Blame view
fs/eventpoll.c
62.8 KB
1da177e4c Linux-2.6.12-rc2 |
1 |
/* |
5071f97ec epoll: fix epoll'... |
2 3 |
* fs/eventpoll.c (Efficient event retrieval implementation) * Copyright (C) 2001,...,2009 Davide Libenzi |
1da177e4c Linux-2.6.12-rc2 |
4 5 6 7 8 9 10 11 12 |
* * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation; either version 2 of the License, or * (at your option) any later version. * * Davide Libenzi <davidel@xmailserver.org> * */ |
1da177e4c Linux-2.6.12-rc2 |
13 14 |
#include <linux/init.h> #include <linux/kernel.h> |
174cd4b1e sched/headers: Pr... |
15 |
#include <linux/sched/signal.h> |
1da177e4c Linux-2.6.12-rc2 |
16 17 18 19 20 21 22 |
#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 Linux-2.6.12-rc2 |
23 24 25 26 27 |
#include <linux/string.h> #include <linux/list.h> #include <linux/hash.h> #include <linux/spinlock.h> #include <linux/syscalls.h> |
1da177e4c Linux-2.6.12-rc2 |
28 29 30 31 32 |
#include <linux/rbtree.h> #include <linux/wait.h> #include <linux/eventpoll.h> #include <linux/mount.h> #include <linux/bitops.h> |
144efe3e3 [PATCH] sem2mutex... |
33 |
#include <linux/mutex.h> |
da66f7cb0 epoll: use anonym... |
34 |
#include <linux/anon_inodes.h> |
4d7e30d98 epoll: Add a flag... |
35 |
#include <linux/device.h> |
7c0f6ba68 Replace <asm/uacc... |
36 |
#include <linux/uaccess.h> |
1da177e4c Linux-2.6.12-rc2 |
37 38 |
#include <asm/io.h> #include <asm/mman.h> |
60063497a atomic: use <linu... |
39 |
#include <linux/atomic.h> |
138d22b58 fs, epoll: add pr... |
40 41 |
#include <linux/proc_fs.h> #include <linux/seq_file.h> |
35280bd4a switch epoll_pwai... |
42 |
#include <linux/compat.h> |
ae10b2b4e epoll: optimize E... |
43 |
#include <linux/rculist.h> |
bf3b9f637 epoll: Add busy p... |
44 |
#include <net/busy_poll.h> |
1da177e4c Linux-2.6.12-rc2 |
45 |
|
1da177e4c Linux-2.6.12-rc2 |
46 47 48 49 |
/* * LOCKING: * There are three level of locking required by epoll : * |
144efe3e3 [PATCH] sem2mutex... |
50 |
* 1) epmutex (mutex) |
c7ea76302 epoll locks chang... |
51 |
* 2) ep->mtx (mutex) |
ee8ef0a4b epoll: use the wa... |
52 |
* 3) ep->wq.lock (spinlock) |
1da177e4c Linux-2.6.12-rc2 |
53 54 |
* * The acquire order is the one listed above, from 1 to 3. |
ee8ef0a4b epoll: use the wa... |
55 |
* We need a spinlock (ep->wq.lock) because we manipulate objects |
1da177e4c Linux-2.6.12-rc2 |
56 57 58 59 60 61 |
* 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 fix epoll single ... |
62 63 64 65 66 |
* 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 Linux-2.6.12-rc2 |
67 68 |
* 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 fs/eventpoll.c: f... |
69 |
* close()d without a previous call to epoll_ctl(EPOLL_CTL_DEL). |
22bacca48 epoll: prevent cr... |
70 71 72 73 74 75 76 |
* 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 epoll: fix spurio... |
77 78 79 80 81 82 83 84 85 |
* 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 fix epoll single ... |
86 |
* It is possible to drop the "ep->mtx" and to use the global |
ee8ef0a4b epoll: use the wa... |
87 |
* mutex "epmutex" (together with "ep->wq.lock") to have it working, |
d47de16c7 fix epoll single ... |
88 |
* but having "ep->mtx" will make the interface more scalable. |
144efe3e3 [PATCH] sem2mutex... |
89 |
* Events that require holding "epmutex" are very rare, while for |
d47de16c7 fix epoll single ... |
90 91 |
* normal operations the epoll private "ep->mtx" will guarantee * a better scalability. |
1da177e4c Linux-2.6.12-rc2 |
92 |
*/ |
1da177e4c Linux-2.6.12-rc2 |
93 |
/* Epoll private bits inside the event mask */ |
df0108c5d epoll: add EPOLLE... |
94 |
#define EP_PRIVATE_BITS (EPOLLWAKEUP | EPOLLONESHOT | EPOLLET | EPOLLEXCLUSIVE) |
1da177e4c Linux-2.6.12-rc2 |
95 |
|
a9a08845e vfs: do bulk POLL... |
96 |
#define EPOLLINOUT_BITS (EPOLLIN | EPOLLOUT) |
b6a515c8a epoll: restrict E... |
97 |
|
a9a08845e vfs: do bulk POLL... |
98 |
#define EPOLLEXCLUSIVE_OK_BITS (EPOLLINOUT_BITS | EPOLLERR | EPOLLHUP | \ |
b6a515c8a epoll: restrict E... |
99 |
EPOLLWAKEUP | EPOLLET | EPOLLEXCLUSIVE) |
5071f97ec epoll: fix epoll'... |
100 101 |
/* Maximum number of nesting allowed inside epoll sets */ #define EP_MAX_NESTS 4 |
1da177e4c Linux-2.6.12-rc2 |
102 |
|
b611967de [PATCH] epoll_pwa... |
103 |
#define EP_MAX_EVENTS (INT_MAX / sizeof(struct epoll_event)) |
d47de16c7 fix epoll single ... |
104 |
#define EP_UNACTIVE_PTR ((void *) -1L) |
7ef9964e6 epoll: introduce ... |
105 |
#define EP_ITEM_COST (sizeof(struct epitem) + sizeof(struct eppoll_entry)) |
1da177e4c Linux-2.6.12-rc2 |
106 107 108 |
struct epoll_filefd { struct file *file; int fd; |
39732ca5a epoll: trim epite... |
109 |
} __packed; |
1da177e4c Linux-2.6.12-rc2 |
110 111 |
/* |
5071f97ec epoll: fix epoll'... |
112 113 |
* Structure used to track possible nested calls, for too deep recursions * and loop cycles. |
1da177e4c Linux-2.6.12-rc2 |
114 |
*/ |
5071f97ec epoll: fix epoll'... |
115 |
struct nested_call_node { |
1da177e4c Linux-2.6.12-rc2 |
116 |
struct list_head llink; |
5071f97ec epoll: fix epoll'... |
117 |
void *cookie; |
3fe4a975d epoll: fix nested... |
118 |
void *ctx; |
1da177e4c Linux-2.6.12-rc2 |
119 120 121 |
}; /* |
5071f97ec epoll: fix epoll'... |
122 123 |
* This structure is used as collector for nested calls, to check for * maximum recursion dept and loop cycles. |
1da177e4c Linux-2.6.12-rc2 |
124 |
*/ |
5071f97ec epoll: fix epoll'... |
125 126 |
struct nested_calls { struct list_head tasks_call_list; |
1da177e4c Linux-2.6.12-rc2 |
127 128 129 130 |
spinlock_t lock; }; /* |
d47de16c7 fix epoll single ... |
131 132 |
* Each file descriptor added to the eventpoll interface will * have an entry of this type linked to the "rbr" RB tree. |
39732ca5a epoll: trim epite... |
133 134 |
* 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 fix epoll single ... |
135 136 |
*/ struct epitem { |
ae10b2b4e epoll: optimize E... |
137 138 139 140 141 142 |
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 fix epoll single ... |
143 144 145 |
/* List header used to link this structure to the eventpoll ready list */ struct list_head rdllink; |
c7ea76302 epoll locks chang... |
146 147 148 149 150 |
/* * Works together "struct eventpoll"->ovflist in keeping the * single linked chain of items. */ struct epitem *next; |
d47de16c7 fix epoll single ... |
151 152 153 154 155 156 157 158 159 160 161 |
/* 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 fix epoll single ... |
162 163 |
/* List header used to link this item to the "struct file" items list */ struct list_head fllink; |
4d7e30d98 epoll: Add a flag... |
164 |
/* wakeup_source used when EPOLLWAKEUP is set */ |
eea1d5859 epoll: use RCU to... |
165 |
struct wakeup_source __rcu *ws; |
4d7e30d98 epoll: Add a flag... |
166 |
|
c7ea76302 epoll locks chang... |
167 168 |
/* The structure that describe the interested events and the source fd */ struct epoll_event event; |
d47de16c7 fix epoll single ... |
169 170 171 |
}; /* |
1da177e4c Linux-2.6.12-rc2 |
172 |
* This structure is stored inside the "private_data" member of the file |
bf6a41db7 fs/eventpoll.c: f... |
173 |
* structure and represents the main data structure for the eventpoll |
1da177e4c Linux-2.6.12-rc2 |
174 |
* interface. |
ee8ef0a4b epoll: use the wa... |
175 176 |
* * Access to it is protected by the lock inside wq. |
1da177e4c Linux-2.6.12-rc2 |
177 178 |
*/ struct eventpoll { |
1da177e4c Linux-2.6.12-rc2 |
179 |
/* |
d47de16c7 fix epoll single ... |
180 181 182 183 |
* 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 Linux-2.6.12-rc2 |
184 |
*/ |
d47de16c7 fix epoll single ... |
185 |
struct mutex mtx; |
1da177e4c Linux-2.6.12-rc2 |
186 187 188 189 190 191 192 193 194 |
/* 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; |
67647d0fb epoll: fix some c... |
195 |
/* RB tree root used to store monitored fd structs */ |
b2ac2ea62 fs/epoll: use fas... |
196 |
struct rb_root_cached rbr; |
d47de16c7 fix epoll single ... |
197 198 199 |
/* * This is a single linked list that chains all the "struct epitem" that |
25985edce Fix common misspe... |
200 |
* happened while transferring ready events to userspace w/out |
ee8ef0a4b epoll: use the wa... |
201 |
* holding ->wq.lock. |
d47de16c7 fix epoll single ... |
202 203 |
*/ struct epitem *ovflist; |
7ef9964e6 epoll: introduce ... |
204 |
|
4d7e30d98 epoll: Add a flag... |
205 206 |
/* wakeup_source used when ep_scan_ready_list is running */ struct wakeup_source *ws; |
7ef9964e6 epoll: introduce ... |
207 208 |
/* The user that created the eventpoll descriptor */ struct user_struct *user; |
28d82dc1c epoll: limit paths |
209 210 211 212 213 214 |
struct file *file; /* used to optimize loop detection check */ int visited; struct list_head visited_list_link; |
bf3b9f637 epoll: Add busy p... |
215 216 217 218 219 |
#ifdef CONFIG_NET_RX_BUSY_POLL /* used to track busy poll napi_id */ unsigned int napi_id; #endif |
1da177e4c Linux-2.6.12-rc2 |
220 221 222 223 224 225 226 227 |
}; /* 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 epoll: use real t... |
228 |
struct epitem *base; |
1da177e4c Linux-2.6.12-rc2 |
229 230 231 232 233 |
/* * Wait queue item that will be linked to the target file wait * queue head. */ |
ac6424b98 sched/wait: Renam... |
234 |
wait_queue_entry_t wait; |
1da177e4c Linux-2.6.12-rc2 |
235 236 237 238 |
/* The wait queue head that linked the "wait" wait queue item */ wait_queue_head_t *whead; }; |
1da177e4c Linux-2.6.12-rc2 |
239 240 241 242 243 |
/* Wrapper struct used by poll queueing */ struct ep_pqueue { poll_table pt; struct epitem *epi; }; |
5071f97ec epoll: fix epoll'... |
244 245 246 247 |
/* Used by the ep_send_events() function as callback private data */ struct ep_send_events_data { int maxevents; struct epoll_event __user *events; |
d7ebbe46f ep_send_events_pr... |
248 |
int res; |
5071f97ec epoll: fix epoll'... |
249 |
}; |
1da177e4c Linux-2.6.12-rc2 |
250 |
/* |
7ef9964e6 epoll: introduce ... |
251 252 |
* Configuration options available inside /proc/sys/fs/epoll/ */ |
7ef9964e6 epoll: introduce ... |
253 |
/* Maximum number of epoll watched descriptors, per user */ |
52bd19f76 epoll: convert ma... |
254 |
static long max_user_watches __read_mostly; |
7ef9964e6 epoll: introduce ... |
255 256 |
/* |
d47de16c7 fix epoll single ... |
257 |
* This mutex is used to serialize ep_free() and eventpoll_release_file(). |
1da177e4c Linux-2.6.12-rc2 |
258 |
*/ |
7ef9964e6 epoll: introduce ... |
259 |
static DEFINE_MUTEX(epmutex); |
1da177e4c Linux-2.6.12-rc2 |
260 |
|
22bacca48 epoll: prevent cr... |
261 262 |
/* Used to check for epoll file descriptor inclusion loops */ static struct nested_calls poll_loop_ncalls; |
1da177e4c Linux-2.6.12-rc2 |
263 |
/* Slab cache used to allocate "struct epitem" */ |
e18b890bb [PATCH] slab: rem... |
