[39e6e65a] | 1 | /* |
---|
| 2 | * This file has undergone several changes to reflect the |
---|
| 3 | * differences between the RTEMS and FreeBSD kernels. |
---|
| 4 | */ |
---|
| 5 | |
---|
| 6 | /* |
---|
| 7 | * Copyright (c) 1982, 1986, 1988, 1990, 1993 |
---|
| 8 | * The Regents of the University of California. All rights reserved. |
---|
| 9 | * |
---|
| 10 | * Redistribution and use in source and binary forms, with or without |
---|
| 11 | * modification, are permitted provided that the following conditions |
---|
| 12 | * are met: |
---|
| 13 | * 1. Redistributions of source code must retain the above copyright |
---|
| 14 | * notice, this list of conditions and the following disclaimer. |
---|
| 15 | * 2. Redistributions in binary form must reproduce the above copyright |
---|
| 16 | * notice, this list of conditions and the following disclaimer in the |
---|
| 17 | * documentation and/or other materials provided with the distribution. |
---|
| 18 | * 3. All advertising materials mentioning features or use of this software |
---|
| 19 | * must display the following acknowledgement: |
---|
| 20 | * This product includes software developed by the University of |
---|
| 21 | * California, Berkeley and its contributors. |
---|
| 22 | * 4. Neither the name of the University nor the names of its contributors |
---|
| 23 | * may be used to endorse or promote products derived from this software |
---|
| 24 | * without specific prior written permission. |
---|
| 25 | * |
---|
| 26 | * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND |
---|
| 27 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
---|
| 28 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
---|
| 29 | * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE |
---|
| 30 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL |
---|
| 31 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS |
---|
| 32 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) |
---|
| 33 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT |
---|
| 34 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY |
---|
| 35 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF |
---|
| 36 | * SUCH DAMAGE. |
---|
| 37 | * |
---|
| 38 | * @(#)uipc_socket2.c 8.1 (Berkeley) 6/10/93 |
---|
| 39 | * $Id$ |
---|
| 40 | */ |
---|
| 41 | |
---|
| 42 | #include <sys/param.h> |
---|
| 43 | #include <sys/systm.h> |
---|
| 44 | #include <sys/kernel.h> |
---|
| 45 | #include <sys/proc.h> |
---|
| 46 | #include <sys/file.h> |
---|
| 47 | #include <sys/buf.h> |
---|
| 48 | #include <sys/malloc.h> |
---|
| 49 | #include <sys/mbuf.h> |
---|
| 50 | #include <sys/protosw.h> |
---|
| 51 | #include <sys/stat.h> |
---|
| 52 | #include <sys/socket.h> |
---|
| 53 | #include <sys/socketvar.h> |
---|
| 54 | #include <sys/signalvar.h> |
---|
| 55 | #include <sys/sysctl.h> |
---|
| 56 | |
---|
| 57 | /* |
---|
| 58 | * Primitive routines for operating on sockets and socket buffers |
---|
| 59 | */ |
---|
| 60 | |
---|
| 61 | u_long sb_max = SB_MAX; /* XXX should be static */ |
---|
[36799d4] | 62 | SYSCTL_INT(_kern, KIPC_MAXSOCKBUF, maxsockbuf, CTLFLAG_RW, &sb_max, 0, ""); |
---|
[39e6e65a] | 63 | |
---|
| 64 | static u_long sb_efficiency = 8; /* parameter for sbreserve() */ |
---|
| 65 | SYSCTL_INT(_kern, OID_AUTO, sockbuf_waste_factor, CTLFLAG_RW, &sb_efficiency, |
---|
| 66 | 0, ""); |
---|
| 67 | |
---|
| 68 | /* |
---|
| 69 | * Procedures to manipulate state flags of socket |
---|
| 70 | * and do appropriate wakeups. Normal sequence from the |
---|
| 71 | * active (originating) side is that soisconnecting() is |
---|
| 72 | * called during processing of connect() call, |
---|
| 73 | * resulting in an eventual call to soisconnected() if/when the |
---|
| 74 | * connection is established. When the connection is torn down |
---|
| 75 | * soisdisconnecting() is called during processing of disconnect() call, |
---|
| 76 | * and soisdisconnected() is called when the connection to the peer |
---|
| 77 | * is totally severed. The semantics of these routines are such that |
---|
| 78 | * connectionless protocols can call soisconnected() and soisdisconnected() |
---|
| 79 | * only, bypassing the in-progress calls when setting up a ``connection'' |
---|
| 80 | * takes no time. |
---|
| 81 | * |
---|
| 82 | * From the passive side, a socket is created with |
---|
| 83 | * two queues of sockets: so_q0 for connections in progress |
---|
| 84 | * and so_q for connections already made and awaiting user acceptance. |
---|
| 85 | * As a protocol is preparing incoming connections, it creates a socket |
---|
| 86 | * structure queued on so_q0 by calling sonewconn(). When the connection |
