Blame view
fs/eventfd.c
10.8 KB
e1ad7468c signal/timer/even... |
1 2 3 4 5 6 7 8 9 10 11 |
/* * fs/eventfd.c * * Copyright (C) 2007 Davide Libenzi <davidel@xmailserver.org> * */ #include <linux/file.h> #include <linux/poll.h> #include <linux/init.h> #include <linux/fs.h> |
174cd4b1e sched/headers: Pr... |
12 |
#include <linux/sched/signal.h> |
e1ad7468c signal/timer/even... |
13 |
#include <linux/kernel.h> |
5a0e3ad6a include cleanup: ... |
14 |
#include <linux/slab.h> |
e1ad7468c signal/timer/even... |
15 16 17 |
#include <linux/list.h> #include <linux/spinlock.h> #include <linux/anon_inodes.h> |
7747cdb2f fs/eventfd.c shou... |
18 |
#include <linux/syscalls.h> |
630d9c472 fs: reduce the us... |
19 |
#include <linux/export.h> |
133890103 eventfd: revised ... |
20 21 |
#include <linux/kref.h> #include <linux/eventfd.h> |
cbac5542d fs, eventfd: add ... |
22 23 |
#include <linux/proc_fs.h> #include <linux/seq_file.h> |
e1ad7468c signal/timer/even... |
24 25 |
struct eventfd_ctx { |
133890103 eventfd: revised ... |
26 |
struct kref kref; |
e1ad7468c signal/timer/even... |
27 28 29 30 31 32 |
wait_queue_head_t wqh; /* * Every time that a write(2) is performed on an eventfd, the * value of the __u64 being written is added to "count" and a * wakeup is performed on "wqh". A read(2) will return the "count" * value to userspace, and will reset "count" to zero. The kernel |
133890103 eventfd: revised ... |
33 |
* side eventfd_signal() also, adds to the "count" counter and |
e1ad7468c signal/timer/even... |
34 35 36 |
* issue a wakeup. */ __u64 count; |
bcd0b235b eventfd: improve ... |
37 |
unsigned int flags; |
e1ad7468c signal/timer/even... |
38 |
}; |
133890103 eventfd: revised ... |
39 40 41 42 43 44 45 46 |
/** * eventfd_signal - Adds @n to the eventfd counter. * @ctx: [in] Pointer to the eventfd context. * @n: [in] Value of the counter to be added to the eventfd internal counter. * The value cannot be negative. * * This function is supposed to be called by the kernel in paths that do not * allow sleeping. In this function we allow the counter to reach the ULLONG_MAX |
a9a08845e vfs: do bulk POLL... |
47 |
* value, and we signal this as overflow condition by returning a EPOLLERR |
133890103 eventfd: revised ... |
48 49 |
* to poll(2). * |
20d5a865e Documentation: fi... |
50 |
* Returns the amount by which the counter was incremented. This will be less |
ee62c6b2d eventfd: change i... |
51 |
* than @n if the counter has overflowed. |
e1ad7468c signal/timer/even... |
52 |
*/ |
ee62c6b2d eventfd: change i... |
53 |
__u64 eventfd_signal(struct eventfd_ctx *ctx, __u64 n) |
e1ad7468c signal/timer/even... |
54 |
{ |
e1ad7468c signal/timer/even... |
55 |
unsigned long flags; |
d48eb2331 eventfd use waitq... |
56 |
spin_lock_irqsave(&ctx->wqh.lock, flags); |
e1ad7468c signal/timer/even... |
57 |
if (ULLONG_MAX - ctx->count < n) |
ee62c6b2d eventfd: change i... |
58 |
n = ULLONG_MAX - ctx->count; |
e1ad7468c signal/timer/even... |
59 60 |
ctx->count += n; if (waitqueue_active(&ctx->wqh)) |
a9a08845e vfs: do bulk POLL... |
61 |
wake_up_locked_poll(&ctx->wqh, EPOLLIN); |
d48eb2331 eventfd use waitq... |
62 |
spin_unlock_irqrestore(&ctx->wqh.lock, flags); |
e1ad7468c signal/timer/even... |
63 64 65 |
return n; } |
5718607bb eventfd: export e... |
66 |
EXPORT_SYMBOL_GPL(eventfd_signal); |
e1ad7468c signal/timer/even... |
67 |
|
562787a5c anonfd: split int... |
68 69 70 71 |
static void eventfd_free_ctx(struct eventfd_ctx *ctx) { kfree(ctx); } |
133890103 eventfd: revised ... |
