_create, int, size) |
a0998b50c
|
1997 |
{ |
bfe3891a5
|
1998 |
if (size <= 0) |
9fe5ad9c8
|
1999 |
return -EINVAL; |
791eb22ee
|
2000 |
return do_epoll_create(0); |
a0998b50c
|
2001 |
} |
39220e8d4
|
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 |
static inline int epoll_mutex_lock(struct mutex *mutex, int depth, bool nonblock) { if (!nonblock) { mutex_lock_nested(mutex, depth); return 0; } if (mutex_trylock(mutex)) return 0; return -EAGAIN; } int do_epoll_ctl(int epfd, int op, int fd, struct epoll_event *epds, bool nonblock) |
7699acd13
|
2016 2017 |
{ int error; |
67347fe4e
|
2018 |
int full_check = 0; |
7e3fb5842
|
2019 |
struct fd f, tf; |
7699acd13
|
2020 2021 |
struct eventpoll *ep; struct epitem *epi; |
67347fe4e
|
2022 |
struct eventpoll *tep = NULL; |
7699acd13
|
2023 |
|
7699acd13
|
2024 |
error = -EBADF; |
7e3fb5842
|
2025 2026 |
f = fdget(epfd); if (!f.file) |
7699acd13
|
2027 2028 2029 |
goto error_return; /* Get the "struct file *" for the target file */ |
7e3fb5842
|
2030 2031 |
tf = fdget(fd); if (!tf.file) |
7699acd13
|
2032 2033 2034 2035 |
goto error_fput; /* The target file descriptor must support poll */ error = -EPERM; |
9965ed174
|
2036 |
if (!file_can_poll(tf.file)) |
7699acd13
|
2037 |
goto error_tgt_fput; |
4d7e30d98
|
2038 |
/* Check if EPOLLWAKEUP is allowed */ |
c680e41b3
|
2039 |
if (ep_op_has_event(op)) |
58e41a44c
|
2040 |
ep_take_care_of_epollwakeup(epds); |
4d7e30d98
|
2041 |
|
7699acd13
|
2042 2043 2044 2045 2046 2047 |
/* * We have to check that the file structure underneath the file descriptor * the user passed to us _is_ an eventpoll file. And also we do not permit * adding an epoll file descriptor inside itself. */ error = -EINVAL; |
7e3fb5842
|
2048 |
if (f.file == tf.file || !is_file_epoll(f.file)) |
7699acd13
|
2049 2050 2051 |
goto error_tgt_fput; /* |
df0108c5d
|
2052 2053 2054 2055 |
* epoll adds to the wakeup queue at EPOLL_CTL_ADD time only, * so EPOLLEXCLUSIVE is not allowed for a EPOLL_CTL_MOD operation. * Also, we do not currently supported nested exclusive wakeups. */ |
58e41a44c
|
2056 |
if (ep_op_has_event(op) && (epds->events & EPOLLEXCLUSIVE)) { |
b6a515c8a
|
2057 2058 2059 |
if (op == EPOLL_CTL_MOD) goto error_tgt_fput; if (op == EPOLL_CTL_ADD && (is_file_epoll(tf.file) || |
58e41a44c
|
2060 |
(epds->events & ~EPOLLEXCLUSIVE_OK_BITS))) |
b6a515c8a
|
2061 2062 |
goto error_tgt_fput; } |
df0108c5d
|
2063 2064 |
/* |
7699acd13
|
2065 2066 2067 |
* At this point it is safe to assume that the "private_data" contains * our own data structure. */ |
7e3fb5842
|
2068 |
ep = f.file->private_data; |
7699acd13
|
2069 |
|
22bacca48
|
2070 2071 2072 |
/* * When we insert an epoll file descriptor, inside another epoll file * descriptor, there is the change of creating closed loops, which are |
28d82dc1c
|
2073 2074 2075 2076 |
