Annotation of src/sys/kern/uipc_socket.c, Revision 1.42
1.42 ! mycroft 1: /* $NetBSD: uipc_socket.c,v 1.41 1999/01/20 09:15:41 mycroft Exp $ */
1.16 cgd 2:
1.1 cgd 3: /*
1.15 mycroft 4: * Copyright (c) 1982, 1986, 1988, 1990, 1993
5: * The Regents of the University of California. All rights reserved.
1.1 cgd 6: *
7: * Redistribution and use in source and binary forms, with or without
8: * modification, are permitted provided that the following conditions
9: * are met:
10: * 1. Redistributions of source code must retain the above copyright
11: * notice, this list of conditions and the following disclaimer.
12: * 2. Redistributions in binary form must reproduce the above copyright
13: * notice, this list of conditions and the following disclaimer in the
14: * documentation and/or other materials provided with the distribution.
15: * 3. All advertising materials mentioning features or use of this software
16: * must display the following acknowledgement:
17: * This product includes software developed by the University of
18: * California, Berkeley and its contributors.
19: * 4. Neither the name of the University nor the names of its contributors
20: * may be used to endorse or promote products derived from this software
21: * without specific prior written permission.
22: *
23: * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24: * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25: * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26: * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27: * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28: * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29: * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30: * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31: * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32: * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33: * SUCH DAMAGE.
34: *
1.32 fvdl 35: * @(#)uipc_socket.c 8.6 (Berkeley) 5/2/95
1.1 cgd 36: */
1.35 thorpej 37:
38: #include "opt_compat_sunos.h"
1.1 cgd 39:
1.9 mycroft 40: #include <sys/param.h>
41: #include <sys/systm.h>
42: #include <sys/proc.h>
43: #include <sys/file.h>
44: #include <sys/malloc.h>
45: #include <sys/mbuf.h>
46: #include <sys/domain.h>
47: #include <sys/kernel.h>
48: #include <sys/protosw.h>
49: #include <sys/socket.h>
50: #include <sys/socketvar.h>
1.21 christos 51: #include <sys/signalvar.h>
1.9 mycroft 52: #include <sys/resourcevar.h>
1.37 thorpej 53: #include <sys/pool.h>
54:
55: struct pool socket_pool;
56:
57: void
58: soinit()
59: {
60:
61: pool_init(&socket_pool, sizeof(struct socket), 0, 0, 0,
62: "sockpl", 0, NULL, NULL, M_SOCKET);
63: }
1.1 cgd 64:
65: /*
66: * Socket operation routines.
67: * These routines are called by the routines in
68: * sys_socket.c or from a system process, and
69: * implement the semantics of socket operations by
70: * switching out to the protocol specific routines.
71: */
72: /*ARGSUSED*/
1.3 andrew 73: int
1.1 cgd 74: socreate(dom, aso, type, proto)
1.11 mycroft 75: int dom;
1.1 cgd 76: struct socket **aso;
77: register int type;
78: int proto;
79: {
80: struct proc *p = curproc; /* XXX */
81: register struct protosw *prp;
82: register struct socket *so;
83: register int error;
1.39 matt 84: int s;
1.1 cgd 85:
86: if (proto)
87: prp = pffindproto(dom, proto, type);
88: else
89: prp = pffindtype(dom, type);
1.15 mycroft 90: if (prp == 0 || prp->pr_usrreq == 0)
1.1 cgd 91: return (EPROTONOSUPPORT);
92: if (prp->pr_type != type)
93: return (EPROTOTYPE);
1.39 matt 94: s = splsoftnet();
1.37 thorpej 95: so = pool_get(&socket_pool, PR_WAITOK);
1.38 perry 96: memset((caddr_t)so, 0, sizeof(*so));
1.31 thorpej 97: TAILQ_INIT(&so->so_q0);
98: TAILQ_INIT(&so->so_q);
1.1 cgd 99: so->so_type = type;
100: so->so_proto = prp;
1.33 matt 101: so->so_send = sosend;
102: so->so_receive = soreceive;
1.22 mycroft 103: error = (*prp->pr_usrreq)(so, PRU_ATTACH, (struct mbuf *)0,
