Blame view
fs/eventpoll.c
64.7 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 |
|
80285b75c
|
106 107 108 109 |
/* Wait structure used by the poll hooks */ struct eppoll_entry { /* List header used to link this structure to the "struct epitem" */ struct eppoll_entry *next; |
1da177e4c
|
110 |
|
80285b75c
|
111 112 113 114 115 116 117 118 119 120 121 |
/* The "base" pointer is set to the container "struct epitem" */ struct epitem *base; /* * Wait queue item that will be linked to the target file wait * queue head. */ wait_queue_entry_t wait; /* The wait queue head that linked the "wait" wait queue item */ wait_queue_head_t *whead; |
1da177e4c
|
122 123 124 |
}; /* |
d47de16c7
|
125 126 |
* Each file descriptor added to the eventpoll interface will * have an entry of this type linked to the "rbr" RB tree. |
39732ca5a
|
127 128 |
* 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
|
129 130 |
*/ struct epitem { |
ae10b2b4e
|
131 132 133 134 135 136 |
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
|
137 138 139 |
/* List header used to link this structure to the eventpoll ready list */ struct list_head rdllink; |
c7ea76302
|
140 141 142 143 144 |
/* * Works together "struct eventpoll"->ovflist in keeping the * single linked chain of items. */ struct epitem *next; |
d47de16c7
|
145 146 |
/* The file descriptor information this item refers to */ struct epoll_filefd ffd; |
d47de16c7
|
147 |
/* List containing poll wait queues */ |
80285b75c
|
148 |
struct eppoll_entry *pwqlist; |
d47de16c7
|
149 150 151 |
/* The "container" of this item */ struct eventpoll *ep; |
d47de16c7
|
152 |
/* List header used to link this item to the "struct file" items list */ |
44cdc1d95
|
153 |
struct hlist_node fllink; |
d47de16c7
|
154 |
|
4d7e30d98
|
155 |
/* wakeup_source used when EPOLLWAKEUP is set */ |
eea1d5859
|
156 |
struct wakeup_source __rcu *ws; |
4d7e30d98
|
157 |
|
c7ea76302
|
158 159 |
/* The structure that describe the interested events and the source fd */ struct epoll_event event; |
d47de16c7
|
160 161 162 |
}; /* |
1da177e4c
|
163 |
* This structure is stored inside the "private_data" member of the file |
bf6a41db7
|
164 |
* structure and represents the main data structure for the eventpoll |
1da177e4c
|
165 166 167 |
* interface. */ struct eventpoll { |
1da177e4c
|
168 |
/* |
d47de16c7
|
169 170 171 172 |
* 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
|
173 |
*/ |
d47de16c7
|
174 |
struct mutex mtx; |
1da177e4c
|
175 176 177 178 179 180 181 182 183 |
/* 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
|
184 185 |
/* Lock which protects rdllist and ovflist */ rwlock_t lock; |
67647d0fb
|
186 |
/* RB tree root used to store monitored fd structs */ |
b2ac2ea62
|
187 |
struct rb_root_cached rbr; |
d47de16c7
|
188 189 190 |
/* * This is a single linked list that chains all the "struct epitem" that |
25985edce
|
191 |
* happened while transferring ready events to userspace w/out |
a218cc491
|
192 |
* holding ->lock. |
d47de16c7
|
193 194 |
*/ struct epitem *ovflist; |
7ef9964e6
|
195 |
|
4d7e30d98
|
196 197 |
/* wakeup_source used when ep_scan_ready_list is running */ struct wakeup_source *ws; |
7ef9964e6
|
198 199 |
/* The user that created the eventpoll descriptor */ struct user_struct *user; |
28d82dc1c
|
200 201 202 203 |
struct file *file; /* used to optimize loop detection check */ |
18306c404
|
204 |
u64 gen; |
319c15174
|
205 |
struct hlist_head refs; |
bf3b9f637
|
206 207 208 209 210 |
#ifdef CONFIG_NET_RX_BUSY_POLL /* used to track busy poll napi_id */ unsigned int napi_id; #endif |
efcdd350d
|
211 212 213 214 215 |
#ifdef CONFIG_DEBUG_LOCK_ALLOC /* tracks wakeup nests for lockdep validation */ u8 nests; #endif |
1da177e4c
|
216 |
}; |
1da177e4c
|
217 218 219 220 221 |
/* Wrapper struct used by poll queueing */ struct ep_pqueue { poll_table pt; struct epitem *epi; }; |
1da177e4c
|
222 |
/* |
7ef9964e6
|
223 224 |
* Configuration options available inside /proc/sys/fs/epoll/ */ |
7ef9964e6
|
225 |
/* Maximum number of epoll watched descriptors, per user */ |
52bd19f76
|
226 |
static long max_user_watches __read_mostly; |
7ef9964e6
|
227 228 |
/* |
d47de16c7
|
229 |
* This mutex is used to serialize ep_free() and eventpoll_release_file(). |
1da177e4c
|
230 |
*/ |
7ef9964e6
|
231 |
static DEFINE_MUTEX(epmutex); |
1da177e4c
|
232 |
|
18306c404
|
233 |
static u64 loop_check_gen = 0; |
22bacca48
|
234 |
/* Used to check for epoll file descriptor inclusion loops */ |
6a3890c47
|
235 |
static struct eventpoll *inserting_into; |
22bacca48
|
236 |
|
1da177e4c
|
237 |
/* Slab cache used to allocate "struct epitem" */ |
e18b890bb
|
238 |
static struct kmem_cache *epi_cache __read_mostly; |
1da177e4c
|
239 240 |
/* Slab cache used to allocate "struct eppoll_entry" */ |
e18b890bb
|
241 |
static struct kmem_cache *pwq_cache __read_mostly; |
1da177e4c
|
242 |
|
28d82dc1c
|
243 244 245 246 |
/* * List of files with newly added links, where we may need to limit the number * of emanating paths. Protected by the epmutex. */ |
319c15174
|
247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 |
struct epitems_head { struct hlist_head epitems; struct epitems_head *next; }; static struct epitems_head *tfile_check_list = EP_UNACTIVE_PTR; static struct kmem_cache *ephead_cache __read_mostly; static inline void free_ephead(struct epitems_head *head) { if (head) kmem_cache_free(ephead_cache, head); } static void list_file(struct file *file) { struct epitems_head *head; head = container_of(file->f_ep, struct epitems_head, epitems); if (!head->next) { head->next = tfile_check_list; tfile_check_list = head; } } static void unlist_file(struct epitems_head *head) { struct epitems_head *to_free = head; struct hlist_node *p = rcu_dereference(hlist_first_rcu(&head->epitems)); if (p) { struct epitem *epi= container_of(p, struct epitem, fllink); spin_lock(&epi->ffd.file->f_lock); if (!hlist_empty(&head->epitems)) to_free = NULL; head->next = NULL; spin_unlock(&epi->ffd.file->f_lock); } free_ephead(to_free); } |
28d82dc1c
|
286 |
|
7ef9964e6
|
287 288 289 |
#ifdef CONFIG_SYSCTL #include <linux/sysctl.h> |
eec4844fa
|
290 |
static long long_zero; |
52bd19f76
|
291 |
static long long_max = LONG_MAX; |
7ef9964e6
|
292 |
|
1f7e0616c
|
293 |
struct ctl_table epoll_table[] = { |
7ef9964e6
|
294 |
{ |
7ef9964e6
|
295 296 |
.procname = "max_user_watches", .data = &max_user_watches, |
52bd19f76
|
297 |
.maxlen = sizeof(max_user_watches), |
7ef9964e6
|
298 |
.mode = 0644, |
52bd19f76
|
299 |
.proc_handler = proc_doulongvec_minmax, |
eec4844fa
|
300 |
.extra1 = &long_zero, |
52bd19f76
|
301 |
.extra2 = &long_max, |
7ef9964e6
|
302 |
}, |
ab09203e3
|
303 |
{ } |
7ef9964e6
|
304 305 |
}; #endif /* CONFIG_SYSCTL */ |
28d82dc1c
|
306 307 308 309 310 311 |
static const struct file_operations eventpoll_fops; static inline int is_file_epoll(struct file *f) { return f->f_op == &eventpoll_fops; } |
b030a4dd6
|
312 |
|
67647d0fb
|
313 |
/* Setup the structure that is used as key for the RB tree */ |
b030a4dd6
|
314 315 316 317 318 319 |
static inline void ep_set_ffd(struct epoll_filefd *ffd, struct file *file, int fd) { ffd->file = file; ffd->fd = fd; } |
67647d0fb
|
320 |
/* Compare RB tree keys */ |
b030a4dd6
|
321 322 323 324 325 326 |
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
|
327 |
/* Tells us if the item is currently linked */ |
992991c03
|
328 |
static inline int ep_is_linked(struct epitem *epi) |
b030a4dd6
|
329 |
{ |
992991c03
|
330 |
return !list_empty(&epi->rdllink); |
b030a4dd6
|
331 |
} |
ac6424b98
|
332 |
static inline struct eppoll_entry *ep_pwq_from_wait(wait_queue_entry_t *p) |
971316f05
|
333 334 335 |
{ return container_of(p, struct eppoll_entry, wait); } |
b030a4dd6
|
336 |
/* Get the "struct epitem" from a wait queue pointer */ |
ac6424b98
|
337 |
static inline struct epitem *ep_item_from_wait(wait_queue_entry_t *p) |
b030a4dd6
|
338 339 340 |
{ return container_of(p, struct eppoll_entry, wait)->base; } |
5071f97ec
|
341 |
/** |
3fb0e584a
|
342 343 344 345 |
* ep_events_available - Checks if ready events might be available. * * @ep: Pointer to the eventpoll context. * |
a6c67fee9
|
346 347 |
* Return: a value different than %zero if ready events are available, * or %zero otherwise. |
3fb0e584a
|
348 349 350 |