264 |
static struct kmem_cache *epi_cache __read_mostly; |
1da177e4c Linux-2.6.12-rc2 |
265 266 |
/* Slab cache used to allocate "struct eppoll_entry" */ |
e18b890bb [PATCH] slab: rem... |
267 |
static struct kmem_cache *pwq_cache __read_mostly; |
1da177e4c Linux-2.6.12-rc2 |
268 |
|
28d82dc1c epoll: limit paths |
269 270 271 272 273 274 275 276 |
/* 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 epoll: introduce ... |
277 278 279 |
#ifdef CONFIG_SYSCTL #include <linux/sysctl.h> |
52bd19f76 epoll: convert ma... |
280 281 |
static long zero; static long long_max = LONG_MAX; |
7ef9964e6 epoll: introduce ... |
282 |
|
1f7e0616c fs: convert use o... |
283 |
struct ctl_table epoll_table[] = { |
7ef9964e6 epoll: introduce ... |
284 |
{ |
7ef9964e6 epoll: introduce ... |
285 286 |
.procname = "max_user_watches", .data = &max_user_watches, |
52bd19f76 epoll: convert ma... |
287 |
.maxlen = sizeof(max_user_watches), |
7ef9964e6 epoll: introduce ... |
288 |
.mode = 0644, |
52bd19f76 epoll: convert ma... |
289 |
.proc_handler = proc_doulongvec_minmax, |
7ef9964e6 epoll: introduce ... |
290 |
.extra1 = &zero, |
52bd19f76 epoll: convert ma... |
291 |
.extra2 = &long_max, |
7ef9964e6 epoll: introduce ... |
292 |
}, |
ab09203e3 sysctl fs: Remove... |
293 |
{ } |
7ef9964e6 epoll: introduce ... |
294 295 |
}; #endif /* CONFIG_SYSCTL */ |
28d82dc1c epoll: limit paths |
296 297 298 299 300 301 |
static const struct file_operations eventpoll_fops; static inline int is_file_epoll(struct file *f) { return f->f_op == &eventpoll_fops; } |
b030a4dd6 [PATCH] Remove ev... |
302 |
|
67647d0fb epoll: fix some c... |
303 |
/* Setup the structure that is used as key for the RB tree */ |
b030a4dd6 [PATCH] Remove ev... |
304 305 306 307 308 309 |
static inline void ep_set_ffd(struct epoll_filefd *ffd, struct file *file, int fd) { ffd->file = file; ffd->fd = fd; } |
67647d0fb epoll: fix some c... |
310 |
/* Compare RB tree keys */ |
b030a4dd6 [PATCH] Remove ev... |
311 312 313 314 315 316 |
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 [PATCH] Remove ev... |
317 |
/* Tells us if the item is currently linked */ |
992991c03 fs/eventpoll.c: s... |
318 |
static inline int ep_is_linked(struct epitem *epi) |
b030a4dd6 [PATCH] Remove ev... |
319 |
{ |
992991c03 fs/eventpoll.c: s... |
320 |
return !list_empty(&epi->rdllink); |
b030a4dd6 [PATCH] Remove ev... |
321 |
} |
ac6424b98 sched/wait: Renam... |
322 |
static inline struct eppoll_entry *ep_pwq_from_wait(wait_queue_entry_t *p) |
971316f05 epoll: ep_unregis... |
323 324 325 |
{ return container_of(p, struct eppoll_entry, wait); } |
b030a4dd6 [PATCH] Remove ev... |
326 |
/* Get the "struct epitem" from a wait queue pointer */ |
ac6424b98 sched/wait: Renam... |
327 |
static inline struct epitem *ep_item_from_wait(wait_queue_entry_t *p) |
b030a4dd6 [PATCH] Remove ev... |
328 329 330 331 332 |
{ return container_of(p, struct eppoll_entry, wait)->base; } /* Get the "struct epitem" from an epoll queue wrapper */ |
cdac75e6f epoll: avoid kmem... |
333 |
static inline struct epitem *ep_item_from_epqueue(poll_table *p) |
b030a4dd6 [PATCH] Remove ev... |
334 335 336 337 338 |
{ return container_of(p, struct ep_pqueue, pt)->epi; } /* Tells if the epoll_ctl(2) operation needs an event copy from userspace */ |
6192bd536 epoll: optimizati... |
339 |
static inline int ep_op_has_event(int op) |
b030a4dd6 [PATCH] Remove ev... |
340 |
{ |
a80a6b85b revert "epoll: su... |
341 |
return op != EPOLL_CTL_DEL; |
b030a4dd6 [PATCH] Remove ev... |
342 |
} |
1da177e4c Linux-2.6.12-rc2 |
343 |
/* Initialize the poll safe wake up structure */ |
5071f97ec epoll: fix epoll'... |
344 |
static void ep_nested_calls_init(struct nested_calls *ncalls) |
1da177e4c Linux-2.6.12-rc2 |
345 |
{ |
5071f97ec epoll: fix epoll'... |
346 347 |
INIT_LIST_HEAD(&ncalls->tasks_call_list); spin_lock_init(&ncalls->lock); |
1da177e4c Linux-2.6.12-rc2 |
348 |
} |
5071f97ec epoll: fix epoll'... |
349 |
/** |
3fb0e584a epoll: move ready... |
350 351 352 353 354 355 356 357 358 359 360 |
* 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) { return !list_empty(&ep->rdllist) || ep->ovflist != EP_UNACTIVE_PTR; } |
bf3b9f637 epoll: Add busy p... |
361 362 363 364 365 366 367 |
#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 epoll: Add busy p... |
368 369 370 371 372 373 374 375 376 |
/* * 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 epoll: Add busy p... |
377 378 379 380 |
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 epoll: Add busy p... |
381 382 383 384 |
} static inline void ep_reset_busy_poll_napi_id(struct eventpoll *ep) { |
bf3b9f637 epoll: Add busy p... |
385 386 |
if (ep->napi_id) ep->napi_id = 0; |
bf3b9f637 epoll: Add busy p... |
387 388 389 390 391 392 393 |
} /* * Set epoll busy poll NAPI ID from sk. */ static inline void ep_set_busy_poll_napi_id(struct epitem *epi) { |
bf3b9f637 epoll: Add busy p... |
394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 |
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 epoll: Add busy p... |
423 |
} |
514056d50 fs/eventpoll.c: s... |
424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 |
#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 epoll: move ready... |
439 |
/** |
5071f97ec epoll: fix epoll'... |
440 441 442 443 444 445 446 447 448 449 |
* 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. * @max_nests: Maximum number of allowed nesting calls. * @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 epoll: fix nested... |
450 |
* @ctx: This instance context. |
5071f97ec epoll: fix epoll'... |
451 452 453 |
* * Returns: Returns the code returned by the @nproc callback, or -1 if * the maximum recursion limit has been exceeded. |
1da177e4c Linux-2.6.12-rc2 |
454 |
*/ |
5071f97ec epoll: fix epoll'... |
455 456 |
static int ep_call_nested(struct nested_calls *ncalls, int max_nests, int (*nproc)(void *, void *, int), void *priv, |
3fe4a975d epoll: fix nested... |
457 |
void *cookie, void *ctx) |
1da177e4c Linux-2.6.12-rc2 |
458 |
{ |
5071f97ec epoll: fix epoll'... |
459 |
int error, call_nests = 0; |
1da177e4c Linux-2.6.12-rc2 |
460 |
unsigned long flags; |
5071f97ec epoll: fix epoll'... |
461 462 463 |
struct list_head *lsthead = &ncalls->tasks_call_list; struct nested_call_node *tncur; struct nested_call_node tnode; |
1da177e4c Linux-2.6.12-rc2 |
464 |
|
5071f97ec epoll: fix epoll'... |
465 |
spin_lock_irqsave(&ncalls->lock, flags); |
1da177e4c Linux-2.6.12-rc2 |
466 |
|
5071f97ec epoll: fix epoll'... |
467 468 469 470 471 |
/* * 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 fs/eventpoll.c: u... |
472 |
list_for_each_entry(tncur, lsthead, llink) { |
3fe4a975d epoll: fix nested... |
473 |
if (tncur->ctx == ctx && |
5071f97ec epoll: fix epoll'... |
474 |
(tncur->cookie == cookie || ++call_nests > max_nests)) { |
1da177e4c Linux-2.6.12-rc2 |
475 476 477 478 |
/* * Ops ... loop detected or maximum nest level reached. * We abort this wake by breaking the cycle itself. */ |
abff55cee epoll: don't use ... |
479 480 |
error = -1; goto out_unlock; |
1da177e4c Linux-2.6.12-rc2 |
481 482 |
} } |
5071f97ec epoll: fix epoll'... |
483 |
/* Add the current task and cookie to the list */ |
3fe4a975d epoll: fix nested... |
484 |
tnode.ctx = ctx; |
5071f97ec epoll: fix epoll'... |
485 |
tnode.cookie = cookie; |
1da177e4c Linux-2.6.12-rc2 |
486 |
list_add(&tnode.llink, lsthead); |
5071f97ec epoll: fix epoll'... |
487 |
spin_unlock_irqrestore(&ncalls->lock, flags); |
1da177e4c Linux-2.6.12-rc2 |
488 |
|
5071f97ec epoll: fix epoll'... |
489 490 |
/* Call the nested function */ error = (*nproc)(priv, cookie, call_nests); |
1da177e4c Linux-2.6.12-rc2 |
491 492 |
/* Remove the current task from the list */ |
5071f97ec epoll: fix epoll'... |
493 |
spin_lock_irqsave(&ncalls->lock, flags); |
1da177e4c Linux-2.6.12-rc2 |
494 |
list_del(&tnode.llink); |
3fe4a975d epoll: fix nested... |
495 |
out_unlock: |
5071f97ec epoll: fix epoll'... |
496 497 498 499 |
spin_unlock_irqrestore(&ncalls->lock, flags); return error; } |
02edc6fc4 epoll: comment th... |
500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 |
/* * 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 epoll keyed wakeu... |
525 |
#ifdef CONFIG_DEBUG_LOCK_ALLOC |
57a173bdf epoll: avoid call... |
526 527 528 529 |
static struct nested_calls poll_safewake_ncalls; static int ep_poll_wakeup_proc(void *priv, void *cookie, int call_nests) |
2dfa4eeab epoll keyed wakeu... |
530 531 |
{ unsigned long flags; |
57a173bdf epoll: avoid call... |
532 |
wait_queue_head_t *wqueue = (wait_queue_head_t *)cookie; |
2dfa4eeab epoll keyed wakeu... |
533 |
|
57a173bdf epoll: avoid call... |
534 |
spin_lock_irqsave_nested(&wqueue->lock, flags, call_nests + 1); |
a9a08845e vfs: do bulk POLL... |
535 |
wake_up_locked_poll(wqueue, EPOLLIN); |
2dfa4eeab epoll keyed wakeu... |
536 |
spin_unlock_irqrestore(&wqueue->lock, flags); |
2dfa4eeab epoll keyed wakeu... |
537 |
|
5071f97ec epoll: fix epoll'... |
538 539 |
return 0; } |
5071f97ec epoll: fix epoll'... |
540 541 |
static void ep_poll_safewake(wait_queue_head_t *wq) { |
3fe4a975d epoll: fix nested... |
542 |
int this_cpu = get_cpu(); |
5071f97ec epoll: fix epoll'... |
543 |
ep_call_nested(&poll_safewake_ncalls, EP_MAX_NESTS, |
3fe4a975d epoll: fix nested... |
544 545 546 |
ep_poll_wakeup_proc, NULL, wq, (void *) (long) this_cpu); put_cpu(); |
1da177e4c Linux-2.6.12-rc2 |
547 |
} |
57a173bdf epoll: avoid call... |
548 549 550 551 |
#else static void ep_poll_safewake(wait_queue_head_t *wq) { |
a9a08845e vfs: do bulk POLL... |
552 |
wake_up_poll(wq, EPOLLIN); |
57a173bdf epoll: avoid call... |
553 554 555 |
} #endif |
971316f05 epoll: ep_unregis... |
556 557 558 559 560 |
static void ep_remove_wait_queue(struct eppoll_entry *pwq) { wait_queue_head_t *whead; rcu_read_lock(); |
138e4ad67 epoll: fix race b... |
561 562 563 564 565 566 567 |
/* * 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 epoll: ep_unregis... |
568 569 570 571 |
if (whead) remove_wait_queue(whead, &pwq->wait); rcu_read_unlock(); } |
1da177e4c Linux-2.6.12-rc2 |
572 |
/* |
d1bc90dd5 epoll: remove unn... |
573 574 575 |
* This function unregisters poll callbacks from the associated file * descriptor. Must be called with "mtx" held (or "epmutex" if called from * ep_free). |
1da177e4c Linux-2.6.12-rc2 |
576 |
*/ |
7699acd13 epoll cleanups: e... |
577 |
static void ep_unregister_pollwait(struct eventpoll *ep, struct epitem *epi) |
1da177e4c Linux-2.6.12-rc2 |
578 |
{ |
7699acd13 epoll cleanups: e... |
579 580 |
struct list_head *lsthead = &epi->pwqlist; struct eppoll_entry *pwq; |
1da177e4c Linux-2.6.12-rc2 |
581 |
|
d1bc90dd5 epoll: remove unn... |
582 583 |
while (!list_empty(lsthead)) { pwq = list_first_entry(lsthead, struct eppoll_entry, llink); |
1da177e4c Linux-2.6.12-rc2 |
