1 /* $NetBSD: uipc_usrreq.c,v 1.179 2015/05/02 17:18:03 rtr Exp $ */ 2 3 /*- 4 * Copyright (c) 1998, 2000, 2004, 2008, 2009 The NetBSD Foundation, Inc. 5 * All rights reserved. 6 * 7 * This code is derived from software contributed to The NetBSD Foundation 8 * by Jason R. Thorpe of the Numerical Aerospace Simulation Facility, 9 * NASA Ames Research Center, and by Andrew Doran. 10 * 11 * Redistribution and use in source and binary forms, with or without 12 * modification, are permitted provided that the following conditions 13 * are met: 14 * 1. Redistributions of source code must retain the above copyright 15 * notice, this list of conditions and the following disclaimer. 16 * 2. Redistributions in binary form must reproduce the above copyright 17 * notice, this list of conditions and the following disclaimer in the 18 * documentation and/or other materials provided with the distribution. 19 * 20 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 21 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 22 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 23 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 24 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 25 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 26 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 27 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 28 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 29 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 30 * POSSIBILITY OF SUCH DAMAGE. 31 */ 32 33 /* 34 * Copyright (c) 1982, 1986, 1989, 1991, 1993 35 * The Regents of the University of California. All rights reserved. 36 * 37 * Redistribution and use in source and binary forms, with or without 38 * modification, are permitted provided that the following conditions 39 * are met: 40 * 1. Redistributions of source code must retain the above copyright 41 * notice, this list of conditions and the following disclaimer. 42 * 2. Redistributions in binary form must reproduce the above copyright 43 * notice, this list of conditions and the following disclaimer in the 44 * documentation and/or other materials provided with the distribution. 45 * 3. Neither the name of the University nor the names of its contributors 46 * may be used to endorse or promote products derived from this software 47 * without specific prior written permission. 48 * 49 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 50 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 51 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 52 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 53 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 54 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 55 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 56 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 57 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 58 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 59 * SUCH DAMAGE. 60 * 61 * @(#)uipc_usrreq.c 8.9 (Berkeley) 5/14/95 62 */ 63 64 /* 65 * Copyright (c) 1997 Christopher G. Demetriou. All rights reserved. 66 * 67 * Redistribution and use in source and binary forms, with or without 68 * modification, are permitted provided that the following conditions 69 * are met: 70 * 1. Redistributions of source code must retain the above copyright 71 * notice, this list of conditions and the following disclaimer. 72 * 2. Redistributions in binary form must reproduce the above copyright 73 * notice, this list of conditions and the following disclaimer in the 74 * documentation and/or other materials provided with the distribution. 75 * 3. All advertising materials mentioning features or use of this software 76 * must display the following acknowledgement: 77 * This product includes software developed by the University of 78 * California, Berkeley and its contributors. 79 * 4. Neither the name of the University nor the names of its contributors 80 * may be used to endorse or promote products derived from this software 81 * without specific prior written permission. 82 * 83 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 84 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 85 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 86 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 87 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 88 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 89 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 90 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 91 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 92 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 93 * SUCH DAMAGE. 94 * 95 * @(#)uipc_usrreq.c 8.9 (Berkeley) 5/14/95 96 */ 97 98 #include <sys/cdefs.h> 99 __KERNEL_RCSID(0, "$NetBSD: uipc_usrreq.c,v 1.179 2015/05/02 17:18:03 rtr Exp $"); 100 101 #include <sys/param.h> 102 #include <sys/systm.h> 103 #include <sys/proc.h> 104 #include <sys/filedesc.h> 105 #include <sys/domain.h> 106 #include <sys/protosw.h> 107 #include <sys/socket.h> 108 #include <sys/socketvar.h> 109 #include <sys/unpcb.h> 110 #include <sys/un.h> 111 #include <sys/namei.h> 112 #include <sys/vnode.h> 113 #include <sys/file.h> 114 #include <sys/stat.h> 115 #include <sys/mbuf.h> 116 #include <sys/kauth.h> 117 #include <sys/kmem.h> 118 #include <sys/atomic.h> 119 #include <sys/uidinfo.h> 120 #include <sys/kernel.h> 121 #include <sys/kthread.h> 122 123 /* 124 * Unix communications domain. 125 * 126 * TODO: 127 * RDM 128 * rethink name space problems 129 * need a proper out-of-band 130 * 131 * Notes on locking: 132 * 133 * The generic rules noted in uipc_socket2.c apply. In addition: 134 * 135 * o We have a global lock, uipc_lock. 136 * 137 * o All datagram sockets are locked by uipc_lock. 138 * 139 * o For stream socketpairs, the two endpoints are created sharing the same 140 * independent lock. Sockets presented to PRU_CONNECT2 must already have 141 * matching locks. 142 * 143 * o Stream sockets created via socket() start life with their own 144 * independent lock. 145 * 146 * o Stream connections to a named endpoint are slightly more complicated. 147 * Sockets that have called listen() have their lock pointer mutated to 148 * the global uipc_lock. When establishing a connection, the connecting 149 * socket also has its lock mutated to uipc_lock, which matches the head 150 * (listening socket). We create a new socket for accept() to return, and 151 * that also shares the head's lock. Until the connection is completely 152 * done on both ends, all three sockets are locked by uipc_lock. Once the 153 * connection is complete, the association with the head's lock is broken. 