Blame view
fs/pipe.c
27.3 KB
1da177e4c
|
1 2 3 4 5 6 7 8 9 10 11 12 13 |
/* * linux/fs/pipe.c * * Copyright (C) 1991, 1992, 1999 Linus Torvalds */ #include <linux/mm.h> #include <linux/file.h> #include <linux/poll.h> #include <linux/slab.h> #include <linux/module.h> #include <linux/init.h> #include <linux/fs.h> |
35f3d14db
|
14 |
#include <linux/log2.h> |
1da177e4c
|
15 |
#include <linux/mount.h> |
b502bd115
|
16 |
#include <linux/magic.h> |
1da177e4c
|
17 18 19 |
#include <linux/pipe_fs_i.h> #include <linux/uio.h> #include <linux/highmem.h> |
5274f052e
|
20 |
#include <linux/pagemap.h> |
db3495099
|
21 |
#include <linux/audit.h> |
ba719baea
|
22 |
#include <linux/syscalls.h> |
b492e95be
|
23 |
#include <linux/fcntl.h> |
d86133bd3
|
24 |
#include <linux/memcontrol.h> |
1da177e4c
|
25 |
|
7c0f6ba68
|
26 |
#include <linux/uaccess.h> |
1da177e4c
|
27 |
#include <asm/ioctls.h> |
599a0ac14
|
28 |
#include "internal.h" |
1da177e4c
|
29 |
/* |
b492e95be
|
30 |
* The max size that a non-root user is allowed to grow the pipe. Can |
ff9da691c
|
31 |
* be set by root in /proc/sys/fs/pipe-max-size |
b492e95be
|
32 |
*/ |
ff9da691c
|
33 34 35 36 37 38 |
unsigned int pipe_max_size = 1048576; /* * Minimum pipe size, as required by POSIX */ unsigned int pipe_min_size = PAGE_SIZE; |
b492e95be
|
39 |
|
759c01142
|
40 41 42 43 44 |
/* Maximum allocatable pages per user. Hard limit is unset by default, soft * matches default values. */ unsigned long pipe_user_pages_hard; unsigned long pipe_user_pages_soft = PIPE_DEF_BUFFERS * INR_OPEN_CUR; |
b492e95be
|
45 |
/* |
1da177e4c
|
46 47 48 49 50 51 52 53 54 55 56 57 58 |
* We use a start+len construction, which provides full use of the * allocated memory. * -- Florian Coosmann (FGC) * * Reads with count = 0 should always return 0. * -- Julian Bradfield 1999-06-07. * * FIFOs and Pipes now generate SIGIO for both readers and writers. * -- Jeremy Elson <jelson@circlemud.org> 2001-08-16 * * pipe_read & write cleanup * -- Manfred Spraul <manfred@colorfullife.com> 2002-05-09 */ |
61e0d47c3
|
59 60 |
static void pipe_lock_nested(struct pipe_inode_info *pipe, int subclass) { |
6447a3cf1
|
61 |
if (pipe->files) |
72b0d9aac
|
62 |
mutex_lock_nested(&pipe->mutex, subclass); |
61e0d47c3
|
63 64 65 66 67 68 69 70 71 72 73 74 75 |
} void pipe_lock(struct pipe_inode_info *pipe) { /* * pipe_lock() nests non-pipe inode locks (for writing to a file) */ pipe_lock_nested(pipe, I_MUTEX_PARENT); } EXPORT_SYMBOL(pipe_lock); void pipe_unlock(struct pipe_inode_info *pipe) { |
6447a3cf1
|
76 |
if (pipe->files) |
72b0d9aac
|
77 |
mutex_unlock(&pipe->mutex); |
61e0d47c3
|
78 79 |
} EXPORT_SYMBOL(pipe_unlock); |
ebec73f47
|
80 81 82 83 84 85 86 87 88 |
static inline void __pipe_lock(struct pipe_inode_info *pipe) { mutex_lock_nested(&pipe->mutex, I_MUTEX_PARENT); } static inline void __pipe_unlock(struct pipe_inode_info *pipe) { mutex_unlock(&pipe->mutex); } |
61e0d47c3
|
89 90 91 92 93 94 95 96 97 |
void pipe_double_lock(struct pipe_inode_info *pipe1, struct pipe_inode_info *pipe2) { BUG_ON(pipe1 == pipe2); if (pipe1 < pipe2) { pipe_lock_nested(pipe1, I_MUTEX_PARENT); pipe_lock_nested(pipe2, I_MUTEX_CHILD); } else { |
023d43c7b
|
98 99 |
pipe_lock_nested(pipe2, I_MUTEX_PARENT); pipe_lock_nested(pipe1, I_MUTEX_CHILD); |
61e0d47c3
|
100 101 |
} } |
1da177e4c
|
102 |
/* Drop the inode semaphore and wait for a pipe event, atomically */ |
3a326a2ce
|
103 |
void pipe_wait(struct pipe_inode_info *pipe) |
1da177e4c
|
104 105 |
{ DEFINE_WAIT(wait); |
d79fc0fc6
|
106 107 108 109 |
/* * Pipes are system-local resources, so sleeping on them * is considered a noninteractive wait: */ |
af9272326
|
110 |
prepare_to_wait(&pipe->wait, &wait, TASK_INTERRUPTIBLE); |
61e0d47c3
|
111 |
pipe_unlock(pipe); |
1da177e4c
|
112 |
schedule(); |
3a326a2ce
|
113 |
finish_wait(&pipe->wait, &wait); |
61e0d47c3
|
114 |
pipe_lock(pipe); |
1da177e4c
|
115 |
} |
341b446bc
|
116 117 |
static void anon_pipe_buf_release(struct pipe_inode_info *pipe, struct pipe_buffer *buf) |
1da177e4c
|
118 119 |
{ struct page *page = buf->page; |
5274f052e
|
120 121 122 |
/* * If nobody else uses this page, and we don't already have a * temporary page, let's keep track of it as a one-deep |
341b446bc
|
123 |
* allocation cache. (Otherwise just release our reference to it) |
5274f052e
|
124 |
*/ |
341b446bc
|
125 |
if (page_count(page) == 1 && !pipe->tmp_page) |
923f4f239
|
126 |
pipe->tmp_page = page; |
341b446bc
|
127 |
else |
09cbfeaf1
|
128 |
put_page(page); |
1da177e4c
|
129 |
} |
d86133bd3
|
130 131 132 133 134 135 |
static int anon_pipe_buf_steal(struct pipe_inode_info *pipe, struct pipe_buffer *buf) { struct page *page = buf->page; if (page_count(page) == 1) { |
c4159a75b
|
136 |
if (memcg_kmem_enabled()) |
d86133bd3
|
137 |
memcg_kmem_uncharge(page, 0); |
d86133bd3
|
138 139 140 141 142 |
__SetPageLocked(page); return 0; } return 1; } |
0845718da
|
143 |
/** |
b51d63c6d
|
144 |
* generic_pipe_buf_steal - attempt to take ownership of a &pipe_buffer |