104: (struct mbuf *)(long)proto, (struct mbuf *)0, p);
1.1 cgd 105: if (error) {
106: so->so_state |= SS_NOFDREF;
107: sofree(so);
1.39 matt 108: splx(s);
1.1 cgd 109: return (error);
110: }
1.10 deraadt 111: #ifdef COMPAT_SUNOS
1.18 christos 112: {
113: extern struct emul emul_sunos;
114: if (p->p_emul == &emul_sunos && type == SOCK_DGRAM)
115: so->so_options |= SO_BROADCAST;
116: }
1.10 deraadt 117: #endif
1.39 matt 118: splx(s);
1.1 cgd 119: *aso = so;
120: return (0);
121: }
122:
1.3 andrew 123: int
1.1 cgd 124: sobind(so, nam)
125: struct socket *so;
126: struct mbuf *nam;
127: {
1.22 mycroft 128: struct proc *p = curproc; /* XXX */
1.20 mycroft 129: int s = splsoftnet();
1.1 cgd 130: int error;
131:
1.22 mycroft 132: error = (*so->so_proto->pr_usrreq)(so, PRU_BIND, (struct mbuf *)0,
133: nam, (struct mbuf *)0, p);
1.1 cgd 134: splx(s);
135: return (error);
136: }
137:
1.3 andrew 138: int
1.1 cgd 139: solisten(so, backlog)
140: register struct socket *so;
141: int backlog;
142: {
1.20 mycroft 143: int s = splsoftnet(), error;
1.1 cgd 144:
1.22 mycroft 145: error = (*so->so_proto->pr_usrreq)(so, PRU_LISTEN, (struct mbuf *)0,
146: (struct mbuf *)0, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 147: if (error) {
148: splx(s);
149: return (error);
150: }
1.31 thorpej 151: if (so->so_q.tqh_first == NULL)
1.1 cgd 152: so->so_options |= SO_ACCEPTCONN;
153: if (backlog < 0)
154: backlog = 0;
155: so->so_qlimit = min(backlog, SOMAXCONN);
156: splx(s);
157: return (0);
158: }
159:
1.21 christos 160: void
1.1 cgd 161: sofree(so)
162: register struct socket *so;
163: {
164:
1.41 mycroft 165: /*
166: * We must not decommission a socket that's on the accept(2) queue.
167: * If we do, then accept(2) may hang even after select(2) indicated
168: * that the listening socket was ready.
169: */
170: if (so->so_pcb || so->so_head || (so->so_state & SS_NOFDREF) == 0)
1.1 cgd 171: return;
172: sbrelease(&so->so_snd);
173: sorflush(so);
1.37 thorpej 174: pool_put(&socket_pool, so);
1.1 cgd 175: }
176:
177: /*
178: * Close a socket on last file table reference removal.
179: * Initiate disconnect if connected.
180: * Free socket when disconnect complete.
181: */
1.3 andrew 182: int
1.1 cgd 183: soclose(so)
184: register struct socket *so;
185: {
1.41 mycroft 186: struct socket *so2;
1.20 mycroft 187: int s = splsoftnet(); /* conservative */
1.1 cgd 188: int error = 0;
189:
190: if (so->so_options & SO_ACCEPTCONN) {
1.41 mycroft 191: while ((so2 = so->so_q0.tqh_first) != 0) {
1.42 ! mycroft 192: (void) soqremque(so2, 0);
1.41 mycroft 193: (void) soabort(so2);
194: }
195: while ((so2 = so->so_q.tqh_first) != 0) {
1.42 ! mycroft 196: (void) soqremque(so2, 1);
1.41 mycroft 197: (void) soabort(so2);
198: }
1.1 cgd 199: }
200: if (so->so_pcb == 0)
201: goto discard;
202: if (so->so_state & SS_ISCONNECTED) {
203: if ((so->so_state & SS_ISDISCONNECTING) == 0) {
204: error = sodisconnect(so);
205: if (error)
206: goto drop;
207: }
208: if (so->so_options & SO_LINGER) {
209: if ((so->so_state & SS_ISDISCONNECTING) &&
210: (so->so_state & SS_NBIO))
211: goto drop;
1.21 christos 212: while (so->so_state & SS_ISCONNECTED) {
213: error = tsleep((caddr_t)&so->so_timeo,
214: PSOCK | PCATCH, netcls,
1.30 thorpej 215: so->so_linger * hz);
1.21 christos 216: if (error)
1.1 cgd 217: break;
1.21 christos 218: }
1.1 cgd 219: }
220: }
221: drop:
222: if (so->so_pcb) {
1.22 mycroft 223: int error2 = (*so->so_proto->pr_usrreq)(so, PRU_DETACH,
224: (struct mbuf *)0, (struct mbuf *)0, (struct mbuf *)0,
225: (struct proc *)0);
1.1 cgd 226: if (error == 0)
227: error = error2;
228: }
229: discard:
230: if (so->so_state & SS_NOFDREF)
231: panic("soclose: NOFDREF");
232: so->so_state |= SS_NOFDREF;
233: sofree(so);
234: splx(s);
235: return (error);
236: }
237:
238: /*
1.20 mycroft 239: * Must be called at splsoftnet...
