| 1 | /* |
| 2 | * Copyright (c) 1982, 1986, 1988, 1990 Regents of the University of California. |
| 3 | * All rights reserved. |
| 4 | * |
| 5 | * Redistribution and use in source and binary forms, with or without |
| 6 | * modification, are permitted provided that the following conditions |
| 7 | * are met: |
| 8 | * 1. Redistributions of source code must retain the above copyright |
| 9 | * notice, this list of conditions and the following disclaimer. |
| 10 | * 2. Redistributions in binary form must reproduce the above copyright |
| 11 | * notice, this list of conditions and the following disclaimer in the |
| 12 | * documentation and/or other materials provided with the distribution. |
| 13 | * 3. All advertising materials mentioning features or use of this software |
| 14 | * must display the following acknowledgement: |
| 15 | * This product includes software developed by the University of |
| 16 | * California, Berkeley and its contributors. |
| 17 | * 4. Neither the name of the University nor the names of its contributors |
| 18 | * may be used to endorse or promote products derived from this software |
| 19 | * without specific prior written permission. |
| 20 | * |
| 21 | * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND |
| 22 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
| 23 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
| 24 | * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE |
| 25 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL |
| 26 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS |
| 27 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) |
| 28 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT |
| 29 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY |
| 30 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF |
| 31 | * SUCH DAMAGE. |
| 32 | * |
| 33 | * @(#)uipc_socket.c 7.28 (Berkeley) 5/4/91 |
| 34 | */ |
| 35 | |
| 36 | #include "param.h" |
| 37 | #include "proc.h" |
| 38 | #include "file.h" |
| 39 | #include "malloc.h" |
| 40 | #include "mbuf.h" |
| 41 | #include "domain.h" |
| 42 | #include "kernel.h" |
| 43 | #include "protosw.h" |
| 44 | #include "socket.h" |
| 45 | #include "socketvar.h" |
| 46 | #include "resourcevar.h" |
| 47 | |
| 48 | /* |
| 49 | * Socket operation routines. |
| 50 | * These routines are called by the routines in |
| 51 | * sys_socket.c or from a system process, and |
| 52 | * implement the semantics of socket operations by |
| 53 | * switching out to the protocol specific routines. |
| 54 | */ |
| 55 | /*ARGSUSED*/ |
| 56 | socreate(dom, aso, type, proto) |
| 57 | struct socket **aso; |
| 58 | register int type; |
| 59 | int proto; |
| 60 | { |
| 61 | struct proc *p = curproc; /* XXX */ |
| 62 | register struct protosw *prp; |
| 63 | register struct socket *so; |
| 64 | register int error; |
| 65 | |
| 66 | if (proto) |
| 67 | prp = pffindproto(dom, proto, type); |
| 68 | else |
| 69 | prp = pffindtype(dom, type); |
| 70 | if (prp == 0) |
| 71 | return (EPROTONOSUPPORT); |
| 72 | if (prp->pr_type != type) |
| 73 | return (EPROTOTYPE); |
| 74 | MALLOC(so, struct socket *, sizeof(*so), M_SOCKET, M_WAIT); |
| 75 | bzero((caddr_t)so, sizeof(*so)); |
| 76 | so->so_type = type; |
| 77 | if (p->p_ucred->cr_uid == 0) |
| 78 | so->so_state = SS_PRIV; |
| 79 | so->so_proto = prp; |
| 80 | error = |
| 81 | (*prp->pr_usrreq)(so, PRU_ATTACH, |
| 82 | (struct mbuf *)0, (struct mbuf *)proto, (struct mbuf *)0); |
| 83 | if (error) { |
| 84 | so->so_state |= SS_NOFDREF; |
| 85 | sofree(so); |
| 86 | return (error); |
| 87 | } |
| 88 | *aso = so; |
| 89 | return (0); |
| 90 | } |
| 91 | |
| 92 | sobind(so, nam) |
| 93 | struct socket *so; |
| 94 | struct mbuf *nam; |
| 95 | { |
| 96 | int s = splnet(); |
| 97 | int error; |
| 98 | |
| 99 | error = |
| 100 | (*so->so_proto->pr_usrreq)(so, PRU_BIND, |
| 101 | (struct mbuf *)0, nam, (struct mbuf *)0); |
