Blame view
fs/eventpoll.c
64.3 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
|
204 205 206 207 208 209 |
struct file *file; /* used to optimize loop detection check */ int visited; struct list_head visited_list_link; |
bf3b9f637
|
210 211 212 213 214 |
#ifdef CONFIG_NET_RX_BUSY_POLL /* used to track busy poll napi_id */ unsigned int napi_id; #endif |
1da177e4c
|
215 216 217 218 219 220 221 222 |
}; /* 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
|
223 |
struct epitem *base; |
1da177e4c
|
224 225 226 227 228 |
/* * Wait queue item that will be linked to the target file wait * queue head. */ |
ac6424b98
|
229 |
wait_queue_entry_t wait; |
1da177e4c
|
230 231 232 233 |
/* The wait queue head that linked the "wait" wait queue item */ wait_queue_head_t *whead; }; |
1da177e4c
|
234 235 236 237 238 |
/* Wrapper struct used by poll queueing */ struct ep_pqueue { poll_table pt; struct epitem *epi; }; |
5071f97ec
|
239 240 241 242 |
/* Used by the ep_send_events() function as callback private data */ struct ep_send_events_data { int maxevents; struct epoll_event __user *events; |
d7ebbe46f
|
243 |
int res; |
5071f97ec
|
244 |
}; |
1da177e4c
|
245 |
/* |
7ef9964e6
|
246 247 |
* Configuration options available inside /proc/sys/fs/epoll/ */ |
7ef9964e6
|
248 |
/* Maximum number of epoll watched descriptors, per user */ |
52bd19f76
|
249 |
static long max_user_watches __read_mostly; |
7ef9964e6
|
250 251 |
/* |
d47de16c7
|
252 |
* This mutex is used to serialize ep_free() and eventpoll_release_file(). |
1da177e4c
|
253 |
*/ |
7ef9964e6
|
254 |
static DEFINE_MUTEX(epmutex); |
1da177e4c
|
255 |
|
22bacca48
|
256 257 |
/* Used to check for epoll file descriptor inclusion loops */ static struct nested_calls poll_loop_ncalls; |
1da177e4c
|
258 |
/* Slab cache used to allocate "struct epitem" */ |
e18b890bb
|
259 |
static struct kmem_cache *epi_cache __read_mostly; |
1da177e4c
|
260 261 |
/* Slab cache used to allocate "struct eppoll_entry" */ |
e18b890bb
|
262 |
static struct kmem_cache *pwq_cache __read_mostly; |
1da177e4c
|
263 |
|
28d82dc1c
|
264 265 266 267 268 269 270 271 |
/* Visited nodes during ep_loop_check(), so we can unset them when we finish */ static LIST_HEAD(visited_list); /* * 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
|
272 273 274 |
#ifdef CONFIG_SYSCTL #include <linux/sysctl.h> |
eec4844fa
|
275 |
static long long_zero; |
52bd19f76
|
276 |
static long long_max = LONG_MAX; |
7ef9964e6
|
277 |
|
1f7e0616c
|
278 |
struct ctl_table epoll_table[] = { |
7ef9964e6
|
279 |
{ |
7ef9964e6
|
280 281 |
.procname = "max_user_watches", .data = &max_user_watches, |
52bd19f76
|
282 |
.maxlen = sizeof(max_user_watches), |
7ef9964e6
|
283 |
.mode = 0644, |
52bd19f76
|
284 |
.proc_handler = proc_doulongvec_minmax, |
eec4844fa
|
285 |
.extra1 = &long_zero, |
52bd19f76
|
286 |
.extra2 = &long_max, |
7ef9964e6
|
287 |
}, |
ab09203e3
|
288 |
{ } |
7ef9964e6
|
289 290 |
}; #endif /* CONFIG_SYSCTL */ |
28d82dc1c
|
291 292 293 294 295 296 |
static const struct file_operations eventpoll_fops; static inline int is_file_epoll(struct file *f) { return f->f_op == &eventpoll_fops; } |
b030a4dd6
|
297 |
|
67647d0fb
|
298 |
/* Setup the structure that is used as key for the RB tree */ |
b030a4dd6
|
299 300 301 302 303 304 |
static inline void ep_set_ffd(struct epoll_filefd *ffd, struct file *file, int fd) { ffd->file = file; ffd->fd = fd; } |
67647d0fb
|
305 |
/* Compare RB tree keys */ |
b030a4dd6
|
306 307 308 309 310 311 |
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
|
312 |
/* Tells us if the item is currently linked */ |
992991c03
|
313 |
static inline int ep_is_linked(struct epitem *epi) |
b030a4dd6
|
314 |
{ |
992991c03
|
315 |
return !list_empty(&epi->rdllink); |
b030a4dd6
|
316 |
} |
ac6424b98
|
317 |
static inline struct eppoll_entry *ep_pwq_from_wait(wait_queue_entry_t *p) |
971316f05
|
318 319 320 |
{ return container_of(p, struct eppoll_entry, wait); } |
b030a4dd6
|
321 |
/* Get the "struct epitem" from a wait queue pointer */ |
ac6424b98
|
322 |
static inline struct epitem *ep_item_from_wait(wait_queue_entry_t *p) |
b030a4dd6
|
323 324 325 326 327 |
{ return container_of(p, struct eppoll_entry, wait)->base; } /* Get the "struct epitem" from an epoll queue wrapper */ |
cdac75e6f
|
328 |
static inline struct epitem *ep_item_from_epqueue(poll_table *p) |
b030a4dd6
|
329 330 331 |
{ return container_of(p, struct ep_pqueue, pt)->epi; } |
1da177e4c
|
332 |
/* Initialize the poll safe wake up structure */ |
5071f97ec
|
333 |
static void ep_nested_calls_init(struct nested_calls *ncalls) |
1da177e4c
|
334 |
{ |
5071f97ec
|
335 336 |
INIT_LIST_HEAD(&ncalls->tasks_call_list); spin_lock_init(&ncalls->lock); |
1da177e4c
|
337 |
} |
5071f97ec
|
338 |
/** |
3fb0e584a
|
339 340 341 342 343 344 345 346 347 |
* 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
|
348 349 |
return !list_empty_careful(&ep->rdllist) || READ_ONCE(ep->ovflist) != EP_UNACTIVE_PTR; |
3fb0e584a
|
350 |
} |
bf3b9f637
|
351 352 353 354 355 356 357 |
#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
|
358 359 360 361 362 363 364 365 366 |
/* * 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
|
367 368 369 370 |
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
|
371 372 373 374 |
} static inline void ep_reset_busy_poll_napi_id(struct eventpoll *ep) { |
bf3b9f637
|
375 376 |
if (ep->napi_id) ep->napi_id = 0; |
bf3b9f637
|
377 378 379 380 381 382 383 |
} /* * Set epoll busy poll NAPI ID from sk. */ static inline void ep_set_busy_poll_napi_id(struct epitem *epi) { |
bf3b9f637
|
384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 |
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
|
413 |
} |
514056d50
|
414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 |
#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
|
429 |
/** |
5071f97ec
|
430 431 432 433 434 435 |
* 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
|
436 437 438 |
* @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
|
439 |
* @ctx: This instance context. |
5071f97ec
|
440 441 442 |
* * Returns: Returns the code returned by the @nproc callback, or -1 if * the maximum recursion limit has been exceeded. |
1da177e4c
|
443 |
*/ |
74bdc1298
|
444 |
static int ep_call_nested(struct nested_calls *ncalls, |
5071f97ec
|
445 |
int (*nproc)(void *, void *, int), void *priv, |
3fe4a975d
|
446 |
void *cookie, void *ctx) |
1da177e4c
|
447 |
{ |
5071f97ec
|
448 |
int error, call_nests = 0; |
1da177e4c
|
449 |
unsigned long flags; |
5071f97ec
|
450 451 452 |
struct list_head *lsthead = &ncalls->tasks_call_list; struct nested_call_node *tncur; struct nested_call_node tnode; |
1da177e4c
|
453 |
|
5071f97ec
|
454 |
spin_lock_irqsave(&ncalls->lock, flags); |
1da177e4c
|
455 |
|
5071f97ec
|
456 457 458 459 460 |
/* * 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
|
461 |
list_for_each_entry(tncur, lsthead, llink) { |
3fe4a975d
|
462 |
if (tncur->ctx == ctx && |
74bdc1298
|
463 |
(tncur->cookie == cookie || ++call_nests > EP_MAX_NESTS)) { |
1da177e4c
|
464 465 466 467 |
/* * Ops ... loop detected or maximum nest level reached. * We abort this wake by breaking the cycle itself. */ |
abff55cee
|
468 469 |
error = -1; goto out_unlock; |
1da177e4c
|
470 471 |
} } |
5071f97ec
|
472 |
/* Add the current task and cookie to the list */ |