1.1 cgd 240: */
1.3 andrew 241: int
1.1 cgd 242: soabort(so)
243: struct socket *so;
244: {
245:
1.22 mycroft 246: return (*so->so_proto->pr_usrreq)(so, PRU_ABORT, (struct mbuf *)0,
247: (struct mbuf *)0, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 248: }
249:
1.3 andrew 250: int
1.1 cgd 251: soaccept(so, nam)
252: register struct socket *so;
253: struct mbuf *nam;
254: {
1.20 mycroft 255: int s = splsoftnet();
1.1 cgd 256: int error;
257:
258: if ((so->so_state & SS_NOFDREF) == 0)
259: panic("soaccept: !NOFDREF");
260: so->so_state &= ~SS_NOFDREF;
1.41 mycroft 261: if ((so->so_state & SS_ISDISCONNECTED) == 0)
262: error = (*so->so_proto->pr_usrreq)(so, PRU_ACCEPT,
263: (struct mbuf *)0, nam, (struct mbuf *)0, (struct proc *)0);
264: else
265: error = 0;
1.1 cgd 266: splx(s);
267: return (error);
268: }
269:
1.3 andrew 270: int
1.1 cgd 271: soconnect(so, nam)
272: register struct socket *so;
273: struct mbuf *nam;
274: {
1.23 mycroft 275: struct proc *p = curproc; /* XXX */
1.1 cgd 276: int s;
277: int error;
278:
279: if (so->so_options & SO_ACCEPTCONN)
280: return (EOPNOTSUPP);
1.20 mycroft 281: s = splsoftnet();
1.1 cgd 282: /*
283: * If protocol is connection-based, can only connect once.
284: * Otherwise, if connected, try to disconnect first.
285: * This allows user to disconnect by connecting to, e.g.,
286: * a null address.
287: */
288: if (so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING) &&
289: ((so->so_proto->pr_flags & PR_CONNREQUIRED) ||
290: (error = sodisconnect(so))))
291: error = EISCONN;
292: else
293: error = (*so->so_proto->pr_usrreq)(so, PRU_CONNECT,
1.23 mycroft 294: (struct mbuf *)0, nam, (struct mbuf *)0, p);
1.1 cgd 295: splx(s);
296: return (error);
297: }
298:
1.3 andrew 299: int
1.1 cgd 300: soconnect2(so1, so2)
301: register struct socket *so1;
302: struct socket *so2;
303: {
1.20 mycroft 304: int s = splsoftnet();
1.1 cgd 305: int error;
306:
1.22 mycroft 307: error = (*so1->so_proto->pr_usrreq)(so1, PRU_CONNECT2,
308: (struct mbuf *)0, (struct mbuf *)so2, (struct mbuf *)0,
309: (struct proc *)0);
1.1 cgd 310: splx(s);
311: return (error);
312: }
313:
1.3 andrew 314: int
1.1 cgd 315: sodisconnect(so)
316: register struct socket *so;
317: {
1.20 mycroft 318: int s = splsoftnet();
1.1 cgd 319: int error;
320:
321: if ((so->so_state & SS_ISCONNECTED) == 0) {
322: error = ENOTCONN;
323: goto bad;
324: }
325: if (so->so_state & SS_ISDISCONNECTING) {
326: error = EALREADY;
327: goto bad;
328: }
1.22 mycroft 329: error = (*so->so_proto->pr_usrreq)(so, PRU_DISCONNECT,
330: (struct mbuf *)0, (struct mbuf *)0, (struct mbuf *)0,
331: (struct proc *)0);
1.1 cgd 332: bad:
333: splx(s);
334: return (error);
335: }
336:
1.15 mycroft 337: #define SBLOCKWAIT(f) (((f) & MSG_DONTWAIT) ? M_NOWAIT : M_WAITOK)
1.1 cgd 338: /*
339: * Send on a socket.
340: * If send must go all at once and message is larger than
341: * send buffering, then hard error.
342: * Lock against other senders.
343: * If must go all at once and not enough room now, then
344: * inform user that this would block and do nothing.
345: * Otherwise, if nonblocking, send as much as possible.
346: * The data to be sent is described by "uio" if nonzero,
347: * otherwise by the mbuf chain "top" (which must be null
348: * if uio is not). Data provided in mbuf chain must be small
349: * enough to send all at once.
350: *
351: * Returns nonzero on error, timeout or signal; callers
352: * must check for short counts if EINTR/ERESTART are returned.
353: * Data and control buffers are freed on return.