| 102 | splx(s); |
| 103 | return (error); |
| 104 | } |
| 105 | |
| 106 | solisten(so, backlog) |
| 107 | register struct socket *so; |
| 108 | int backlog; |
| 109 | { |
| 110 | int s = splnet(), error; |
| 111 | |
| 112 | error = |
| 113 | (*so->so_proto->pr_usrreq)(so, PRU_LISTEN, |
| 114 | (struct mbuf *)0, (struct mbuf *)0, (struct mbuf *)0); |
| 115 | if (error) { |
| 116 | splx(s); |
| 117 | return (error); |
| 118 | } |
| 119 | if (so->so_q == 0) |
| 120 | so->so_options |= SO_ACCEPTCONN; |
| 121 | if (backlog < 0) |
| 122 | backlog = 0; |
| 123 | so->so_qlimit = min(backlog, SOMAXCONN); |
| 124 | splx(s); |
| 125 | return (0); |
| 126 | } |
| 127 | |
| 128 | sofree(so) |
| 129 | register struct socket *so; |
| 130 | { |
| 131 | |
| 132 | if (so->so_pcb || (so->so_state & SS_NOFDREF) == 0) |
| 133 | return; |
| 134 | if (so->so_head) { |
| 135 | if (!soqremque(so, 0) && !soqremque(so, 1)) |
| 136 | panic("sofree dq"); |
| 137 | so->so_head = 0; |
| 138 | } |
| 139 | sbrelease(&so->so_snd); |
| 140 | sorflush(so); |
| 141 | FREE(so, M_SOCKET); |
| 142 | } |
| 143 | |
| 144 | /* |
| 145 | * Close a socket on last file table reference removal. |
| 146 | * Initiate disconnect if connected. |
| 147 | * Free socket when disconnect complete. |
| 148 | */ |
| 149 | soclose(so) |
| 150 | register struct socket *so; |
| 151 | { |
| 152 | int s = splnet(); /* conservative */ |
| 153 | int error = 0; |
| 154 | |
| 155 | if (so->so_options & SO_ACCEPTCONN) { |
| 156 | while (so->so_q0) |
| 157 | (void) soabort(so->so_q0); |
| 158 | while (so->so_q) |
| 159 | (void) soabort(so->so_q); |
| 160 | } |
| 161 | if (so->so_pcb == 0) |
| 162 | goto discard; |
| 163 | if (so->so_state & SS_ISCONNECTED) { |
| 164 | if ((so->so_state & SS_ISDISCONNECTING) == 0) { |
| 165 | error = sodisconnect(so); |
| 166 | if (error) |
| 167 | goto drop; |
| 168 | } |
| 169 | if (so->so_options & SO_LINGER) { |
| 170 | if ((so->so_state & SS_ISDISCONNECTING) && |
| 171 | (so->so_state & SS_NBIO)) |
| 172 | goto drop; |
| 173 | while (so->so_state & SS_ISCONNECTED) |
| 174 | if (error = tsleep((caddr_t)&so->so_timeo, |
| 175 | PSOCK | PCATCH, netcls, so->so_linger)) |
| 176 | break; |
| 177 | } |
| 178 | } |
| 179 | drop: |
| 180 | if (so->so_pcb) { |
| 181 | int error2 = |
| 182 | (*so->so_proto->pr_usrreq)(so, PRU_DETACH, |
| 183 | (struct mbuf *)0, (struct mbuf *)0, (struct mbuf *)0); |
| 184 | if (error == 0) |
| 185 | error = error2; |
| 186 | } |
| 187 | discard: |
| 188 | if (so->so_state & SS_NOFDREF) |
| 189 | panic("soclose: NOFDREF"); |
| 190 | so->so_state |= SS_NOFDREF; |
| 191 | sofree(so); |
| 192 | splx(s); |
| 193 | return (error); |
| 194 | } |
| 195 | |
| 196 | /* |
| 197 | * Must be called at splnet... |
| 198 | */ |
| 199 | soabort(so) |
| 200 | struct socket *so; |
| 201 | { |
| 202 | |
| 203 | return ( |
| 204 | (*so->so_proto->pr_usrreq)(so, PRU_ABORT, |
| 205 | (struct mbuf *)0, (struct mbuf *)0, (struct mbuf *)0)); |
| 206 | } |
| 207 | |
| 208 | soaccept(so, nam) |
| 209 | register struct socket *so; |
| 210 | struct mbuf *nam; |
| 211 | { |
| 212 | int s = splnet(); |
| 213 | int error; |
| 214 | |
| 215 | if ((so->so_state & SS_NOFDREF) == 0) |
| 216 | panic("soaccept: !NOFDREF"); |
| 217 | so->so_state &= ~SS_NOFDREF; |
| 218 | error = (*so->so_proto->pr_usrreq)(so, PRU_ACCEPT, |
| 219 | (struct mbuf *)0, nam, (struct mbuf *)0); |
| 220 | splx(s); |
| 221 | return (error); |
| 222 | } |
| 223 | |
| 224 | soconnect(so, nam) |
| 225 | register struct socket *so; |
| 226 | struct mbuf *nam; |
| 227 | { |
| 228 | int s; |
| 229 | int error; |
| 230 | |
| 231 | if (so->so_options & SO_ACCEPTCONN) |
| 232 | return (EOPNOTSUPP); |
| 233 | s = splnet(); |
| 234 | /* |
| 235 | * If protocol is connection-based, can only connect once. |
| 236 | * Otherwise, if connected, try to disconnect first. |
| 237 | * This allows user to disconnect by connecting to, e.g., |