---|
| 87 | * is established, soisconnected() is called, and transfers the |
---|
| 88 | * socket structure to so_q, making it available to accept(). |
---|
| 89 | * |
---|
| 90 | * If a socket is closed with sockets on either |
---|
| 91 | * so_q0 or so_q, these sockets are dropped. |
---|
| 92 | * |
---|
| 93 | * If higher level protocols are implemented in |
---|
| 94 | * the kernel, the wakeups done here will sometimes |
---|
| 95 | * cause software-interrupt process scheduling. |
---|
| 96 | */ |
---|
| 97 | |
---|
| 98 | void |
---|
| 99 | soisconnecting(so) |
---|
| 100 | register struct socket *so; |
---|
| 101 | { |
---|
| 102 | |
---|
| 103 | so->so_state &= ~(SS_ISCONNECTED|SS_ISDISCONNECTING); |
---|
| 104 | so->so_state |= SS_ISCONNECTING; |
---|
| 105 | } |
---|
| 106 | |
---|
| 107 | void |
---|
| 108 | soisconnected(so) |
---|
| 109 | register struct socket *so; |
---|
| 110 | { |
---|
| 111 | register struct socket *head = so->so_head; |
---|
| 112 | |
---|
| 113 | so->so_state &= ~(SS_ISCONNECTING|SS_ISDISCONNECTING|SS_ISCONFIRMING); |
---|
| 114 | so->so_state |= SS_ISCONNECTED; |
---|
| 115 | if (head && (so->so_state & SS_INCOMP)) { |
---|
| 116 | TAILQ_REMOVE(&head->so_incomp, so, so_list); |
---|
| 117 | head->so_incqlen--; |
---|
| 118 | so->so_state &= ~SS_INCOMP; |
---|
| 119 | TAILQ_INSERT_TAIL(&head->so_comp, so, so_list); |
---|
| 120 | so->so_state |= SS_COMP; |
---|
| 121 | sorwakeup(head); |
---|
| 122 | soconnwakeup(head); |
---|
| 123 | } else { |
---|
| 124 | soconnwakeup(so); |
---|
| 125 | sorwakeup(so); |
---|
| 126 | sowwakeup(so); |
---|
| 127 | } |
---|
| 128 | } |
---|
| 129 | |
---|
| 130 | void |
---|
| 131 | soisdisconnecting(so) |
---|
| 132 | register struct socket *so; |
---|
| 133 | { |
---|
| 134 | |
---|
| 135 | so->so_state &= ~SS_ISCONNECTING; |
---|
| 136 | so->so_state |= (SS_ISDISCONNECTING|SS_CANTRCVMORE|SS_CANTSENDMORE); |
---|
| 137 | soconnwakeup(so); |
---|
| 138 | sowwakeup(so); |
---|
| 139 | sorwakeup(so); |
---|
| 140 | } |
---|
| 141 | |
---|
| 142 | void |
---|
| 143 | soisdisconnected(so) |
---|
| 144 | register struct socket *so; |
---|
| 145 | { |
---|
| 146 | |
---|
| 147 | so->so_state &= ~(SS_ISCONNECTING|SS_ISCONNECTED|SS_ISDISCONNECTING); |
---|
| 148 | so->so_state |= (SS_CANTRCVMORE|SS_CANTSENDMORE); |
---|
| 149 | soconnwakeup(so); |
---|
| 150 | sowwakeup(so); |
---|
| 151 | sorwakeup(so); |
---|
| 152 | } |
---|
| 153 | |
---|
| 154 | /* |
---|
| 155 | * Return a random connection that hasn't been serviced yet and |
---|
| 156 | * is eligible for discard. There is a one in qlen chance that |
---|
| 157 | * we will return a null, saying that there are no dropable |
---|
| 158 | * requests. In this case, the protocol specific code should drop |
---|
| 159 | * the new request. This insures fairness. |
---|
| 160 | * |
---|
| 161 | * This may be used in conjunction with protocol specific queue |
---|
| 162 | * congestion routines. |
---|
| 163 | */ |
---|
| 164 | struct socket * |
---|
| 165 | sodropablereq(head) |
---|
| 166 | register struct socket *head; |
---|
| 167 | { |
---|
| 168 | register struct socket *so; |
---|
[ae5a79c3] | 169 | uint32_t i, j, qlen, m; |
---|
[39e6e65a] | 170 | |
---|
| 171 | static int rnd; |
---|
| 172 | static long old_mono_secs; |
---|
| 173 | static unsigned int cur_cnt, old_cnt; |
---|
| 174 | |
---|
| 175 | if ((i = (m = rtems_bsdnet_seconds_since_boot()) - old_mono_secs) != 0) { |
---|
| 176 | old_mono_secs = m; |
---|
| 177 | old_cnt = cur_cnt / i; |
---|
| 178 | cur_cnt = 0; |
---|
| 179 | } |
---|
| 180 | |
---|
| 181 | so = TAILQ_FIRST(&head->so_incomp); |
---|
| 182 | if (!so) |
---|
| 183 | return (so); |
---|
| 184 | |
---|
| 185 | qlen = head->so_incqlen; |
---|
| 186 | if (++cur_cnt > qlen || old_cnt > qlen) { |
---|
| 187 | rnd = (314159 * rnd + 66329) & 0xffff; |
---|
| 188 | j = ((qlen + 1) * rnd) >> 16; |
---|
| 189 | |
---|
| 190 | while (j-- && so) |
---|
| 191 | so = TAILQ_NEXT(so, so_list); |
---|
| 192 | } |
---|
| 193 | |
---|
| 194 | return (so); |
---|
| 195 | } |
---|
| 196 | |
---|
| 197 | /* |
---|
| 198 | * When an attempt at a new connection is noted on a socket |
---|
| 199 | * which accepts connections, sonewconn is called. If the |
---|
| 200 | * connection is possible (subject to space constraints, etc.) |
---|
| 201 | * then we allocate a new structure, propoerly linked into the |
---|
| 202 | * data structure of the original socket, and return this. |
---|
| 203 | * Connstatus may be 0, or SO_ISCONFIRMING, or SO_ISCONNECTED. |
---|
| 204 | * |
---|
| 205 | * Currently, sonewconn() is defined as sonewconn1() in socketvar.h |
