| 1 | /* |
| 2 | * Copyright (c) 2000-2007 Apple Inc. All rights reserved. |
| 3 | * |
| 4 | * @APPLE_OSREFERENCE_LICENSE_HEADER_START@ |
| 5 | * |
| 6 | * This file contains Original Code and/or Modifications of Original Code |
| 7 | * as defined in and that are subject to the Apple Public Source License |
| 8 | * Version 2.0 (the 'License'). You may not use this file except in |
| 9 | * compliance with the License. The rights granted to you under the License |
| 10 | * may not be used to create, or enable the creation or redistribution of, |
| 11 | * unlawful or unlicensed copies of an Apple operating system, or to |
| 12 | * circumvent, violate, or enable the circumvention or violation of, any |
| 13 | * terms of an Apple operating system software license agreement. |
| 14 | * |
| 15 | * Please obtain a copy of the License at |
| 16 | * http://www.opensource.apple.com/apsl/ and read it before using this file. |
| 17 | * |
| 18 | * The Original Code and all software distributed under the License are |
| 19 | * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER |
| 20 | * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES, |
| 21 | * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY, |
| 22 | * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT. |
| 23 | * Please see the License for the specific language governing rights and |
| 24 | * limitations under the License. |
| 25 | * |
| 26 | * @APPLE_OSREFERENCE_LICENSE_HEADER_END@ |
| 27 | */ |
| 28 | /* |
| 29 | * @OSF_FREE_COPYRIGHT@ |
| 30 | */ |
| 31 | /* |
| 32 | * Mach Operating System |
| 33 | * Copyright (c) 1991,1990,1989 Carnegie Mellon University |
| 34 | * All Rights Reserved. |
| 35 | * |
| 36 | * Permission to use, copy, modify and distribute this software and its |
| 37 | * documentation is hereby granted, provided that both the copyright |
| 38 | * notice and this permission notice appear in all copies of the |
| 39 | * software, derivative works or modified versions, and any portions |
| 40 | * thereof, and that both notices appear in supporting documentation. |
| 41 | * |
| 42 | * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS" |
| 43 | * CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND FOR |
| 44 | * ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE. |
| 45 | * |
| 46 | * Carnegie Mellon requests users of this software to return to |
| 47 | * |
| 48 | * Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU |
| 49 | * School of Computer Science |
| 50 | * Carnegie Mellon University |
| 51 | * Pittsburgh PA 15213-3890 |
| 52 | * |
| 53 | * any improvements or extensions that they make and grant Carnegie Mellon |
| 54 | * the rights to redistribute these changes. |
| 55 | */ |
| 56 | /* |
| 57 | */ |
| 58 | /* |
| 59 | * File: ipc/ipc_mqueue.c |
| 60 | * Author: Rich Draves |
| 61 | * Date: 1989 |
| 62 | * |
| 63 | * Functions to manipulate IPC message queues. |
| 64 | */ |
| 65 | /* |
| 66 | * NOTICE: This file was modified by SPARTA, Inc. in 2006 to introduce |
| 67 | * support for mandatory and extensible security protections. This notice |
| 68 | * is included in support of clause 2.2 (b) of the Apple Public License, |
| 69 | * Version 2.0. |
| 70 | */ |
| 71 | |
| 72 | |
| 73 | #include <mach/port.h> |
| 74 | #include <mach/message.h> |
| 75 | #include <mach/sync_policy.h> |
| 76 | |
| 77 | #include <kern/assert.h> |
| 78 | #include <kern/counter.h> |
| 79 | #include <kern/sched_prim.h> |
| 80 | #include <kern/ipc_kobject.h> |
| 81 | #include <kern/ipc_mig.h> /* XXX - for mach_msg_receive_continue */ |
| 82 | #include <kern/misc_protos.h> |
| 83 | #include <kern/task.h> |
| 84 | #include <kern/thread.h> |
| 85 | #include <kern/waitq.h> |
| 86 | |
| 87 | #include <ipc/port.h> |
| 88 | #include <ipc/ipc_mqueue.h> |
| 89 | #include <ipc/ipc_kmsg.h> |
| 90 | #include <ipc/ipc_right.h> |
| 91 | #include <ipc/ipc_port.h> |
| 92 | #include <ipc/ipc_pset.h> |
| 93 | #include <ipc/ipc_space.h> |
| 94 | |
| 95 | #if MACH_FLIPC |
| 96 | #include <ipc/flipc.h> |
| 97 | #endif |
| 98 | |
| 99 | #ifdef __LP64__ |
| 100 | #include <vm/vm_map.h> |
| 101 | #endif |
| 102 | |
| 103 | #include <sys/event.h> |
| 104 | |
| 105 | extern char *proc_name_address(void *p); |
| 106 | |
| 107 | int ipc_mqueue_full; /* address is event for queue space */ |
| 108 | int ipc_mqueue_rcv; /* address is event for message arrival */ |
| 109 | |
| 110 | /* forward declarations */ |
| 111 | static void ipc_mqueue_receive_results(wait_result_t result); |
| 112 | |
| 113 | #if MACH_FLIPC |
| 114 | static void ipc_mqueue_peek_on_thread_locked( |
| 115 | ipc_mqueue_t port_mq, |
| 116 | mach_msg_option64_t option, |
| 117 | thread_t thread); |
| 118 | #endif /* MACH_FLIPC */ |
| 119 | |
| 120 | /* Deliver message to message queue or waiting receiver */ |
| 121 | static void ipc_mqueue_post( |
| 122 | ipc_mqueue_t mqueue, |
| 123 | ipc_kmsg_t kmsg, |
| 124 | mach_msg_option_t option); |
| 125 | |
| 126 | /* |
| 127 | * Routine: ipc_mqueue_init |
| 128 | * Purpose: |
| 129 | * Initialize a newly-allocated message queue. |
| 130 | */ |
| 131 | void |
| 132 | ipc_mqueue_init( |
| 133 | ipc_mqueue_t mqueue) |
| 134 | { |
| 135 | ipc_kmsg_queue_init(&mqueue->imq_messages); |
| 136 | mqueue->imq_qlimit = MACH_PORT_QLIMIT_DEFAULT; |
| 137 | klist_init(list: &mqueue->imq_klist); |
| 138 | } |
| 139 | |
| 140 | /* |
| 141 | * Routine: ipc_mqueue_add_locked. |
| 142 | * Purpose: |
| 143 | * Associate the portset's mqueue with the port's mqueue. |
| 144 | * This has to be done so that posting the port will wakeup |
| 145 | * a portset waiter. If there are waiters on the portset |
| 146 | * mqueue and messages on the port mqueue, try to match them |
| 147 | * up now. |
| 148 | * Conditions: |
| 149 | * Port and Pset both locked. |
| 150 | */ |
| 151 | kern_return_t |
| 152 | ipc_mqueue_add_locked( |
| 153 | ipc_mqueue_t port_mqueue, |
| 154 | ipc_pset_t pset, |
| 155 | waitq_link_t *linkp) |
| 156 | { |
| 157 | ipc_port_t port = ip_from_mq(port_mqueue); |
| 158 | struct waitq_set *wqset = &pset->ips_wqset; |
| 159 | circle_queue_t kmsgq = &port_mqueue->imq_messages; |
| 160 | kern_return_t kr = KERN_SUCCESS; |
| 161 | ipc_kmsg_t kmsg; |
| 162 | |
| 163 | kr = waitq_link_locked(waitq: &port->ip_waitq, wqset, link: linkp); |
| 164 | if (kr != KERN_SUCCESS) { |
| 165 | return kr; |
| 166 | } |
| 167 | |
| 168 | /* |
| 169 | * Now that the set has been added to the port, there may be |
| 170 | * messages queued on the port and threads waiting on the set |
| 171 | * waitq. Lets get them together. |
| 172 | * |
| 173 | * Only consider this set however, as the other ones have been |
| 174 | * posted to already. |
| 175 | */ |
| 176 | while ((kmsg = ipc_kmsg_queue_first(kmsgq)) != IKM_NULL) { |
| 177 | thread_t th; |
| 178 | mach_msg_size_t msize, asize; |
| 179 | |
| 180 | th = waitq_wakeup64_identify_locked(waitq: wqset, IPC_MQUEUE_RECEIVE, |
| 181 | THREAD_AWAKENED, flags: WAITQ_KEEP_LOCKED); |
| 182 | /* port and pset still locked, thread not runnable */ |
| 183 | |
| 184 | if (th == THREAD_NULL) { |