| 238 | * a null address. |
| 239 | */ |
| 240 | if (so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING) && |
| 241 | ((so->so_proto->pr_flags & PR_CONNREQUIRED) || |
| 242 | (error = sodisconnect(so)))) |
| 243 | error = EISCONN; |
| 244 | else |
| 245 | error = (*so->so_proto->pr_usrreq)(so, PRU_CONNECT, |
| 246 | (struct mbuf *)0, nam, (struct mbuf *)0); |
| 247 | splx(s); |
| 248 | return (error); |
| 249 | } |
| 250 | |
| 251 | soconnect2(so1, so2) |
| 252 | register struct socket *so1; |
| 253 | struct socket *so2; |
| 254 | { |
| 255 | int s = splnet(); |
| 256 | int error; |
| 257 | |
| 258 | error = (*so1->so_proto->pr_usrreq)(so1, PRU_CONNECT2, |
| 259 | (struct mbuf *)0, (struct mbuf *)so2, (struct mbuf *)0); |
| 260 | splx(s); |
| 261 | return (error); |
| 262 | } |
| 263 | |
| 264 | sodisconnect(so) |
| 265 | register struct socket *so; |
| 266 | { |
| 267 | int s = splnet(); |
| 268 | int error; |
| 269 | |
| 270 | if ((so->so_state & SS_ISCONNECTED) == 0) { |
| 271 | error = ENOTCONN; |
| 272 | goto bad; |
| 273 | } |
| 274 | if (so->so_state & SS_ISDISCONNECTING) { |
| 275 | error = EALREADY; |
| 276 | goto bad; |
| 277 | } |
| 278 | error = (*so->so_proto->pr_usrreq)(so, PRU_DISCONNECT, |
| 279 | (struct mbuf *)0, (struct mbuf *)0, (struct mbuf *)0); |
| 280 | bad: |
| 281 | splx(s); |
| 282 | return (error); |
| 283 | } |
| 284 | |
| 285 | /* |
| 286 | * Send on a socket. |
| 287 | * If send must go all at once and message is larger than |
| 288 | * send buffering, then hard error. |
| 289 | * Lock against other senders. |
| 290 | * If must go all at once and not enough room now, then |
| 291 | * inform user that this would block and do nothing. |
| 292 | * Otherwise, if nonblocking, send as much as possible. |
| 293 | * The data to be sent is described by "uio" if nonzero, |
| 294 | * otherwise by the mbuf chain "top" (which must be null |
| 295 | * if uio is not). Data provided in mbuf chain must be small |
| 296 | * enough to send all at once. |
| 297 | * |
| 298 | * Returns nonzero on error, timeout or signal; callers |
| 299 | * must check for short counts if EINTR/ERESTART are returned. |
| 300 | * Data and control buffers are freed on return. |
| 301 | */ |
| 302 | sosend(so, addr, uio, top, control, flags) |
| 303 | register struct socket *so; |
| 304 | struct mbuf *addr; |
| 305 | struct uio *uio; |
| 306 | struct mbuf *top; |
| 307 | struct mbuf *control; |
| 308 | int flags; |
| 309 | { |
| 310 | struct proc *p = curproc; /* XXX */ |
| 311 | struct mbuf **mp; |
| 312 | register struct mbuf *m; |
| 313 | register long space, len, resid; |
| 314 | int clen = 0, error, s, dontroute, mlen; |
| 315 | int atomic = sosendallatonce(so) || top; |
| 316 | |
| 317 | if (uio) |
| 318 | resid = uio->uio_resid; |
| 319 | else |
| 320 | resid = top->m_pkthdr.len; |
| 321 | dontroute = |
| 322 | (flags & MSG_DONTROUTE) && (so->so_options & SO_DONTROUTE) == 0 && |
| 323 | (so->so_proto->pr_flags & PR_ATOMIC); |
| 324 | p->p_stats->p_ru.ru_msgsnd++; |
| 325 | if (control) |
| 326 | clen = control->m_len; |
| 327 | #define snderr(errno) { error = errno; splx(s); goto release; } |
| 328 | |
| 329 | restart: |
| 330 | if (error = sblock(&so->so_snd)) |
| 331 | goto out; |
| 332 | do { |
| 333 | s = splnet(); |
| 334 | if (so->so_state & SS_CANTSENDMORE) |
| 335 | snderr(EPIPE); |
| 336 | if (so->so_error) |
| 337 | snderr(so->so_error); |
| 338 | if ((so->so_state & SS_ISCONNECTED) == 0) { |
| 339 | if (so->so_proto->pr_flags & PR_CONNREQUIRED) { |
| 340 | if ((so->so_state & SS_ISCONFIRMING) == 0 && |
| 341 | !(resid == 0 && clen != 0)) |
| 342 | snderr(ENOTCONN); |
| 343 | } else if (addr == 0) |
| 344 | snderr(EDESTADDRREQ); |
| 345 | } |
| 346 | space = sbspace(&so->so_snd); |
| 347 | if (flags & MSG_OOB) |
| 348 | space += 1024; |
| 349 | if (space < resid + clen && |