---|
| 206 | * to catch calls that are missing the (new) second parameter. |
---|
| 207 | */ |
---|
| 208 | struct socket * |
---|
| 209 | sonewconn1(head, connstatus) |
---|
| 210 | register struct socket *head; |
---|
| 211 | int connstatus; |
---|
| 212 | { |
---|
| 213 | register struct socket *so; |
---|
| 214 | |
---|
| 215 | if (head->so_qlen > 3 * head->so_qlimit / 2) |
---|
| 216 | return ((struct socket *)0); |
---|
| 217 | MALLOC(so, struct socket *, sizeof(*so), M_SOCKET, M_DONTWAIT); |
---|
| 218 | if (so == NULL) |
---|
| 219 | return ((struct socket *)0); |
---|
| 220 | bzero((caddr_t)so, sizeof(*so)); |
---|
| 221 | so->so_head = head; |
---|
| 222 | so->so_type = head->so_type; |
---|
| 223 | so->so_options = head->so_options &~ SO_ACCEPTCONN; |
---|
| 224 | so->so_linger = head->so_linger; |
---|
| 225 | so->so_state = head->so_state | SS_NOFDREF; |
---|
| 226 | so->so_proto = head->so_proto; |
---|
| 227 | so->so_timeo = head->so_timeo; |
---|
| 228 | so->so_pgid = head->so_pgid; |
---|
| 229 | so->so_uid = head->so_uid; |
---|
| 230 | (void) soreserve(so, head->so_snd.sb_hiwat, head->so_rcv.sb_hiwat); |
---|
| 231 | if (connstatus) { |
---|
| 232 | TAILQ_INSERT_TAIL(&head->so_comp, so, so_list); |
---|
| 233 | so->so_state |= SS_COMP; |
---|
| 234 | } else { |
---|
| 235 | TAILQ_INSERT_TAIL(&head->so_incomp, so, so_list); |
---|
| 236 | so->so_state |= SS_INCOMP; |
---|
| 237 | head->so_incqlen++; |
---|
| 238 | } |
---|
| 239 | head->so_qlen++; |
---|
| 240 | if ((*so->so_proto->pr_usrreqs->pru_attach)(so, 0)) { |
---|
| 241 | if (so->so_state & SS_COMP) { |
---|
| 242 | TAILQ_REMOVE(&head->so_comp, so, so_list); |
---|
| 243 | } else { |
---|
| 244 | TAILQ_REMOVE(&head->so_incomp, so, so_list); |
---|
| 245 | head->so_incqlen--; |
---|
| 246 | } |
---|
| 247 | head->so_qlen--; |
---|
| 248 | (void) free((caddr_t)so, M_SOCKET); |
---|
| 249 | return ((struct socket *)0); |
---|
| 250 | } |
---|
| 251 | if (connstatus) { |
---|
| 252 | sorwakeup(head); |
---|
| 253 | soconnwakeup(head); |
---|
| 254 | so->so_state |= connstatus; |
---|
| 255 | } |
---|
| 256 | return (so); |
---|
| 257 | } |
---|
| 258 | |
---|
| 259 | /* |
---|
| 260 | * Socantsendmore indicates that no more data will be sent on the |
---|
| 261 | * socket; it would normally be applied to a socket when the user |
---|
| 262 | * informs the system that no more data is to be sent, by the protocol |
---|
| 263 | * code (in case PRU_SHUTDOWN). Socantrcvmore indicates that no more data |
---|
| 264 | * will be received, and will normally be applied to the socket by a |
---|
| 265 | * protocol when it detects that the peer will send no more data. |
---|
| 266 | * Data queued for reading in the socket may yet be read. |
---|
| 267 | */ |
---|
| 268 | |
---|
| 269 | void |
---|
| 270 | socantsendmore(so) |
---|
| 271 | struct socket *so; |
---|
| 272 | { |
---|
| 273 | |
---|
| 274 | so->so_state |= SS_CANTSENDMORE; |
---|
| 275 | sowwakeup(so); |
---|
| 276 | } |
---|
| 277 | |
---|
| 278 | void |
---|
| 279 | socantrcvmore(so) |
---|
| 280 | struct socket *so; |
---|
| 281 | { |
---|
| 282 | |
---|
| 283 | so->so_state |= SS_CANTRCVMORE; |
---|
| 284 | sorwakeup(so); |
---|
| 285 | } |
---|
| 286 | |
---|
| 287 | /* |
---|
| 288 | * Socket buffer (struct sockbuf) utility routines. |
---|
| 289 | * |
---|
| 290 | * Each socket contains two socket buffers: one for sending data and |
---|
| 291 | * one for receiving data. Each buffer contains a queue of mbufs, |
---|
| 292 | * information about the number of mbufs and amount of data in the |
---|
| 293 | * queue, and other fields allowing select() statements and notification |
---|
| 294 | * on data availability to be implemented. |
---|
| 295 | * |
---|
| 296 | * Data stored in a socket buffer is maintained as a list of records. |
---|
| 297 | * Each record is a list of mbufs chained together with the m_next |
---|
| 298 | * field. Records are chained together with the m_nextpkt field. The upper |
---|
| 299 | * level routine soreceive() expects the following conventions to be |
---|
| 300 | * observed when placing information in the receive buffer: |
---|
| 301 | * |
---|
| 302 | * 1. If the protocol requires each message be preceded by the sender's |
---|
| 303 | * name, then a record containing that name must be present before |
---|
| 304 | * any associated data (mbuf's must be of type MT_SONAME). |
---|
| 305 | * 2. If the protocol supports the exchange of ``access rights'' (really |
---|
| 306 | * just additional data associated with the message), and there are |
---|
| 307 | * ``rights'' to be received, then a record containing this data |