72 73 74 |
static void eventfd_free(struct kref *kref) { struct eventfd_ctx *ctx = container_of(kref, struct eventfd_ctx, kref); |
562787a5c anonfd: split int... |
75 |
eventfd_free_ctx(ctx); |
133890103 eventfd: revised ... |
76 77 78 |
} /** |
133890103 eventfd: revised ... |
79 80 81 82 |
* eventfd_ctx_put - Releases a reference to the internal eventfd context. * @ctx: [in] Pointer to eventfd context. * * The eventfd context reference must have been previously acquired either |
105f2b709 eventfd: fold eve... |
83 |
* with eventfd_ctx_fdget() or eventfd_ctx_fileget(). |
133890103 eventfd: revised ... |
84 85 86 87 88 89 |
*/ void eventfd_ctx_put(struct eventfd_ctx *ctx) { kref_put(&ctx->kref, eventfd_free); } EXPORT_SYMBOL_GPL(eventfd_ctx_put); |
e1ad7468c signal/timer/even... |
90 91 |
static int eventfd_release(struct inode *inode, struct file *file) { |
133890103 eventfd: revised ... |
92 |
struct eventfd_ctx *ctx = file->private_data; |
a9a08845e vfs: do bulk POLL... |
93 |
wake_up_poll(&ctx->wqh, EPOLLHUP); |
133890103 eventfd: revised ... |
94 |
eventfd_ctx_put(ctx); |
e1ad7468c signal/timer/even... |
95 96 |
return 0; } |
a11e1d432 Revert changes to... |
97 |
static __poll_t eventfd_poll(struct file *file, poll_table *wait) |
e1ad7468c signal/timer/even... |
98 99 |
{ struct eventfd_ctx *ctx = file->private_data; |
076ccb76e fs: annotate ->po... |
100 |
__poll_t events = 0; |
e22553e2a eventfd: don't ta... |
101 |
u64 count; |
e1ad7468c signal/timer/even... |
102 |
|
a11e1d432 Revert changes to... |
103 |
poll_wait(file, &ctx->wqh, wait); |
a484c3dd9 eventfd: document... |
104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 |
/* * All writes to ctx->count occur within ctx->wqh.lock. This read * can be done outside ctx->wqh.lock because we know that poll_wait * takes that lock (through add_wait_queue) if our caller will sleep. * * The read _can_ therefore seep into add_wait_queue's critical * section, but cannot move above it! add_wait_queue's spin_lock acts * as an acquire barrier and ensures that the read be ordered properly * against the writes. The following CAN happen and is safe: * * poll write * ----------------- ------------ * lock ctx->wqh.lock (in poll_wait) * count = ctx->count * __add_wait_queue * unlock ctx->wqh.lock * lock ctx->qwh.lock * ctx->count += n * if (waitqueue_active) * wake_up_locked_poll * unlock ctx->qwh.lock * eventfd_poll returns 0 * * but the following, which would miss a wakeup, cannot happen: * * poll write * ----------------- ------------ * count = ctx->count (INVALID!) * lock ctx->qwh.lock * ctx->count += n * **waitqueue_active is false** * **no wake_up_locked_poll!** * unlock ctx->qwh.lock * lock ctx->wqh.lock (in poll_wait) * __add_wait_queue * unlock ctx->wqh.lock * eventfd_poll returns 0 */ count = READ_ONCE(ctx->count); |
e1ad7468c signal/timer/even... |
143 |
|
e22553e2a eventfd: don't ta... |
144 |
if (count > 0) |
a11e1d432 Revert changes to... |
145 |
events |= EPOLLIN; |
e22553e2a eventfd: don't ta... |
146 |
if (count == ULLONG_MAX) |
a9a08845e vfs: do bulk POLL... |
147 |
events |= EPOLLERR; |
e22553e2a eventfd: don't ta... |
148 |
if (ULLONG_MAX - 1 > count) |
a11e1d432 Revert changes to... |
149 |
events |= EPOLLOUT; |
e1ad7468c signal/timer/even... |
150 151 152 |
return events; } |
cb289d624 eventfd - allow a... |
153 154 155 156 157 158 159 160 161 162 |