3fe4a975d
|
473 |
tnode.ctx = ctx; |
5071f97ec
|
474 |
tnode.cookie = cookie; |
1da177e4c
|
475 |
list_add(&tnode.llink, lsthead); |
5071f97ec
|
476 |
spin_unlock_irqrestore(&ncalls->lock, flags); |
1da177e4c
|
477 |
|
5071f97ec
|
478 479 |
/* Call the nested function */ error = (*nproc)(priv, cookie, call_nests); |
1da177e4c
|
480 481 |
/* Remove the current task from the list */ |
5071f97ec
|
482 |
spin_lock_irqsave(&ncalls->lock, flags); |
1da177e4c
|
483 |
list_del(&tnode.llink); |
3fe4a975d
|
484 |
out_unlock: |
5071f97ec
|
485 486 487 488 |
spin_unlock_irqrestore(&ncalls->lock, flags); return error; } |
02edc6fc4
|
489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 |
/* * 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
|
514 |
#ifdef CONFIG_DEBUG_LOCK_ALLOC |
57a173bdf
|
515 |
|
f6520c520
|
516 |
static DEFINE_PER_CPU(int, wakeup_nest); |
5071f97ec
|
517 |
|
5071f97ec
|
518 519 |
static void ep_poll_safewake(wait_queue_head_t *wq) { |
f6520c520
|
520 521 |
unsigned long flags; int subclass; |
3fe4a975d
|
522 |
|
f6520c520
|
523 524 525 526 527 528 529 530 531 532 |
local_irq_save(flags); preempt_disable(); subclass = __this_cpu_read(wakeup_nest); spin_lock_nested(&wq->lock, subclass + 1); __this_cpu_inc(wakeup_nest); wake_up_locked_poll(wq, POLLIN); __this_cpu_dec(wakeup_nest); spin_unlock(&wq->lock); local_irq_restore(flags); preempt_enable(); |
1da177e4c
|
533 |
} |
57a173bdf
|
534 535 536 537 |
#else static void ep_poll_safewake(wait_queue_head_t *wq) { |
a9a08845e
|
538 |
wake_up_poll(wq, EPOLLIN); |
57a173bdf
|
539 540 541 |
} #endif |
971316f05
|
542 543 544 545 546 |
static void ep_remove_wait_queue(struct eppoll_entry *pwq) { wait_queue_head_t *whead; rcu_read_lock(); |
138e4ad67
|
547 548 549 550 551 552 553 |
/* * 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
|
554 555 556 557 |
if (whead) remove_wait_queue(whead, &pwq->wait); rcu_read_unlock(); } |
1da177e4c
|
558 |
/* |
d1bc90dd5
|
559 560 561 |
* This function unregisters poll callbacks from the associated file * descriptor. Must be called with "mtx" held (or "epmutex" if called from * ep_free). |
1da177e4c
|
562 |
*/ |
7699acd13
|
563 |
static void ep_unregister_pollwait(struct eventpoll *ep, struct epitem *epi) |
1da177e4c
|
564 |
{ |
7699acd13
|
565 566 |
struct list_head *lsthead = &epi->pwqlist; struct eppoll_entry *pwq; |
1da177e4c
|
567 |
|
d1bc90dd5
|
568 569 |
while (!list_empty(lsthead)) { pwq = list_first_entry(lsthead, struct eppoll_entry, llink); |
1da177e4c
|
570 |
|
d1bc90dd5
|
571 |
list_del(&pwq->llink); |
971316f05
|
572 |
ep_remove_wait_queue(pwq); |
d1bc90dd5
|
573 |
kmem_cache_free(pwq_cache, pwq); |
1da177e4c
|
574 |
} |
1da177e4c
|
575 |
} |
eea1d5859
|
576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 |
/* 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
|
607 608 609 610 611 612 613 614 |
/** * 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
|
615 |
* @depth: The current depth of recursive f_op->poll calls. |
67347fe4e
|
616 |
* @ep_locked: caller already holds ep->mtx |
5071f97ec
|
617 618 619 |
* * Returns: The same integer error code returned by the @sproc callback. */ |
d85e2aa2e
|
620 621 |
static __poll_t ep_scan_ready_list(struct eventpoll *ep, __poll_t (*sproc)(struct eventpoll *, |
5071f97ec
|
622 |
struct list_head *, void *), |
67347fe4e
|
623 |
void *priv, int depth, bool ep_locked) |
5071f97ec
|
624 |
{ |
d85e2aa2e
|
625 |
__poll_t res; |
5071f97ec
|
626 |
struct epitem *epi, *nepi; |
296e236e9
|
627 |
LIST_HEAD(txlist); |
5071f97ec
|
628 |
|
92e641784
|
629 |
lockdep_assert_irqs_enabled(); |
5071f97ec
|
630 631 |
/* * We need to lock this because we could be hit by |
e057e15ff
|
632 |
* eventpoll_release_file() and epoll_ctl(). |
5071f97ec
|
633 |
*/ |
67347fe4e
|
634 635 636 |
if (!ep_locked) mutex_lock_nested(&ep->mtx, depth); |
5071f97ec
|
637 638 639 640 641 642 643 644 645 |
/* * 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
|
646 |
write_lock_irq(&ep->lock); |
296e236e9
|
647 |
list_splice_init(&ep->rdllist, &txlist); |
c5a282e96
|
648 |
WRITE_ONCE(ep->ovflist, NULL); |
a218cc491
|
649 |
write_unlock_irq(&ep->lock); |
5071f97ec
|
650 651 652 653 |
/* * Now call the callback function. */ |
d85e2aa2e
|
654 |
res = (*sproc)(ep, &txlist, priv); |
5071f97ec
|
655 |
|
a218cc491
|
656 |
write_lock_irq(&ep->lock); |
5071f97ec
|
657 658 659 660 661 |
/* * 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
|
662 |
for (nepi = READ_ONCE(ep->ovflist); (epi = nepi) != NULL; |
5071f97ec
|
663 664 665 666 667 668 669 |
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
|
670 |
if (!ep_is_linked(epi)) { |
c141175d0
|
671 672 673 674 675 |
/* * ->ovflist is LIFO, so we have to reverse it in order * to keep in FIFO. */ list_add(&epi->rdllink, &ep->rdllist); |
eea1d5859
|
676 |
ep_pm_stay_awake(epi); |
4d7e30d98
|
677 |
} |
5071f97ec
|
678 679 680 681 682 683 |
} /* * 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
|
684 |
WRITE_ONCE(ep->ovflist, EP_UNACTIVE_PTR); |
5071f97ec
|
685 686 687 688 689 |
/* * Quickly re-inject items left on "txlist". */ list_splice(&txlist, &ep->rdllist); |
4d7e30d98
|
690 |
__pm_relax(ep->ws); |
a218cc491
|
691 |
write_unlock_irq(&ep->lock); |
5071f97ec
|
692 |
|
67347fe4e
|
693 694 |
if (!ep_locked) mutex_unlock(&ep->mtx); |
5071f97ec
|
695 |
|
d85e2aa2e
|
696 |
return res; |
5071f97ec
|
697 |
} |
ae10b2b4e
|
698 699 700 701 702 |
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
|
703 |
/* |
7699acd13
|
704 |
* Removes a "struct epitem" from the eventpoll RB tree and deallocates |
c7ea76302
|
705 |
* all the associated resources. Must be called with "mtx" held. |
7699acd13
|
706 707 708 |
*/ static int ep_remove(struct eventpoll *ep, struct epitem *epi) { |
7699acd13
|
709 |
struct file *file = epi->ffd.file; |
1da177e4c
|
710 |
|
92e641784
|
711 |
lockdep_assert_irqs_enabled(); |
1da177e4c
|
712 |
/* |
ee8ef0a4b
|
713 |
* Removes poll wait queue hooks. |
1da177e4c
|
714 |
*/ |
7699acd13
|
715 |
ep_unregister_pollwait(ep, epi); |
1da177e4c
|
716 |
|
7699acd13
|
717 |
/* Remove the current item from the list of epoll hooks */ |
684999149
|
718 |
spin_lock(&file->f_lock); |
ae10b2b4e
|
719 |
list_del_rcu(&epi->fllink); |
684999149
|
720 |
spin_unlock(&file->f_lock); |
1da177e4c
|
721 |
|
b2ac2ea62
|
722 |
rb_erase_cached(&epi->rbn, &ep->rbr); |
1da177e4c
|
723 |
|
a218cc491
|
724 |
write_lock_irq(&ep->lock); |
992991c03
|
725 |
if (ep_is_linked(epi)) |
c7ea76302
|
726 |
list_del_init(&epi->rdllink); |
a218cc491
|
727 |
write_unlock_irq(&ep->lock); |
1da177e4c
|
728 |
|
eea1d5859
|
729 |
wakeup_source_unregister(ep_wakeup_source(epi)); |
ae10b2b4e
|
730 731 732 733 734 735 736 737 |
/* * 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
|
738 |
|
52bd19f76
|
739 |
atomic_long_dec(&ep->user->epoll_watches); |
7ef9964e6
|
740 |
|
c7ea76302
|
741 |
return 0; |
1da177e4c
|
742 |
} |
7699acd13
|
743 |
static void ep_free(struct eventpoll *ep) |
1da177e4c
|
744 |
{ |
7699acd13
|
745 746 |
struct rb_node *rbp; struct epitem *epi; |
1da177e4c
|
747 |
|
7699acd13
|
748 749 |
/* We need to release all tasks waiting for these file */ if (waitqueue_active(&ep->poll_wait)) |
5071f97ec
|
750 |
ep_poll_safewake(&ep->poll_wait); |
1da177e4c
|
751 |
|
7699acd13
|
752 753 754 |