---|
| 308 | * should be present (mbuf's must be of type MT_RIGHTS). |
---|
| 309 | * 3. If a name or rights record exists, then it must be followed by |
---|
| 310 | * a data record, perhaps of zero length. |
---|
| 311 | * |
---|
| 312 | * Before using a new socket structure it is first necessary to reserve |
---|
| 313 | * buffer space to the socket, by calling sbreserve(). This should commit |
---|
| 314 | * some of the available buffer space in the system buffer pool for the |
---|
| 315 | * socket (currently, it does nothing but enforce limits). The space |
---|
| 316 | * should be released by calling sbrelease() when the socket is destroyed. |
---|
| 317 | */ |
---|
| 318 | |
---|
| 319 | int |
---|
| 320 | soreserve(so, sndcc, rcvcc) |
---|
| 321 | register struct socket *so; |
---|
| 322 | u_long sndcc, rcvcc; |
---|
| 323 | { |
---|
| 324 | |
---|
| 325 | if (sbreserve(&so->so_snd, sndcc) == 0) |
---|
| 326 | goto bad; |
---|
| 327 | if (sbreserve(&so->so_rcv, rcvcc) == 0) |
---|
| 328 | goto bad2; |
---|
| 329 | if (so->so_rcv.sb_lowat == 0) |
---|
| 330 | so->so_rcv.sb_lowat = 1; |
---|
| 331 | if (so->so_snd.sb_lowat == 0) |
---|
| 332 | so->so_snd.sb_lowat = MCLBYTES; |
---|
| 333 | if (so->so_snd.sb_lowat > so->so_snd.sb_hiwat) |
---|
| 334 | so->so_snd.sb_lowat = so->so_snd.sb_hiwat; |
---|
| 335 | return (0); |
---|
| 336 | bad2: |
---|
| 337 | sbrelease(&so->so_snd); |
---|
| 338 | bad: |
---|
| 339 | return (ENOBUFS); |
---|
| 340 | } |
---|
| 341 | |
---|
| 342 | /* |
---|
| 343 | * Allot mbufs to a sockbuf. |
---|
| 344 | * Attempt to scale mbmax so that mbcnt doesn't become limiting |
---|
| 345 | * if buffering efficiency is near the normal case. |
---|
| 346 | */ |
---|
| 347 | int |
---|
| 348 | sbreserve(sb, cc) |
---|
| 349 | struct sockbuf *sb; |
---|
| 350 | u_long cc; |
---|
| 351 | { |
---|
| 352 | |
---|
| 353 | if (cc > sb_max * MCLBYTES / (MSIZE + MCLBYTES)) |
---|
| 354 | return (0); |
---|
| 355 | sb->sb_hiwat = cc; |
---|
| 356 | sb->sb_mbmax = min(cc * sb_efficiency, sb_max); |
---|
| 357 | if (sb->sb_lowat > sb->sb_hiwat) |
---|
| 358 | sb->sb_lowat = sb->sb_hiwat; |
---|
| 359 | return (1); |
---|
| 360 | } |
---|
| 361 | |
---|
| 362 | /* |
---|
| 363 | * Free mbufs held by a socket, and reserved mbuf space. |
---|
| 364 | */ |
---|
| 365 | void |
---|
| 366 | sbrelease(sb) |
---|
| 367 | struct sockbuf *sb; |
---|
| 368 | { |
---|
| 369 | |
---|
| 370 | sbflush(sb); |
---|
| 371 | sb->sb_hiwat = sb->sb_mbmax = 0; |
---|
| 372 | } |
---|
| 373 | |
---|
| 374 | /* |
---|
| 375 | * Routines to add and remove |
---|
| 376 | * data from an mbuf queue. |
---|
| 377 | * |
---|
| 378 | * The routines sbappend() or sbappendrecord() are normally called to |
---|
| 379 | * append new mbufs to a socket buffer, after checking that adequate |
---|
| 380 | * space is available, comparing the function sbspace() with the amount |
---|
| 381 | * of data to be added. sbappendrecord() differs from sbappend() in |
---|
| 382 | * that data supplied is treated as the beginning of a new record. |
---|
| 383 | * To place a sender's address, optional access rights, and data in a |
---|
| 384 | * socket receive buffer, sbappendaddr() should be used. To place |
---|
| 385 | * access rights and data in a socket receive buffer, sbappendrights() |
---|
| 386 | * should be used. In either case, the new data begins a new record. |
---|
| 387 | * Note that unlike sbappend() and sbappendrecord(), these routines check |
---|
| 388 | * for the caller that there will be enough space to store the data. |
---|
| 389 | * Each fails if there is not enough space, or if it cannot find mbufs |
---|
| 390 | * to store additional information in. |
---|
| 391 | * |
---|
| 392 | * Reliable protocols may use the socket send buffer to hold data |
---|
| 393 | * awaiting acknowledgement. Data is normally copied from a socket |
---|
| 394 | * send buffer in a protocol with m_copy for output to a peer, |
---|
| 395 | * and then removing the data from the socket buffer with sbdrop() |
---|
| 396 | * or sbdroprecord() when the data is acknowledged by the peer. |
---|
| 397 | */ |
---|
| 398 | |
---|
| 399 | /* |
---|
| 400 | * Append mbuf chain m to the last record in the |
---|
| 401 | * socket buffer sb. The additional space associated |
---|
| 402 | * the mbuf chain is recorded in sb. Empty mbufs are |
---|
| 403 | * discarded and mbufs are compacted where possible. |
---|
| 404 | */ |
---|
| 405 | void |
---|
| 406 | sbappend(sb, m) |
---|
| 407 | struct sockbuf *sb; |
---|
| 408 | struct mbuf *m; |
---|
| 409 | { |
---|
| 410 | register struct mbuf *n; |