154 * The connecting socket and the socket returned from accept() have their 155 * lock pointers mutated away from uipc_lock, and back to the connecting 156 * socket's original, independent lock. The head continues to be locked 157 * by uipc_lock. 158 * 159 * o If uipc_lock is determined to be a significant source of contention, 160 * it could easily be hashed out. It is difficult to simply make it an 161 * independent lock because of visibility / garbage collection issues: 162 * if a socket has been associated with a lock at any point, that lock 163 * must remain valid until the socket is no longer visible in the system. 164 * The lock must not be freed or otherwise destroyed until any sockets 165 * that had referenced it have also been destroyed. 166 */ 167 const struct sockaddr_un sun_noname = { 168 .sun_len = offsetof(struct sockaddr_un, sun_path), 169 .sun_family = AF_LOCAL, 170 }; 171 ino_t unp_ino; /* prototype for fake inode numbers */ 172 173 static struct mbuf * unp_addsockcred(struct lwp *, struct mbuf *); 174 static void unp_discard_later(file_t *); 175 static void unp_discard_now(file_t *); 176 static void unp_disconnect1(struct unpcb *); 177 static bool unp_drop(struct unpcb *, int); 178 static int unp_internalize(struct mbuf **); 179 static void unp_mark(file_t *); 180 static void unp_scan(struct mbuf *, void (*)(file_t *), int); 181 static void unp_shutdown1(struct unpcb *); 182 static void unp_thread(void *); 183 static void unp_thread_kick(void); 184 185 static kmutex_t *uipc_lock; 186 187 static kcondvar_t unp_thread_cv; 188 static lwp_t *unp_thread_lwp; 189 static SLIST_HEAD(,file) unp_thread_discard; 190 static int unp_defer; 191 192 /* 193 * Initialize Unix protocols. 194 */ 195 void 196 uipc_init(void) 197 { 198 int error; 199 200 uipc_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NONE); 201 cv_init(&unp_thread_cv, "unpgc"); 202 203 error = kthread_create(PRI_NONE, KTHREAD_MPSAFE, NULL, unp_thread, 204 NULL, &unp_thread_lwp, "unpgc"); 205 if (error != 0) 206 panic("uipc_init %d", error); 207 } 208 209 /* 210 * A connection succeeded: disassociate both endpoints from the head's 211 * lock, and make them share their own lock. There is a race here: for 212 * a very brief time one endpoint will be locked by a different lock 213 * than the other end. However, since the current thread holds the old 214 * lock (the listening socket's lock, the head) access can still only be 215 * made to one side of the connection. 216 */ 217 static void 218 unp_setpeerlocks(struct socket *so, struct socket *so2) 219 { 220 struct unpcb *unp; 221 kmutex_t *lock; 222 223 KASSERT(solocked2(so, so2)); 224 225 /* 226 * Bail out if either end of the socket is not yet fully 227 * connected or accepted. We only break the lock association 228 * with the head when the pair of sockets stand completely 229 * on their own. 230 */ 231 KASSERT(so->so_head == NULL); 232 if (so2->so_head != NULL) 233 return; 234 235 /* 236 * Drop references to old lock. A third reference (from the 237 * queue head) must be held as we still hold its lock. Bonus: 238 * we don't need to worry about garbage collecting the lock. 239 */ 240 lock = so->so_lock; 241 KASSERT(lock == uipc_lock); 242 mutex_obj_free(lock); 243 mutex_obj_free(lock); 244 245 /* 246 * Grab stream lock from the initiator and share between the two 247 * endpoints. Issue memory barrier to ensure all modifications 248 * become globally visible before the lock change. so2 is 249 * assumed not to have a stream lock, because it was created 250 * purely for the server side to accept this connection and 251 * started out life using the domain-wide lock. 252 */ 253 unp = sotounpcb(so); 254 KASSERT(unp->unp_streamlock != NULL); 255 KASSERT(sotounpcb(so2)->unp_streamlock == NULL); 256 lock = unp->unp_streamlock; 257 unp->unp_streamlock = NULL; 258 mutex_obj_hold(lock); 259 membar_exit(); 260 /* 261 * possible race if lock is not held - see comment in 262 * uipc_usrreq(PRU_ACCEPT). 263 */ 264 KASSERT(mutex_owned(lock)); 265 solockreset(so, lock); 266 solockreset(so2, lock); 267 } 268 269 /* 270 * Reset a socket's lock back to the domain-wide lock. 271 */ 272 static void 273 unp_resetlock(struct socket *so) 274 { 275 kmutex_t *olock, *nlock; 276 struct unpcb *unp; 277 278 KASSERT(solocked(so)); 279 280 olock = so->so_lock; 281 nlock = uipc_lock; 282 if (olock == nlock) 283 return; 284 unp = sotounpcb(so); 285 KASSERT(unp->unp_streamlock == NULL); 286 unp->unp_streamlock = olock; 287 mutex_obj_hold(nlock); 288 mutex_enter(nlock); 289 solockreset(so, nlock); 290 mutex_exit(olock); 291 } 292 293 static void 294 unp_free(struct unpcb *unp) 295 { 296 if (unp->unp_addr) 297 free(unp->unp_addr, M_SONAME); 298 if (unp->unp_streamlock != NULL) 299 mutex_obj_free(unp->unp_streamlock); 300 kmem_free(unp, sizeof(*unp)); 301 } 302 303 static int 304 unp_output(struct mbuf *m, struct mbuf *control, struct unpcb *unp) 305 { 306 struct socket *so2; 307 const struct sockaddr_un *sun; 308 309 /* XXX: server side closed the socket */ 310 if (unp->unp_conn == NULL) 311 return ECONNREFUSED; 312 so2 = unp->unp_conn->unp_socket; 313 314 KASSERT(solocked(so2)); 315 316 if (unp->unp_addr) 317 sun = unp->unp_addr; 318 else 319 sun = &sun_noname; 320 if (unp->unp_conn->unp_flags & UNP_WANTCRED) 321 control = unp_addsockcred(curlwp, control); 322 if (sbappendaddr(&so2->so_rcv, (const struct sockaddr *)sun, m, 323 control) == 0) { 324 so2->so_rcv.sb_overflowed++; 325 unp_dispose(control); 326 m_freem(control); 327 m_freem(m); 328 return (ENOBUFS); 329 } else { 330 sorwakeup(so2); 331 return (0); 332 } 333 } 334 335 static void 336 unp_setaddr(struct socket *so, struct sockaddr *nam, bool peeraddr) 337 { 338 const struct sockaddr_un *sun = NULL; 339 struct unpcb *unp; 340 341 KASSERT(solocked(so)); 342 unp = sotounpcb(so); 343 344 if (peeraddr) { 345 if (unp->unp_conn && unp->unp_conn->unp_addr) 346 sun = unp->unp_conn->unp_addr; 347 } else { 348 if (unp->unp_addr) 349 sun = unp->unp_addr; 350 } 351 if (sun == NULL) 352 sun = &sun_noname; 353 354 memcpy(nam, sun, sun->sun_len); 355 } 356 357 static int 358 unp_rcvd(struct socket *so, int flags, struct lwp *l) 359 { 360 struct unpcb *unp = sotounpcb(so); 361 struct socket *so2; 362 u_int newhiwat; 363 364 KASSERT(solocked(so)); 365 KASSERT(unp != NULL); 366 367 switch (so->so_type) { 368 369 case SOCK_DGRAM: 370 panic("uipc 1"); 371 /*NOTREACHED*/ 372 373 case SOCK_SEQPACKET: /* FALLTHROUGH */ 374 case SOCK_STREAM: 375 #define rcv (&so->so_rcv) 376 #define snd (&so2->so_snd) 377 if (unp->unp_conn == 0) 378 break; 379 so2 = unp->unp_conn->unp_socket; 380 KASSERT(solocked2(so, so2)); 381 /* 382 * Adjust backpressure on sender 383 * and wakeup any waiting to write. 