0845718da
|
145 146 147 148 |
* @pipe: the pipe that the buffer belongs to * @buf: the buffer to attempt to steal * * Description: |
b51d63c6d
|
149 |
* This function attempts to steal the &struct page attached to |
0845718da
|
150 151 |
* @buf. If successful, this function returns 0 and returns with * the page locked. The caller may then reuse the page for whatever |
b51d63c6d
|
152 |
* he wishes; the typical use is insertion into a different file |
0845718da
|
153 154 |
* page cache. */ |
330ab7161
|
155 156 |
int generic_pipe_buf_steal(struct pipe_inode_info *pipe, struct pipe_buffer *buf) |
5abc97aa2
|
157 |
{ |
46e678c96
|
158 |
struct page *page = buf->page; |
0845718da
|
159 160 161 162 163 |
/* * A reference of one is golden, that means that the owner of this * page is the only one holding a reference to it. lock the page * and return OK. */ |
46e678c96
|
164 |
if (page_count(page) == 1) { |
46e678c96
|
165 166 167 168 169 |
lock_page(page); return 0; } return 1; |
5abc97aa2
|
170 |
} |
51921cb74
|
171 |
EXPORT_SYMBOL(generic_pipe_buf_steal); |
5abc97aa2
|
172 |
|
0845718da
|
173 |
/** |
b51d63c6d
|
174 |
* generic_pipe_buf_get - get a reference to a &struct pipe_buffer |
0845718da
|
175 176 177 178 179 180 181 182 183 |
* @pipe: the pipe that the buffer belongs to * @buf: the buffer to get a reference to * * Description: * This function grabs an extra reference to @buf. It's used in * in the tee() system call, when we duplicate the buffers in one * pipe into another. */ void generic_pipe_buf_get(struct pipe_inode_info *pipe, struct pipe_buffer *buf) |
70524490e
|
184 |
{ |
09cbfeaf1
|
185 |
get_page(buf->page); |
70524490e
|
186 |
} |
51921cb74
|
187 |
EXPORT_SYMBOL(generic_pipe_buf_get); |
70524490e
|
188 |
|
0845718da
|
189 190 |
/** * generic_pipe_buf_confirm - verify contents of the pipe buffer |
79685b8de
|
191 |
* @info: the pipe that the buffer belongs to |
0845718da
|
192 193 194 195 196 197 |
* @buf: the buffer to confirm * * Description: * This function does nothing, because the generic pipe code uses * pages that are always good when inserted into the pipe. */ |
cac36bb06
|
198 199 |
int generic_pipe_buf_confirm(struct pipe_inode_info *info, struct pipe_buffer *buf) |
f84d75199
|
200 201 202 |
{ return 0; } |
51921cb74
|
203 |
EXPORT_SYMBOL(generic_pipe_buf_confirm); |
f84d75199
|
204 |
|
6818173bd
|
205 206 207 208 209 210 211 212 213 214 215 |
/** * generic_pipe_buf_release - put a reference to a &struct pipe_buffer * @pipe: the pipe that the buffer belongs to * @buf: the buffer to put a reference to * * Description: * This function releases a reference to @buf. */ void generic_pipe_buf_release(struct pipe_inode_info *pipe, struct pipe_buffer *buf) { |
09cbfeaf1
|
216 |
put_page(buf->page); |
6818173bd
|
217 |
} |
51921cb74
|
218 |
EXPORT_SYMBOL(generic_pipe_buf_release); |
6818173bd
|
219 |
|
d4c3cca94
|
220 |
static const struct pipe_buf_operations anon_pipe_buf_ops = { |
1da177e4c
|
221 |
.can_merge = 1, |
cac36bb06
|
222 |
.confirm = generic_pipe_buf_confirm, |
1da177e4c
|
223 |
.release = anon_pipe_buf_release, |
d86133bd3
|
224 |
.steal = anon_pipe_buf_steal, |
f84d75199
|
225 |
.get = generic_pipe_buf_get, |
1da177e4c
|
226 |
}; |
9883035ae
|
227 228 |
static const struct pipe_buf_operations packet_pipe_buf_ops = { .can_merge = 0, |
9883035ae
|
229 230 |
.confirm = generic_pipe_buf_confirm, .release = anon_pipe_buf_release, |
d86133bd3
|
231 |
.steal = anon_pipe_buf_steal, |
9883035ae
|
232 233 |
.get = generic_pipe_buf_get, }; |
1da177e4c
|
234 |
static ssize_t |
fb9096a34
|
235 |
pipe_read(struct kiocb *iocb, struct iov_iter *to) |
1da177e4c
|
236 |
{ |
fb9096a34
|
237 |
size_t total_len = iov_iter_count(to); |
ee0b3e671
|
238 |
struct file *filp = iocb->ki_filp; |
de32ec4cf
|
239 |
struct pipe_inode_info *pipe = filp->private_data; |
1da177e4c
|
240 241 |
int do_wakeup; ssize_t ret; |
1da177e4c
|
242 |
|
1da177e4c
|
243 244 245 246 247 248 |
/* Null read succeeds. */ if (unlikely(total_len == 0)) return 0; do_wakeup = 0; ret = 0; |
ebec73f47
|
249 |
__pipe_lock(pipe); |
1da177e4c
|
250 |
for (;;) { |
923f4f239
|
251 |
int bufs = pipe->nrbufs; |
1da177e4c
|
252 |
if (bufs) { |
923f4f239
|
253 254 |
int curbuf = pipe->curbuf; struct pipe_buffer *buf = pipe->bufs + curbuf; |
1da177e4c
|
255 |
size_t chars = buf->len; |
637b58c28
|
256 257 |
size_t written; int error; |
1da177e4c
|
258 259 260 |
if (chars > total_len) chars = total_len; |
fba597db4
|
261 |
error = pipe_buf_confirm(pipe, buf); |
f84d75199
|
262 |
if (error) { |
5274f052e
|
263 |
if (!ret) |
e5953cbdf
|
264 |
ret = error; |
5274f052e
|
265 266 |
break; } |
f84d75199
|
267 |
|
fb9096a34
|
268 |
written = copy_page_to_iter(buf->page, buf->offset, chars, to); |
637b58c28
|
269 |
if (unlikely(written < chars)) { |
341b446bc
|
270 |
if (!ret) |
637b58c28
|
271 |
ret = -EFAULT; |
1da177e4c
|
272 273 274 275 276 |
break; } ret += chars; buf->offset += chars; buf->len -= chars; |