---|
| 411 | |
---|
| 412 | if (m == 0) |
---|
| 413 | return; |
---|
| 414 | n = sb->sb_mb; |
---|
| 415 | if (n) { |
---|
| 416 | while (n->m_nextpkt) |
---|
| 417 | n = n->m_nextpkt; |
---|
| 418 | do { |
---|
| 419 | if (n->m_flags & M_EOR) { |
---|
| 420 | sbappendrecord(sb, m); /* XXXXXX!!!! */ |
---|
| 421 | return; |
---|
| 422 | } |
---|
| 423 | } while (n->m_next && (n = n->m_next)); |
---|
| 424 | } |
---|
| 425 | sbcompress(sb, m, n); |
---|
| 426 | } |
---|
| 427 | |
---|
| 428 | #ifdef SOCKBUF_DEBUG |
---|
| 429 | void |
---|
| 430 | sbcheck(sb) |
---|
| 431 | register struct sockbuf *sb; |
---|
| 432 | { |
---|
| 433 | register struct mbuf *m; |
---|
| 434 | register int len = 0, mbcnt = 0; |
---|
| 435 | |
---|
| 436 | for (m = sb->sb_mb; m; m = m->m_next) { |
---|
| 437 | len += m->m_len; |
---|
| 438 | mbcnt += MSIZE; |
---|
| 439 | if (m->m_flags & M_EXT) /*XXX*/ /* pretty sure this is bogus */ |
---|
| 440 | mbcnt += m->m_ext.ext_size; |
---|
| 441 | if (m->m_nextpkt) |
---|
| 442 | panic("sbcheck nextpkt"); |
---|
| 443 | } |
---|
| 444 | if (len != sb->sb_cc || mbcnt != sb->sb_mbcnt) { |
---|
| 445 | printf("cc %d != %d || mbcnt %d != %d\n", len, sb->sb_cc, |
---|
| 446 | mbcnt, sb->sb_mbcnt); |
---|
| 447 | panic("sbcheck"); |
---|
| 448 | } |
---|
| 449 | } |
---|
| 450 | #endif |
---|
| 451 | |
---|
| 452 | /* |
---|
| 453 | * As above, except the mbuf chain |
---|
| 454 | * begins a new record. |
---|
| 455 | */ |
---|
| 456 | void |
---|
| 457 | sbappendrecord(sb, m0) |
---|
| 458 | register struct sockbuf *sb; |
---|
| 459 | register struct mbuf *m0; |
---|
| 460 | { |
---|
| 461 | register struct mbuf *m; |
---|
| 462 | |
---|
| 463 | if (m0 == 0) |
---|
| 464 | return; |
---|
| 465 | m = sb->sb_mb; |
---|
| 466 | if (m) |
---|
| 467 | while (m->m_nextpkt) |
---|
| 468 | m = m->m_nextpkt; |
---|
| 469 | /* |
---|
| 470 | * Put the first mbuf on the queue. |
---|
| 471 | * Note this permits zero length records. |
---|
| 472 | */ |
---|
| 473 | sballoc(sb, m0); |
---|
| 474 | if (m) |
---|
| 475 | m->m_nextpkt = m0; |
---|
| 476 | else |
---|
| 477 | sb->sb_mb = m0; |
---|
| 478 | m = m0->m_next; |
---|
| 479 | m0->m_next = 0; |
---|
| 480 | if (m && (m0->m_flags & M_EOR)) { |
---|
| 481 | m0->m_flags &= ~M_EOR; |
---|
| 482 | m->m_flags |= M_EOR; |
---|
| 483 | } |
---|
| 484 | sbcompress(sb, m, m0); |
---|
| 485 | } |
---|
| 486 | |
---|
| 487 | /* |
---|
| 488 | * As above except that OOB data |
---|
| 489 | * is inserted at the beginning of the sockbuf, |
---|
| 490 | * but after any other OOB data. |
---|
| 491 | */ |
---|
| 492 | void |
---|
| 493 | sbinsertoob(sb, m0) |
---|
| 494 | register struct sockbuf *sb; |
---|
| 495 | register struct mbuf *m0; |
---|
| 496 | { |
---|
| 497 | register struct mbuf *m; |
---|
| 498 | register struct mbuf **mp; |
---|
| 499 | |
---|
| 500 | if (m0 == 0) |
---|
| 501 | return; |
---|
| 502 | for (mp = &sb->sb_mb; *mp ; mp = &((*mp)->m_nextpkt)) { |
---|
| 503 | m = *mp; |
---|
| 504 | again: |
---|
| 505 | switch (m->m_type) { |
---|
| 506 | |
---|
| 507 | case MT_OOBDATA: |
---|
| 508 | continue; /* WANT next train */ |
---|
| 509 | |
---|
| 510 | case MT_CONTROL: |
---|
| 511 | m = m->m_next; |
---|
| 512 | if (m) |
---|
| 513 | goto again; /* inspect THIS train further */ |
---|
| 514 | } |
---|
| 515 | break; |
---|
| 516 | } |
---|
| 517 | /* |
---|
| 518 | * Put the first mbuf on the queue. |
---|
| 519 | * Note this permits zero length records. |
---|
| 520 | */ |
---|
| 521 | sballoc(sb, m0); |
---|
| 522 | m0->m_nextpkt = *mp; |
---|
| 523 | *mp = m0; |
---|
| 524 | m = m0->m_next; |
---|
| 525 | m0->m_next = 0; |
---|
| 526 | if (m && (m0->m_flags & M_EOR)) { |
---|
| 527 | m0->m_flags &= ~M_EOR; |
---|
| 528 | m->m_flags |= M_EOR; |
---|
| 529 | } |
---|
| 530 | sbcompress(sb, m, m0); |
---|
| 531 | } |
---|
| 532 | |
---|
| 533 | /* |
---|
| 534 | * Append address and data, and optionally, control (ancillary) data |
---|
| 535 | * to the receive queue of a socket. If present, |
---|
| 536 | * m0 must include a packet header with total length. |
---|
| 537 | * Returns 0 if no space in sockbuf or insufficient mbufs. |
---|
| 538 | */ |
---|
| 539 | int |
---|
| 540 | sbappendaddr(sb, asa, m0, control) |
---|
| 541 | register struct sockbuf *sb; |
---|
| 542 | struct sockaddr *asa; |
---|
| 543 | struct mbuf *m0, *control; |
---|
| 544 | { |
---|
| 545 | register struct mbuf *m, *n; |
---|
| 546 | int space = asa->sa_len; |
---|
| 547 | |
---|
| 548 | if (m0 && (m0->m_flags & M_PKTHDR) == 0) |
---|
| 549 | panic("sbappendaddr"); |
---|
| 550 | if (m0) |
---|
| 551 | space += m0->m_pkthdr.len; |