*/ static inline int ep_events_available(struct eventpoll *ep) { |
c5a282e96
|
351 352 |
return !list_empty_careful(&ep->rdllist) || READ_ONCE(ep->ovflist) != EP_UNACTIVE_PTR; |
3fb0e584a
|
353 |
} |
bf3b9f637
|
354 355 356 357 358 359 360 |
#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
|
361 362 363 364 365 366 367 |
/* * 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 */ |
1493c47fb
|
368 |
static bool ep_busy_loop(struct eventpoll *ep, int nonblock) |
bf3b9f637
|
369 |
{ |
bf3b9f637
|
370 |
unsigned int napi_id = READ_ONCE(ep->napi_id); |
1493c47fb
|
371 |
if ((napi_id >= MIN_NAPI_ID) && net_busy_loop_on()) { |
7c951cafc
|
372 373 |
napi_busy_loop(napi_id, nonblock ? NULL : ep_busy_loop_end, ep, false, BUSY_POLL_BUDGET); |
1493c47fb
|
374 375 376 377 378 379 380 |
if (ep_events_available(ep)) return true; /* * 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. */ |
bf3b9f637
|
381 |
ep->napi_id = 0; |
1493c47fb
|
382 383 384 |
return false; } return false; |
bf3b9f637
|
385 386 387 388 389 390 391 |
} /* * Set epoll busy poll NAPI ID from sk. */ static inline void ep_set_busy_poll_napi_id(struct epitem *epi) { |
bf3b9f637
|
392 393 394 395 |
struct eventpoll *ep; unsigned int napi_id; struct socket *sock; struct sock *sk; |
bf3b9f637
|
396 397 398 |
if (!net_busy_loop_on()) return; |
dba4a9256
|
399 |
sock = sock_from_file(epi->ffd.file); |
bf3b9f637
|
400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 |
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
|
419 |
} |
514056d50
|
420 |
#else |
1493c47fb
|
421 |
static inline bool ep_busy_loop(struct eventpoll *ep, int nonblock) |
514056d50
|
422 |
{ |
1493c47fb
|
423 |
return false; |
514056d50
|
424 425 426 427 428 429 430 |
} static inline void ep_set_busy_poll_napi_id(struct epitem *epi) { } #endif /* CONFIG_NET_RX_BUSY_POLL */ |
02edc6fc4
|
431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 |
/* * 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
|
456 |
#ifdef CONFIG_DEBUG_LOCK_ALLOC |
57a173bdf
|
457 |
|
efcdd350d
|
458 |
static void ep_poll_safewake(struct eventpoll *ep, struct epitem *epi) |
5071f97ec
|
459 |
{ |
efcdd350d
|
460 |
struct eventpoll *ep_src; |
f6520c520
|
461 |
unsigned long flags; |
efcdd350d
|
462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 |
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
|
492 |
} |
57a173bdf
|
493 |
#else |
efcdd350d
|
494 |
static void ep_poll_safewake(struct eventpoll *ep, struct epitem *epi) |
57a173bdf
|
495 |
{ |
efcdd350d
|
496 |
wake_up_poll(&ep->poll_wait, EPOLLIN); |
57a173bdf
|
497 498 499 |
} #endif |
971316f05
|
500 501 502 503 504 |
static void ep_remove_wait_queue(struct eppoll_entry *pwq) { wait_queue_head_t *whead; rcu_read_lock(); |
138e4ad67
|
505 506 507 508 509 510 511 |
/* * 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
|
512 513 514 515 |
if (whead) remove_wait_queue(whead, &pwq->wait); rcu_read_unlock(); } |
1da177e4c
|
516 |
/* |
d1bc90dd5
|
517 518 519 |
* This function unregisters poll callbacks from the associated file * descriptor. Must be called with "mtx" held (or "epmutex" if called from * ep_free). |
1da177e4c
|
520 |
*/ |
7699acd13
|
521 |
static void ep_unregister_pollwait(struct eventpoll *ep, struct epitem *epi) |
1da177e4c
|
522 |
{ |
80285b75c
|
523 |
struct eppoll_entry **p = &epi->pwqlist; |
7699acd13
|
524 |
struct eppoll_entry *pwq; |
1da177e4c
|
525 |
|
80285b75c
|
526 527 |
while ((pwq = *p) != NULL) { *p = pwq->next; |
971316f05
|
528 |
ep_remove_wait_queue(pwq); |
d1bc90dd5
|
529 |
kmem_cache_free(pwq_cache, pwq); |
1da177e4c
|
530 |
} |
1da177e4c
|
531 |
} |
eea1d5859
|
532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 |
/* 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
|
563 |
|
57804b1cc
|
564 565 566 |
/* * ep->mutex needs to be held because we could be hit by * eventpoll_release_file() and epoll_ctl(). |
5071f97ec
|
567 |
*/ |
57804b1cc
|
568 |
static void ep_start_scan(struct eventpoll *ep, struct list_head *txlist) |
5071f97ec
|
569 |
{ |
5071f97ec
|
570 571 572 573 574 575 576 577 |
/* * 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. */ |
57804b1cc
|
578 |
lockdep_assert_irqs_enabled(); |
a218cc491
|
579 |
write_lock_irq(&ep->lock); |
db502f8a3
|
580 |
list_splice_init(&ep->rdllist, txlist); |
c5a282e96
|
581 |
WRITE_ONCE(ep->ovflist, NULL); |
a218cc491
|
582 |
write_unlock_irq(&ep->lock); |
db502f8a3
|
583 |
} |
5071f97ec
|
584 |
|
db502f8a3
|
585 |
static void ep_done_scan(struct eventpoll *ep, |
db502f8a3
|
586 587 588 |
struct list_head *txlist) { struct epitem *epi, *nepi; |
5071f97ec
|
589 |
|
a218cc491
|
590 |
write_lock_irq(&ep->lock); |
5071f97ec
|
591 592 593 594 595 |
/* * 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
|
596 |
for (nepi = READ_ONCE(ep->ovflist); (epi = nepi) != NULL; |
5071f97ec
|
597 598 599 600 601 602 603 |
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
|
604 |
if (!ep_is_linked(epi)) { |
c141175d0
|
605 606 607 608 609 |
/* * ->ovflist is LIFO, so we have to reverse it in order * to keep in FIFO. */ list_add(&epi->rdllink, &ep->rdllist); |
eea1d5859
|
610 |
ep_pm_stay_awake(epi); |
4d7e30d98
|
611 |
} |
5071f97ec
|
612 613 614 615 616 617 |
} /* * 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
|
618 |
WRITE_ONCE(ep->ovflist, EP_UNACTIVE_PTR); |
5071f97ec
|
619 620 621 622 |
/* * Quickly re-inject items left on "txlist". */ |
db502f8a3
|
623 |
list_splice(txlist, &ep->rdllist); |
4d7e30d98
|
624 |
__pm_relax(ep->ws); |
7fab29e35
|
625 626 627 628 629 |
if (!list_empty(&ep->rdllist)) { if (waitqueue_active(&ep->wq)) wake_up(&ep->wq); } |
a218cc491
|
630 |
write_unlock_irq(&ep->lock); |
5071f97ec
|
631 |
} |
ae10b2b4e
|
632 633 634 635 636 |
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
|
637 |
/* |
7699acd13
|
638 |
* Removes a "struct epitem" from the eventpoll RB tree and deallocates |
c7ea76302
|
639 |
* all the associated resources. Must be called with "mtx" held. |
7699acd13
|
640 641 642 |
*/ static int ep_remove(struct eventpoll *ep, struct epitem *epi) { |
7699acd13
|
643 |
struct file *file = epi->ffd.file; |
319c15174
|
644 645 |
struct epitems_head *to_free; struct hlist_head *head; |
1da177e4c
|
646 |
|
92e641784
|
647 |
lockdep_assert_irqs_enabled(); |
1da177e4c
|
648 |
/* |
ee8ef0a4b
|
649 |
* Removes poll wait queue hooks. |
1da177e4c
|
650 |
*/ |
7699acd13
|
651 |
ep_unregister_pollwait(ep, epi); |
1da177e4c
|
652 |
|
7699acd13
|
653 |
/* Remove the current item from the list of epoll hooks */ |
684999149
|
654 |
spin_lock(&file->f_lock); |
319c15174
|
655 656 657 658 659 660 661 662 663 664 665 |
to_free = NULL; head = file->f_ep; if (head->first == &epi->fllink && !epi->fllink.next) { file->f_ep = NULL; if (!is_file_epoll(file)) { struct epitems_head *v; v = container_of(head, struct epitems_head, epitems); if (!smp_load_acquire(&v->next)) to_free = v; } } |
44cdc1d95
|
666 |
hlist_del_rcu(&epi->fllink); |
684999149
|
667 |
spin_unlock(&file->f_lock); |
319c15174
|
668 |
free_ephead(to_free); |
1da177e4c
|
669 |
|
b2ac2ea62
|
670 |
rb_erase_cached(&epi->rbn, &ep->rbr); |
1da177e4c
|
671 |
|
a218cc491
|
672 |
write_lock_irq(&ep->lock); |
992991c03
|
673 |
if (ep_is_linked(epi)) |
c7ea76302
|
674 |
list_del_init(&epi->rdllink); |
a218cc491
|
675 |
write_unlock_irq(&ep->lock); |
1da177e4c
|
676 |
|
eea1d5859
|
677 |
wakeup_source_unregister(ep_wakeup_source(epi)); |
ae10b2b4e
|
678 679 680 681 682 683 684 685 |
/* * 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
|
686 |
|
1e1c15839
|
687 |
percpu_counter_dec(&ep->user->epoll_watches); |
7ef9964e6
|
688 |
|
c7ea76302
|
689 |
return 0; |
1da177e4c
|
690 |
} |
7699acd13
|
691 |
static void ep_free(struct eventpoll *ep) |
1da177e4c
|
692 |
{ |
7699acd13
|
693 694 |
struct rb_node *rbp; struct epitem *epi; |
1da177e4c
|
695 |
|
7699acd13
|
696 697 |
/* We need to release all tasks waiting for these file */ if (waitqueue_active(&ep->poll_wait)) |
efcdd350d
|
698 |
ep_poll_safewake(ep, NULL); |
1da177e4c
|
699 |
|
7699acd13
|
700 701 702 |
/* * We need to lock this because we could be hit by * eventpoll_release_file() while we're freeing the "struct eventpoll". |
d47de16c7
|
703 |
* We do not need to hold "ep->mtx" here because the epoll file |
