uipc_usrreq.c revision 1.87
1 /* $NetBSD: uipc_usrreq.c,v 1.87 2006/03/01 02:06:11 christos Exp $ */ 2 3 /*- 4 * Copyright (c) 1998, 2000, 2004 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. 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 * 3. All advertising materials mentioning features or use of this software 20 * must display the following acknowledgement: 21 * This product includes software developed by the NetBSD 22 * Foundation, Inc. and its contributors. 23 * 4. Neither the name of The NetBSD Foundation nor the names of its 24 * contributors may be used to endorse or promote products derived 25 * from this software without specific prior written permission. 26 * 27 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 28 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 29 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 30 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 31 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 32 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 33 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 34 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 35 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 36 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 37 * POSSIBILITY OF SUCH DAMAGE. 38 */ 39 40 /* 41 * Copyright (c) 1982, 1986, 1989, 1991, 1993 42 * The Regents of the University of California. All rights reserved. 43 * 44 * Redistribution and use in source and binary forms, with or without 45 * modification, are permitted provided that the following conditions 46 * are met: 47 * 1. Redistributions of source code must retain the above copyright 48 * notice, this list of conditions and the following disclaimer. 49 * 2. Redistributions in binary form must reproduce the above copyright 50 * notice, this list of conditions and the following disclaimer in the 51 * documentation and/or other materials provided with the distribution. 52 * 3. Neither the name of the University nor the names of its contributors 53 * may be used to endorse or promote products derived from this software 54 * without specific prior written permission. 55 * 56 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 57 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 58 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 59 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 60 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 61 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 62 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 63 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 64 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 65 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 66 * SUCH DAMAGE. 67 * 68 * @(#)uipc_usrreq.c 8.9 (Berkeley) 5/14/95 69 */ 70 71 /* 72 * Copyright (c) 1997 Christopher G. Demetriou. All rights reserved. 73 * 74 * Redistribution and use in source and binary forms, with or without 75 * modification, are permitted provided that the following conditions 76 * are met: 77 * 1. Redistributions of source code must retain the above copyright 78 * notice, this list of conditions and the following disclaimer. 79 * 2. Redistributions in binary form must reproduce the above copyright 80 * notice, this list of conditions and the following disclaimer in the 81 * documentation and/or other materials provided with the distribution. 82 * 3. All advertising materials mentioning features or use of this software 83 * must display the following acknowledgement: 84 * This product includes software developed by the University of 85 * California, Berkeley and its contributors. 86 * 4. Neither the name of the University nor the names of its contributors 87 * may be used to endorse or promote products derived from this software 88 * without specific prior written permission. 89 * 90 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 91 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 92 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 93 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 94 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 95 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 96 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 97 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 98 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 99 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 100 * SUCH DAMAGE. 