384 */ 385 snd->sb_mbmax += unp->unp_mbcnt - rcv->sb_mbcnt; 386 unp->unp_mbcnt = rcv->sb_mbcnt; 387 newhiwat = snd->sb_hiwat + unp->unp_cc - rcv->sb_cc; 388 (void)chgsbsize(so2->so_uidinfo, 389 &snd->sb_hiwat, newhiwat, RLIM_INFINITY); 390 unp->unp_cc = rcv->sb_cc; 391 sowwakeup(so2); 392 #undef snd 393 #undef rcv 394 break; 395 396 default: 397 panic("uipc 2"); 398 } 399 400 return 0; 401 } 402 403 static int 404 unp_recvoob(struct socket *so, struct mbuf *m, int flags) 405 { 406 KASSERT(solocked(so)); 407 408 return EOPNOTSUPP; 409 } 410 411 static int 412 unp_send(struct socket *so, struct mbuf *m, struct sockaddr *nam, 413 struct mbuf *control, struct lwp *l) 414 { 415 struct unpcb *unp = sotounpcb(so); 416 int error = 0; 417 u_int newhiwat; 418 struct socket *so2; 419 420 KASSERT(solocked(so)); 421 KASSERT(unp != NULL); 422 KASSERT(m != NULL); 423 424 /* 425 * Note: unp_internalize() rejects any control message 426 * other than SCM_RIGHTS, and only allows one. This 427 * has the side-effect of preventing a caller from 428 * forging SCM_CREDS. 429 */ 430 if (control) { 431 sounlock(so); 432 error = unp_internalize(&control); 433 solock(so); 434 if (error != 0) { 435 m_freem(control); 436 m_freem(m); 437 return error; 438 } 439 } 440 441 switch (so->so_type) { 442 443 case SOCK_DGRAM: { 444 KASSERT(so->so_lock == uipc_lock); 445 if (nam) { 446 if ((so->so_state & SS_ISCONNECTED) != 0) 447 error = EISCONN; 448 else { 449 /* 450 * Note: once connected, the 451 * socket's lock must not be 452 * dropped until we have sent 453 * the message and disconnected. 454 * This is necessary to prevent 455 * intervening control ops, like 456 * another connection. 457 */ 458 error = unp_connect(so, nam, l); 459 } 460 } else { 461 if ((so->so_state & SS_ISCONNECTED) == 0) 462 error = ENOTCONN; 463 } 464 if (error) { 465 unp_dispose(control); 466 m_freem(control); 467 m_freem(m); 468 return error; 469 } 470 error = unp_output(m, control, unp); 471 if (nam) 472 unp_disconnect1(unp); 473 break; 474 } 475 476 case SOCK_SEQPACKET: /* FALLTHROUGH */ 477 case SOCK_STREAM: 478 #define rcv (&so2->so_rcv) 479 #define snd (&so->so_snd) 480 if (unp->unp_conn == NULL) { 481 error = ENOTCONN; 482 break; 483 } 484 so2 = unp->unp_conn->unp_socket; 485 KASSERT(solocked2(so, so2)); 486 if (unp->unp_conn->unp_flags & UNP_WANTCRED) { 487 /* 488 * Credentials are passed only once on 489 * SOCK_STREAM and SOCK_SEQPACKET. 490 */ 491 unp->unp_conn->unp_flags &= ~UNP_WANTCRED; 492 control = unp_addsockcred(l, control); 493 } 494 /* 495 * Send to paired receive port, and then reduce 496 * send buffer hiwater marks to maintain backpressure. 497 * Wake up readers. 498 */ 499 if (control) { 500 if (sbappendcontrol(rcv, m, control) != 0) 501 control = NULL; 502 } else { 503 switch(so->so_type) { 504 case SOCK_SEQPACKET: 505 sbappendrecord(rcv, m); 506 break; 507 case SOCK_STREAM: 508 sbappend(rcv, m); 509 break; 510 default: 511 panic("uipc_usrreq"); 512 break; 513 } 514 } 515 snd->sb_mbmax -= 516 rcv->sb_mbcnt - unp->unp_conn->unp_mbcnt; 517 unp->unp_conn->unp_mbcnt = rcv->sb_mbcnt; 518 newhiwat = snd->sb_hiwat - 519 (rcv->sb_cc - unp->unp_conn->unp_cc); 520 (void)chgsbsize(so->so_uidinfo, 521 &snd->sb_hiwat, newhiwat, RLIM_INFINITY); 522 unp->unp_conn->unp_cc = rcv->sb_cc; 523 sorwakeup(so2); 524 #undef snd 525 #undef rcv 526 if (control != NULL) { 527 unp_dispose(control); 528 m_freem(control); 529 } 530 break; 531 532 default: 533 panic("uipc 4"); 534 } 535 536 return error; 537 } 538 539 static int 540 unp_sendoob(struct socket *so, struct mbuf *m, struct mbuf * control) 541 { 542 KASSERT(solocked(so)); 543 544 m_freem(m); 545 m_freem(control); 546 547 return EOPNOTSUPP; 548 } 549 550 /* 551 * Unix domain socket option processing. 552 */ 553 int 554 uipc_ctloutput(int op, struct socket *so, struct sockopt *sopt) 555 { 556 struct unpcb *unp = sotounpcb(so); 557 int optval = 0, error = 0; 558 559 KASSERT(solocked(so)); 560 561 if (sopt->sopt_level != 0) { 562 error = ENOPROTOOPT; 563 } else switch (op) { 564 565 case PRCO_SETOPT: 566 switch (sopt->sopt_name) { 567 case LOCAL_CREDS: 568 case LOCAL_CONNWAIT: 569 error = sockopt_getint(sopt, &optval); 570 if (error) 571 break; 572 switch (sopt->sopt_name) { 573 #define OPTSET(bit) \ 574 if (optval) \ 575 unp->unp_flags |= (bit); \ 576 else \ 577 unp->unp_flags &= ~(bit); 578 579 case LOCAL_CREDS: 580 OPTSET(UNP_WANTCRED); 581 break; 582 case LOCAL_CONNWAIT: 583 OPTSET(UNP_CONNWAIT); 584 break; 585 } 586 break; 587 #undef OPTSET 588 589 default: 590 error = ENOPROTOOPT; 591 break; 592 } 593 break; 594 595 case PRCO_GETOPT: 596 sounlock(so); 597 switch (sopt->sopt_name) { 598 case LOCAL_PEEREID: 599 if (unp->unp_flags & UNP_EIDSVALID) { 600 error = sockopt_set(sopt, 601 &unp->unp_connid, sizeof(unp->unp_connid)); 602 } else { 603 error = EINVAL; 604 } 605 break; 606 case LOCAL_CREDS: 607 #define OPTBIT(bit) (unp->unp_flags & (bit) ? 1 : 0) 608 609 optval = OPTBIT(UNP_WANTCRED); 610 error = sockopt_setint(sopt, optval); 611 break; 612 #undef OPTBIT 613 614 default: 615 error = ENOPROTOOPT; 616 break; 617 } 618 solock(so); 619 break; 620 } 621 return (error); 622 } 623 624 /* 625 * Both send and receive buffers are allocated PIPSIZ bytes of buffering 626 * for stream sockets, although the total for sender and receiver is 627 * actually only PIPSIZ. 628 * Datagram sockets really use the sendspace as the maximum datagram size, 629 * and don't really want to reserve the sendspace. Their recvspace should 630 * be large enough for at least one max-size datagram plus address. 