7699acd13
|
704 705 |
* 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
|
706 |
* holding "epmutex" is sufficient here. |
7699acd13
|
707 708 |
*/ mutex_lock(&epmutex); |
1da177e4c
|
709 710 |
/* |
7699acd13
|
711 |
* Walks through the whole tree by unregistering poll callbacks. |
1da177e4c
|
712 |
*/ |
b2ac2ea62
|
713 |
for (rbp = rb_first_cached(&ep->rbr); rbp; rbp = rb_next(rbp)) { |
7699acd13
|
714 715 716 |
epi = rb_entry(rbp, struct epitem, rbn); ep_unregister_pollwait(ep, epi); |
91cf5ab60
|
717 |
cond_resched(); |
7699acd13
|
718 |
} |
1da177e4c
|
719 720 |
/* |
7699acd13
|
721 722 |
* 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
|
723 |
* holding "epmutex" we can be sure that no file cleanup code will hit |
a218cc491
|
724 |
* us during this operation. So we can avoid the lock on "ep->lock". |
ddf676c38
|
725 726 |
* We do not need to lock ep->mtx, either, we only do it to prevent * a lockdep warning. |
1da177e4c
|
727 |
*/ |
ddf676c38
|
728 |
mutex_lock(&ep->mtx); |
b2ac2ea62
|
729 |
while ((rbp = rb_first_cached(&ep->rbr)) != NULL) { |
7699acd13
|
730 731 |
epi = rb_entry(rbp, struct epitem, rbn); ep_remove(ep, epi); |
91cf5ab60
|
732 |
cond_resched(); |
7699acd13
|
733 |
} |
ddf676c38
|
734 |
mutex_unlock(&ep->mtx); |
1da177e4c
|
735 |
|
7699acd13
|
736 |
mutex_unlock(&epmutex); |
d47de16c7
|
737 |
mutex_destroy(&ep->mtx); |
7ef9964e6
|
738 |
free_uid(ep->user); |
4d7e30d98
|
739 |
wakeup_source_unregister(ep->ws); |
f0ee9aabb
|
740 |
kfree(ep); |
7699acd13
|
741 |
} |
1da177e4c
|
742 |
|
7699acd13
|
743 744 745 |
static int ep_eventpoll_release(struct inode *inode, struct file *file) { struct eventpoll *ep = file->private_data; |
1da177e4c
|
746 |
|
f0ee9aabb
|
747 |
if (ep) |
7699acd13
|
748 |
ep_free(ep); |
7699acd13
|
749 |
|
7699acd13
|
750 |
return 0; |
1da177e4c
|
751 |
} |
2c0b71c1e
|
752 |
static __poll_t ep_item_poll(const struct epitem *epi, poll_table *pt, int depth); |
37b5e5212
|
753 |
|
ad9366b13
|
754 |
static __poll_t __ep_eventpoll_poll(struct file *file, poll_table *wait, int depth) |
5071f97ec
|
755 |
{ |
ad9366b13
|
756 757 |
struct eventpoll *ep = file->private_data; LIST_HEAD(txlist); |
5071f97ec
|
758 |
struct epitem *epi, *tmp; |
626cf2366
|
759 |
poll_table pt; |
2c0b71c1e
|
760 |
__poll_t res = 0; |
5071f97ec
|
761 |
|
626cf2366
|
762 |
init_poll_funcptr(&pt, NULL); |
450d89ec0
|
763 |
|
ad9366b13
|
764 765 766 767 768 769 770 771 772 |
/* Insert inside our poll wait queue */ poll_wait(file, &ep->poll_wait, wait); /* * Proceed to find out if wanted events are really available inside * the ready list. */ mutex_lock_nested(&ep->mtx, depth); ep_start_scan(ep, &txlist); |
2c0b71c1e
|
773 774 775 776 |
list_for_each_entry_safe(epi, tmp, &txlist, rdllink) { if (ep_item_poll(epi, &pt, depth + 1)) { res = EPOLLIN | EPOLLRDNORM; break; |
37b5e5212
|
777 |
} else { |
5071f97ec
|
778 779 780 781 782 |
/* * 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
|
783 |
__pm_relax(ep_wakeup_source(epi)); |
5071f97ec
|
784 |
list_del_init(&epi->rdllink); |
296e236e9
|
785 |
} |
5071f97ec
|
786 |
} |
ad9366b13
|
787 788 789 |
ep_done_scan(ep, &txlist); mutex_unlock(&ep->mtx); return res; |
5071f97ec
|
790 |
} |
37b5e5212
|
791 792 793 794 795 |
/* * 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
|
796 |
static __poll_t ep_item_poll(const struct epitem *epi, poll_table *pt, |
bec1a502d
|
797 |
int depth) |
11c5ad0ec
|
798 |
{ |
ad9366b13
|
799 |
struct file *file = epi->ffd.file; |
1ec09974d
|
800 |
__poll_t res; |
7699acd13
|
801 |
|
450d89ec0
|
802 |
pt->_key = epi->event.events; |
ad9366b13
|
803 804 805 806 |
if (!is_file_epoll(file)) res = vfs_poll(file, pt); else res = __ep_eventpoll_poll(file, pt, depth); |
1ec09974d
|
807 |
return res & epi->event.events; |
450d89ec0
|
808 |
} |
a11e1d432
|
809 |
|
a11e1d432
|
810 |
static __poll_t ep_eventpoll_poll(struct file *file, poll_table *wait) |
11c5ad0ec
|
811 |
{ |
ad9366b13
|
812 |
return __ep_eventpoll_poll(file, wait, 0); |
7699acd13
|
813 |
} |
138d22b58
|
814 |
#ifdef CONFIG_PROC_FS |
a3816ab0e
|
815 |
static void ep_show_fdinfo(struct seq_file *m, struct file *f) |
138d22b58
|
816 817 818 |
{ struct eventpoll *ep = f->private_data; struct rb_node *rbp; |
138d22b58
|
819 820 |
mutex_lock(&ep->mtx); |
b2ac2ea62
|
821 |
for (rbp = rb_first_cached(&ep->rbr); rbp; rbp = rb_next(rbp)) { |
138d22b58
|
822 |
struct epitem *epi = rb_entry(rbp, struct epitem, rbn); |
77493f04b
|
823 |
struct inode *inode = file_inode(epi->ffd.file); |
138d22b58
|
824 |
|
77493f04b
|
825 826 827 |
seq_printf(m, "tfd: %8d events: %8x data: %16llx " " pos:%lli ino:%lx sdev:%x ", |
a3816ab0e
|
828 |
epi->ffd.fd, epi->event.events, |
77493f04b
|
829 830 831 |
(long long)epi->event.data, (long long)epi->ffd.file->f_pos, inode->i_ino, inode->i_sb->s_dev); |
a3816ab0e
|
832 |
if (seq_has_overflowed(m)) |
138d22b58
|
833 834 835 |
break; } mutex_unlock(&ep->mtx); |
138d22b58
|
836 837 |
} #endif |
7699acd13
|
838 839 |
/* File callbacks that implement the eventpoll file behaviour */ static const struct file_operations eventpoll_fops = { |
138d22b58
|
840 841 842 |
#ifdef CONFIG_PROC_FS .show_fdinfo = ep_show_fdinfo, #endif |
7699acd13
|
843 |
.release = ep_eventpoll_release, |
a11e1d432
|
844 |
.poll = ep_eventpoll_poll, |
6038f373a
|
845 |
.llseek = noop_llseek, |
7699acd13
|
846 |
}; |
b611967de
|
847 |
/* |
7699acd13
|
848 849 850 |
* 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
|
851 |
*/ |
7699acd13
|
852 |
void eventpoll_release_file(struct file *file) |
b611967de
|
853 |
{ |
7699acd13
|
854 |
struct eventpoll *ep; |
44cdc1d95
|
855 856 |
struct epitem *epi; struct hlist_node *next; |
b611967de
|
857 858 |
/* |
684999149
|
859 |
* We don't want to get "file->f_lock" because it is not |
7699acd13
|
860 |
* necessary. It is not necessary because we're in the "struct file" |
25985edce
|
861 |
* cleanup path, and this means that no one is using this file anymore. |
5071f97ec
|
862 |
* So, for example, epoll_ctl() cannot hit here since if we reach this |
67647d0fb
|
863 |
* point, the file counter already went to zero and fget() would fail. |
d47de16c7
|
864 |
* The only hit might come from ep_free() but by holding the mutex |
7699acd13
|
865 |
* will correctly serialize the operation. We do need to acquire |
d47de16c7
|
866 |
* "ep->mtx" after "epmutex" because ep_remove() requires it when called |
7699acd13
|
867 |
* from anywhere but ep_free(). |
684999149
|
868 869 |
* * Besides, ep_remove() acquires the lock, so we can't hold it here. |
b611967de
|
870 |
*/ |
7699acd13
|
871 |
mutex_lock(&epmutex); |
319c15174
|
872 873 874 875 876 |
if (unlikely(!file->f_ep)) { mutex_unlock(&epmutex); return; } hlist_for_each_entry_safe(epi, next, file->f_ep, fllink) { |
7699acd13
|
877 |
ep = epi->ep; |
d8805e633
|
878 |
mutex_lock_nested(&ep->mtx, 0); |
7699acd13
|
879 |
ep_remove(ep, epi); |
d47de16c7
|
880 |
mutex_unlock(&ep->mtx); |
b611967de
|
881 |
} |
7699acd13
|
882 |
mutex_unlock(&epmutex); |
b611967de
|
883 |
} |
53d2be79d
|
884 |
static int ep_alloc(struct eventpoll **pep) |
1da177e4c
|
885 |
{ |
7ef9964e6
|
886 887 888 |
int error; struct user_struct *user; struct eventpoll *ep; |
1da177e4c
|
889 |
|
7ef9964e6
|
890 |
user = get_current_user(); |
7ef9964e6
|
891 892 893 894 |
error = -ENOMEM; ep = kzalloc(sizeof(*ep), GFP_KERNEL); if (unlikely(!ep)) goto free_uid; |
1da177e4c
|
895 |
|
d47de16c7
|
896 |
mutex_init(&ep->mtx); |
a218cc491
|
897 |
rwlock_init(&ep->lock); |
1da177e4c
|
898 899 900 |
init_waitqueue_head(&ep->wq); init_waitqueue_head(&ep->poll_wait); INIT_LIST_HEAD(&ep->rdllist); |
b2ac2ea62
|
901 |
ep->rbr = RB_ROOT_CACHED; |
d47de16c7
|
902 |
ep->ovflist = EP_UNACTIVE_PTR; |
7ef9964e6
|
903 |
ep->user = user; |
1da177e4c
|
904 |
|
53d2be79d
|
905 |
*pep = ep; |
1da177e4c
|
906 |
|
1da177e4c
|
907 |
return 0; |
7ef9964e6
|
908 909 910 911 |
free_uid: free_uid(user); return error; |
1da177e4c
|
912 |
} |
1da177e4c
|
913 |
/* |
c7ea76302
|
914 915 916 |
* 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
|
917 918 919 920 |
*/ static struct epitem *ep_find(struct eventpoll *ep, struct file *file, int fd) { int kcmp; |
1da177e4c
|
921 922 923 |
struct rb_node *rbp; struct epitem *epi, *epir = NULL; struct epoll_filefd ffd; |