584 |
|
d1bc90dd5 epoll: remove unn... |
585 |
list_del(&pwq->llink); |
971316f05 epoll: ep_unregis... |
586 |
ep_remove_wait_queue(pwq); |
d1bc90dd5 epoll: remove unn... |
587 |
kmem_cache_free(pwq_cache, pwq); |
1da177e4c Linux-2.6.12-rc2 |
588 |
} |
1da177e4c Linux-2.6.12-rc2 |
589 |
} |
eea1d5859 epoll: use RCU to... |
590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 |
/* 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 epoll: fix epoll'... |
621 622 623 624 625 626 627 628 |
/** * 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 epoll: fix spurio... |
629 |
* @depth: The current depth of recursive f_op->poll calls. |
67347fe4e epoll: do not tak... |
630 |
* @ep_locked: caller already holds ep->mtx |
5071f97ec epoll: fix epoll'... |
631 632 633 |
* * Returns: The same integer error code returned by the @sproc callback. */ |
d85e2aa2e annotate ep_scan_... |
634 635 |
static __poll_t ep_scan_ready_list(struct eventpoll *ep, __poll_t (*sproc)(struct eventpoll *, |
5071f97ec epoll: fix epoll'... |
636 |
struct list_head *, void *), |
67347fe4e epoll: do not tak... |
637 |
void *priv, int depth, bool ep_locked) |
5071f97ec epoll: fix epoll'... |
638 |
{ |
d85e2aa2e annotate ep_scan_... |
639 640 |
__poll_t res; int pwake = 0; |
5071f97ec epoll: fix epoll'... |
641 |
struct epitem *epi, *nepi; |
296e236e9 epoll: fix epoll'... |
642 |
LIST_HEAD(txlist); |
5071f97ec epoll: fix epoll'... |
643 |
|
92e641784 s/epoll: robustif... |
644 |
lockdep_assert_irqs_enabled(); |
5071f97ec epoll: fix epoll'... |
645 646 |
/* * We need to lock this because we could be hit by |
e057e15ff epoll: clean up e... |
647 |
* eventpoll_release_file() and epoll_ctl(). |
5071f97ec epoll: fix epoll'... |
648 |
*/ |
67347fe4e epoll: do not tak... |
649 650 651 |
if (!ep_locked) mutex_lock_nested(&ep->mtx, depth); |
5071f97ec epoll: fix epoll'... |
652 653 654 655 656 657 658 659 660 |
/* * 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. */ |
002b34366 fs/epoll: loosen ... |
661 |
spin_lock_irq(&ep->wq.lock); |
296e236e9 epoll: fix epoll'... |
662 |
list_splice_init(&ep->rdllist, &txlist); |
5071f97ec epoll: fix epoll'... |
663 |
ep->ovflist = NULL; |
002b34366 fs/epoll: loosen ... |
664 |
spin_unlock_irq(&ep->wq.lock); |
5071f97ec epoll: fix epoll'... |
665 666 667 668 |
/* * Now call the callback function. */ |
d85e2aa2e annotate ep_scan_... |
669 |
res = (*sproc)(ep, &txlist, priv); |
5071f97ec epoll: fix epoll'... |
670 |
|
002b34366 fs/epoll: loosen ... |
671 |
spin_lock_irq(&ep->wq.lock); |
5071f97ec epoll: fix epoll'... |
672 673 674 675 676 677 678 679 680 681 682 683 684 |
/* * 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. */ for (nepi = ep->ovflist; (epi = nepi) != NULL; 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 fs/eventpoll.c: s... |
685 |
if (!ep_is_linked(epi)) { |
5071f97ec epoll: fix epoll'... |
686 |
list_add_tail(&epi->rdllink, &ep->rdllist); |
eea1d5859 epoll: use RCU to... |
687 |
ep_pm_stay_awake(epi); |
4d7e30d98 epoll: Add a flag... |
688 |
} |
5071f97ec epoll: fix epoll'... |
689 690 691 692 693 694 695 696 697 698 699 700 |
} /* * 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. */ ep->ovflist = EP_UNACTIVE_PTR; /* * Quickly re-inject items left on "txlist". */ list_splice(&txlist, &ep->rdllist); |
4d7e30d98 epoll: Add a flag... |
701 |
__pm_relax(ep->ws); |
5071f97ec epoll: fix epoll'... |
702 703 704 |
if (!list_empty(&ep->rdllist)) { /* |
296e236e9 epoll: fix epoll'... |
705 706 |
* Wake up (if active) both the eventpoll wait list and * the ->poll() wait list (delayed after we release the lock). |
5071f97ec epoll: fix epoll'... |
707 708 709 710 711 712 |
*/ if (waitqueue_active(&ep->wq)) wake_up_locked(&ep->wq); if (waitqueue_active(&ep->poll_wait)) pwake++; } |
002b34366 fs/epoll: loosen ... |
713 |
spin_unlock_irq(&ep->wq.lock); |
5071f97ec epoll: fix epoll'... |
714 |
|
67347fe4e epoll: do not tak... |
715 716 |
if (!ep_locked) mutex_unlock(&ep->mtx); |
5071f97ec epoll: fix epoll'... |
717 718 719 720 |
/* We have to call this outside the lock */ if (pwake) ep_poll_safewake(&ep->poll_wait); |
d85e2aa2e annotate ep_scan_... |
721 |
return res; |
5071f97ec epoll: fix epoll'... |
722 |
} |
ae10b2b4e epoll: optimize E... |
723 724 725 726 727 |
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 Linux-2.6.12-rc2 |
728 |
/* |
7699acd13 epoll cleanups: e... |
729 |
* Removes a "struct epitem" from the eventpoll RB tree and deallocates |
c7ea76302 epoll locks chang... |
730 |
* all the associated resources. Must be called with "mtx" held. |
7699acd13 epoll cleanups: e... |
731 732 733 |
*/ static int ep_remove(struct eventpoll *ep, struct epitem *epi) { |
7699acd13 epoll cleanups: e... |
734 |
struct file *file = epi->ffd.file; |
1da177e4c Linux-2.6.12-rc2 |
735 |
|
92e641784 s/epoll: robustif... |
736 |
lockdep_assert_irqs_enabled(); |
1da177e4c Linux-2.6.12-rc2 |
737 |
/* |
ee8ef0a4b epoll: use the wa... |
738 |
* Removes poll wait queue hooks. |
1da177e4c Linux-2.6.12-rc2 |
739 |
*/ |
7699acd13 epoll cleanups: e... |
740 |
ep_unregister_pollwait(ep, epi); |
1da177e4c Linux-2.6.12-rc2 |
741 |
|
7699acd13 epoll cleanups: e... |
742 |
/* Remove the current item from the list of epoll hooks */ |
684999149 Rename struct fil... |
743 |
spin_lock(&file->f_lock); |
ae10b2b4e epoll: optimize E... |
744 |
list_del_rcu(&epi->fllink); |
684999149 Rename struct fil... |
745 |
spin_unlock(&file->f_lock); |
1da177e4c Linux-2.6.12-rc2 |
746 |
|
b2ac2ea62 fs/epoll: use fas... |
747 |
rb_erase_cached(&epi->rbn, &ep->rbr); |
1da177e4c Linux-2.6.12-rc2 |
748 |
|
304b18b8d fs/epoll: loosen ... |
749 |
spin_lock_irq(&ep->wq.lock); |
992991c03 fs/eventpoll.c: s... |
750 |
if (ep_is_linked(epi)) |
c7ea76302 epoll locks chang... |
751 |
list_del_init(&epi->rdllink); |
304b18b8d fs/epoll: loosen ... |
752 |
spin_unlock_irq(&ep->wq.lock); |
1da177e4c Linux-2.6.12-rc2 |
753 |
|
eea1d5859 epoll: use RCU to... |
754 |
wakeup_source_unregister(ep_wakeup_source(epi)); |
ae10b2b4e epoll: optimize E... |
755 756 757 758 759 760 761 762 |
/* * 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 Linux-2.6.12-rc2 |
763 |
|
52bd19f76 epoll: convert ma... |
764 |
atomic_long_dec(&ep->user->epoll_watches); |
7ef9964e6 epoll: introduce ... |
765 |
|
c7ea76302 epoll locks chang... |
766 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
767 |
} |
7699acd13 epoll cleanups: e... |
768 |
static void ep_free(struct eventpoll *ep) |
1da177e4c Linux-2.6.12-rc2 |
769 |
{ |
7699acd13 epoll cleanups: e... |
770 771 |
struct rb_node *rbp; struct epitem *epi; |
1da177e4c Linux-2.6.12-rc2 |
772 |
|
7699acd13 epoll cleanups: e... |
773 774 |
/* We need to release all tasks waiting for these file */ if (waitqueue_active(&ep->poll_wait)) |
5071f97ec epoll: fix epoll'... |
775 |
ep_poll_safewake(&ep->poll_wait); |
1da177e4c Linux-2.6.12-rc2 |
776 |
|
7699acd13 epoll cleanups: e... |
777 778 779 |
/* * We need to lock this because we could be hit by * eventpoll_release_file() while we're freeing the "struct eventpoll". |
d47de16c7 fix epoll single ... |
780 |
* We do not need to hold "ep->mtx" here because the epoll file |
7699acd13 epoll cleanups: e... |
781 782 |
* 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 Fix common misspe... |
783 |
* holding "epmutex" is sufficient here. |
7699acd13 epoll cleanups: e... |
784 785 |
*/ mutex_lock(&epmutex); |
1da177e4c Linux-2.6.12-rc2 |
786 787 |
/* |
7699acd13 epoll cleanups: e... |
788 |
* Walks through the whole tree by unregistering poll callbacks. |
1da177e4c Linux-2.6.12-rc2 |
789 |
*/ |
b2ac2ea62 fs/epoll: use fas... |
790 |
for (rbp = rb_first_cached(&ep->rbr); rbp; rbp = rb_next(rbp)) { |
7699acd13 epoll cleanups: e... |
791 792 793 |
epi = rb_entry(rbp, struct epitem, rbn); ep_unregister_pollwait(ep, epi); |
91cf5ab60 epoll: add a resc... |
794 |
cond_resched(); |
7699acd13 epoll cleanups: e... |
795 |
} |
1da177e4c Linux-2.6.12-rc2 |
796 797 |
/* |
7699acd13 epoll cleanups: e... |
798 799 |
* 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 fix epoll single ... |
800 |
* holding "epmutex" we can be sure that no file cleanup code will hit |
ee8ef0a4b epoll: use the wa... |
801 |
* us during this operation. So we can avoid the lock on "ep->wq.lock". |
ddf676c38 epoll: lock ep->m... |
802 803 |
* We do not need to lock ep->mtx, either, we only do it to prevent * a lockdep warning. |
1da177e4c Linux-2.6.12-rc2 |
804 |
*/ |
ddf676c38 epoll: lock ep->m... |
805 |
mutex_lock(&ep->mtx); |
b2ac2ea62 fs/epoll: use fas... |
806 |
while ((rbp = rb_first_cached(&ep->rbr)) != NULL) { |
7699acd13 epoll cleanups: e... |
807 808 |
epi = rb_entry(rbp, struct epitem, rbn); ep_remove(ep, epi); |
91cf5ab60 epoll: add a resc... |
809 |
cond_resched(); |
7699acd13 epoll cleanups: e... |
810 |
} |
ddf676c38 epoll: lock ep->m... |
811 |
mutex_unlock(&ep->mtx); |
1da177e4c Linux-2.6.12-rc2 |
812 |
|
7699acd13 epoll cleanups: e... |
813 |
mutex_unlock(&epmutex); |
d47de16c7 fix epoll single ... |
814 |
mutex_destroy(&ep->mtx); |
7ef9964e6 epoll: introduce ... |
815 |
free_uid(ep->user); |
4d7e30d98 epoll: Add a flag... |
816 |
wakeup_source_unregister(ep->ws); |
f0ee9aabb epoll: move kfree... |
817 |
kfree(ep); |
7699acd13 epoll cleanups: e... |
818 |
} |
1da177e4c Linux-2.6.12-rc2 |
819 |
|
7699acd13 epoll cleanups: e... |
820 821 822 |
static int ep_eventpoll_release(struct inode *inode, struct file *file) { struct eventpoll *ep = file->private_data; |
1da177e4c Linux-2.6.12-rc2 |
823 |
|
f0ee9aabb epoll: move kfree... |
824 |
if (ep) |
7699acd13 epoll cleanups: e... |
825 |
ep_free(ep); |
7699acd13 epoll cleanups: e... |
826 |
|
7699acd13 epoll cleanups: e... |
827 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
828 |
} |
d85e2aa2e annotate ep_scan_... |
829 |
static __poll_t ep_read_events_proc(struct eventpoll *ep, struct list_head *head, |
37b5e5212 epoll: remove ep_... |
830 831 832 833 834 835 836 837 838 |
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 annotate ep_scan_... |
839 |
static __poll_t ep_item_poll(const struct epitem *epi, poll_table *pt, |
bec1a502d eventpoll: consti... |
840 |
int depth) |
450d89ec0 epoll: cleanup: h... |
841 |
{ |
37b5e5212 epoll: remove ep_... |
842 843 |
struct eventpoll *ep; bool locked; |
450d89ec0 epoll: cleanup: h... |
844 |
pt->_key = epi->event.events; |
37b5e5212 epoll: remove ep_... |
845 |
if (!is_file_epoll(epi->ffd.file)) |
9965ed174 fs: add new vfs_p... |
846 |
return vfs_poll(epi->ffd.file, pt) & epi->event.events; |
450d89ec0 epoll: cleanup: h... |
847 |
|
37b5e5212 epoll: remove ep_... |
848 849 850 |
ep = epi->ffd.file->private_data; poll_wait(epi->ffd.file, &ep->poll_wait, pt); locked = pt && (pt->_qproc == ep_ptable_queue_proc); |
450d89ec0 epoll: cleanup: h... |
851 |