static void eventfd_ctx_do_read(struct eventfd_ctx *ctx, __u64 *cnt) { *cnt = (ctx->flags & EFD_SEMAPHORE) ? 1 : ctx->count; ctx->count -= *cnt; } /** * eventfd_ctx_remove_wait_queue - Read the current counter and removes wait queue. * @ctx: [in] Pointer to eventfd context. * @wait: [in] Wait queue to be removed. |
361821854 Docbook: add fs/e... |
163 |
* @cnt: [out] Pointer to the 64-bit counter value. |
cb289d624 eventfd - allow a... |
164 |
* |
361821854 Docbook: add fs/e... |
165 |
* Returns %0 if successful, or the following error codes: |
cb289d624 eventfd - allow a... |
166 167 168 169 170 171 |
* * -EAGAIN : The operation would have blocked. * * This is used to atomically remove a wait queue entry from the eventfd wait * queue head, and read/reset the counter value. */ |
ac6424b98 sched/wait: Renam... |
172 |
int eventfd_ctx_remove_wait_queue(struct eventfd_ctx *ctx, wait_queue_entry_t *wait, |
cb289d624 eventfd - allow a... |
173 174 175 176 177 178 179 180 |
__u64 *cnt) { unsigned long flags; spin_lock_irqsave(&ctx->wqh.lock, flags); eventfd_ctx_do_read(ctx, cnt); __remove_wait_queue(&ctx->wqh, wait); if (*cnt != 0 && waitqueue_active(&ctx->wqh)) |
a9a08845e vfs: do bulk POLL... |
181 |
wake_up_locked_poll(&ctx->wqh, EPOLLOUT); |
cb289d624 eventfd - allow a... |
182 183 184 185 186 |
spin_unlock_irqrestore(&ctx->wqh.lock, flags); return *cnt != 0 ? 0 : -EAGAIN; } EXPORT_SYMBOL_GPL(eventfd_ctx_remove_wait_queue); |
b6364572d eventfd: fold eve... |
187 188 |
static ssize_t eventfd_read(struct file *file, char __user *buf, size_t count, loff_t *ppos) |
e1ad7468c signal/timer/even... |
189 |
{ |
b6364572d eventfd: fold eve... |
190 |
struct eventfd_ctx *ctx = file->private_data; |
e1ad7468c signal/timer/even... |
191 |
ssize_t res; |
b6364572d eventfd: fold eve... |
192 |
__u64 ucnt = 0; |
e1ad7468c signal/timer/even... |
193 |
DECLARE_WAITQUEUE(wait, current); |
b6364572d eventfd: fold eve... |
194 195 |
if (count < sizeof(ucnt)) return -EINVAL; |
d48eb2331 eventfd use waitq... |
196 |
spin_lock_irq(&ctx->wqh.lock); |
e1ad7468c signal/timer/even... |
197 |
res = -EAGAIN; |
bcd0b235b eventfd: improve ... |
198 |
if (ctx->count > 0) |
b6364572d eventfd: fold eve... |
199 200 |
res = sizeof(ucnt); else if (!(file->f_flags & O_NONBLOCK)) { |
e1ad7468c signal/timer/even... |
201 |
__add_wait_queue(&ctx->wqh, &wait); |
cb289d624 eventfd - allow a... |
202 |
for (;;) { |
e1ad7468c signal/timer/even... |
203 204 |
set_current_state(TASK_INTERRUPTIBLE); if (ctx->count > 0) { |
b6364572d eventfd: fold eve... |
205 |
res = sizeof(ucnt); |
e1ad7468c signal/timer/even... |
206 207 208 209 210 211 |
break; } if (signal_pending(current)) { res = -ERESTARTSYS; break; } |
d48eb2331 eventfd use waitq... |
212 |
spin_unlock_irq(&ctx->wqh.lock); |
e1ad7468c signal/timer/even... |
213 |
schedule(); |
d48eb2331 eventfd use waitq... |
214 |
spin_lock_irq(&ctx->wqh.lock); |
e1ad7468c signal/timer/even... |
215 216 217 218 |
} __remove_wait_queue(&ctx->wqh, &wait); __set_current_state(TASK_RUNNING); } |
b6364572d eventfd: fold eve... |
219 220 |
if (likely(res > 0)) { eventfd_ctx_do_read(ctx, &ucnt); |
e1ad7468c signal/timer/even... |
221 |
if (waitqueue_active(&ctx->wqh)) |
a9a08845e vfs: do bulk POLL... |
222 |
wake_up_locked_poll(&ctx->wqh, EPOLLOUT); |
e1ad7468c signal/timer/even... |
223 |
} |
d48eb2331 eventfd use waitq... |
224 |
spin_unlock_irq(&ctx->wqh.lock); |
e1ad7468c signal/timer/even... |
225 |
|
b6364572d eventfd: fold eve... |
226 227 |
if (res > 0 && put_user(ucnt, (__u64 __user *)buf)) return -EFAULT; |
cb289d624 eventfd - allow a... |
228 |
|
b6364572d eventfd: fold eve... |