b030a4dd6
|
924 |
ep_set_ffd(&ffd, file, fd); |
b2ac2ea62
|
925 |
for (rbp = ep->rbr.rb_root.rb_node; rbp; ) { |
1da177e4c
|
926 |
epi = rb_entry(rbp, struct epitem, rbn); |
b030a4dd6
|
927 |
kcmp = ep_cmp_ffd(&ffd, &epi->ffd); |
1da177e4c
|
928 929 930 931 932 |
if (kcmp > 0) rbp = rbp->rb_right; else if (kcmp < 0) rbp = rbp->rb_left; else { |
1da177e4c
|
933 934 935 936 |
epir = epi; break; } } |
1da177e4c
|
937 |
|
1da177e4c
|
938 939 |
return epir; } |
bfe3911a9
|
940 |
#ifdef CONFIG_KCMP |
0791e3644
|
941 942 943 944 |
static struct epitem *ep_find_tfd(struct eventpoll *ep, int tfd, unsigned long toff) { struct rb_node *rbp; struct epitem *epi; |
b2ac2ea62
|
945 |
for (rbp = rb_first_cached(&ep->rbr); rbp; rbp = rb_next(rbp)) { |
0791e3644
|
946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 |
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; } |
bfe3911a9
|
981 |
#endif /* CONFIG_KCMP */ |
0791e3644
|
982 |
|
a6c67fee9
|
983 |
/* |
a218cc491
|
984 985 986 987 988 989 990 991 992 993 994 |
* 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 |
a6c67fee9
|
995 |
* direction i.e. either to the tail or to the head, otherwise |
a218cc491
|
996 997 |
* concurrent access will corrupt the list. * |
a6c67fee9
|
998 |
* Return: %false if element has been already added to the list, %true |
a218cc491
|
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 1033 1034 |
* 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; } |
a6c67fee9
|
1035 |
/* |
a218cc491
|
1036 1037 1038 |
* 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. * |
a6c67fee9
|
1039 |
* Return: %false if epi element has been already chained, %true otherwise. |
a218cc491
|
1040 1041 1042 1043 |
*/ static inline bool chain_epi_lockless(struct epitem *epi) { struct eventpoll *ep = epi->ep; |
0c54a6a44
|
1044 1045 1046 |
/* Fast preliminary check */ if (epi->next != EP_UNACTIVE_PTR) return false; |
a218cc491
|
1047 1048 1049 1050 1051 1052 1053 1054 1055 |
/* 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
|
1056 |
/* |
7699acd13
|
1057 |
* This is the callback that is passed to the wait queue wakeup |
bf6a41db7
|
1058 |
* mechanism. It is called by the stored file descriptors when they |
7699acd13
|
1059 |
* have events to report. |
a218cc491
|
1060 |
* |
a6c67fee9
|
1061 1062 |
* This callback takes a read lock in order not to contend with concurrent * events from another file descriptor, thus all modifications to ->rdllist |
a218cc491
|
1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 |
* 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
|
1073 |
*/ |
ac6424b98
|
1074 |
static int ep_poll_callback(wait_queue_entry_t *wait, unsigned mode, int sync, void *key) |
1da177e4c
|
1075 |
{ |
7699acd13
|
1076 |
int pwake = 0; |
7699acd13
|
1077 1078 |
struct epitem *epi = ep_item_from_wait(wait); struct eventpoll *ep = epi->ep; |
3ad6f93e9
|
1079 |
__poll_t pollflags = key_to_poll(key); |
a218cc491
|
1080 |
unsigned long flags; |
df0108c5d
|
1081 |
int ewake = 0; |
1da177e4c
|
1082 |
|
a218cc491
|
1083 |
read_lock_irqsave(&ep->lock, flags); |
1da177e4c
|
1084 |
|
bf3b9f637
|
1085 |
ep_set_busy_poll_napi_id(epi); |
7699acd13
|
1086 1087 1088 1089 1090 1091 1092 |
/* * 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
|
1093 1094 1095 |
goto out_unlock; /* |
2dfa4eeab
|
1096 1097 1098 1099 1100 |
* 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
|
1101 |
if (pollflags && !(pollflags & epi->event.events)) |
2dfa4eeab
|
1102 1103 1104 |
goto out_unlock; /* |
bf6a41db7
|
1105 |
* If we are transferring events to userspace, we can hold no locks |
d47de16c7
|
1106 |
* (because we're accessing user memory, and because of linux f_op->poll() |
bf6a41db7
|
1107 |
* semantics). All the events that happen during that period of time are |
d47de16c7
|
1108 1109 |
* chained in ep->ovflist and requeued later on. */ |
c5a282e96
|
1110 |
if (READ_ONCE(ep->ovflist) != EP_UNACTIVE_PTR) { |
0c54a6a44
|
1111 1112 1113 1114 1115 |
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
|
1116 |
ep_pm_stay_awake_rcu(epi); |
4d7e30d98
|
1117 |
} |
7699acd13
|
1118 |
|
7699acd13
|
1119 1120 1121 1122 |
/* * Wake up ( if active ) both the eventpoll wait list and the ->poll() * wait list. */ |
df0108c5d
|
1123 |
if (waitqueue_active(&ep->wq)) { |
b6a515c8a
|
1124 |
if ((epi->event.events & EPOLLEXCLUSIVE) && |
3ad6f93e9
|
1125 1126 |
!(pollflags & POLLFREE)) { switch (pollflags & EPOLLINOUT_BITS) { |
a9a08845e
|
1127 1128 |
case EPOLLIN: if (epi->event.events & EPOLLIN) |
b6a515c8a
|
1129 1130 |
ewake = 1; break; |
a9a08845e
|
1131 1132 |
case EPOLLOUT: if (epi->event.events & EPOLLOUT) |
b6a515c8a
|
1133 1134 1135 1136 1137 1138 1139 |
ewake = 1; break; case 0: ewake = 1; break; } } |
a218cc491
|
1140 |
wake_up(&ep->wq); |
df0108c5d
|
1141 |
} |
7699acd13
|
1142 1143 |
if (waitqueue_active(&ep->poll_wait)) pwake++; |
d47de16c7
|
1144 |
out_unlock: |
a218cc491
|
1145 |
read_unlock_irqrestore(&ep->lock, flags); |
1da177e4c
|
1146 |
|
7699acd13
|
1147 1148 |
/* We have to call this outside the lock */ if (pwake) |
efcdd350d
|
1149 |
ep_poll_safewake(ep, epi); |
7699acd13
|
1150 |
|
138e4ad67
|
1151 1152 |
if (!(epi->event.events & EPOLLEXCLUSIVE)) ewake = 1; |
3ad6f93e9
|
1153 |
if (pollflags & POLLFREE) { |
138e4ad67
|
1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 |
/* * 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
|
1168 |
|
138e4ad67
|
1169 |
return ewake; |
7699acd13
|
1170 |
} |
1da177e4c
|
1171 1172 1173 1174 1175 1176 1177 1178 |
/* * 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) { |
364f374f2
|
1179 1180 |
struct ep_pqueue *epq = container_of(pt, struct ep_pqueue, pt); struct epitem *epi = epq->epi; |
1da177e4c
|
1181 |
struct eppoll_entry *pwq; |
364f374f2
|
1182 1183 1184 1185 1186 1187 1188 |
if (unlikely(!epi)) // an earlier allocation has failed return; pwq = kmem_cache_alloc(pwq_cache, GFP_KERNEL); if (unlikely(!pwq)) { epq->epi = NULL; return; |
296e236e9
|
1189 |
} |
364f374f2
|
1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 |
init_waitqueue_func_entry(&pwq->wait, ep_poll_callback); pwq->whead = whead; pwq->base = epi; if (epi->event.events & EPOLLEXCLUSIVE) add_wait_queue_exclusive(whead, &pwq->wait); else add_wait_queue(whead, &pwq->wait); pwq->next = epi->pwqlist; epi->pwqlist = pwq; |
1da177e4c
|
1200 |
} |
1da177e4c
|
1201 1202 1203 |
static void ep_rbtree_insert(struct eventpoll *ep, struct epitem *epi) { int kcmp; |
b2ac2ea62
|
1204 |
struct rb_node **p = &ep->rbr.rb_root.rb_node, *parent = NULL; |
1da177e4c
|
1205 |
struct epitem *epic; |
b2ac2ea62
|
1206 |
bool leftmost = true; |
1da177e4c
|
1207 1208 1209 1210 |
while (*p) { parent = *p; epic = rb_entry(parent, struct epitem, rbn); |
b030a4dd6
|
1211 |
kcmp = ep_cmp_ffd(&epi->ffd, &epic->ffd); |
b2ac2ea62
|
1212 |
if (kcmp > 0) { |
1da177e4c
|
1213 |
p = &parent->rb_right; |
b2ac2ea62
|
1214 1215 |
leftmost = false; } else |
1da177e4c
|
1216 1217 1218 |
p = &parent->rb_left; } rb_link_node(&epi->rbn, parent, p); |
b2ac2ea62
|
1219 |
rb_insert_color_cached(&epi->rbn, &ep->rbr, leftmost); |
1da177e4c
|
1220 |
} |
a80a6b85b
|
1221 |
|
28d82dc1c
|
1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 |
#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
|
1239 1240 1241 |
/* Allow an arbitrary number of depth 1 paths */ if (nests == 0) return 0; |
28d82dc1c
|
1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 |
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; } |
319c15174
|
1254 |
static int reverse_path_check_proc(struct hlist_head *refs, int depth) |
28d82dc1c
|
1255 1256 |
{ int error = 0; |
28d82dc1c
|
1257 |
struct epitem *epi; |
0c320f776
|
1258 |
if (depth > EP_MAX_NESTS) /* too deep nesting */ |
99d84d433
|
1259 |
return -1; |
ae10b2b4e
|
1260 |
/* CTL_DEL can remove links here, but that can't increase our count */ |
319c15174
|
1261 1262 1263 |
hlist_for_each_entry_rcu(epi, refs, fllink) { struct hlist_head *refs = &epi->ep->refs; if (hlist_empty(refs)) |
d16312a46
|
1264 1265 |
error = path_count_inc(depth); else |
319c15174
|
1266 |
error = reverse_path_check_proc(refs, depth + 1); |
d16312a46
|
1267 1268 |
if (error != 0) break; |
28d82dc1c
|
1269 1270 1271 1272 1273 |
} return error; } /** |
319c15174
|