9883035ae
|
277 278 279 280 281 282 |
/* Was it a packet buffer? Clean up and exit */ if (buf->flags & PIPE_BUF_FLAG_PACKET) { total_len = chars; buf->len = 0; } |
1da177e4c
|
283 |
if (!buf->len) { |
a779638cf
|
284 |
pipe_buf_release(pipe, buf); |
35f3d14db
|
285 |
curbuf = (curbuf + 1) & (pipe->buffers - 1); |
923f4f239
|
286 287 |
pipe->curbuf = curbuf; pipe->nrbufs = --bufs; |
1da177e4c
|
288 289 290 291 292 293 294 295 |
do_wakeup = 1; } total_len -= chars; if (!total_len) break; /* common path: read succeeded */ } if (bufs) /* More to do? */ continue; |
923f4f239
|
296 |
if (!pipe->writers) |
1da177e4c
|
297 |
break; |
923f4f239
|
298 |
if (!pipe->waiting_writers) { |
1da177e4c
|
299 300 301 302 303 304 305 306 307 308 309 310 311 |
/* syscall merging: Usually we must not sleep * if O_NONBLOCK is set, or if we got some data. * But if a writer sleeps in kernel space, then * we can wait for that data without violating POSIX. */ if (ret) break; if (filp->f_flags & O_NONBLOCK) { ret = -EAGAIN; break; } } if (signal_pending(current)) { |
341b446bc
|
312 313 |
if (!ret) ret = -ERESTARTSYS; |
1da177e4c
|
314 315 316 |
break; } if (do_wakeup) { |
28e58ee8c
|
317 |
wake_up_interruptible_sync_poll(&pipe->wait, POLLOUT | POLLWRNORM); |
923f4f239
|
318 |
kill_fasync(&pipe->fasync_writers, SIGIO, POLL_OUT); |
1da177e4c
|
319 |
} |
923f4f239
|
320 |
pipe_wait(pipe); |
1da177e4c
|
321 |
} |
ebec73f47
|
322 |
__pipe_unlock(pipe); |
341b446bc
|
323 324 |
/* Signal writers asynchronously that there is more room. */ |
1da177e4c
|
325 |
if (do_wakeup) { |
28e58ee8c
|
326 |
wake_up_interruptible_sync_poll(&pipe->wait, POLLOUT | POLLWRNORM); |
923f4f239
|
327 |
kill_fasync(&pipe->fasync_writers, SIGIO, POLL_OUT); |
1da177e4c
|
328 329 330 331 332 |
} if (ret > 0) file_accessed(filp); return ret; } |
9883035ae
|
333 334 335 336 |
static inline int is_packetized(struct file *file) { return (file->f_flags & O_DIRECT) != 0; } |
1da177e4c
|
337 |
static ssize_t |
f0d1bec9d
|
338 |
pipe_write(struct kiocb *iocb, struct iov_iter *from) |
1da177e4c
|
339 |
{ |
ee0b3e671
|
340 |
struct file *filp = iocb->ki_filp; |
de32ec4cf
|
341 |
struct pipe_inode_info *pipe = filp->private_data; |
f0d1bec9d
|
342 343 344 |
ssize_t ret = 0; int do_wakeup = 0; size_t total_len = iov_iter_count(from); |
1da177e4c
|
345 |
ssize_t chars; |
1da177e4c
|
346 347 348 |
/* Null write succeeds. */ if (unlikely(total_len == 0)) return 0; |
ebec73f47
|
349 |
__pipe_lock(pipe); |
1da177e4c
|
350 |
|
923f4f239
|
351 |
if (!pipe->readers) { |
1da177e4c
|
352 353 354 355 356 357 358 |
send_sig(SIGPIPE, current, 0); ret = -EPIPE; goto out; } /* We try to merge small writes */ chars = total_len & (PAGE_SIZE-1); /* size of the last buffer */ |
923f4f239
|
359 |
if (pipe->nrbufs && chars != 0) { |
341b446bc
|
360 |
int lastbuf = (pipe->curbuf + pipe->nrbufs - 1) & |
35f3d14db
|
361 |
(pipe->buffers - 1); |
923f4f239
|
362 |
struct pipe_buffer *buf = pipe->bufs + lastbuf; |
1da177e4c
|
363 |
int offset = buf->offset + buf->len; |
341b446bc
|
364 |
|
fba597db4
|
365 366 |
if (buf->ops->can_merge && offset + chars <= PAGE_SIZE) { ret = pipe_buf_confirm(pipe, buf); |
6ae080699
|
367 |
if (ret) |
5274f052e
|
368 |
goto out; |
f84d75199
|
369 |
|
f0d1bec9d
|
370 371 |
ret = copy_page_from_iter(buf->page, offset, chars, from); if (unlikely(ret < chars)) { |
6ae080699
|
372 |
ret = -EFAULT; |
1da177e4c
|
373 |
goto out; |
f6762b7ad
|
374 |
} |
f0d1bec9d
|
375 |
do_wakeup = 1; |
6ae080699
|
376 |
buf->len += ret; |
f0d1bec9d
|
377 |
if (!iov_iter_count(from)) |
1da177e4c
|
378 379 380 381 382 383 |
goto out; } } for (;;) { int bufs; |
341b446bc
|
384 |
|
923f4f239
|
385 |
if (!pipe->readers) { |
1da177e4c
|
386 |
send_sig(SIGPIPE, current, 0); |
341b446bc
|
387 388 |
if (!ret) ret = -EPIPE; |
1da177e4c
|
389 390 |
break; } |
923f4f239
|
391 |
bufs = pipe->nrbufs; |
35f3d14db
|
392 393 |
if (bufs < pipe->buffers) { int newbuf = (pipe->curbuf + bufs) & (pipe->buffers-1); |
923f4f239
|
394 395 |
struct pipe_buffer *buf = pipe->bufs + newbuf; struct page *page = pipe->tmp_page; |
f0d1bec9d
|
396 |
int copied; |
1da177e4c
|
397 398 |
if (!page) { |
d86133bd3
|
399 |
page = alloc_page(GFP_HIGHUSER | __GFP_ACCOUNT); |
1da177e4c
|
400 401 402 403 |
if (unlikely(!page)) { ret = ret ? : -ENOMEM; break; } |
923f4f239
|
404 |
pipe->tmp_page = page; |
1da177e4c
|
405 |
} |
341b446bc
|
406 |
/* Always wake up, even if the copy fails. Otherwise |
1da177e4c
|
407 408 409 410 411 |
* we lock up (O_NONBLOCK-)readers that sleep due to * syscall merging. * FIXME! Is this really true? */ do_wakeup = 1; |
f0d1bec9d
|
412 413 |
copied = copy_page_from_iter(page, 0, PAGE_SIZE, from); if (unlikely(copied < PAGE_SIZE && iov_iter_count(from))) { |
341b446bc
|
414 |
if (!ret) |
f0d1bec9d
|
415 |
ret = -EFAULT; |
1da177e4c
|
416 417 |
break; } |
f0d1bec9d
|
418 |
ret += copied; |
1da177e4c
|
419 420 421 422 423 |
