1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854
|
/* $Id: ioqueue_epoll.c 5194 2015-11-06 04:18:46Z nanang $ */
/*
* Copyright (C) 2008-2011 Teluu Inc. (http://www.teluu.com)
* Copyright (C) 2003-2008 Benny Prijono <benny@prijono.org>
*
* 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.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/
/*
* ioqueue_epoll.c
*
* This is the implementation of IOQueue framework using /dev/epoll
* API in _both_ Linux user-mode and kernel-mode.
*/
#include <pj/ioqueue.h>
#include <pj/os.h>
#include <pj/lock.h>
#include <pj/log.h>
#include <pj/list.h>
#include <pj/pool.h>
#include <pj/string.h>
#include <pj/assert.h>
#include <pj/errno.h>
#include <pj/sock.h>
#include <pj/compat/socket.h>
#include <pj/rand.h>
#if !defined(PJ_LINUX_KERNEL) || PJ_LINUX_KERNEL==0
/*
* Linux user mode
*/
# include <sys/epoll.h>
# include <errno.h>
# include <unistd.h>
# define epoll_data data.ptr
# define epoll_data_type void*
# define ioctl_val_type unsigned long
# define getsockopt_val_ptr int*
# define os_getsockopt getsockopt
# define os_ioctl ioctl
# define os_read read
# define os_close close
# define os_epoll_create epoll_create
# define os_epoll_ctl epoll_ctl
# define os_epoll_wait epoll_wait
#else
/*
* Linux kernel mode.
*/
# include <linux/config.h>
# include <linux/version.h>
# if defined(MODVERSIONS)
# include <linux/modversions.h>
# endif
# include <linux/kernel.h>
# include <linux/poll.h>
# include <linux/eventpoll.h>
# include <linux/syscalls.h>
# include <linux/errno.h>
# include <linux/unistd.h>
# include <asm/ioctls.h>
enum EPOLL_EVENTS
{
EPOLLIN = 0x001,
EPOLLOUT = 0x004,
EPOLLERR = 0x008,
};
# define os_epoll_create sys_epoll_create
static int os_epoll_ctl(int epfd, int op, int fd, struct epoll_event *event)
{
long rc;
mm_segment_t oldfs = get_fs();
set_fs(KERNEL_DS);
rc = sys_epoll_ctl(epfd, op, fd, event);
set_fs(oldfs);
if (rc) {
errno = -rc;
return -1;
} else {
return 0;
}
}
static int os_epoll_wait(int epfd, struct epoll_event *events,
int maxevents, int timeout)
{
int count;
mm_segment_t oldfs = get_fs();
set_fs(KERNEL_DS);
count = sys_epoll_wait(epfd, events, maxevents, timeout);
set_fs(oldfs);
return count;
}
# define os_close sys_close
# define os_getsockopt pj_sock_getsockopt
static int os_read(int fd, void *buf, size_t len)
{
long rc;
mm_segment_t oldfs = get_fs();
set_fs(KERNEL_DS);
rc = sys_read(fd, buf, len);
set_fs(oldfs);
if (rc) {
errno = -rc;
return -1;
} else {
return 0;
}
}
# define socklen_t unsigned
# define ioctl_val_type unsigned long
int ioctl(int fd, int opt, ioctl_val_type value);
static int os_ioctl(int fd, int opt, ioctl_val_type value)
{
int rc;
mm_segment_t oldfs = get_fs();
set_fs(KERNEL_DS);
rc = ioctl(fd, opt, value);
set_fs(oldfs);
if (rc < 0) {
errno = -rc;
return rc;
} else
return rc;
}
# define getsockopt_val_ptr char*
# define epoll_data data
# define epoll_data_type __u32
#endif
#define THIS_FILE "ioq_epoll"
//#define TRACE_(expr) PJ_LOG(3,expr)
#define TRACE_(expr)
/*
* Include common ioqueue abstraction.
*/
#include "ioqueue_common_abs.h"
/*
* This describes each key.
*/
struct pj_ioqueue_key_t
{
DECLARE_COMMON_KEY
};
struct queue
{
pj_ioqueue_key_t *key;
enum ioqueue_event_type event_type;
};
/*
* This describes the I/O queue.