354: */
1.3 andrew 355: int
1.1 cgd 356: sosend(so, addr, uio, top, control, flags)
357: register struct socket *so;
358: struct mbuf *addr;
359: struct uio *uio;
360: struct mbuf *top;
361: struct mbuf *control;
362: int flags;
363: {
1.15 mycroft 364: struct proc *p = curproc; /* XXX */
1.1 cgd 365: struct mbuf **mp;
366: register struct mbuf *m;
367: register long space, len, resid;
368: int clen = 0, error, s, dontroute, mlen;
369: int atomic = sosendallatonce(so) || top;
370:
371: if (uio)
372: resid = uio->uio_resid;
373: else
374: resid = top->m_pkthdr.len;
1.7 cgd 375: /*
376: * In theory resid should be unsigned.
377: * However, space must be signed, as it might be less than 0
378: * if we over-committed, and we must use a signed comparison
379: * of space and resid. On the other hand, a negative resid
380: * causes us to loop sending 0-length segments to the protocol.
381: */
1.29 mycroft 382: if (resid < 0) {
383: error = EINVAL;
384: goto out;
385: }
1.1 cgd 386: dontroute =
387: (flags & MSG_DONTROUTE) && (so->so_options & SO_DONTROUTE) == 0 &&
388: (so->so_proto->pr_flags & PR_ATOMIC);
1.12 mycroft 389: p->p_stats->p_ru.ru_msgsnd++;
1.1 cgd 390: if (control)
391: clen = control->m_len;
392: #define snderr(errno) { error = errno; splx(s); goto release; }
393:
394: restart:
1.21 christos 395: if ((error = sblock(&so->so_snd, SBLOCKWAIT(flags))) != 0)
1.1 cgd 396: goto out;
397: do {
1.20 mycroft 398: s = splsoftnet();
1.1 cgd 399: if (so->so_state & SS_CANTSENDMORE)
400: snderr(EPIPE);
401: if (so->so_error)
402: snderr(so->so_error);
403: if ((so->so_state & SS_ISCONNECTED) == 0) {
404: if (so->so_proto->pr_flags & PR_CONNREQUIRED) {
405: if ((so->so_state & SS_ISCONFIRMING) == 0 &&
406: !(resid == 0 && clen != 0))
407: snderr(ENOTCONN);
408: } else if (addr == 0)
409: snderr(EDESTADDRREQ);
410: }
411: space = sbspace(&so->so_snd);
412: if (flags & MSG_OOB)
413: space += 1024;
1.21 christos 414: if ((atomic && resid > so->so_snd.sb_hiwat) ||
1.11 mycroft 415: clen > so->so_snd.sb_hiwat)
416: snderr(EMSGSIZE);
417: if (space < resid + clen && uio &&
1.1 cgd 418: (atomic || space < so->so_snd.sb_lowat || space < clen)) {
419: if (so->so_state & SS_NBIO)
420: snderr(EWOULDBLOCK);
421: sbunlock(&so->so_snd);
422: error = sbwait(&so->so_snd);
423: splx(s);
424: if (error)
425: goto out;
426: goto restart;
427: }
428: splx(s);
429: mp = ⊤
430: space -= clen;
431: do {
432: if (uio == NULL) {
433: /*
434: * Data is prepackaged in "top".
435: */
436: resid = 0;
437: if (flags & MSG_EOR)
438: top->m_flags |= M_EOR;
439: } else do {
440: if (top == 0) {
441: MGETHDR(m, M_WAIT, MT_DATA);
442: mlen = MHLEN;
443: m->m_pkthdr.len = 0;
444: m->m_pkthdr.rcvif = (struct ifnet *)0;
445: } else {
446: MGET(m, M_WAIT, MT_DATA);
447: mlen = MLEN;
448: }
1.15 mycroft 449: if (resid >= MINCLSIZE && space >= MCLBYTES) {
1.1 cgd 450: MCLGET(m, M_WAIT);
451: if ((m->m_flags & M_EXT) == 0)
452: goto nopages;
453: mlen = MCLBYTES;
1.15 mycroft 454: #ifdef MAPPED_MBUFS
455: len = min(MCLBYTES, resid);
456: #else
457: if (atomic && top == 0) {
458: len = min(MCLBYTES - max_hdr, resid);
459: m->m_data += max_hdr;
460: } else
461: len = min(MCLBYTES, resid);
462: #endif
1.25 explorer 463: space -= len;
1.1 cgd 464: } else {
465: nopages:
466: len = min(min(mlen, resid), space);
1.15 mycroft 467: space -= len;
1.1 cgd 468: /*
469: * For datagram protocols, leave room
470: * for protocol headers in first mbuf.