| 185 | /* |
| 186 | * Didn't find a thread to wake up but messages |
| 187 | * are enqueued, prepost the set instead, |
| 188 | * as calling waitq_wakeup64_identify_locked() |
| 189 | * on the set directly will not take care of it. |
| 190 | */ |
| 191 | waitq_link_prepost_locked(waitq: &port->ip_waitq, wqset); |
| 192 | break; |
| 193 | } |
| 194 | |
| 195 | /* |
| 196 | * Because we hold the thread off the runqueue at this point, |
| 197 | * it's safe to modify ith_ fields on the thread, as |
| 198 | * until it is resumed, it must be off core or in between |
| 199 | * the assert wait and returning from the continuation. |
| 200 | */ |
| 201 | |
| 202 | /* |
| 203 | * If the receiver waited with a facility not directly |
| 204 | * related to Mach messaging, then it isn't prepared to get |
| 205 | * handed the message directly. Just set it running, and |
| 206 | * go look for another thread that can. |
| 207 | */ |
| 208 | if (th->ith_state != MACH_RCV_IN_PROGRESS) { |
| 209 | #if MACH_FLIPC |
| 210 | if (th->ith_state == MACH_PEEK_IN_PROGRESS) { |
| 211 | /* |
| 212 | * wakeup the peeking thread, but |
| 213 | * continue to loop over the threads |
| 214 | * waiting on the port's mqueue to see |
| 215 | * if there are any actual receivers |
| 216 | */ |
| 217 | ipc_mqueue_peek_on_thread_locked(port_mqueue, |
| 218 | th->ith_option, th); |
| 219 | } |
| 220 | #endif /* MACH_FLIPC */ |
| 221 | |
| 222 | waitq_resume_identified_thread(waitq: wqset, thread: th, |
| 223 | THREAD_AWAKENED, flags: WAITQ_WAKEUP_DEFAULT); |
| 224 | continue; |
| 225 | } |
| 226 | |
| 227 | /* |
| 228 | * Found a receiver. see if they can handle the message |
| 229 | * correctly (the message is not too large for them, or |
| 230 | * they didn't care to be informed that the message was |
| 231 | * too large). If they can't handle it, take them off |
| 232 | * the list and let them go back and figure it out and |
| 233 | * just move onto the next. |
| 234 | */ |
| 235 | msize = ipc_kmsg_copyout_size(kmsg, map: th->map); |
| 236 | asize = ipc_kmsg_aux_data_size(kmsg); |
| 237 | |
| 238 | if (ipc_kmsg_too_large(msg_size: msize, aux_size: asize, options: th->ith_option, |
| 239 | max_msg_size: th->ith_max_msize, max_aux_size: th->ith_max_asize, receiver: th)) { |
| 240 | th->ith_state = MACH_RCV_TOO_LARGE; |
| 241 | th->ith_msize = msize; |
| 242 | th->ith_asize = asize; |
| 243 | if (th->ith_option & MACH_RCV_LARGE) { |
| 244 | /* |
| 245 | * let him go without message |
| 246 | */ |
| 247 | th->ith_receiver_name = port_mqueue->imq_receiver_name; |
| 248 | th->ith_kmsg = IKM_NULL; |
| 249 | th->ith_seqno = 0; |
| 250 | |
| 251 | waitq_resume_identified_thread(waitq: wqset, thread: th, |
| 252 | THREAD_AWAKENED, flags: WAITQ_WAKEUP_DEFAULT); |
| 253 | |
| 254 | continue; /* find another thread */ |
| 255 | } |
| 256 | } else { |
| 257 | th->ith_state = MACH_MSG_SUCCESS; |
| 258 | } |
| 259 | |
| 260 | /* |
| 261 | * This thread is going to take this message, |
| 262 | * so give it the message. |
| 263 | */ |
| 264 | ipc_kmsg_rmqueue(kmsgq, kmsg); |
| 265 | |
| 266 | #if MACH_FLIPC |
| 267 | mach_node_t node = kmsg->ikm_node; |
| 268 | #endif |
| 269 | |
| 270 | ipc_mqueue_release_msgcount(port_mq: port_mqueue); |
| 271 | |
| 272 | th->ith_kmsg = kmsg; |
| 273 | th->ith_seqno = port_mqueue->imq_seqno++; |
| 274 | |
| 275 | waitq_resume_identified_thread(waitq: wqset, thread: th, |
| 276 | THREAD_AWAKENED, flags: WAITQ_WAKEUP_DEFAULT); |
| 277 | |
| 278 | #if MACH_FLIPC |
| 279 | if (MACH_NODE_VALID(node) && FPORT_VALID(port_mqueue->imq_fport)) { |
| 280 | flipc_msg_ack(node, port_mqueue, TRUE); |
| 281 | } |
| 282 | #endif |
| 283 | } |
| 284 | |
| 285 | return KERN_SUCCESS; |
| 286 | } |
| 287 | |
| 288 | |
| 289 | /* |
| 290 | * Routine: ipc_port_has_klist |
| 291 | * Purpose: |
| 292 | * Returns whether the given port imq_klist field can be used as a klist. |
| 293 | */ |
| 294 | bool |
| 295 | ipc_port_has_klist(ipc_port_t port) |
| 296 | { |
| 297 | return !port->ip_specialreply && |
| 298 | port->ip_sync_link_state == PORT_SYNC_LINK_ANY; |
| 299 | } |
| 300 | |
| 301 | static inline struct klist * |
| 302 | ipc_object_klist(ipc_object_t object) |
| 303 | { |
| 304 | if (io_otype(object) == IOT_PORT) { |
| 305 | ipc_port_t port = ip_object_to_port(object); |
| 306 | |
| 307 | return ipc_port_has_klist(port) ? &port->ip_klist : NULL; |
| 308 | } |
| 309 | return &ips_object_to_pset(object)->ips_klist; |
| 310 | } |
| 311 | |
| 312 | /* |
| 313 | * Routine: ipc_mqueue_changed |
| 314 | * Purpose: |
| 315 | * Wake up receivers waiting in a message queue. |
| 316 | * Conditions: |
| 317 | * The object containing the message queue is locked. |
| 318 | */ |
| 319 | void |
| 320 | ipc_mqueue_changed( |
| 321 | ipc_space_t space, |
| 322 | struct waitq *waitq) |
| 323 | { |
| 324 | ipc_object_t object = io_from_waitq(waitq); |
| 325 | struct klist *klist = ipc_object_klist(object); |
| 326 | |
| 327 | if (klist && SLIST_FIRST(klist)) { |
| 328 | /* |
| 329 | * Indicate that this message queue is vanishing |
| 330 | * |
| 331 | * When this is called, the associated receive right may be in flight |
| 332 | * between two tasks: the one it used to live in, and the one that armed |
| 333 | * a port destroyed notification for it. |
| 334 | * |
| 335 | * The new process may want to register the port it gets back with an |
| 336 | * EVFILT_MACHPORT filter again, and may have pending sync IPC on this |
| 337 | * port pending already, in which case we want the imq_klist field to be |
| 338 | * reusable for nefarious purposes. |
| 339 | * |
| 340 | * Fortunately, we really don't need this linkage anymore after this |
| 341 | * point as EV_VANISHED / EV_EOF will be the last thing delivered ever. |
| 342 | * |
| 343 | * Note: we don't have the space lock here, however, this covers the |
| 344 | * case of when a task is terminating the space, triggering |
| 345 | * several knote_vanish() calls. |
| 346 | * |
| 347 | * We don't need the lock to observe that the space is inactive as |
| 348 | * we just deactivated it on the same thread. |
| 349 | * |
| 350 | * We still need to call knote_vanish() so that the knote is |
| 351 | * marked with EV_VANISHED or EV_EOF so that the detach step |
| 352 | * in filt_machportdetach is skipped correctly. |
| 353 | */ |
| 354 | assert(space); |
| 355 | knote_vanish(list: klist, is_active(space)); |
| 356 | } |
| 357 | |
| 358 | if (io_otype(object) == IOT_PORT) { |
| 359 | ipc_port_t port = ip_object_to_port(object); |
| 360 | if (!port->ip_specialreply) { |
| 361 | ipc_port_adjust_sync_link_state_locked(port, |
| 362 | PORT_SYNC_LINK_ANY, NULL); |
| 363 | } |
| 364 | } else { |
| 365 | klist_init(list: klist); |
| 366 | } |
| 367 | |
| 368 | /* |
| 369 | * do not pass WAITQ_UPDATE_INHERITOR, ipc_port_destroy() |
| 370 | * needs to handle this manually, and the port lock |
| 371 | * is the waitq lock, so there's really no inefficiency there. |
| 372 | */ |
| 373 | waitq_wakeup64_all_locked(waitq: waitq, IPC_MQUEUE_RECEIVE, |
| 374 | THREAD_RESTART, flags: WAITQ_KEEP_LOCKED); |