631 */ 632 #define PIPSIZ 4096 633 u_long unpst_sendspace = PIPSIZ; 634 u_long unpst_recvspace = PIPSIZ; 635 u_long unpdg_sendspace = 2*1024; /* really max datagram size */ 636 u_long unpdg_recvspace = 4*1024; 637 638 u_int unp_rights; /* files in flight */ 639 u_int unp_rights_ratio = 2; /* limit, fraction of maxfiles */ 640 641 static int 642 unp_attach(struct socket *so, int proto) 643 { 644 struct unpcb *unp = sotounpcb(so); 645 u_long sndspc, rcvspc; 646 int error; 647 648 KASSERT(unp == NULL); 649 650 switch (so->so_type) { 651 case SOCK_SEQPACKET: 652 /* FALLTHROUGH */ 653 case SOCK_STREAM: 654 if (so->so_lock == NULL) { 655 so->so_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NONE); 656 solock(so); 657 } 658 sndspc = unpst_sendspace; 659 rcvspc = unpst_recvspace; 660 break; 661 662 case SOCK_DGRAM: 663 if (so->so_lock == NULL) { 664 mutex_obj_hold(uipc_lock); 665 so->so_lock = uipc_lock; 666 solock(so); 667 } 668 sndspc = unpdg_sendspace; 669 rcvspc = unpdg_recvspace; 670 break; 671 672 default: 673 panic("unp_attach"); 674 } 675 676 if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) { 677 error = soreserve(so, sndspc, rcvspc); 678 if (error) { 679 return error; 680 } 681 } 682 683 unp = kmem_zalloc(sizeof(*unp), KM_SLEEP); 684 nanotime(&unp->unp_ctime); 685 unp->unp_socket = so; 686 so->so_pcb = unp; 687 688 KASSERT(solocked(so)); 689 return 0; 690 } 691 692 static void 693 unp_detach(struct socket *so) 694 { 695 struct unpcb *unp; 696 vnode_t *vp; 697 698 unp = sotounpcb(so); 699 KASSERT(unp != NULL); 700 KASSERT(solocked(so)); 701 retry: 702 if ((vp = unp->unp_vnode) != NULL) { 703 sounlock(so); 704 /* Acquire v_interlock to protect against unp_connect(). */ 705 /* XXXAD racy */ 706 mutex_enter(vp->v_interlock); 707 vp->v_socket = NULL; 708 mutex_exit(vp->v_interlock); 709 vrele(vp); 710 solock(so); 711 unp->unp_vnode = NULL; 712 } 713 if (unp->unp_conn) 714 unp_disconnect1(unp); 715 while (unp->unp_refs) { 716 KASSERT(solocked2(so, unp->unp_refs->unp_socket)); 717 if (unp_drop(unp->unp_refs, ECONNRESET)) { 718 solock(so); 719 goto retry; 720 } 721 } 722 soisdisconnected(so); 723 so->so_pcb = NULL; 724 if (unp_rights) { 725 /* 726 * Normally the receive buffer is flushed later, in sofree, 727 * but if our receive buffer holds references to files that 728 * are now garbage, we will enqueue those file references to 729 * the garbage collector and kick it into action. 730 */ 731 sorflush(so); 732 unp_free(unp); 733 unp_thread_kick(); 734 } else 735 unp_free(unp); 736 } 737 738 static int 739 unp_accept(struct socket *so, struct sockaddr *nam) 740 { 741 struct unpcb *unp = sotounpcb(so); 742 struct socket *so2; 743 744 KASSERT(solocked(so)); 745 KASSERT(nam != NULL); 746 747 /* XXX code review required to determine if unp can ever be NULL */ 748 if (unp == NULL) 749 return EINVAL; 750 751 KASSERT(so->so_lock == uipc_lock); 752 /* 753 * Mark the initiating STREAM socket as connected *ONLY* 754 * after it's been accepted. This prevents a client from 755 * overrunning a server and receiving ECONNREFUSED. 756 */ 757 if (unp->unp_conn == NULL) { 758 /* 759 * This will use the empty socket and will not 760 * allocate. 761 */ 762 unp_setaddr(so, nam, true); 763 return 0; 764 } 765 so2 = unp->unp_conn->unp_socket; 766 if (so2->so_state & SS_ISCONNECTING) { 767 KASSERT(solocked2(so, so->so_head)); 768 KASSERT(solocked2(so2, so->so_head)); 769 soisconnected(so2); 770 } 771 /* 772 * If the connection is fully established, break the 773 * association with uipc_lock and give the connected 774 * pair a separate lock to share. 775 * There is a race here: sotounpcb(so2)->unp_streamlock 776 * is not locked, so when changing so2->so_lock 777 * another thread can grab it while so->so_lock is still 778 * pointing to the (locked) uipc_lock. 779 * this should be harmless, except that this makes 780 * solocked2() and solocked() unreliable. 781 * Another problem is that unp_setaddr() expects the 782 * the socket locked. Grabing sotounpcb(so2)->unp_streamlock 783 * fixes both issues. 784 */ 785 mutex_enter(sotounpcb(so2)->unp_streamlock); 786 unp_setpeerlocks(so2, so); 787 /* 788 * Only now return peer's address, as we may need to 789 * block in order to allocate memory. 790 * 791 * XXX Minor race: connection can be broken while 792 * lock is dropped in unp_setaddr(). We will return 793 * error == 0 and sun_noname as the peer address. 794 */ 795 unp_setaddr(so, nam, true); 796 /* so_lock now points to unp_streamlock */ 797 mutex_exit(so2->so_lock); 798 return 0; 799 } 800 801 static int 802 unp_ioctl(struct socket *so, u_long cmd, void *nam, struct ifnet *ifp) 803 { 804 return EOPNOTSUPP; 805 } 806 807 static int 808 unp_stat(struct socket *so, struct stat *ub) 809 { 810 struct unpcb *unp; 811 struct socket *so2; 812 813 KASSERT(solocked(so)); 814 815 unp = sotounpcb(so); 816 if (unp == NULL) 817 return EINVAL; 818 819 ub->st_blksize = so->so_snd.sb_hiwat; 820 switch (so->so_type) { 821 case SOCK_SEQPACKET: /* FALLTHROUGH */ 822 case SOCK_STREAM: 823 if (unp->unp_conn == 0) 824 break; 825 826 so2 = unp->unp_conn->unp_socket; 827 KASSERT(solocked2(so, so2)); 828 ub->st_blksize += so2->so_rcv.sb_cc; 829 break; 830 default: 831 break; 832 } 833 ub->st_dev = NODEV; 834 if (unp->unp_ino == 0) 835 unp->unp_ino = unp_ino++; 836 ub->st_atimespec = ub->st_mtimespec = ub->st_ctimespec = unp->unp_ctime; 837 ub->st_ino = unp->unp_ino; 838 return (0); 839 } 840 841 static int 842 unp_peeraddr(struct socket *so, struct sockaddr *nam) 843 { 844 KASSERT(solocked(so)); 845 KASSERT(sotounpcb(so) != NULL); 846 KASSERT(nam != NULL); 847 848 unp_setaddr(so, nam, true); 849 return 0; 850 } 851 852 static int 853 unp_sockaddr(struct socket *so, struct sockaddr *nam) 854 { 855 KASSERT(solocked(so)); 856 KASSERT(sotounpcb(so) != NULL); 857 KASSERT(nam != NULL); 858 859 unp_setaddr(so, nam, false); 860 return 0; 861 } 862 863 /* 864 * we only need to perform this allocation until syscalls other than 865 * bind are adjusted to use sockaddr_big. 866 */ 867 static struct sockaddr_un * 868 makeun_sb(struct sockaddr *nam, size_t *addrlen) 869 { 870 struct sockaddr_un *sun; 871 872 *addrlen = nam->sa_len + 1; 873 sun = malloc(*addrlen, M_SONAME, M_WAITOK); 874 memcpy(sun, nam, nam->sa_len); 875 *(((char *)sun) + nam->sa_len) = '\0'; 876 return sun; 877 } 878 879 static int 880 unp_bind(struct socket *so, struct sockaddr *nam, struct lwp *l) 881 { 882 struct sockaddr_un *sun; 883 struct unpcb *unp; 884 vnode_t *vp; 885 struct vattr vattr; 886 size_t addrlen; 887 int error; 888 struct pathbuf *pb; 889 struct nameidata nd; 890 proc_t *p; 891 892 unp = sotounpcb(so); 893 894 KASSERT(solocked(so)); 895 KASSERT(unp != NULL); 896 KASSERT(nam != NULL); 897 898 if (unp->unp_vnode != NULL) 899 return (EINVAL); 900 if ((unp->unp_flags & UNP_BUSY) != 0) { 901 /* 902 * EALREADY may not be strictly accurate, but since this 903 * is a major application error it's hardly a big deal. 904 */ 905 return (EALREADY); 906 } 907 unp->unp_flags |= UNP_BUSY; 908 sounlock(so); 909 910 p = l->l_proc; 911 sun = makeun_sb(nam, &addrlen); 912 913 pb = pathbuf_create(sun->sun_path); 914 if (pb == NULL) { 915 error = ENOMEM; 916 goto bad; 917 } 918 NDINIT(&nd, CREATE, FOLLOW | LOCKPARENT | TRYEMULROOT, pb); 919 920 /* SHOULD BE ABLE TO ADOPT EXISTING AND wakeup() ALA FIFO's */ 921 if ((error = namei(&nd)) != 0) { 922 pathbuf_destroy(pb); 923 goto