101 * 102 * @(#)uipc_usrreq.c 8.9 (Berkeley) 5/14/95 103 */ 104 105 #include <sys/cdefs.h> 106 __KERNEL_RCSID(0, "$NetBSD: uipc_usrreq.c,v 1.87 2006/03/01 02:06:11 christos Exp $"); 107 108 #include <sys/param.h> 109 #include <sys/systm.h> 110 #include <sys/proc.h> 111 #include <sys/filedesc.h> 112 #include <sys/domain.h> 113 #include <sys/protosw.h> 114 #include <sys/socket.h> 115 #include <sys/socketvar.h> 116 #include <sys/unpcb.h> 117 #include <sys/un.h> 118 #include <sys/namei.h> 119 #include <sys/vnode.h> 120 #include <sys/file.h> 121 #include <sys/stat.h> 122 #include <sys/mbuf.h> 123 124 /* 125 * Unix communications domain. 126 * 127 * TODO: 128 * SEQPACKET, RDM 129 * rethink name space problems 130 * need a proper out-of-band 131 */ 132 const struct sockaddr_un sun_noname = { sizeof(sun_noname), AF_LOCAL }; 133 ino_t unp_ino; /* prototype for fake inode numbers */ 134 135 struct mbuf *unp_addsockcred(struct proc *, struct mbuf *); 136 137 int 138 unp_output(struct mbuf *m, struct mbuf *control, struct unpcb *unp, 139 struct proc *p) 140 { 141 struct socket *so2; 142 const struct sockaddr_un *sun; 143 144 so2 = unp->unp_conn->unp_socket; 145 if (unp->unp_addr) 146 sun = unp->unp_addr; 147 else 148 sun = &sun_noname; 149 if (unp->unp_conn->unp_flags & UNP_WANTCRED) 150 control = unp_addsockcred(p, control); 151 if (sbappendaddr(&so2->so_rcv, (const struct sockaddr *)sun, m, 152 control) == 0) { 153 m_freem(control); 154 m_freem(m); 155 so2->so_rcv.sb_overflowed++; 156 return (ENOBUFS); 157 } else { 158 sorwakeup(so2); 159 return (0); 160 } 161 } 162 163 void 164 unp_setsockaddr(struct unpcb *unp, struct mbuf *nam) 165 { 166 const struct sockaddr_un *sun; 167 168 if (unp->unp_addr) 169 sun = unp->unp_addr; 170 else 171 sun = &sun_noname; 172 nam->m_len = sun->sun_len; 173 if (nam->m_len > MLEN) 174 MEXTMALLOC(nam, nam->m_len, M_WAITOK); 175 memcpy(mtod(nam, caddr_t), sun, (size_t)nam->m_len); 176 } 177 178 void 179 unp_setpeeraddr(struct unpcb *unp, struct mbuf *nam) 180 { 181 const struct sockaddr_un *sun; 182 183 if (unp->unp_conn && unp->unp_conn->unp_addr) 184 sun = unp->unp_conn->unp_addr; 185 else 186 sun = &sun_noname; 187 nam->m_len = sun->sun_len; 188 if (nam->m_len > MLEN) 189 MEXTMALLOC(nam, nam->m_len, M_WAITOK); 190 memcpy(mtod(nam, caddr_t), sun, (size_t)nam->m_len); 191 } 192 193 /*ARGSUSED*/ 194 int 195 uipc_usrreq(struct socket *so, int req, struct mbuf *m, struct mbuf *nam, 196 struct mbuf *control, struct lwp *l) 197 { 198 struct unpcb *unp = sotounpcb(so); 199 struct socket *so2; 200 struct proc *p; 201 u_int newhiwat; 202 int error = 0; 203 204 if (req == PRU_CONTROL) 205 return (EOPNOTSUPP); 206 207 #ifdef DIAGNOSTIC 208 if (req != PRU_SEND && req != PRU_SENDOOB && control) 209 panic("uipc_usrreq: unexpected control mbuf"); 210 #endif 211 p = l ? l->l_proc : NULL; 212 if (unp == 0 && req != PRU_ATTACH) { 213 error = EINVAL; 214 goto release; 215 } 216 217 switch (req) { 218 219 case PRU_ATTACH: 220 if (unp != 0) { 221 error = EISCONN; 222 break; 223 } 224 error = unp_attach(so); 225 break; 226 227 case PRU_DETACH: 228 unp_detach(unp); 229 break; 230 231 case PRU_BIND: 232 error = unp_bind(unp, nam, l); 233 break; 234 235 case PRU_LISTEN: 236 if (unp->unp_vnode == 0) 237 error = EINVAL; 238 break; 239 240 case PRU_CONNECT: 241 error = unp_connect(so, nam, l); 242 break; 243 244 case PRU_CONNECT2: 245 error = unp_connect2(so, (struct socket *)nam, PRU_CONNECT2); 246 break; 247 248 case PRU_DISCONNECT: 249 unp_disconnect(unp); 250 break; 251 252 case PRU_ACCEPT: 253 unp_setpeeraddr(unp, nam); 254 /* 255 * Mark the initiating STREAM socket as connected *ONLY* 256 * after it's been accepted. This prevents a client from 257 * overrunning a server and receiving ECONNREFUSED. 258 */ 259 if (unp->unp_conn != NULL && 260 (unp->unp_conn->unp_socket->so_state & SS_ISCONNECTING)) 261 soisconnected(unp->unp_conn->unp_socket); 262 break; 263 264 case PRU_SHUTDOWN: 265 socantsendmore(so); 266 unp_shutdown(unp); 267 break; 268 269 case PRU_RCVD: 270 switch (so->so_type) { 271 272 case SOCK_DGRAM: 273 panic("uipc 1"); 274 /*NOTREACHED*/ 275 276 case SOCK_STREAM: 277 #define rcv (&so->so_rcv) 278 #define snd (&so2->so_snd) 279 if (unp->unp_conn == 0) 280 break; 281 so2 = unp->unp_conn->unp_socket; 282 /* 283 * Adjust backpressure on sender 284 * and wakeup any waiting to write. 285 */ 286 snd->sb_mbmax += unp->unp_mbcnt - rcv->sb_mbcnt; 287 unp->unp_mbcnt = rcv->sb_mbcnt; 288 newhiwat = snd->sb_hiwat + unp->unp_cc - rcv->sb_cc; 289 (void)chgsbsize(so2->so_uidinfo, 290 &snd->sb_hiwat, newhiwat, RLIM_INFINITY); 291 unp->unp_cc = rcv->sb_cc; 292 sowwakeup(so2); 293 #undef snd 294 #undef rcv 295 break; 296 297 default: 298 panic("uipc 2"); 299 } 300 break; 301 302 case PRU_SEND: 303 /* 304 * Note: unp_internalize() rejects any control message 305 * other than SCM_RIGHTS, and only allows one. This 306 * has the side-effect of preventing a caller from 307 * forging SCM_CREDS. 