| 375 | } |
| 376 | |
| 377 | |
| 378 | |
| 379 | |
| 380 | /* |
| 381 | * Routine: ipc_mqueue_send |
| 382 | * Purpose: |
| 383 | * Send a message to a message queue. The message holds a reference |
| 384 | * for the destination port for this message queue in the |
| 385 | * msgh_remote_port field. |
| 386 | * |
| 387 | * If unsuccessful, the caller still has possession of |
| 388 | * the message and must do something with it. If successful, |
| 389 | * the message is queued, given to a receiver, or destroyed. |
| 390 | * Conditions: |
| 391 | * port is locked. |
| 392 | * Returns: |
| 393 | * MACH_MSG_SUCCESS The message was accepted. |
| 394 | * MACH_SEND_TIMED_OUT Caller still has message. |
| 395 | * MACH_SEND_INTERRUPTED Caller still has message. |
| 396 | */ |
| 397 | mach_msg_return_t |
| 398 | ipc_mqueue_send_locked( |
| 399 | ipc_mqueue_t mqueue, |
| 400 | ipc_kmsg_t kmsg, |
| 401 | mach_msg_option_t option, |
| 402 | mach_msg_timeout_t send_timeout) |
| 403 | { |
| 404 | ipc_port_t port = ip_from_mq(mqueue); |
| 405 | int wresult; |
| 406 | |
| 407 | /* |
| 408 | * Don't block if: |
| 409 | * 1) We're under the queue limit. |
| 410 | * 2) Caller used the MACH_SEND_ALWAYS internal option. |
| 411 | * 3) Message is sent to a send-once right. |
| 412 | */ |
| 413 | if (!imq_full(mqueue) || |
| 414 | (!imq_full_kernel(mqueue) && |
| 415 | ((option & MACH_SEND_ALWAYS) || |
| 416 | (MACH_MSGH_BITS_REMOTE(ikm_header(kmsg)->msgh_bits) == |
| 417 | MACH_MSG_TYPE_PORT_SEND_ONCE)))) { |
| 418 | mqueue->imq_msgcount++; |
| 419 | assert(mqueue->imq_msgcount > 0); |
| 420 | ip_mq_unlock(port); |
| 421 | } else { |
| 422 | thread_t cur_thread = current_thread(); |
| 423 | struct turnstile *send_turnstile = TURNSTILE_NULL; |
| 424 | uint64_t deadline; |
| 425 | |
| 426 | /* |
| 427 | * We have to wait for space to be granted to us. |
| 428 | */ |
| 429 | if ((option & MACH_SEND_TIMEOUT) && (send_timeout == 0)) { |
| 430 | ip_mq_unlock(port); |
| 431 | return MACH_SEND_TIMED_OUT; |
| 432 | } |
| 433 | if (imq_full_kernel(mqueue)) { |
| 434 | ip_mq_unlock(port); |
| 435 | return MACH_SEND_NO_BUFFER; |
| 436 | } |
| 437 | port->ip_fullwaiters = true; |
| 438 | |
| 439 | if (option & MACH_SEND_TIMEOUT) { |
| 440 | clock_interval_to_deadline(interval: send_timeout, scale_factor: 1000 * NSEC_PER_USEC, result: &deadline); |
| 441 | } else { |
| 442 | deadline = 0; |
| 443 | } |
| 444 | |
| 445 | thread_set_pending_block_hint(thread: cur_thread, block_hint: kThreadWaitPortSend); |
| 446 | |
| 447 | send_turnstile = turnstile_prepare(proprietor: (uintptr_t)port, |
| 448 | port_send_turnstile_address(port), |
| 449 | TURNSTILE_NULL, type: TURNSTILE_SYNC_IPC); |
| 450 | |
| 451 | ipc_port_send_update_inheritor(port, turnstile: send_turnstile, |
| 452 | flags: TURNSTILE_DELAYED_UPDATE); |
| 453 | |
| 454 | wresult = waitq_assert_wait64_leeway( |
| 455 | waitq: &send_turnstile->ts_waitq, |
| 456 | IPC_MQUEUE_FULL, |
| 457 | THREAD_ABORTSAFE, |
| 458 | TIMEOUT_URGENCY_USER_NORMAL, |
| 459 | deadline, |
| 460 | TIMEOUT_NO_LEEWAY); |
| 461 | |
| 462 | ip_mq_unlock(port); |
| 463 | turnstile_update_inheritor_complete(turnstile: send_turnstile, |
| 464 | flags: TURNSTILE_INTERLOCK_NOT_HELD); |
| 465 | |
| 466 | if (wresult == THREAD_WAITING) { |
| 467 | wresult = thread_block(THREAD_CONTINUE_NULL); |
| 468 | } |
| 469 | |
| 470 | /* Call turnstile complete with interlock held */ |
| 471 | ip_mq_lock(port); |
| 472 | turnstile_complete(proprietor: (uintptr_t)port, port_send_turnstile_address(port), NULL, type: TURNSTILE_SYNC_IPC); |
| 473 | ip_mq_unlock(port); |
| 474 | |
| 475 | /* Call cleanup after dropping the interlock */ |
| 476 | turnstile_cleanup(); |
| 477 | |
| 478 | switch (wresult) { |
| 479 | case THREAD_AWAKENED: |
| 480 | /* |
| 481 | * we can proceed - inherited msgcount from waker |
| 482 | * or the message queue has been destroyed and the msgcount |
| 483 | * has been reset to zero (will detect in ipc_mqueue_post()). |
| 484 | */ |
| 485 | break; |
| 486 | |
| 487 | case THREAD_TIMED_OUT: |
| 488 | assert(option & MACH_SEND_TIMEOUT); |
| 489 | return MACH_SEND_TIMED_OUT; |
| 490 | |
| 491 | case THREAD_INTERRUPTED: |
| 492 | return MACH_SEND_INTERRUPTED; |
| 493 | |
| 494 | case THREAD_RESTART: |
| 495 | /* mqueue is being destroyed */ |
| 496 | return MACH_SEND_INVALID_DEST; |
| 497 | default: |
| 498 | panic("ipc_mqueue_send" ); |
| 499 | } |
| 500 | } |
| 501 | |
| 502 | ipc_mqueue_post(mqueue, kmsg, option); |
| 503 | return MACH_MSG_SUCCESS; |
| 504 | } |
| 505 | |
| 506 | /* |
| 507 | * Routine: ipc_mqueue_override_send_locked |
| 508 | * Purpose: |
| 509 | * Set an override qos on the first message in the queue |
| 510 | * (if the queue is full). This is a send-possible override |
| 511 | * that will go away as soon as we drain a message from the |
| 512 | * queue. |
| 513 | * |
| 514 | * Conditions: |
| 515 | * The port corresponding to mqueue is locked. |
| 516 | * The caller holds a reference on the message queue. |
| 517 | */ |
| 518 | void |
| 519 | ipc_mqueue_override_send_locked( |
| 520 | ipc_mqueue_t mqueue, |
| 521 | mach_msg_qos_t qos_ovr) |
| 522 | { |
| 523 | ipc_port_t port = ip_from_mq(mqueue); |
| 524 | |
| 525 | assert(waitq_is_valid(&port->ip_waitq)); |
| 526 | |
| 527 | if (imq_full(mqueue)) { |
| 528 | ipc_kmsg_t first = ipc_kmsg_queue_first(&mqueue->imq_messages); |
| 529 | |
| 530 | if (first && ipc_kmsg_override_qos(queue: &mqueue->imq_messages, kmsg: first, qos_ovr)) { |
| 531 | if (ip_in_a_space(port) && |
| 532 | is_active(ip_get_receiver(port)) && |
| 533 | ipc_port_has_klist(port)) { |
| 534 | KNOTE(&port->ip_klist, 0); |
| 535 | } |
| 536 | } |
| 537 | } |
| 538 | } |
| 539 | |
| 540 | /* |
| 541 | * Routine: ipc_mqueue_release_msgcount |
| 542 | * Purpose: |
| 543 | * Release a message queue reference in the case where we |
| 544 | * found a waiter. |
| 545 | * |
| 546 | * Conditions: |
| 547 | * The port corresponding to message queue is locked. |
| 548 | * The message corresponding to this reference is off the queue. |
| 549 | * There is no need to pass reserved preposts because this will |
| 550 | * never prepost to anyone |
| 551 | */ |
| 552 | void |
| 553 | ipc_mqueue_release_msgcount(ipc_mqueue_t port_mq) |
| 554 | { |
| 555 | ipc_port_t port = ip_from_mq(port_mq); |
| 556 | struct turnstile *send_turnstile = port_send_turnstile(port); |
| 557 | |
| 558 | ip_mq_lock_held(port); |
| 559 | assert(port_mq->imq_msgcount > 1 || ipc_kmsg_queue_empty(&port_mq->imq_messages)); |
| 560 | |
| 561 | port_mq->imq_msgcount--; |
| 562 | |
| 563 | if (!imq_full(port_mq) && port->ip_fullwaiters && |
| 564 | send_turnstile != TURNSTILE_NULL) { |
| 565 | /* |
| 566 | * boost the priority of the awoken thread |
| 567 | * (WAITQ_PROMOTE_PRIORITY) to ensure it uses |
| 568 | * the message queue slot we've just reserved. |
| 569 | * |
| 570 | * NOTE: this will never prepost |
| 571 | * |
| 572 | * The wakeup happens on a turnstile waitq |
| 573 | * which will wakeup the highest priority waiter. |