|
37b5e5212 epoll: remove ep_... |
852 853 854 |
return ep_scan_ready_list(epi->ffd.file->private_data, ep_read_events_proc, &depth, depth, locked) & epi->event.events; |
450d89ec0 epoll: cleanup: h... |
855 |
} |
d85e2aa2e annotate ep_scan_... |
856 |
static __poll_t ep_read_events_proc(struct eventpoll *ep, struct list_head *head, |
296e236e9 epoll: fix epoll'... |
857 |
void *priv) |
5071f97ec epoll: fix epoll'... |
858 859 |
{ struct epitem *epi, *tmp; |
626cf2366 poll: add poll_re... |
860 |
poll_table pt; |
37b5e5212 epoll: remove ep_... |
861 |
int depth = *(int *)priv; |
5071f97ec epoll: fix epoll'... |
862 |
|
626cf2366 poll: add poll_re... |
863 |
init_poll_funcptr(&pt, NULL); |
37b5e5212 epoll: remove ep_... |
864 |
depth++; |
450d89ec0 epoll: cleanup: h... |
865 |
|
5071f97ec epoll: fix epoll'... |
866 |
list_for_each_entry_safe(epi, tmp, head, rdllink) { |
37b5e5212 epoll: remove ep_... |
867 |
if (ep_item_poll(epi, &pt, depth)) { |
a9a08845e vfs: do bulk POLL... |
868 |
return EPOLLIN | EPOLLRDNORM; |
37b5e5212 epoll: remove ep_... |
869 |
} else { |
5071f97ec epoll: fix epoll'... |
870 871 872 873 874 |
/* * 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 epoll: use RCU to... |
875 |
__pm_relax(ep_wakeup_source(epi)); |
5071f97ec epoll: fix epoll'... |
876 |
list_del_init(&epi->rdllink); |
296e236e9 epoll: fix epoll'... |
877 |
} |
5071f97ec epoll: fix epoll'... |
878 879 880 881 |
} return 0; } |
a11e1d432 Revert changes to... |
882 |
static __poll_t ep_eventpoll_poll(struct file *file, poll_table *wait) |
11c5ad0ec eventpoll: switch... |
883 884 885 |
{ struct eventpoll *ep = file->private_data; int depth = 0; |
7699acd13 epoll cleanups: e... |
886 |
|
a11e1d432 Revert changes to... |
887 888 |
/* Insert inside our poll wait queue */ poll_wait(file, &ep->poll_wait, wait); |
5071f97ec epoll: fix epoll'... |
889 890 |
/* * Proceed to find out if wanted events are really available inside |
37b5e5212 epoll: remove ep_... |
891 |
* the ready list. |
5071f97ec epoll: fix epoll'... |
892 |
*/ |
37b5e5212 epoll: remove ep_... |
893 894 |
return ep_scan_ready_list(ep, ep_read_events_proc, &depth, depth, false); |
7699acd13 epoll cleanups: e... |
895 |
} |
138d22b58 fs, epoll: add pr... |
896 |
#ifdef CONFIG_PROC_FS |
a3816ab0e fs: Convert show_... |
897 |
static void ep_show_fdinfo(struct seq_file *m, struct file *f) |
138d22b58 fs, epoll: add pr... |
898 899 900 |
{ struct eventpoll *ep = f->private_data; struct rb_node *rbp; |
138d22b58 fs, epoll: add pr... |
901 902 |
mutex_lock(&ep->mtx); |
b2ac2ea62 fs/epoll: use fas... |
903 |
for (rbp = rb_first_cached(&ep->rbr); rbp; rbp = rb_next(rbp)) { |
138d22b58 fs, epoll: add pr... |
904 |
struct epitem *epi = rb_entry(rbp, struct epitem, rbn); |
77493f04b procfs: fdinfo: e... |
905 |
struct inode *inode = file_inode(epi->ffd.file); |
138d22b58 fs, epoll: add pr... |
906 |
|
77493f04b procfs: fdinfo: e... |
907 908 909 |
seq_printf(m, "tfd: %8d events: %8x data: %16llx " " pos:%lli ino:%lx sdev:%x ", |
a3816ab0e fs: Convert show_... |
910 |
epi->ffd.fd, epi->event.events, |
77493f04b procfs: fdinfo: e... |
911 912 913 |
(long long)epi->event.data, (long long)epi->ffd.file->f_pos, inode->i_ino, inode->i_sb->s_dev); |
a3816ab0e fs: Convert show_... |
914 |
if (seq_has_overflowed(m)) |
138d22b58 fs, epoll: add pr... |
915 916 917 |
break; } mutex_unlock(&ep->mtx); |
138d22b58 fs, epoll: add pr... |
918 919 |
} #endif |
7699acd13 epoll cleanups: e... |
920 921 |
/* File callbacks that implement the eventpoll file behaviour */ static const struct file_operations eventpoll_fops = { |
138d22b58 fs, epoll: add pr... |
922 923 924 |
#ifdef CONFIG_PROC_FS .show_fdinfo = ep_show_fdinfo, #endif |
7699acd13 epoll cleanups: e... |
925 |
.release = ep_eventpoll_release, |
a11e1d432 Revert changes to... |
926 |
.poll = ep_eventpoll_poll, |
6038f373a llseek: automatic... |
927 |
.llseek = noop_llseek, |
7699acd13 epoll cleanups: e... |
928 |
}; |
b611967de [PATCH] epoll_pwa... |
929 |
/* |
7699acd13 epoll cleanups: e... |
930 931 932 |
* 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 [PATCH] epoll_pwa... |
933 |
*/ |
7699acd13 epoll cleanups: e... |
934 |
void eventpoll_release_file(struct file *file) |
b611967de [PATCH] epoll_pwa... |
935 |
{ |
7699acd13 epoll cleanups: e... |
936 |
struct eventpoll *ep; |
ebe06187b epoll: fix use-af... |
937 |
struct epitem *epi, *next; |
b611967de [PATCH] epoll_pwa... |
938 939 |
/* |
684999149 Rename struct fil... |
940 |
* We don't want to get "file->f_lock" because it is not |
7699acd13 epoll cleanups: e... |
941 |
* necessary. It is not necessary because we're in the "struct file" |
25985edce Fix common misspe... |
942 |
* cleanup path, and this means that no one is using this file anymore. |
5071f97ec epoll: fix epoll'... |
943 |
* So, for example, epoll_ctl() cannot hit here since if we reach this |
67647d0fb epoll: fix some c... |
944 |
* point, the file counter already went to zero and fget() would fail. |
d47de16c7 fix epoll single ... |
945 |
* The only hit might come from ep_free() but by holding the mutex |
7699acd13 epoll cleanups: e... |
946 |
* will correctly serialize the operation. We do need to acquire |
d47de16c7 fix epoll single ... |
947 |
* "ep->mtx" after "epmutex" because ep_remove() requires it when called |
7699acd13 epoll cleanups: e... |
948 |
* from anywhere but ep_free(). |
684999149 Rename struct fil... |
949 950 |
* * Besides, ep_remove() acquires the lock, so we can't hold it here. |
b611967de [PATCH] epoll_pwa... |
951 |
*/ |
7699acd13 epoll cleanups: e... |
952 |
mutex_lock(&epmutex); |
ebe06187b epoll: fix use-af... |
953 |
list_for_each_entry_safe(epi, next, &file->f_ep_links, fllink) { |
7699acd13 epoll cleanups: e... |
954 |
ep = epi->ep; |
d8805e633 epoll: fix spurio... |
955 |
mutex_lock_nested(&ep->mtx, 0); |
7699acd13 epoll cleanups: e... |
956 |
ep_remove(ep, epi); |
d47de16c7 fix epoll single ... |
957 |
mutex_unlock(&ep->mtx); |
b611967de [PATCH] epoll_pwa... |
958 |
} |
7699acd13 epoll cleanups: e... |
959 |
mutex_unlock(&epmutex); |
b611967de [PATCH] epoll_pwa... |
960 |
} |
53d2be79d [PATCH] epoll: fi... |
961 |
static int ep_alloc(struct eventpoll **pep) |
1da177e4c Linux-2.6.12-rc2 |
962 |
{ |
7ef9964e6 epoll: introduce ... |
963 964 965 |
int error; struct user_struct *user; struct eventpoll *ep; |
1da177e4c Linux-2.6.12-rc2 |
966 |
|
7ef9964e6 epoll: introduce ... |
967 |
user = get_current_user(); |
7ef9964e6 epoll: introduce ... |
968 969 970 971 |
error = -ENOMEM; ep = kzalloc(sizeof(*ep), GFP_KERNEL); if (unlikely(!ep)) goto free_uid; |
1da177e4c Linux-2.6.12-rc2 |
972 |
|
d47de16c7 fix epoll single ... |
973 |
mutex_init(&ep->mtx); |
1da177e4c Linux-2.6.12-rc2 |
974 975 976 |
init_waitqueue_head(&ep->wq); init_waitqueue_head(&ep->poll_wait); INIT_LIST_HEAD(&ep->rdllist); |
b2ac2ea62 fs/epoll: use fas... |
977 |
ep->rbr = RB_ROOT_CACHED; |
d47de16c7 fix epoll single ... |
978 |
ep->ovflist = EP_UNACTIVE_PTR; |
7ef9964e6 epoll: introduce ... |
979 |
ep->user = user; |
1da177e4c Linux-2.6.12-rc2 |
980 |
|
53d2be79d [PATCH] epoll: fi... |
981 |
*pep = ep; |
1da177e4c Linux-2.6.12-rc2 |
982 |
|
1da177e4c Linux-2.6.12-rc2 |
983 |
return 0; |
7ef9964e6 epoll: introduce ... |
984 985 986 987 |
free_uid: free_uid(user); return error; |
1da177e4c Linux-2.6.12-rc2 |
988 |
} |
1da177e4c Linux-2.6.12-rc2 |
989 |
/* |
c7ea76302 epoll locks chang... |
990 991 992 |
* 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 Linux-2.6.12-rc2 |
993 994 995 996 |
*/ static struct epitem *ep_find(struct eventpoll *ep, struct file *file, int fd) { int kcmp; |
1da177e4c Linux-2.6.12-rc2 |
997 998 999 |
struct rb_node *rbp; struct epitem *epi, *epir = NULL; struct epoll_filefd ffd; |
b030a4dd6 [PATCH] Remove ev... |
1000 |
ep_set_ffd(&ffd, file, fd); |
b2ac2ea62 fs/epoll: use fas... |
1001 |
for (rbp = ep->rbr.rb_root.rb_node; rbp; ) { |
1da177e4c Linux-2.6.12-rc2 |
1002 |
epi = rb_entry(rbp, struct epitem, rbn); |
b030a4dd6 [PATCH] Remove ev... |
1003 |
kcmp = ep_cmp_ffd(&ffd, &epi->ffd); |
1da177e4c Linux-2.6.12-rc2 |
1004 1005 1006 1007 1008 |
if (kcmp > 0) rbp = rbp->rb_right; else if (kcmp < 0) rbp = rbp->rb_left; else { |
1da177e4c Linux-2.6.12-rc2 |
1009 1010 1011 1012 |
epir = epi; break; } } |
1da177e4c Linux-2.6.12-rc2 |
1013 |
|
1da177e4c Linux-2.6.12-rc2 |
1014 1015 |
return epir; } |
92ef6da3d kcmp: fs/epoll: w... |
1016 |
#ifdef CONFIG_CHECKPOINT_RESTORE |
0791e3644 kcmp: add KCMP_EP... |
1017 1018 1019 1020 |
static struct epitem *ep_find_tfd(struct eventpoll *ep, int tfd, unsigned long toff) { struct rb_node *rbp; struct epitem *epi; |
b2ac2ea62 fs/epoll: use fas... |
1021 |
for (rbp = rb_first_cached(&ep->rbr); rbp; rbp = rb_next(rbp)) { |
0791e3644 kcmp: add KCMP_EP... |
1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 |
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 kcmp: fs/epoll: w... |
1057 |
#endif /* CONFIG_CHECKPOINT_RESTORE */ |
0791e3644 kcmp: add KCMP_EP... |
1058 |
|
1da177e4c Linux-2.6.12-rc2 |
1059 |
/* |
7699acd13 epoll cleanups: e... |
1060 |
* This is the callback that is passed to the wait queue wakeup |
bf6a41db7 fs/eventpoll.c: f... |
1061 |
* mechanism. It is called by the stored file descriptors when they |
7699acd13 epoll cleanups: e... |
1062 |
* have events to report. |
1da177e4c Linux-2.6.12-rc2 |
1063 |
*/ |
ac6424b98 sched/wait: Renam... |
1064 |
static int ep_poll_callback(wait_queue_entry_t *wait, unsigned mode, int sync, void *key) |
1da177e4c Linux-2.6.12-rc2 |
1065 |
{ |
7699acd13 epoll cleanups: e... |
1066 1067 1068 1069 |
int pwake = 0; unsigned long flags; struct epitem *epi = ep_item_from_wait(wait); struct eventpoll *ep = epi->ep; |
3ad6f93e9 annotate poll-rel... |
1070 |
__poll_t pollflags = key_to_poll(key); |
df0108c5d epoll: add EPOLLE... |
1071 |
int ewake = 0; |
1da177e4c Linux-2.6.12-rc2 |
1072 |
|
ee8ef0a4b epoll: use the wa... |
1073 |
spin_lock_irqsave(&ep->wq.lock, flags); |
1da177e4c Linux-2.6.12-rc2 |
1074 |
|
bf3b9f637 epoll: Add busy p... |
1075 |
ep_set_busy_poll_napi_id(epi); |
7699acd13 epoll cleanups: e... |
1076 1077 1078 1079 1080 1081 1082 |
/* * 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 fix epoll single ... |
1083 1084 1085 |
goto out_unlock; /* |
2dfa4eeab epoll keyed wakeu... |
1086 1087 1088 1089 1090 |
* 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 annotate poll-rel... |
1091 |
if (pollflags && !(pollflags & epi->event.events)) |
2dfa4eeab epoll keyed wakeu... |
1092 1093 1094 |
goto out_unlock; /* |
bf6a41db7 fs/eventpoll.c: f... |
1095 |
* If we are transferring events to userspace, we can hold no locks |
d47de16c7 fix epoll single ... |
1096 |
* (because we're accessing user memory, and because of linux f_op->poll() |
bf6a41db7 fs/eventpoll.c: f... |