*/
struct pj_ioqueue_t
{
DECLARE_COMMON_IOQUEUE
unsigned max, count;
//pj_ioqueue_key_t hlist;
pj_ioqueue_key_t active_list;
int epfd;
//struct epoll_event *events;
//struct queue *queue;
#if PJ_IOQUEUE_HAS_SAFE_UNREG
pj_mutex_t *ref_cnt_mutex;
pj_ioqueue_key_t closing_list;
pj_ioqueue_key_t free_list;
#endif
};
/* Include implementation for common abstraction after we declare
* pj_ioqueue_key_t and pj_ioqueue_t.
*/
#include "ioqueue_common_abs.c"
#if PJ_IOQUEUE_HAS_SAFE_UNREG
/* Scan closing keys to be put to free list again */
static void scan_closing_keys(pj_ioqueue_t *ioqueue);
#endif
/*
* pj_ioqueue_name()
*/
PJ_DEF(const char*) pj_ioqueue_name(void)
{
#if defined(PJ_LINUX_KERNEL) && PJ_LINUX_KERNEL!=0
return "epoll-kernel";
#else
return "epoll";
#endif
}
/*
* pj_ioqueue_create()
*
* Create select ioqueue.
*/
PJ_DEF(pj_status_t) pj_ioqueue_create( pj_pool_t *pool,
pj_size_t max_fd,
pj_ioqueue_t **p_ioqueue)
{
pj_ioqueue_t *ioqueue;
pj_status_t rc;
pj_lock_t *lock;
int i;
/* Check that arguments are valid. */
PJ_ASSERT_RETURN(pool != NULL && p_ioqueue != NULL &&
max_fd > 0, PJ_EINVAL);
/* Check that size of pj_ioqueue_op_key_t is sufficient */
PJ_ASSERT_RETURN(sizeof(pj_ioqueue_op_key_t)-sizeof(void*) >=
sizeof(union operation_key), PJ_EBUG);
ioqueue = pj_pool_alloc(pool, sizeof(pj_ioqueue_t));
ioqueue_init(ioqueue);
ioqueue->max = max_fd;
ioqueue->count = 0;
pj_list_init(&ioqueue->active_list);
#if PJ_IOQUEUE_HAS_SAFE_UNREG
/* When safe unregistration is used (the default), we pre-create
* all keys and put them in the free list.
*/
/* Mutex to protect key's reference counter
* We don't want to use key's mutex or ioqueue's mutex because
* that would create deadlock situation in some cases.
*/
rc = pj_mutex_create_simple(pool, NULL, &ioqueue->ref_cnt_mutex);
if (rc != PJ_SUCCESS)
return rc;
/* Init key list */
pj_list_init(&ioqueue->free_list);
pj_list_init(&ioqueue->closing_list);
/* Pre-create all keys according to max_fd */
for ( i=0; i<max_fd; ++i) {
pj_ioqueue_key_t *key;
key = PJ_POOL_ALLOC_T(pool, pj_ioqueue_key_t);
key->ref_count = 0;
rc = pj_lock_create_recursive_mutex(pool, NULL, &key->lock);
if (rc != PJ_SUCCESS) {
key = ioqueue->free_list.next;
while (key != &ioqueue->free_list) {
pj_lock_destroy(key->lock);
key = key->next;
}
pj_mutex_destroy(ioqueue->ref_cnt_mutex);
return rc;
}
pj_list_push_back(&ioqueue->free_list, key);
}
#endif
rc = pj_lock_create_simple_mutex(pool, "ioq%p", &lock);
if (rc != PJ_SUCCESS)
return rc;
rc = pj_ioqueue_set_lock(ioqueue, lock, PJ_TRUE);
if (rc != PJ_SUCCESS)
return rc;
ioqueue->epfd = os_epoll_create(max_fd);
if (ioqueue->epfd < 0) {
ioqueue_destroy(ioqueue);
return PJ_RETURN_OS_ERROR(pj_get_native_os_error());
}
/*ioqueue->events = pj_pool_calloc(pool, max_fd, sizeof(struct epoll_event));
PJ_ASSERT_RETURN(ioqueue->events != NULL, PJ_ENOMEM);
ioqueue->queue = pj_pool_calloc(pool, max_fd, sizeof(struct queue));
PJ_ASSERT_RETURN(ioqueue->queue != NULL, PJ_ENOMEM);
*/
PJ_LOG(4, ("pjlib", "epoll I/O Queue created (%p)", ioqueue));
*p_ioqueue = ioqueue;
return PJ_SUCCESS;
}
/*
* pj_ioqueue_destroy()
*
* Destroy ioqueue.