---|
| 552 | for (n = control; n; n = n->m_next) { |
---|
| 553 | space += n->m_len; |
---|
| 554 | if (n->m_next == 0) /* keep pointer to last control buf */ |
---|
| 555 | break; |
---|
| 556 | } |
---|
| 557 | if (space > sbspace(sb)) |
---|
| 558 | return (0); |
---|
| 559 | if (asa->sa_len > MLEN) |
---|
| 560 | return (0); |
---|
| 561 | MGET(m, M_DONTWAIT, MT_SONAME); |
---|
| 562 | if (m == 0) |
---|
| 563 | return (0); |
---|
| 564 | m->m_len = asa->sa_len; |
---|
| 565 | bcopy((caddr_t)asa, mtod(m, caddr_t), asa->sa_len); |
---|
| 566 | if (n) |
---|
| 567 | n->m_next = m0; /* concatenate data to control */ |
---|
| 568 | else |
---|
| 569 | control = m0; |
---|
| 570 | m->m_next = control; |
---|
| 571 | for (n = m; n; n = n->m_next) |
---|
| 572 | sballoc(sb, n); |
---|
| 573 | n = sb->sb_mb; |
---|
| 574 | if (n) { |
---|
| 575 | while (n->m_nextpkt) |
---|
| 576 | n = n->m_nextpkt; |
---|
| 577 | n->m_nextpkt = m; |
---|
| 578 | } else |
---|
| 579 | sb->sb_mb = m; |
---|
| 580 | return (1); |
---|
| 581 | } |
---|
| 582 | |
---|
| 583 | int |
---|
| 584 | sbappendcontrol(sb, m0, control) |
---|
| 585 | struct sockbuf *sb; |
---|
| 586 | struct mbuf *control, *m0; |
---|
| 587 | { |
---|
| 588 | register struct mbuf *m, *n; |
---|
| 589 | int space = 0; |
---|
| 590 | |
---|
| 591 | if (control == 0) |
---|
| 592 | panic("sbappendcontrol"); |
---|
| 593 | for (m = control; ; m = m->m_next) { |
---|
| 594 | space += m->m_len; |
---|
| 595 | if (m->m_next == 0) |
---|
| 596 | break; |
---|
| 597 | } |
---|
| 598 | n = m; /* save pointer to last control buffer */ |
---|
| 599 | for (m = m0; m; m = m->m_next) |
---|
| 600 | space += m->m_len; |
---|
| 601 | if (space > sbspace(sb)) |
---|
| 602 | return (0); |
---|
| 603 | n->m_next = m0; /* concatenate data to control */ |
---|
| 604 | for (m = control; m; m = m->m_next) |
---|
| 605 | sballoc(sb, m); |
---|
| 606 | n = sb->sb_mb; |
---|
| 607 | if (n) { |
---|
| 608 | while (n->m_nextpkt) |
---|
| 609 | n = n->m_nextpkt; |
---|
| 610 | n->m_nextpkt = control; |
---|
| 611 | } else |
---|
| 612 | sb->sb_mb = control; |
---|
| 613 | return (1); |
---|
| 614 | } |
---|
| 615 | |
---|
| 616 | /* |
---|
| 617 | * Compress mbuf chain m into the socket |
---|
| 618 | * buffer sb following mbuf n. If n |
---|
| 619 | * is null, the buffer is presumed empty. |
---|
| 620 | */ |
---|
| 621 | void |
---|
| 622 | sbcompress(sb, m, n) |
---|
| 623 | register struct sockbuf *sb; |
---|
| 624 | register struct mbuf *m, *n; |
---|
| 625 | { |
---|
| 626 | register int eor = 0; |
---|
| 627 | register struct mbuf *o; |
---|
| 628 | |
---|
| 629 | while (m) { |
---|
| 630 | eor |= m->m_flags & M_EOR; |
---|
| 631 | if (m->m_len == 0 && |
---|
| 632 | (eor == 0 || |
---|
| 633 | (((o = m->m_next) || (o = n)) && |
---|
| 634 | o->m_type == m->m_type))) { |
---|
| 635 | m = m_free(m); |
---|
| 636 | continue; |
---|
| 637 | } |
---|
| 638 | if (n && (n->m_flags & (M_EXT | M_EOR)) == 0 && |
---|
| 639 | (n->m_data + n->m_len + m->m_len) < &n->m_dat[MLEN] && |
---|
| 640 | n->m_type == m->m_type) { |
---|
| 641 | bcopy(mtod(m, caddr_t), mtod(n, caddr_t) + n->m_len, |
---|
| 642 | (unsigned)m->m_len); |
---|
| 643 | n->m_len += m->m_len; |
---|
| 644 | sb->sb_cc += m->m_len; |
---|
| 645 | m = m_free(m); |
---|
| 646 | continue; |
---|
| 647 | } |
---|
| 648 | if (n) |
---|
| 649 | n->m_next = m; |
---|
| 650 | else |
---|
| 651 | sb->sb_mb = m; |
---|
| 652 | sballoc(sb, m); |
---|
| 653 | n = m; |
---|
| 654 | m->m_flags &= ~M_EOR; |
---|
| 655 | m = m->m_next; |
---|
| 656 | n->m_next = 0; |
---|
| 657 | } |
---|
| 658 | if (eor) { |
---|
| 659 | if (n) |
---|
| 660 | n->m_flags |= eor; |
---|
| 661 | else |
---|
| 662 | printf("semi-panic: sbcompress\n"); |
---|
| 663 | } |
---|
| 664 | } |
---|
| 665 | |
---|
| 666 | /* |
---|
| 667 | * Free all mbufs in a sockbuf. |
---|
| 668 | * Check that all resources are reclaimed. |
---|
| 669 | */ |
---|
| 670 | void |
---|
| 671 | sbflush(sb) |
---|
| 672 | register struct sockbuf *sb; |
---|
| 673 | { |
---|
| 674 | |
---|
| 675 | if (sb->sb_flags & SB_LOCK) |
---|
| 676 | panic("sbflush"); |
---|
| 677 | while (sb->sb_mbcnt) |
---|
| 678 | sbdrop(sb, (int)sb->sb_cc); |
---|
| 679 | if (sb->sb_cc || sb->sb_mb) |
---|
| 680 | panic("sbflush 2"); |
---|
| 681 | } |
---|
| 682 | |
---|
| 683 | /* |
---|
| 684 | * Drop data from (the front of) a sockbuf. |
---|
| 685 | */ |
---|
| 686 | void |
---|
| 687 | sbdrop(sb, len) |
---|
| 688 | register struct sockbuf *sb; |
---|
| 689 | register int len; |
---|
| 690 | { |
---|
| 691 | register struct mbuf *m, *mn; |
---|