| 574 | * A potential downside of this would be starving low |
| 575 | * priority senders if there is a constant churn of |
| 576 | * high priority threads trying to send to this port. |
| 577 | */ |
| 578 | if (waitq_wakeup64_one(waitq: &send_turnstile->ts_waitq, |
| 579 | IPC_MQUEUE_FULL, |
| 580 | THREAD_AWAKENED, |
| 581 | flags: WAITQ_PROMOTE_PRIORITY) != KERN_SUCCESS) { |
| 582 | port->ip_fullwaiters = false; |
| 583 | } else { |
| 584 | /* gave away our slot - add reference back */ |
| 585 | port_mq->imq_msgcount++; |
| 586 | } |
| 587 | } |
| 588 | |
| 589 | if (ipc_kmsg_queue_empty(&port_mq->imq_messages)) { |
| 590 | waitq_clear_prepost_locked(waitq: &port->ip_waitq); |
| 591 | } |
| 592 | } |
| 593 | |
| 594 | /* |
| 595 | * Routine: ipc_mqueue_post |
| 596 | * Purpose: |
| 597 | * Post a message to a waiting receiver or enqueue it. If a |
| 598 | * receiver is waiting, we can release our reserved space in |
| 599 | * the message queue. |
| 600 | * |
| 601 | * Conditions: |
| 602 | * port is unlocked |
| 603 | * If we need to queue, our space in the message queue is reserved. |
| 604 | */ |
| 605 | static void |
| 606 | ipc_mqueue_post( |
| 607 | ipc_mqueue_t mqueue, |
| 608 | ipc_kmsg_t kmsg, |
| 609 | mach_msg_option_t __unused option) |
| 610 | { |
| 611 | ipc_port_t port = ip_from_mq(mqueue); |
| 612 | struct waitq *waitq = &port->ip_waitq; |
| 613 | boolean_t destroy_msg = FALSE; |
| 614 | |
| 615 | ipc_kmsg_trace_send(kmsg, option); |
| 616 | |
| 617 | /* |
| 618 | * While the msg queue is locked, we have control of the |
| 619 | * kmsg, so the ref in it for the port is still good. |
| 620 | * |
| 621 | * Check for a receiver for the message. |
| 622 | */ |
| 623 | ip_mq_lock(port); |
| 624 | |
| 625 | /* we may have raced with port destruction! */ |
| 626 | if (!waitq_is_valid(wq: &port->ip_waitq)) { |
| 627 | destroy_msg = TRUE; |
| 628 | goto out_unlock; |
| 629 | } |
| 630 | |
| 631 | for (;;) { |
| 632 | thread_t receiver; |
| 633 | mach_msg_size_t msize, asize; |
| 634 | |
| 635 | receiver = waitq_wakeup64_identify_locked(waitq: waitq, |
| 636 | IPC_MQUEUE_RECEIVE, THREAD_AWAKENED, flags: WAITQ_KEEP_LOCKED); |
| 637 | /* waitq still locked, thread not runnable */ |
| 638 | |
| 639 | if (receiver == THREAD_NULL) { |
| 640 | /* |
| 641 | * no receivers; queue kmsg if space still reserved |
| 642 | * Reservations are cancelled when the port goes inactive. |
| 643 | * note that this will enqueue the message for any |
| 644 | * "peeking" receivers. |
| 645 | * |
| 646 | * Also, post the knote to wake up any threads waiting |
| 647 | * on that style of interface if this insertion is of |
| 648 | * note (first insertion, or adjusted override qos all |
| 649 | * the way to the head of the queue). |
| 650 | * |
| 651 | * This is just for ports. port-sets knotes are being |
| 652 | * posted to by the waitq_wakeup64_identify_locked() |
| 653 | * above already. |
| 654 | */ |
| 655 | if (mqueue->imq_msgcount == 0) { |
| 656 | /* |
| 657 | * The message queue must belong |
| 658 | * to an inactive port, so just destroy |
| 659 | * the message and pretend it was posted. |
| 660 | */ |
| 661 | destroy_msg = TRUE; |
| 662 | } else if (!ipc_kmsg_enqueue_qos(queue: &mqueue->imq_messages, kmsg)) { |
| 663 | /* |
| 664 | * queue was not empty and qos |
| 665 | * didn't change, nothing to do. |
| 666 | */ |
| 667 | } else if (ip_in_a_space(port) && |
| 668 | is_active(ip_get_receiver(port)) && |
| 669 | ipc_port_has_klist(port)) { |
| 670 | /* |
| 671 | * queue was empty or qos changed |
| 672 | * we need to tell kqueue, unless |
| 673 | * the space is getting torn down |
| 674 | */ |
| 675 | KNOTE(&port->ip_klist, 0); |
| 676 | } |
| 677 | break; |
| 678 | } |
| 679 | |
| 680 | #if MACH_FLIPC |
| 681 | /* |
| 682 | * If a thread is attempting a "peek" into the message queue |
| 683 | * (MACH_PEEK_IN_PROGRESS), then we enqueue the message and set the |
| 684 | * thread running. A successful peek is essentially the same as |
| 685 | * message delivery since the peeking thread takes responsibility |
| 686 | * for delivering the message and (eventually) removing it from |
| 687 | * the mqueue. Only one thread can successfully use the peek |
| 688 | * facility on any given port, so we exit the waitq loop after |
| 689 | * encountering such a thread. |
| 690 | */ |
| 691 | if (receiver->ith_state == MACH_PEEK_IN_PROGRESS && mqueue->imq_msgcount > 0) { |
| 692 | ipc_kmsg_enqueue_qos(&mqueue->imq_messages, kmsg); |
| 693 | ipc_mqueue_peek_on_thread_locked(mqueue, receiver->ith_option, receiver); |
| 694 | waitq_resume_identified_thread(waitq, receiver, |
| 695 | THREAD_AWAKENED, WAITQ_WAKEUP_DEFAULT); |
| 696 | break; /* Message was posted, so break out of loop */ |
| 697 | } |
| 698 | #endif /* MACH_FLIPC */ |
| 699 | |
| 700 | /* |
| 701 | * If the receiver waited with a facility not directly related |
| 702 | * to Mach messaging, then it isn't prepared to get handed the |
| 703 | * message directly. Just set it running, and go look for |
| 704 | * another thread that can. |
| 705 | */ |
| 706 | if (receiver->ith_state != MACH_RCV_IN_PROGRESS) { |
| 707 | waitq_resume_identified_thread(waitq: waitq, thread: receiver, |
| 708 | THREAD_AWAKENED, flags: WAITQ_WAKEUP_DEFAULT); |
| 709 | |
| 710 | continue; |
| 711 | } |
| 712 | |
| 713 | |
| 714 | /* |
| 715 | * We found a waiting thread. |
| 716 | * If the message is too large or the scatter list is too small |
| 717 | * the thread we wake up will get that as its status. |
| 718 | */ |
| 719 | msize = ipc_kmsg_copyout_size(kmsg, map: receiver->map); |
| 720 | asize = ipc_kmsg_aux_data_size(kmsg); |
| 721 | |
| 722 | if (ipc_kmsg_too_large(msg_size: msize, aux_size: asize, options: receiver->ith_option, |
| 723 | max_msg_size: receiver->ith_max_msize, max_aux_size: receiver->ith_max_asize, receiver)) { |
| 724 | receiver->ith_msize = msize; |
| 725 | receiver->ith_asize = asize; |
| 726 | receiver->ith_state = MACH_RCV_TOO_LARGE; |
| 727 | } else { |
| 728 | receiver->ith_state = MACH_MSG_SUCCESS; |
| 729 | } |
| 730 | |
| 731 | /* |
| 732 | * If there is no problem with the upcoming receive, or the |
| 733 | * receiver thread didn't specifically ask for special too |
| 734 | * large error condition, go ahead and select it anyway. |
| 735 | */ |
| 736 | if ((receiver->ith_state == MACH_MSG_SUCCESS) || |
| 737 | !(receiver->ith_option & MACH_RCV_LARGE)) { |
| 738 | receiver->ith_kmsg = kmsg; |
| 739 | receiver->ith_seqno = mqueue->imq_seqno++; |
| 740 | #if MACH_FLIPC |
| 741 | mach_node_t node = kmsg->ikm_node; |
| 742 | #endif |
| 743 | waitq_resume_identified_thread(waitq: waitq, thread: receiver, |
| 744 | THREAD_AWAKENED, flags: WAITQ_WAKEUP_DEFAULT); |
| 745 | |
| 746 | /* we didn't need our reserved spot in the queue */ |
| 747 | ipc_mqueue_release_msgcount(port_mq: mqueue); |
| 748 | |
| 749 | #if MACH_FLIPC |
| 750 | if (MACH_NODE_VALID(node) && FPORT_VALID(mqueue->imq_fport)) { |
| 751 | flipc_msg_ack(node, mqueue, TRUE); |
| 752 | } |
| 753 | #endif |
| 754 | break; |
| 755 | } |
| 756 | |
| 757 | /* |