471: */
472: if (atomic && top == 0 && len < mlen)
473: MH_ALIGN(m, len);
474: }
475: error = uiomove(mtod(m, caddr_t), (int)len, uio);
476: resid = uio->uio_resid;
477: m->m_len = len;
478: *mp = m;
479: top->m_pkthdr.len += len;
480: if (error)
481: goto release;
482: mp = &m->m_next;
483: if (resid <= 0) {
484: if (flags & MSG_EOR)
485: top->m_flags |= M_EOR;
486: break;
487: }
488: } while (space > 0 && atomic);
489: if (dontroute)
490: so->so_options |= SO_DONTROUTE;
1.40 thorpej 491: if (resid > 0)
492: so->so_state |= SS_MORETOCOME;
1.20 mycroft 493: s = splsoftnet(); /* XXX */
1.22 mycroft 494: error = (*so->so_proto->pr_usrreq)(so,
1.24 mycroft 495: (flags & MSG_OOB) ? PRU_SENDOOB : PRU_SEND,
496: top, addr, control, p);
1.1 cgd 497: splx(s);
498: if (dontroute)
499: so->so_options &= ~SO_DONTROUTE;
1.40 thorpej 500: if (resid > 0)
501: so->so_state &= ~SS_MORETOCOME;
1.1 cgd 502: clen = 0;
503: control = 0;
504: top = 0;
505: mp = ⊤
506: if (error)
507: goto release;
508: } while (resid && space > 0);
509: } while (resid);
510:
511: release:
512: sbunlock(&so->so_snd);
513: out:
514: if (top)
515: m_freem(top);
516: if (control)
517: m_freem(control);
518: return (error);
519: }
520:
521: /*
522: * Implement receive operations on a socket.
523: * We depend on the way that records are added to the sockbuf
524: * by sbappend*. In particular, each record (mbufs linked through m_next)
525: * must begin with an address if the protocol so specifies,
526: * followed by an optional mbuf or mbufs containing ancillary data,
527: * and then zero or more mbufs of data.
528: * In order to avoid blocking network interrupts for the entire time here,
529: * we splx() while doing the actual copy to user space.
530: * Although the sockbuf is locked, new data may still be appended,
531: * and thus we must maintain consistency of the sockbuf during that time.
532: *
533: * The caller may receive the data as a single mbuf chain by supplying
534: * an mbuf **mp0 for use in returning the chain. The uio is then used
535: * only for the count in uio_resid.
536: */
1.3 andrew 537: int
1.1 cgd 538: soreceive(so, paddr, uio, mp0, controlp, flagsp)
539: register struct socket *so;
540: struct mbuf **paddr;
541: struct uio *uio;
542: struct mbuf **mp0;
543: struct mbuf **controlp;
544: int *flagsp;
545: {
546: register struct mbuf *m, **mp;
547: register int flags, len, error, s, offset;
548: struct protosw *pr = so->so_proto;
549: struct mbuf *nextrecord;
1.21 christos 550: int moff, type = 0;
1.3 andrew 551: int orig_resid = uio->uio_resid;
1.1 cgd 552:
553: mp = mp0;
554: if (paddr)
555: *paddr = 0;
556: if (controlp)
557: *controlp = 0;
558: if (flagsp)
559: flags = *flagsp &~ MSG_EOR;
560: else
561: flags = 0;
562: if (flags & MSG_OOB) {
563: m = m_get(M_WAIT, MT_DATA);
1.17 cgd 564: error = (*pr->pr_usrreq)(so, PRU_RCVOOB, m,
1.22 mycroft 565: (struct mbuf *)(long)(flags & MSG_PEEK), (struct mbuf *)0,
566: (struct proc *)0);
1.1 cgd 567: if (error)
568: goto bad;
569: do {
570: error = uiomove(mtod(m, caddr_t),
571: (int) min(uio->uio_resid, m->m_len), uio);
572: m = m_free(m);
573: } while (uio->uio_resid && error == 0 && m);
574: bad:
575: if (m)
576: m_freem(m);
577: return (error);
578: }
579: if (mp)
580: *mp = (struct mbuf *)0;
581: if (so->so_state & SS_ISCONFIRMING && uio->uio_resid)
1.22 mycroft 582: (*pr->pr_usrreq)(so, PRU_RCVD, (struct mbuf *)0,
583: (struct mbuf *)0, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 584:
585: restart:
1.21 christos 586: if ((error = sblock(&so->so_rcv, SBLOCKWAIT(flags))) != 0)
1.1 cgd 587: return (error);
1.20 mycroft 588: s = splsoftnet();
1.1 cgd 589:
590: m = so->so_rcv.sb_mb;
591: /*
592: * If we have less data than requested, block awaiting more
593: * (subject to any timeout) if:
1.15 mycroft 594: * 1. the current count is less than the low water mark,
1.1 cgd 595: * 2. MSG_WAITALL is set, and it is possible to do the entire
1.15 mycroft 596: * receive operation at once if we block (resid <= hiwat), or
597: * 3. MSG_DONTWAIT is not set.