1097 |
* semantics). All the events that happen during that period of time are |
d47de16c7 fix epoll single ... |
1098 1099 |
* chained in ep->ovflist and requeued later on. */ |
9cb8f8088 fs/epoll: drop ov... |
1100 |
if (ep->ovflist != EP_UNACTIVE_PTR) { |
d47de16c7 fix epoll single ... |
1101 1102 1103 |
if (epi->next == EP_UNACTIVE_PTR) { epi->next = ep->ovflist; ep->ovflist = epi; |
4d7e30d98 epoll: Add a flag... |
1104 1105 1106 1107 1108 1109 1110 |
if (epi->ws) { /* * Activate ep->ws since epi->ws may get * deactivated at any time. */ __pm_stay_awake(ep->ws); } |
d47de16c7 fix epoll single ... |
1111 1112 1113 |
} goto out_unlock; } |
1da177e4c Linux-2.6.12-rc2 |
1114 |
|
7699acd13 epoll cleanups: e... |
1115 |
/* If this file is already in the ready list we exit soon */ |
992991c03 fs/eventpoll.c: s... |
1116 |
if (!ep_is_linked(epi)) { |
5071f97ec epoll: fix epoll'... |
1117 |
list_add_tail(&epi->rdllink, &ep->rdllist); |
eea1d5859 epoll: use RCU to... |
1118 |
ep_pm_stay_awake_rcu(epi); |
4d7e30d98 epoll: Add a flag... |
1119 |
} |
7699acd13 epoll cleanups: e... |
1120 |
|
7699acd13 epoll cleanups: e... |
1121 1122 1123 1124 |
/* * Wake up ( if active ) both the eventpoll wait list and the ->poll() * wait list. */ |
df0108c5d epoll: add EPOLLE... |
1125 |
if (waitqueue_active(&ep->wq)) { |
b6a515c8a epoll: restrict E... |
1126 |
if ((epi->event.events & EPOLLEXCLUSIVE) && |
3ad6f93e9 annotate poll-rel... |
1127 1128 |
!(pollflags & POLLFREE)) { switch (pollflags & EPOLLINOUT_BITS) { |
a9a08845e vfs: do bulk POLL... |
1129 1130 |
case EPOLLIN: if (epi->event.events & EPOLLIN) |
b6a515c8a epoll: restrict E... |
1131 1132 |
ewake = 1; break; |
a9a08845e vfs: do bulk POLL... |
1133 1134 |
case EPOLLOUT: if (epi->event.events & EPOLLOUT) |
b6a515c8a epoll: restrict E... |
1135 1136 1137 1138 1139 1140 1141 |
ewake = 1; break; case 0: ewake = 1; break; } } |
4a6e9e2ce Use wake_up_locke... |
1142 |
wake_up_locked(&ep->wq); |
df0108c5d epoll: add EPOLLE... |
1143 |
} |
7699acd13 epoll cleanups: e... |
1144 1145 |
if (waitqueue_active(&ep->poll_wait)) pwake++; |
d47de16c7 fix epoll single ... |
1146 |
out_unlock: |
ee8ef0a4b epoll: use the wa... |
1147 |
spin_unlock_irqrestore(&ep->wq.lock, flags); |
1da177e4c Linux-2.6.12-rc2 |
1148 |
|
7699acd13 epoll cleanups: e... |
1149 1150 |
/* We have to call this outside the lock */ if (pwake) |
5071f97ec epoll: fix epoll'... |
1151 |
ep_poll_safewake(&ep->poll_wait); |
7699acd13 epoll cleanups: e... |
1152 |
|
138e4ad67 epoll: fix race b... |
1153 1154 |
if (!(epi->event.events & EPOLLEXCLUSIVE)) ewake = 1; |
3ad6f93e9 annotate poll-rel... |
1155 |
if (pollflags & POLLFREE) { |
138e4ad67 epoll: fix race b... |
1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 |
/* * 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 epoll: add EPOLLE... |
1170 |
|
138e4ad67 epoll: fix race b... |
1171 |
return ewake; |
7699acd13 epoll cleanups: e... |
1172 |
} |
1da177e4c Linux-2.6.12-rc2 |
1173 1174 1175 1176 1177 1178 1179 1180 |
/* * 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 [PATCH] Remove ev... |
1181 |
struct epitem *epi = ep_item_from_epqueue(pt); |
1da177e4c Linux-2.6.12-rc2 |
1182 |
struct eppoll_entry *pwq; |
e94b17660 [PATCH] slab: rem... |
1183 |
if (epi->nwait >= 0 && (pwq = kmem_cache_alloc(pwq_cache, GFP_KERNEL))) { |
1da177e4c Linux-2.6.12-rc2 |
1184 1185 1186 |
init_waitqueue_func_entry(&pwq->wait, ep_poll_callback); pwq->whead = whead; pwq->base = epi; |
df0108c5d epoll: add EPOLLE... |
1187 1188 1189 1190 |
if (epi->event.events & EPOLLEXCLUSIVE) add_wait_queue_exclusive(whead, &pwq->wait); else add_wait_queue(whead, &pwq->wait); |
1da177e4c Linux-2.6.12-rc2 |
1191 1192 |
list_add_tail(&pwq->llink, &epi->pwqlist); epi->nwait++; |
296e236e9 epoll: fix epoll'... |
1193 |
} else { |
1da177e4c Linux-2.6.12-rc2 |
1194 1195 |
/* We have to signal that an error occurred */ epi->nwait = -1; |
296e236e9 epoll: fix epoll'... |
1196 |
} |
1da177e4c Linux-2.6.12-rc2 |
1197 |
} |
1da177e4c Linux-2.6.12-rc2 |
1198 1199 1200 |
static void ep_rbtree_insert(struct eventpoll *ep, struct epitem *epi) { int kcmp; |
b2ac2ea62 fs/epoll: use fas... |
1201 |
struct rb_node **p = &ep->rbr.rb_root.rb_node, *parent = NULL; |
1da177e4c Linux-2.6.12-rc2 |
1202 |
struct epitem *epic; |
b2ac2ea62 fs/epoll: use fas... |
1203 |
bool leftmost = true; |
1da177e4c Linux-2.6.12-rc2 |
1204 1205 1206 1207 |
while (*p) { parent = *p; epic = rb_entry(parent, struct epitem, rbn); |
b030a4dd6 [PATCH] Remove ev... |
1208 |
kcmp = ep_cmp_ffd(&epi->ffd, &epic->ffd); |
b2ac2ea62 fs/epoll: use fas... |
1209 |
if (kcmp > 0) { |
1da177e4c Linux-2.6.12-rc2 |
1210 |
p = &parent->rb_right; |
b2ac2ea62 fs/epoll: use fas... |
1211 1212 |
leftmost = false; } else |
1da177e4c Linux-2.6.12-rc2 |
1213 1214 1215 |
p = &parent->rb_left; } rb_link_node(&epi->rbn, parent, p); |
b2ac2ea62 fs/epoll: use fas... |
1216 |
rb_insert_color_cached(&epi->rbn, &ep->rbr, leftmost); |
1da177e4c Linux-2.6.12-rc2 |
1217 |
} |
a80a6b85b revert "epoll: su... |
1218 |
|
28d82dc1c epoll: limit paths |
1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 |
#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 Don't limit non-n... |
1236 1237 1238 |
/* Allow an arbitrary number of depth 1 paths */ if (nests == 0) return 0; |
28d82dc1c epoll: limit paths |
1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 |
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 epoll: optimize E... |
1258 1259 1260 |
/* 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 epoll: limit paths |
1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 |
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, EP_MAX_NESTS, 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 epoll: optimize E... |
1283 |
rcu_read_unlock(); |
28d82dc1c epoll: limit paths |
1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 |
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 epoll: limit paths |
1299 1300 1301 1302 1303 |
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 epoll: limit paths |
1304 1305 1306 1307 1308 1309 1310 1311 1312 |
path_count_init(); error = ep_call_nested(&poll_loop_ncalls, EP_MAX_NESTS, reverse_path_check_proc, current_file, current_file, current); if (error) break; } return error; } |
4d7e30d98 epoll: Add a flag... |
1313 1314 1315 |
static int ep_create_wakeup_source(struct epitem *epi) { const char *name; |
eea1d5859 epoll: use RCU to... |
1316 |
struct wakeup_source *ws; |
4d7e30d98 epoll: Add a flag... |
1317 1318 1319 1320 1321 1322 1323 1324 |
if (!epi->ep->ws) { epi->ep->ws = wakeup_source_register("eventpoll"); if (!epi->ep->ws) return -ENOMEM; } name = epi->ffd.file->f_path.dentry->d_name.name; |
eea1d5859 epoll: use RCU to... |
1325 1326 1327 |
ws = wakeup_source_register(name); if (!ws) |
4d7e30d98 epoll: Add a flag... |
1328 |
return -ENOMEM; |
eea1d5859 epoll: use RCU to... |
1329 |
rcu_assign_pointer(epi->ws, ws); |
4d7e30d98 epoll: Add a flag... |
1330 1331 1332 |
return 0; } |
eea1d5859 epoll: use RCU to... |
1333 1334 |
/* rare code path, only used when EPOLL_CTL_MOD removes a wakeup source */ static noinline void ep_destroy_wakeup_source(struct epitem *epi) |
4d7e30d98 epoll: Add a flag... |
1335 |
{ |
eea1d5859 epoll: use RCU to... |
1336 |
struct wakeup_source *ws = ep_wakeup_source(epi); |
d6d67e723 epoll: cleanup: u... |
1337 |
RCU_INIT_POINTER(epi->ws, NULL); |
eea1d5859 epoll: use RCU to... |
1338 1339 1340 1341 1342 1343 1344 1345 |
/* * 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 epoll: Add a flag... |
1346 |
} |
c7ea76302 epoll locks chang... |
1347 1348 1349 |
/* * Must be called with "mtx" held. */ |
bec1a502d eventpoll: consti... |
1350 |
static int ep_insert(struct eventpoll *ep, const struct epoll_event *event, |
67347fe4e epoll: do not tak... |
1351 |
struct file *tfile, int fd, int full_check) |
1da177e4c Linux-2.6.12-rc2 |
1352 |
{ |
d85e2aa2e annotate ep_scan_... |
1353 1354 |
int error, pwake = 0; __poll_t revents; |
52bd19f76 epoll: convert ma... |
1355 |
long user_watches; |
1da177e4c Linux-2.6.12-rc2 |
1356 1357 |
struct epitem *epi; struct ep_pqueue epq; |
92e641784 s/epoll: robustif... |
1358 |
lockdep_assert_irqs_enabled(); |
52bd19f76 epoll: convert ma... |
1359 1360 |
user_watches = atomic_long_read(&ep->user->epoll_watches); if (unlikely(user_watches >= max_user_watches)) |
7ef9964e6 epoll: introduce ... |
1361 |
return -ENOSPC; |
e94b17660 [PATCH] slab: rem... |
1362 |
if (!(epi = kmem_cache_alloc(epi_cache, GFP_KERNEL))) |
7ef9964e6 epoll: introduce ... |
1363 |
return -ENOMEM; |
1da177e4c Linux-2.6.12-rc2 |
1364 1365 |
/* Item initialization follow here ... */ |
1da177e4c Linux-2.6.12-rc2 |
1366 1367 |
INIT_LIST_HEAD(&epi->rdllink); INIT_LIST_HEAD(&epi->fllink); |
1da177e4c Linux-2.6.12-rc2 |
1368 1369 |
INIT_LIST_HEAD(&epi->pwqlist); epi->ep = ep; |
b030a4dd6 [PATCH] Remove ev... |
1370 |
ep_set_ffd(&epi->ffd, tfile, fd); |
1da177e4c Linux-2.6.12-rc2 |
1371 |
epi->event = *event; |
1da177e4c Linux-2.6.12-rc2 |
1372 |
epi->nwait = 0; |
d47de16c7 fix epoll single ... |
1373 |
epi->next = EP_UNACTIVE_PTR; |
4d7e30d98 epoll: Add a flag... |
1374 1375 1376 1377 1378 |
if (epi->event.events & EPOLLWAKEUP) { error = ep_create_wakeup_source(epi); if (error) goto error_create_wakeup_source; } else { |
eea1d5859 epoll: use RCU to... |
1379 |
RCU_INIT_POINTER(epi->ws, NULL); |
4d7e30d98 epoll: Add a flag... |
1380 |
} |
1da177e4c Linux-2.6.12-rc2 |
1381 1382 1383 1384 1385 1386 1387 1388 |
/* 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 epoll locks chang... |
1389 1390 1391 |
* been increased by the caller of this function. Note that after * this operation completes, the poll callback can start hitting * the new item. |
1da177e4c Linux-2.6.12-rc2 |
1392 |
*/ |
37b5e5212 epoll: remove ep_... |
1393 |
revents = ep_item_poll(epi, &epq.pt, 1); |
1da177e4c Linux-2.6.12-rc2 |
1394 1395 1396 1397 1398 1399 |
/* * 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 epoll: introduce ... |
1400 |
error = -ENOMEM; |
1da177e4c Linux-2.6.12-rc2 |
1401 |
if (epi->nwait < 0) |
7699acd13 epoll cleanups: e... |
1402 |
goto error_unregister; |
1da177e4c Linux-2.6.12-rc2 |
1403 1404 |
/* Add the current item to the list of active epoll hook for this file */ |
684999149 Rename struct fil... |
1405 |
spin_lock(&tfile->f_lock); |
ae10b2b4e epoll: optimize E... |
1406 |
list_add_tail_rcu(&epi->fllink, &tfile->f_ep_links); |
684999149 Rename struct fil... |
1407 |
spin_unlock(&tfile->f_lock); |
1da177e4c Linux-2.6.12-rc2 |
1408 |
|
c7ea76302 epoll locks chang... |
1409 1410 1411 1412 |
/* * 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 Linux-2.6.12-rc2 |
1413 |
ep_rbtree_insert(ep, epi); |
28d82dc1c epoll: limit paths |
1414 1415 |