| 350 | (atomic || space < so->so_snd.sb_lowat || space < clen)) { |
| 351 | if (atomic && resid > so->so_snd.sb_hiwat || |
| 352 | clen > so->so_snd.sb_hiwat) |
| 353 | snderr(EMSGSIZE); |
| 354 | if (so->so_state & SS_NBIO) |
| 355 | snderr(EWOULDBLOCK); |
| 356 | sbunlock(&so->so_snd); |
| 357 | error = sbwait(&so->so_snd); |
| 358 | splx(s); |
| 359 | if (error) |
| 360 | goto out; |
| 361 | goto restart; |
| 362 | } |
| 363 | splx(s); |
| 364 | mp = ⊤ |
| 365 | space -= clen; |
| 366 | do { |
| 367 | if (uio == NULL) { |
| 368 | /* |
| 369 | * Data is prepackaged in "top". |
| 370 | */ |
| 371 | resid = 0; |
| 372 | if (flags & MSG_EOR) |
| 373 | top->m_flags |= M_EOR; |
| 374 | } else do { |
| 375 | if (top == 0) { |
| 376 | MGETHDR(m, M_WAIT, MT_DATA); |
| 377 | mlen = MHLEN; |
| 378 | m->m_pkthdr.len = 0; |
| 379 | m->m_pkthdr.rcvif = (struct ifnet *)0; |
| 380 | } else { |
| 381 | MGET(m, M_WAIT, MT_DATA); |
| 382 | mlen = MLEN; |
| 383 | } |
| 384 | if (resid >= MINCLSIZE && space >= MCLBYTES) { |
| 385 | MCLGET(m, M_WAIT); |
| 386 | if ((m->m_flags & M_EXT) == 0) |
| 387 | goto nopages; |
| 388 | mlen = MCLBYTES; |
| 389 | #ifdef MAPPED_MBUFS |
| 390 | len = min(MCLBYTES, resid); |
| 391 | #else |
| 392 | if (top == 0) { |
| 393 | len = min(MCLBYTES - max_hdr, resid); |
| 394 | m->m_data += max_hdr; |
| 395 | } else |
| 396 | len = min(MCLBYTES, resid); |
| 397 | #endif |
| 398 | space -= MCLBYTES; |
| 399 | } else { |
| 400 | nopages: |
| 401 | len = min(min(mlen, resid), space); |
| 402 | space -= len; |
| 403 | /* |
| 404 | * For datagram protocols, leave room |
| 405 | * for protocol headers in first mbuf. |
| 406 | */ |
| 407 | if (atomic && top == 0 && len < mlen) |
| 408 | MH_ALIGN(m, len); |
| 409 | } |
| 410 | error = uiomove(mtod(m, caddr_t), (int)len, uio); |
| 411 | resid = uio->uio_resid; |
| 412 | m->m_len = len; |
| 413 | *mp = m; |
| 414 | top->m_pkthdr.len += len; |
| 415 | if (error) |
| 416 | goto release; |
| 417 | mp = &m->m_next; |
| 418 | if (resid <= 0) { |
| 419 | if (flags & MSG_EOR) |
| 420 | top->m_flags |= M_EOR; |
| 421 | break; |
| 422 | } |
| 423 | } while (space > 0 && atomic); |
| 424 | if (dontroute) |
| 425 | so->so_options |= SO_DONTROUTE; |
| 426 | s = splnet(); /* XXX */ |
| 427 | error = (*so->so_proto->pr_usrreq)(so, |
| 428 | (flags & MSG_OOB) ? PRU_SENDOOB : PRU_SEND, |
| 429 | top, addr, control); |
| 430 | splx(s); |
| 431 | if (dontroute) |
| 432 | so->so_options &= ~SO_DONTROUTE; |
| 433 | clen = 0; |
| 434 | control = 0; |
| 435 | top = 0; |
| 436 | mp = ⊤ |
| 437 | if (error) |
| 438 | goto release; |
| 439 | } while (resid && space > 0); |
| 440 | } while (resid); |
| 441 | |
| 442 | release: |
| 443 | sbunlock(&so->so_snd); |
| 444 | out: |
| 445 | if (top) |
| 446 | m_freem(top); |
| 447 | if (control) |
| 448 | m_freem(control); |
| 449 | return (error); |
| 450 | } |
| 451 | |
| 452 | /* |
| 453 | * Implement receive operations on a socket. |
| 454 | * We depend on the way that records are added to the sockbuf |
| 455 | * by sbappend*. In particular, each record (mbufs linked through m_next) |
| 456 | * must begin with an address if the protocol so specifies, |
| 457 | * followed by an optional mbuf or mbufs containing ancillary data, |
| 458 | * and then zero or more mbufs of data. |
| 459 | * In order to avoid blocking network interrupts for the entire time here, |
| 460 | * we splx() while doing the actual copy to user space. |
| 461 | * Although the sockbuf is locked, new data may still be appended, |
| 462 | * and thus we must maintain consistency of the sockbuf during that time. |
| 463 | * |
| 464 | * The caller may receive the data as a single mbuf chain by supplying |
| 465 | * an mbuf **mp0 for use in returning the chain. The uio is then used |
| 466 | * only for the count in uio_resid. |
| 467 | */ |
| 468 | soreceive(so, paddr, uio, mp0, controlp, flagsp) |