308 */ 309 if (control && (error = unp_internalize(control, l))) { 310 goto die; 311 } 312 switch (so->so_type) { 313 314 case SOCK_DGRAM: { 315 if (nam) { 316 if ((so->so_state & SS_ISCONNECTED) != 0) { 317 error = EISCONN; 318 goto die; 319 } 320 error = unp_connect(so, nam, l); 321 if (error) { 322 die: 323 m_freem(control); 324 m_freem(m); 325 break; 326 } 327 } else { 328 if ((so->so_state & SS_ISCONNECTED) == 0) { 329 error = ENOTCONN; 330 goto die; 331 } 332 } 333 error = unp_output(m, control, unp, p); 334 if (nam) 335 unp_disconnect(unp); 336 break; 337 } 338 339 case SOCK_STREAM: 340 #define rcv (&so2->so_rcv) 341 #define snd (&so->so_snd) 342 if (unp->unp_conn == NULL) { 343 error = ENOTCONN; 344 break; 345 } 346 so2 = unp->unp_conn->unp_socket; 347 if (unp->unp_conn->unp_flags & UNP_WANTCRED) { 348 /* 349 * Credentials are passed only once on 350 * SOCK_STREAM. 351 */ 352 unp->unp_conn->unp_flags &= ~UNP_WANTCRED; 353 control = unp_addsockcred(p, control); 354 } 355 /* 356 * Send to paired receive port, and then reduce 357 * send buffer hiwater marks to maintain backpressure. 358 * Wake up readers. 359 */ 360 if (control) { 361 if (sbappendcontrol(rcv, m, control) == 0) 362 m_freem(control); 363 } else 364 sbappend(rcv, m); 365 snd->sb_mbmax -= 366 rcv->sb_mbcnt - unp->unp_conn->unp_mbcnt; 367 unp->unp_conn->unp_mbcnt = rcv->sb_mbcnt; 368 newhiwat = snd->sb_hiwat - 369 (rcv->sb_cc - unp->unp_conn->unp_cc); 370 (void)chgsbsize(so->so_uidinfo, 371 &snd->sb_hiwat, newhiwat, RLIM_INFINITY); 372 unp->unp_conn->unp_cc = rcv->sb_cc; 373 sorwakeup(so2); 374 #undef snd 375 #undef rcv 376 break; 377 378 default: 379 panic("uipc 4"); 380 } 381 break; 382 383 case PRU_ABORT: 384 unp_drop(unp, ECONNABORTED); 385 386 #ifdef DIAGNOSTIC 387 if (so->so_pcb == 0) 388 panic("uipc 5: drop killed pcb"); 389 #endif 390 unp_detach(unp); 391 break; 392 393 case PRU_SENSE: 394 ((struct stat *) m)->st_blksize = so->so_snd.sb_hiwat; 395 if (so->so_type == SOCK_STREAM && unp->unp_conn != 0) { 396 so2 = unp->unp_conn->unp_socket; 397 ((struct stat *) m)->st_blksize += so2->so_rcv.sb_cc; 398 } 399 ((struct stat *) m)->st_dev = NODEV; 400 if (unp->unp_ino == 0) 401 unp->unp_ino = unp_ino++; 402 ((struct stat *) m)->st_atimespec = 403 ((struct stat *) m)->st_mtimespec = 404 ((struct stat *) m)->st_ctimespec = unp->unp_ctime; 405 ((struct stat *) m)->st_ino = unp->unp_ino; 406 return (0); 407 408 case PRU_RCVOOB: 409 error = EOPNOTSUPP; 410 break; 411 412 case PRU_SENDOOB: 413 m_freem(control); 414 m_freem(m); 415 error = EOPNOTSUPP; 416 break; 417 418 case PRU_SOCKADDR: 419 unp_setsockaddr(unp, nam); 420 break; 421 422 case PRU_PEERADDR: 423 unp_setpeeraddr(unp, nam); 424 break; 425 426 default: 427 panic("piusrreq"); 428 } 429 430 release: 431 return (error); 432 } 433 434 /* 435 * Unix domain socket option processing. 436 */ 437 int 438 uipc_ctloutput(int op, struct socket *so, int level, int optname, 439 struct mbuf **mp) 440 { 441 struct unpcb *unp = sotounpcb(so); 442 struct mbuf *m = *mp; 443 int optval = 0, error = 0; 444 445 if (level != 0) { 446 error = EINVAL; 447 if (op == PRCO_SETOPT && m) 448 (void) m_free(m); 449 } else switch (op) { 450 451 case PRCO_SETOPT: 452 switch (optname) { 453 case LOCAL_CREDS: 454 case LOCAL_CONNWAIT: 455 if (m == NULL || m->m_len != sizeof(int)) 456 error = EINVAL; 457 else { 458 optval = *mtod(m, int *); 459 switch (optname) { 460 #define OPTSET(bit) \ 461 if (optval) \ 462 unp->unp_flags |= (bit); \ 463 else \ 464 unp->unp_flags &= ~(bit); 465 466 case LOCAL_CREDS: 467 OPTSET(UNP_WANTCRED); 468 break; 469 case LOCAL_CONNWAIT: 470 OPTSET(UNP_CONNWAIT); 471 break; 472 } 473 } 474 break; 475 #undef OPTSET 476 477 default: 478 error = ENOPROTOOPT; 479 break; 480 } 481 if (m) 482 (void) m_free(m); 483 break; 484 485 case PRCO_GETOPT: 486 switch (optname) { 487 case LOCAL_CREDS: 488 *mp = m = m_get(M_WAIT, MT_SOOPTS); 489 m->m_len = sizeof(int); 490 switch (optname) { 491 492 #define OPTBIT(bit) (unp->unp_flags & (bit) ? 1 : 0) 493 494 case LOCAL_CREDS: 495 optval = OPTBIT(UNP_WANTCRED); 496 break; 497 } 498 *mtod(m, int *) = optval; 499 break; 500 #undef OPTBIT 501 502 default: 503 error = ENOPROTOOPT; 504 break; 505 } 506 break; 507 } 508 return (error); 509 } 510 511 /* 512 * Both send and receive buffers are allocated PIPSIZ bytes of buffering 513 * for stream sockets, although the total for sender and receiver is 514 * actually only PIPSIZ. 