*/
PJ_DEF(pj_status_t) pj_ioqueue_destroy(pj_ioqueue_t *ioqueue)
{
pj_ioqueue_key_t *key;
PJ_ASSERT_RETURN(ioqueue, PJ_EINVAL);
PJ_ASSERT_RETURN(ioqueue->epfd > 0, PJ_EINVALIDOP);
pj_lock_acquire(ioqueue->lock);
os_close(ioqueue->epfd);
ioqueue->epfd = 0;
#if PJ_IOQUEUE_HAS_SAFE_UNREG
/* Destroy reference counters */
key = ioqueue->active_list.next;
while (key != &ioqueue->active_list) {
pj_lock_destroy(key->lock);
key = key->next;
}
key = ioqueue->closing_list.next;
while (key != &ioqueue->closing_list) {
pj_lock_destroy(key->lock);
key = key->next;
}
key = ioqueue->free_list.next;
while (key != &ioqueue->free_list) {
pj_lock_destroy(key->lock);
key = key->next;
}
pj_mutex_destroy(ioqueue->ref_cnt_mutex);
#endif
return ioqueue_destroy(ioqueue);
}
/*
* pj_ioqueue_register_sock()
*
* Register a socket to ioqueue.
*/
PJ_DEF(pj_status_t) pj_ioqueue_register_sock2(pj_pool_t *pool,
pj_ioqueue_t *ioqueue,
pj_sock_t sock,
pj_grp_lock_t *grp_lock,
void *user_data,
const pj_ioqueue_callback *cb,
pj_ioqueue_key_t **p_key)
{
pj_ioqueue_key_t *key = NULL;
pj_uint32_t value;
struct epoll_event ev;
int status;
pj_status_t rc = PJ_SUCCESS;
PJ_ASSERT_RETURN(pool && ioqueue && sock != PJ_INVALID_SOCKET &&
cb && p_key, PJ_EINVAL);
pj_lock_acquire(ioqueue->lock);
if (ioqueue->count >= ioqueue->max) {
rc = PJ_ETOOMANY;
TRACE_((THIS_FILE, "pj_ioqueue_register_sock error: too many files"));
goto on_return;
}
/* Set socket to nonblocking. */
value = 1;
if ((rc=os_ioctl(sock, FIONBIO, (ioctl_val_type)&value))) {
TRACE_((THIS_FILE, "pj_ioqueue_register_sock error: ioctl rc=%d",
rc));
rc = pj_get_netos_error();
goto on_return;
}
/* If safe unregistration (PJ_IOQUEUE_HAS_SAFE_UNREG) is used, get
* the key from the free list. Otherwise allocate a new one.
*/
#if PJ_IOQUEUE_HAS_SAFE_UNREG
/* Scan closing_keys first to let them come back to free_list */
scan_closing_keys(ioqueue);
pj_assert(!pj_list_empty(&ioqueue->free_list));
if (pj_list_empty(&ioqueue->free_list)) {
rc = PJ_ETOOMANY;
goto on_return;
}
key = ioqueue->free_list.next;
pj_list_erase(key);
#else
/* Create key. */
key = (pj_ioqueue_key_t*)pj_pool_zalloc(pool, sizeof(pj_ioqueue_key_t));
#endif
rc = ioqueue_init_key(pool, ioqueue, key, sock, grp_lock, user_data, cb);
if (rc != PJ_SUCCESS) {
key = NULL;
goto on_return;
}
/* Create key's mutex */
/* rc = pj_mutex_create_recursive(pool, NULL, &key->mutex);
if (rc != PJ_SUCCESS) {
key = NULL;
goto on_return;
}
*/
/* os_epoll_ctl. */
ev.events = EPOLLIN | EPOLLERR;
ev.epoll_data = (epoll_data_type)key;
status = os_epoll_ctl(ioqueue->epfd, EPOLL_CTL_ADD, sock, &ev);
if (status < 0) {
rc = pj_get_os_error();
pj_lock_destroy(key->lock);
key = NULL;
TRACE_((THIS_FILE,
"pj_ioqueue_register_sock error: os_epoll_ctl rc=%d",
status));
goto on_return;
}
/* Register */
pj_list_insert_before(&ioqueue->active_list, key);
++ioqueue->count;
//TRACE_((THIS_FILE, "socket registered, count=%d", ioqueue->count));
on_return:
if (rc != PJ_SUCCESS) {
if (key && key->grp_lock)
pj_grp_lock_dec_ref_dbg(key->grp_lock, "ioqueue", 0);
}
*p_key = key;
pj_lock_release(ioqueue->lock);
return rc;
}
PJ_DEF(pj_status_t) pj_ioqueue_register_sock( pj_pool_t *pool,
pj_ioqueue_t *ioqueue,
pj_sock_t sock,
void *user_data,
const pj_ioqueue_callback *cb,
pj_ioqueue_key_t **p_key)
{
return pj_ioqueue_register_sock2(pool, ioqueue, sock, NULL, user_data,
cb, p_key);
}
#if PJ_IOQUEUE_HAS_SAFE_UNREG
/* Increment key's reference counter */
static void increment_counter(pj_ioqueue_key_t *key)
{
pj_mutex_lock(key->ioqueue->ref_cnt_mutex);
++key->ref_count;
pj_mutex_unlock(key->ioqueue->ref_cnt_mutex);
}
/* Decrement the key's reference counter, and when the counter reach zero,
* destroy the key.