| 469 | register struct socket *so; |
| 470 | struct mbuf **paddr; |
| 471 | struct uio *uio; |
| 472 | struct mbuf **mp0; |
| 473 | struct mbuf **controlp; |
| 474 | int *flagsp; |
| 475 | { |
| 476 | struct proc *p = curproc; /* XXX */ |
| 477 | register struct mbuf *m, **mp; |
| 478 | register int flags, len, error, s, offset; |
| 479 | struct protosw *pr = so->so_proto; |
| 480 | struct mbuf *nextrecord; |
| 481 | int moff, type; |
| 482 | |
| 483 | mp = mp0; |
| 484 | if (paddr) |
| 485 | *paddr = 0; |
| 486 | if (controlp) |
| 487 | *controlp = 0; |
| 488 | if (flagsp) |
| 489 | flags = *flagsp &~ MSG_EOR; |
| 490 | else |
| 491 | flags = 0; |
| 492 | if (flags & MSG_OOB) { |
| 493 | m = m_get(M_WAIT, MT_DATA); |
| 494 | error = (*pr->pr_usrreq)(so, PRU_RCVOOB, |
| 495 | m, (struct mbuf *)(flags & MSG_PEEK), (struct mbuf *)0); |
| 496 | if (error) |
| 497 | goto bad; |
| 498 | do { |
| 499 | error = uiomove(mtod(m, caddr_t), |
| 500 | (int) min(uio->uio_resid, m->m_len), uio); |
| 501 | m = m_free(m); |
| 502 | } while (uio->uio_resid && error == 0 && m); |
| 503 | bad: |
| 504 | if (m) |
| 505 | m_freem(m); |
| 506 | return (error); |
| 507 | } |
| 508 | if (mp) |
| 509 | *mp = (struct mbuf *)0; |
| 510 | if (so->so_state & SS_ISCONFIRMING && uio->uio_resid) |
| 511 | (*pr->pr_usrreq)(so, PRU_RCVD, (struct mbuf *)0, |
| 512 | (struct mbuf *)0, (struct mbuf *)0); |
| 513 | |
| 514 | restart: |
| 515 | if (error = sblock(&so->so_rcv)) |
| 516 | return (error); |
| 517 | s = splnet(); |
| 518 | |
| 519 | m = so->so_rcv.sb_mb; |
| 520 | /* |
| 521 | * If we have less data than requested, block awaiting more |
| 522 | * (subject to any timeout) if: |
| 523 | * 1. the current count is less than the low water mark, or |
| 524 | * 2. MSG_WAITALL is set, and it is possible to do the entire |
| 525 | * receive operation at once if we block (resid <= hiwat). |
| 526 | * If MSG_WAITALL is set but resid is larger than the receive buffer, |
| 527 | * we have to do the receive in sections, and thus risk returning |
| 528 | * a short count if a timeout or signal occurs after we start. |
| 529 | */ |
| 530 | while (m == 0 || so->so_rcv.sb_cc < uio->uio_resid && |
| 531 | (so->so_rcv.sb_cc < so->so_rcv.sb_lowat || |
| 532 | ((flags & MSG_WAITALL) && uio->uio_resid <= so->so_rcv.sb_hiwat)) && |
| 533 | m->m_nextpkt == 0) { |
| 534 | #ifdef DIAGNOSTIC |
| 535 | if (m == 0 && so->so_rcv.sb_cc) |
| 536 | panic("receive 1"); |
| 537 | #endif |
| 538 | if (so->so_error) { |
| 539 | if (m) |
| 540 | break; |
| 541 | error = so->so_error; |
| 542 | if ((flags & MSG_PEEK) == 0) |
| 543 | so->so_error = 0; |
| 544 | goto release; |
| 545 | } |
| 546 | if (so->so_state & SS_CANTRCVMORE) { |
| 547 | if (m) |
| 548 | break; |
| 549 | else |
| 550 | goto release; |
| 551 | } |
| 552 | for (; m; m = m->m_next) |
| 553 | if (m->m_type == MT_OOBDATA || (m->m_flags & M_EOR)) { |
| 554 | m = so->so_rcv.sb_mb; |
| 555 | goto dontblock; |
| 556 | } |
| 557 | if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) == 0 && |
| 558 | (so->so_proto->pr_flags & PR_CONNREQUIRED)) { |
| 559 | error = ENOTCONN; |
| 560 | goto release; |
| 561 | } |
| 562 | if (uio->uio_resid == 0) |
| 563 | goto release; |
| 564 | if (so->so_state & SS_NBIO) { |
| 565 | error = EWOULDBLOCK; |
| 566 | goto release; |
| 567 | } |
| 568 | sbunlock(&so->so_rcv); |
| 569 | error = sbwait(&so->so_rcv); |
| 570 | splx(s); |
| 571 | if (error) |
| 572 | return (error); |
| 573 | goto restart; |
| 574 | } |
| 575 | dontblock: |
| 576 | p->p_stats->p_ru.ru_msgrcv++; |
| 577 | nextrecord = m->m_nextpkt; |
| 578 | if (pr->pr_flags & PR_ADDR) { |
| 579 | #ifdef DIAGNOSTIC |
| 580 | if (m->m_type != MT_SONAME) |
| 581 | panic("receive 1a"); |
| 582 | #endif |
| 583 | if (flags & MSG_PEEK) { |
| 584 | if (paddr) |
| 585 | *paddr = m_copy(m, 0, m->m_len); |
| 586 | m = m->m_next; |
| 587 | } else { |