515 * Datagram sockets really use the sendspace as the maximum datagram size, 516 * and don't really want to reserve the sendspace. Their recvspace should 517 * be large enough for at least one max-size datagram plus address. 518 */ 519 #define PIPSIZ 4096 520 u_long unpst_sendspace = PIPSIZ; 521 u_long unpst_recvspace = PIPSIZ; 522 u_long unpdg_sendspace = 2*1024; /* really max datagram size */ 523 u_long unpdg_recvspace = 4*1024; 524 525 int unp_rights; /* file descriptors in flight */ 526 527 int 528 unp_attach(struct socket *so) 529 { 530 struct unpcb *unp; 531 int error; 532 533 if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) { 534 switch (so->so_type) { 535 536 case SOCK_STREAM: 537 error = soreserve(so, unpst_sendspace, unpst_recvspace); 538 break; 539 540 case SOCK_DGRAM: 541 error = soreserve(so, unpdg_sendspace, unpdg_recvspace); 542 break; 543 544 default: 545 panic("unp_attach"); 546 } 547 if (error) 548 return (error); 549 } 550 unp = malloc(sizeof(*unp), M_PCB, M_NOWAIT); 551 if (unp == NULL) 552 return (ENOBUFS); 553 memset((caddr_t)unp, 0, sizeof(*unp)); 554 unp->unp_socket = so; 555 so->so_pcb = unp; 556 nanotime(&unp->unp_ctime); 557 return (0); 558 } 559 560 void 561 unp_detach(struct unpcb *unp) 562 { 563 564 if (unp->unp_vnode) { 565 unp->unp_vnode->v_socket = 0; 566 vrele(unp->unp_vnode); 567 unp->unp_vnode = 0; 568 } 569 if (unp->unp_conn) 570 unp_disconnect(unp); 571 while (unp->unp_refs) 572 unp_drop(unp->unp_refs, ECONNRESET); 573 soisdisconnected(unp->unp_socket); 574 unp->unp_socket->so_pcb = 0; 575 if (unp->unp_addr) 576 free(unp->unp_addr, M_SONAME); 577 if (unp_rights) { 578 /* 579 * Normally the receive buffer is flushed later, 580 * in sofree, but if our receive buffer holds references 581 * to descriptors that are now garbage, we will dispose 582 * of those descriptor references after the garbage collector 583 * gets them (resulting in a "panic: closef: count < 0"). 584 */ 585 sorflush(unp->unp_socket); 586 free(unp, M_PCB); 587 unp_gc(); 588 } else 589 free(unp, M_PCB); 590 } 591 592 int 593 unp_bind(struct unpcb *unp, struct mbuf *nam, struct lwp *l) 594 { 595 struct sockaddr_un *sun; 596 struct vnode *vp; 597 struct mount *mp; 598 struct vattr vattr; 599 size_t addrlen; 600 struct proc *p; 601 int error; 602 struct nameidata nd; 603 604 if (unp->unp_vnode != 0) 605 return (EINVAL); 606 607 p = l->l_proc; 608 /* 609 * Allocate the new sockaddr. We have to allocate one 610 * extra byte so that we can ensure that the pathname 611 * is nul-terminated. 612 */ 613 addrlen = nam->m_len + 1; 614 sun = malloc(addrlen, M_SONAME, M_WAITOK); 615 m_copydata(nam, 0, nam->m_len, (caddr_t)sun); 616 *(((char *)sun) + nam->m_len) = '\0'; 617 618 restart: 619 NDINIT(&nd, CREATE, FOLLOW | LOCKPARENT, UIO_SYSSPACE, 620 sun->sun_path, l); 621 622 /* SHOULD BE ABLE TO ADOPT EXISTING AND wakeup() ALA FIFO's */ 623 if ((error = namei(&nd)) != 0) 624 goto bad; 625 vp = nd.ni_vp; 626 if (vp != NULL || vn_start_write(nd.ni_dvp, &mp, V_NOWAIT) != 0) { 627 VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd); 628 if (nd.ni_dvp == vp) 629 vrele(nd.ni_dvp); 630 else 631 vput(nd.ni_dvp); 632 vrele(vp); 633 if (vp != NULL) { 634 error = EADDRINUSE; 635 goto bad; 636 } 637 error = vn_start_write(NULL, &mp, 638 V_WAIT | V_SLEEPONLY | V_PCATCH); 639 if (error) 640 goto bad; 641 goto restart; 642 } 643 VATTR_NULL(&vattr); 644 vattr.va_type = VSOCK; 645 vattr.va_mode = ACCESSPERMS & ~(p->p_cwdi->cwdi_cmask); 646 VOP_LEASE(nd.ni_dvp, l, p->p_ucred, LEASE_WRITE); 647 error = VOP_CREATE(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr); 648 vn_finished_write(mp, 0); 649 if (error) 650 goto bad; 651 vp = nd.ni_vp; 652 vp->v_socket = unp->unp_socket; 653 unp->unp_vnode = vp; 654 unp->unp_addrlen = addrlen; 655 unp->unp_addr = sun; 656 VOP_UNLOCK(vp, 0); 657 return (0); 658 659 bad: 660 free(sun, M_SONAME); 661 return (error); 662 } 663 664 int 665 unp_connect(struct socket *so, struct mbuf *nam, struct lwp *l) 666 { 667 struct sockaddr_un *sun; 668 struct vnode *vp; 669 struct socket *so2, *so3; 670 struct unpcb *unp2, *unp3; 671 size_t addrlen; 672 int error; 673 struct nameidata nd; 674 675 /* 676 * Allocate a temporary sockaddr. We have to allocate one extra 677 * byte so that we can ensure that the pathname is nul-terminated. 