* better be handled here, than in more critical paths. While we are * checking for loops we also determine the list of files reachable * and hang them on the tfile_check_list, so we can check that we * haven't created too many possible wakeup paths. |
22bacca48
|
2077 |
* |
67347fe4e
|
2078 2079 2080 2081 2082 2083 |
* We do not need to take the global 'epumutex' on EPOLL_CTL_ADD when * the epoll file descriptor is attaching directly to a wakeup source, * unless the epoll file descriptor is nested. The purpose of taking the * 'epmutex' on add is to prevent complex toplogies such as loops and * deep wakeup paths from forming in parallel through multiple * EPOLL_CTL_ADD operations. |
22bacca48
|
2084 |
*/ |
39220e8d4
|
2085 2086 2087 |
error = epoll_mutex_lock(&ep->mtx, 0, nonblock); if (error) goto error_tgt_fput; |
28d82dc1c
|
2088 |
if (op == EPOLL_CTL_ADD) { |
67347fe4e
|
2089 |
if (!list_empty(&f.file->f_ep_links) || |
fe0a916c1
|
2090 |
ep->gen == loop_check_gen || |
67347fe4e
|
2091 |
is_file_epoll(tf.file)) { |
67347fe4e
|
2092 |
mutex_unlock(&ep->mtx); |
39220e8d4
|
2093 2094 2095 |
error = epoll_mutex_lock(&epmutex, 0, nonblock); if (error) goto error_tgt_fput; |
18306c404
|
2096 |
loop_check_gen++; |
39220e8d4
|
2097 |
full_check = 1; |
67347fe4e
|
2098 2099 |
if (is_file_epoll(tf.file)) { error = -ELOOP; |
52c479697
|
2100 |
if (ep_loop_check(ep, tf.file) != 0) |
67347fe4e
|
2101 |
goto error_tgt_fput; |
a9ed4a656
|
2102 2103 |
} else { get_file(tf.file); |
67347fe4e
|
2104 2105 |
list_add(&tf.file->f_tfile_llink, &tfile_check_list); |
a9ed4a656
|
2106 |
} |
39220e8d4
|
2107 |
error = epoll_mutex_lock(&ep->mtx, 0, nonblock); |
52c479697
|
2108 |
if (error) |
39220e8d4
|
2109 |
goto error_tgt_fput; |
67347fe4e
|
2110 2111 |
if (is_file_epoll(tf.file)) { tep = tf.file->private_data; |
39220e8d4
|
2112 2113 2114 |
error = epoll_mutex_lock(&tep->mtx, 1, nonblock); if (error) { mutex_unlock(&ep->mtx); |
52c479697
|
2115 |
goto error_tgt_fput; |
39220e8d4
|
2116 |
} |
13d518074
|
2117 |
} |
67347fe4e
|
2118 2119 |
} } |
7699acd13
|
2120 |
|
67647d0fb
|
2121 2122 2123 2124 2125 |
/* * Try to lookup the file inside our RB tree, Since we grabbed "mtx" * above, we can be sure to be able to use the item looked up by * ep_find() till we release the mutex. */ |
7e3fb5842
|
2126 |
epi = ep_find(ep, tf.file, fd); |
7699acd13
|
2127 2128 2129 2130 2131 |
error = -EINVAL; switch (op) { case EPOLL_CTL_ADD: if (!epi) { |
58e41a44c
|
2132 2133 |
epds->events |= EPOLLERR | EPOLLHUP; error = ep_insert(ep, epds, tf.file, fd, full_check); |
7699acd13
|
2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 |
} else error = -EEXIST; break; case EPOLL_CTL_DEL: if (epi) error = ep_remove(ep, epi); else error = -ENOENT; break; case EPOLL_CTL_MOD: if (epi) { |
b6a515c8a
|
2145 |
if (!(epi->event.events & EPOLLEXCLUSIVE)) { |
58e41a44c
|
2146 2147 |
epds->events |= EPOLLERR | EPOLLHUP; error = ep_modify(ep, epi, epds); |
b6a515c8a
|
2148 |
} |
7699acd13
|
2149 2150 2151 2152 |
} else error = -ENOENT; break; } |
67347fe4e
|