/* now check if we've created too many backpaths */ error = -EINVAL; |
67347fe4e epoll: do not tak... |
1416 |
if (full_check && reverse_path_check()) |
28d82dc1c epoll: limit paths |
1417 |
goto error_remove_epi; |
c7ea76302 epoll locks chang... |
1418 |
/* We have to drop the new item inside our item list to keep track of it */ |
304b18b8d fs/epoll: loosen ... |
1419 |
spin_lock_irq(&ep->wq.lock); |
c7ea76302 epoll locks chang... |
1420 |
|
bf3b9f637 epoll: Add busy p... |
1421 1422 |
/* record NAPI ID of new item if present */ ep_set_busy_poll_napi_id(epi); |
1da177e4c Linux-2.6.12-rc2 |
1423 |
/* If the file is already "ready" we drop it inside the ready list */ |
992991c03 fs/eventpoll.c: s... |
1424 |
if (revents && !ep_is_linked(epi)) { |
1da177e4c Linux-2.6.12-rc2 |
1425 |
list_add_tail(&epi->rdllink, &ep->rdllist); |
eea1d5859 epoll: use RCU to... |
1426 |
ep_pm_stay_awake(epi); |
1da177e4c Linux-2.6.12-rc2 |
1427 1428 1429 |
/* Notify waiting tasks that events are available */ if (waitqueue_active(&ep->wq)) |
4a6e9e2ce Use wake_up_locke... |
1430 |
wake_up_locked(&ep->wq); |
1da177e4c Linux-2.6.12-rc2 |
1431 1432 1433 |
if (waitqueue_active(&ep->poll_wait)) pwake++; } |
304b18b8d fs/epoll: loosen ... |
1434 |
spin_unlock_irq(&ep->wq.lock); |
1da177e4c Linux-2.6.12-rc2 |
1435 |
|
52bd19f76 epoll: convert ma... |
1436 |
atomic_long_inc(&ep->user->epoll_watches); |
7ef9964e6 epoll: introduce ... |
1437 |
|
1da177e4c Linux-2.6.12-rc2 |
1438 1439 |
/* We have to call this outside the lock */ if (pwake) |
5071f97ec epoll: fix epoll'... |
1440 |
ep_poll_safewake(&ep->poll_wait); |
1da177e4c Linux-2.6.12-rc2 |
1441 |
|
1da177e4c Linux-2.6.12-rc2 |
1442 |
return 0; |
28d82dc1c epoll: limit paths |
1443 1444 |
error_remove_epi: spin_lock(&tfile->f_lock); |
ae10b2b4e epoll: optimize E... |
1445 |
list_del_rcu(&epi->fllink); |
28d82dc1c epoll: limit paths |
1446 |
spin_unlock(&tfile->f_lock); |
b2ac2ea62 fs/epoll: use fas... |
1447 |
rb_erase_cached(&epi->rbn, &ep->rbr); |
28d82dc1c epoll: limit paths |
1448 |
|
7699acd13 epoll cleanups: e... |
1449 |
error_unregister: |
1da177e4c Linux-2.6.12-rc2 |
1450 1451 1452 1453 |
ep_unregister_pollwait(ep, epi); /* * We need to do this because an event could have been arrived on some |
67647d0fb epoll: fix some c... |
1454 1455 1456 |
* 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 Linux-2.6.12-rc2 |
1457 |
*/ |
304b18b8d fs/epoll: loosen ... |
1458 |
spin_lock_irq(&ep->wq.lock); |
992991c03 fs/eventpoll.c: s... |
1459 |
if (ep_is_linked(epi)) |
6192bd536 epoll: optimizati... |
1460 |
list_del_init(&epi->rdllink); |
304b18b8d fs/epoll: loosen ... |
1461 |
spin_unlock_irq(&ep->wq.lock); |
1da177e4c Linux-2.6.12-rc2 |
1462 |
|
eea1d5859 epoll: use RCU to... |
1463 |
wakeup_source_unregister(ep_wakeup_source(epi)); |
4d7e30d98 epoll: Add a flag... |
1464 1465 |
error_create_wakeup_source: |
b030a4dd6 [PATCH] Remove ev... |
1466 |
kmem_cache_free(epi_cache, epi); |
7ef9964e6 epoll: introduce ... |
1467 |
|
1da177e4c Linux-2.6.12-rc2 |
1468 1469 |
return error; } |
1da177e4c Linux-2.6.12-rc2 |
1470 1471 |
/* * Modify the interest event mask by dropping an event if the new mask |
c7ea76302 epoll locks chang... |
1472 |
* has a match in the current file status. Must be called with "mtx" held. |
1da177e4c Linux-2.6.12-rc2 |
1473 |
*/ |
bec1a502d eventpoll: consti... |
1474 1475 |
static int ep_modify(struct eventpoll *ep, struct epitem *epi, const struct epoll_event *event) |
1da177e4c Linux-2.6.12-rc2 |
1476 1477 |
{ int pwake = 0; |
626cf2366 poll: add poll_re... |
1478 |
poll_table pt; |
92e641784 s/epoll: robustif... |
1479 |
lockdep_assert_irqs_enabled(); |
626cf2366 poll: add poll_re... |
1480 |
init_poll_funcptr(&pt, NULL); |
1da177e4c Linux-2.6.12-rc2 |
1481 1482 |
/* |
e057e15ff epoll: clean up e... |
1483 1484 1485 |
* 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 Linux-2.6.12-rc2 |
1486 |
*/ |
128dd1759 epoll: prevent mi... |
1487 |
epi->event.events = event->events; /* need barrier below */ |
e057e15ff epoll: clean up e... |
1488 |
epi->event.data = event->data; /* protected by mtx */ |
4d7e30d98 epoll: Add a flag... |
1489 |
if (epi->event.events & EPOLLWAKEUP) { |
eea1d5859 epoll: use RCU to... |
1490 |
if (!ep_has_wakeup_source(epi)) |
4d7e30d98 epoll: Add a flag... |
1491 |
ep_create_wakeup_source(epi); |
eea1d5859 epoll: use RCU to... |
1492 |
} else if (ep_has_wakeup_source(epi)) { |
4d7e30d98 epoll: Add a flag... |
1493 1494 |
ep_destroy_wakeup_source(epi); } |
1da177e4c Linux-2.6.12-rc2 |
1495 1496 |
/* |
128dd1759 epoll: prevent mi... |
1497 1498 1499 1500 1501 |
* 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(). |
ee8ef0a4b epoll: use the wa... |
1502 |
* We need this because we did not take ep->wq.lock while |
128dd1759 epoll: prevent mi... |
1503 |
* changing epi above (but ep_poll_callback does take |
ee8ef0a4b epoll: use the wa... |
1504 |
* ep->wq.lock). |
128dd1759 epoll: prevent mi... |
1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 |
* * 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 Linux-2.6.12-rc2 |
1517 1518 |
* 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 epoll locks chang... |
1519 |
* If the item is "hot" and it is not registered inside the ready |
67647d0fb epoll: fix some c... |
1520 |
* list, push it inside. |
1da177e4c Linux-2.6.12-rc2 |
1521 |
*/ |
69112736e eventpoll: no nee... |
1522 |
if (ep_item_poll(epi, &pt, 1)) { |
ee8ef0a4b epoll: use the wa... |
1523 |
spin_lock_irq(&ep->wq.lock); |
992991c03 fs/eventpoll.c: s... |
1524 |
if (!ep_is_linked(epi)) { |
c7ea76302 epoll locks chang... |
1525 |
list_add_tail(&epi->rdllink, &ep->rdllist); |
eea1d5859 epoll: use RCU to... |
1526 |
ep_pm_stay_awake(epi); |
c7ea76302 epoll locks chang... |
1527 1528 1529 |
/* Notify waiting tasks that events are available */ if (waitqueue_active(&ep->wq)) |
4a6e9e2ce Use wake_up_locke... |
1530 |
wake_up_locked(&ep->wq); |
c7ea76302 epoll locks chang... |
1531 1532 |
if (waitqueue_active(&ep->poll_wait)) pwake++; |
7699acd13 epoll cleanups: e... |
1533 |
} |
ee8ef0a4b epoll: use the wa... |
1534 |
spin_unlock_irq(&ep->wq.lock); |
7699acd13 epoll cleanups: e... |
1535 |
} |
1da177e4c Linux-2.6.12-rc2 |
1536 |
|
7699acd13 epoll cleanups: e... |
1537 1538 |
/* We have to call this outside the lock */ if (pwake) |
5071f97ec epoll: fix epoll'... |
1539 |
ep_poll_safewake(&ep->poll_wait); |
1da177e4c Linux-2.6.12-rc2 |
1540 |
|
7699acd13 epoll cleanups: e... |
1541 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
1542 |
} |
d85e2aa2e annotate ep_scan_... |
1543 |
static __poll_t ep_send_events_proc(struct eventpoll *ep, struct list_head *head, |
296e236e9 epoll: fix epoll'... |
1544 |
void *priv) |
1da177e4c Linux-2.6.12-rc2 |
1545 |
{ |
5071f97ec epoll: fix epoll'... |
1546 |
struct ep_send_events_data *esed = priv; |
d85e2aa2e annotate ep_scan_... |
1547 |
__poll_t revents; |
5071f97ec epoll: fix epoll'... |
1548 1549 |
struct epitem *epi; struct epoll_event __user *uevent; |
eea1d5859 epoll: use RCU to... |
1550 |
struct wakeup_source *ws; |
626cf2366 poll: add poll_re... |
1551 1552 1553 |
poll_table pt; init_poll_funcptr(&pt, NULL); |
1da177e4c Linux-2.6.12-rc2 |
1554 |
|
296e236e9 epoll: fix epoll'... |
1555 |
/* |
5071f97ec epoll: fix epoll'... |
1556 1557 1558 |
* 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 epoll: fix epoll'... |
1559 |
*/ |
d7ebbe46f ep_send_events_pr... |
1560 1561 |
for (esed->res = 0, uevent = esed->events; !list_empty(head) && esed->res < esed->maxevents;) { |
5071f97ec epoll: fix epoll'... |
1562 |
epi = list_first_entry(head, struct epitem, rdllink); |
d47de16c7 fix epoll single ... |
1563 |
|
4d7e30d98 epoll: Add a flag... |
1564 1565 1566 1567 1568 1569 1570 1571 1572 |
/* * 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 epoll: use RCU to... |
1573 1574 1575 1576 1577 1578 |
ws = ep_wakeup_source(epi); if (ws) { if (ws->active) __pm_stay_awake(ep->ws); __pm_relax(ws); } |
d47de16c7 fix epoll single ... |
1579 |
list_del_init(&epi->rdllink); |
1da177e4c Linux-2.6.12-rc2 |
1580 |
|
37b5e5212 epoll: remove ep_... |
1581 |
revents = ep_item_poll(epi, &pt, 1); |
5071f97ec epoll: fix epoll'... |
1582 |
|
296e236e9 epoll: fix epoll'... |
1583 |
/* |
5071f97ec epoll: fix epoll'... |
1584 1585 1586 1587 |
* If the event mask intersect the caller-requested one, * deliver the event to userspace. Again, ep_scan_ready_list() * is holding "mtx", so no operations coming from userspace * can change the item. |
296e236e9 epoll: fix epoll'... |
1588 1589 |
*/ if (revents) { |
5071f97ec epoll: fix epoll'... |
1590 |
if (__put_user(revents, &uevent->events) || |
d03058828 epoll: remember t... |
1591 1592 |
__put_user(epi->event.data, &uevent->data)) { list_add(&epi->rdllink, head); |
eea1d5859 epoll: use RCU to... |
1593 |
ep_pm_stay_awake(epi); |
d7ebbe46f ep_send_events_pr... |
1594 1595 1596 |
if (!esed->res) esed->res = -EFAULT; return 0; |
d03058828 epoll: remember t... |
1597 |
} |
d7ebbe46f ep_send_events_pr... |
1598 |
esed->res++; |
5071f97ec epoll: fix epoll'... |
1599 |
uevent++; |
296e236e9 epoll: fix epoll'... |
1600 1601 1602 1603 1604 1605 1606 1607 |
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 |
25985edce Fix common misspe... |
1608 |
* availability. At this point, no one can insert |
296e236e9 epoll: fix epoll'... |
1609 1610 1611 1612 1613 1614 |
* 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); |
eea1d5859 epoll: use RCU to... |
1615 |
ep_pm_stay_awake(epi); |
296e236e9 epoll: fix epoll'... |
1616 1617 1618 |
} } } |
5071f97ec epoll: fix epoll'... |
1619 |
|
d7ebbe46f ep_send_events_pr... |
1620 |
return 0; |
5071f97ec epoll: fix epoll'... |
1621 |
} |
d47de16c7 fix epoll single ... |
1622 |
|
296e236e9 epoll: fix epoll'... |
1623 1624 |
static int ep_send_events(struct eventpoll *ep, struct epoll_event __user *events, int maxevents) |
5071f97ec epoll: fix epoll'... |
1625 1626 |
{ struct ep_send_events_data esed; |
1da177e4c Linux-2.6.12-rc2 |
1627 |
|
5071f97ec epoll: fix epoll'... |
1628 1629 |
esed.maxevents = maxevents; esed.events = events; |
6192bd536 epoll: optimizati... |
1630 |
|
d7ebbe46f ep_send_events_pr... |
1631 1632 |
ep_scan_ready_list(ep, ep_send_events_proc, &esed, 0, false); return esed.res; |
1da177e4c Linux-2.6.12-rc2 |
1633 |
} |
766b9f928 fs: poll/select/r... |
1634 |
static inline struct timespec64 ep_set_mstimeout(long ms) |
0781b909b epoll: epoll_wait... |
1635 |
{ |
766b9f928 fs: poll/select/r... |
1636 |
struct timespec64 now, ts = { |
0781b909b epoll: epoll_wait... |
1637 1638 1639 |
.tv_sec = ms / MSEC_PER_SEC, .tv_nsec = NSEC_PER_MSEC * (ms % MSEC_PER_SEC), }; |
766b9f928 fs: poll/select/r... |
1640 1641 |
ktime_get_ts64(&now); return timespec64_add_safe(now, ts); |
0781b909b epoll: epoll_wait... |