678 * When we establish the connection, we copy the other PCB's 679 * sockaddr to our own. 680 */ 681 addrlen = nam->m_len + 1; 682 sun = malloc(addrlen, M_SONAME, M_WAITOK); 683 m_copydata(nam, 0, nam->m_len, (caddr_t)sun); 684 *(((char *)sun) + nam->m_len) = '\0'; 685 686 NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_SYSSPACE, sun->sun_path, l); 687 688 if ((error = namei(&nd)) != 0) 689 goto bad2; 690 vp = nd.ni_vp; 691 if (vp->v_type != VSOCK) { 692 error = ENOTSOCK; 693 goto bad; 694 } 695 if ((error = VOP_ACCESS(vp, VWRITE, l->l_proc->p_ucred, l)) != 0) 696 goto bad; 697 so2 = vp->v_socket; 698 if (so2 == 0) { 699 error = ECONNREFUSED; 700 goto bad; 701 } 702 if (so->so_type != so2->so_type) { 703 error = EPROTOTYPE; 704 goto bad; 705 } 706 if (so->so_proto->pr_flags & PR_CONNREQUIRED) { 707 if ((so2->so_options & SO_ACCEPTCONN) == 0 || 708 (so3 = sonewconn(so2, 0)) == 0) { 709 error = ECONNREFUSED; 710 goto bad; 711 } 712 unp2 = sotounpcb(so2); 713 unp3 = sotounpcb(so3); 714 if (unp2->unp_addr) { 715 unp3->unp_addr = malloc(unp2->unp_addrlen, 716 M_SONAME, M_WAITOK); 717 memcpy(unp3->unp_addr, unp2->unp_addr, 718 unp2->unp_addrlen); 719 unp3->unp_addrlen = unp2->unp_addrlen; 720 } 721 unp3->unp_flags = unp2->unp_flags; 722 so2 = so3; 723 } 724 error = unp_connect2(so, so2, PRU_CONNECT); 725 bad: 726 vput(vp); 727 bad2: 728 free(sun, M_SONAME); 729 return (error); 730 } 731 732 int 733 unp_connect2(struct socket *so, struct socket *so2, int req) 734 { 735 struct unpcb *unp = sotounpcb(so); 736 struct unpcb *unp2; 737 738 if (so2->so_type != so->so_type) 739 return (EPROTOTYPE); 740 unp2 = sotounpcb(so2); 741 unp->unp_conn = unp2; 742 switch (so->so_type) { 743 744 case SOCK_DGRAM: 745 unp->unp_nextref = unp2->unp_refs; 746 unp2->unp_refs = unp; 747 soisconnected(so); 748 break; 749 750 case SOCK_STREAM: 751 unp2->unp_conn = unp; 752 if (req == PRU_CONNECT && 753 ((unp->unp_flags | unp2->unp_flags) & UNP_CONNWAIT)) 754 soisconnecting(so); 755 else 756 soisconnected(so); 757 soisconnected(so2); 758 break; 759 760 default: 761 panic("unp_connect2"); 762 } 763 return (0); 764 } 765 766 void 767 unp_disconnect(struct unpcb *unp) 768 { 769 struct unpcb *unp2 = unp->unp_conn; 770 771 if (unp2 == 0) 772 return; 773 unp->unp_conn = 0; 774 switch (unp->unp_socket->so_type) { 775 776 case SOCK_DGRAM: 777 if (unp2->unp_refs == unp) 778 unp2->unp_refs = unp->unp_nextref; 779 else { 780 unp2 = unp2->unp_refs; 781 for (;;) { 782 if (unp2 == 0) 783 panic("unp_disconnect"); 784 if (unp2->unp_nextref == unp) 785 break; 786 unp2 = unp2->unp_nextref; 787 } 788 unp2->unp_nextref = unp->unp_nextref; 789 } 790 unp->unp_nextref = 0; 791 unp->unp_socket->so_state &= ~SS_ISCONNECTED; 792 break; 793 794 case SOCK_STREAM: 795 soisdisconnected(unp->unp_socket); 796 unp2->unp_conn = 0; 797 soisdisconnected(unp2->unp_socket); 798 break; 799 } 800 } 801 802 #ifdef notdef 803 unp_abort(struct unpcb *unp) 804 { 805 unp_detach(unp); 806 } 807 #endif 808 809 void 810 unp_shutdown(struct unpcb *unp) 811 { 812 struct socket *so; 813 814 if (unp->unp_socket->so_type == SOCK_STREAM && unp->unp_conn && 815 (so = unp->unp_conn->unp_socket)) 816 socantrcvmore(so); 817 } 818 819 void 820 unp_drop(struct unpcb *unp, int errno) 821 { 822 struct socket *so = unp->unp_socket; 823 824 so->so_error = errno; 825 unp_disconnect(unp); 826 if (so->so_head) { 827 so->so_pcb = 0; 828 sofree(so); 829 if (unp->unp_addr) 830 free(unp->unp_addr, M_SONAME); 831 free(unp, M_PCB); 832 } 833 } 834 835 #ifdef notdef 836 unp_drain(void) 837 { 838 839 } 840 #endif 841 842 int 843 unp_externalize(struct mbuf *rights, struct lwp *l) 844 { 845 struct cmsghdr *cm = mtod(rights, struct cmsghdr *); 846 struct proc *p = l->l_proc; 847 int i, *fdp; 848 struct file **rp; 849 struct file *fp; 850 int nfds, error = 0; 851 852 nfds = (cm->cmsg_len - CMSG_ALIGN(sizeof(*cm))) / 853 sizeof(struct file *); 854 rp = (struct file **)CMSG_DATA(cm); 855 856 fdp = malloc(nfds * sizeof(int), M_TEMP, M_WAITOK); 857 858 /* Make sure the recipient should be able to see the descriptors.. */ 859 if (p->p_cwdi->cwdi_rdir != NULL) { 860 rp = (struct file **)CMSG_DATA(cm); 861 for (i = 0; i < nfds; i++) { 862 fp = *rp++; 863 /* 864 * If we are in a chroot'ed directory, and 865 * someone wants to pass us a directory, make 866 * sure it's inside the subtree we're allowed 867 * to access. 868 */ 869 if (fp->f_type == DTYPE_VNODE) { 870 struct vnode *vp = (struct vnode *)fp->f_data; 871 if ((vp->v_type == VDIR) && 872 !vn_isunder(vp, p->p_cwdi->cwdi_rdir, l)) { 873 error = EPERM; 874 break; 875 } 876 } 877 } 878 } 879 880 restart: 881 rp = (struct file **)CMSG_DATA(cm); 882 if (error != 0) { 883 for (i = 0; i < nfds; i++) { 884 fp = *rp; 885 /* 886 * zero the pointer before calling unp_discard, 887 * since it may end up in unp_gc().. 888 */ 889 *rp++ = 0; 890 unp_discard(fp); 891 } 892 goto out; 893 } 894 895 /* 896 * First loop -- allocate file descriptor table slots for the 897 * new descriptors. 