bad; 924 } 925 vp = nd.ni_vp; 926 if (vp != NULL) { 927 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd); 928 if (nd.ni_dvp == vp) 929 vrele(nd.ni_dvp); 930 else 931 vput(nd.ni_dvp); 932 vrele(vp); 933 pathbuf_destroy(pb); 934 error = EADDRINUSE; 935 goto bad; 936 } 937 vattr_null(&vattr); 938 vattr.va_type = VSOCK; 939 vattr.va_mode = ACCESSPERMS & ~(p->p_cwdi->cwdi_cmask); 940 error = VOP_CREATE(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr); 941 if (error) { 942 vput(nd.ni_dvp); 943 pathbuf_destroy(pb); 944 goto bad; 945 } 946 vp = nd.ni_vp; 947 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 948 solock(so); 949 vp->v_socket = unp->unp_socket; 950 unp->unp_vnode = vp; 951 unp->unp_addrlen = addrlen; 952 unp->unp_addr = sun; 953 unp->unp_connid.unp_pid = p->p_pid; 954 unp->unp_connid.unp_euid = kauth_cred_geteuid(l->l_cred); 955 unp->unp_connid.unp_egid = kauth_cred_getegid(l->l_cred); 956 unp->unp_flags |= UNP_EIDSBIND; 957 VOP_UNLOCK(vp); 958 vput(nd.ni_dvp); 959 unp->unp_flags &= ~UNP_BUSY; 960 pathbuf_destroy(pb); 961 return (0); 962 963 bad: 964 free(sun, M_SONAME); 965 solock(so); 966 unp->unp_flags &= ~UNP_BUSY; 967 return (error); 968 } 969 970 static int 971 unp_listen(struct socket *so, struct lwp *l) 972 { 973 struct unpcb *unp = sotounpcb(so); 974 975 KASSERT(solocked(so)); 976 KASSERT(unp != NULL); 977 978 /* 979 * If the socket can accept a connection, it must be 980 * locked by uipc_lock. 981 */ 982 unp_resetlock(so); 983 if (unp->unp_vnode == NULL) 984 return EINVAL; 985 986 return 0; 987 } 988 989 static int 990 unp_disconnect(struct socket *so) 991 { 992 KASSERT(solocked(so)); 993 KASSERT(sotounpcb(so) != NULL); 994 995 unp_disconnect1(sotounpcb(so)); 996 return 0; 997 } 998 999 static int 1000 unp_shutdown(struct socket *so) 1001 { 1002 KASSERT(solocked(so)); 1003 KASSERT(sotounpcb(so) != NULL); 1004 1005 socantsendmore(so); 1006 unp_shutdown1(sotounpcb(so)); 1007 return 0; 1008 } 1009 1010 static int 1011 unp_abort(struct socket *so) 1012 { 1013 KASSERT(solocked(so)); 1014 KASSERT(sotounpcb(so) != NULL); 1015 1016 (void)unp_drop(sotounpcb(so), ECONNABORTED); 1017 KASSERT(so->so_head == NULL); 1018 KASSERT(so->so_pcb != NULL); 1019 unp_detach(so); 1020 return 0; 1021 } 1022 1023 static int 1024 unp_connect1(struct socket *so, struct socket *so2, struct lwp *l) 1025 { 1026 struct unpcb *unp = sotounpcb(so); 1027 struct unpcb *unp2; 1028 1029 if (so2->so_type != so->so_type) 1030 return EPROTOTYPE; 1031 1032 /* 1033 * All three sockets involved must be locked by same lock: 1034 * 1035 * local endpoint (so) 1036 * remote endpoint (so2) 1037 * queue head (so2->so_head, only if PR_CONNREQUIRED) 1038 */ 1039 KASSERT(solocked2(so, so2)); 1040 KASSERT(so->so_head == NULL); 1041 if (so2->so_head != NULL) { 1042 KASSERT(so2->so_lock == uipc_lock); 1043 KASSERT(solocked2(so2, so2->so_head)); 1044 } 1045 1046 unp2 = sotounpcb(so2); 1047 unp->unp_conn = unp2; 1048 1049 if ((so->so_proto->pr_flags & PR_CONNREQUIRED) != 0) { 1050 unp2->unp_connid.unp_pid = l->l_proc->p_pid; 1051 unp2->unp_connid.unp_euid = kauth_cred_geteuid(l->l_cred); 1052 unp2->unp_connid.unp_egid = kauth_cred_getegid(l->l_cred); 1053 unp2->unp_flags |= UNP_EIDSVALID; 1054 if (unp2->unp_flags & UNP_EIDSBIND) { 1055 unp->unp_connid = unp2->unp_connid; 1056 unp->unp_flags |= UNP_EIDSVALID; 1057 } 1058 } 1059 1060 switch (so->so_type) { 1061 1062 case SOCK_DGRAM: 1063 unp->unp_nextref = unp2->unp_refs; 1064 unp2->unp_refs = unp; 1065 soisconnected(so); 1066 break; 1067 1068 case SOCK_SEQPACKET: /* FALLTHROUGH */ 1069 case SOCK_STREAM: 1070 1071 /* 1072 * SOCK_SEQPACKET and SOCK_STREAM cases are handled by callers 1073 * which are unp_connect() or unp_connect2(). 1074 */ 1075 1076 break; 1077 1078 default: 1079 panic("unp_connect1"); 1080 } 1081 1082 return 0; 1083 } 1084 1085 int 1086 unp_connect(struct socket *so, struct sockaddr *nam, struct lwp *l) 1087 { 1088 struct sockaddr_un *sun; 1089 vnode_t *vp; 1090 struct socket *so2, *so3; 1091 struct unpcb *unp, *unp2, *unp3; 1092 size_t addrlen; 1093 int error; 1094 struct pathbuf *pb; 1095 struct nameidata nd; 1096 1097 unp = sotounpcb(so); 1098 if ((unp->unp_flags & UNP_BUSY) != 0) { 1099 /* 1100 * EALREADY may not be strictly accurate, but since this 1101 * is a major application error it's hardly a big deal. 1102 */ 1103 return (EALREADY); 1104 } 1105 unp->unp_flags |= UNP_BUSY; 1106 sounlock(so); 1107 1108 sun = makeun_sb(nam, &addrlen); 1109 pb = pathbuf_create(sun->sun_path); 1110 if (pb == NULL) { 1111 error = ENOMEM; 1112 goto bad2; 1113 } 1114 1115 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | TRYEMULROOT, pb); 1116 1117 if ((error = namei(&nd)) != 0) { 1118 pathbuf_destroy(pb); 1119 goto bad2; 1120 } 1121 vp = nd.ni_vp; 1122 if (vp->v_type != VSOCK) { 1123 error = ENOTSOCK; 1124 goto bad; 1125 } 1126 pathbuf_destroy(pb); 1127 if ((error = VOP_ACCESS(vp, VWRITE, l->l_cred)) != 0) 1128 goto bad; 1129 /* Acquire v_interlock to protect against unp_detach(). */ 1130 mutex_enter(vp->v_interlock); 1131 so2 = vp->v_socket; 1132 if (so2 == NULL) { 1133 mutex_exit(vp->v_interlock); 1134 error = ECONNREFUSED; 1135 goto bad; 1136 } 1137 if (so->so_type != so2->so_type) { 1138 mutex_exit(vp->v_interlock); 1139 error = EPROTOTYPE; 1140 goto bad; 1141 } 1142 solock(so); 1143 unp_resetlock(so); 1144 mutex_exit(vp->v_interlock); 1145 if ((so->so_proto->pr_flags & PR_CONNREQUIRED) != 0) { 1146 /* 1147 * This may seem somewhat fragile but is OK: if we can 1148 * see SO_ACCEPTCONN set on the endpoint, then it must 1149 * be locked by the domain-wide uipc_lock. 1150 */ 1151 KASSERT((so2->so_options & SO_ACCEPTCONN) == 0 || 1152 so2->so_lock == uipc_lock); 1153 if ((so2->so_options & SO_ACCEPTCONN) == 0 || 1154 (so3 = sonewconn(so2, false)) == NULL) { 1155 error = ECONNREFUSED; 1156 sounlock(so); 1157 goto bad; 1158 } 1159 unp2 = sotounpcb(so2); 1160 unp3 = sotounpcb(so3); 1161 if (unp2->unp_addr) { 1162 unp3->unp_addr = malloc(unp2->unp_addrlen, 1163 M_SONAME, M_WAITOK); 1164 memcpy(unp3->unp_addr, unp2->unp_addr, 1165 unp2->unp_addrlen); 1166 unp3->unp_addrlen = unp2->unp_addrlen; 1167 } 1168 unp3->unp_flags = unp2->unp_flags; 1169 so2 = so3; 1170 } 1171 error = unp_connect1(so, so2, l); 1172 if (error) { 1173 sounlock(so); 1174 goto bad; 1175 } 1176 unp2 = sotounpcb(so2); 1177 switch (so->so_type) { 1178 1179 /* 1180 * SOCK_DGRAM and default cases are handled in prior call to 1181 * unp_connect1(), do not add a default case without fixing 1182 * unp_connect1(). 