1274 |
* reverse_path_check - The tfile_check_list is list of epitem_head, which have |
28d82dc1c
|
1275 1276 1277 1278 1279 |
* 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. * |
a6c67fee9
|
1280 1281 |
* Return: %zero if the proposed links don't create too many paths, * %-1 otherwise. |
28d82dc1c
|
1282 1283 1284 |
*/ static int reverse_path_check(void) { |
319c15174
|
1285 |
struct epitems_head *p; |
28d82dc1c
|
1286 |
|
319c15174
|
1287 1288 |
for (p = tfile_check_list; p != EP_UNACTIVE_PTR; p = p->next) { int error; |
28d82dc1c
|
1289 |
path_count_init(); |
b62d2706a
|
1290 |
rcu_read_lock(); |
319c15174
|
1291 |
error = reverse_path_check_proc(&p->epitems, 0); |
b62d2706a
|
1292 |
rcu_read_unlock(); |
28d82dc1c
|
1293 |
if (error) |
319c15174
|
1294 |
return error; |
28d82dc1c
|
1295 |
} |
319c15174
|
1296 |
return 0; |
28d82dc1c
|
1297 |
} |
4d7e30d98
|
1298 1299 |
static int ep_create_wakeup_source(struct epitem *epi) { |
3701cb59d
|
1300 |
struct name_snapshot n; |
eea1d5859
|
1301 |
struct wakeup_source *ws; |
4d7e30d98
|
1302 1303 |
if (!epi->ep->ws) { |
c8377adfa
|
1304 |
epi->ep->ws = wakeup_source_register(NULL, "eventpoll"); |
4d7e30d98
|
1305 1306 1307 |
if (!epi->ep->ws) return -ENOMEM; } |
3701cb59d
|
1308 1309 1310 |
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
|
1311 1312 |
if (!ws) |
4d7e30d98
|
1313 |
return -ENOMEM; |
eea1d5859
|
1314 |
rcu_assign_pointer(epi->ws, ws); |
4d7e30d98
|
1315 1316 1317 |
return 0; } |
eea1d5859
|
1318 1319 |
/* rare code path, only used when EPOLL_CTL_MOD removes a wakeup source */ static noinline void ep_destroy_wakeup_source(struct epitem *epi) |
4d7e30d98
|
1320 |
{ |
eea1d5859
|
1321 |
struct wakeup_source *ws = ep_wakeup_source(epi); |
d6d67e723
|
1322 |
RCU_INIT_POINTER(epi->ws, NULL); |
eea1d5859
|
1323 1324 1325 1326 1327 1328 1329 1330 |
/* * 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
|
1331 |
} |
319c15174
|
1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 |
static int attach_epitem(struct file *file, struct epitem *epi) { struct epitems_head *to_free = NULL; struct hlist_head *head = NULL; struct eventpoll *ep = NULL; if (is_file_epoll(file)) ep = file->private_data; if (ep) { head = &ep->refs; } else if (!READ_ONCE(file->f_ep)) { allocate: to_free = kmem_cache_zalloc(ephead_cache, GFP_KERNEL); if (!to_free) return -ENOMEM; head = &to_free->epitems; } spin_lock(&file->f_lock); if (!file->f_ep) { if (unlikely(!head)) { spin_unlock(&file->f_lock); goto allocate; } file->f_ep = head; to_free = NULL; } hlist_add_head_rcu(&epi->fllink, file->f_ep); spin_unlock(&file->f_lock); free_ephead(to_free); return 0; } |
c7ea76302
|
1364 1365 1366 |
/* * Must be called with "mtx" held. */ |
bec1a502d
|
1367 |
static int ep_insert(struct eventpoll *ep, const struct epoll_event *event, |
67347fe4e
|
1368 |
struct file *tfile, int fd, int full_check) |
1da177e4c
|
1369 |
{ |
d85e2aa2e
|
1370 1371 |
int error, pwake = 0; __poll_t revents; |
1da177e4c
|
1372 1373 |
struct epitem *epi; struct ep_pqueue epq; |
85353e919
|
1374 1375 1376 1377 |
struct eventpoll *tep = NULL; if (is_file_epoll(tfile)) tep = tfile->private_data; |
1da177e4c
|
1378 |
|
92e641784
|
1379 |
lockdep_assert_irqs_enabled(); |
1e1c15839
|
1380 1381 |
if (unlikely(percpu_counter_compare(&ep->user->epoll_watches, max_user_watches) >= 0)) |
7ef9964e6
|
1382 |
return -ENOSPC; |
1e1c15839
|
1383 1384 1385 1386 |
percpu_counter_inc(&ep->user->epoll_watches); if (!(epi = kmem_cache_zalloc(epi_cache, GFP_KERNEL))) { percpu_counter_dec(&ep->user->epoll_watches); |
7ef9964e6
|
1387 |
return -ENOMEM; |
1e1c15839
|
1388 |
} |
1da177e4c
|
1389 1390 |
/* Item initialization follow here ... */ |
1da177e4c
|
1391 |
INIT_LIST_HEAD(&epi->rdllink); |
1da177e4c
|
1392 |
epi->ep = ep; |
b030a4dd6
|
1393 |
ep_set_ffd(&epi->ffd, tfile, fd); |
1da177e4c
|
1394 |
epi->event = *event; |
d47de16c7
|
1395 |
epi->next = EP_UNACTIVE_PTR; |
1da177e4c
|
1396 |
|
85353e919
|
1397 1398 |
if (tep) mutex_lock_nested(&tep->mtx, 1); |
f8d4f44df
|
1399 |
/* Add the current item to the list of active epoll hook for this file */ |
319c15174
|
1400 |
if (unlikely(attach_epitem(tfile, epi) < 0)) { |
319c15174
|
1401 1402 |
if (tep) mutex_unlock(&tep->mtx); |
1e1c15839
|
1403 1404 |
kmem_cache_free(epi_cache, epi); percpu_counter_dec(&ep->user->epoll_watches); |
319c15174
|
1405 |
return -ENOMEM; |
d9f41e3c9
|
1406 |
} |
f8d4f44df
|
1407 |
|
319c15174
|
1408 1409 |
if (full_check && !tep) list_file(tfile); |
f8d4f44df
|
1410 1411 1412 1413 1414 |
/* * 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); |
85353e919
|
1415 1416 |
if (tep) mutex_unlock(&tep->mtx); |
f8d4f44df
|
1417 1418 |
/* now check if we've created too many backpaths */ |
e3e096e7f
|
1419 1420 1421 1422 |
if (unlikely(full_check && reverse_path_check())) { ep_remove(ep, epi); return -EINVAL; } |
f8d4f44df
|
1423 |
|
d1ec50adb
|
1424 1425 1426 1427 1428 1429 1430 |
if (epi->event.events & EPOLLWAKEUP) { error = ep_create_wakeup_source(epi); if (error) { ep_remove(ep, epi); return error; } } |
f8d4f44df
|
1431 |
|
1da177e4c
|
1432 1433 1434 1435 1436 1437 1438 |
/* 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
|
1439 1440 1441 |
* been increased by the caller of this function. Note that after * this operation completes, the poll callback can start hitting * the new item. |
1da177e4c
|
1442 |
*/ |
37b5e5212
|
1443 |
revents = ep_item_poll(epi, &epq.pt, 1); |
1da177e4c
|
1444 1445 1446 1447 1448 1449 |
/* * 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. */ |
e3e096e7f
|
1450 1451 1452 1453 |
if (unlikely(!epq.epi)) { ep_remove(ep, epi); return -ENOMEM; } |
1da177e4c
|
1454 |
|
c7ea76302
|
1455 |
/* We have to drop the new item inside our item list to keep track of it */ |
a218cc491
|
1456 |
write_lock_irq(&ep->lock); |
c7ea76302
|
1457 |
|
bf3b9f637
|
1458 1459 |
/* record NAPI ID of new item if present */ ep_set_busy_poll_napi_id(epi); |
1da177e4c
|
1460 |
/* If the file is already "ready" we drop it inside the ready list */ |
992991c03
|
1461 |
if (revents && !ep_is_linked(epi)) { |
1da177e4c
|
1462 |
list_add_tail(&epi->rdllink, &ep->rdllist); |
eea1d5859
|
1463 |
ep_pm_stay_awake(epi); |
1da177e4c
|
1464 1465 1466 |
/* Notify waiting tasks that events are available */ if (waitqueue_active(&ep->wq)) |
a218cc491
|
1467 |
wake_up(&ep->wq); |
1da177e4c
|
1468 1469 1470 |
if (waitqueue_active(&ep->poll_wait)) pwake++; } |
a218cc491
|
1471 |
write_unlock_irq(&ep->lock); |
1da177e4c
|
1472 1473 1474 |
/* We have to call this outside the lock */ if (pwake) |
efcdd350d
|
1475 |
ep_poll_safewake(ep, NULL); |
1da177e4c
|
1476 |
|
1da177e4c
|
1477 |
return 0; |
1da177e4c
|
1478 |
} |
1da177e4c
|
1479 1480 |
/* * Modify the interest event mask by dropping an event if the new mask |
c7ea76302
|
1481 |
* has a match in the current file status. Must be called with "mtx" held. |
1da177e4c
|
1482 |
*/ |
bec1a502d
|
1483 1484 |
static int ep_modify(struct eventpoll *ep, struct epitem *epi, const struct epoll_event *event) |
1da177e4c
|
1485 1486 |
{ int pwake = 0; |
626cf2366
|
1487 |
poll_table pt; |
92e641784
|
1488 |
lockdep_assert_irqs_enabled(); |
626cf2366
|
1489 |
init_poll_funcptr(&pt, NULL); |
1da177e4c
|
1490 1491 |
/* |
e057e15ff
|
1492 1493 1494 |
* 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
|
1495 |
*/ |
128dd1759
|
1496 |
epi->event.events = event->events; /* need barrier below */ |
e057e15ff
|
1497 |
epi->event.data = event->data; /* protected by mtx */ |
4d7e30d98
|
1498 |
if (epi->event.events & EPOLLWAKEUP) { |
eea1d5859
|
1499 |
if (!ep_has_wakeup_source(epi)) |
4d7e30d98
|
1500 |
ep_create_wakeup_source(epi); |
eea1d5859
|
1501 |
} else if (ep_has_wakeup_source(epi)) { |
4d7e30d98
|
1502 1503 |
ep_destroy_wakeup_source(epi); } |
1da177e4c
|
1504 1505 |
/* |
128dd1759
|
1506 1507 1508 1509 1510 |
* 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
|
1511 |
* We need this because we did not take ep->lock while |
128dd1759
|
1512 |
* changing epi above (but ep_poll_callback does take |
a218cc491
|
1513 |
* ep->lock). |
128dd1759
|
1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 |
* * 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
|
1526 1527 |
* 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