229 |
return res; |
cb289d624 eventfd - allow a... |
230 |
} |
e1ad7468c signal/timer/even... |
231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 |
static ssize_t eventfd_write(struct file *file, const char __user *buf, size_t count, loff_t *ppos) { struct eventfd_ctx *ctx = file->private_data; ssize_t res; __u64 ucnt; DECLARE_WAITQUEUE(wait, current); if (count < sizeof(ucnt)) return -EINVAL; if (copy_from_user(&ucnt, buf, sizeof(ucnt))) return -EFAULT; if (ucnt == ULLONG_MAX) return -EINVAL; |
d48eb2331 eventfd use waitq... |
246 |
spin_lock_irq(&ctx->wqh.lock); |
e1ad7468c signal/timer/even... |
247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 |
res = -EAGAIN; if (ULLONG_MAX - ctx->count > ucnt) res = sizeof(ucnt); else if (!(file->f_flags & O_NONBLOCK)) { __add_wait_queue(&ctx->wqh, &wait); for (res = 0;;) { set_current_state(TASK_INTERRUPTIBLE); if (ULLONG_MAX - ctx->count > ucnt) { res = sizeof(ucnt); break; } if (signal_pending(current)) { res = -ERESTARTSYS; break; } |
d48eb2331 eventfd use waitq... |
262 |
spin_unlock_irq(&ctx->wqh.lock); |
e1ad7468c signal/timer/even... |
263 |
schedule(); |
d48eb2331 eventfd use waitq... |
264 |
spin_lock_irq(&ctx->wqh.lock); |
e1ad7468c signal/timer/even... |
265 266 267 268 |
} __remove_wait_queue(&ctx->wqh, &wait); __set_current_state(TASK_RUNNING); } |
bcd0b235b eventfd: improve ... |
269 |
if (likely(res > 0)) { |
e1ad7468c signal/timer/even... |
270 271 |
ctx->count += ucnt; if (waitqueue_active(&ctx->wqh)) |
a9a08845e vfs: do bulk POLL... |
272 |
wake_up_locked_poll(&ctx->wqh, EPOLLIN); |
e1ad7468c signal/timer/even... |
273 |
} |
d48eb2331 eventfd use waitq... |
274 |
spin_unlock_irq(&ctx->wqh.lock); |
e1ad7468c signal/timer/even... |
275 276 277 |
return res; } |
cbac5542d fs, eventfd: add ... |
278 |
#ifdef CONFIG_PROC_FS |
a3816ab0e fs: Convert show_... |
279 |
static void eventfd_show_fdinfo(struct seq_file *m, struct file *f) |
cbac5542d fs, eventfd: add ... |
280 281 |
{ struct eventfd_ctx *ctx = f->private_data; |
cbac5542d fs, eventfd: add ... |
282 283 |
spin_lock_irq(&ctx->wqh.lock); |
a3816ab0e fs: Convert show_... |
284 285 286 |
seq_printf(m, "eventfd-count: %16llx ", (unsigned long long)ctx->count); |
cbac5542d fs, eventfd: add ... |
287 |
spin_unlock_irq(&ctx->wqh.lock); |
cbac5542d fs, eventfd: add ... |
288 289 |
} #endif |
e1ad7468c signal/timer/even... |
290 |
static const struct file_operations eventfd_fops = { |
cbac5542d fs, eventfd: add ... |
291 292 293 |
#ifdef CONFIG_PROC_FS .show_fdinfo = eventfd_show_fdinfo, #endif |
e1ad7468c signal/timer/even... |
294 |
.release = eventfd_release, |
a11e1d432 Revert changes to... |
295 |
.poll = eventfd_poll, |
e1ad7468c signal/timer/even... |
296 297 |
.read = eventfd_read, .write = eventfd_write, |
6038f373a llseek: automatic... |
298 |
.llseek = noop_llseek, |
e1ad7468c signal/timer/even... |
299 |
}; |
133890103 eventfd: revised ... |
300 301 302 303 304 305 306 307 308 309 |
/** * eventfd_fget - Acquire a reference of an eventfd file descriptor. * @fd: [in] Eventfd file descriptor. * * Returns a pointer to the eventfd file structure in case of success, or the * following error pointer: * * -EBADF : Invalid @fd file descriptor. * -EINVAL : The @fd file descriptor is not an eventfd file. */ |
e1ad7468c signal/timer/even... |
310 311 312 313 314 315 316 317 318 319 320 321 322 323 |
struct file *eventfd_fget(int fd) { struct file *file; file = fget(fd); if (!file) return ERR_PTR(-EBADF); if (file->f_op != &eventfd_fops) { fput(file); return ERR_PTR(-EINVAL); } return file; } |