1.1 cgd 598: * If MSG_WAITALL is set but resid is larger than the receive buffer,
599: * we have to do the receive in sections, and thus risk returning
600: * a short count if a timeout or signal occurs after we start.
601: */
1.21 christos 602: if (m == 0 || (((flags & MSG_DONTWAIT) == 0 &&
1.15 mycroft 603: so->so_rcv.sb_cc < uio->uio_resid) &&
1.1 cgd 604: (so->so_rcv.sb_cc < so->so_rcv.sb_lowat ||
605: ((flags & MSG_WAITALL) && uio->uio_resid <= so->so_rcv.sb_hiwat)) &&
1.21 christos 606: m->m_nextpkt == 0 && (pr->pr_flags & PR_ATOMIC) == 0)) {
1.1 cgd 607: #ifdef DIAGNOSTIC
608: if (m == 0 && so->so_rcv.sb_cc)
609: panic("receive 1");
610: #endif
611: if (so->so_error) {
612: if (m)
1.15 mycroft 613: goto dontblock;
1.1 cgd 614: error = so->so_error;
615: if ((flags & MSG_PEEK) == 0)
616: so->so_error = 0;
617: goto release;
618: }
619: if (so->so_state & SS_CANTRCVMORE) {
620: if (m)
1.15 mycroft 621: goto dontblock;
1.1 cgd 622: else
623: goto release;
624: }
625: for (; m; m = m->m_next)
626: if (m->m_type == MT_OOBDATA || (m->m_flags & M_EOR)) {
627: m = so->so_rcv.sb_mb;
628: goto dontblock;
629: }
630: if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) == 0 &&
631: (so->so_proto->pr_flags & PR_CONNREQUIRED)) {
632: error = ENOTCONN;
633: goto release;
634: }
635: if (uio->uio_resid == 0)
636: goto release;
1.15 mycroft 637: if ((so->so_state & SS_NBIO) || (flags & MSG_DONTWAIT)) {
1.1 cgd 638: error = EWOULDBLOCK;
639: goto release;
640: }
641: sbunlock(&so->so_rcv);
642: error = sbwait(&so->so_rcv);
643: splx(s);
644: if (error)
645: return (error);
646: goto restart;
647: }
648: dontblock:
1.15 mycroft 649: #ifdef notyet /* XXXX */
650: if (uio->uio_procp)
651: uio->uio_procp->p_stats->p_ru.ru_msgrcv++;
652: #endif
1.1 cgd 653: nextrecord = m->m_nextpkt;
654: if (pr->pr_flags & PR_ADDR) {
655: #ifdef DIAGNOSTIC
656: if (m->m_type != MT_SONAME)
657: panic("receive 1a");
658: #endif
1.3 andrew 659: orig_resid = 0;
1.1 cgd 660: if (flags & MSG_PEEK) {
661: if (paddr)
662: *paddr = m_copy(m, 0, m->m_len);
663: m = m->m_next;
664: } else {
665: sbfree(&so->so_rcv, m);
666: if (paddr) {
667: *paddr = m;
668: so->so_rcv.sb_mb = m->m_next;
669: m->m_next = 0;
670: m = so->so_rcv.sb_mb;
671: } else {
672: MFREE(m, so->so_rcv.sb_mb);
673: m = so->so_rcv.sb_mb;
674: }
675: }
676: }
677: while (m && m->m_type == MT_CONTROL && error == 0) {
678: if (flags & MSG_PEEK) {
679: if (controlp)
680: *controlp = m_copy(m, 0, m->m_len);
681: m = m->m_next;
682: } else {
683: sbfree(&so->so_rcv, m);
684: if (controlp) {
685: if (pr->pr_domain->dom_externalize &&
686: mtod(m, struct cmsghdr *)->cmsg_type ==
687: SCM_RIGHTS)
688: error = (*pr->pr_domain->dom_externalize)(m);
689: *controlp = m;
690: so->so_rcv.sb_mb = m->m_next;
691: m->m_next = 0;
692: m = so->so_rcv.sb_mb;
693: } else {
694: MFREE(m, so->so_rcv.sb_mb);
695: m = so->so_rcv.sb_mb;
696: }
697: }
1.3 andrew 698: if (controlp) {
699: orig_resid = 0;
1.1 cgd 700: controlp = &(*controlp)->m_next;
1.3 andrew 701: }
1.1 cgd 702: }
703: if (m) {
704: if ((flags & MSG_PEEK) == 0)
705: m->m_nextpkt = nextrecord;
706: type = m->m_type;
707: if (type == MT_OOBDATA)
708: flags |= MSG_OOB;
709: }
710: moff = 0;
711: offset = 0;
712: while (m && uio->uio_resid > 0 && error == 0) {
713: if (m->m_type == MT_OOBDATA) {
714: if (type != MT_OOBDATA)
715: break;
716: } else if (type == MT_OOBDATA)
717: break;
718: #ifdef DIAGNOSTIC
719: else if (m->m_type != MT_DATA && m->m_type != MT_HEADER)
720: panic("receive 3");
721: #endif
722: so->so_state &= ~SS_RCVATMARK;
723: len = uio->uio_resid;
724: if (so->so_oobmark && len > so->so_oobmark - offset)
725: len = so->so_oobmark - offset;
726: if (len > m->m_len - moff)
727: len = m->m_len - moff;
728: /*
729: * If mp is set, just pass back the mbufs.