1183 */ 1184 1185 case SOCK_SEQPACKET: /* FALLTHROUGH */ 1186 case SOCK_STREAM: 1187 unp2->unp_conn = unp; 1188 if ((unp->unp_flags | unp2->unp_flags) & UNP_CONNWAIT) 1189 soisconnecting(so); 1190 else 1191 soisconnected(so); 1192 soisconnected(so2); 1193 /* 1194 * If the connection is fully established, break the 1195 * association with uipc_lock and give the connected 1196 * pair a seperate lock to share. 1197 */ 1198 KASSERT(so2->so_head != NULL); 1199 unp_setpeerlocks(so, so2); 1200 break; 1201 1202 } 1203 sounlock(so); 1204 bad: 1205 vput(vp); 1206 bad2: 1207 free(sun, M_SONAME); 1208 solock(so); 1209 unp->unp_flags &= ~UNP_BUSY; 1210 return (error); 1211 } 1212 1213 int 1214 unp_connect2(struct socket *so, struct socket *so2) 1215 { 1216 struct unpcb *unp = sotounpcb(so); 1217 struct unpcb *unp2; 1218 int error = 0; 1219 1220 KASSERT(solocked2(so, so2)); 1221 1222 error = unp_connect1(so, so2, curlwp); 1223 if (error) 1224 return error; 1225 1226 unp2 = sotounpcb(so2); 1227 switch (so->so_type) { 1228 1229 /* 1230 * SOCK_DGRAM and default cases are handled in prior call to 1231 * unp_connect1(), do not add a default case without fixing 1232 * unp_connect1(). 1233 */ 1234 1235 case SOCK_SEQPACKET: /* FALLTHROUGH */ 1236 case SOCK_STREAM: 1237 unp2->unp_conn = unp; 1238 if ((so->so_proto->pr_flags & PR_CONNREQUIRED) != 0) { 1239 unp->unp_connid = unp2->unp_connid; 1240 unp->unp_flags |= UNP_EIDSVALID; 1241 } 1242 soisconnected(so); 1243 soisconnected(so2); 1244 break; 1245 1246 } 1247 return error; 1248 } 1249 1250 static void 1251 unp_disconnect1(struct unpcb *unp) 1252 { 1253 struct unpcb *unp2 = unp->unp_conn; 1254 struct socket *so; 1255 1256 if (unp2 == 0) 1257 return; 1258 unp->unp_conn = 0; 1259 so = unp->unp_socket; 1260 switch (so->so_type) { 1261 case SOCK_DGRAM: 1262 if (unp2->unp_refs == unp) 1263 unp2->unp_refs = unp->unp_nextref; 1264 else { 1265 unp2 = unp2->unp_refs; 1266 for (;;) { 1267 KASSERT(solocked2(so, unp2->unp_socket)); 1268 if (unp2 == 0) 1269 panic("unp_disconnect1"); 1270 if (unp2->unp_nextref == unp) 1271 break; 1272 unp2 = unp2->unp_nextref; 1273 } 1274 unp2->unp_nextref = unp->unp_nextref; 1275 } 1276 unp->unp_nextref = 0; 1277 so->so_state &= ~SS_ISCONNECTED; 1278 break; 1279 1280 case SOCK_SEQPACKET: /* FALLTHROUGH */ 1281 case SOCK_STREAM: 1282 KASSERT(solocked2(so, unp2->unp_socket)); 1283 soisdisconnected(so); 1284 unp2->unp_conn = 0; 1285 soisdisconnected(unp2->unp_socket); 1286 break; 1287 } 1288 } 1289 1290 static void 1291 unp_shutdown1(struct unpcb *unp) 1292 { 1293 struct socket *so; 1294 1295 switch(unp->unp_socket->so_type) { 1296 case SOCK_SEQPACKET: /* FALLTHROUGH */ 1297 case SOCK_STREAM: 1298 if (unp->unp_conn && (so = unp->unp_conn->unp_socket)) 1299 socantrcvmore(so); 1300 break; 1301 default: 1302 break; 1303 } 1304 } 1305 1306 static bool 1307 unp_drop(struct unpcb *unp, int errno) 1308 { 1309 struct socket *so = unp->unp_socket; 1310 1311 KASSERT(solocked(so)); 1312 1313 so->so_error = errno; 1314 unp_disconnect1(unp); 1315 if (so->so_head) { 1316 so->so_pcb = NULL; 1317 /* sofree() drops the socket lock */ 1318 sofree(so); 1319 unp_free(unp); 1320 return true; 1321 } 1322 return false; 1323 } 1324 1325 #ifdef notdef 1326 unp_drain(void) 1327 { 1328 1329 } 1330 #endif 1331 1332 int 1333 unp_externalize(struct mbuf *rights, struct lwp *l, int flags) 1334 { 1335 struct cmsghdr * const cm = mtod(rights, struct cmsghdr *); 1336 struct proc * const p = l->l_proc; 1337 file_t **rp; 1338 int error = 0; 1339 1340 const size_t nfds = (cm->cmsg_len - CMSG_ALIGN(sizeof(*cm))) / 1341 sizeof(file_t *); 1342 if (nfds == 0) 1343 goto noop; 1344 1345 int * const fdp = kmem_alloc(nfds * sizeof(int), KM_SLEEP); 1346 rw_enter(&p->p_cwdi->cwdi_lock, RW_READER); 1347 1348 /* Make sure the recipient should be able to see the files.. */ 1349 rp = (file_t **)CMSG_DATA(cm); 1350 for (size_t i = 0; i < nfds; i++) { 1351 file_t * const fp = *rp++; 1352 if (fp == NULL) { 1353 error = EINVAL; 1354 goto out; 1355 } 1356 /* 1357 * If we are in a chroot'ed directory, and 1358 * someone wants to pass us a directory, make 1359 * sure it's inside the subtree we're allowed 1360 * to access. 1361 */ 1362 if (p->p_cwdi->cwdi_rdir != NULL && fp->f_type == DTYPE_VNODE) { 1363 vnode_t *vp = fp->f_vnode; 1364 if ((vp->v_type == VDIR) && 1365 !vn_isunder(vp, p->p_cwdi->cwdi_rdir, l)) { 1366 error = EPERM; 1367 goto out; 1368 } 1369 } 1370 } 1371 1372 restart: 1373 /* 1374 * First loop -- allocate file descriptor table slots for the 1375 * new files. 1376 */ 1377 for (size_t i = 0; i < nfds; i++) { 1378 if ((error = fd_alloc(p, 0, &fdp[i])) != 0) { 1379 /* 1380 * Back out what we've done so far. 1381 */ 1382 while (i-- > 0) { 1383 fd_abort(p, NULL, fdp[i]); 1384 } 1385 if (error == ENOSPC) { 1386 fd_tryexpand(p); 1387 error = 0; 1388 goto restart; 1389 } 1390 /* 1391 * This is the error that has historically 1392 * been returned, and some callers may 1393 * expect it. 1394 */ 1395 error = EMSGSIZE; 1396 goto out; 1397 } 1398 } 1399 1400 /* 1401 * Now that adding them has succeeded, update all of the 1402 * file passing state and affix the descriptors. 1403 */ 1404 rp = (file_t **)CMSG_DATA(cm); 1405 int *ofdp = (int *)CMSG_DATA(cm); 1406 for (size_t i = 0; i < nfds; i++) { 1407 file_t * const fp = *rp++; 1408 const int fd = fdp[i]; 1409 atomic_dec_uint(&unp_rights); 1410 fd_set_exclose(l, fd, (flags & O_CLOEXEC) != 0); 1411 fd_affix(p, fp, fd); 1412 /* 1413 * Done with this file pointer, replace it with a fd; 1414 */ 1415 *ofdp++ = fd; 1416 mutex_enter(&fp->f_lock); 1417 fp->f_msgcount--; 1418 mutex_exit(&fp->f_lock); 1419 /* 1420 * Note that fd_affix() adds a reference to the file. 1421 * The file may already have been closed by another 1422 * LWP in the process, so we must drop the reference 1423 * added by unp_internalize() with closef(). 1424 */ 1425 closef(fp); 1426 } 1427 1428 /* 1429 * Adjust length, in case of transition from large file_t 1430 * pointers to ints. 1431 */ 1432 if (sizeof(file_t *) != sizeof(int)) { 1433 cm->cmsg_len = CMSG_LEN(nfds * sizeof(int)); 1434 rights->m_len = CMSG_SPACE(nfds * sizeof(int)); 1435 } 1436 out: 1437 if (__predict_false(error != 0)) { 1438 file_t **const fpp = (file_t **)CMSG_DATA(cm); 1439 for (size_t i = 0; i < nfds; i++) 1440 unp_discard_now(fpp[i]); 1441 /* 1442 * Truncate the array so that nobody will try to interpret 1443 * what is now garbage in it. 1444 */ 1445 cm->cmsg_len = CMSG_LEN(0); 1446 rights->m_len = CMSG_SPACE(0); 1447 } 1448 rw_exit(&p->p_cwdi->cwdi_lock); 1449 kmem_free(fdp, nfds * sizeof(int)); 1450 1451 noop: 1452 /* 1453 * Don't disclose kernel memory in the alignment space. 