1642 |
} |
f4d93ad74 epoll: fix compil... |
1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 |
/** * 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 Linux-2.6.12-rc2 |
1660 1661 1662 |
static int ep_poll(struct eventpoll *ep, struct epoll_event __user *events, int maxevents, long timeout) { |
f4d93ad74 epoll: fix compil... |
1663 |
int res = 0, eavail, timed_out = 0; |
da8b44d5a timer: convert ti... |
1664 |
u64 slack = 0; |
ac6424b98 sched/wait: Renam... |
1665 |
wait_queue_entry_t wait; |
95aac7b1c epoll: make epoll... |
1666 |
ktime_t expires, *to = NULL; |
679abf381 fs/eventpoll.c: l... |
1667 |
lockdep_assert_irqs_enabled(); |
95aac7b1c epoll: make epoll... |
1668 |
if (timeout > 0) { |
766b9f928 fs: poll/select/r... |
1669 |
struct timespec64 end_time = ep_set_mstimeout(timeout); |
0781b909b epoll: epoll_wait... |
1670 |
|
95aac7b1c epoll: make epoll... |
1671 1672 |
slack = select_estimate_accuracy(&end_time); to = &expires; |
766b9f928 fs: poll/select/r... |
1673 |
*to = timespec64_to_ktime(end_time); |
95aac7b1c epoll: make epoll... |
1674 |
} else if (timeout == 0) { |
f4d93ad74 epoll: fix compil... |
1675 1676 1677 1678 |
/* * Avoid the unnecessary trip to the wait queue loop, if the * caller specified a non blocking operation. */ |
95aac7b1c epoll: make epoll... |
1679 |
timed_out = 1; |
679abf381 fs/eventpoll.c: l... |
1680 |
spin_lock_irq(&ep->wq.lock); |
f4d93ad74 epoll: fix compil... |
1681 |
goto check_events; |
95aac7b1c epoll: make epoll... |
1682 |
} |
1da177e4c Linux-2.6.12-rc2 |
1683 |
|
f4d93ad74 epoll: fix compil... |
1684 |
fetch_events: |
bf3b9f637 epoll: Add busy p... |
1685 1686 1687 |
if (!ep_events_available(ep)) ep_busy_loop(ep, timed_out); |
679abf381 fs/eventpoll.c: l... |
1688 |
spin_lock_irq(&ep->wq.lock); |
1da177e4c Linux-2.6.12-rc2 |
1689 |
|
3fb0e584a epoll: move ready... |
1690 |
if (!ep_events_available(ep)) { |
1da177e4c Linux-2.6.12-rc2 |
1691 |
/* |
bf3b9f637 epoll: Add busy p... |
1692 1693 1694 1695 1696 1697 1698 |
* 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); /* |
1da177e4c Linux-2.6.12-rc2 |
1699 1700 1701 1702 1703 |
* We don't have any available event to return to the caller. * We need to sleep here, and we will be wake up by * ep_poll_callback() when events will become available. */ init_waitqueue_entry(&wait, current); |
a93d2f174 sched, wait: Use ... |
1704 |
__add_wait_queue_exclusive(&ep->wq, &wait); |
1da177e4c Linux-2.6.12-rc2 |
1705 1706 1707 1708 1709 1710 1711 1712 |
for (;;) { /* * 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. */ set_current_state(TASK_INTERRUPTIBLE); |
c257a340e fs, epoll: short ... |
1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 |
/* * 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)) { res = -EINTR; break; } |
3fb0e584a epoll: move ready... |
1723 |
if (ep_events_available(ep) || timed_out) |
1da177e4c Linux-2.6.12-rc2 |
1724 1725 1726 1727 1728 |
break; if (signal_pending(current)) { res = -EINTR; break; } |
679abf381 fs/eventpoll.c: l... |
1729 |
spin_unlock_irq(&ep->wq.lock); |
c511851de Revert "epoll: us... |
1730 |
if (!schedule_hrtimeout_range(to, slack, HRTIMER_MODE_ABS)) |
95aac7b1c epoll: make epoll... |
1731 |
timed_out = 1; |
679abf381 fs/eventpoll.c: l... |
1732 |
spin_lock_irq(&ep->wq.lock); |
1da177e4c Linux-2.6.12-rc2 |
1733 |
} |
1da177e4c Linux-2.6.12-rc2 |
1734 |
|
4d5755b14 epoll: optimize s... |
1735 1736 |
__remove_wait_queue(&ep->wq, &wait); __set_current_state(TASK_RUNNING); |
1da177e4c Linux-2.6.12-rc2 |
1737 |
} |
f4d93ad74 epoll: fix compil... |
1738 |
check_events: |
1da177e4c Linux-2.6.12-rc2 |
1739 |
/* Is it worth to try to dig for events ? */ |
3fb0e584a epoll: move ready... |
1740 |
eavail = ep_events_available(ep); |
1da177e4c Linux-2.6.12-rc2 |
1741 |
|
679abf381 fs/eventpoll.c: l... |
1742 |
spin_unlock_irq(&ep->wq.lock); |
1da177e4c Linux-2.6.12-rc2 |
1743 1744 1745 1746 1747 1748 1749 |
/* * 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 epoll: make epoll... |
1750 |
!(res = ep_send_events(ep, events, maxevents)) && !timed_out) |
f4d93ad74 epoll: fix compil... |
1751 |
goto fetch_events; |
1da177e4c Linux-2.6.12-rc2 |
1752 1753 1754 |
return res; } |
22bacca48 epoll: prevent cr... |
1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 |
/** * 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 epoll: limit paths |
1775 |
struct eventpoll *ep_tovisit; |
22bacca48 epoll: prevent cr... |
1776 1777 |
struct rb_node *rbp; struct epitem *epi; |
d8805e633 epoll: fix spurio... |
1778 |
mutex_lock_nested(&ep->mtx, call_nests + 1); |
28d82dc1c epoll: limit paths |
1779 1780 |
ep->visited = 1; list_add(&ep->visited_list_link, &visited_list); |
b2ac2ea62 fs/epoll: use fas... |
1781 |
for (rbp = rb_first_cached(&ep->rbr); rbp; rbp = rb_next(rbp)) { |
22bacca48 epoll: prevent cr... |
1782 1783 |
epi = rb_entry(rbp, struct epitem, rbn); if (unlikely(is_file_epoll(epi->ffd.file))) { |
28d82dc1c epoll: limit paths |
1784 1785 1786 |
ep_tovisit = epi->ffd.file->private_data; if (ep_tovisit->visited) continue; |
22bacca48 epoll: prevent cr... |
1787 |
error = ep_call_nested(&poll_loop_ncalls, EP_MAX_NESTS, |
28d82dc1c epoll: limit paths |
1788 1789 |
ep_loop_check_proc, epi->ffd.file, ep_tovisit, current); |
22bacca48 epoll: prevent cr... |
1790 1791 |
if (error != 0) break; |
28d82dc1c epoll: limit paths |
1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 |
} 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 epoll: prevent cr... |
1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 |
} } 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 epoll: limit paths |
1824 1825 1826 1827 |
int ret; struct eventpoll *ep_cur, *ep_next; ret = ep_call_nested(&poll_loop_ncalls, EP_MAX_NESTS, |
22bacca48 epoll: prevent cr... |
1828 |
ep_loop_check_proc, file, ep, current); |
28d82dc1c epoll: limit paths |
1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 |
/* 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 epoll: prevent cr... |
1849 |
} |
7699acd13 epoll cleanups: e... |
1850 |
/* |
523723bb5 fs/eventpoll.c: f... |
1851 |
* Open an eventpoll file descriptor. |
7699acd13 epoll cleanups: e... |
1852 |
*/ |
791eb22ee fs: add do_epoll_... |
1853 |
static int do_epoll_create(int flags) |
7699acd13 epoll cleanups: e... |
1854 |
{ |
28d82dc1c epoll: limit paths |
1855 |
int error, fd; |
bb57c3edc epoll: remove deb... |
1856 |
struct eventpoll *ep = NULL; |
28d82dc1c epoll: limit paths |
1857 |
struct file *file; |
7699acd13 epoll cleanups: e... |
1858 |
|
e38b36f32 flag parameters: ... |
1859 1860 |
/* Check the EPOLL_* constant for consistency. */ BUILD_BUG_ON(EPOLL_CLOEXEC != O_CLOEXEC); |
296e236e9 epoll: fix epoll'... |
1861 1862 |
if (flags & ~EPOLL_CLOEXEC) return -EINVAL; |
7699acd13 epoll cleanups: e... |
1863 |
/* |
bb57c3edc epoll: remove deb... |
1864 |
* Create the internal data structure ("struct eventpoll"). |
7699acd13 epoll cleanups: e... |
1865 |
*/ |
9fe5ad9c8 flag parameters a... |
1866 |
error = ep_alloc(&ep); |
bb57c3edc epoll: remove deb... |
1867 1868 |
if (error < 0) return error; |
7699acd13 epoll cleanups: e... |
1869 1870 |
/* * Creates all the items needed to setup an eventpoll file. That is, |
2030a42ce [PATCH] sanitize ... |
1871 |
* a file structure and a free file descriptor. |
7699acd13 epoll cleanups: e... |
1872 |
*/ |
28d82dc1c epoll: limit paths |
1873 1874 1875 1876 1877 1878 |
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 anonfd: Allow mak... |
1879 |
O_RDWR | (flags & O_CLOEXEC)); |
28d82dc1c epoll: limit paths |
1880 1881 1882 1883 |
if (IS_ERR(file)) { error = PTR_ERR(file); goto out_free_fd; } |
28d82dc1c epoll: limit paths |
1884 |
ep->file = file; |
98022748f eventpoll: use-af... |
1885 |
fd_install(fd, file); |
28d82dc1c epoll: limit paths |
1886 1887 1888 1889 1890 1891 |
return fd; out_free_fd: put_unused_fd(fd); out_free_ep: ep_free(ep); |
bb57c3edc epoll: remove deb... |
1892 |
return error; |
7699acd13 epoll cleanups: e... |
1893 |
} |
791eb22ee fs: add do_epoll_... |
1894 1895 1896 1897 |
SYSCALL_DEFINE1(epoll_create1, int, flags) { return do_epoll_create(flags); } |
5a8a82b1d [CVE-2009-0029] S... |
1898 |
SYSCALL_DEFINE1(epoll_create, int, size) |
a0998b50c flag parameters: ... |
1899 |
{ |
bfe3891a5 epoll: fix size c... |
1900 |
if (size <= 0) |
9fe5ad9c8 flag parameters a... |
1901 |
return -EINVAL; |
791eb22ee fs: add do_epoll_... |
1902 |
return do_epoll_create(0); |
a0998b50c flag parameters: ... |
1903 |
} |
7699acd13 epoll cleanups: e... |
1904 1905 1906 |
/* * The following function implements the controller interface for * the eventpoll file that enables the insertion/removal/change of |
67647d0fb epoll: fix some c... |
1907 |
* file descriptors inside the interest set. |
7699acd13 epoll cleanups: e... |
1908 |
*/ |
5a8a82b1d [CVE-2009-0029] S... |
1909 1910 |
SYSCALL_DEFINE4(epoll_ctl, int, epfd, int, op, int, fd, struct epoll_event __user *, event) |
7699acd13 epoll cleanups: e... |
1911 1912 |
{ int error; |
67347fe4e epoll: do not tak... |
1913 |
int full_check = 0; |
7e3fb5842 switch epoll_ctl(... |
1914 |
struct fd f, tf; |
7699acd13 epoll cleanups: e... |
1915 1916 1917 |
struct eventpoll *ep; struct epitem *epi; struct epoll_event epds; |
67347fe4e epoll: do not tak... |
1918 |
struct eventpoll *tep = NULL; |
7699acd13 epoll cleanups: e... |
1919 |
|
7699acd13 epoll cleanups: e... |
1920 1921 1922 1923 |
error = -EFAULT; if (ep_op_has_event(op) && copy_from_user(&epds, event, sizeof(struct epoll_event))) goto error_return; |
7699acd13 epoll cleanups: e... |
1924 |
error = -EBADF; |
7e3fb5842 switch epoll_ctl(... |
1925 1926 |
f = fdget(epfd); if (!f.file) |
7699acd13 epoll cleanups: e... |
1927 1928 1929 |
goto error_return; /* Get the "struct file *" for the target file */ |
7e3fb5842 switch epoll_ctl(... |
1930 1931 |
tf = fdget(fd); if (!tf.file) |
7699acd13 epoll cleanups: e... |
1932 1933 1934 1935 |
goto error_fput; /* The target file descriptor must support poll */ error = -EPERM; |
9965ed174 fs: add new vfs_p... |
1936 |
if (!file_can_poll(tf.file)) |
7699acd13 epoll cleanups: e... |
1937 |
goto error_tgt_fput; |
4d7e30d98 epoll: Add a flag... |
1938 |
/* Check if EPOLLWAKEUP is allowed */ |
c680e41b3 eventpoll: fix un... |
1939 1940 |
if (ep_op_has_event(op)) ep_take_care_of_epollwakeup(&epds); |
4d7e30d98 epoll: Add a flag... |
1941 |
|
7699acd13 epoll cleanups: e... |
1942 1943 1944 1945 1946 1947 |
/* * 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 switch epoll_ctl(... |
1948 |
if (f.file == tf.file || !is_file_epoll(f.file)) |
7699acd13 epoll cleanups: e... |
1949 1950 1951 |
goto error_tgt_fput; /* |
df0108c5d epoll: add EPOLLE... |
1952 1953 1954 1955 |
* 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. */ |