730: * Otherwise copy them out via the uio, then free.
731: * Sockbuf must be consistent here (points to current mbuf,
732: * it points to next record) when we drop priority;
733: * we must note any additions to the sockbuf when we
734: * block interrupts again.
735: */
736: if (mp == 0) {
737: splx(s);
738: error = uiomove(mtod(m, caddr_t) + moff, (int)len, uio);
1.20 mycroft 739: s = splsoftnet();
1.1 cgd 740: } else
741: uio->uio_resid -= len;
742: if (len == m->m_len - moff) {
743: if (m->m_flags & M_EOR)
744: flags |= MSG_EOR;
745: if (flags & MSG_PEEK) {
746: m = m->m_next;
747: moff = 0;
748: } else {
749: nextrecord = m->m_nextpkt;
750: sbfree(&so->so_rcv, m);
751: if (mp) {
752: *mp = m;
753: mp = &m->m_next;
754: so->so_rcv.sb_mb = m = m->m_next;
755: *mp = (struct mbuf *)0;
756: } else {
757: MFREE(m, so->so_rcv.sb_mb);
758: m = so->so_rcv.sb_mb;
759: }
760: if (m)
761: m->m_nextpkt = nextrecord;
762: }
763: } else {
764: if (flags & MSG_PEEK)
765: moff += len;
766: else {
767: if (mp)
768: *mp = m_copym(m, 0, len, M_WAIT);
769: m->m_data += len;
770: m->m_len -= len;
771: so->so_rcv.sb_cc -= len;
772: }
773: }
774: if (so->so_oobmark) {
775: if ((flags & MSG_PEEK) == 0) {
776: so->so_oobmark -= len;
777: if (so->so_oobmark == 0) {
778: so->so_state |= SS_RCVATMARK;
779: break;
780: }
1.7 cgd 781: } else {
1.1 cgd 782: offset += len;
1.7 cgd 783: if (offset == so->so_oobmark)
784: break;
785: }
1.1 cgd 786: }
787: if (flags & MSG_EOR)
788: break;
789: /*
790: * If the MSG_WAITALL flag is set (for non-atomic socket),
791: * we must not quit until "uio->uio_resid == 0" or an error
792: * termination. If a signal/timeout occurs, return
793: * with a short count but without error.
794: * Keep sockbuf locked against other readers.