1454 */ 1455 KASSERT(cm->cmsg_len <= rights->m_len); 1456 memset(&mtod(rights, char *)[cm->cmsg_len], 0, rights->m_len - 1457 cm->cmsg_len); 1458 return error; 1459 } 1460 1461 static int 1462 unp_internalize(struct mbuf **controlp) 1463 { 1464 filedesc_t *fdescp = curlwp->l_fd; 1465 struct mbuf *control = *controlp; 1466 struct cmsghdr *newcm, *cm = mtod(control, struct cmsghdr *); 1467 file_t **rp, **files; 1468 file_t *fp; 1469 int i, fd, *fdp; 1470 int nfds, error; 1471 u_int maxmsg; 1472 1473 error = 0; 1474 newcm = NULL; 1475 1476 /* Sanity check the control message header. */ 1477 if (cm->cmsg_type != SCM_RIGHTS || cm->cmsg_level != SOL_SOCKET || 1478 cm->cmsg_len > control->m_len || 1479 cm->cmsg_len < CMSG_ALIGN(sizeof(*cm))) 1480 return (EINVAL); 1481 1482 /* 1483 * Verify that the file descriptors are valid, and acquire 1484 * a reference to each. 1485 */ 1486 nfds = (cm->cmsg_len - CMSG_ALIGN(sizeof(*cm))) / sizeof(int); 1487 fdp = (int *)CMSG_DATA(cm); 1488 maxmsg = maxfiles / unp_rights_ratio; 1489 for (i = 0; i < nfds; i++) { 1490 fd = *fdp++; 1491 if (atomic_inc_uint_nv(&unp_rights) > maxmsg) { 1492 atomic_dec_uint(&unp_rights); 1493 nfds = i; 1494 error = EAGAIN; 1495 goto out; 1496 } 1497 if ((fp = fd_getfile(fd)) == NULL 1498 || fp->f_type == DTYPE_KQUEUE) { 1499 if (fp) 1500 fd_putfile(fd); 1501 atomic_dec_uint(&unp_rights); 1502 nfds = i; 1503 error = EBADF; 1504 goto out; 1505 } 1506 } 1507 1508 /* Allocate new space and copy header into it. */ 1509 newcm = malloc(CMSG_SPACE(nfds * sizeof(file_t *)), M_MBUF, M_WAITOK); 1510 if (newcm == NULL) { 1511 error = E2BIG; 1512 goto out; 1513 } 1514 memcpy(newcm, cm, sizeof(struct cmsghdr)); 1515 files = (file_t **)CMSG_DATA(newcm); 1516 1517 /* 1518 * Transform the file descriptors into file_t pointers, in 1519 * reverse order so that if pointers are bigger than ints, the 1520 * int won't get until we're done. No need to lock, as we have 1521 * already validated the descriptors with fd_getfile(). 1522 */ 1523 fdp = (int *)CMSG_DATA(cm) + nfds; 1524 rp = files + nfds; 1525 for (i = 0; i < nfds; i++) { 1526 fp = fdescp->fd_dt->dt_ff[*--fdp]->ff_file; 1527 KASSERT(fp != NULL); 1528 mutex_enter(&fp->f_lock); 1529 *--rp = fp; 1530 fp->f_count++; 1531 fp->f_msgcount++; 1532 mutex_exit(&fp->f_lock); 1533 } 1534 1535 out: 1536 /* Release descriptor references. */ 1537 fdp = (int *)CMSG_DATA(cm); 1538 for (i = 0; i < nfds; i++) { 1539 fd_putfile(*fdp++); 1540 if (error != 0) { 1541 atomic_dec_uint(&unp_rights); 1542 } 1543 } 1544 1545 if (error == 0) { 1546 if (control->m_flags & M_EXT) { 1547 m_freem(control); 1548 *controlp = control = m_get(M_WAIT, MT_CONTROL); 1549 } 1550 MEXTADD(control, newcm, CMSG_SPACE(nfds * sizeof(file_t *)), 1551 M_MBUF, NULL, NULL); 1552 cm = newcm; 1553 /* 1554 * Adjust message & mbuf to note amount of space 1555 * actually used. 1556 */ 1557 cm->cmsg_len = CMSG_LEN(nfds * sizeof(file_t *)); 1558 control->m_len = CMSG_SPACE(nfds * sizeof(file_t *)); 1559 } 1560 1561 return error; 1562 } 1563 1564 struct mbuf * 1565 unp_addsockcred(struct lwp *l, struct mbuf *control) 1566 { 1567 struct sockcred *sc; 1568 struct mbuf *m; 1569 void *p; 1570 1571 m = sbcreatecontrol1(&p, SOCKCREDSIZE(kauth_cred_ngroups(l->l_cred)), 1572 SCM_CREDS, SOL_SOCKET, M_WAITOK); 1573 if (m == NULL) 1574 return control; 1575 1576 sc = p; 1577 sc->sc_uid = kauth_cred_getuid(l->l_cred); 1578 sc->sc_euid = kauth_cred_geteuid(l->l_cred); 1579 sc->sc_gid = kauth_cred_getgid(l->l_cred); 1580 sc->sc_egid = kauth_cred_getegid(l->l_cred); 1581 sc->sc_ngroups = kauth_cred_ngroups(l->l_cred); 1582 1583 for (int i = 0; i < sc->sc_ngroups; i++) 1584 sc->sc_groups[i] = kauth_cred_group(l->l_cred, i); 1585 1586 return m_add(control, m); 1587 } 1588 1589 /* 1590 * Do a mark-sweep GC of files in the system, to free up any which are 1591 * caught in flight to an about-to-be-closed socket. Additionally, 1592 * process deferred file closures. 1593 */ 1594 static void 1595 unp_gc(file_t *dp) 1596 { 1597 extern struct domain unixdomain; 1598 file_t *fp, *np; 1599 struct socket *so, *so1; 1600 u_int i, oflags, rflags; 1601 bool didwork; 1602 1603 KASSERT(curlwp == unp_thread_lwp); 1604 KASSERT(mutex_owned(&filelist_lock)); 1605 1606 /* 1607 * First, process deferred file closures. 1608 */ 1609 while (!SLIST_EMPTY(&unp_thread_discard)) { 1610 fp = SLIST_FIRST(&unp_thread_discard); 1611 KASSERT(fp->f_unpcount > 0); 1612 KASSERT(fp->f_count > 0); 1613 KASSERT(fp->f_msgcount > 0); 1614 KASSERT(fp->f_count >= fp->f_unpcount); 1615 KASSERT(fp->f_count >= fp->f_msgcount); 1616 KASSERT(fp->f_msgcount >= fp->f_unpcount); 1617 SLIST_REMOVE_HEAD(&unp_thread_discard, f_unplist); 1618 i = fp->f_unpcount; 1619 fp->f_unpcount = 0; 1620 mutex_exit(&filelist_lock); 1621 for (; i != 0; i--) { 1622 unp_discard_now(fp); 1623 } 1624 mutex_enter(&filelist_lock); 1625 } 1626 1627 /* 1628 * Clear mark bits. Ensure that we don't consider new files 1629 * entering the file table during this loop (they will not have 1630 * FSCAN set). 1631 */ 1632 unp_defer = 0; 1633 LIST_FOREACH(fp, &filehead, f_list) { 1634 for (oflags = fp->f_flag;; oflags = rflags) { 1635 rflags = atomic_cas_uint(&fp->f_flag, oflags, 1636 (oflags | FSCAN) & ~(FMARK|FDEFER)); 1637 if (__predict_true(oflags == rflags)) { 1638 break; 1639 } 1640 } 1641 } 1642 1643 /* 1644 * Iterate over the set of sockets, marking ones believed (based on 1645 * refcount) to be referenced from a process, and marking for rescan 1646 * sockets which are queued on a socket. Recan continues descending 1647 * and searching for sockets referenced by sockets (FDEFER), until 1648 * there are no more socket->socket references to be discovered. 