|
1528 |
* If the item is "hot" and it is not registered inside the ready |
67647d0fb
|
1529 |
* list, push it inside. |
1da177e4c
|
1530 |
*/ |
69112736e
|
1531 |
if (ep_item_poll(epi, &pt, 1)) { |
a218cc491
|
1532 |
write_lock_irq(&ep->lock); |
992991c03
|
1533 |
if (!ep_is_linked(epi)) { |
c7ea76302
|
1534 |
list_add_tail(&epi->rdllink, &ep->rdllist); |
eea1d5859
|
1535 |
ep_pm_stay_awake(epi); |
c7ea76302
|
1536 1537 1538 |
/* Notify waiting tasks that events are available */ if (waitqueue_active(&ep->wq)) |
a218cc491
|
1539 |
wake_up(&ep->wq); |
c7ea76302
|
1540 1541 |
if (waitqueue_active(&ep->poll_wait)) pwake++; |
7699acd13
|
1542 |
} |
a218cc491
|
1543 |
write_unlock_irq(&ep->lock); |
7699acd13
|
1544 |
} |
1da177e4c
|
1545 |
|
7699acd13
|
1546 1547 |
/* We have to call this outside the lock */ if (pwake) |
efcdd350d
|
1548 |
ep_poll_safewake(ep, NULL); |
1da177e4c
|
1549 |
|
7699acd13
|
1550 |
return 0; |
1da177e4c
|
1551 |
} |
ff07952ae
|
1552 1553 |
static int ep_send_events(struct eventpoll *ep, struct epoll_event __user *events, int maxevents) |
1da177e4c
|
1554 |
{ |
4e0982a00
|
1555 |
struct epitem *epi, *tmp; |
ff07952ae
|
1556 |
LIST_HEAD(txlist); |
626cf2366
|
1557 |
poll_table pt; |
ff07952ae
|
1558 |
int res = 0; |
626cf2366
|
1559 |
|
cccd29bf0
|
1560 1561 1562 1563 1564 1565 1566 |
/* * 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. */ if (fatal_signal_pending(current)) return -EINTR; |
626cf2366
|
1567 |
init_poll_funcptr(&pt, NULL); |
ff07952ae
|
1568 |
|
57804b1cc
|
1569 1570 |
mutex_lock(&ep->mtx); ep_start_scan(ep, &txlist); |
1da177e4c
|
1571 |
|
296e236e9
|
1572 |
/* |
5071f97ec
|
1573 |
* We can loop without lock because we are passed a task private list. |
57804b1cc
|
1574 |
* Items cannot vanish during the loop we are holding ep->mtx. |
296e236e9
|
1575 |
*/ |
ff07952ae
|
1576 1577 1578 |
list_for_each_entry_safe(epi, tmp, &txlist, rdllink) { struct wakeup_source *ws; __poll_t revents; |
21877e1a5
|
1579 |
|
ff07952ae
|
1580 |
if (res >= maxevents) |
4e0982a00
|
1581 |
break; |
d47de16c7
|
1582 |
|
4d7e30d98
|
1583 1584 1585 1586 1587 1588 1589 1590 1591 |
/* * 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
|
1592 1593 1594 1595 1596 1597 |
ws = ep_wakeup_source(epi); if (ws) { if (ws->active) __pm_stay_awake(ep->ws); __pm_relax(ws); } |
d47de16c7
|
1598 |
list_del_init(&epi->rdllink); |
1da177e4c
|
1599 |
|
296e236e9
|
1600 |
/* |
5071f97ec
|
1601 |
* If the event mask intersect the caller-requested one, |
57804b1cc
|
1602 1603 |
* deliver the event to userspace. Again, we are holding ep->mtx, * so no operations coming from userspace can change the item. |
296e236e9
|
1604 |
*/ |
4e0982a00
|
1605 1606 1607 |
revents = ep_item_poll(epi, &pt, 1); if (!revents) continue; |
249dbe74d
|
1608 1609 |
events = epoll_put_uevent(revents, epi->event.data, events); if (!events) { |
ff07952ae
|
1610 |
list_add(&epi->rdllink, &txlist); |
4e0982a00
|
1611 |
ep_pm_stay_awake(epi); |
ff07952ae
|
1612 1613 1614 |
if (!res) res = -EFAULT; break; |
4e0982a00
|
1615 |
} |
ff07952ae
|
1616 |
res++; |
4e0982a00
|
1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 |
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
|
1633 1634 |
} } |
57804b1cc
|
1635 1636 |
ep_done_scan(ep, &txlist); mutex_unlock(&ep->mtx); |
5071f97ec
|
1637 |
|
ff07952ae
|
1638 |
return res; |
1da177e4c
|
1639 |
} |
7cdf7c20e
|
1640 |
static struct timespec64 *ep_timeout_to_timespec(struct timespec64 *to, long ms) |
0781b909b
|
1641 |
{ |
7cdf7c20e
|
1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 |
struct timespec64 now; if (ms < 0) return NULL; if (!ms) { to->tv_sec = 0; to->tv_nsec = 0; return to; } to->tv_sec = ms / MSEC_PER_SEC; to->tv_nsec = NSEC_PER_MSEC * (ms % MSEC_PER_SEC); |
0781b909b
|
1655 |
|
766b9f928
|
1656 |
ktime_get_ts64(&now); |
7cdf7c20e
|
1657 1658 |
*to = timespec64_add_safe(now, *to); return to; |
0781b909b
|
1659 |
} |
9ccb5d391
|
1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 |
/* * autoremove_wake_function, but remove even on failure to wake up, because we * know that default_wake_function/ttwu will only fail if the thread is already * woken, and in that case the ep_poll loop will remove the entry anyways, not * try to reuse it. */ static int ep_autoremove_wake_function(struct wait_queue_entry *wq_entry, unsigned int mode, int sync, void *key) { int ret = default_wake_function(wq_entry, mode, sync, key); list_del_init(&wq_entry->entry); return ret; } |
f4d93ad74
|
1674 |
/** |
a6c67fee9
|
1675 |
* ep_poll - Retrieves ready events, and delivers them to the caller-supplied |
f4d93ad74
|
1676 1677 1678 1679 1680 1681 1682 |
* 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 |
7cdf7c20e
|
1683 1684 |
* timespec. If the timeout is zero, the function will not block, * while if the @timeout ptr is NULL, the function will block |
f4d93ad74
|
1685 1686 1687 |
* until at least one event has been retrieved (or an error * occurred). * |
a6c67fee9
|
1688 |
* Return: the number of ready events which have been fetched, or an |
f4d93ad74
|
1689 1690 |
* error code, in case of error. */ |
1da177e4c
|
1691 |
static int ep_poll(struct eventpoll *ep, struct epoll_event __user *events, |
7cdf7c20e
|
1692 |
int maxevents, struct timespec64 *timeout) |
1da177e4c
|
1693 |
{ |
e59d3c64c
|
1694 |
int res, eavail, timed_out = 0; |
da8b44d5a
|
1695 |
u64 slack = 0; |
ac6424b98
|
1696 |
wait_queue_entry_t wait; |
95aac7b1c
|
1697 |
ktime_t expires, *to = NULL; |
679abf381
|
1698 |
lockdep_assert_irqs_enabled(); |
7cdf7c20e
|
1699 1700 |
if (timeout && (timeout->tv_sec | timeout->tv_nsec)) { slack = select_estimate_accuracy(timeout); |
95aac7b1c
|
1701 |
to = &expires; |
7cdf7c20e
|
1702 1703 |
*to = timespec64_to_ktime(*timeout); } else if (timeout) { |
f4d93ad74
|
1704 1705 |
/* * Avoid the unnecessary trip to the wait queue loop, if the |
e59d3c64c
|
1706 |
* caller specified a non blocking operation. |
f4d93ad74
|
1707 |
*/ |
95aac7b1c
|
1708 1709 |
timed_out = 1; } |
1da177e4c
|
1710 |
|
e59d3c64c
|
1711 1712 |
/* * This call is racy: We may or may not see events that are being added |
a6c67fee9
|
1713 |
* to the ready list under the lock (e.g., in IRQ callbacks). For cases |
e59d3c64c
|
1714 |
* with a non-zero timeout, this thread will check the ready list under |
a6c67fee9
|
1715 |
* lock and will add to the wait queue. For cases with a zero |
e59d3c64c
|
1716 1717 1718 1719 |
* timeout, the user by definition should not care and will have to * recheck again. */ eavail = ep_events_available(ep); |
00b27634b
|
1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 |
while (1) { if (eavail) { /* * 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. */ res = ep_send_events(ep, events, maxevents); if (res) return res; } if (timed_out) return 0; |
00b27634b
|
1734 |
eavail = ep_busy_loop(ep, timed_out); |
e8c85328b
|
1735 |
if (eavail) |
00b27634b
|
1736 |
continue; |
1da177e4c
|
1737 |
|
2efdaf766
|
1738 1739 |
if (signal_pending(current)) return -EINTR; |
412895f03
|
1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 |
/* * 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. |
9ccb5d391
|
1750 1751 1752 1753 1754 1755 |
* * In fact, we now use an even more aggressive function that * unconditionally removes, because we don't reuse the wait * entry between loop iterations. This lets us also avoid the * performance issue if a process is killed, causing all of its * threads to wake up without being removed normally. |
412895f03
|
1756 1757 |
*/ init_wait(&wait); |
9ccb5d391
|
1758 |
wait.func = ep_autoremove_wake_function; |
1da177e4c
|
1759 |
|
65759097d
|
1760 |
write_lock_irq(&ep->lock); |
1da177e4c
|
1761 |
/* |
65759097d
|
1762 1763 1764 |
* 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
|
1765 |
*/ |
65759097d
|
1766 |
__set_current_state(TASK_INTERRUPTIBLE); |
bf3b9f637
|
1767 |
/* |
65759097d
|
1768 1769 1770 1771 1772 |
* 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
|
1773 |
*/ |
65759097d
|
1774 |
eavail = ep_events_available(ep); |
2efdaf766
|
1775 1776 |
if (!eavail) __add_wait_queue_exclusive(&ep->wq, &wait); |
65759097d
|
1777 |
write_unlock_irq(&ep->lock); |
95aac7b1c
|
1778 |
|
2efdaf766
|
1779 |
if (!eavail) |
289caf5d8
|
1780 1781 |