/* * We need to lock this because we could be hit by * eventpoll_release_file() while we're freeing the "struct eventpoll". |
d47de16c7
|
755 |
* We do not need to hold "ep->mtx" here because the epoll file |
7699acd13
|
756 757 |
* 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
|
758 |
* holding "epmutex" is sufficient here. |
7699acd13
|
759 760 |
*/ mutex_lock(&epmutex); |
1da177e4c
|
761 762 |
/* |
7699acd13
|
763 |
* Walks through the whole tree by unregistering poll callbacks. |
1da177e4c
|
764 |
*/ |
b2ac2ea62
|
765 |
for (rbp = rb_first_cached(&ep->rbr); rbp; rbp = rb_next(rbp)) { |
7699acd13
|
766 767 768 |
epi = rb_entry(rbp, struct epitem, rbn); ep_unregister_pollwait(ep, epi); |
91cf5ab60
|
769 |
cond_resched(); |
7699acd13
|
770 |
} |
1da177e4c
|
771 772 |
/* |
7699acd13
|
773 774 |
* 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
|
775 |
* holding "epmutex" we can be sure that no file cleanup code will hit |
a218cc491
|
776 |
* us during this operation. So we can avoid the lock on "ep->lock". |
ddf676c38
|
777 778 |
* We do not need to lock ep->mtx, either, we only do it to prevent * a lockdep warning. |
1da177e4c
|
779 |
*/ |
ddf676c38
|
780 |
mutex_lock(&ep->mtx); |
b2ac2ea62
|
781 |
while ((rbp = rb_first_cached(&ep->rbr)) != NULL) { |
7699acd13
|
782 783 |
epi = rb_entry(rbp, struct epitem, rbn); ep_remove(ep, epi); |
91cf5ab60
|
784 |
cond_resched(); |
7699acd13
|
785 |
} |
ddf676c38
|
786 |
mutex_unlock(&ep->mtx); |
1da177e4c
|
787 |
|
7699acd13
|
788 |
mutex_unlock(&epmutex); |
d47de16c7
|
789 |
mutex_destroy(&ep->mtx); |
7ef9964e6
|
790 |
free_uid(ep->user); |
4d7e30d98
|
791 |
wakeup_source_unregister(ep->ws); |
f0ee9aabb
|
792 |
kfree(ep); |
7699acd13
|
793 |
} |
1da177e4c
|
794 |
|
7699acd13
|
795 796 797 |
static int ep_eventpoll_release(struct inode *inode, struct file *file) { struct eventpoll *ep = file->private_data; |
1da177e4c
|
798 |
|
f0ee9aabb
|
799 |
if (ep) |
7699acd13
|
800 |
ep_free(ep); |
7699acd13
|
801 |
|
7699acd13
|
802 |
return 0; |
1da177e4c
|
803 |
} |
d85e2aa2e
|
804 |
static __poll_t ep_read_events_proc(struct eventpoll *ep, struct list_head *head, |
37b5e5212
|
805 806 807 808 809 810 811 812 813 |
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
|
814 |
static __poll_t ep_item_poll(const struct epitem *epi, poll_table *pt, |
bec1a502d
|
815 |
int depth) |
450d89ec0
|
816 |
{ |
37b5e5212
|
817 818 |
struct eventpoll *ep; bool locked; |
450d89ec0
|
819 |
pt->_key = epi->event.events; |
37b5e5212
|
820 |
if (!is_file_epoll(epi->ffd.file)) |
9965ed174
|
821 |
return vfs_poll(epi->ffd.file, pt) & epi->event.events; |
450d89ec0
|
822 |
|
37b5e5212
|
823 824 825 |
ep = epi->ffd.file->private_data; poll_wait(epi->ffd.file, &ep->poll_wait, pt); locked = pt && (pt->_qproc == ep_ptable_queue_proc); |
450d89ec0
|
826 |
|
37b5e5212
|
827 828 829 |
return ep_scan_ready_list(epi->ffd.file->private_data, ep_read_events_proc, &depth, depth, locked) & epi->event.events; |
450d89ec0
|
830 |
} |
d85e2aa2e
|
831 |
static __poll_t ep_read_events_proc(struct eventpoll *ep, struct list_head *head, |
296e236e9
|
832 |
void *priv) |
5071f97ec
|
833 834 |
{ struct epitem *epi, *tmp; |
626cf2366
|
835 |
poll_table pt; |
37b5e5212
|
836 |
int depth = *(int *)priv; |
5071f97ec
|
837 |
|
626cf2366
|
838 |
init_poll_funcptr(&pt, NULL); |
37b5e5212
|
839 |
depth++; |
450d89ec0
|
840 |
|
5071f97ec
|
841 |
list_for_each_entry_safe(epi, tmp, head, rdllink) { |
37b5e5212
|
842 |
if (ep_item_poll(epi, &pt, depth)) { |
a9a08845e
|
843 |
return EPOLLIN | EPOLLRDNORM; |
37b5e5212
|
844 |
} else { |
5071f97ec
|
845 846 847 848 849 |
/* * 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
|
850 |
__pm_relax(ep_wakeup_source(epi)); |
5071f97ec
|
851 |
list_del_init(&epi->rdllink); |
296e236e9
|
852 |
} |
5071f97ec
|
853 854 855 856 |
} return 0; } |
a11e1d432
|
857 |
static __poll_t ep_eventpoll_poll(struct file *file, poll_table *wait) |
11c5ad0ec
|
858 859 860 |
{ struct eventpoll *ep = file->private_data; int depth = 0; |
7699acd13
|
861 |
|
a11e1d432
|
862 863 |
/* Insert inside our poll wait queue */ poll_wait(file, &ep->poll_wait, wait); |
5071f97ec
|
864 865 |
/* * Proceed to find out if wanted events are really available inside |
37b5e5212
|
866 |
* the ready list. |
5071f97ec
|
867 |
*/ |
37b5e5212
|
868 869 |
return ep_scan_ready_list(ep, ep_read_events_proc, &depth, depth, false); |
7699acd13
|
870 |
} |
138d22b58
|
871 |
#ifdef CONFIG_PROC_FS |
a3816ab0e
|
872 |
static void ep_show_fdinfo(struct seq_file *m, struct file *f) |
138d22b58
|
873 874 875 |
{ struct eventpoll *ep = f->private_data; struct rb_node *rbp; |
138d22b58
|
876 877 |
mutex_lock(&ep->mtx); |
b2ac2ea62
|
878 |
for (rbp = rb_first_cached(&ep->rbr); rbp; rbp = rb_next(rbp)) { |
138d22b58
|
879 |
struct epitem *epi = rb_entry(rbp, struct epitem, rbn); |
77493f04b
|
880 |
struct inode *inode = file_inode(epi->ffd.file); |
138d22b58
|
881 |
|
77493f04b
|
882 883 884 |
seq_printf(m, "tfd: %8d events: %8x data: %16llx " " pos:%lli ino:%lx sdev:%x ", |
a3816ab0e
|
885 |
epi->ffd.fd, epi->event.events, |
77493f04b
|
886 887 888 |
(long long)epi->event.data, (long long)epi->ffd.file->f_pos, inode->i_ino, inode->i_sb->s_dev); |
a3816ab0e
|
889 |
if (seq_has_overflowed(m)) |
138d22b58
|
890 891 892 |
break; } mutex_unlock(&ep->mtx); |
138d22b58
|
893 894 |
} #endif |
7699acd13
|
895 896 |
/* File callbacks that implement the eventpoll file behaviour */ static const struct file_operations eventpoll_fops = { |
138d22b58
|
897 898 899 |
#ifdef CONFIG_PROC_FS .show_fdinfo = ep_show_fdinfo, #endif |
7699acd13
|
900 |
.release = ep_eventpoll_release, |
a11e1d432
|
901 |
.poll = ep_eventpoll_poll, |
6038f373a
|
902 |
.llseek = noop_llseek, |
7699acd13
|
903 |
}; |
b611967de
|
904 |
/* |
7699acd13
|
905 906 907 |
* 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
|
908 |
*/ |
7699acd13
|
909 |
void eventpoll_release_file(struct file *file) |
b611967de
|
910 |
{ |
7699acd13
|
911 |
struct eventpoll *ep; |
ebe06187b
|
912 |
struct epitem *epi, *next; |
b611967de
|
913 914 |
/* |
684999149
|
915 |
* We don't want to get "file->f_lock" because it is not |
7699acd13
|
916 |
* necessary. It is not necessary because we're in the "struct file" |
25985edce
|
917 |
* cleanup path, and this means that no one is using this file anymore. |
5071f97ec
|
918 |
* So, for example, epoll_ctl() cannot hit here since if we reach this |
67647d0fb
|
919 |
* point, the file counter already went to zero and fget() would fail. |
d47de16c7
|
920 |
* The only hit might come from ep_free() but by holding the mutex |
7699acd13
|
921 |
* will correctly serialize the operation. We do need to acquire |
d47de16c7
|
922 |
* "ep->mtx" after "epmutex" because ep_remove() requires it when called |
7699acd13
|
923 |
* from anywhere but ep_free(). |
684999149
|
924 925 |
* * Besides, ep_remove() acquires the lock, so we can't hold it here. |
b611967de
|
926 |
*/ |
7699acd13
|
927 |
mutex_lock(&epmutex); |
ebe06187b
|
928 |
list_for_each_entry_safe(epi, next, &file->f_ep_links, fllink) { |
7699acd13
|
929 |
ep = epi->ep; |
d8805e633
|
930 |
mutex_lock_nested(&ep->mtx, 0); |
7699acd13
|
931 |
ep_remove(ep, epi); |
d47de16c7
|
932 |
mutex_unlock(&ep->mtx); |
b611967de
|
933 |
} |
7699acd13
|
934 |
mutex_unlock(&epmutex); |
b611967de
|
935 |
} |
53d2be79d