5718607bb eventfd: export e... |
324 |
EXPORT_SYMBOL_GPL(eventfd_fget); |
e1ad7468c signal/timer/even... |
325 |
|
133890103 eventfd: revised ... |
326 327 328 329 330 331 332 333 334 335 336 |
/** * eventfd_ctx_fdget - Acquires a reference to the internal eventfd context. * @fd: [in] Eventfd file descriptor. * * Returns a pointer to the internal eventfd context, otherwise the error * pointers returned by the following functions: * * eventfd_fget */ struct eventfd_ctx *eventfd_ctx_fdget(int fd) { |
133890103 eventfd: revised ... |
337 |
struct eventfd_ctx *ctx; |
36a741172 eventfd_ctx_fdget... |
338 339 340 341 342 |
struct fd f = fdget(fd); if (!f.file) return ERR_PTR(-EBADF); ctx = eventfd_ctx_fileget(f.file); fdput(f); |
133890103 eventfd: revised ... |
343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 |
return ctx; } EXPORT_SYMBOL_GPL(eventfd_ctx_fdget); /** * eventfd_ctx_fileget - Acquires a reference to the internal eventfd context. * @file: [in] Eventfd file pointer. * * Returns a pointer to the internal eventfd context, otherwise the error * pointer: * * -EINVAL : The @fd file descriptor is not an eventfd file. */ struct eventfd_ctx *eventfd_ctx_fileget(struct file *file) { |
105f2b709 eventfd: fold eve... |
358 |
struct eventfd_ctx *ctx; |
133890103 eventfd: revised ... |
359 360 |
if (file->f_op != &eventfd_fops) return ERR_PTR(-EINVAL); |
105f2b709 eventfd: fold eve... |
361 362 363 |
ctx = file->private_data; kref_get(&ctx->kref); return ctx; |
133890103 eventfd: revised ... |
364 365 |
} EXPORT_SYMBOL_GPL(eventfd_ctx_fileget); |
2fc96f833 fs: add do_eventf... |
366 |
static int do_eventfd(unsigned int count, int flags) |
e1ad7468c signal/timer/even... |
367 |
{ |
e1ad7468c signal/timer/even... |
368 |
struct eventfd_ctx *ctx; |
7d815165c eventfd: convert ... |
369 |
int fd; |
e1ad7468c signal/timer/even... |
370 |
|
e38b36f32 flag parameters: ... |
371 372 373 |
/* Check the EFD_* constants for consistency. */ BUILD_BUG_ON(EFD_CLOEXEC != O_CLOEXEC); BUILD_BUG_ON(EFD_NONBLOCK != O_NONBLOCK); |
bcd0b235b eventfd: improve ... |
374 |
if (flags & ~EFD_FLAGS_SET) |
7d815165c eventfd: convert ... |
375 |
return -EINVAL; |
b087498eb flag parameters: ... |
376 |
|
e1ad7468c signal/timer/even... |
377 378 |
ctx = kmalloc(sizeof(*ctx), GFP_KERNEL); if (!ctx) |
7d815165c eventfd: convert ... |
379 |
return -ENOMEM; |
e1ad7468c signal/timer/even... |
380 |
|
133890103 eventfd: revised ... |
381 |
kref_init(&ctx->kref); |
e1ad7468c signal/timer/even... |
382 |
init_waitqueue_head(&ctx->wqh); |
e1ad7468c signal/timer/even... |
383 |
ctx->count = count; |
bcd0b235b eventfd: improve ... |
384 |
ctx->flags = flags; |
e1ad7468c signal/timer/even... |
385 |
|
7d815165c eventfd: convert ... |
386 387 388 |
fd = anon_inode_getfd("[eventfd]", &eventfd_fops, ctx, O_RDWR | (flags & EFD_SHARED_FCNTL_FLAGS)); if (fd < 0) |
562787a5c anonfd: split int... |
389 |
eventfd_free_ctx(ctx); |
2030a42ce [PATCH] sanitize ... |
390 |
return fd; |
e1ad7468c signal/timer/even... |
391 |
} |
2fc96f833 fs: add do_eventf... |
392 393 394 395 |
SYSCALL_DEFINE2(eventfd2, unsigned int, count, int, flags) { return do_eventfd(count, flags); } |
d4e82042c [CVE-2009-0029] S... |
396 |
SYSCALL_DEFINE1(eventfd, unsigned int, count) |
b087498eb flag parameters: ... |
397 |
{ |
2fc96f833 fs: add do_eventf... |
398 |
return do_eventfd(count, 0); |
b087498eb flag parameters: ... |
399 |
} |
bcd0b235b eventfd: improve ... |
400 |