| 758 | * Otherwise, this thread needs to be released to run |
| 759 | * and handle its error without getting the message. We |
| 760 | * need to go back and pick another one. |
| 761 | */ |
| 762 | receiver->ith_receiver_name = mqueue->imq_receiver_name; |
| 763 | receiver->ith_kmsg = IKM_NULL; |
| 764 | receiver->ith_seqno = 0; |
| 765 | |
| 766 | waitq_resume_identified_thread(waitq: waitq, thread: receiver, |
| 767 | THREAD_AWAKENED, flags: WAITQ_WAKEUP_DEFAULT); |
| 768 | } |
| 769 | |
| 770 | out_unlock: |
| 771 | /* clear the waitq boost we may have been given */ |
| 772 | waitq_clear_promotion_locked(waitq: waitq, thread: current_thread()); |
| 773 | waitq_unlock(wq: waitq); |
| 774 | |
| 775 | if (destroy_msg) { |
| 776 | ipc_kmsg_destroy(kmsg, flags: IPC_KMSG_DESTROY_ALL); |
| 777 | } |
| 778 | |
| 779 | counter_inc(¤t_task()->messages_sent); |
| 780 | return; |
| 781 | } |
| 782 | |
| 783 | |
| 784 | static void |
| 785 | ipc_mqueue_receive_results(wait_result_t saved_wait_result) |
| 786 | { |
| 787 | thread_t self = current_thread(); |
| 788 | mach_msg_option64_t option64 = self->ith_option; |
| 789 | |
| 790 | /* |
| 791 | * why did we wake up? |
| 792 | */ |
| 793 | switch (saved_wait_result) { |
| 794 | case THREAD_TIMED_OUT: |
| 795 | self->ith_state = MACH_RCV_TIMED_OUT; |
| 796 | return; |
| 797 | |
| 798 | case THREAD_INTERRUPTED: |
| 799 | self->ith_state = MACH_RCV_INTERRUPTED; |
| 800 | return; |
| 801 | |
| 802 | case THREAD_RESTART: |
| 803 | /* something bad happened to the port/set */ |
| 804 | self->ith_state = MACH_RCV_PORT_CHANGED; |
| 805 | return; |
| 806 | |
| 807 | case THREAD_AWAKENED: |
| 808 | /* |
| 809 | * We do not need to go select a message, somebody |
| 810 | * handed us one (or a too-large indication). |
| 811 | */ |
| 812 | switch (self->ith_state) { |
| 813 | case MACH_RCV_SCATTER_SMALL: |
| 814 | case MACH_RCV_TOO_LARGE: |
| 815 | /* |
| 816 | * Somebody tried to give us a too large |
| 817 | * message. If we indicated that we cared, |
| 818 | * then they only gave us the indication, |
| 819 | * otherwise they gave us the indication |
| 820 | * AND the message anyway. |
| 821 | */ |
| 822 | if (option64 & MACH_RCV_LARGE) { |
| 823 | return; |
| 824 | } |
| 825 | return; |
| 826 | case MACH_MSG_SUCCESS: |
| 827 | return; |
| 828 | #if MACH_FLIPC |
| 829 | case MACH_PEEK_READY: |
| 830 | return; |
| 831 | #endif /* MACH_FLIPC */ |
| 832 | |
| 833 | default: |
| 834 | panic("ipc_mqueue_receive_results: strange ith_state %d" , self->ith_state); |
| 835 | } |
| 836 | |
| 837 | default: |
| 838 | panic("ipc_mqueue_receive_results: strange wait_result %d" , saved_wait_result); |
| 839 | } |
| 840 | } |
| 841 | |
| 842 | void |
| 843 | ipc_mqueue_receive_continue( |
| 844 | __unused void *param, |
| 845 | wait_result_t wresult) |
| 846 | { |
| 847 | ipc_mqueue_receive_results(saved_wait_result: wresult); |
| 848 | mach_msg_receive_continue(); /* hard-coded for now */ |
| 849 | } |
| 850 | |
| 851 | /* |
| 852 | * Routine: ipc_mqueue_receive |
| 853 | * Purpose: |
| 854 | * Receive a message from a message queue. |
| 855 | * |
| 856 | * Conditions: |
| 857 | * Our caller must hold a reference for the port or port set |
| 858 | * to which this queue belongs, to keep the queue |
| 859 | * from being deallocated. |
| 860 | * |
| 861 | * The kmsg is returned with clean header fields |
| 862 | * and with the circular bit turned off through the ith_kmsg |
| 863 | * field of the thread's receive continuation state. |
| 864 | * Returns: |
| 865 | * MACH_MSG_SUCCESS Message returned in ith_kmsg. |
| 866 | * MACH_RCV_TOO_LARGE Message size returned in ith_msize, |
| 867 | * Auxiliary data size returned in ith_asize |
| 868 | * MACH_RCV_TIMED_OUT No message obtained. |
| 869 | * MACH_RCV_INTERRUPTED No message obtained. |
| 870 | * MACH_RCV_PORT_DIED Port/set died; no message. |
| 871 | * MACH_RCV_PORT_CHANGED Port moved into set; no msg. |
| 872 | * |
| 873 | */ |
| 874 | |
| 875 | void |
| 876 | ipc_mqueue_receive( |
| 877 | struct waitq *waitq, |
| 878 | mach_msg_option64_t option64, |
| 879 | mach_msg_size_t max_size, |
| 880 | mach_msg_size_t max_aux_size, /* 0 if no aux buffer */ |
| 881 | mach_msg_timeout_t rcv_timeout, |
| 882 | int interruptible, |
| 883 | bool has_continuation) |
| 884 | { |
| 885 | wait_result_t wresult; |
| 886 | thread_t self = current_thread(); |
| 887 | |
| 888 | waitq_lock(wq: waitq); |
| 889 | |
| 890 | wresult = ipc_mqueue_receive_on_thread_and_unlock(waitq: waitq, option64, max_size, |
| 891 | max_aux_size, rcv_timeout, interruptible, thread: self); |
| 892 | /* object unlocked */ |
| 893 | if (wresult == THREAD_NOT_WAITING) { |
| 894 | return; |
| 895 | } |
| 896 | |
| 897 | if (wresult == THREAD_WAITING) { |
| 898 | if (has_continuation) { |
| 899 | wresult = thread_block(continuation: ipc_mqueue_receive_continue); |
| 900 | /* NOTREACHED */ |
| 901 | } |
| 902 | wresult = thread_block(THREAD_CONTINUE_NULL); |
| 903 | } |
| 904 | ipc_mqueue_receive_results(saved_wait_result: wresult); |
| 905 | } |
| 906 | |
| 907 | /* |
| 908 | * Routine: ipc_mqueue_receive_on_thread_and_unlock |
| 909 | * Purpose: |
| 910 | * Receive a message from a message queue using a specified thread. |
| 911 | * If no message available, assert_wait on the appropriate waitq. |
| 912 | * |
| 913 | * Conditions: |
| 914 | * Assumes thread is self. |
| 915 | * The port/port-set waitq is locked on entry, unlocked on return. |
| 916 | * May have assert-waited. Caller must block in those cases. |
| 917 | */ |
| 918 | wait_result_t |
| 919 | ipc_mqueue_receive_on_thread_and_unlock( |
| 920 | struct waitq *waitq, |
| 921 | mach_msg_option64_t option64, |
| 922 | mach_msg_size_t max_msg_size, |
| 923 | mach_msg_size_t max_aux_size, |
| 924 | mach_msg_timeout_t rcv_timeout, |
| 925 | int interruptible, |
| 926 | thread_t thread) |
| 927 | { |
| 928 | ipc_object_t object = io_from_waitq(waitq); |
| 929 | ipc_port_t port = IP_NULL; |
| 930 | wait_result_t wresult; |
| 931 | uint64_t deadline; |
| 932 | struct turnstile *rcv_turnstile = TURNSTILE_NULL; |
| 933 | |
| 934 | if (waitq_type(wq: waitq) == WQT_PORT_SET) { |
| 935 | ipc_pset_t pset = ips_object_to_pset(object); |
| 936 | wqs_prepost_flags_t wqs_flags = WQS_PREPOST_LOCK; |
| 937 | struct waitq *port_wq; |
| 938 | |
| 939 | /* |
| 940 | * Put the message at the back of the prepost list |
| 941 | * if it's not a PEEK. |
| 942 | * |
| 943 | * Might drop the pset lock temporarily. |
| 944 | */ |
| 945 | #if MACH_FLIPC |
| 946 | if (option64 & MACH64_PEEK_MSG) { |
| 947 | wqs_flags |= WQS_PREPOST_PEEK; |
| 948 | } |
| 949 | #endif /* MACH_FLIPC */ |
| 950 | port_wq = waitq_set_first_prepost(wqset: &pset->ips_wqset, flags: wqs_flags); |
| 951 | |
| 952 | /* Returns with port locked */ |
| 953 | |
| 954 | if (port_wq != NULL) { |
| 955 | /* |
| 956 | * We get here if there is at least one message |
| 957 | * waiting on port_wq. We have instructed the prepost |
| 958 | * iteration logic to leave both the port_wq and the |
| 959 | * set waitq locked. |
| 960 | * |
| 961 | * Continue on to handling the message with just |