| 692 | struct mbuf *next; |
---|
| 693 | |
---|
| 694 | next = (m = sb->sb_mb) ? m->m_nextpkt : 0; |
---|
| 695 | while (len > 0) { |
---|
| 696 | if (m == 0) { |
---|
| 697 | if (next == 0) |
---|
| 698 | panic("sbdrop"); |
---|
| 699 | m = next; |
---|
| 700 | next = m->m_nextpkt; |
---|
| 701 | continue; |
---|
| 702 | } |
---|
| 703 | if (m->m_len > len) { |
---|
| 704 | m->m_len -= len; |
---|
| 705 | m->m_data += len; |
---|
| 706 | sb->sb_cc -= len; |
---|
| 707 | break; |
---|
| 708 | } |
---|
| 709 | len -= m->m_len; |
---|
| 710 | sbfree(sb, m); |
---|
| 711 | MFREE(m, mn); |
---|
| 712 | m = mn; |
---|
| 713 | } |
---|
| 714 | while (m && m->m_len == 0) { |
---|
| 715 | sbfree(sb, m); |
---|
| 716 | MFREE(m, mn); |
---|
| 717 | m = mn; |
---|
| 718 | } |
---|
| 719 | if (m) { |
---|
| 720 | sb->sb_mb = m; |
---|
| 721 | m->m_nextpkt = next; |
---|
| 722 | } else |
---|
| 723 | sb->sb_mb = next; |
---|
| 724 | } |
---|
| 725 | |
---|
| 726 | /* |
---|
| 727 | * Drop a record off the front of a sockbuf |
---|
| 728 | * and move the next record to the front. |
---|
| 729 | */ |
---|
| 730 | void |
---|
| 731 | sbdroprecord(sb) |
---|
| 732 | register struct sockbuf *sb; |
---|
| 733 | { |
---|
| 734 | register struct mbuf *m, *mn; |
---|
| 735 | |
---|
| 736 | m = sb->sb_mb; |
---|
| 737 | if (m) { |
---|
| 738 | sb->sb_mb = m->m_nextpkt; |
---|
| 739 | do { |
---|
| 740 | sbfree(sb, m); |
---|
| 741 | MFREE(m, mn); |
---|
| 742 | m = mn; |
---|
| 743 | } while (m); |
---|
| 744 | } |
---|
| 745 | } |
---|
| 746 | |
---|
| 747 | /* |
---|
| 748 | * Create a "control" mbuf containing the specified data |
---|
| 749 | * with the specified type for presentation on a socket buffer. |
---|
| 750 | */ |
---|
| 751 | struct mbuf * |
---|
| 752 | sbcreatecontrol(p, size, type, level) |
---|
| 753 | caddr_t p; |
---|
| 754 | register int size; |
---|
| 755 | int type, level; |
---|
| 756 | { |
---|
| 757 | register struct cmsghdr *cp; |
---|
| 758 | struct mbuf *m; |
---|
| 759 | |
---|
| 760 | if ((m = m_get(M_DONTWAIT, MT_CONTROL)) == NULL) |
---|
| 761 | return ((struct mbuf *) NULL); |
---|
| 762 | cp = mtod(m, struct cmsghdr *); |
---|
| 763 | /* XXX check size? */ |
---|
| 764 | (void)memcpy(CMSG_DATA(cp), p, size); |
---|
| 765 | size += sizeof(*cp); |
---|
| 766 | m->m_len = size; |
---|
| 767 | cp->cmsg_len = size; |
---|
| 768 | cp->cmsg_level = level; |
---|
| 769 | cp->cmsg_type = type; |
---|
| 770 | return (m); |
---|
| 771 | } |
---|
| 772 | |
---|
| 773 | #ifdef PRU_OLDSTYLE |
---|
| 774 | /* |
---|
| 775 | * The following routines mediate between the old-style `pr_usrreq' |
---|
| 776 | * protocol implementations and the new-style `struct pr_usrreqs' |
---|
| 777 | * calling convention. |
---|
| 778 | */ |
---|
| 779 | |
---|
| 780 | /* syntactic sugar */ |
---|
| 781 | #define nomb (struct mbuf *)0 |
---|
| 782 | |
---|
| 783 | static int |
---|
| 784 | old_abort(struct socket *so) |
---|
| 785 | { |
---|
| 786 | return so->so_proto->pr_ousrreq(so, PRU_ABORT, nomb, nomb, nomb); |
---|
| 787 | } |
---|
| 788 | |
---|
| 789 | static int |
---|
| 790 | old_accept(struct socket *so, struct mbuf *nam) |
---|
| 791 | { |
---|
| 792 | return so->so_proto->pr_ousrreq(so, PRU_ACCEPT, nomb, nam, nomb); |
---|
| 793 | } |
---|
| 794 | |
---|
| 795 | static int |
---|
[d506dff] | 796 | old_attach(struct socket *so, intptr_t proto) |
---|
[39e6e65a] | 797 | { |
---|
| 798 | return so->so_proto->pr_ousrreq(so, PRU_ATTACH, nomb, |
---|
| 799 | (struct mbuf *)proto, /* XXX */ |
---|
| 800 | nomb); |
---|
| 801 | } |
---|
| 802 | |
---|
| 803 | static int |
---|
| 804 | old_bind(struct socket *so, struct mbuf *nam) |
---|
| 805 | { |
---|
| 806 | return so->so_proto->pr_ousrreq(so, PRU_BIND, nomb, nam, nomb); |
---|
| 807 | } |
---|
| 808 | |
---|
| 809 | static int |
---|
| 810 | old_connect(struct socket *so, struct mbuf *nam) |
---|
| 811 | { |
---|
| 812 | return so->so_proto->pr_ousrreq(so, PRU_CONNECT, nomb, nam, nomb); |
---|
| 813 | } |
---|
| 814 | |
---|
| 815 | static int |
---|
| 816 | old_connect2(struct socket *so1, struct socket *so2) |
---|
| 817 | { |
---|
| 818 | return so1->so_proto->pr_ousrreq(so1, PRU_CONNECT2, nomb, |
---|
| 819 | (struct mbuf *)so2, nomb); |
---|
| 820 | } |
---|
| 821 | |
---|
| 822 | static int |
---|
[d506dff] | 823 | old_control(struct socket *so, intptr_t cmd, caddr_t data, struct ifnet *ifp) |
---|
[39e6e65a] | 824 | { |
---|
| 825 | return so->so_proto->pr_ousrreq(so, PRU_CONTROL, (struct mbuf *)cmd, |
---|
| 826 | (struct mbuf *)data, |
---|
| 827 | (struct mbuf *)ifp); |
---|
| 828 | } |
---|
| 829 | |
---|
| 830 | static int |