2153 2154 |
if (tep != NULL) mutex_unlock(&tep->mtx); |
d47de16c7
|
2155 |
mutex_unlock(&ep->mtx); |
7699acd13
|
2156 2157 |
error_tgt_fput: |
52c479697
|
2158 2159 |
if (full_check) { clear_tfile_check_list(); |
18306c404
|
2160 |
loop_check_gen++; |
22bacca48
|
2161 |
mutex_unlock(&epmutex); |
52c479697
|
2162 |
} |
22bacca48
|
2163 |
|
7e3fb5842
|
2164 |
fdput(tf); |
7699acd13
|
2165 |
error_fput: |
7e3fb5842
|
2166 |
fdput(f); |
7699acd13
|
2167 |
error_return: |
7699acd13
|
2168 2169 2170 2171 2172 |
return error; } /* |
58e41a44c
|
2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 |
* The following function implements the controller interface for * the eventpoll file that enables the insertion/removal/change of * file descriptors inside the interest set. */ SYSCALL_DEFINE4(epoll_ctl, int, epfd, int, op, int, fd, struct epoll_event __user *, event) { struct epoll_event epds; if (ep_op_has_event(op) && copy_from_user(&epds, event, sizeof(struct epoll_event))) return -EFAULT; |
39220e8d4
|
2185 |
return do_epoll_ctl(epfd, op, fd, &epds, false); |
58e41a44c
|
2186 2187 2188 |
} /* |
7699acd13
|
2189 2190 2191 |
* Implement the event wait interface for the eventpoll file. It is the kernel * part of the user space epoll_wait(2). */ |
791eb22ee
|
2192 2193 |
static int do_epoll_wait(int epfd, struct epoll_event __user *events, int maxevents, int timeout) |
7699acd13
|
2194 |
{ |
2903ff019
|
2195 2196 |
int error; struct fd f; |
7699acd13
|
2197 |
struct eventpoll *ep; |
7699acd13
|
2198 2199 2200 2201 2202 |
/* The maximum number of event must be greater than zero */ if (maxevents <= 0 || maxevents > EP_MAX_EVENTS) return -EINVAL; /* Verify that the area passed by the user is writeable */ |
96d4f267e
|
2203 |
if (!access_ok(events, maxevents * sizeof(struct epoll_event))) |
2903ff019
|
2204 |
return -EFAULT; |
7699acd13
|
2205 2206 |
/* Get the "struct file *" for the eventpoll file */ |
2903ff019
|
2207 2208 2209 |
f = fdget(epfd); if (!f.file) return -EBADF; |
7699acd13
|
2210 2211 2212 2213 2214 2215 |
/* * We have to check that the file structure underneath the fd * the user passed to us _is_ an eventpoll file. */ error = -EINVAL; |
2903ff019
|
2216 |
if (!is_file_epoll(f.file)) |
7699acd13
|
2217 2218 2219 2220 2221 2222 |
goto error_fput; /* * At this point it is safe to assume that the "private_data" contains * our own data structure. */ |
2903ff019
|
2223 |
ep = f.file->private_data; |
7699acd13
|
2224 2225 2226 2227 2228 |
/* Time to fish for events ... */ error = ep_poll(ep, events, maxevents, timeout); error_fput: |
2903ff019
|
2229 |
fdput(f); |
7699acd13
|
2230 2231 |
return error; } |
791eb22ee
|
2232 2233 2234 2235 2236 |
SYSCALL_DEFINE4(epoll_wait, int, epfd, struct epoll_event __user *, events, int, maxevents, int, timeout) { return do_epoll_wait(epfd, events, maxevents, timeout); } |
7699acd13
|
2237 2238 2239 2240 |