timed_out = !schedule_hrtimeout_range(to, slack, HRTIMER_MODE_ABS); |
e411596d4
|
1782 |
__set_current_state(TASK_RUNNING); |
1da177e4c
|
1783 |
|
289caf5d8
|
1784 1785 1786 1787 1788 |
/* * 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
|
1789 |
eavail = 1; |
1da177e4c
|
1790 |
|
e8c85328b
|
1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 |
if (!list_empty_careful(&wait.entry)) { write_lock_irq(&ep->lock); /* * 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); __remove_wait_queue(&ep->wq, &wait); write_unlock_irq(&ep->lock); } |
00b27634b
|
1805 |
} |
1da177e4c
|
1806 |
} |
22bacca48
|
1807 |
/** |
773318edd
|
1808 |
* ep_loop_check_proc - verify that adding an epoll file inside another |
a6c67fee9
|
1809 |
* epoll structure does not violate the constraints, in |
22bacca48
|
1810 1811 1812 |
* terms of closed loops, or too deep chains (which can * result in excessive stack usage). * |
a6c67fee9
|
1813 |
* @ep: the &struct eventpoll to be currently checked. |
bde03c4c1
|
1814 |
* @depth: Current depth of the path being checked. |
22bacca48
|
1815 |
* |
a6c67fee9
|
1816 1817 |
* Return: %zero if adding the epoll @file inside current epoll * structure @ep does not violate the constraints, or %-1 otherwise. |
22bacca48
|
1818 |
*/ |
bde03c4c1
|
1819 |
static int ep_loop_check_proc(struct eventpoll *ep, int depth) |
22bacca48
|
1820 1821 |
{ int error = 0; |
22bacca48
|
1822 1823 |
struct rb_node *rbp; struct epitem *epi; |
773318edd
|
1824 |
mutex_lock_nested(&ep->mtx, depth + 1); |
18306c404
|
1825 |
ep->gen = loop_check_gen; |
b2ac2ea62
|
1826 |
for (rbp = rb_first_cached(&ep->rbr); rbp; rbp = rb_next(rbp)) { |
22bacca48
|
1827 1828 |
epi = rb_entry(rbp, struct epitem, rbn); if (unlikely(is_file_epoll(epi->ffd.file))) { |
bde03c4c1
|
1829 |
struct eventpoll *ep_tovisit; |
28d82dc1c
|
1830 |
ep_tovisit = epi->ffd.file->private_data; |
18306c404
|
1831 |
if (ep_tovisit->gen == loop_check_gen) |
28d82dc1c
|
1832 |
continue; |
bde03c4c1
|
1833 |
if (ep_tovisit == inserting_into || depth > EP_MAX_NESTS) |
56c428cac
|
1834 |
error = -1; |
bde03c4c1
|
1835 1836 |
else error = ep_loop_check_proc(ep_tovisit, depth + 1); |
22bacca48
|
1837 1838 |
if (error != 0) break; |
28d82dc1c
|
1839 1840 1841 1842 1843 1844 1845 1846 1847 |
} 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(). */ |
319c15174
|
1848 |
list_file(epi->ffd.file); |
22bacca48
|
1849 1850 1851 1852 1853 1854 1855 1856 |
} } mutex_unlock(&ep->mtx); return error; } /** |
bde03c4c1
|
1857 |
* ep_loop_check - Performs a check to verify that adding an epoll file (@to) |
a6c67fee9
|
1858 |
* into another epoll file (represented by @ep) does not create |
22bacca48
|
1859 1860 |
* closed loops or too deep chains. * |
a6c67fee9
|
1861 |
* @ep: Pointer to the epoll we are inserting into. |
bde03c4c1
|
1862 |
* @to: Pointer to the epoll to be inserted. |
22bacca48
|
1863 |
* |
a6c67fee9
|
1864 1865 |
* Return: %zero if adding the epoll @to inside the epoll @from * does not violate the constraints, or %-1 otherwise. |
22bacca48
|
1866 |
*/ |
bde03c4c1
|
1867 |
static int ep_loop_check(struct eventpoll *ep, struct eventpoll *to) |
22bacca48
|
1868 |
{ |
6a3890c47
|
1869 |
inserting_into = ep; |
bde03c4c1
|
1870 |
return ep_loop_check_proc(to, 0); |
28d82dc1c
|
1871 1872 1873 1874 |
} static void clear_tfile_check_list(void) { |
319c15174
|
1875 1876 1877 1878 1879 |
rcu_read_lock(); while (tfile_check_list != EP_UNACTIVE_PTR) { struct epitems_head *head = tfile_check_list; tfile_check_list = head->next; unlist_file(head); |
28d82dc1c
|
1880 |
} |
319c15174
|
1881 |
rcu_read_unlock(); |
22bacca48
|
1882 |
} |
7699acd13
|
1883 |
/* |
523723bb5
|
1884 |
* Open an eventpoll file descriptor. |
7699acd13
|
1885 |
*/ |
791eb22ee
|
1886 |
static int do_epoll_create(int flags) |
7699acd13
|
1887 |
{ |
28d82dc1c
|
1888 |
int error, fd; |
bb57c3edc
|
1889 |
struct eventpoll *ep = NULL; |
28d82dc1c
|
1890 |
struct file *file; |
7699acd13
|
1891 |
|
e38b36f32
|
1892 1893 |
/* Check the EPOLL_* constant for consistency. */ BUILD_BUG_ON(EPOLL_CLOEXEC != O_CLOEXEC); |
296e236e9
|
1894 1895 |
if (flags & ~EPOLL_CLOEXEC) return -EINVAL; |
7699acd13
|
1896 |
/* |
bb57c3edc
|
1897 |
* Create the internal data structure ("struct eventpoll"). |
7699acd13
|
1898 |
*/ |
9fe5ad9c8
|
1899 |
error = ep_alloc(&ep); |
bb57c3edc
|
1900 1901 |
if (error < 0) return error; |
7699acd13
|
1902 1903 |
/* * Creates all the items needed to setup an eventpoll file. That is, |
2030a42ce
|
1904 |
* a file structure and a free file descriptor. |
7699acd13
|
1905 |
*/ |
28d82dc1c
|
1906 1907 1908 1909 1910 1911 |
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
|
1912 |
O_RDWR | (flags & O_CLOEXEC)); |
28d82dc1c
|
1913 1914 1915 1916 |
if (IS_ERR(file)) { error = PTR_ERR(file); goto out_free_fd; } |
28d82dc1c
|
1917 |
ep->file = file; |
98022748f
|
1918 |
fd_install(fd, file); |
28d82dc1c
|
1919 1920 1921 1922 1923 1924 |
return fd; out_free_fd: put_unused_fd(fd); out_free_ep: ep_free(ep); |
bb57c3edc
|
1925 |
return error; |
7699acd13
|
1926 |
} |
791eb22ee
|
1927 1928 1929 1930 |
SYSCALL_DEFINE1(epoll_create1, int, flags) { return do_epoll_create(flags); } |
5a8a82b1d
|
1931 |
SYSCALL_DEFINE1(epoll_create, int, size) |
a0998b50c
|
1932 |
{ |
bfe3891a5
|
1933 |
if (size <= 0) |
9fe5ad9c8
|
1934 |
return -EINVAL; |
791eb22ee
|
1935 |
return do_epoll_create(0); |
a0998b50c
|
1936 |
} |
39220e8d4
|
1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 |
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
|
1951 1952 |
{ int error; |
67347fe4e
|
1953 |
int full_check = 0; |
7e3fb5842
|
1954 |
struct fd f, tf; |
7699acd13
|
1955 1956 |
struct eventpoll *ep; struct epitem *epi; |
67347fe4e
|
1957 |
struct eventpoll *tep = NULL; |
7699acd13
|
1958 |
|
7699acd13
|
1959 |
error = -EBADF; |
7e3fb5842
|
1960 1961 |
f = fdget(epfd); if (!f.file) |
7699acd13
|
1962 1963 1964 |
goto error_return; /* Get the "struct file *" for the target file */ |
7e3fb5842
|
1965 1966 |
tf = fdget(fd); if (!tf.file) |
7699acd13
|
1967 1968 1969 1970 |
goto error_fput; /* The target file descriptor must support poll */ error = -EPERM; |
9965ed174
|
1971 |
if (!file_can_poll(tf.file)) |
7699acd13
|
1972 |
goto error_tgt_fput; |
4d7e30d98
|
1973 |
/* Check if EPOLLWAKEUP is allowed */ |
c680e41b3
|
1974 |
if (ep_op_has_event(op)) |
58e41a44c
|
1975 |
ep_take_care_of_epollwakeup(epds); |
4d7e30d98
|
1976 |
|
7699acd13
|
1977 1978 1979 1980 1981 1982 |
/* * 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
|
1983 |
if (f.file == tf.file || !is_file_epoll(f.file)) |
7699acd13
|
1984 1985 1986 |
goto error_tgt_fput; /* |
df0108c5d
|
1987 1988 1989 1990 |
* 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
|
1991 |
if (ep_op_has_event(op) && (epds->events & EPOLLEXCLUSIVE)) { |
b6a515c8a
|
1992 1993 1994 |
if (op == EPOLL_CTL_MOD) goto error_tgt_fput; if (op == EPOLL_CTL_ADD && (is_file_epoll(tf.file) || |
58e41a44c
|
1995 |
(epds->events & ~EPOLLEXCLUSIVE_OK_BITS))) |
b6a515c8a
|
1996 1997 |
goto error_tgt_fput; } |
df0108c5d
|
1998 1999 |
/* |
7699acd13
|
2000 2001 2002 |
* At this point it is safe to assume that the "private_data" contains * our own data structure. */ |
7e3fb5842
|
2003 |
ep = f.file->private_data; |
7699acd13
|
2004 |
|
22bacca48
|
2005 |
/* |
a6c67fee9
|
2006 2007 |
* When we insert an epoll file descriptor inside another epoll file * descriptor, there is the chance of creating closed loops, which are |
28d82dc1c
|
2008 2009 2010 2011 |
* 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
|
2012 |
* |
67347fe4e
|
2013 2014 2015 2016 2017 2018 |
* 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
|
2019 |
*/ |
39220e8d4
|
2020 2021 2022 |
error = epoll_mutex_lock(&ep->mtx, 0, nonblock); if (error) goto error_tgt_fput; |
28d82dc1c
|
2023 |
if (op == EPOLL_CTL_ADD) { |
319c15174
|
2024 2025 |
if (READ_ONCE(f.file->f_ep) || ep->gen == loop_check_gen || is_file_epoll(tf.file)) { |
67347fe4e
|
2026 |
mutex_unlock(&ep->mtx); |
39220e8d4
|
2027 2028 2029 |
error = epoll_mutex_lock(&epmutex, 0, nonblock); if (error) goto error_tgt_fput; |
18306c404
|
2030 |
loop_check_gen++; |
39220e8d4
|
2031 |
full_check = 1; |
67347fe4e
|
2032 |
if (is_file_epoll(tf.file)) { |
bde03c4c1
|
2033 |