| 588 | sbfree(&so->so_rcv, m); |
| 589 | if (paddr) { |
| 590 | *paddr = m; |
| 591 | so->so_rcv.sb_mb = m->m_next; |
| 592 | m->m_next = 0; |
| 593 | m = so->so_rcv.sb_mb; |
| 594 | } else { |
| 595 | MFREE(m, so->so_rcv.sb_mb); |
| 596 | m = so->so_rcv.sb_mb; |
| 597 | } |
| 598 | } |
| 599 | } |
| 600 | while (m && m->m_type == MT_CONTROL && error == 0) { |
| 601 | if (flags & MSG_PEEK) { |
| 602 | if (controlp) |
| 603 | *controlp = m_copy(m, 0, m->m_len); |
| 604 | m = m->m_next; |
| 605 | } else { |
| 606 | sbfree(&so->so_rcv, m); |
| 607 | if (controlp) { |
| 608 | if (pr->pr_domain->dom_externalize && |
| 609 | mtod(m, struct cmsghdr *)->cmsg_type == |
| 610 | SCM_RIGHTS) |
| 611 | error = (*pr->pr_domain->dom_externalize)(m); |
| 612 | *controlp = m; |
| 613 | so->so_rcv.sb_mb = m->m_next; |
| 614 | m->m_next = 0; |
| 615 | m = so->so_rcv.sb_mb; |
| 616 | } else { |
| 617 | MFREE(m, so->so_rcv.sb_mb); |
| 618 | m = so->so_rcv.sb_mb; |
| 619 | } |
| 620 | } |
| 621 | if (controlp) |
| 622 | controlp = &(*controlp)->m_next; |
| 623 | } |
| 624 | if (m) { |
| 625 | if ((flags & MSG_PEEK) == 0) |
| 626 | m->m_nextpkt = nextrecord; |
| 627 | type = m->m_type; |
| 628 | if (type == MT_OOBDATA) |
| 629 | flags |= MSG_OOB; |
| 630 | } |
| 631 | moff = 0; |
| 632 | offset = 0; |
| 633 | while (m && uio->uio_resid > 0 && error == 0) { |
| 634 | if (m->m_type == MT_OOBDATA) { |
| 635 | if (type != MT_OOBDATA) |
| 636 | break; |
| 637 | } else if (type == MT_OOBDATA) |
| 638 | break; |
| 639 | #ifdef DIAGNOSTIC |
| 640 | else if (m->m_type != MT_DATA && m->m_type != MT_HEADER) |
| 641 | panic("receive 3"); |
| 642 | #endif |
| 643 | so->so_state &= ~SS_RCVATMARK; |
| 644 | len = uio->uio_resid; |
| 645 | if (so->so_oobmark && len > so->so_oobmark - offset) |
| 646 | len = so->so_oobmark - offset; |
| 647 | if (len > m->m_len - moff) |
| 648 | len = m->m_len - moff; |
| 649 | /* |
| 650 | * If mp is set, just pass back the mbufs. |
| 651 | * Otherwise copy them out via the uio, then free. |
| 652 | * Sockbuf must be consistent here (points to current mbuf, |
| 653 | * it points to next record) when we drop priority; |
| 654 | * we must note any additions to the sockbuf when we |
| 655 | * block interrupts again. |
| 656 | */ |
| 657 | if (mp == 0) { |
| 658 | splx(s); |
| 659 | error = uiomove(mtod(m, caddr_t) + moff, (int)len, uio); |
| 660 | s = splnet(); |
| 661 | } else |
| 662 | uio->uio_resid -= len; |
| 663 | if (len == m->m_len - moff) { |
| 664 | if (m->m_flags & M_EOR) |
| 665 | flags |= MSG_EOR; |
| 666 | if (flags & MSG_PEEK) { |
| 667 | m = m->m_next; |
| 668 | moff = 0; |
| 669 | } else { |
| 670 | nextrecord = m->m_nextpkt; |
| 671 | sbfree(&so->so_rcv, m); |
| 672 | if (mp) { |
| 673 | *mp = m; |
| 674 | mp = &m->m_next; |
| 675 | so->so_rcv.sb_mb = m = m->m_next; |
| 676 | *mp = (struct mbuf *)0; |
| 677 | } else { |
| 678 | MFREE(m, so->so_rcv.sb_mb); |
| 679 | m = so->so_rcv.sb_mb; |
| 680 | } |
| 681 | if (m) |
| 682 | m->m_nextpkt = nextrecord; |
| 683 | } |
| 684 | } else { |
| 685 | if (flags & MSG_PEEK) |
| 686 | moff += len; |
| 687 | else { |
| 688 | if (mp) |
| 689 | *mp = m_copym(m, 0, len, M_WAIT); |
| 690 | m->m_data += len; |
| 691 | m->m_len -= len; |
| 692 | so->so_rcv.sb_cc -= len; |
| 693 | } |
| 694 | } |
| 695 | if (so->so_oobmark) { |
| 696 | if ((flags & MSG_PEEK) == 0) { |
| 697 | so->so_oobmark -= len; |
| 698 | if (so->so_oobmark == 0) { |
| 699 | so->so_state |= SS_RCVATMARK; |
| 700 | break; |
| 701 | } |
| 702 | } else |
| 703 | offset += len; |
| 704 | } |
| 705 | if (flags & MSG_EOR) |
| 706 | break; |
| 707 | /* |
| 708 | * If the MSG_WAITALL flag is set (for non-atomic socket), |
| 709 | * we must not quit until "uio->uio_resid == 0" or an error |
| 710 | * termination. If a signal/timeout occurs, return |
| 711 | * with a short count but without error. |
| 712 | * Keep sockbuf locked against other readers. |