/* Insert it into the buffer array */ buf->page = page; buf->ops = &anon_pipe_buf_ops; buf->offset = 0; |
f0d1bec9d
|
424 |
buf->len = copied; |
9883035ae
|
425 426 427 428 429 |
buf->flags = 0; if (is_packetized(filp)) { buf->ops = &packet_pipe_buf_ops; buf->flags = PIPE_BUF_FLAG_PACKET; } |
923f4f239
|
430 431 |
pipe->nrbufs = ++bufs; pipe->tmp_page = NULL; |
1da177e4c
|
432 |
|
f0d1bec9d
|
433 |
if (!iov_iter_count(from)) |
1da177e4c
|
434 435 |
break; } |
35f3d14db
|
436 |
if (bufs < pipe->buffers) |
1da177e4c
|
437 438 |
continue; if (filp->f_flags & O_NONBLOCK) { |
341b446bc
|
439 440 |
if (!ret) ret = -EAGAIN; |
1da177e4c
|
441 442 443 |
break; } if (signal_pending(current)) { |
341b446bc
|
444 445 |
if (!ret) ret = -ERESTARTSYS; |
1da177e4c
|
446 447 448 |
break; } if (do_wakeup) { |
28e58ee8c
|
449 |
wake_up_interruptible_sync_poll(&pipe->wait, POLLIN | POLLRDNORM); |
923f4f239
|
450 |
kill_fasync(&pipe->fasync_readers, SIGIO, POLL_IN); |
1da177e4c
|
451 452 |
do_wakeup = 0; } |
923f4f239
|
453 454 455 |
pipe->waiting_writers++; pipe_wait(pipe); pipe->waiting_writers--; |
1da177e4c
|
456 457 |
} out: |
ebec73f47
|
458 |
__pipe_unlock(pipe); |
1da177e4c
|
459 |
if (do_wakeup) { |
28e58ee8c
|
460 |
wake_up_interruptible_sync_poll(&pipe->wait, POLLIN | POLLRDNORM); |
923f4f239
|
461 |
kill_fasync(&pipe->fasync_readers, SIGIO, POLL_IN); |
1da177e4c
|
462 |
} |
7e775f46a
|
463 |
if (ret > 0 && sb_start_write_trylock(file_inode(filp)->i_sb)) { |
c3b2da314
|
464 465 466 |
int err = file_update_time(filp); if (err) ret = err; |
7e775f46a
|
467 |
sb_end_write(file_inode(filp)->i_sb); |
c3b2da314
|
468 |
} |
1da177e4c
|
469 470 |
return ret; } |
d59d0b1b8
|
471 |
static long pipe_ioctl(struct file *filp, unsigned int cmd, unsigned long arg) |
1da177e4c
|
472 |
{ |
de32ec4cf
|
473 |
struct pipe_inode_info *pipe = filp->private_data; |
1da177e4c
|
474 475 476 477 |
int count, buf, nrbufs; switch (cmd) { case FIONREAD: |
ebec73f47
|
478 |
__pipe_lock(pipe); |
1da177e4c
|
479 |
count = 0; |
923f4f239
|
480 481 |
buf = pipe->curbuf; nrbufs = pipe->nrbufs; |
1da177e4c
|
482 |
while (--nrbufs >= 0) { |
923f4f239
|
483 |
count += pipe->bufs[buf].len; |
35f3d14db
|
484 |
buf = (buf+1) & (pipe->buffers - 1); |
1da177e4c
|
485 |
} |
ebec73f47
|
486 |
__pipe_unlock(pipe); |
923f4f239
|
487 |
|
1da177e4c
|
488 489 |
return put_user(count, (int __user *)arg); default: |
46ce341b2
|
490 |
return -ENOIOCTLCMD; |
1da177e4c
|
491 492 493 494 495 496 497 498 |
} } /* No kernel lock held - fine */ static unsigned int pipe_poll(struct file *filp, poll_table *wait) { unsigned int mask; |
de32ec4cf
|
499 |
struct pipe_inode_info *pipe = filp->private_data; |
1da177e4c
|
500 |
int nrbufs; |
923f4f239
|
501 |
poll_wait(filp, &pipe->wait, wait); |
1da177e4c
|
502 503 |
/* Reading only -- no need for acquiring the semaphore. */ |
923f4f239
|
504 |
nrbufs = pipe->nrbufs; |
1da177e4c
|
505 506 507 |
mask = 0; if (filp->f_mode & FMODE_READ) { mask = (nrbufs > 0) ? POLLIN | POLLRDNORM : 0; |
923f4f239
|
508 |
if (!pipe->writers && filp->f_version != pipe->w_counter) |
1da177e4c
|
509 510 511 512 |
mask |= POLLHUP; } if (filp->f_mode & FMODE_WRITE) { |
35f3d14db
|
513 |
mask |= (nrbufs < pipe->buffers) ? POLLOUT | POLLWRNORM : 0; |
5e5d7a222
|
514 515 516 517 |
/* * Most Unices do not set POLLERR for FIFOs but on Linux they * behave exactly like pipes for poll(). */ |
923f4f239
|
518 |
if (!pipe->readers) |
1da177e4c
|
519 520 521 522 523 |
mask |= POLLERR; } return mask; } |
b0d8d2292
|
524 525 526 527 528 529 530 531 532 533 534 535 536 537 |
static void put_pipe_info(struct inode *inode, struct pipe_inode_info *pipe) { int kill = 0; spin_lock(&inode->i_lock); if (!--pipe->files) { inode->i_pipe = NULL; kill = 1; } spin_unlock(&inode->i_lock); if (kill) free_pipe_info(pipe); } |
1da177e4c
|
538 |
static int |
599a0ac14
|
539 |
pipe_release(struct inode *inode, struct file *file) |
1da177e4c
|
540 |
{ |
b0d8d2292
|
541 |
struct pipe_inode_info *pipe = file->private_data; |
923f4f239
|
542 |
|
ebec73f47
|
543 |
__pipe_lock(pipe); |
599a0ac14
|
544 545 546 547 |
if (file->f_mode & FMODE_READ) pipe->readers--; if (file->f_mode & FMODE_WRITE) pipe->writers--; |
341b446bc
|
548 |
|
ba5bb1473
|
549 |
if (pipe->readers || pipe->writers) { |
28e58ee8c
|
550 |
wake_up_interruptible_sync_poll(&pipe->wait, POLLIN | POLLOUT | POLLRDNORM | POLLWRNORM | POLLERR | POLLHUP); |
923f4f239
|
551 552 |
kill_fasync(&pipe->fasync_readers, SIGIO, POLL_IN); kill_fasync(&pipe->fasync_writers, SIGIO, POLL_OUT); |
1da177e4c
|
553 |
} |
ebec73f47
|
554 |
__pipe_unlock(pipe); |
ba5bb1473
|
555 |
|
b0d8d2292
|
556 |
put_pipe_info(inode, pipe); |
1da177e4c
|
557 558 559 560 |
return 0; } static int |
599a0ac14
|
561 |
pipe_fasync(int fd, struct file *filp, int on) |
1da177e4c
|
562 |
{ |
de32ec4cf
|
563 |
struct pipe_inode_info *pipe = filp->private_data; |
599a0ac14
|
564 |
int retval = 0; |
1da177e4c
|
565 |
|
ebec73f47
|
566 |
__pipe_lock(pipe); |
599a0ac14
|
567 568 569 |