|
936 |
static int ep_alloc(struct eventpoll **pep) |
1da177e4c
|
937 |
{ |
7ef9964e6
|
938 939 940 |
int error; struct user_struct *user; struct eventpoll *ep; |
1da177e4c
|
941 |
|
7ef9964e6
|
942 |
user = get_current_user(); |
7ef9964e6
|
943 944 945 946 |
error = -ENOMEM; ep = kzalloc(sizeof(*ep), GFP_KERNEL); if (unlikely(!ep)) goto free_uid; |
1da177e4c
|
947 |
|
d47de16c7
|
948 |
mutex_init(&ep->mtx); |
a218cc491
|
949 |
rwlock_init(&ep->lock); |
1da177e4c
|
950 951 952 |
init_waitqueue_head(&ep->wq); init_waitqueue_head(&ep->poll_wait); INIT_LIST_HEAD(&ep->rdllist); |
b2ac2ea62
|
953 |
ep->rbr = RB_ROOT_CACHED; |
d47de16c7
|
954 |
ep->ovflist = EP_UNACTIVE_PTR; |
7ef9964e6
|
955 |
ep->user = user; |
1da177e4c
|
956 |
|
53d2be79d
|
957 |
*pep = ep; |
1da177e4c
|
958 |
|
1da177e4c
|
959 |
return 0; |
7ef9964e6
|
960 961 962 963 |
free_uid: free_uid(user); return error; |
1da177e4c
|
964 |
} |
1da177e4c
|
965 |
/* |
c7ea76302
|
966 967 968 |
* 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
|
969 970 971 972 |
*/ static struct epitem *ep_find(struct eventpoll *ep, struct file *file, int fd) { int kcmp; |
1da177e4c
|
973 974 975 |
struct rb_node *rbp; struct epitem *epi, *epir = NULL; struct epoll_filefd ffd; |
b030a4dd6
|
976 |
ep_set_ffd(&ffd, file, fd); |
b2ac2ea62
|
977 |
for (rbp = ep->rbr.rb_root.rb_node; rbp; ) { |
1da177e4c
|
978 |
epi = rb_entry(rbp, struct epitem, rbn); |
b030a4dd6
|
979 |
kcmp = ep_cmp_ffd(&ffd, &epi->ffd); |
1da177e4c
|
980 981 982 983 984 |
if (kcmp > 0) rbp = rbp->rb_right; else if (kcmp < 0) rbp = rbp->rb_left; else { |
1da177e4c
|
985 986 987 988 |
epir = epi; break; } } |
1da177e4c
|
989 |
|
1da177e4c
|
990 991 |
return epir; } |
92ef6da3d
|
992 |
#ifdef CONFIG_CHECKPOINT_RESTORE |
0791e3644
|
993 994 995 996 |
static struct epitem *ep_find_tfd(struct eventpoll *ep, int tfd, unsigned long toff) { struct rb_node *rbp; struct epitem *epi; |
b2ac2ea62
|
997 |
for (rbp = rb_first_cached(&ep->rbr); rbp; rbp = rb_next(rbp)) { |
0791e3644
|
998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 |
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
|
1033 |
#endif /* CONFIG_CHECKPOINT_RESTORE */ |
0791e3644
|
1034 |
|
a218cc491
|
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 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 |
/** * 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; /* 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
|
1107 |
/* |
7699acd13
|
1108 |
* This is the callback that is passed to the wait queue wakeup |
bf6a41db7
|
1109 |
* mechanism. It is called by the stored file descriptors when they |
7699acd13
|
1110 |
* have events to report. |
a218cc491
|
1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 |
* * 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
|
1124 |
*/ |
ac6424b98
|
1125 |
static int ep_poll_callback(wait_queue_entry_t *wait, unsigned mode, int sync, void *key) |
1da177e4c
|
1126 |
{ |
7699acd13
|
1127 |
int pwake = 0; |
7699acd13
|
1128 1129 |
struct epitem *epi = ep_item_from_wait(wait); struct eventpoll *ep = epi->ep; |
3ad6f93e9
|
1130 |
__poll_t pollflags = key_to_poll(key); |
a218cc491
|
1131 |
unsigned long flags; |
df0108c5d
|
1132 |
int ewake = 0; |
1da177e4c
|
1133 |
|
a218cc491
|
1134 |
read_lock_irqsave(&ep->lock, flags); |
1da177e4c
|
1135 |
|
bf3b9f637
|
1136 |
ep_set_busy_poll_napi_id(epi); |
7699acd13
|
1137 1138 1139 1140 1141 1142 1143 |
/* * 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
|
1144 1145 1146 |
goto out_unlock; /* |
2dfa4eeab
|
1147 1148 1149 1150 1151 |
* 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
|
1152 |
if (pollflags && !(pollflags & epi->event.events)) |
2dfa4eeab
|
1153 1154 1155 |
goto out_unlock; /* |
bf6a41db7
|
1156 |
* If we are transferring events to userspace, we can hold no locks |
d47de16c7
|
1157 |
* (because we're accessing user memory, and because of linux f_op->poll() |
bf6a41db7
|
1158 |
* semantics). All the events that happen during that period of time are |
d47de16c7
|
1159 1160 |
* chained in ep->ovflist and requeued later on. */ |
c5a282e96
|
1161 |
if (READ_ONCE(ep->ovflist) != EP_UNACTIVE_PTR) { |
a218cc491
|
1162 1163 |
if (epi->next == EP_UNACTIVE_PTR && chain_epi_lockless(epi)) |
c3e320b61
|
1164 |
ep_pm_stay_awake_rcu(epi); |
d47de16c7
|
1165 1166 |
goto out_unlock; } |
1da177e4c
|
1167 |
|
7699acd13
|
1168 |
/* If this file is already in the ready list we exit soon */ |
a218cc491
|
1169 1170 |
if (!ep_is_linked(epi) && list_add_tail_lockless(&epi->rdllink, &ep->rdllist)) { |
eea1d5859
|
1171 |
ep_pm_stay_awake_rcu(epi); |
4d7e30d98
|
1172 |
} |
7699acd13
|
1173 |
|
7699acd13
|
1174 1175 1176 1177 |
/* * Wake up ( if active ) both the eventpoll wait list and the ->poll() * wait list. */ |
df0108c5d
|
1178 |
if (waitqueue_active(&ep->wq)) { |
b6a515c8a
|
1179 |
if ((epi->event.events & EPOLLEXCLUSIVE) && |
3ad6f93e9
|
1180 1181 |
!(pollflags & POLLFREE)) { switch (pollflags & EPOLLINOUT_BITS) { |
a9a08845e
|
1182 1183 |
case EPOLLIN: if (epi->event.events & EPOLLIN) |
b6a515c8a
|
1184 1185 |
ewake = 1; break; |
a9a08845e
|
1186 1187 |
case EPOLLOUT: if (epi->event.events & EPOLLOUT) |
b6a515c8a
|
1188 1189 1190 1191 1192 1193 1194 |
ewake = 1; break; case 0: ewake = 1; break; } } |
a218cc491
|
1195 |
wake_up(&ep->wq); |
df0108c5d
|
1196 |
} |
7699acd13
|
1197 1198 |
if (waitqueue_active(&ep->poll_wait)) pwake++; |
d47de16c7
|
1199 |
out_unlock: |
a218cc491
|
1200 |
read_unlock_irqrestore(&ep->lock, flags); |
1da177e4c
|
1201 |
|
7699acd13
|
1202 1203 |
/* We have to call this outside the lock */ if (pwake) |
5071f97ec
|
1204 |
ep_poll_safewake(&ep->poll_wait); |
7699acd13
|
1205 |
|
138e4ad67
|
1206 1207 |
if (!(epi->event.events & EPOLLEXCLUSIVE)) ewake = 1; |
3ad6f93e9
|
1208 |
if (pollflags & POLLFREE) { |
138e4ad67
|
1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 |
/* * 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
|
1223 |
|
138e4ad67
|
1224 |
return ewake; |
7699acd13
|
1225 |
} |
1da177e4c
|
1226 1227 1228 1229 1230 1231 1232 1233 |
/* * 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
|
1234 |
struct epitem *epi = ep_item_from_epqueue(pt); |
1da177e4c
|
1235 |
struct eppoll_entry *pwq; |
e94b17660
|
1236 |
if (epi->nwait >= 0 && (pwq = kmem_cache_alloc(pwq_cache, GFP_KERNEL))) { |
1da177e4c
|
1237 1238 1239 |
init_waitqueue_func_entry(&pwq->wait, ep_poll_callback); pwq->whead = whead; pwq->base = epi; |
df0108c5d
|
1240 1241 1242 1243 |
if (epi->event.events & EPOLLEXCLUSIVE) add_wait_queue_exclusive(whead, &pwq->wait); else add_wait_queue(whead, &pwq->wait); |
1da177e4c
|
1244 1245 |
list_add_tail(&pwq->llink, &epi->pwqlist); epi->nwait++; |
296e236e9
|
1246 |
} else { |
1da177e4c
|
1247 1248 |
/* We have to signal that an error occurred */ epi->nwait = -1; |
296e236e9
|
1249 |
} |
1da177e4c
|
1250 |
} |
1da177e4c
|
1251 1252 1253 |
static void ep_rbtree_insert(struct eventpoll *ep, struct epitem *epi) { int kcmp; |
b2ac2ea62
|
1254 |
struct rb_node **p = &ep->rbr.rb_root.rb_node, *parent = NULL; |
1da177e4c
|
1255 |
struct epitem *epic; |
b2ac2ea62
|
1256 |
bool leftmost = true; |
1da177e4c
|
1257 1258 1259 1260 |
while (*p) { parent = *p; epic = rb_entry(parent, struct epitem, rbn); |
b030a4dd6
|
1261 |
kcmp = ep_cmp_ffd(&epi->ffd, &epic->ffd); |
b2ac2ea62
|
1262 |