*
* Note: MUST NOT CALL THIS FUNCTION WHILE HOLDING ioqueue's LOCK.
*/
static void decrement_counter(pj_ioqueue_key_t *key)
{
pj_lock_acquire(key->ioqueue->lock);
pj_mutex_lock(key->ioqueue->ref_cnt_mutex);
--key->ref_count;
if (key->ref_count == 0) {
pj_assert(key->closing == 1);
pj_gettickcount(&key->free_time);
key->free_time.msec += PJ_IOQUEUE_KEY_FREE_DELAY;
pj_time_val_normalize(&key->free_time);
pj_list_erase(key);
pj_list_push_back(&key->ioqueue->closing_list, key);
}
pj_mutex_unlock(key->ioqueue->ref_cnt_mutex);
pj_lock_release(key->ioqueue->lock);
}
#endif
/*
* pj_ioqueue_unregister()
*
* Unregister handle from ioqueue.
*/
PJ_DEF(pj_status_t) pj_ioqueue_unregister( pj_ioqueue_key_t *key)
{
pj_ioqueue_t *ioqueue;
struct epoll_event ev;
int status;
PJ_ASSERT_RETURN(key != NULL, PJ_EINVAL);
ioqueue = key->ioqueue;
/* Lock the key to make sure no callback is simultaneously modifying
* the key. We need to lock the key before ioqueue here to prevent
* deadlock.
*/
pj_ioqueue_lock_key(key);
/* Best effort to avoid double key-unregistration */
if (IS_CLOSING(key)) {
pj_ioqueue_unlock_key(key);
return PJ_SUCCESS;
}
/* Also lock ioqueue */
pj_lock_acquire(ioqueue->lock);
/* Avoid "negative" ioqueue count */
if (ioqueue->count > 0) {
--ioqueue->count;
} else {
/* If this happens, very likely there is double unregistration
* of a key.
*/
pj_assert(!"Bad ioqueue count in key unregistration!");
PJ_LOG(1,(THIS_FILE, "Bad ioqueue count in key unregistration!"));
}
#if !PJ_IOQUEUE_HAS_SAFE_UNREG
pj_list_erase(key);
#endif
ev.events = 0;
ev.epoll_data = (epoll_data_type)key;
status = os_epoll_ctl( ioqueue->epfd, EPOLL_CTL_DEL, key->fd, &ev);
if (status != 0) {
pj_status_t rc = pj_get_os_error();
pj_lock_release(ioqueue->lock);
pj_ioqueue_unlock_key(key);
return rc;
}
/* Destroy the key. */
pj_sock_close(key->fd);
pj_lock_release(ioqueue->lock);
#if PJ_IOQUEUE_HAS_SAFE_UNREG
/* Mark key is closing. */
key->closing = 1;
/* Decrement counter. */
decrement_counter(key);
/* Done. */
if (key->grp_lock) {
/* just dec_ref and unlock. we will set grp_lock to NULL
* elsewhere */
pj_grp_lock_t *grp_lock = key->grp_lock;
// Don't set grp_lock to NULL otherwise the other thread
// will crash. Just leave it as dangling pointer, but this
// should be safe
//key->grp_lock = NULL;
pj_grp_lock_dec_ref_dbg(grp_lock, "ioqueue", 0);
pj_grp_lock_release(grp_lock);
} else {
pj_ioqueue_unlock_key(key);
}
#else
if (key->grp_lock) {
/* set grp_lock to NULL and unlock */
pj_grp_lock_t *grp_lock = key->grp_lock;
// Don't set grp_lock to NULL otherwise the other thread
// will crash. Just leave it as dangling pointer, but this
// should be safe
//key->grp_lock = NULL;
pj_grp_lock_dec_ref_dbg(grp_lock, "ioqueue", 0);
pj_grp_lock_release(grp_lock);
} else {
pj_ioqueue_unlock_key(key);
}
pj_lock_destroy(key->lock);
#endif
return PJ_SUCCESS;
}
/* ioqueue_remove_from_set()
* This function is called from ioqueue_dispatch_event() to instruct
* the ioqueue to remove the specified descriptor from ioqueue's descriptor
* set for the specified event.