if (filp->f_mode & FMODE_READ) retval = fasync_helper(fd, filp, on, &pipe->fasync_readers); if ((filp->f_mode & FMODE_WRITE) && retval >= 0) { |
341b446bc
|
570 |
retval = fasync_helper(fd, filp, on, &pipe->fasync_writers); |
599a0ac14
|
571 572 |
if (retval < 0 && (filp->f_mode & FMODE_READ)) /* this can happen only if on == T */ |
e5bc49ba7
|
573 574 |
fasync_helper(-1, filp, 0, &pipe->fasync_readers); } |
ebec73f47
|
575 |
__pipe_unlock(pipe); |
60aa49243
|
576 |
return retval; |
1da177e4c
|
577 |
} |
9c87bcf0a
|
578 |
static unsigned long account_pipe_buffers(struct user_struct *user, |
759c01142
|
579 580 |
unsigned long old, unsigned long new) { |
9c87bcf0a
|
581 |
return atomic_long_add_return(new - old, &user->pipe_bufs); |
759c01142
|
582 |
} |
9c87bcf0a
|
583 |
static bool too_many_pipe_buffers_soft(unsigned long user_bufs) |
759c01142
|
584 |
{ |
9c87bcf0a
|
585 |
return pipe_user_pages_soft && user_bufs >= pipe_user_pages_soft; |
759c01142
|
586 |
} |
9c87bcf0a
|
587 |
static bool too_many_pipe_buffers_hard(unsigned long user_bufs) |
759c01142
|
588 |
{ |
9c87bcf0a
|
589 |
return pipe_user_pages_hard && user_bufs >= pipe_user_pages_hard; |
759c01142
|
590 |
} |
7bee130e2
|
591 |
struct pipe_inode_info *alloc_pipe_info(void) |
3a326a2ce
|
592 |
{ |
923f4f239
|
593 |
struct pipe_inode_info *pipe; |
09b4d1990
|
594 595 |
unsigned long pipe_bufs = PIPE_DEF_BUFFERS; struct user_struct *user = get_current_user(); |
9c87bcf0a
|
596 |
unsigned long user_bufs; |
3a326a2ce
|
597 |
|
d86133bd3
|
598 |
pipe = kzalloc(sizeof(struct pipe_inode_info), GFP_KERNEL_ACCOUNT); |
09b4d1990
|
599 600 |
if (pipe == NULL) goto out_free_uid; |
086e774a5
|
601 602 |
if (pipe_bufs * PAGE_SIZE > pipe_max_size && !capable(CAP_SYS_RESOURCE)) pipe_bufs = pipe_max_size >> PAGE_SHIFT; |
9c87bcf0a
|
603 |
user_bufs = account_pipe_buffers(user, 0, pipe_bufs); |
a005ca0e6
|
604 |
|
9c87bcf0a
|
605 606 |
if (too_many_pipe_buffers_soft(user_bufs)) { user_bufs = account_pipe_buffers(user, pipe_bufs, 1); |
a005ca0e6
|
607 |
pipe_bufs = 1; |
09b4d1990
|
608 |
} |
759c01142
|
609 |
|
9c87bcf0a
|
610 |
if (too_many_pipe_buffers_hard(user_bufs)) |
a005ca0e6
|
611 612 613 614 |
goto out_revert_acct; pipe->bufs = kcalloc(pipe_bufs, sizeof(struct pipe_buffer), GFP_KERNEL_ACCOUNT); |
09b4d1990
|
615 616 617 618 619 |
if (pipe->bufs) { init_waitqueue_head(&pipe->wait); pipe->r_counter = pipe->w_counter = 1; pipe->buffers = pipe_bufs; pipe->user = user; |
09b4d1990
|
620 621 |
mutex_init(&pipe->mutex); return pipe; |
3a326a2ce
|
622 |
} |
a005ca0e6
|
623 |
out_revert_acct: |
9c87bcf0a
|
624 |
(void) account_pipe_buffers(user, pipe_bufs, 0); |
09b4d1990
|
625 626 627 |
kfree(pipe); out_free_uid: free_uid(user); |
35f3d14db
|
628 |
return NULL; |
3a326a2ce
|
629 |
} |
4b8a8f1e4
|
630 |
void free_pipe_info(struct pipe_inode_info *pipe) |
1da177e4c
|
631 632 |
{ int i; |
1da177e4c
|
633 |
|
9c87bcf0a
|
634 |
(void) account_pipe_buffers(pipe->user, pipe->buffers, 0); |
759c01142
|
635 |
free_uid(pipe->user); |
35f3d14db
|
636 |
for (i = 0; i < pipe->buffers; i++) { |
923f4f239
|
637 |
struct pipe_buffer *buf = pipe->bufs + i; |
1da177e4c
|
638 |
if (buf->ops) |
a779638cf
|
639 |
pipe_buf_release(pipe, buf); |
1da177e4c
|
640 |
} |
923f4f239
|
641 642 |
if (pipe->tmp_page) __free_page(pipe->tmp_page); |
35f3d14db
|
643 |
kfree(pipe->bufs); |
923f4f239
|
644 |
kfree(pipe); |
1da177e4c
|
645 |
} |
fa3536cc1
|
646 |
static struct vfsmount *pipe_mnt __read_mostly; |
341b446bc
|
647 |
|
c23fbb6bc
|
648 649 650 651 652 653 |
/* * pipefs_dname() is called from d_path(). */ static char *pipefs_dname(struct dentry *dentry, char *buffer, int buflen) { return dynamic_dname(dentry, buffer, buflen, "pipe:[%lu]", |
75c3cfa85
|
654 |
d_inode(dentry)->i_ino); |
c23fbb6bc
|
655 |
} |
3ba13d179
|
656 |
static const struct dentry_operations pipefs_dentry_operations = { |
c23fbb6bc
|
657 |
.d_dname = pipefs_dname, |
1da177e4c
|
658 659 660 661 |
}; static struct inode * get_pipe_inode(void) { |
a209dfc7b
|
662 |
struct inode *inode = new_inode_pseudo(pipe_mnt->mnt_sb); |
923f4f239
|
663 |
struct pipe_inode_info *pipe; |
1da177e4c
|
664 665 666 |
if (!inode) goto fail_inode; |
85fe4025c
|
667 |
inode->i_ino = get_next_ino(); |
7bee130e2
|
668 |
pipe = alloc_pipe_info(); |
923f4f239
|
669 |
if (!pipe) |
1da177e4c
|
670 |
goto fail_iput; |
3a326a2ce
|
671 |
|
ba5bb1473
|
672 673 |
inode->i_pipe = pipe; pipe->files = 2; |
923f4f239
|
674 |
pipe->readers = pipe->writers = 1; |
599a0ac14
|
675 |
inode->i_fop = &pipefifo_fops; |
1da177e4c
|
676 677 678 679 680 681 682 683 684 |
/* * Mark the inode dirty from the very beginning, * that way it will never be moved to the dirty * list because "mark_inode_dirty()" will think * that it already _is_ on the dirty list. */ inode->i_state = I_DIRTY; inode->i_mode = S_IFIFO | S_IRUSR | S_IWUSR; |
da9592ede
|
685 686 |
inode->i_uid = current_fsuid(); inode->i_gid = current_fsgid(); |
078cd8279
|
687 |