if (kcmp > 0) { |
1da177e4c
|
1263 |
p = &parent->rb_right; |
b2ac2ea62
|
1264 1265 |
leftmost = false; } else |
1da177e4c
|
1266 1267 1268 |
p = &parent->rb_left; } rb_link_node(&epi->rbn, parent, p); |
b2ac2ea62
|
1269 |
rb_insert_color_cached(&epi->rbn, &ep->rbr, leftmost); |
1da177e4c
|
1270 |
} |
a80a6b85b
|
1271 |
|
28d82dc1c
|
1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 |
#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
|
1289 1290 1291 |
/* Allow an arbitrary number of depth 1 paths */ if (nests == 0) return 0; |
28d82dc1c
|
1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 |
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
|
1311 1312 1313 |
/* 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
|
1314 1315 1316 1317 1318 1319 1320 1321 1322 |
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
|
1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 |
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
|
1335 |
rcu_read_unlock(); |
28d82dc1c
|
1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 |
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
|
1351 1352 1353 1354 1355 |
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
|
1356 |
path_count_init(); |
74bdc1298
|
1357 |
error = ep_call_nested(&poll_loop_ncalls, |
28d82dc1c
|
1358 1359 1360 1361 1362 1363 1364 |
reverse_path_check_proc, current_file, current_file, current); if (error) break; } return error; } |
4d7e30d98
|
1365 1366 1367 |
static int ep_create_wakeup_source(struct epitem *epi) { const char *name; |
eea1d5859
|
1368 |
struct wakeup_source *ws; |
4d7e30d98
|
1369 1370 |
if (!epi->ep->ws) { |
c8377adfa
|
1371 |
epi->ep->ws = wakeup_source_register(NULL, "eventpoll"); |
4d7e30d98
|
1372 1373 1374 1375 1376 |
if (!epi->ep->ws) return -ENOMEM; } name = epi->ffd.file->f_path.dentry->d_name.name; |
c8377adfa
|
1377 |
ws = wakeup_source_register(NULL, name); |
eea1d5859
|
1378 1379 |
if (!ws) |
4d7e30d98
|
1380 |
return -ENOMEM; |
eea1d5859
|
1381 |
rcu_assign_pointer(epi->ws, ws); |
4d7e30d98
|
1382 1383 1384 |
return 0; } |
eea1d5859
|
1385 1386 |
/* rare code path, only used when EPOLL_CTL_MOD removes a wakeup source */ static noinline void ep_destroy_wakeup_source(struct epitem *epi) |
4d7e30d98
|
1387 |
{ |
eea1d5859
|
1388 |
struct wakeup_source *ws = ep_wakeup_source(epi); |
d6d67e723
|
1389 |
RCU_INIT_POINTER(epi->ws, NULL); |
eea1d5859
|
1390 1391 1392 1393 1394 1395 1396 1397 |
/* * 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
|
1398 |
} |
c7ea76302
|
1399 1400 1401 |
/* * Must be called with "mtx" held. */ |
bec1a502d
|
1402 |
static int ep_insert(struct eventpoll *ep, const struct epoll_event *event, |
67347fe4e
|
1403 |
struct file *tfile, int fd, int full_check) |
1da177e4c
|
1404 |
{ |
d85e2aa2e
|
1405 1406 |
int error, pwake = 0; __poll_t revents; |
52bd19f76
|
1407 |
long user_watches; |
1da177e4c
|
1408 1409 |
struct epitem *epi; struct ep_pqueue epq; |
92e641784
|
1410 |
lockdep_assert_irqs_enabled(); |
52bd19f76
|
1411 1412 |
user_watches = atomic_long_read(&ep->user->epoll_watches); if (unlikely(user_watches >= max_user_watches)) |
7ef9964e6
|
1413 |
return -ENOSPC; |
e94b17660
|
1414 |
if (!(epi = kmem_cache_alloc(epi_cache, GFP_KERNEL))) |
7ef9964e6
|
1415 |
return -ENOMEM; |
1da177e4c
|
1416 1417 |
/* Item initialization follow here ... */ |
1da177e4c
|
1418 1419 |
INIT_LIST_HEAD(&epi->rdllink); INIT_LIST_HEAD(&epi->fllink); |
1da177e4c
|
1420 1421 |
INIT_LIST_HEAD(&epi->pwqlist); epi->ep = ep; |
b030a4dd6
|
1422 |
ep_set_ffd(&epi->ffd, tfile, fd); |
1da177e4c
|
1423 |
epi->event = *event; |
1da177e4c
|
1424 |
epi->nwait = 0; |
d47de16c7
|
1425 |
epi->next = EP_UNACTIVE_PTR; |
4d7e30d98
|
1426 1427 1428 1429 1430 |
if (epi->event.events & EPOLLWAKEUP) { error = ep_create_wakeup_source(epi); if (error) goto error_create_wakeup_source; } else { |
eea1d5859
|
1431 |
RCU_INIT_POINTER(epi->ws, NULL); |
4d7e30d98
|
1432 |
} |
1da177e4c
|
1433 1434 1435 1436 1437 1438 1439 1440 |
/* 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
|
1441 1442 1443 |
* been increased by the caller of this function. Note that after * this operation completes, the poll callback can start hitting * the new item. |
1da177e4c
|
1444 |
*/ |
37b5e5212
|
1445 |
revents = ep_item_poll(epi, &epq.pt, 1); |
1da177e4c
|
1446 1447 1448 1449 1450 1451 |
/* * 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
|
1452 |
error = -ENOMEM; |
1da177e4c
|
1453 |
if (epi->nwait < 0) |
7699acd13
|
1454 |
goto error_unregister; |
1da177e4c
|
1455 1456 |
/* Add the current item to the list of active epoll hook for this file */ |
684999149
|
1457 |
spin_lock(&tfile->f_lock); |
ae10b2b4e
|
1458 |
list_add_tail_rcu(&epi->fllink, &tfile->f_ep_links); |
684999149
|
1459 |
spin_unlock(&tfile->f_lock); |
1da177e4c
|
1460 |
|
c7ea76302
|
1461 1462 1463 1464 |
/* * Add the current item to the RB tree. All RB tree operations are * protected by "mtx", and ep_insert() is called with "mtx" held. */ |
1da177e4c
|
1465 |
ep_rbtree_insert(ep, epi); |
28d82dc1c
|
1466 1467 |
/* now check if we've created too many backpaths */ error = -EINVAL; |
67347fe4e
|
1468 |
if (full_check && reverse_path_check()) |
28d82dc1c
|
1469 |
goto error_remove_epi; |
c7ea76302
|
1470 |
/* We have to drop the new item inside our item list to keep track of it */ |
a218cc491
|
1471 |
write_lock_irq(&ep->lock); |
c7ea76302
|
1472 |
|
bf3b9f637
|
1473 1474 |
/* record NAPI ID of new item if present */ ep_set_busy_poll_napi_id(epi); |
1da177e4c
|
1475 |
/* If the file is already "ready" we drop it inside the ready list */ |
992991c03
|
1476 |
if (revents && !ep_is_linked(epi)) { |
1da177e4c
|
1477 |
list_add_tail(&epi->rdllink, &ep->rdllist); |
eea1d5859
|
1478 |
ep_pm_stay_awake(epi); |
1da177e4c
|
1479 1480 1481 |
/* Notify waiting tasks that events are available */ if (waitqueue_active(&ep->wq)) |
a218cc491
|
1482 |
wake_up(&ep->wq); |
1da177e4c
|
1483 1484 1485 |
if (waitqueue_active(&ep->poll_wait)) pwake++; } |
a218cc491
|
1486 |
write_unlock_irq(&ep->lock); |
1da177e4c
|
1487 |
|
52bd19f76
|
1488 |
atomic_long_inc(&ep->user->epoll_watches); |
7ef9964e6
|
1489 |
|
1da177e4c
|
1490 1491 |
/* We have to call this outside the lock */ if (pwake) |
5071f97ec
|
1492 |
ep_poll_safewake(&ep->poll_wait); |
1da177e4c
|
1493 |
|
1da177e4c
|
1494 |
return 0; |
28d82dc1c
|
1495 1496 |
error_remove_epi: spin_lock(&tfile->f_lock); |
ae10b2b4e
|
1497 |
list_del_rcu(&epi->fllink); |
28d82dc1c
|
1498 |
spin_unlock(&tfile->f_lock); |
b2ac2ea62
|
1499 |
rb_erase_cached(&epi->rbn, &ep->rbr); |
28d82dc1c
|
1500 |
|
7699acd13
|
1501 |
error_unregister: |
1da177e4c
|
1502 1503 1504 1505 |
ep_unregister_pollwait(ep, epi); /* * We need to do this because an event could have been arrived on some |
67647d0fb
|
1506 1507 1508 |
* 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
|
1509 |
*/ |
a218cc491
|
1510 |
write_lock_irq(&ep->lock); |
992991c03
|
1511 |
if (ep_is_linked(epi)) |
6192bd536
|
1512 |
list_del_init(&epi->rdllink); |
a218cc491
|
1513 |
write_unlock_irq(&ep->lock); |
1da177e4c
|
1514 |
|
eea1d5859
|
1515 |
wakeup_source_unregister(ep_wakeup_source(epi)); |
4d7e30d98
|
1516 1517 |
error_create_wakeup_source: |
b030a4dd6
|
1518 |
kmem_cache_free(epi_cache, epi); |
7ef9964e6
|
1519 |
|
1da177e4c
|
1520 1521 |
return error; } |