| 962 | * the port waitq locked. |
| 963 | */ |
| 964 | io_unlock(object); |
| 965 | port = ip_from_waitq(port_wq); |
| 966 | } |
| 967 | } else if (waitq_type(wq: waitq) == WQT_PORT) { |
| 968 | port = ip_from_waitq(waitq); |
| 969 | if (ipc_kmsg_queue_empty(&port->ip_messages.imq_messages)) { |
| 970 | port = IP_NULL; |
| 971 | } |
| 972 | } else { |
| 973 | panic("Unknown waitq type (%p/0x%x)" , waitq, waitq_type(waitq)); |
| 974 | } |
| 975 | |
| 976 | if (port) { |
| 977 | #if MACH_FLIPC |
| 978 | if (option64 & MACH64_PEEK_MSG) { |
| 979 | ipc_mqueue_peek_on_thread_locked(&port->ip_messages, |
| 980 | option64, thread); |
| 981 | } else |
| 982 | #endif /* MACH_FLIPC */ |
| 983 | { |
| 984 | ipc_mqueue_select_on_thread_locked(port_mq: &port->ip_messages, |
| 985 | option64, max_size: max_msg_size, max_aux_size, thread); |
| 986 | } |
| 987 | ip_mq_unlock(port); |
| 988 | return THREAD_NOT_WAITING; |
| 989 | } |
| 990 | |
| 991 | if (!waitq_is_valid(wq: waitq)) { |
| 992 | /* someone raced us to destroy this mqueue/port! */ |
| 993 | io_unlock(object); |
| 994 | /* |
| 995 | * ipc_mqueue_receive_results updates the thread's ith_state |
| 996 | * TODO: differentiate between rights being moved and |
| 997 | * rights/ports being destroyed (21885327) |
| 998 | */ |
| 999 | return THREAD_RESTART; |
| 1000 | } |
| 1001 | |
| 1002 | /* |
| 1003 | * Looks like we'll have to block. The waitq we will |
| 1004 | * block on (whether the set's or the local port's) is |
| 1005 | * still locked. |
| 1006 | */ |
| 1007 | if ((option64 & MACH_RCV_TIMEOUT) && rcv_timeout == 0) { |
| 1008 | io_unlock(object); |
| 1009 | thread->ith_state = MACH_RCV_TIMED_OUT; |
| 1010 | return THREAD_NOT_WAITING; |
| 1011 | } |
| 1012 | |
| 1013 | thread->ith_option = option64; |
| 1014 | thread->ith_max_msize = max_msg_size; |
| 1015 | thread->ith_msize = 0; |
| 1016 | |
| 1017 | thread->ith_max_asize = max_aux_size; |
| 1018 | thread->ith_asize = 0; |
| 1019 | |
| 1020 | thread->ith_state = MACH_RCV_IN_PROGRESS; |
| 1021 | #if MACH_FLIPC |
| 1022 | if (option64 & MACH64_PEEK_MSG) { |
| 1023 | thread->ith_state = MACH_PEEK_IN_PROGRESS; |
| 1024 | } |
| 1025 | #endif /* MACH_FLIPC */ |
| 1026 | |
| 1027 | if (option64 & MACH_RCV_TIMEOUT) { |
| 1028 | clock_interval_to_deadline(interval: rcv_timeout, scale_factor: 1000 * NSEC_PER_USEC, result: &deadline); |
| 1029 | } else { |
| 1030 | deadline = 0; |
| 1031 | } |
| 1032 | |
| 1033 | /* |
| 1034 | * Threads waiting on a reply port (not portset) |
| 1035 | * will wait on its receive turnstile. |
| 1036 | * |
| 1037 | * Donate waiting thread's turnstile and |
| 1038 | * setup inheritor for special reply port. |
| 1039 | * Based on the state of the special reply |
| 1040 | * port, the inheritor would be the send |
| 1041 | * turnstile of the connection port on which |
| 1042 | * the send of sync ipc would happen or |
| 1043 | * workloop's turnstile who would reply to |
| 1044 | * the sync ipc message. |
| 1045 | * |
| 1046 | * Pass in mqueue wait in waitq_assert_wait to |
| 1047 | * support port set wakeup. The mqueue waitq of port |
| 1048 | * will be converted to to turnstile waitq |
| 1049 | * in waitq_assert_wait instead of global waitqs. |
| 1050 | */ |
| 1051 | if (waitq_type(wq: waitq) == WQT_PORT) { |
| 1052 | port = ip_from_waitq(waitq); |
| 1053 | rcv_turnstile = turnstile_prepare(proprietor: (uintptr_t)port, |
| 1054 | port_rcv_turnstile_address(port), |
| 1055 | TURNSTILE_NULL, type: TURNSTILE_SYNC_IPC); |
| 1056 | |
| 1057 | ipc_port_recv_update_inheritor(port, turnstile: rcv_turnstile, |
| 1058 | flags: TURNSTILE_DELAYED_UPDATE); |
| 1059 | } |
| 1060 | |
| 1061 | thread_set_pending_block_hint(thread, block_hint: kThreadWaitPortReceive); |
| 1062 | wresult = waitq_assert_wait64_locked(waitq: waitq, |
| 1063 | IPC_MQUEUE_RECEIVE, |
| 1064 | interruptible, |
| 1065 | TIMEOUT_URGENCY_USER_NORMAL, |
| 1066 | deadline, |
| 1067 | TIMEOUT_NO_LEEWAY, |
| 1068 | thread); |
| 1069 | if (wresult == THREAD_AWAKENED) { |
| 1070 | /* |
| 1071 | * The first thing we did was to look for preposts |
| 1072 | * (using waitq_set_first_prepost() for sets, or looking |
| 1073 | * at the port's queue for ports). |
| 1074 | * |
| 1075 | * Since we found none, we kept the waitq locked. |
| 1076 | * |
| 1077 | * It ensures that waitq_assert_wait64_locked() can't |
| 1078 | * find pre-posts either, won't drop the waitq lock |
| 1079 | * either (even for a set), and can't return THREAD_AWAKENED. |
| 1080 | */ |
| 1081 | panic("ipc_mqueue_receive_on_thread: sleep walking" ); |
| 1082 | } |
| 1083 | |
| 1084 | io_unlock(object); |
| 1085 | |
| 1086 | /* |
| 1087 | * After this point, a waiting thread could be found by the wakeup |
| 1088 | * identify path, and the other side now owns the ith_ fields until |
| 1089 | * this thread blocks and resumes in the continuation |
| 1090 | */ |
| 1091 | |
| 1092 | /* Check if its a port mqueue and if it needs to call turnstile_update_inheritor_complete */ |
| 1093 | if (rcv_turnstile != TURNSTILE_NULL) { |
| 1094 | turnstile_update_inheritor_complete(turnstile: rcv_turnstile, flags: TURNSTILE_INTERLOCK_NOT_HELD); |
| 1095 | } |
| 1096 | /* Its callers responsibility to call turnstile_complete to get the turnstile back */ |
| 1097 | |
| 1098 | return wresult; |
| 1099 | } |
| 1100 | |
| 1101 | #if MACH_FLIPC |
| 1102 | /* |
| 1103 | * Routine: ipc_mqueue_peek_on_thread_locked |
| 1104 | * Purpose: |
| 1105 | * A receiver discovered that there was a message on the queue |
| 1106 | * before he had to block. Tell a thread about the message queue, |
| 1107 | * but don't pick off any messages. |
| 1108 | * Conditions: |
| 1109 | * port_mq locked |
| 1110 | * at least one message on port_mq's message queue |
| 1111 | * |
| 1112 | * Returns: (on thread->ith_state) |
| 1113 | * MACH_PEEK_READY ith_peekq contains a message queue |
| 1114 | */ |
| 1115 | void |
| 1116 | ipc_mqueue_peek_on_thread_locked( |
| 1117 | ipc_mqueue_t port_mq, |
| 1118 | __assert_only mach_msg_option64_t option64, |
| 1119 | thread_t thread) |
| 1120 | { |
| 1121 | assert(option64 & MACH64_PEEK_MSG); |
| 1122 | assert(ipc_kmsg_queue_first(&port_mq->imq_messages) != IKM_NULL); |
| 1123 | |
| 1124 | /* |
| 1125 | * Take a reference on the mqueue's associated port: |
| 1126 | * the peeking thread will be responsible to release this reference |
| 1127 | */ |
| 1128 | ip_validate(ip_from_mq(port_mq)); |
| 1129 | ip_reference(ip_from_mq(port_mq)); |
| 1130 | thread->ith_peekq = port_mq; |
| 1131 | thread->ith_state = MACH_PEEK_READY; |
| 1132 | } |
| 1133 | #endif /* MACH_FLIPC */ |
| 1134 | |
| 1135 | /* |
| 1136 | * Routine: ipc_mqueue_select_on_thread_locked |
| 1137 | * Purpose: |
| 1138 | * A receiver discovered that there was a message on the queue |
| 1139 | * before he had to block. Pick the message off the queue and |
| 1140 | * "post" it to thread. |
| 1141 | * Conditions: |
| 1142 | * port locked. |
| 1143 | * thread not locked. |
| 1144 | * There is a message. |
| 1145 | * No need to reserve prepost objects - it will never prepost |