inode->i_atime = inode->i_mtime = inode->i_ctime = current_time(inode); |
923f4f239
|
688 |
|
1da177e4c
|
689 690 691 692 |
return inode; fail_iput: iput(inode); |
341b446bc
|
693 |
|
1da177e4c
|
694 695 696 |
fail_inode: return NULL; } |
e4fad8e5d
|
697 |
int create_pipe_files(struct file **res, int flags) |
1da177e4c
|
698 |
{ |
d6cbd281d
|
699 |
int err; |
e4fad8e5d
|
700 |
struct inode *inode = get_pipe_inode(); |
d6cbd281d
|
701 |
struct file *f; |
2c48b9c45
|
702 |
struct path path; |
1da177e4c
|
703 |
|
1da177e4c
|
704 |
if (!inode) |
e4fad8e5d
|
705 |
return -ENFILE; |
1da177e4c
|
706 |
|
d6cbd281d
|
707 |
err = -ENOMEM; |
cdf01226b
|
708 |
path.dentry = d_alloc_pseudo(pipe_mnt->mnt_sb, &empty_name); |
2c48b9c45
|
709 |
if (!path.dentry) |
d6cbd281d
|
710 |
goto err_inode; |
2c48b9c45
|
711 |
path.mnt = mntget(pipe_mnt); |
341b446bc
|
712 |
|
2c48b9c45
|
713 |
d_instantiate(path.dentry, inode); |
430e285e0
|
714 |
|
599a0ac14
|
715 |
f = alloc_file(&path, FMODE_WRITE, &pipefifo_fops); |
e9bb1f9b1
|
716 717 |
if (IS_ERR(f)) { err = PTR_ERR(f); |
430e285e0
|
718 |
goto err_dentry; |
e9bb1f9b1
|
719 |
} |
341b446bc
|
720 |
|
9883035ae
|
721 |
f->f_flags = O_WRONLY | (flags & (O_NONBLOCK | O_DIRECT)); |
de32ec4cf
|
722 |
f->private_data = inode->i_pipe; |
d6cbd281d
|
723 |
|
599a0ac14
|
724 |
res[0] = alloc_file(&path, FMODE_READ, &pipefifo_fops); |
e9bb1f9b1
|
725 726 |
if (IS_ERR(res[0])) { err = PTR_ERR(res[0]); |
e4fad8e5d
|
727 |
goto err_file; |
e9bb1f9b1
|
728 |
} |
e4fad8e5d
|
729 730 |
path_get(&path); |
de32ec4cf
|
731 |
res[0]->private_data = inode->i_pipe; |
e4fad8e5d
|
732 733 734 |
res[0]->f_flags = O_RDONLY | (flags & O_NONBLOCK); res[1] = f; return 0; |
1da177e4c
|
735 |
|
e4fad8e5d
|
736 737 738 |
err_file: put_filp(f); err_dentry: |
4b8a8f1e4
|
739 |
free_pipe_info(inode->i_pipe); |
2c48b9c45
|
740 |
path_put(&path); |
e4fad8e5d
|
741 |
return err; |
ed1524371
|
742 |
|
e4fad8e5d
|
743 |
err_inode: |
4b8a8f1e4
|
744 |
free_pipe_info(inode->i_pipe); |
1da177e4c
|
745 |
iput(inode); |
e4fad8e5d
|
746 |
return err; |
d6cbd281d
|
747 |
} |
5b249b1b0
|
748 |
static int __do_pipe_flags(int *fd, struct file **files, int flags) |
d6cbd281d
|
749 |
{ |
d6cbd281d
|
750 751 |
int error; int fdw, fdr; |
9883035ae
|
752 |
if (flags & ~(O_CLOEXEC | O_NONBLOCK | O_DIRECT)) |
ed8cae8ba
|
753 |
return -EINVAL; |
e4fad8e5d
|
754 755 756 |
error = create_pipe_files(files, flags); if (error) return error; |
d6cbd281d
|
757 |
|
ed8cae8ba
|
758 |
error = get_unused_fd_flags(flags); |
d6cbd281d
|
759 760 761 |
if (error < 0) goto err_read_pipe; fdr = error; |
ed8cae8ba
|
762 |
error = get_unused_fd_flags(flags); |
d6cbd281d
|
763 764 765 |
if (error < 0) goto err_fdr; fdw = error; |
157cf649a
|
766 |
audit_fd_pair(fdr, fdw); |
d6cbd281d
|
767 768 |
fd[0] = fdr; fd[1] = fdw; |
d6cbd281d
|
769 770 771 772 773 |
return 0; err_fdr: put_unused_fd(fdr); err_read_pipe: |
e4fad8e5d
|
774 775 |
fput(files[0]); fput(files[1]); |
d6cbd281d
|
776 |
return error; |
1da177e4c
|
777 |
} |
5b249b1b0
|
778 779 780 781 782 783 784 785 786 787 |
int do_pipe_flags(int *fd, int flags) { struct file *files[2]; int error = __do_pipe_flags(fd, files, flags); if (!error) { fd_install(fd[0], files[0]); fd_install(fd[1], files[1]); } return error; } |
1da177e4c
|
788 |
/* |
d35c7b0e5
|
789 790 791 |
* sys_pipe() is the normal C calling standard for creating * a pipe. It's not the way Unix traditionally does this, though. */ |
d4e82042c
|
792 |
SYSCALL_DEFINE2(pipe2, int __user *, fildes, int, flags) |
d35c7b0e5
|
793 |
{ |
5b249b1b0
|
794 |
struct file *files[2]; |
d35c7b0e5
|
795 796 |
int fd[2]; int error; |
5b249b1b0
|
797 |
error = __do_pipe_flags(fd, files, flags); |
d35c7b0e5
|
798 |
if (!error) { |
5b249b1b0
|
799 800 801 802 803 |
if (unlikely(copy_to_user(fildes, fd, sizeof(fd)))) { fput(files[0]); fput(files[1]); put_unused_fd(fd[0]); put_unused_fd(fd[1]); |
d35c7b0e5
|
804 |
error = -EFAULT; |
5b249b1b0
|
805 806 807 |
} else { fd_install(fd[0], files[0]); fd_install(fd[1], files[1]); |
ba719baea
|
808 |
} |
d35c7b0e5
|
809 810 811 |
} return error; } |
2b6642199
|
812 |
SYSCALL_DEFINE1(pipe, int __user *, fildes) |
ed8cae8ba
|
813 814 815 |
{ return sys_pipe2(fildes, 0); } |
fc7478a2b
|
816 |
static int wait_for_partner(struct pipe_inode_info *pipe, unsigned int *cnt) |
f776c7388
|
817 818 819 820 |
{ int cur = *cnt; while (cur == *cnt) { |
fc7478a2b
|
821 |
pipe_wait(pipe); |
f776c7388
|
822 823 824 825 826 |
if (signal_pending(current)) break; } return cur == *cnt ? -ERESTARTSYS : 0; } |
fc7478a2b
|
827 |
static void wake_up_partner(struct pipe_inode_info *pipe) |
f776c7388
|
828 |
{ |
fc7478a2b
|
829 |
wake_up_interruptible(&pipe->wait); |
f776c7388
|
830 831 832 833 834 |
} static int fifo_open(struct inode *inode, struct file *filp) { struct pipe_inode_info *pipe; |
599a0ac14
|
835 |
bool is_pipe = inode->i_sb->s_magic == PIPEFS_MAGIC; |
f776c7388
|
836 |
int ret; |
ba5bb1473
|