*/
static void ioqueue_remove_from_set( pj_ioqueue_t *ioqueue,
pj_ioqueue_key_t *key,
enum ioqueue_event_type event_type)
{
if (event_type == WRITEABLE_EVENT) {
struct epoll_event ev;
ev.events = EPOLLIN | EPOLLERR;
ev.epoll_data = (epoll_data_type)key;
os_epoll_ctl( ioqueue->epfd, EPOLL_CTL_MOD, key->fd, &ev);
}
}
/*
* ioqueue_add_to_set()
* This function is called from pj_ioqueue_recv(), pj_ioqueue_send() etc
* to instruct the ioqueue to add the specified handle to ioqueue's descriptor
* set for the specified event.
*/
static void ioqueue_add_to_set( pj_ioqueue_t *ioqueue,
pj_ioqueue_key_t *key,
enum ioqueue_event_type event_type )
{
if (event_type == WRITEABLE_EVENT) {
struct epoll_event ev;
ev.events = EPOLLIN | EPOLLOUT | EPOLLERR;
ev.epoll_data = (epoll_data_type)key;
os_epoll_ctl( ioqueue->epfd, EPOLL_CTL_MOD, key->fd, &ev);
}
}
#if PJ_IOQUEUE_HAS_SAFE_UNREG
/* Scan closing keys to be put to free list again */
static void scan_closing_keys(pj_ioqueue_t *ioqueue)
{
pj_time_val now;
pj_ioqueue_key_t *h;
pj_gettickcount(&now);
h = ioqueue->closing_list.next;
while (h != &ioqueue->closing_list) {
pj_ioqueue_key_t *next = h->next;
pj_assert(h->closing != 0);
if (PJ_TIME_VAL_GTE(now, h->free_time)) {
pj_list_erase(h);
// Don't set grp_lock to NULL otherwise the other thread
// will crash. Just leave it as dangling pointer, but this
// should be safe
//h->grp_lock = NULL;
pj_list_push_back(&ioqueue->free_list, h);
}
h = next;
}
}
#endif
/*
* pj_ioqueue_poll()
*
*/
PJ_DEF(int) pj_ioqueue_poll( pj_ioqueue_t *ioqueue, const pj_time_val *timeout)
{
int i, count, event_cnt, processed_cnt;
int msec;
//struct epoll_event *events = ioqueue->events;
//struct queue *queue = ioqueue->queue;
enum { MAX_EVENTS = PJ_IOQUEUE_MAX_CAND_EVENTS };
struct epoll_event events[MAX_EVENTS];
struct queue queue[MAX_EVENTS];
pj_timestamp t1, t2;
PJ_CHECK_STACK();
msec = timeout ? PJ_TIME_VAL_MSEC(*timeout) : 9000;
TRACE_((THIS_FILE, "start os_epoll_wait, msec=%d", msec));
pj_get_timestamp(&t1);
//count = os_epoll_wait( ioqueue->epfd, events, ioqueue->max, msec);
count = os_epoll_wait( ioqueue->epfd, events, MAX_EVENTS, msec);
if (count == 0) {
#if PJ_IOQUEUE_HAS_SAFE_UNREG
/* Check the closing keys only when there's no activity and when there are
* pending closing keys.
*/
if (count == 0 && !pj_list_empty(&ioqueue->closing_list)) {
pj_lock_acquire(ioqueue->lock);
scan_closing_keys(ioqueue);
pj_lock_release(ioqueue->lock);
}
#endif
TRACE_((THIS_FILE, "os_epoll_wait timed out"));
return count;
}
else if (count < 0) {
TRACE_((THIS_FILE, "os_epoll_wait error"));
return -pj_get_netos_error();
}
pj_get_timestamp(&t2);
TRACE_((THIS_FILE, "os_epoll_wait returns %d, time=%d usec",
count, pj_elapsed_usec(&t1, &t2)));
/* Lock ioqueue. */
pj_lock_acquire(ioqueue->lock);
for (event_cnt=0, i=0; i<count; ++i) {
pj_ioqueue_key_t *h = (pj_ioqueue_key_t*)(epoll_data_type)
events[i].epoll_data;
TRACE_((THIS_FILE, "event %d: events=%d", i, events[i].events));
/*
* Check readability.