| 1146 | * |
| 1147 | * Returns: |
| 1148 | * MACH_MSG_SUCCESS Actually selected a message for ourselves. |
| 1149 | * MACH_RCV_TOO_LARGE May or may not have pull it, but it is large |
| 1150 | */ |
| 1151 | void |
| 1152 | ipc_mqueue_select_on_thread_locked( |
| 1153 | ipc_mqueue_t port_mq, |
| 1154 | mach_msg_option64_t option64, |
| 1155 | mach_msg_size_t max_msg_size, |
| 1156 | mach_msg_size_t max_aux_size, |
| 1157 | thread_t thread) |
| 1158 | { |
| 1159 | ipc_kmsg_t kmsg; |
| 1160 | mach_msg_size_t msize, asize; |
| 1161 | |
| 1162 | mach_msg_return_t mr = MACH_MSG_SUCCESS; |
| 1163 | |
| 1164 | /* |
| 1165 | * Do some sanity checking of our ability to receive |
| 1166 | * before pulling the message off the queue. |
| 1167 | */ |
| 1168 | kmsg = ipc_kmsg_queue_first(&port_mq->imq_messages); |
| 1169 | assert(kmsg != IKM_NULL); |
| 1170 | |
| 1171 | /* |
| 1172 | * If we really can't receive it, but we had the |
| 1173 | * MACH_RCV_LARGE option set, then don't take it off |
| 1174 | * the queue, instead return the appropriate error |
| 1175 | * (and size needed). |
| 1176 | */ |
| 1177 | msize = ipc_kmsg_copyout_size(kmsg, map: thread->map); |
| 1178 | asize = ipc_kmsg_aux_data_size(kmsg); |
| 1179 | |
| 1180 | if (ipc_kmsg_too_large(msg_size: msize, aux_size: asize, options: option64, |
| 1181 | max_msg_size, max_aux_size, receiver: thread)) { |
| 1182 | mr = MACH_RCV_TOO_LARGE; |
| 1183 | if (option64 & MACH_RCV_LARGE) { |
| 1184 | ipc_kmsg_validate_partial_sig(kmsg); |
| 1185 | thread->ith_receiver_name = port_mq->imq_receiver_name; |
| 1186 | thread->ith_kmsg = IKM_NULL; |
| 1187 | thread->ith_msize = msize; |
| 1188 | thread->ith_asize = asize; |
| 1189 | thread->ith_seqno = 0; |
| 1190 | thread->ith_state = mr; |
| 1191 | return; |
| 1192 | } |
| 1193 | } |
| 1194 | |
| 1195 | ipc_kmsg_rmqueue(&port_mq->imq_messages, kmsg); |
| 1196 | #if MACH_FLIPC |
| 1197 | if (MACH_NODE_VALID(kmsg->ikm_node) && FPORT_VALID(port_mq->imq_fport)) { |
| 1198 | flipc_msg_ack(kmsg->ikm_node, port_mq, TRUE); |
| 1199 | } |
| 1200 | #endif |
| 1201 | ipc_mqueue_release_msgcount(port_mq); |
| 1202 | thread->ith_seqno = port_mq->imq_seqno++; |
| 1203 | thread->ith_kmsg = kmsg; |
| 1204 | thread->ith_state = mr; |
| 1205 | |
| 1206 | counter_inc(¤t_task()->messages_received); |
| 1207 | return; |
| 1208 | } |
| 1209 | |
| 1210 | /* |
| 1211 | * Routine: ipc_mqueue_peek_locked |
| 1212 | * Purpose: |
| 1213 | * Peek at a (non-set) message queue to see if it has a message |
| 1214 | * matching the sequence number provided (if zero, then the |
| 1215 | * first message in the queue) and return vital info about the |
| 1216 | * message. |
| 1217 | * |
| 1218 | * Conditions: |
| 1219 | * The io object corresponding to mq is locked by callers. |
| 1220 | * Other locks may be held by callers, so this routine cannot block. |
| 1221 | * Caller holds reference on the message queue. |
| 1222 | */ |
| 1223 | unsigned |
| 1224 | ipc_mqueue_peek_locked(ipc_mqueue_t mq, |
| 1225 | mach_port_seqno_t * seqnop, |
| 1226 | mach_msg_size_t * msg_sizep, |
| 1227 | mach_msg_id_t * msg_idp, |
| 1228 | mach_msg_max_trailer_t * msg_trailerp, |
| 1229 | ipc_kmsg_t *kmsgp) |
| 1230 | { |
| 1231 | ipc_kmsg_queue_t kmsgq; |
| 1232 | ipc_kmsg_t kmsg; |
| 1233 | mach_port_seqno_t seqno, msgoff; |
| 1234 | unsigned res = 0; |
| 1235 | mach_msg_header_t *hdr; |
| 1236 | |
| 1237 | seqno = 0; |
| 1238 | if (seqnop != NULL) { |
| 1239 | seqno = *seqnop; |
| 1240 | } |
| 1241 | |
| 1242 | if (seqno == 0) { |
| 1243 | seqno = mq->imq_seqno; |
| 1244 | msgoff = 0; |
| 1245 | } else if (seqno >= mq->imq_seqno && |
| 1246 | seqno < mq->imq_seqno + mq->imq_msgcount) { |
| 1247 | msgoff = seqno - mq->imq_seqno; |
| 1248 | } else { |
| 1249 | goto out; |
| 1250 | } |
| 1251 | |
| 1252 | /* look for the message that would match that seqno */ |
| 1253 | kmsgq = &mq->imq_messages; |
| 1254 | kmsg = ipc_kmsg_queue_first(kmsgq); |
| 1255 | while (msgoff-- && kmsg != IKM_NULL) { |
| 1256 | kmsg = ipc_kmsg_queue_next(kmsgq, kmsg); |
| 1257 | } |
| 1258 | if (kmsg == IKM_NULL) { |
| 1259 | goto out; |
| 1260 | } |
| 1261 | |
| 1262 | /* |
| 1263 | * Validate kmsg signature before doing anything with it. Since we are holding |
| 1264 | * the mqueue lock here, and only header + trailer will be peeked on, just |
| 1265 | * do a partial validation to finish quickly. |
| 1266 | * |
| 1267 | * Partial kmsg signature is only supported on PAC devices. |
| 1268 | */ |
| 1269 | ipc_kmsg_validate_partial_sig(kmsg); |
| 1270 | |
| 1271 | hdr = ikm_header(kmsg); |
| 1272 | /* found one - return the requested info */ |
| 1273 | if (seqnop != NULL) { |
| 1274 | *seqnop = seqno; |
| 1275 | } |
| 1276 | if (msg_sizep != NULL) { |
| 1277 | *msg_sizep = hdr->msgh_size; |
| 1278 | } |
| 1279 | if (msg_idp != NULL) { |
| 1280 | *msg_idp = hdr->msgh_id; |
| 1281 | } |
| 1282 | if (msg_trailerp != NULL) { |
| 1283 | memcpy(dst: msg_trailerp, src: ipc_kmsg_get_trailer(kmsg, false), n: sizeof(mach_msg_max_trailer_t)); |
| 1284 | } |
| 1285 | if (kmsgp != NULL) { |
| 1286 | *kmsgp = kmsg; |
| 1287 | } |
| 1288 | |
| 1289 | res = 1; |
| 1290 | |
| 1291 | out: |
| 1292 | return res; |
| 1293 | } |
| 1294 | |
| 1295 | |
| 1296 | /* |
| 1297 | * Routine: ipc_mqueue_peek |
| 1298 | * Purpose: |
| 1299 | * Peek at a (non-set) message queue to see if it has a message |
| 1300 | * matching the sequence number provided (if zero, then the |
| 1301 | * first message in the queue) and return vital info about the |
| 1302 | * message. |
| 1303 | * |
| 1304 | * Conditions: |
| 1305 | * The ipc_mqueue_t is unlocked. |
| 1306 | * Locks may be held by callers, so this routine cannot block. |
| 1307 | * Caller holds reference on the message queue. |
| 1308 | */ |
| 1309 | unsigned |
| 1310 | ipc_mqueue_peek(ipc_mqueue_t mq, |
| 1311 | mach_port_seqno_t * seqnop, |
| 1312 | mach_msg_size_t * msg_sizep, |
| 1313 | mach_msg_id_t * msg_idp, |
| 1314 | mach_msg_max_trailer_t * msg_trailerp, |
| 1315 | ipc_kmsg_t *kmsgp) |
| 1316 | { |
| 1317 | ipc_port_t port = ip_from_mq(mq); |
| 1318 | unsigned res; |
| 1319 | |
| 1320 | ip_mq_lock(port); |
| 1321 | |
| 1322 | res = ipc_mqueue_peek_locked(mq, seqnop, msg_sizep, msg_idp, |
| 1323 | msg_trailerp, kmsgp); |
| 1324 | |
| 1325 | ip_mq_unlock(port); |
| 1326 | return res; |
| 1327 | } |
| 1328 | |
| 1329 | #if MACH_FLIPC |
| 1330 | /* |
| 1331 | * Routine: ipc_mqueue_release_peek_ref |
| 1332 | * Purpose: |
| 1333 | * Release the reference on an mqueue's associated port which was |
| 1334 | * granted to a thread in ipc_mqueue_peek_on_thread (on the |
| 1335 | * MACH64_PEEK_MSG thread wakeup path). |
| 1336 | * |
| 1337 | * Conditions: |
| 1338 | * The ipc_mqueue_t should be locked on entry. |
| 1339 | * The ipc_mqueue_t will be _unlocked_ on return |
| 1340 | * (and potentially invalid!) |
| 1341 | * |
| 1342 | */ |
| 1343 | void |
| 1344 | ipc_mqueue_release_peek_ref(ipc_mqueue_t mqueue) |
| 1345 | { |
| 1346 | ipc_port_t port = ip_from_mq(mqueue); |
| 1347 | |
| 1348 | ip_mq_lock_held(port); |