898 */ 899 for (i = 0; i < nfds; i++) { 900 fp = *rp++; 901 if ((error = fdalloc(p, 0, &fdp[i])) != 0) { 902 /* 903 * Back out what we've done so far. 904 */ 905 for (--i; i >= 0; i--) 906 fdremove(p->p_fd, fdp[i]); 907 908 if (error == ENOSPC) { 909 fdexpand(p); 910 error = 0; 911 } else { 912 /* 913 * This is the error that has historically 914 * been returned, and some callers may 915 * expect it. 916 */ 917 error = EMSGSIZE; 918 } 919 goto restart; 920 } 921 922 /* 923 * Make the slot reference the descriptor so that 924 * fdalloc() works properly.. We finalize it all 925 * in the loop below. 926 */ 927 p->p_fd->fd_ofiles[fdp[i]] = fp; 928 } 929 930 /* 931 * Now that adding them has succeeded, update all of the 932 * descriptor passing state. 933 */ 934 rp = (struct file **)CMSG_DATA(cm); 935 for (i = 0; i < nfds; i++) { 936 fp = *rp++; 937 fp->f_msgcount--; 938 unp_rights--; 939 } 940 941 /* 942 * Copy temporary array to message and adjust length, in case of 943 * transition from large struct file pointers to ints. 944 */ 945 memcpy(CMSG_DATA(cm), fdp, nfds * sizeof(int)); 946 cm->cmsg_len = CMSG_LEN(nfds * sizeof(int)); 947 rights->m_len = CMSG_SPACE(nfds * sizeof(int)); 948 out: 949 free(fdp, M_TEMP); 950 return (error); 951 } 952 953 int 954 unp_internalize(struct mbuf *control, struct lwp *l) 955 { 956 struct proc *p = l->l_proc; 957 struct filedesc *fdescp = p->p_fd; 958 struct cmsghdr *newcm, *cm = mtod(control, struct cmsghdr *); 959 struct file **rp, **files; 960 struct file *fp; 961 int i, fd, *fdp; 962 int nfds; 963 u_int neededspace; 964 965 /* Sanity check the control message header */ 966 if (cm->cmsg_type != SCM_RIGHTS || cm->cmsg_level != SOL_SOCKET || 967 cm->cmsg_len != control->m_len) 968 return (EINVAL); 969 970 /* Verify that the file descriptors are valid */ 971 nfds = (cm->cmsg_len - CMSG_ALIGN(sizeof(*cm))) / sizeof(int); 972 fdp = (int *)CMSG_DATA(cm); 973 for (i = 0; i < nfds; i++) { 974 fd = *fdp++; 975 if ((fp = fd_getfile(fdescp, fd)) == NULL) 976 return (EBADF); 977 simple_unlock(&fp->f_slock); 978 } 979 980 /* Make sure we have room for the struct file pointers */ 981 neededspace = CMSG_SPACE(nfds * sizeof(struct file *)) - 982 control->m_len; 983 if (neededspace > M_TRAILINGSPACE(control)) { 984 985 /* allocate new space and copy header into it */ 986 newcm = malloc( 987 CMSG_SPACE(nfds * sizeof(struct file *)), 988 M_MBUF, M_WAITOK); 989 if (newcm == NULL) 990 return (E2BIG); 991 memcpy(newcm, cm, sizeof(struct cmsghdr)); 992 files = (struct file **)CMSG_DATA(newcm); 993 } else { 994 /* we can convert in-place */ 995 newcm = NULL; 996 files = (struct file **)CMSG_DATA(cm); 997 } 998 999 /* 1000 * Transform the file descriptors into struct file pointers, in 1001 * reverse order so that if pointers are bigger than ints, the 1002 * int won't get until we're done. 1003 */ 1004 fdp = (int *)CMSG_DATA(cm) + nfds - 1; 1005 rp = files + nfds - 1; 1006 for (i = 0; i < nfds; i++) { 1007 fp = fdescp->fd_ofiles[*fdp--]; 1008 simple_lock(&fp->f_slock); 1009 #ifdef DIAGNOSTIC 1010 if (fp->f_iflags & FIF_WANTCLOSE) 1011 panic("unp_internalize: file already closed"); 1012 #endif 1013 *rp-- = fp; 1014 fp->f_count++; 1015 fp->f_msgcount++; 1016 simple_unlock(&fp->f_slock); 1017 unp_rights++; 1018 } 1019 1020 if (newcm) { 1021 if (control->m_flags & M_EXT) 1022 MEXTREMOVE(control); 1023 MEXTADD(control, newcm, 1024 CMSG_SPACE(nfds * sizeof(struct file *)), 1025 M_MBUF, NULL, NULL); 1026 cm = newcm; 1027 } 1028 1029 /* adjust message & mbuf to note amount of space actually used. */ 1030 cm->cmsg_len = CMSG_LEN(nfds * sizeof(struct file *)); 1031 control->m_len = CMSG_SPACE(nfds * sizeof(struct file *)); 1032 1033 return (0); 1034 } 1035 1036 struct mbuf * 1037 unp_addsockcred(struct proc *p, struct mbuf *control) 1038 { 1039 struct cmsghdr *cmp; 1040 struct sockcred *sc; 1041 struct mbuf *m, *n; 1042 int len, space, i; 1043 1044 len = CMSG_LEN(SOCKCREDSIZE(p->p_ucred->cr_ngroups)); 1045 space = CMSG_SPACE(SOCKCREDSIZE(p->p_ucred->cr_ngroups)); 1046 1047 m = m_get(M_WAIT, MT_CONTROL); 1048 if (space > MLEN) { 1049 if (space > MCLBYTES) 1050 MEXTMALLOC(m, space, M_WAITOK); 1051 else 1052 m_clget(m, M_WAIT); 1053 if ((m->m_flags & M_EXT) == 0) { 1054 m_free(m); 1055 return (control); 1056 } 1057 } 1058 1059 m->m_len = space; 1060 m->m_next = NULL; 1061 cmp = mtod(m, struct cmsghdr *); 1062 sc = (struct sockcred *)CMSG_DATA(cmp); 1063 cmp->cmsg_len = len; 1064 cmp->cmsg_level = SOL_SOCKET; 1065 cmp->cmsg_type = SCM_CREDS; 1066 sc->sc_uid = p->p_cred->p_ruid; 1067 sc->sc_euid = p->p_ucred->cr_uid; 