| 713 | */ |
| 714 | while (flags & MSG_WAITALL && m == 0 && uio->uio_resid > 0 && |
| 715 | !sosendallatonce(so)) { |
| 716 | if (so->so_error || so->so_state & SS_CANTRCVMORE) |
| 717 | break; |
| 718 | error = sbwait(&so->so_rcv); |
| 719 | if (error) { |
| 720 | sbunlock(&so->so_rcv); |
| 721 | splx(s); |
| 722 | return (0); |
| 723 | } |
| 724 | if (m = so->so_rcv.sb_mb) |
| 725 | nextrecord = m->m_nextpkt; |
| 726 | } |
| 727 | } |
| 728 | if ((flags & MSG_PEEK) == 0) { |
| 729 | if (m == 0) |
| 730 | so->so_rcv.sb_mb = nextrecord; |
| 731 | else if (pr->pr_flags & PR_ATOMIC) { |
| 732 | flags |= MSG_TRUNC; |
| 733 | (void) sbdroprecord(&so->so_rcv); |
| 734 | } |
| 735 | if (pr->pr_flags & PR_WANTRCVD && so->so_pcb) |
| 736 | (*pr->pr_usrreq)(so, PRU_RCVD, (struct mbuf *)0, |
| 737 | (struct mbuf *)flags, (struct mbuf *)0, |
| 738 | (struct mbuf *)0); |
| 739 | } |
| 740 | if (flagsp) |
| 741 | *flagsp |= flags; |
| 742 | release: |
| 743 | sbunlock(&so->so_rcv); |
| 744 | splx(s); |
| 745 | return (error); |
| 746 | } |
| 747 | |
| 748 | soshutdown(so, how) |
| 749 | register struct socket *so; |
| 750 | register int how; |
| 751 | { |
| 752 | register struct protosw *pr = so->so_proto; |
| 753 | |
| 754 | how++; |
| 755 | if (how & FREAD) |
| 756 | sorflush(so); |
| 757 | if (how & FWRITE) |
| 758 | return ((*pr->pr_usrreq)(so, PRU_SHUTDOWN, |
| 759 | (struct mbuf *)0, (struct mbuf *)0, (struct mbuf *)0)); |
| 760 | return (0); |
| 761 | } |
| 762 | |
| 763 | sorflush(so) |
| 764 | register struct socket *so; |
| 765 | { |
| 766 | register struct sockbuf *sb = &so->so_rcv; |
| 767 | register struct protosw *pr = so->so_proto; |
| 768 | register int s; |
| 769 | struct sockbuf asb; |
| 770 | |
| 771 | sb->sb_flags |= SB_NOINTR; |
| 772 | (void) sblock(sb); |
| 773 | s = splimp(); |
| 774 | socantrcvmore(so); |
| 775 | sbunlock(sb); |
| 776 | asb = *sb; |
| 777 | bzero((caddr_t)sb, sizeof (*sb)); |
| 778 | splx(s); |
| 779 | if (pr->pr_flags & PR_RIGHTS && pr->pr_domain->dom_dispose) |
| 780 | (*pr->pr_domain->dom_dispose)(asb.sb_mb); |
| 781 | sbrelease(&asb); |
| 782 | } |
| 783 | |
| 784 | sosetopt(so, level, optname, m0) |
| 785 | register struct socket *so; |
| 786 | int level, optname; |
| 787 | struct mbuf *m0; |
| 788 | { |
| 789 | int error = 0; |
| 790 | register struct mbuf *m = m0; |
| 791 | |
| 792 | if (level != SOL_SOCKET) { |
| 793 | if (so->so_proto && so->so_proto->pr_ctloutput) |
| 794 | return ((*so->so_proto->pr_ctloutput) |
| 795 | (PRCO_SETOPT, so, level, optname, &m0)); |
| 796 | error = ENOPROTOOPT; |
| 797 | } else { |
| 798 | switch (optname) { |
| 799 | |
| 800 | case SO_LINGER: |
| 801 | if (m == NULL || m->m_len != sizeof (struct linger)) { |
| 802 | error = EINVAL; |
| 803 | goto bad; |
| 804 | } |
| 805 | so->so_linger = mtod(m, struct linger *)->l_linger; |
| 806 | /* fall thru... */ |
| 807 | |
| 808 | case SO_DEBUG: |
| 809 | case SO_KEEPALIVE: |
| 810 | case SO_DONTROUTE: |
| 811 | case SO_USELOOPBACK: |
| 812 | case SO_BROADCAST: |
| 813 | case SO_REUSEADDR: |
| 814 | case SO_OOBINLINE: |
| 815 | if (m == NULL || m->m_len < sizeof (int)) { |
| 816 | error = EINVAL; |
| 817 | goto bad; |
| 818 | } |
| 819 | if (*mtod(m, int *)) |
| 820 | so->so_options |= optname; |
| 821 | else |
| 822 | so->so_options &= ~optname; |
| 823 | break; |
| 824 | |
| 825 | case SO_SNDBUF: |
| 826 | case SO_RCVBUF: |
| 827 | case SO_SNDLOWAT: |
| 828 | case SO_RCVLOWAT: |
| 829 | if (m == NULL || m->m_len < sizeof (int)) { |
| 830 | error = EINVAL; |
| 831 | goto bad; |
| 832 | } |
| 833 | switch (optname) { |
| 834 | |
| 835 | case SO_SNDBUF: |
| 836 | case SO_RCVBUF: |
| 837 | if (sbreserve(optname == SO_SNDBUF ? |
| 838 | &so->so_snd : &so->so_rcv, |
| 839 | (u_long) *mtod(m, int *)) == 0) { |
| 840 | error = ENOBUFS; |
| 841 | goto bad; |
| 842 | } |
| 843 | break; |
| 844 | |
| 845 | case SO_SNDLOWAT: |