/* * Implement the event wait interface for the eventpoll file. It is the kernel * part of the user space epoll_pwait(2). */ |
5a8a82b1d
|
2241 2242 2243 |
SYSCALL_DEFINE6(epoll_pwait, int, epfd, struct epoll_event __user *, events, int, maxevents, int, timeout, const sigset_t __user *, sigmask, size_t, sigsetsize) |
7699acd13
|
2244 2245 |
{ int error; |
7699acd13
|
2246 2247 2248 2249 2250 |
/* * If the caller wants a certain signal mask to be set during the wait, * we apply it here. */ |
b772434be
|
2251 |
error = set_user_sigmask(sigmask, sigsetsize); |
ded653ccb
|
2252 2253 |
if (error) return error; |
7699acd13
|
2254 |
|
791eb22ee
|
2255 |
error = do_epoll_wait(epfd, events, maxevents, timeout); |
b772434be
|
2256 |
restore_saved_sigmask_unless(error == -EINTR); |
7699acd13
|
2257 2258 2259 |
return error; } |
35280bd4a
|
2260 2261 2262 2263 2264 2265 2266 2267 |
#ifdef CONFIG_COMPAT COMPAT_SYSCALL_DEFINE6(epoll_pwait, int, epfd, struct epoll_event __user *, events, int, maxevents, int, timeout, const compat_sigset_t __user *, sigmask, compat_size_t, sigsetsize) { long err; |
35280bd4a
|
2268 2269 2270 2271 2272 |
/* * If the caller wants a certain signal mask to be set during the wait, * we apply it here. */ |
b772434be
|
2273 |
err = set_compat_user_sigmask(sigmask, sigsetsize); |
ded653ccb
|
2274 2275 |
if (err) return err; |
35280bd4a
|
2276 |
|
791eb22ee
|
2277 |
err = do_epoll_wait(epfd, events, maxevents, timeout); |
b772434be
|
2278 |
restore_saved_sigmask_unless(err == -EINTR); |
35280bd4a
|
2279 2280 2281 2282 |
return err; } #endif |
1da177e4c
|
2283 2284 |
static int __init eventpoll_init(void) { |
7ef9964e6
|
2285 2286 2287 |
struct sysinfo si; si_meminfo(&si); |
9df04e1f2
|
2288 2289 2290 2291 |
/* * Allows top 4% of lomem to be allocated for epoll watches (per user). */ max_user_watches = (((si.totalram - si.totalhigh) / 25) << PAGE_SHIFT) / |
7ef9964e6
|
2292 |
EP_ITEM_COST; |
52bd19f76
|
2293 |
BUG_ON(max_user_watches < 0); |
1da177e4c
|
2294 |
|
22bacca48
|
2295 2296 2297 2298 2299 |
/* * Initialize the structure used to perform epoll file descriptor * inclusion loops checks. */ ep_nested_calls_init(&poll_loop_ncalls); |
39732ca5a
|
2300 2301 2302 2303 2304 |
/* * We can have many thousands of epitems, so prevent this from * using an extra cache line on 64-bit (and smaller) CPUs */ BUILD_BUG_ON(sizeof(void *) <= 8 && sizeof(struct epitem) > 128); |
1da177e4c
|
2305 2306 |
/* Allocates slab cache used to allocate "struct epitem" items */ epi_cache = kmem_cache_create("eventpoll_epi", sizeof(struct epitem), |
2ae928a94
|
2307 |
0, SLAB_HWCACHE_ALIGN|SLAB_PANIC|SLAB_ACCOUNT, NULL); |
1da177e4c
|
2308 2309 2310 |
/* Allocates slab cache used to allocate "struct eppoll_entry" */ pwq_cache = kmem_cache_create("eventpoll_pwq", |
2ae928a94
|
2311 |
sizeof(struct eppoll_entry), 0, SLAB_PANIC|SLAB_ACCOUNT, NULL); |
1da177e4c
|
2312 |
|
1da177e4c
|
2313 |
return 0; |
1da177e4c
|
2314 |
} |
cea692418
|
2315 |
fs_initcall(eventpoll_init); |