---|
| 831 | old_detach(struct socket *so) |
---|
| 832 | { |
---|
| 833 | return so->so_proto->pr_ousrreq(so, PRU_DETACH, nomb, nomb, nomb); |
---|
| 834 | } |
---|
| 835 | |
---|
| 836 | static int |
---|
| 837 | old_disconnect(struct socket *so) |
---|
| 838 | { |
---|
| 839 | return so->so_proto->pr_ousrreq(so, PRU_DISCONNECT, nomb, nomb, nomb); |
---|
| 840 | } |
---|
| 841 | |
---|
| 842 | static int |
---|
| 843 | old_listen(struct socket *so) |
---|
| 844 | { |
---|
| 845 | return so->so_proto->pr_ousrreq(so, PRU_LISTEN, nomb, nomb, nomb); |
---|
| 846 | } |
---|
| 847 | |
---|
| 848 | static int |
---|
| 849 | old_peeraddr(struct socket *so, struct mbuf *nam) |
---|
| 850 | { |
---|
| 851 | return so->so_proto->pr_ousrreq(so, PRU_PEERADDR, nomb, nam, nomb); |
---|
| 852 | } |
---|
| 853 | |
---|
| 854 | static int |
---|
[d506dff] | 855 | old_rcvd(struct socket *so, intptr_t flags) |
---|
[39e6e65a] | 856 | { |
---|
| 857 | return so->so_proto->pr_ousrreq(so, PRU_RCVD, nomb, |
---|
| 858 | (struct mbuf *)flags, /* XXX */ |
---|
| 859 | nomb); |
---|
| 860 | } |
---|
| 861 | |
---|
| 862 | static int |
---|
[d506dff] | 863 | old_rcvoob(struct socket *so, struct mbuf *m, intptr_t flags) |
---|
[39e6e65a] | 864 | { |
---|
| 865 | return so->so_proto->pr_ousrreq(so, PRU_RCVOOB, m, |
---|
| 866 | (struct mbuf *)flags, /* XXX */ |
---|
| 867 | nomb); |
---|
| 868 | } |
---|
| 869 | |
---|
| 870 | static int |
---|
| 871 | old_send(struct socket *so, int flags, struct mbuf *m, struct mbuf *addr, |
---|
| 872 | struct mbuf *control) |
---|
| 873 | { |
---|
| 874 | int req; |
---|
| 875 | |
---|
| 876 | if (flags & PRUS_OOB) { |
---|
| 877 | req = PRU_SENDOOB; |
---|
| 878 | } else if(flags & PRUS_EOF) { |
---|
| 879 | req = PRU_SEND_EOF; |
---|
| 880 | } else { |
---|
| 881 | req = PRU_SEND; |
---|
| 882 | } |
---|
| 883 | return so->so_proto->pr_ousrreq(so, req, m, addr, control); |
---|
| 884 | } |
---|
| 885 | |
---|
| 886 | static int |
---|
| 887 | old_sense(struct socket *so, struct stat *sb) |
---|
| 888 | { |
---|
| 889 | return so->so_proto->pr_ousrreq(so, PRU_SENSE, (struct mbuf *)sb, |
---|
| 890 | nomb, nomb); |
---|
| 891 | } |
---|
| 892 | |
---|
| 893 | static int |
---|
| 894 | old_shutdown(struct socket *so) |
---|
| 895 | { |
---|
| 896 | return so->so_proto->pr_ousrreq(so, PRU_SHUTDOWN, nomb, nomb, nomb); |
---|
| 897 | } |
---|
| 898 | |
---|
| 899 | static int |
---|
| 900 | old_sockaddr(struct socket *so, struct mbuf *nam) |
---|
| 901 | { |
---|
| 902 | return so->so_proto->pr_ousrreq(so, PRU_SOCKADDR, nomb, nam, nomb); |
---|
| 903 | } |
---|
| 904 | |
---|
| 905 | struct pr_usrreqs pru_oldstyle = { |
---|
| 906 | old_abort, old_accept, old_attach, old_bind, old_connect, |
---|
| 907 | old_connect2, old_control, old_detach, old_disconnect, |
---|
| 908 | old_listen, old_peeraddr, old_rcvd, old_rcvoob, old_send, |
---|
| 909 | old_sense, old_shutdown, old_sockaddr |
---|
| 910 | }; |
---|
| 911 | |
---|
| 912 | #endif /* PRU_OLDSTYLE */ |
---|
| 913 | |
---|
| 914 | /* |
---|
| 915 | * Some routines that return EOPNOTSUPP for entry points that are not |
---|
| 916 | * supported by a protocol. Fill in as needed. |
---|
| 917 | */ |
---|
| 918 | int |
---|
| 919 | pru_accept_notsupp(struct socket *so, struct mbuf *nam) |
---|
| 920 | { |
---|
| 921 | return EOPNOTSUPP; |
---|
| 922 | } |
---|
| 923 | |
---|
| 924 | int |
---|
| 925 | pru_connect2_notsupp(struct socket *so1, struct socket *so2) |
---|
| 926 | { |
---|
| 927 | return EOPNOTSUPP; |
---|
| 928 | } |
---|
| 929 | |
---|
| 930 | int |
---|
| 931 | pru_control_notsupp(struct socket *so, int cmd, caddr_t data, |
---|
| 932 | struct ifnet *ifp) |
---|
| 933 | { |
---|
| 934 | return EOPNOTSUPP; |
---|
| 935 | } |
---|
| 936 | |
---|
| 937 | int |
---|
| 938 | pru_listen_notsupp(struct socket *so) |
---|
| 939 | { |
---|
| 940 | return EOPNOTSUPP; |
---|
| 941 | } |
---|
| 942 | |
---|
| 943 | int |
---|
| 944 | pru_rcvd_notsupp(struct socket *so, int flags) |
---|
| 945 | { |
---|
| 946 | return EOPNOTSUPP; |
---|
| 947 | } |
---|
| 948 | |
---|
| 949 | int |
---|
| 950 | pru_rcvoob_notsupp(struct socket *so, struct mbuf *m, int flags) |
---|
| 951 | { |
---|
| 952 | return EOPNOTSUPP; |
---|
| 953 | } |
---|
| 954 | |
---|
| 955 | /* |
---|
| 956 | * This isn't really a ``null'' operation, but it's the default one |
---|
| 957 | * and doesn't do anything destructive. |
---|
| 958 | */ |
---|
| 959 | int |
---|
| 960 | pru_sense_null(struct socket *so, struct stat *sb) |
---|
| 961 | { |
---|
| 962 | sb->st_blksize = so->so_snd.sb_hiwat; |
---|
| 963 | return 0; |
---|
| 964 | } |
---|