1da177e4c
|
1522 1523 |
/* * Modify the interest event mask by dropping an event if the new mask |
c7ea76302
|
1524 |
* has a match in the current file status. Must be called with "mtx" held. |
1da177e4c
|
1525 |
*/ |
bec1a502d
|
1526 1527 |
static int ep_modify(struct eventpoll *ep, struct epitem *epi, const struct epoll_event *event) |
1da177e4c
|
1528 1529 |
{ int pwake = 0; |
626cf2366
|
1530 |
poll_table pt; |
92e641784
|
1531 |
lockdep_assert_irqs_enabled(); |
626cf2366
|
1532 |
init_poll_funcptr(&pt, NULL); |
1da177e4c
|
1533 1534 |
/* |
e057e15ff
|
1535 1536 1537 |
* 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
|
1538 |
*/ |
128dd1759
|
1539 |
epi->event.events = event->events; /* need barrier below */ |
e057e15ff
|
1540 |
epi->event.data = event->data; /* protected by mtx */ |
4d7e30d98
|
1541 |
if (epi->event.events & EPOLLWAKEUP) { |
eea1d5859
|
1542 |
if (!ep_has_wakeup_source(epi)) |
4d7e30d98
|
1543 |
ep_create_wakeup_source(epi); |
eea1d5859
|
1544 |
} else if (ep_has_wakeup_source(epi)) { |
4d7e30d98
|
1545 1546 |
ep_destroy_wakeup_source(epi); } |
1da177e4c
|
1547 1548 |
/* |
128dd1759
|
1549 1550 1551 1552 1553 |
* 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
|
1554 |
* We need this because we did not take ep->lock while |
128dd1759
|
1555 |
* changing epi above (but ep_poll_callback does take |
a218cc491
|
1556 |
* ep->lock). |
128dd1759
|
1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 |
* * 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
|
1569 1570 |
* 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
|
1571 |
* If the item is "hot" and it is not registered inside the ready |
67647d0fb
|
1572 |
* list, push it inside. |
1da177e4c
|
1573 |
*/ |
69112736e
|
1574 |
if (ep_item_poll(epi, &pt, 1)) { |
a218cc491
|
1575 |
write_lock_irq(&ep->lock); |
992991c03
|
1576 |
if (!ep_is_linked(epi)) { |
c7ea76302
|
1577 |
list_add_tail(&epi->rdllink, &ep->rdllist); |
eea1d5859
|
1578 |
ep_pm_stay_awake(epi); |
c7ea76302
|
1579 1580 1581 |
/* Notify waiting tasks that events are available */ if (waitqueue_active(&ep->wq)) |
a218cc491
|
1582 |
wake_up(&ep->wq); |
c7ea76302
|
1583 1584 |
if (waitqueue_active(&ep->poll_wait)) pwake++; |
7699acd13
|
1585 |
} |
a218cc491
|
1586 |
write_unlock_irq(&ep->lock); |
7699acd13
|
1587 |
} |
1da177e4c
|
1588 |
|
7699acd13
|
1589 1590 |
/* We have to call this outside the lock */ if (pwake) |
5071f97ec
|
1591 |
ep_poll_safewake(&ep->poll_wait); |
1da177e4c
|
1592 |
|
7699acd13
|
1593 |
return 0; |
1da177e4c
|
1594 |
} |
d85e2aa2e
|
1595 |
static __poll_t ep_send_events_proc(struct eventpoll *ep, struct list_head *head, |
296e236e9
|
1596 |
void *priv) |
1da177e4c
|
1597 |
{ |
5071f97ec
|
1598 |
struct ep_send_events_data *esed = priv; |
d85e2aa2e
|
1599 |
__poll_t revents; |
4e0982a00
|
1600 1601 |
struct epitem *epi, *tmp; struct epoll_event __user *uevent = esed->events; |
eea1d5859
|
1602 |
struct wakeup_source *ws; |
626cf2366
|
1603 1604 1605 |
poll_table pt; init_poll_funcptr(&pt, NULL); |
4e0982a00
|
1606 |
esed->res = 0; |
1da177e4c
|
1607 |
|
296e236e9
|
1608 |
/* |
5071f97ec
|
1609 1610 1611 |
* 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
|
1612 |
*/ |
21877e1a5
|
1613 |
lockdep_assert_held(&ep->mtx); |
4e0982a00
|
1614 1615 1616 |
list_for_each_entry_safe(epi, tmp, head, rdllink) { if (esed->res >= esed->maxevents) break; |
d47de16c7
|
1617 |
|
4d7e30d98
|
1618 1619 1620 1621 1622 1623 1624 1625 1626 |
/* * 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
|
1627 1628 1629 1630 1631 1632 |
ws = ep_wakeup_source(epi); if (ws) { if (ws->active) __pm_stay_awake(ep->ws); __pm_relax(ws); } |
d47de16c7
|
1633 |
list_del_init(&epi->rdllink); |
1da177e4c
|
1634 |
|
296e236e9
|
1635 |
/* |
5071f97ec
|
1636 1637 |
* If the event mask intersect the caller-requested one, * deliver the event to userspace. Again, ep_scan_ready_list() |
4e0982a00
|
1638 |
* is holding ep->mtx, so no operations coming from userspace |
5071f97ec
|
1639 |
* can change the item. |
296e236e9
|
1640 |
*/ |
4e0982a00
|
1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 |
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
|
1671 1672 |
} } |
5071f97ec
|
1673 |
|
d7ebbe46f
|
1674 |
return 0; |
5071f97ec
|
1675 |
} |
d47de16c7
|
1676 |
|
296e236e9
|
1677 1678 |
static int ep_send_events(struct eventpoll *ep, struct epoll_event __user *events, int maxevents) |
5071f97ec
|
1679 1680 |
{ struct ep_send_events_data esed; |
1da177e4c
|
1681 |
|
5071f97ec
|
1682 1683 |
esed.maxevents = maxevents; esed.events = events; |
6192bd536
|
1684 |
|
d7ebbe46f
|
1685 1686 |
ep_scan_ready_list(ep, ep_send_events_proc, &esed, 0, false); return esed.res; |
1da177e4c
|
1687 |
} |
766b9f928
|
1688 |
static inline struct timespec64 ep_set_mstimeout(long ms) |
0781b909b
|
1689 |
{ |
766b9f928
|
1690 |
struct timespec64 now, ts = { |
0781b909b
|
1691 1692 1693 |
.tv_sec = ms / MSEC_PER_SEC, .tv_nsec = NSEC_PER_MSEC * (ms % MSEC_PER_SEC), }; |
766b9f928
|
1694 1695 |
ktime_get_ts64(&now); return timespec64_add_safe(now, ts); |
0781b909b
|
1696 |
} |
f4d93ad74
|
1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 |
/** * 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
|
1714 1715 1716 |
static int ep_poll(struct eventpoll *ep, struct epoll_event __user *events, int maxevents, long timeout) { |
f4d93ad74
|
1717 |
int res = 0, eavail, timed_out = 0; |
da8b44d5a
|
1718 |
u64 slack = 0; |
86c051793
|
1719 |
bool waiter = false; |
ac6424b98
|
1720 |
wait_queue_entry_t wait; |
95aac7b1c
|
1721 |
ktime_t expires, *to = NULL; |
679abf381
|
1722 |
lockdep_assert_irqs_enabled(); |
95aac7b1c
|
1723 |
if (timeout > 0) { |
766b9f928
|
1724 |
struct timespec64 end_time = ep_set_mstimeout(timeout); |
0781b909b
|
1725 |
|
95aac7b1c
|
1726 1727 |
slack = select_estimate_accuracy(&end_time); to = &expires; |
766b9f928
|
1728 |
*to = timespec64_to_ktime(end_time); |
95aac7b1c
|
1729 |
} else if (timeout == 0) { |
f4d93ad74
|
1730 1731 |
/* * Avoid the unnecessary trip to the wait queue loop, if the |
c5a282e96
|
1732 1733 1734 1735 |
* 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
|
1736 |
*/ |
95aac7b1c
|
1737 |
timed_out = 1; |
c5a282e96
|
1738 |
|
a218cc491
|
1739 |
write_lock_irq(&ep->lock); |
c5a282e96
|
1740 |
eavail = ep_events_available(ep); |
a218cc491
|
1741 |
write_unlock_irq(&ep->lock); |
c5a282e96
|
1742 |
|
35cff1a6e
|
1743 |
goto send_events; |
95aac7b1c
|
1744 |
} |
1da177e4c
|
1745 |
|
f4d93ad74
|
1746 |
fetch_events: |
bf3b9f637
|
1747 1748 1749 |
if (!ep_events_available(ep)) ep_busy_loop(ep, timed_out); |
c5a282e96
|
1750 1751 |
eavail = ep_events_available(ep); if (eavail) |
35cff1a6e
|
1752 |
goto send_events; |
1da177e4c
|
1753 |
|
c5a282e96
|
1754 1755 1756 1757 1758 1759 |
/* * 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
|
1760 |
|
c5a282e96
|
1761 |
/* |
86c051793
|
1762 1763 1764 |
* We don't have any available event to return to the caller. We need * to sleep here, and we will be woken by ep_poll_callback() when events * become available. |
c5a282e96
|
1765 |
*/ |
86c051793
|
1766 1767 |
if (!waiter) { waiter = true; |
1da177e4c
|
1768 |
init_waitqueue_entry(&wait, current); |
1da177e4c
|
1769 |
|
1b53734bd
|
1770 |
write_lock_irq(&ep->lock); |
86c051793
|
1771 |
__add_wait_queue_exclusive(&ep->wq, &wait); |
1b53734bd