837 838 839 840 841 842 843 844 845 |
filp->f_version = 0; spin_lock(&inode->i_lock); if (inode->i_pipe) { pipe = inode->i_pipe; pipe->files++; spin_unlock(&inode->i_lock); } else { spin_unlock(&inode->i_lock); |
7bee130e2
|
846 |
pipe = alloc_pipe_info(); |
f776c7388
|
847 |
if (!pipe) |
ba5bb1473
|
848 849 850 851 852 853 |
return -ENOMEM; pipe->files = 1; spin_lock(&inode->i_lock); if (unlikely(inode->i_pipe)) { inode->i_pipe->files++; spin_unlock(&inode->i_lock); |
4b8a8f1e4
|
854 |
free_pipe_info(pipe); |
ba5bb1473
|
855 856 857 858 859 |
pipe = inode->i_pipe; } else { inode->i_pipe = pipe; spin_unlock(&inode->i_lock); } |
f776c7388
|
860 |
} |
de32ec4cf
|
861 |
filp->private_data = pipe; |
ba5bb1473
|
862 |
/* OK, we have a pipe and it's pinned down */ |
ebec73f47
|
863 |
__pipe_lock(pipe); |
f776c7388
|
864 865 866 867 868 869 870 871 872 873 874 |
/* We can only do regular read/write on fifos */ filp->f_mode &= (FMODE_READ | FMODE_WRITE); switch (filp->f_mode) { case FMODE_READ: /* * O_RDONLY * POSIX.1 says that O_NONBLOCK means return with the FIFO * opened, even when there is no process writing the FIFO. */ |
f776c7388
|
875 876 |
pipe->r_counter++; if (pipe->readers++ == 0) |
fc7478a2b
|
877 |
wake_up_partner(pipe); |
f776c7388
|
878 |
|
599a0ac14
|
879 |
if (!is_pipe && !pipe->writers) { |
f776c7388
|
880 881 882 883 884 |
if ((filp->f_flags & O_NONBLOCK)) { /* suppress POLLHUP until we have * seen a writer */ filp->f_version = pipe->w_counter; } else { |
fc7478a2b
|
885 |
if (wait_for_partner(pipe, &pipe->w_counter)) |
f776c7388
|
886 887 888 889 890 891 892 893 894 895 896 897 |
goto err_rd; } } break; case FMODE_WRITE: /* * O_WRONLY * POSIX.1 says that O_NONBLOCK means return -1 with * errno=ENXIO when there is no process reading the FIFO. */ ret = -ENXIO; |
599a0ac14
|
898 |
if (!is_pipe && (filp->f_flags & O_NONBLOCK) && !pipe->readers) |
f776c7388
|
899 |
goto err; |
f776c7388
|
900 901 |
pipe->w_counter++; if (!pipe->writers++) |
fc7478a2b
|
902 |
wake_up_partner(pipe); |
f776c7388
|
903 |
|
599a0ac14
|
904 |
if (!is_pipe && !pipe->readers) { |
fc7478a2b
|
905 |
if (wait_for_partner(pipe, &pipe->r_counter)) |
f776c7388
|
906 907 908 909 910 911 912 913 914 915 916 |
goto err_wr; } break; case FMODE_READ | FMODE_WRITE: /* * O_RDWR * POSIX.1 leaves this case "undefined" when O_NONBLOCK is set. * This implementation will NEVER block on a O_RDWR open, since * the process can at least talk to itself. */ |
f776c7388
|
917 918 919 920 921 922 |
pipe->readers++; pipe->writers++; pipe->r_counter++; pipe->w_counter++; if (pipe->readers == 1 || pipe->writers == 1) |
fc7478a2b
|
923 |
wake_up_partner(pipe); |
f776c7388
|
924 925 926 927 928 929 930 931 |
break; default: ret = -EINVAL; goto err; } /* Ok! */ |
ebec73f47
|
932 |
__pipe_unlock(pipe); |
f776c7388
|
933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 |
return 0; err_rd: if (!--pipe->readers) wake_up_interruptible(&pipe->wait); ret = -ERESTARTSYS; goto err; err_wr: if (!--pipe->writers) wake_up_interruptible(&pipe->wait); ret = -ERESTARTSYS; goto err; err: |
ebec73f47
|
948 |
__pipe_unlock(pipe); |
b0d8d2292
|
949 950 |
put_pipe_info(inode, pipe); |
f776c7388
|
951 952 |
return ret; } |
599a0ac14
|
953 954 955 |
const struct file_operations pipefifo_fops = { .open = fifo_open, .llseek = no_llseek, |
fb9096a34
|
956 |
.read_iter = pipe_read, |
f0d1bec9d
|
957 |
.write_iter = pipe_write, |
599a0ac14
|
958 959 960 961 |
.poll = pipe_poll, .unlocked_ioctl = pipe_ioctl, .release = pipe_release, .fasync = pipe_fasync, |
f776c7388
|
962 |
}; |
d35c7b0e5
|
963 |
/* |
f491bd711
|
964 965 966 967 968 969 970 971 972 973 974 975 |
* Currently we rely on the pipe array holding a power-of-2 number * of pages. */ static inline unsigned int round_pipe_size(unsigned int size) { unsigned long nr_pages; nr_pages = (size + PAGE_SIZE - 1) >> PAGE_SHIFT; return roundup_pow_of_two(nr_pages) << PAGE_SHIFT; } /* |
35f3d14db
|
976 977 978 |
* Allocate a new array of pipe buffers and copy the info over. Returns the * pipe size if successful, or return -ERROR on error. */ |
d37d41666
|
979 |
static long pipe_set_size(struct pipe_inode_info *pipe, unsigned long arg) |
35f3d14db
|
980 981 |
{ struct pipe_buffer *bufs; |
d37d41666
|
982 |
unsigned int size, nr_pages; |
9c87bcf0a
|
983 |
unsigned long user_bufs; |
b0b91d18e
|
984 |
long ret = 0; |
d37d41666
|
985 986 987 988 989 990 |
size = round_pipe_size(arg); nr_pages = size >> PAGE_SHIFT; if (!nr_pages) return -EINVAL; |
b0b91d18e
|
991 992 993 994 995 996 997 998 999 |
/* * If trying to increase the pipe capacity, check that an * unprivileged user is not trying to exceed various limits * (soft limit check here, hard limit check just below). * Decreasing the pipe capacity is always permitted, even * if the user is currently over a limit. */ if (nr_pages > pipe->buffers && size > pipe_max_size && !capable(CAP_SYS_RESOURCE)) |
d37d41666
|
1000 |
return -EPERM; |
9c87bcf0a
|
1001 |