| 1349 | |
| 1350 | /* |
| 1351 | * clear any preposts this mq may have generated |
| 1352 | * (which would cause subsequent immediate wakeups) |
| 1353 | */ |
| 1354 | waitq_clear_prepost_locked(&port->ip_waitq); |
| 1355 | |
| 1356 | ip_mq_unlock(port); |
| 1357 | |
| 1358 | /* |
| 1359 | * release the port reference: we need to do this outside the lock |
| 1360 | * because we might be holding the last port reference! |
| 1361 | **/ |
| 1362 | ip_release(port); |
| 1363 | } |
| 1364 | #endif /* MACH_FLIPC */ |
| 1365 | |
| 1366 | /* |
| 1367 | * Routine: ipc_mqueue_destroy_locked |
| 1368 | * Purpose: |
| 1369 | * Destroy a message queue. |
| 1370 | * Set any blocked senders running. |
| 1371 | * Destroy the kmsgs in the queue. |
| 1372 | * Conditions: |
| 1373 | * port locked |
| 1374 | * Receivers were removed when the receive right was "changed" |
| 1375 | */ |
| 1376 | boolean_t |
| 1377 | ipc_mqueue_destroy_locked(ipc_mqueue_t mqueue, waitq_link_list_t *free_l) |
| 1378 | { |
| 1379 | ipc_port_t port = ip_from_mq(mqueue); |
| 1380 | boolean_t reap = FALSE; |
| 1381 | struct turnstile *send_turnstile = port_send_turnstile(port); |
| 1382 | |
| 1383 | /* |
| 1384 | * rouse all blocked senders |
| 1385 | * (don't boost anyone - we're tearing this queue down) |
| 1386 | * (never preposts) |
| 1387 | */ |
| 1388 | port->ip_fullwaiters = false; |
| 1389 | |
| 1390 | if (send_turnstile != TURNSTILE_NULL) { |
| 1391 | waitq_wakeup64_all(waitq: &send_turnstile->ts_waitq, |
| 1392 | IPC_MQUEUE_FULL, |
| 1393 | THREAD_RESTART, flags: WAITQ_WAKEUP_DEFAULT); |
| 1394 | } |
| 1395 | |
| 1396 | #if MACH_FLIPC |
| 1397 | ipc_kmsg_t kmsg; |
| 1398 | |
| 1399 | cqe_foreach_element_safe(kmsg, &mqueue->imq_messages, ikm_link) { |
| 1400 | if (MACH_NODE_VALID(kmsg->ikm_node) && |
| 1401 | FPORT_VALID(mqueue->imq_fport)) { |
| 1402 | flipc_msg_ack(kmsg->ikm_node, mqueue, TRUE); |
| 1403 | } |
| 1404 | } |
| 1405 | #endif |
| 1406 | |
| 1407 | /* |
| 1408 | * Move messages from the specified queue to the per-thread |
| 1409 | * clean/drain queue while we have the mqueue lock. |
| 1410 | */ |
| 1411 | reap = ipc_kmsg_delayed_destroy_queue(queue: &mqueue->imq_messages); |
| 1412 | |
| 1413 | /* |
| 1414 | * Wipe out message count, both for messages about to be |
| 1415 | * reaped and for reserved space for (previously) woken senders. |
| 1416 | * This is the indication to them that their reserved space is gone |
| 1417 | * (the mqueue was destroyed). |
| 1418 | */ |
| 1419 | mqueue->imq_msgcount = 0; |
| 1420 | |
| 1421 | /* |
| 1422 | * invalidate the waitq for subsequent mqueue operations, |
| 1423 | * the port lock could be dropped after invalidating the mqueue. |
| 1424 | */ |
| 1425 | |
| 1426 | waitq_invalidate(wq: &port->ip_waitq); |
| 1427 | |
| 1428 | waitq_unlink_all_locked(waitq: &port->ip_waitq, NULL, free_l); |
| 1429 | |
| 1430 | return reap; |
| 1431 | } |
| 1432 | |
| 1433 | /* |
| 1434 | * Routine: ipc_mqueue_set_qlimit_locked |
| 1435 | * Purpose: |
| 1436 | * Changes a message queue limit; the maximum number |
| 1437 | * of messages which may be queued. |
| 1438 | * Conditions: |
| 1439 | * Port locked. |
| 1440 | */ |
| 1441 | |
| 1442 | void |
| 1443 | ipc_mqueue_set_qlimit_locked( |
| 1444 | ipc_mqueue_t mqueue, |
| 1445 | mach_port_msgcount_t qlimit) |
| 1446 | { |
| 1447 | ipc_port_t port = ip_from_mq(mqueue); |
| 1448 | |
| 1449 | assert(qlimit <= MACH_PORT_QLIMIT_MAX); |
| 1450 | |
| 1451 | /* wake up senders allowed by the new qlimit */ |
| 1452 | if (qlimit > mqueue->imq_qlimit) { |
| 1453 | mach_port_msgcount_t i, wakeup; |
| 1454 | struct turnstile *send_turnstile = port_send_turnstile(port); |
| 1455 | |
| 1456 | /* caution: wakeup, qlimit are unsigned */ |
| 1457 | wakeup = qlimit - mqueue->imq_qlimit; |
| 1458 | |
| 1459 | for (i = 0; i < wakeup; i++) { |
| 1460 | /* |
| 1461 | * boost the priority of the awoken thread |
| 1462 | * (WAITQ_PROMOTE_PRIORITY) to ensure it uses |
| 1463 | * the message queue slot we've just reserved. |
| 1464 | * |
| 1465 | * NOTE: this will never prepost |
| 1466 | */ |
| 1467 | if (send_turnstile == TURNSTILE_NULL || |
| 1468 | waitq_wakeup64_one(waitq: &send_turnstile->ts_waitq, |
| 1469 | IPC_MQUEUE_FULL, |
| 1470 | THREAD_AWAKENED, |
| 1471 | flags: WAITQ_PROMOTE_PRIORITY) == KERN_NOT_WAITING) { |
| 1472 | port->ip_fullwaiters = false; |
| 1473 | break; |
| 1474 | } |
| 1475 | mqueue->imq_msgcount++; /* give it to the awakened thread */ |
| 1476 | } |
| 1477 | } |
| 1478 | mqueue->imq_qlimit = (uint16_t)qlimit; |
| 1479 | } |
| 1480 | |
| 1481 | /* |
| 1482 | * Routine: ipc_mqueue_set_seqno_locked |
| 1483 | * Purpose: |
| 1484 | * Changes an mqueue's sequence number. |
| 1485 | * Conditions: |
| 1486 | * Caller holds a reference to the queue's containing object. |
| 1487 | */ |
| 1488 | void |
| 1489 | ipc_mqueue_set_seqno_locked( |
| 1490 | ipc_mqueue_t mqueue, |
| 1491 | mach_port_seqno_t seqno) |
| 1492 | { |
| 1493 | mqueue->imq_seqno = seqno; |
| 1494 | } |
| 1495 | |
| 1496 | |
| 1497 | /* |
| 1498 | * Routine: ipc_mqueue_copyin |
| 1499 | * Purpose: |
| 1500 | * Convert a name in a space to a message queue. |
| 1501 | * Conditions: |
| 1502 | * Nothing locked. If successful, the caller gets a ref for |
| 1503 | * for the object. This ref ensures the continued existence of |
| 1504 | * the queue. |
| 1505 | * Returns: |
| 1506 | * MACH_MSG_SUCCESS Found a message queue. |
| 1507 | * MACH_RCV_INVALID_NAME The space is dead. |
| 1508 | * MACH_RCV_INVALID_NAME The name doesn't denote a right. |
| 1509 | * MACH_RCV_INVALID_NAME |
| 1510 | * The denoted right is not receive or port set. |
| 1511 | * MACH_RCV_IN_SET Receive right is a member of a set. |
| 1512 | */ |
| 1513 | |
| 1514 | mach_msg_return_t |
| 1515 | ipc_mqueue_copyin( |
| 1516 | ipc_space_t space, |
| 1517 | mach_port_name_t name, |
| 1518 | ipc_object_t *objectp) |
| 1519 | { |
| 1520 | ipc_entry_bits_t bits; |
| 1521 | ipc_object_t object; |
| 1522 | kern_return_t kr; |
| 1523 | |
| 1524 | kr = ipc_right_lookup_read(space, name, bitsp: &bits, objectp: &object); |
| 1525 | if (kr != KERN_SUCCESS) { |
| 1526 | return MACH_RCV_INVALID_NAME; |
| 1527 | } |
| 1528 | /* object is locked and active */ |
| 1529 | |
| 1530 | if (bits & MACH_PORT_TYPE_RECEIVE) { |
| 1531 | __assert_only ipc_port_t port = ip_object_to_port(object); |
| 1532 | assert(ip_get_receiver_name(port) == name); |
| 1533 | assert(ip_in_space(port, space)); |
| 1534 | } |
| 1535 | if (bits & (MACH_PORT_TYPE_RECEIVE | MACH_PORT_TYPE_PORT_SET)) { |
| 1536 | io_reference(object); |
| 1537 | io_unlock(object); |
| 1538 | } else { |
| 1539 | io_unlock(object); |
| 1540 | /* guard exception if we never held the receive right in this entry */ |
| 1541 | if ((bits & MACH_PORT_TYPE_EX_RECEIVE) == 0) { |
| 1542 | mach_port_guard_exception(name, inguard: 0, portguard: 0, reason: kGUARD_EXC_RCV_INVALID_NAME); |
| 1543 | } |
| 1544 | return MACH_RCV_INVALID_NAME; |
| 1545 | } |
| 1546 | |
| 1547 | *objectp = object; |
| 1548 | return MACH_MSG_SUCCESS; |
| 1549 | } |
| 1550 | |