1068 sc->sc_gid = p->p_cred->p_rgid; 1069 sc->sc_egid = p->p_ucred->cr_gid; 1070 sc->sc_ngroups = p->p_ucred->cr_ngroups; 1071 for (i = 0; i < sc->sc_ngroups; i++) 1072 sc->sc_groups[i] = p->p_ucred->cr_groups[i]; 1073 1074 /* 1075 * If a control message already exists, append us to the end. 1076 */ 1077 if (control != NULL) { 1078 for (n = control; n->m_next != NULL; n = n->m_next) 1079 ; 1080 n->m_next = m; 1081 } else 1082 control = m; 1083 1084 return (control); 1085 } 1086 1087 int unp_defer, unp_gcing; 1088 extern struct domain unixdomain; 1089 1090 /* 1091 * Comment added long after the fact explaining what's going on here. 1092 * Do a mark-sweep GC of file descriptors on the system, to free up 1093 * any which are caught in flight to an about-to-be-closed socket. 1094 * 1095 * Traditional mark-sweep gc's start at the "root", and mark 1096 * everything reachable from the root (which, in our case would be the 1097 * process table). The mark bits are cleared during the sweep. 1098 * 1099 * XXX For some inexplicable reason (perhaps because the file 1100 * descriptor tables used to live in the u area which could be swapped 1101 * out and thus hard to reach), we do multiple scans over the set of 1102 * descriptors, using use *two* mark bits per object (DEFER and MARK). 1103 * Whenever we find a descriptor which references other descriptors, 1104 * the ones it references are marked with both bits, and we iterate 1105 * over the whole file table until there are no more DEFER bits set. 1106 * We also make an extra pass *before* the GC to clear the mark bits, 1107 * which could have been cleared at almost no cost during the previous 1108 * sweep. 1109 * 1110 * XXX MP: this needs to run with locks such that no other thread of 1111 * control can create or destroy references to file descriptors. it 1112 * may be necessary to defer the GC until later (when the locking 1113 * situation is more hospitable); it may be necessary to push this 1114 * into a separate thread. 1115 */ 1116 void 1117 unp_gc(void) 1118 { 1119 struct file *fp, *nextfp; 1120 struct socket *so, *so1; 1121 struct file **extra_ref, **fpp; 1122 int nunref, i; 1123 1124 if (unp_gcing) 1125 return; 1126 unp_gcing = 1; 1127 unp_defer = 0; 1128 1129 /* Clear mark bits */ 1130 LIST_FOREACH(fp, &filehead, f_list) 1131 fp->f_flag &= ~(FMARK|FDEFER); 1132 1133 /* 1134 * Iterate over the set of descriptors, marking ones believed 1135 * (based on refcount) to be referenced from a process, and 1136 * marking for rescan descriptors which are queued on a socket. 1137 */ 1138 do { 1139 LIST_FOREACH(fp, &filehead, f_list) { 1140 if (fp->f_flag & FDEFER) { 1141 fp->f_flag &= ~FDEFER; 1142 unp_defer--; 1143 #ifdef DIAGNOSTIC 1144 if (fp->f_count == 0) 1145 panic("unp_gc: deferred unreferenced socket"); 1146 #endif 1147 } else { 1148 if (fp->f_count == 0) 1149 continue; 1150 if (fp->f_flag & FMARK) 1151 continue; 1152 if (fp->f_count == fp->f_msgcount) 1153 continue; 1154 } 1155 fp->f_flag |= FMARK; 1156 1157 if (fp->f_type != DTYPE_SOCKET || 1158 (so = (struct socket *)fp->f_data) == 0) 1159 continue; 1160 if (so->so_proto->pr_domain != &unixdomain || 1161 (so->so_proto->pr_flags&PR_RIGHTS) == 0) 1162 continue; 1163 #ifdef notdef 1164 if (so->so_rcv.sb_flags & SB_LOCK) { 1165 /* 1166 * This is problematical; it's not clear 1167 * we need to wait for the sockbuf to be 1168 * unlocked (on a uniprocessor, at least), 1169 * and it's also not clear what to do 1170 * if sbwait returns an error due to receipt 1171 * of a signal. If sbwait does return 1172 * an error, we'll go into an infinite 1173 * loop. Delete all of this for now. 1174 */ 1175 (void) sbwait(&so->so_rcv); 1176 goto restart; 1177 } 1178 #endif 1179 unp_scan(so->so_rcv.sb_mb, unp_mark, 0); 1180 /* 1181 * mark descriptors referenced from sockets queued on the accept queue as well. 1182 */ 1183 if (so->so_options & SO_ACCEPTCONN) { 1184 TAILQ_FOREACH(so1, &so->so_q0, so_qe) { 1185 unp_scan(so1->so_rcv.sb_mb, unp_mark, 0); 1186 } 1187 TAILQ_FOREACH(so1, &so->so_q, so_qe) { 1188 unp_scan(so1->so_rcv.sb_mb, unp_mark, 0); 1189 } 1190 } 1191 1192 } 1193 } while (unp_defer); 1194 /* 1195 * Sweep pass. Find unmarked descriptors, and free them. 1196 * 1197 * We grab an extra reference to each of the file table entries 1198 * that are not otherwise accessible and then free the rights 1199 * that are stored in messages on them. 1200 * 1201 * The bug in the original code is a little tricky, so I'll describe 1202 * what's wrong with it here. 1203 * 1204 * It is incorrect to simply unp_discard each entry for f_msgcount 1205 * times -- consider the case