tep = tf.file->private_data; |
67347fe4e
|
2034 |
error = -ELOOP; |
bde03c4c1
|
2035 |
if (ep_loop_check(ep, tep) != 0) |
67347fe4e
|
2036 |
goto error_tgt_fput; |
a9ed4a656
|
2037 |
} |
39220e8d4
|
2038 |
error = epoll_mutex_lock(&ep->mtx, 0, nonblock); |
52c479697
|
2039 |
if (error) |
39220e8d4
|
2040 |
goto error_tgt_fput; |
67347fe4e
|
2041 2042 |
} } |
7699acd13
|
2043 |
|
67647d0fb
|
2044 |
/* |
a6c67fee9
|
2045 |
* Try to lookup the file inside our RB tree. Since we grabbed "mtx" |
67647d0fb
|
2046 2047 2048 |
* above, we can be sure to be able to use the item looked up by * ep_find() till we release the mutex. */ |
7e3fb5842
|
2049 |
epi = ep_find(ep, tf.file, fd); |
7699acd13
|
2050 2051 2052 2053 2054 |
error = -EINVAL; switch (op) { case EPOLL_CTL_ADD: if (!epi) { |
58e41a44c
|
2055 2056 |
epds->events |= EPOLLERR | EPOLLHUP; error = ep_insert(ep, epds, tf.file, fd, full_check); |
7699acd13
|
2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 |
} 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
|
2068 |
if (!(epi->event.events & EPOLLEXCLUSIVE)) { |
58e41a44c
|
2069 2070 |
epds->events |= EPOLLERR | EPOLLHUP; error = ep_modify(ep, epi, epds); |
b6a515c8a
|
2071 |
} |
7699acd13
|
2072 2073 2074 2075 |
} else error = -ENOENT; break; } |
d47de16c7
|
2076 |
mutex_unlock(&ep->mtx); |
7699acd13
|
2077 2078 |
error_tgt_fput: |
52c479697
|
2079 2080 |
if (full_check) { clear_tfile_check_list(); |
18306c404
|
2081 |
loop_check_gen++; |
22bacca48
|
2082 |
mutex_unlock(&epmutex); |
52c479697
|
2083 |
} |
22bacca48
|
2084 |
|
7e3fb5842
|
2085 |
fdput(tf); |
7699acd13
|
2086 |
error_fput: |
7e3fb5842
|
2087 |
fdput(f); |
7699acd13
|
2088 |
error_return: |
7699acd13
|
2089 2090 2091 2092 2093 |
return error; } /* |
58e41a44c
|
2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 |
* 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
|
2106 |
return do_epoll_ctl(epfd, op, fd, &epds, false); |
58e41a44c
|
2107 2108 2109 |
} /* |
7699acd13
|
2110 2111 2112 |
* Implement the event wait interface for the eventpoll file. It is the kernel * part of the user space epoll_wait(2). */ |
791eb22ee
|
2113 |
static int do_epoll_wait(int epfd, struct epoll_event __user *events, |
7cdf7c20e
|
2114 |
int maxevents, struct timespec64 *to) |
7699acd13
|
2115 |
{ |
2903ff019
|
2116 2117 |
int error; struct fd f; |
7699acd13
|
2118 |
struct eventpoll *ep; |
7699acd13
|
2119 2120 2121 2122 2123 |
/* 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
|
2124 |
if (!access_ok(events, maxevents * sizeof(struct epoll_event))) |
2903ff019
|
2125 |
return -EFAULT; |
7699acd13
|
2126 2127 |
/* Get the "struct file *" for the eventpoll file */ |
2903ff019
|
2128 2129 2130 |
f = fdget(epfd); if (!f.file) return -EBADF; |
7699acd13
|
2131 2132 2133 2134 2135 2136 |
/* * We have to check that the file structure underneath the fd * the user passed to us _is_ an eventpoll file. */ error = -EINVAL; |
2903ff019
|
2137 |
if (!is_file_epoll(f.file)) |
7699acd13
|
2138 2139 2140 2141 2142 2143 |
goto error_fput; /* * At this point it is safe to assume that the "private_data" contains * our own data structure. */ |
2903ff019
|
2144 |
ep = f.file->private_data; |
7699acd13
|
2145 2146 |
/* Time to fish for events ... */ |
7cdf7c20e
|
2147 |
error = ep_poll(ep, events, maxevents, to); |
7699acd13
|
2148 2149 |
error_fput: |
2903ff019
|
2150 |
fdput(f); |
7699acd13
|
2151 2152 |
return error; } |
791eb22ee
|
2153 2154 2155 |
SYSCALL_DEFINE4(epoll_wait, int, epfd, struct epoll_event __user *, events, int, maxevents, int, timeout) { |
7cdf7c20e
|
2156 2157 2158 2159 |
struct timespec64 to; return do_epoll_wait(epfd, events, maxevents, ep_timeout_to_timespec(&to, timeout)); |
791eb22ee
|
2160 |
} |
7699acd13
|
2161 2162 2163 2164 |
/* * Implement the event wait interface for the eventpoll file. It is the kernel * part of the user space epoll_pwait(2). */ |
58169a52e
|
2165 2166 2167 |
static int do_epoll_pwait(int epfd, struct epoll_event __user *events, int maxevents, struct timespec64 *to, const sigset_t __user *sigmask, size_t sigsetsize) |
7699acd13
|
2168 2169 |
{ int error; |
7699acd13
|
2170 2171 2172 2173 2174 |
/* * If the caller wants a certain signal mask to be set during the wait, * we apply it here. */ |
b772434be
|
2175 |
error = set_user_sigmask(sigmask, sigsetsize); |
ded653ccb
|
2176 2177 |
if (error) return error; |
7699acd13
|
2178 |
|
58169a52e
|
2179 |
error = do_epoll_wait(epfd, events, maxevents, to); |
7cdf7c20e
|
2180 |
|
b772434be
|
2181 |
restore_saved_sigmask_unless(error == -EINTR); |
7699acd13
|
2182 2183 2184 |
return error; } |
58169a52e
|
2185 2186 2187 |
SYSCALL_DEFINE6(epoll_pwait, int, epfd, struct epoll_event __user *, events, int, maxevents, int, timeout, const sigset_t __user *, sigmask, size_t, sigsetsize) |
35280bd4a
|
2188 |
{ |
7cdf7c20e
|
2189 |
struct timespec64 to; |
58169a52e
|
2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 |
return do_epoll_pwait(epfd, events, maxevents, ep_timeout_to_timespec(&to, timeout), sigmask, sigsetsize); } SYSCALL_DEFINE6(epoll_pwait2, int, epfd, struct epoll_event __user *, events, int, maxevents, const struct __kernel_timespec __user *, timeout, const sigset_t __user *, sigmask, size_t, sigsetsize) { struct timespec64 ts, *to = NULL; if (timeout) { if (get_timespec64(&ts, timeout)) return -EFAULT; to = &ts; if (poll_select_set_timeout(to, ts.tv_sec, ts.tv_nsec)) return -EINVAL; } return do_epoll_pwait(epfd, events, maxevents, to, sigmask, sigsetsize); } #ifdef CONFIG_COMPAT static int do_compat_epoll_pwait(int epfd, struct epoll_event __user *events, int maxevents, struct timespec64 *timeout, const compat_sigset_t __user *sigmask, compat_size_t sigsetsize) { |
35280bd4a
|
2220 |
long err; |
35280bd4a
|
2221 2222 2223 2224 2225 |
/* * If the caller wants a certain signal mask to be set during the wait, * we apply it here. */ |
b772434be
|
2226 |
err = set_compat_user_sigmask(sigmask, sigsetsize); |
ded653ccb
|
2227 2228 |
if (err) return err; |
35280bd4a
|
2229 |
|
58169a52e
|
2230 |
err = do_epoll_wait(epfd, events, maxevents, timeout); |
7cdf7c20e
|
2231 |
|
b772434be
|
2232 |
restore_saved_sigmask_unless(err == -EINTR); |
35280bd4a
|
2233 2234 2235 |
return err; } |
58169a52e
|
2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 |
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) { struct timespec64 to; return do_compat_epoll_pwait(epfd, events, maxevents, ep_timeout_to_timespec(&to, timeout), sigmask, sigsetsize); } COMPAT_SYSCALL_DEFINE6(epoll_pwait2, int, epfd, struct epoll_event __user *, events, int, maxevents, const struct __kernel_timespec __user *, timeout, const compat_sigset_t __user *, sigmask, compat_size_t, sigsetsize) { struct timespec64 ts, *to = NULL; if (timeout) { if (get_timespec64(&ts, timeout)) return -EFAULT; to = &ts; if (poll_select_set_timeout(to, ts.tv_sec, ts.tv_nsec)) return -EINVAL; } return do_compat_epoll_pwait(epfd, events, maxevents, to, sigmask, sigsetsize); } |
35280bd4a
|
2270 |
#endif |
1da177e4c
|
2271 2272 |
static int __init eventpoll_init(void) { |
7ef9964e6
|
2273 2274 2275 |
struct sysinfo si; si_meminfo(&si); |
9df04e1f2
|
2276 2277 2278 2279 |
/* * Allows top 4% of lomem to be allocated for epoll watches (per user). */ max_user_watches = (((si.totalram - si.totalhigh) / 25) << PAGE_SHIFT) / |
7ef9964e6
|
2280 |
EP_ITEM_COST; |
52bd19f76
|
2281 |
BUG_ON(max_user_watches < 0); |
1da177e4c
|
2282 |
|
22bacca48
|
2283 |
/* |
39732ca5a
|
2284 2285 2286 2287 |
* 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
|
2288 2289 |
/* Allocates slab cache used to allocate "struct epitem" items */ epi_cache = kmem_cache_create("eventpoll_epi", sizeof(struct epitem), |
2ae928a94
|
2290 |
0, SLAB_HWCACHE_ALIGN|SLAB_PANIC|SLAB_ACCOUNT, NULL); |
1da177e4c
|
2291 2292 2293 |
/* Allocates slab cache used to allocate "struct eppoll_entry" */ pwq_cache = kmem_cache_create("eventpoll_pwq", |
2ae928a94
|
2294 |
sizeof(struct eppoll_entry), 0, SLAB_PANIC|SLAB_ACCOUNT, NULL); |
1da177e4c
|
2295 |
|
319c15174
|
2296 2297 |
ephead_cache = kmem_cache_create("ep_head", sizeof(struct epitems_head), 0, SLAB_PANIC|SLAB_ACCOUNT, NULL); |
1da177e4c
|
2298 |
return 0; |
1da177e4c
|
2299 |
} |
cea692418
|
2300 |
fs_initcall(eventpoll_init); |