user_bufs = account_pipe_buffers(pipe->user, pipe->buffers, nr_pages); |
b0b91d18e
|
1002 1003 |
if (nr_pages > pipe->buffers && |
9c87bcf0a
|
1004 1005 |
(too_many_pipe_buffers_hard(user_bufs) || too_many_pipe_buffers_soft(user_bufs)) && |
b0b91d18e
|
1006 1007 1008 1009 |
!capable(CAP_SYS_RESOURCE) && !capable(CAP_SYS_ADMIN)) { ret = -EPERM; goto out_revert_acct; } |
35f3d14db
|
1010 1011 |
/* |
35f3d14db
|
1012 1013 1014 1015 1016 |
* We can shrink the pipe, if arg >= pipe->nrbufs. Since we don't * expect a lot of shrink+grow operations, just free and allocate * again like we would do for growing. If the pipe currently * contains more buffers than arg, then return busy. */ |
b0b91d18e
|
1017 1018 1019 1020 |
if (nr_pages < pipe->nrbufs) { ret = -EBUSY; goto out_revert_acct; } |
35f3d14db
|
1021 |
|
d86133bd3
|
1022 1023 |
bufs = kcalloc(nr_pages, sizeof(*bufs), GFP_KERNEL_ACCOUNT | __GFP_NOWARN); |
b0b91d18e
|
1024 1025 1026 1027 |
if (unlikely(!bufs)) { ret = -ENOMEM; goto out_revert_acct; } |
35f3d14db
|
1028 1029 1030 1031 1032 1033 |
/* * The pipe array wraps around, so just start the new one at zero * and adjust the indexes. */ if (pipe->nrbufs) { |
1d862f412
|
1034 1035 |
unsigned int tail; unsigned int head; |
35f3d14db
|
1036 |
|
1d862f412
|
1037 1038 1039 1040 1041 1042 1043 |
tail = pipe->curbuf + pipe->nrbufs; if (tail < pipe->buffers) tail = 0; else tail &= (pipe->buffers - 1); head = pipe->nrbufs - tail; |
35f3d14db
|
1044 1045 1046 |
if (head) memcpy(bufs, pipe->bufs + pipe->curbuf, head * sizeof(struct pipe_buffer)); if (tail) |
1d862f412
|
1047 |
memcpy(bufs + head, pipe->bufs, tail * sizeof(struct pipe_buffer)); |
35f3d14db
|
1048 1049 1050 1051 1052 |
} pipe->curbuf = 0; kfree(pipe->bufs); pipe->bufs = bufs; |
b9598db34
|
1053 1054 |
pipe->buffers = nr_pages; return nr_pages * PAGE_SIZE; |
b0b91d18e
|
1055 1056 |
out_revert_acct: |
9c87bcf0a
|
1057 |
(void) account_pipe_buffers(pipe->user, nr_pages, pipe->buffers); |
b0b91d18e
|
1058 |
return ret; |
35f3d14db
|
1059 |
} |
ff9da691c
|
1060 |
/* |
ff9da691c
|
1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 |
* This should work even if CONFIG_PROC_FS isn't set, as proc_dointvec_minmax * will return an error. */ int pipe_proc_fn(struct ctl_table *table, int write, void __user *buf, size_t *lenp, loff_t *ppos) { int ret; ret = proc_dointvec_minmax(table, write, buf, lenp, ppos); if (ret < 0 || !write) return ret; pipe_max_size = round_pipe_size(pipe_max_size); return ret; } |
720836465
|
1076 1077 1078 1079 1080 1081 1082 |
/* * After the inode slimming patch, i_pipe/i_bdev/i_cdev share the same * location, so checking ->i_pipe is not enough to verify that this is a * pipe. */ struct pipe_inode_info *get_pipe_info(struct file *file) { |
de32ec4cf
|
1083 |
return file->f_op == &pipefifo_fops ? file->private_data : NULL; |
720836465
|
1084 |
} |
35f3d14db
|
1085 1086 1087 1088 |
long pipe_fcntl(struct file *file, unsigned int cmd, unsigned long arg) { struct pipe_inode_info *pipe; long ret; |
c66fb3479
|
1089 |
pipe = get_pipe_info(file); |
35f3d14db
|
1090 1091 |
if (!pipe) return -EBADF; |
ebec73f47
|
1092 |
__pipe_lock(pipe); |
35f3d14db
|
1093 1094 |
switch (cmd) { |
d37d41666
|
1095 1096 |
case F_SETPIPE_SZ: ret = pipe_set_size(pipe, arg); |
35f3d14db
|
1097 1098 |
break; case F_GETPIPE_SZ: |
b9598db34
|
1099 |
ret = pipe->buffers * PAGE_SIZE; |
35f3d14db
|
1100 1101 1102 1103 1104 |
break; default: ret = -EINVAL; break; } |
ebec73f47
|
1105 |
__pipe_unlock(pipe); |
35f3d14db
|
1106 1107 |
return ret; } |
ff0c7d15f
|
1108 1109 |
static const struct super_operations pipefs_ops = { .destroy_inode = free_inode_nonrcu, |
d70ef97ba
|
1110 |
.statfs = simple_statfs, |
ff0c7d15f
|
1111 |
}; |
35f3d14db
|
1112 |
/* |
1da177e4c
|
1113 1114 1115 1116 1117 |
* pipefs should _never_ be mounted by userland - too much of security hassle, * no real gain from having the whole whorehouse mounted. So we don't need * any operations on the root directory. However, we need a non-trivial * d_name - pipe: will go nicely and kill the special-casing in procfs. */ |
51139adac
|
1118 1119 |
static struct dentry *pipefs_mount(struct file_system_type *fs_type, int flags, const char *dev_name, void *data) |
1da177e4c
|
1120 |
{ |
c74a1cbb3
|
1121 1122 |
return mount_pseudo(fs_type, "pipe:", &pipefs_ops, &pipefs_dentry_operations, PIPEFS_MAGIC); |
1da177e4c
|
1123 1124 1125 1126 |
} static struct file_system_type pipe_fs_type = { .name = "pipefs", |
51139adac
|
1127 |
.mount = pipefs_mount, |
1da177e4c
|
1128 1129 1130 1131 1132 1133 |
.kill_sb = kill_anon_super, }; static int __init init_pipe_fs(void) { int err = register_filesystem(&pipe_fs_type); |
341b446bc
|
1134 |
|
1da177e4c
|
1135 1136 1137 1138 1139 1140 1141 1142 1143 |
if (!err) { pipe_mnt = kern_mount(&pipe_fs_type); if (IS_ERR(pipe_mnt)) { err = PTR_ERR(pipe_mnt); unregister_filesystem(&pipe_fs_type); } } return err; } |
1da177e4c
|
1144 |
fs_initcall(init_pipe_fs); |