|
1772 |
write_unlock_irq(&ep->lock); |
86c051793
|
1773 |
} |
1da177e4c
|
1774 |
|
c5a282e96
|
1775 |
for (;;) { |
1da177e4c
|
1776 |
/* |
c5a282e96
|
1777 1778 1779 |
* We don't want to sleep if the ep_poll_callback() sends us * a wakeup in between. That's why we set the task state * to TASK_INTERRUPTIBLE before doing the checks. |
bf3b9f637
|
1780 |
*/ |
c5a282e96
|
1781 |
set_current_state(TASK_INTERRUPTIBLE); |
bf3b9f637
|
1782 |
/* |
c5a282e96
|
1783 1784 1785 1786 |
* 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. |
1da177e4c
|
1787 |
*/ |
c5a282e96
|
1788 1789 1790 1791 |
if (fatal_signal_pending(current)) { res = -EINTR; break; } |
95aac7b1c
|
1792 |
|
abc610e01
|
1793 1794 |
eavail = ep_events_available(ep); if (eavail) |
c5a282e96
|
1795 1796 1797 1798 |
break; if (signal_pending(current)) { res = -EINTR; break; |
1da177e4c
|
1799 |
} |
1da177e4c
|
1800 |
|
abc610e01
|
1801 |
if (!schedule_hrtimeout_range(to, slack, HRTIMER_MODE_ABS)) { |
c5a282e96
|
1802 |
timed_out = 1; |
abc610e01
|
1803 1804 |
break; } |
1da177e4c
|
1805 |
} |
1da177e4c
|
1806 |
|
c5a282e96
|
1807 |
__set_current_state(TASK_RUNNING); |
1da177e4c
|
1808 |
|
35cff1a6e
|
1809 |
send_events: |
1da177e4c
|
1810 1811 1812 1813 1814 1815 |
/* * 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
|
1816 |
!(res = ep_send_events(ep, events, maxevents)) && !timed_out) |
f4d93ad74
|
1817 |
goto fetch_events; |
1da177e4c
|
1818 |
|
86c051793
|
1819 |
if (waiter) { |
1b53734bd
|
1820 |
write_lock_irq(&ep->lock); |
86c051793
|
1821 |
__remove_wait_queue(&ep->wq, &wait); |
1b53734bd
|
1822 |
write_unlock_irq(&ep->lock); |
86c051793
|
1823 |
} |
1da177e4c
|
1824 1825 |
return res; } |
22bacca48
|
1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 |
/** * 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
|
1846 |
struct eventpoll *ep_tovisit; |
22bacca48
|
1847 1848 |
struct rb_node *rbp; struct epitem *epi; |
d8805e633
|
1849 |
mutex_lock_nested(&ep->mtx, call_nests + 1); |
28d82dc1c
|
1850 1851 |
ep->visited = 1; list_add(&ep->visited_list_link, &visited_list); |
b2ac2ea62
|
1852 |
for (rbp = rb_first_cached(&ep->rbr); rbp; rbp = rb_next(rbp)) { |
22bacca48
|
1853 1854 |
epi = rb_entry(rbp, struct epitem, rbn); if (unlikely(is_file_epoll(epi->ffd.file))) { |
28d82dc1c
|
1855 1856 1857 |
ep_tovisit = epi->ffd.file->private_data; if (ep_tovisit->visited) continue; |
74bdc1298
|
1858 |
error = ep_call_nested(&poll_loop_ncalls, |
28d82dc1c
|
1859 1860 |
ep_loop_check_proc, epi->ffd.file, ep_tovisit, current); |
22bacca48
|
1861 1862 |
if (error != 0) break; |
28d82dc1c
|
1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 |
} 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(). */ if (list_empty(&epi->ffd.file->f_tfile_llink)) list_add(&epi->ffd.file->f_tfile_llink, &tfile_check_list); |
22bacca48
|
1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 |
} } 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) { |
28d82dc1c
|
1895 1896 |
int ret; struct eventpoll *ep_cur, *ep_next; |
74bdc1298
|
1897 |
ret = ep_call_nested(&poll_loop_ncalls, |
22bacca48
|
1898 |
ep_loop_check_proc, file, ep, current); |
28d82dc1c
|
1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 |
/* clear visited list */ list_for_each_entry_safe(ep_cur, ep_next, &visited_list, visited_list_link) { ep_cur->visited = 0; list_del(&ep_cur->visited_list_link); } return ret; } 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); } INIT_LIST_HEAD(&tfile_check_list); |
22bacca48
|
1919 |
} |
7699acd13
|
1920 |
/* |
523723bb5
|
1921 |
* Open an eventpoll file descriptor. |
7699acd13
|
1922 |
*/ |
791eb22ee
|
1923 |
static int do_epoll_create(int flags) |
7699acd13
|
1924 |
{ |
28d82dc1c
|
1925 |
int error, fd; |
bb57c3edc
|
1926 |
struct eventpoll *ep = NULL; |
28d82dc1c
|
1927 |
struct file *file; |
7699acd13
|
1928 |
|
e38b36f32
|
1929 1930 |
/* Check the EPOLL_* constant for consistency. */ BUILD_BUG_ON(EPOLL_CLOEXEC != O_CLOEXEC); |
296e236e9
|
1931 1932 |
if (flags & ~EPOLL_CLOEXEC) return -EINVAL; |
7699acd13
|
1933 |
/* |
bb57c3edc
|
1934 |
* Create the internal data structure ("struct eventpoll"). |
7699acd13
|
1935 |
*/ |
9fe5ad9c8
|
1936 |
error = ep_alloc(&ep); |
bb57c3edc
|
1937 1938 |
if (error < 0) return error; |
7699acd13
|
1939 1940 |
/* * Creates all the items needed to setup an eventpoll file. That is, |
2030a42ce
|
1941 |
* a file structure and a free file descriptor. |
7699acd13
|
1942 |
*/ |
28d82dc1c
|
1943 1944 1945 1946 1947 1948 |
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
|
1949 |
O_RDWR | (flags & O_CLOEXEC)); |
28d82dc1c
|
1950 1951 1952 1953 |
if (IS_ERR(file)) { error = PTR_ERR(file); goto out_free_fd; } |
28d82dc1c
|
1954 |
ep->file = file; |
98022748f
|
1955 |
fd_install(fd, file); |
28d82dc1c
|
1956 1957 1958 1959 1960 1961 |
return fd; out_free_fd: put_unused_fd(fd); out_free_ep: ep_free(ep); |
bb57c3edc
|
1962 |
return error; |
7699acd13
|
1963 |
} |
791eb22ee
|
1964 1965 1966 1967 |
SYSCALL_DEFINE1(epoll_create1, int, flags) { return do_epoll_create(flags); } |
5a8a82b1d
|
1968 |
SYSCALL_DEFINE1(epoll_create, int, size) |
a0998b50c
|
1969 |
{ |
bfe3891a5
|
1970 |
if (size <= 0) |
9fe5ad9c8
|
1971 |
return -EINVAL; |
791eb22ee
|
1972 |
return do_epoll_create(0); |
a0998b50c
|
1973 |
} |
39220e8d4
|
1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 |
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
|
1988 1989 |
{ int error; |
67347fe4e
|
1990 |
int full_check = 0; |
7e3fb5842
|
1991 |
struct fd f, tf; |
7699acd13
|
1992 1993 |
struct eventpoll *ep; struct epitem *epi; |
67347fe4e
|
1994 |
struct eventpoll *tep = NULL; |
7699acd13
|
1995 |
|
7699acd13
|
1996 |
error = -EBADF; |
7e3fb5842
|
1997 1998 |
f = fdget(epfd); if (!f.file) |
7699acd13
|
1999 2000 2001 |
goto error_return; /* Get the "struct file *" for the target file */ |
7e3fb5842
|
2002 2003 |
tf = fdget(fd); if (!tf.file) |
7699acd13
|
2004 2005 2006 2007 |
goto error_fput; /* The target file descriptor must support poll */ error = -EPERM; |
9965ed174
|
2008 |
if (!file_can_poll(tf.file)) |
7699acd13
|
2009 |
goto error_tgt_fput; |
4d7e30d98
|
2010 |
/* Check if EPOLLWAKEUP is allowed */ |
c680e41b3
|
2011 |
if (ep_op_has_event(op)) |
58e41a44c
|
2012 |
ep_take_care_of_epollwakeup(epds); |
4d7e30d98
|
2013 |
|
7699acd13
|
2014 2015 2016 2017 2018 2019 |
/* * 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
|
2020 |
if (f.file == tf.file || !is_file_epoll(f.file)) |
7699acd13
|
2021 2022 2023 |
goto error_tgt_fput; /* |
df0108c5d
|
2024 2025 2026 2027 |
* 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
|
2028 |
if (ep_op_has_event(op) && (epds->events & EPOLLEXCLUSIVE)) { |
b6a515c8a
|
2029 2030 2031 |
if (op == EPOLL_CTL_MOD) goto error_tgt_fput; if (op == EPOLL_CTL_ADD && (is_file_epoll(tf.file) || |
58e41a44c
|
2032 |
(epds->events & ~EPOLLEXCLUSIVE_OK_BITS))) |
b6a515c8a
|
2033 2034 |
goto error_tgt_fput; } |
df0108c5d
|
2035 2036 |
/* |
7699acd13
|
2037 2038 2039 |
* At this point it is safe to assume that the "private_data" contains * our own data structure. */ |
7e3fb5842
|
2040 |
ep = f.file->private_data; |
7699acd13
|
2041 |
|
22bacca48
|
2042 2043 2044 |
/* * When we insert an epoll file descriptor, inside another epoll file * descriptor, there is the change of creating closed loops, which are |