| 846 | so->so_snd.sb_lowat = *mtod(m, int *); |
| 847 | break; |
| 848 | case SO_RCVLOWAT: |
| 849 | so->so_rcv.sb_lowat = *mtod(m, int *); |
| 850 | break; |
| 851 | } |
| 852 | break; |
| 853 | |
| 854 | case SO_SNDTIMEO: |
| 855 | case SO_RCVTIMEO: |
| 856 | { |
| 857 | struct timeval *tv; |
| 858 | short val; |
| 859 | |
| 860 | if (m == NULL || m->m_len < sizeof (*tv)) { |
| 861 | error = EINVAL; |
| 862 | goto bad; |
| 863 | } |
| 864 | tv = mtod(m, struct timeval *); |
| 865 | if (tv->tv_sec > SHRT_MAX / hz - hz) { |
| 866 | error = EDOM; |
| 867 | goto bad; |
| 868 | } |
| 869 | val = tv->tv_sec * hz + tv->tv_usec / tick; |
| 870 | |
| 871 | switch (optname) { |
| 872 | |
| 873 | case SO_SNDTIMEO: |
| 874 | so->so_snd.sb_timeo = val; |
| 875 | break; |
| 876 | case SO_RCVTIMEO: |
| 877 | so->so_rcv.sb_timeo = val; |
| 878 | break; |
| 879 | } |
| 880 | break; |
| 881 | } |
| 882 | |
| 883 | default: |
| 884 | error = ENOPROTOOPT; |
| 885 | break; |
| 886 | } |
| 887 | } |
| 888 | bad: |
| 889 | if (m) |
| 890 | (void) m_free(m); |
| 891 | return (error); |
| 892 | } |
| 893 | |
| 894 | sogetopt(so, level, optname, mp) |
| 895 | register struct socket *so; |
| 896 | int level, optname; |
| 897 | struct mbuf **mp; |
| 898 | { |
| 899 | register struct mbuf *m; |
| 900 | |
| 901 | if (level != SOL_SOCKET) { |
| 902 | if (so->so_proto && so->so_proto->pr_ctloutput) { |
| 903 | return ((*so->so_proto->pr_ctloutput) |
| 904 | (PRCO_GETOPT, so, level, optname, mp)); |
| 905 | } else |
| 906 | return (ENOPROTOOPT); |
| 907 | } else { |
| 908 | m = m_get(M_WAIT, MT_SOOPTS); |
| 909 | m->m_len = sizeof (int); |
| 910 | |
| 911 | switch (optname) { |
| 912 | |
| 913 | case SO_LINGER: |
| 914 | m->m_len = sizeof (struct linger); |
| 915 | mtod(m, struct linger *)->l_onoff = |
| 916 | so->so_options & SO_LINGER; |
| 917 | mtod(m, struct linger *)->l_linger = so->so_linger; |
| 918 | break; |
| 919 | |
| 920 | case SO_USELOOPBACK: |
| 921 | case SO_DONTROUTE: |
| 922 | case SO_DEBUG: |
| 923 | case SO_KEEPALIVE: |
| 924 | case SO_REUSEADDR: |
| 925 | case SO_BROADCAST: |
| 926 | case SO_OOBINLINE: |
| 927 | *mtod(m, int *) = so->so_options & optname; |
| 928 | break; |
| 929 | |
| 930 | case SO_TYPE: |
| 931 | *mtod(m, int *) = so->so_type; |
| 932 | break; |
| 933 | |
| 934 | case SO_ERROR: |
| 935 | *mtod(m, int *) = so->so_error; |
| 936 | so->so_error = 0; |
| 937 | break; |
| 938 | |
| 939 | case SO_SNDBUF: |
| 940 | *mtod(m, int *) = so->so_snd.sb_hiwat; |
| 941 | break; |
| 942 | |
| 943 | case SO_RCVBUF: |
| 944 | *mtod(m, int *) = so->so_rcv.sb_hiwat; |
| 945 | break; |
| 946 | |
| 947 | case SO_SNDLOWAT: |
| 948 | *mtod(m, int *) = so->so_snd.sb_lowat; |
| 949 | break; |
| 950 | |
| 951 | case SO_RCVLOWAT: |
| 952 | *mtod(m, int *) = so->so_rcv.sb_lowat; |
| 953 | break; |
| 954 | |
| 955 | case SO_SNDTIMEO: |
| 956 | case SO_RCVTIMEO: |
| 957 | { |
| 958 | int val = (optname == SO_SNDTIMEO ? |
| 959 | so->so_snd.sb_timeo : so->so_rcv.sb_timeo); |
| 960 | |
| 961 | m->m_len = sizeof(struct timeval); |
| 962 | mtod(m, struct timeval *)->tv_sec = val / hz; |
| 963 | mtod(m, struct timeval *)->tv_usec = |
| 964 | (val % hz) / tick; |
| 965 | break; |
| 966 | } |
| 967 | |
| 968 | default: |
| 969 | (void)m_free(m); |
| 970 | return (ENOPROTOOPT); |
| 971 | } |
| 972 | *mp = m; |
| 973 | return (0); |
| 974 | } |
| 975 | } |
| 976 | |
| 977 | sohasoutofband(so) |
| 978 | register struct socket *so; |
| 979 | { |
| 980 | struct proc *p; |
| 981 | |
| 982 | if (so->so_pgid < 0) |
| 983 | gsignal(-so->so_pgid, SIGURG); |
| 984 | else if (so->so_pgid > 0 && (p = pfind(so->so_pgid)) != 0) |
| 985 | psignal(p, SIGURG); |
| 986 | if (so->so_rcv.sb_sel) { |
| 987 | selwakeup(so->so_rcv.sb_sel, so->so_rcv.sb_flags & SB_COLL); |
| 988 | so->so_rcv.sb_sel = 0; |
| 989 | so->so_rcv.sb_flags &= ~SB_COLL; |
| 990 | } |
| 991 | } |