c857ab640 fs,eventpoll: don... |
1956 |
if (ep_op_has_event(op) && (epds.events & EPOLLEXCLUSIVE)) { |
b6a515c8a epoll: restrict E... |
1957 1958 1959 1960 1961 1962 |
if (op == EPOLL_CTL_MOD) goto error_tgt_fput; if (op == EPOLL_CTL_ADD && (is_file_epoll(tf.file) || (epds.events & ~EPOLLEXCLUSIVE_OK_BITS))) goto error_tgt_fput; } |
df0108c5d epoll: add EPOLLE... |
1963 1964 |
/* |
7699acd13 epoll cleanups: e... |
1965 1966 1967 |
* At this point it is safe to assume that the "private_data" contains * our own data structure. */ |
7e3fb5842 switch epoll_ctl(... |
1968 |
ep = f.file->private_data; |
7699acd13 epoll cleanups: e... |
1969 |
|
22bacca48 epoll: prevent cr... |
1970 1971 1972 |
/* * When we insert an epoll file descriptor, inside another epoll file * descriptor, there is the change of creating closed loops, which are |
28d82dc1c epoll: limit paths |
1973 1974 1975 1976 |
* 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 epoll: prevent cr... |
1977 |
* |
67347fe4e epoll: do not tak... |
1978 1979 1980 1981 1982 1983 |
* 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 epoll: prevent cr... |
1984 |
*/ |
67347fe4e epoll: do not tak... |
1985 |
mutex_lock_nested(&ep->mtx, 0); |
28d82dc1c epoll: limit paths |
1986 |
if (op == EPOLL_CTL_ADD) { |
67347fe4e epoll: do not tak... |
1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 |
if (!list_empty(&f.file->f_ep_links) || is_file_epoll(tf.file)) { full_check = 1; mutex_unlock(&ep->mtx); mutex_lock(&epmutex); 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); mutex_lock_nested(&ep->mtx, 0); if (is_file_epoll(tf.file)) { tep = tf.file->private_data; mutex_lock_nested(&tep->mtx, 1); |
13d518074 epoll: clear the ... |
2005 |
} |
67347fe4e epoll: do not tak... |
2006 2007 |
} } |
7699acd13 epoll cleanups: e... |
2008 |
|
67647d0fb epoll: fix some c... |
2009 2010 2011 2012 2013 |
/* * 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 switch epoll_ctl(... |
2014 |
epi = ep_find(ep, tf.file, fd); |
7699acd13 epoll cleanups: e... |
2015 2016 2017 2018 2019 |
error = -EINVAL; switch (op) { case EPOLL_CTL_ADD: if (!epi) { |
a9a08845e vfs: do bulk POLL... |
2020 |
epds.events |= EPOLLERR | EPOLLHUP; |
67347fe4e epoll: do not tak... |
2021 |
error = ep_insert(ep, &epds, tf.file, fd, full_check); |
7699acd13 epoll cleanups: e... |
2022 2023 |
} else error = -EEXIST; |
67347fe4e epoll: do not tak... |
2024 2025 |
if (full_check) clear_tfile_check_list(); |
7699acd13 epoll cleanups: e... |
2026 2027 2028 2029 2030 2031 2032 2033 2034 |
break; case EPOLL_CTL_DEL: if (epi) error = ep_remove(ep, epi); else error = -ENOENT; break; case EPOLL_CTL_MOD: if (epi) { |
b6a515c8a epoll: restrict E... |
2035 |
if (!(epi->event.events & EPOLLEXCLUSIVE)) { |
a9a08845e vfs: do bulk POLL... |
2036 |
epds.events |= EPOLLERR | EPOLLHUP; |
b6a515c8a epoll: restrict E... |
2037 2038 |
error = ep_modify(ep, epi, &epds); } |
7699acd13 epoll cleanups: e... |
2039 2040 2041 2042 |
} else error = -ENOENT; break; } |
67347fe4e epoll: do not tak... |
2043 2044 |
if (tep != NULL) mutex_unlock(&tep->mtx); |
d47de16c7 fix epoll single ... |
2045 |
mutex_unlock(&ep->mtx); |
7699acd13 epoll cleanups: e... |
2046 2047 |
error_tgt_fput: |
67347fe4e epoll: do not tak... |
2048 |
if (full_check) |
22bacca48 epoll: prevent cr... |
2049 |
mutex_unlock(&epmutex); |
7e3fb5842 switch epoll_ctl(... |
2050 |
fdput(tf); |
7699acd13 epoll cleanups: e... |
2051 |
error_fput: |
7e3fb5842 switch epoll_ctl(... |
2052 |
fdput(f); |
7699acd13 epoll cleanups: e... |
2053 |
error_return: |
7699acd13 epoll cleanups: e... |
2054 2055 2056 2057 2058 2059 2060 2061 |
return error; } /* * Implement the event wait interface for the eventpoll file. It is the kernel * part of the user space epoll_wait(2). */ |
791eb22ee fs: add do_epoll_... |
2062 2063 |
static int do_epoll_wait(int epfd, struct epoll_event __user *events, int maxevents, int timeout) |
7699acd13 epoll cleanups: e... |
2064 |
{ |
2903ff019 switch simple cas... |
2065 2066 |
int error; struct fd f; |
7699acd13 epoll cleanups: e... |
2067 |
struct eventpoll *ep; |
7699acd13 epoll cleanups: e... |
2068 2069 2070 2071 2072 |
/* 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 */ |
2903ff019 switch simple cas... |
2073 2074 |
if (!access_ok(VERIFY_WRITE, events, maxevents * sizeof(struct epoll_event))) return -EFAULT; |
7699acd13 epoll cleanups: e... |
2075 2076 |
/* Get the "struct file *" for the eventpoll file */ |
2903ff019 switch simple cas... |
2077 2078 2079 |
f = fdget(epfd); if (!f.file) return -EBADF; |
7699acd13 epoll cleanups: e... |
2080 2081 2082 2083 2084 2085 |
/* * We have to check that the file structure underneath the fd * the user passed to us _is_ an eventpoll file. */ error = -EINVAL; |
2903ff019 switch simple cas... |
2086 |
if (!is_file_epoll(f.file)) |
7699acd13 epoll cleanups: e... |
2087 2088 2089 2090 2091 2092 |
goto error_fput; /* * At this point it is safe to assume that the "private_data" contains * our own data structure. */ |
2903ff019 switch simple cas... |
2093 |
ep = f.file->private_data; |
7699acd13 epoll cleanups: e... |
2094 2095 2096 2097 2098 |
/* Time to fish for events ... */ error = ep_poll(ep, events, maxevents, timeout); error_fput: |
2903ff019 switch simple cas... |
2099 |
fdput(f); |
7699acd13 epoll cleanups: e... |
2100 2101 |
return error; } |
791eb22ee fs: add do_epoll_... |
2102 2103 2104 2105 2106 |
SYSCALL_DEFINE4(epoll_wait, int, epfd, struct epoll_event __user *, events, int, maxevents, int, timeout) { return do_epoll_wait(epfd, events, maxevents, timeout); } |
7699acd13 epoll cleanups: e... |
2107 2108 2109 2110 |
/* * Implement the event wait interface for the eventpoll file. It is the kernel * part of the user space epoll_pwait(2). */ |
5a8a82b1d [CVE-2009-0029] S... |
2111 2112 2113 |
SYSCALL_DEFINE6(epoll_pwait, int, epfd, struct epoll_event __user *, events, int, maxevents, int, timeout, const sigset_t __user *, sigmask, size_t, sigsetsize) |
7699acd13 epoll cleanups: e... |
2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 |
{ int error; sigset_t ksigmask, sigsaved; /* * If the caller wants a certain signal mask to be set during the wait, * we apply it here. */ if (sigmask) { if (sigsetsize != sizeof(sigset_t)) return -EINVAL; if (copy_from_user(&ksigmask, sigmask, sizeof(ksigmask))) return -EFAULT; |
77d559180 signals: eventpol... |
2127 2128 |
sigsaved = current->blocked; set_current_blocked(&ksigmask); |
7699acd13 epoll cleanups: e... |
2129 |
} |
791eb22ee fs: add do_epoll_... |
2130 |
error = do_epoll_wait(epfd, events, maxevents, timeout); |
7699acd13 epoll cleanups: e... |
2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 |
/* * If we changed the signal mask, we need to restore the original one. * In case we've got a signal while waiting, we do not restore the * signal mask yet, and we allow do_signal() to deliver the signal on * the way back to userspace, before the signal mask is restored. */ if (sigmask) { if (error == -EINTR) { memcpy(¤t->saved_sigmask, &sigsaved, |
c7ea76302 epoll locks chang... |
2141 |
sizeof(sigsaved)); |
4e4c22c71 signals: add set_... |
2142 |
set_restore_sigmask(); |
7699acd13 epoll cleanups: e... |
2143 |
} else |
77d559180 signals: eventpol... |
2144 |
set_current_blocked(&sigsaved); |
7699acd13 epoll cleanups: e... |
2145 2146 2147 2148 |
} return error; } |
35280bd4a switch epoll_pwai... |
2149 2150 2151 2152 2153 2154 2155 2156 |
#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 switch epoll_pwai... |
2157 2158 2159 2160 2161 2162 2163 2164 2165 |
sigset_t ksigmask, sigsaved; /* * If the caller wants a certain signal mask to be set during the wait, * we apply it here. */ if (sigmask) { if (sigsetsize != sizeof(compat_sigset_t)) return -EINVAL; |
3968cf623 get_compat_sigset() |
2166 |
if (get_compat_sigset(&ksigmask, sigmask)) |
35280bd4a switch epoll_pwai... |
2167 |
return -EFAULT; |
77d559180 signals: eventpol... |
2168 2169 |
sigsaved = current->blocked; set_current_blocked(&ksigmask); |
35280bd4a switch epoll_pwai... |
2170 |
} |
791eb22ee fs: add do_epoll_... |
2171 |
err = do_epoll_wait(epfd, events, maxevents, timeout); |
35280bd4a switch epoll_pwai... |
2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 |
/* * If we changed the signal mask, we need to restore the original one. * In case we've got a signal while waiting, we do not restore the * signal mask yet, and we allow do_signal() to deliver the signal on * the way back to userspace, before the signal mask is restored. */ if (sigmask) { if (err == -EINTR) { memcpy(¤t->saved_sigmask, &sigsaved, sizeof(sigsaved)); set_restore_sigmask(); } else |
77d559180 signals: eventpol... |
2185 |
set_current_blocked(&sigsaved); |
35280bd4a switch epoll_pwai... |
2186 2187 2188 2189 2190 |
} return err; } #endif |
1da177e4c Linux-2.6.12-rc2 |
2191 2192 |
static int __init eventpoll_init(void) { |
7ef9964e6 epoll: introduce ... |
2193 2194 2195 |
struct sysinfo si; si_meminfo(&si); |
9df04e1f2 epoll: drop max_u... |
2196 2197 2198 2199 |
/* * Allows top 4% of lomem to be allocated for epoll watches (per user). */ max_user_watches = (((si.totalram - si.totalhigh) / 25) << PAGE_SHIFT) / |
7ef9964e6 epoll: introduce ... |
2200 |
EP_ITEM_COST; |
52bd19f76 epoll: convert ma... |
2201 |
BUG_ON(max_user_watches < 0); |
1da177e4c Linux-2.6.12-rc2 |
2202 |
|
22bacca48 epoll: prevent cr... |
2203 2204 2205 2206 2207 |
/* * Initialize the structure used to perform epoll file descriptor * inclusion loops checks. */ ep_nested_calls_init(&poll_loop_ncalls); |
57a173bdf epoll: avoid call... |
2208 |
#ifdef CONFIG_DEBUG_LOCK_ALLOC |
1da177e4c Linux-2.6.12-rc2 |
2209 |
/* Initialize the structure used to perform safe poll wait head wake ups */ |
5071f97ec epoll: fix epoll'... |
2210 |
ep_nested_calls_init(&poll_safewake_ncalls); |
57a173bdf epoll: avoid call... |
2211 |
#endif |
1da177e4c Linux-2.6.12-rc2 |
2212 |
|
39732ca5a epoll: trim epite... |
2213 2214 2215 2216 2217 |
/* * 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 Linux-2.6.12-rc2 |
2218 2219 |
/* Allocates slab cache used to allocate "struct epitem" items */ epi_cache = kmem_cache_create("eventpoll_epi", sizeof(struct epitem), |
2ae928a94 epoll: account ep... |
2220 |
0, SLAB_HWCACHE_ALIGN|SLAB_PANIC|SLAB_ACCOUNT, NULL); |
1da177e4c Linux-2.6.12-rc2 |
2221 2222 2223 |
/* Allocates slab cache used to allocate "struct eppoll_entry" */ pwq_cache = kmem_cache_create("eventpoll_pwq", |
2ae928a94 epoll: account ep... |
2224 |
sizeof(struct eppoll_entry), 0, SLAB_PANIC|SLAB_ACCOUNT, NULL); |
1da177e4c Linux-2.6.12-rc2 |
2225 |
|
1da177e4c Linux-2.6.12-rc2 |
2226 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
2227 |
} |
cea692418 epoll cleanups: e... |
2228 |
fs_initcall(eventpoll_init); |