28d82dc1c
|
2045 2046 2047 2048 |
* 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
|
2049 |
* |
67347fe4e
|
2050 2051 2052 2053 2054 2055 |
* 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
|
2056 |
*/ |
39220e8d4
|
2057 2058 2059 |
error = epoll_mutex_lock(&ep->mtx, 0, nonblock); if (error) goto error_tgt_fput; |
28d82dc1c
|
2060 |
if (op == EPOLL_CTL_ADD) { |
67347fe4e
|
2061 2062 |
if (!list_empty(&f.file->f_ep_links) || is_file_epoll(tf.file)) { |
67347fe4e
|
2063 |
mutex_unlock(&ep->mtx); |
39220e8d4
|
2064 2065 2066 2067 |
error = epoll_mutex_lock(&epmutex, 0, nonblock); if (error) goto error_tgt_fput; full_check = 1; |
67347fe4e
|
2068 2069 2070 2071 2072 2073 2074 2075 2076 |
if (is_file_epoll(tf.file)) { error = -ELOOP; if (ep_loop_check(ep, tf.file) != 0) { clear_tfile_check_list(); goto error_tgt_fput; } } else list_add(&tf.file->f_tfile_llink, &tfile_check_list); |
39220e8d4
|
2077 2078 2079 2080 2081 2082 |
error = epoll_mutex_lock(&ep->mtx, 0, nonblock); if (error) { out_del: list_del(&tf.file->f_tfile_llink); goto error_tgt_fput; } |
67347fe4e
|
2083 2084 |
if (is_file_epoll(tf.file)) { tep = tf.file->private_data; |
39220e8d4
|
2085 2086 2087 2088 2089 |
error = epoll_mutex_lock(&tep->mtx, 1, nonblock); if (error) { mutex_unlock(&ep->mtx); goto out_del; } |
13d518074
|
2090 |
} |
67347fe4e
|
2091 2092 |
} } |
7699acd13
|
2093 |
|
67647d0fb
|
2094 2095 2096 2097 2098 |
/* * 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
|
2099 |
epi = ep_find(ep, tf.file, fd); |
7699acd13
|
2100 2101 2102 2103 2104 |
error = -EINVAL; switch (op) { case EPOLL_CTL_ADD: if (!epi) { |
58e41a44c
|
2105 2106 |
epds->events |= EPOLLERR | EPOLLHUP; error = ep_insert(ep, epds, tf.file, fd, full_check); |
7699acd13
|
2107 2108 |
} else error = -EEXIST; |
67347fe4e
|
2109 2110 |
if (full_check) clear_tfile_check_list(); |
7699acd13
|
2111 2112 2113 2114 2115 2116 2117 2118 2119 |
break; case EPOLL_CTL_DEL: if (epi) error = ep_remove(ep, epi); else error = -ENOENT; break; case EPOLL_CTL_MOD: if (epi) { |
b6a515c8a
|
2120 |
if (!(epi->event.events & EPOLLEXCLUSIVE)) { |
58e41a44c
|
2121 2122 |
epds->events |= EPOLLERR | EPOLLHUP; error = ep_modify(ep, epi, epds); |
b6a515c8a
|
2123 |
} |
7699acd13
|
2124 2125 2126 2127 |
} else error = -ENOENT; break; } |
67347fe4e
|
2128 2129 |
if (tep != NULL) mutex_unlock(&tep->mtx); |
d47de16c7
|
2130 |
mutex_unlock(&ep->mtx); |
7699acd13
|
2131 2132 |
error_tgt_fput: |
67347fe4e
|
2133 |
if (full_check) |
22bacca48
|
2134 |
mutex_unlock(&epmutex); |
7e3fb5842
|
2135 |
fdput(tf); |
7699acd13
|
2136 |
error_fput: |
7e3fb5842
|
2137 |
fdput(f); |
7699acd13
|
2138 |
error_return: |
7699acd13
|
2139 2140 2141 2142 2143 |
return error; } /* |
58e41a44c
|
2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 |
* 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
|
2156 |
return do_epoll_ctl(epfd, op, fd, &epds, false); |
58e41a44c
|
2157 2158 2159 |
} /* |
7699acd13
|
2160 2161 2162 |
* Implement the event wait interface for the eventpoll file. It is the kernel * part of the user space epoll_wait(2). */ |
791eb22ee
|
2163 2164 |
static int do_epoll_wait(int epfd, struct epoll_event __user *events, int maxevents, int timeout) |
7699acd13
|
2165 |
{ |
2903ff019
|
2166 2167 |
int error; struct fd f; |
7699acd13
|
2168 |
struct eventpoll *ep; |
7699acd13
|
2169 2170 2171 2172 2173 |
/* 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
|
2174 |
if (!access_ok(events, maxevents * sizeof(struct epoll_event))) |
2903ff019
|
2175 |
return -EFAULT; |
7699acd13
|
2176 2177 |
/* Get the "struct file *" for the eventpoll file */ |
2903ff019
|
2178 2179 2180 |
f = fdget(epfd); if (!f.file) return -EBADF; |
7699acd13
|
2181 2182 2183 2184 2185 2186 |
/* * We have to check that the file structure underneath the fd * the user passed to us _is_ an eventpoll file. */ error = -EINVAL; |
2903ff019
|
2187 |
if (!is_file_epoll(f.file)) |
7699acd13
|
2188 2189 2190 2191 2192 2193 |
goto error_fput; /* * At this point it is safe to assume that the "private_data" contains * our own data structure. */ |
2903ff019
|
2194 |
ep = f.file->private_data; |
7699acd13
|
2195 2196 2197 2198 2199 |
/* Time to fish for events ... */ error = ep_poll(ep, events, maxevents, timeout); error_fput: |
2903ff019
|
2200 |
fdput(f); |
7699acd13
|
2201 2202 |
return error; } |
791eb22ee
|
2203 2204 2205 2206 2207 |
SYSCALL_DEFINE4(epoll_wait, int, epfd, struct epoll_event __user *, events, int, maxevents, int, timeout) { return do_epoll_wait(epfd, events, maxevents, timeout); } |
7699acd13
|
2208 2209 2210 2211 |
/* * Implement the event wait interface for the eventpoll file. It is the kernel * part of the user space epoll_pwait(2). */ |
5a8a82b1d
|
2212 2213 2214 |
SYSCALL_DEFINE6(epoll_pwait, int, epfd, struct epoll_event __user *, events, int, maxevents, int, timeout, const sigset_t __user *, sigmask, size_t, sigsetsize) |
7699acd13
|
2215 2216 |
{ int error; |
7699acd13
|
2217 2218 2219 2220 2221 |
/* * If the caller wants a certain signal mask to be set during the wait, * we apply it here. */ |
b772434be
|
2222 |
error = set_user_sigmask(sigmask, sigsetsize); |
ded653ccb
|
2223 2224 |
if (error) return error; |
7699acd13
|
2225 |
|
791eb22ee
|
2226 |
error = do_epoll_wait(epfd, events, maxevents, timeout); |
b772434be
|
2227 |
restore_saved_sigmask_unless(error == -EINTR); |
7699acd13
|
2228 2229 2230 |
return error; } |
35280bd4a
|
2231 2232 2233 2234 2235 2236 2237 2238 |
#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
|
2239 2240 2241 2242 2243 |
/* * If the caller wants a certain signal mask to be set during the wait, * we apply it here. */ |
b772434be
|
2244 |
err = set_compat_user_sigmask(sigmask, sigsetsize); |
ded653ccb
|
2245 2246 |
if (err) return err; |
35280bd4a
|
2247 |
|
791eb22ee
|
2248 |
err = do_epoll_wait(epfd, events, maxevents, timeout); |
b772434be
|
2249 |
restore_saved_sigmask_unless(err == -EINTR); |
35280bd4a
|
2250 2251 2252 2253 |
return err; } #endif |
1da177e4c
|
2254 2255 |
static int __init eventpoll_init(void) { |
7ef9964e6
|
2256 2257 2258 |
struct sysinfo si; si_meminfo(&si); |
9df04e1f2
|
2259 2260 2261 2262 |
/* * Allows top 4% of lomem to be allocated for epoll watches (per user). */ max_user_watches = (((si.totalram - si.totalhigh) / 25) << PAGE_SHIFT) / |
7ef9964e6
|
2263 |
EP_ITEM_COST; |
52bd19f76
|
2264 |
BUG_ON(max_user_watches < 0); |
1da177e4c
|
2265 |
|
22bacca48
|
2266 2267 2268 2269 2270 |
/* * Initialize the structure used to perform epoll file descriptor * inclusion loops checks. */ ep_nested_calls_init(&poll_loop_ncalls); |
39732ca5a
|
2271 2272 2273 2274 2275 |
/* * 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
|
2276 2277 |
/* Allocates slab cache used to allocate "struct epitem" items */ epi_cache = kmem_cache_create("eventpoll_epi", sizeof(struct epitem), |
2ae928a94
|
2278 |
0, SLAB_HWCACHE_ALIGN|SLAB_PANIC|SLAB_ACCOUNT, NULL); |
1da177e4c
|
2279 2280 2281 |
/* Allocates slab cache used to allocate "struct eppoll_entry" */ pwq_cache = kmem_cache_create("eventpoll_pwq", |
2ae928a94
|
2282 |
sizeof(struct eppoll_entry), 0, SLAB_PANIC|SLAB_ACCOUNT, NULL); |
1da177e4c
|
2283 |
|
1da177e4c
|
2284 |
return 0; |
1da177e4c
|
2285 |
} |
cea692418
|
2286 |
fs_initcall(eventpoll_init); |