*/
if ((events[i].events & EPOLLIN) &&
(key_has_pending_read(h) || key_has_pending_accept(h)) && !IS_CLOSING(h) ) {
#if PJ_IOQUEUE_HAS_SAFE_UNREG
increment_counter(h);
#endif
queue[event_cnt].key = h;
queue[event_cnt].event_type = READABLE_EVENT;
++event_cnt;
continue;
}
/*
* Check for writeability.
*/
if ((events[i].events & EPOLLOUT) && key_has_pending_write(h) && !IS_CLOSING(h)) {
#if PJ_IOQUEUE_HAS_SAFE_UNREG
increment_counter(h);
#endif
queue[event_cnt].key = h;
queue[event_cnt].event_type = WRITEABLE_EVENT;
++event_cnt;
continue;
}
#if PJ_HAS_TCP
/*
* Check for completion of connect() operation.
*/
if ((events[i].events & EPOLLOUT) && (h->connecting) && !IS_CLOSING(h)) {
#if PJ_IOQUEUE_HAS_SAFE_UNREG
increment_counter(h);
#endif
queue[event_cnt].key = h;
queue[event_cnt].event_type = WRITEABLE_EVENT;
++event_cnt;
continue;
}
#endif /* PJ_HAS_TCP */
/*
* Check for error condition.
*/
if ((events[i].events & EPOLLERR) && !IS_CLOSING(h)) {
/*
* We need to handle this exception event. If it's related to us
* connecting, report it as such. If not, just report it as a
* read event and the higher layers will handle it.
*/
if (h->connecting) {
#if PJ_IOQUEUE_HAS_SAFE_UNREG
increment_counter(h);
#endif
queue[event_cnt].key = h;
queue[event_cnt].event_type = EXCEPTION_EVENT;
++event_cnt;
} else if (key_has_pending_read(h) || key_has_pending_accept(h)) {
#if PJ_IOQUEUE_HAS_SAFE_UNREG
increment_counter(h);
#endif
queue[event_cnt].key = h;
queue[event_cnt].event_type = READABLE_EVENT;
++event_cnt;
}
continue;
}
}
for (i=0; i<event_cnt; ++i) {
if (queue[i].key->grp_lock)
pj_grp_lock_add_ref_dbg(queue[i].key->grp_lock, "ioqueue", 0);
}
PJ_RACE_ME(5);
pj_lock_release(ioqueue->lock);
PJ_RACE_ME(5);
processed_cnt = 0;
/* Now process the events. */
for (i=0; i<event_cnt; ++i) {
/* Just do not exceed PJ_IOQUEUE_MAX_EVENTS_IN_SINGLE_POLL */
if (processed_cnt < PJ_IOQUEUE_MAX_EVENTS_IN_SINGLE_POLL) {
switch (queue[i].event_type) {
case READABLE_EVENT:
if (ioqueue_dispatch_read_event(ioqueue, queue[i].key))
++processed_cnt;
break;
case WRITEABLE_EVENT:
if (ioqueue_dispatch_write_event(ioqueue, queue[i].key))
++processed_cnt;
break;
case EXCEPTION_EVENT:
if (ioqueue_dispatch_exception_event(ioqueue, queue[i].key))
++processed_cnt;
break;
case NO_EVENT:
pj_assert(!"Invalid event!");
break;
}
}
#if PJ_IOQUEUE_HAS_SAFE_UNREG
decrement_counter(queue[i].key);
#endif
if (queue[i].key->grp_lock)
pj_grp_lock_dec_ref_dbg(queue[i].key->grp_lock,
"ioqueue", 0);
}
/* Special case:
* When epoll returns > 0 but no descriptors are actually set!
*/
if (count > 0 && !event_cnt && msec > 0) {
pj_thread_sleep(msec);
}
TRACE_((THIS_FILE, " poll: count=%d events=%d processed=%d",
count, event_cnt, processed_cnt));
pj_get_timestamp(&t1);
TRACE_((THIS_FILE, "ioqueue_poll() returns %d, time=%d usec",
processed, pj_elapsed_usec(&t2, &t1)));
return processed_cnt;
}
|