1649 */ 1650 do { 1651 didwork = false; 1652 for (fp = LIST_FIRST(&filehead); fp != NULL; fp = np) { 1653 KASSERT(mutex_owned(&filelist_lock)); 1654 np = LIST_NEXT(fp, f_list); 1655 mutex_enter(&fp->f_lock); 1656 if ((fp->f_flag & FDEFER) != 0) { 1657 atomic_and_uint(&fp->f_flag, ~FDEFER); 1658 unp_defer--; 1659 if (fp->f_count == 0) { 1660 /* 1661 * XXX: closef() doesn't pay attention 1662 * to FDEFER 1663 */ 1664 mutex_exit(&fp->f_lock); 1665 continue; 1666 } 1667 } else { 1668 if (fp->f_count == 0 || 1669 (fp->f_flag & FMARK) != 0 || 1670 fp->f_count == fp->f_msgcount || 1671 fp->f_unpcount != 0) { 1672 mutex_exit(&fp->f_lock); 1673 continue; 1674 } 1675 } 1676 atomic_or_uint(&fp->f_flag, FMARK); 1677 1678 if (fp->f_type != DTYPE_SOCKET || 1679 (so = fp->f_socket) == NULL || 1680 so->so_proto->pr_domain != &unixdomain || 1681 (so->so_proto->pr_flags & PR_RIGHTS) == 0) { 1682 mutex_exit(&fp->f_lock); 1683 continue; 1684 } 1685 1686 /* Gain file ref, mark our position, and unlock. */ 1687 didwork = true; 1688 LIST_INSERT_AFTER(fp, dp, f_list); 1689 fp->f_count++; 1690 mutex_exit(&fp->f_lock); 1691 mutex_exit(&filelist_lock); 1692 1693 /* 1694 * Mark files referenced from sockets queued on the 1695 * accept queue as well. 1696 */ 1697 solock(so); 1698 unp_scan(so->so_rcv.sb_mb, unp_mark, 0); 1699 if ((so->so_options & SO_ACCEPTCONN) != 0) { 1700 TAILQ_FOREACH(so1, &so->so_q0, so_qe) { 1701 unp_scan(so1->so_rcv.sb_mb, unp_mark, 0); 1702 } 1703 TAILQ_FOREACH(so1, &so->so_q, so_qe) { 1704 unp_scan(so1->so_rcv.sb_mb, unp_mark, 0); 1705 } 1706 } 1707 sounlock(so); 1708 1709 /* Re-lock and restart from where we left off. */ 1710 closef(fp); 1711 mutex_enter(&filelist_lock); 1712 np = LIST_NEXT(dp, f_list); 1713 LIST_REMOVE(dp, f_list); 1714 } 1715 /* 1716 * Bail early if we did nothing in the loop above. Could 1717 * happen because of concurrent activity causing unp_defer 1718 * to get out of sync. 1719 */ 1720 } while (unp_defer != 0 && didwork); 1721 1722 /* 1723 * Sweep pass. 1724 * 1725 * We grab an extra reference to each of the files that are 1726 * not otherwise accessible and then free the rights that are 1727 * stored in messages on them. 1728 */ 1729 for (fp = LIST_FIRST(&filehead); fp != NULL; fp = np) { 1730 KASSERT(mutex_owned(&filelist_lock)); 1731 np = LIST_NEXT(fp, f_list); 1732 mutex_enter(&fp->f_lock); 1733 1734 /* 1735 * Ignore non-sockets. 1736 * Ignore dead sockets, or sockets with pending close. 1737 * Ignore sockets obviously referenced elsewhere. 1738 * Ignore sockets marked as referenced by our scan. 1739 * Ignore new sockets that did not exist during the scan. 1740 */ 1741 if (fp->f_type != DTYPE_SOCKET || 1742 fp->f_count == 0 || fp->f_unpcount != 0 || 1743 fp->f_count != fp->f_msgcount || 1744 (fp->f_flag & (FMARK | FSCAN)) != FSCAN) { 1745 mutex_exit(&fp->f_lock); 1746 continue; 1747 } 1748 1749 /* Gain file ref, mark our position, and unlock. */ 1750 LIST_INSERT_AFTER(fp, dp, f_list); 1751 fp->f_count++; 1752 mutex_exit(&fp->f_lock); 1753 mutex_exit(&filelist_lock); 1754 1755 /* 1756 * Flush all data from the socket's receive buffer. 1757 * This will cause files referenced only by the 1758 * socket to be queued for close. 1759 */ 1760 so = fp->f_socket; 1761 solock(so); 1762 sorflush(so); 1763 sounlock(so); 1764 1765 /* Re-lock and restart from where we left off. */ 1766 closef(fp); 1767 mutex_enter(&filelist_lock); 1768 np = LIST_NEXT(dp, f_list); 1769 LIST_REMOVE(dp, f_list); 1770 } 1771 } 1772 1773 /* 1774 * Garbage collector thread. While SCM_RIGHTS messages are in transit, 1775 * wake once per second to garbage collect. Run continually while we 1776 * have deferred closes to process. 1777 */ 1778 static void 1779 unp_thread(void *cookie) 1780 { 1781 file_t *dp; 1782 1783 /* Allocate a dummy file for our scans. */ 1784 if ((dp = fgetdummy()) == NULL) { 1785 panic("unp_thread"); 1786 } 1787 1788 mutex_enter(&filelist_lock); 1789 for (;;) { 1790 KASSERT(mutex_owned(&filelist_lock)); 1791 if (SLIST_EMPTY(&unp_thread_discard)) { 1792 if (unp_rights != 0) { 1793 (void)cv_timedwait(&unp_thread_cv, 1794 &filelist_lock, hz); 1795 } else { 1796 cv_wait(&unp_thread_cv, &filelist_lock); 1797 } 1798 } 1799 unp_gc(dp); 1800 } 1801 /* NOTREACHED */ 1802 } 1803 1804 /* 1805 * Kick the garbage collector into action if there is something for 1806 * it to process. 1807 */ 1808 static void 1809 unp_thread_kick(void) 1810 { 1811 1812 if (!SLIST_EMPTY(&unp_thread_discard) || unp_rights != 0) { 1813 mutex_enter(&filelist_lock); 1814 cv_signal(&unp_thread_cv); 1815 mutex_exit(&filelist_lock); 1816 } 1817 } 1818 1819 void 1820 unp_dispose(struct mbuf *m) 1821 { 1822 1823 if (m) 1824 unp_scan(m, unp_discard_later, 1); 1825 } 1826 1827 void 1828 unp_scan(struct mbuf *m0, void (*op)(file_t *), int discard) 1829 { 1830 struct mbuf *m; 1831 file_t **rp, *fp; 1832 struct cmsghdr *cm; 1833 int i, qfds; 1834 1835 while (m0) { 1836 for (m = m0; m; m = m->m_next) { 1837 if (m->m_type != MT_CONTROL || 1838 m->m_len < sizeof(*cm)) { 1839 continue; 1840 } 1841 cm = mtod(m, struct cmsghdr *); 1842 if (cm->cmsg_level != SOL_SOCKET || 1843 cm->cmsg_type != SCM_RIGHTS) 1844 continue; 1845 qfds = (cm->cmsg_len - CMSG_ALIGN(sizeof(*cm))) 1846 / sizeof(file_t *); 1847 rp = (file_t **)CMSG_DATA(cm); 1848 for (i = 0; i < qfds; i++) { 1849 fp = *rp; 1850 if (discard) { 1851 *rp = 0; 1852 } 1853 (*op)(fp); 1854 rp++; 1855 } 1856 } 1857 m0 = m0->m_nextpkt; 1858 } 1859 } 1860 1861 void 1862 unp_mark(file_t *fp) 1863 { 1864 1865 if (fp == NULL) 1866 return; 1867 1868 /* If we're already deferred, don't screw up the defer count */ 1869 mutex_enter(&fp->f_lock); 1870 if (fp->f_flag & (FMARK | FDEFER)) { 1871 mutex_exit(&fp->f_lock); 1872 return; 1873 } 1874 1875 /* 1876 * Minimize the number of deferrals... Sockets are the only type of 1877 * file which can hold references to another file, so just mark 1878 * other files, and defer unmarked sockets for the next pass. 1879 */ 1880 if (fp->f_type == DTYPE_SOCKET) { 1881 unp_defer++; 1882 KASSERT(fp->f_count != 0); 1883 atomic_or_uint(&fp->f_flag, FDEFER); 1884 } else { 1885 atomic_or_uint(&fp->f_flag, FMARK); 1886 } 1887 mutex_exit(&fp->f_lock); 1888 } 1889 1890 static void 1891 unp_discard_now(file_t *fp) 1892 { 1893 1894 if (fp == NULL) 1895 return; 1896 1897 KASSERT(fp->f_count > 0); 1898 KASSERT(fp->f_msgcount > 0); 1899 1900 mutex_enter(&fp->f_lock); 1901 fp->f_msgcount--; 1902 mutex_exit(&fp->f_lock); 1903 atomic_dec_uint(&unp_rights); 1904 (void)closef(fp); 1905 } 1906 1907 static void 1908 unp_discard_later(file_t *fp) 1909 { 1910 1911 if (fp == NULL) 1912 return; 1913 1914 KASSERT(fp->f_count > 0); 1915 KASSERT(fp->f_msgcount > 0); 1916 1917 mutex_enter(&filelist_lock); 1918 if (fp->f_unpcount++ == 0) { 1919 SLIST_INSERT_HEAD(&unp_thread_discard, fp, f_unplist); 1920 } 1921 mutex_exit(&filelist_lock); 1922 } 1923 1924 const struct pr_usrreqs unp_usrreqs = { 1925 .pr_attach = unp_attach, 1926 .pr_detach = unp_detach, 1927 .pr_accept = unp_accept, 1928 .pr_bind = unp_bind, 1929 .pr_listen = unp_listen, 1930 .pr_connect = unp_connect, 1931 .pr_connect2 = unp_connect2, 1932 .pr_disconnect = unp_disconnect, 1933 .pr_shutdown = unp_shutdown, 1934 .pr_abort = unp_abort, 1935 .pr_ioctl = unp_ioctl, 1936 .pr_stat = unp_stat, 1937 .pr_peeraddr = unp_peeraddr, 1938 .pr_sockaddr = unp_sockaddr, 1939 .pr_rcvd = unp_rcvd, 1940 .pr_recvoob = unp_recvoob, 1941 .pr_send = unp_send, 1942 .pr_sendoob = unp_sendoob, 1943 }; 1944