of sockets A and B that contain 1206 * references to each other. On a last close of some other socket, 1207 * we trigger a gc since the number of outstanding rights (unp_rights) 1208 * is non-zero. If during the sweep phase the gc code un_discards, 1209 * we end up doing a (full) closef on the descriptor. A closef on A 1210 * results in the following chain. Closef calls soo_close, which 1211 * calls soclose. Soclose calls first (through the switch 1212 * uipc_usrreq) unp_detach, which re-invokes unp_gc. Unp_gc simply 1213 * returns because the previous instance had set unp_gcing, and 1214 * we return all the way back to soclose, which marks the socket 1215 * with SS_NOFDREF, and then calls sofree. Sofree calls sorflush 1216 * to free up the rights that are queued in messages on the socket A, 1217 * i.e., the reference on B. The sorflush calls via the dom_dispose 1218 * switch unp_dispose, which unp_scans with unp_discard. This second 1219 * instance of unp_discard just calls closef on B. 1220 * 1221 * Well, a similar chain occurs on B, resulting in a sorflush on B, 1222 * which results in another closef on A. Unfortunately, A is already 1223 * being closed, and the descriptor has already been marked with 1224 * SS_NOFDREF, and soclose panics at this point. 1225 * 1226 * Here, we first take an extra reference to each inaccessible 1227 * descriptor. Then, if the inaccessible descriptor is a 1228 * socket, we call sorflush in case it is a Unix domain 1229 * socket. After we destroy all the rights carried in 1230 * messages, we do a last closef to get rid of our extra 1231 * reference. This is the last close, and the unp_detach etc 1232 * will shut down the socket. 1233 * 1234 * 91/09/19, bsy (at) cs.cmu.edu 1235 */ 1236 extra_ref = malloc(nfiles * sizeof(struct file *), M_FILE, M_WAITOK); 1237 for (nunref = 0, fp = LIST_FIRST(&filehead), fpp = extra_ref; fp != 0; 1238 fp = nextfp) { 1239 nextfp = LIST_NEXT(fp, f_list); 1240 simple_lock(&fp->f_slock); 1241 if (fp->f_count != 0 && 1242 fp->f_count == fp->f_msgcount && !(fp->f_flag & FMARK)) { 1243 *fpp++ = fp; 1244 nunref++; 1245 fp->f_count++; 1246 } 1247 simple_unlock(&fp->f_slock); 1248 } 1249 for (i = nunref, fpp = extra_ref; --i >= 0; ++fpp) { 1250 fp = *fpp; 1251 simple_lock(&fp->f_slock); 1252 FILE_USE(fp); 1253 if (fp->f_type == DTYPE_SOCKET) 1254 sorflush((struct socket *)fp->f_data); 1255 FILE_UNUSE(fp, NULL); 1256 } 1257 for (i = nunref, fpp = extra_ref; --i >= 0; ++fpp) { 1258 fp = *fpp; 1259 simple_lock(&fp->f_slock); 1260 FILE_USE(fp); 1261 (void) closef(fp, (struct lwp *)0); 1262 } 1263 free((caddr_t)extra_ref, M_FILE); 1264 unp_gcing = 0; 1265 } 1266 1267 void 1268 unp_dispose(struct mbuf *m) 1269 { 1270 1271 if (m) 1272 unp_scan(m, unp_discard, 1); 1273 } 1274 1275 void 1276 unp_scan(struct mbuf *m0, void (*op)(struct file *), int discard) 1277 { 1278 struct mbuf *m; 1279 struct file **rp; 1280 struct cmsghdr *cm; 1281 int i; 1282 int qfds; 1283 1284 while (m0) { 1285 for (m = m0; m; m = m->m_next) { 1286 if (m->m_type == MT_CONTROL && 1287 m->m_len >= sizeof(*cm)) { 1288 cm = mtod(m, struct cmsghdr *); 1289 if (cm->cmsg_level != SOL_SOCKET || 1290 cm->cmsg_type != SCM_RIGHTS) 1291 continue; 1292 qfds = (cm->cmsg_len - CMSG_ALIGN(sizeof(*cm))) 1293 / sizeof(struct file *); 1294 rp = (struct file **)CMSG_DATA(cm); 1295 for (i = 0; i < qfds; i++) { 1296 struct file *fp = *rp; 1297 if (discard) 1298 *rp = 0; 1299 (*op)(fp); 1300 rp++; 1301 } 1302 break; /* XXX, but saves time */ 1303 } 1304 } 1305 m0 = m0->m_nextpkt; 1306 } 1307 } 1308 1309 void 1310 unp_mark(struct file *fp) 1311 { 1312 if (fp == NULL) 1313 return; 1314 1315 if (fp->f_flag & FMARK) 1316 return; 1317 1318 /* If we're already deferred, don't screw up the defer count */ 1319 if (fp->f_flag & FDEFER) 1320 return; 1321 1322 /* 1323 * Minimize the number of deferrals... Sockets are the only 1324 * type of descriptor which can hold references to another 1325 * descriptor, so just mark other descriptors, and defer 1326 * unmarked sockets for the next pass. 1327 */ 1328 if (fp->f_type == DTYPE_SOCKET) { 1329 unp_defer++; 1330 if (fp->f_count == 0) 1331 panic("unp_mark: queued unref"); 1332 fp->f_flag |= FDEFER; 1333 } else { 1334 fp->f_flag |= FMARK; 1335 } 1336 return; 1337 } 1338 1339 void 1340 unp_discard(struct file *fp) 1341 { 1342 if (fp == NULL) 1343 return; 1344 simple_lock(&fp->f_slock); 1345 fp->f_usecount++; /* i.e. FILE_USE(fp) sans locking */ 1346 fp->f_msgcount--; 1347 simple_unlock(&fp->f_slock); 1348 unp_rights--; 1349 (void) closef(fp, (struct lwp *)0); 1350 } 1351
Indexes created Sun Sep 21 18:09:52 GMT 2025