795: */
796: while (flags & MSG_WAITALL && m == 0 && uio->uio_resid > 0 &&
1.3 andrew 797: !sosendallatonce(so) && !nextrecord) {
1.1 cgd 798: if (so->so_error || so->so_state & SS_CANTRCVMORE)
799: break;
800: error = sbwait(&so->so_rcv);
801: if (error) {
802: sbunlock(&so->so_rcv);
803: splx(s);
804: return (0);
805: }
1.21 christos 806: if ((m = so->so_rcv.sb_mb) != NULL)
1.1 cgd 807: nextrecord = m->m_nextpkt;
808: }
809: }
1.3 andrew 810:
811: if (m && pr->pr_flags & PR_ATOMIC) {
812: flags |= MSG_TRUNC;
813: if ((flags & MSG_PEEK) == 0)
814: (void) sbdroprecord(&so->so_rcv);
815: }
1.1 cgd 816: if ((flags & MSG_PEEK) == 0) {
817: if (m == 0)
818: so->so_rcv.sb_mb = nextrecord;
819: if (pr->pr_flags & PR_WANTRCVD && so->so_pcb)
1.22 mycroft 820: (*pr->pr_usrreq)(so, PRU_RCVD, (struct mbuf *)0,
821: (struct mbuf *)(long)flags, (struct mbuf *)0,
822: (struct proc *)0);
1.1 cgd 823: }
1.3 andrew 824: if (orig_resid == uio->uio_resid && orig_resid &&
825: (flags & MSG_EOR) == 0 && (so->so_state & SS_CANTRCVMORE) == 0) {
826: sbunlock(&so->so_rcv);
827: splx(s);
828: goto restart;
829: }
830:
1.1 cgd 831: if (flagsp)
832: *flagsp |= flags;
833: release:
834: sbunlock(&so->so_rcv);
835: splx(s);
836: return (error);
837: }
838:
1.14 mycroft 839: int
1.1 cgd 840: soshutdown(so, how)
1.34 kleink 841: struct socket *so;
842: int how;
1.1 cgd 843: {
1.34 kleink 844: struct protosw *pr = so->so_proto;
845:
846: if (!(how == SHUT_RD || how == SHUT_WR || how == SHUT_RDWR))
847: return (EINVAL);
1.1 cgd 848:
1.34 kleink 849: if (how == SHUT_RD || how == SHUT_RDWR)
1.1 cgd 850: sorflush(so);
1.34 kleink 851: if (how == SHUT_WR || how == SHUT_RDWR)
1.22 mycroft 852: return (*pr->pr_usrreq)(so, PRU_SHUTDOWN, (struct mbuf *)0,
853: (struct mbuf *)0, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 854: return (0);
855: }
856:
1.14 mycroft 857: void
1.1 cgd 858: sorflush(so)
859: register struct socket *so;
860: {
861: register struct sockbuf *sb = &so->so_rcv;
862: register struct protosw *pr = so->so_proto;
863: register int s;
864: struct sockbuf asb;
865:
866: sb->sb_flags |= SB_NOINTR;
1.15 mycroft 867: (void) sblock(sb, M_WAITOK);
1.1 cgd 868: s = splimp();
869: socantrcvmore(so);
870: sbunlock(sb);
871: asb = *sb;
1.38 perry 872: memset((caddr_t)sb, 0, sizeof(*sb));
1.1 cgd 873: splx(s);
874: if (pr->pr_flags & PR_RIGHTS && pr->pr_domain->dom_dispose)
875: (*pr->pr_domain->dom_dispose)(asb.sb_mb);
876: sbrelease(&asb);
877: }
878:
1.14 mycroft 879: int
1.1 cgd 880: sosetopt(so, level, optname, m0)
881: register struct socket *so;
882: int level, optname;
883: struct mbuf *m0;
884: {
885: int error = 0;
886: register struct mbuf *m = m0;
887:
888: if (level != SOL_SOCKET) {
889: if (so->so_proto && so->so_proto->pr_ctloutput)
890: return ((*so->so_proto->pr_ctloutput)
891: (PRCO_SETOPT, so, level, optname, &m0));
892: error = ENOPROTOOPT;
893: } else {
894: switch (optname) {
895:
896: case SO_LINGER:
1.36 perry 897: if (m == NULL || m->m_len != sizeof(struct linger)) {
1.1 cgd 898: error = EINVAL;
899: goto bad;
900: }
901: so->so_linger = mtod(m, struct linger *)->l_linger;
902: /* fall thru... */
903:
904: case SO_DEBUG:
905: case SO_KEEPALIVE:
906: case SO_DONTROUTE:
907: case SO_USELOOPBACK:
908: case SO_BROADCAST:
909: case SO_REUSEADDR:
1.15 mycroft 910: case SO_REUSEPORT:
1.1 cgd 911: case SO_OOBINLINE:
1.26 thorpej 912: case SO_TIMESTAMP:
1.36 perry 913: if (m == NULL || m->m_len < sizeof(int)) {
1.1 cgd 914: error = EINVAL;
915: goto bad;
916: }
917: if (*mtod(m, int *))
918: so->so_options |= optname;
919: else
920: so->so_options &= ~optname;
921: break;
922:
923: case SO_SNDBUF:
924: case SO_RCVBUF:
925: case SO_SNDLOWAT:
926: case SO_RCVLOWAT:
1.28 thorpej 927: {
928: int optval;
929:
1.36 perry 930: if (m == NULL || m->m_len < sizeof(int)) {
1.1 cgd 931: error = EINVAL;
932: goto bad;
933: }
1.28 thorpej 934:
935: /*
936: * Values < 1 make no sense for any of these
937: * options, so disallow them.
938: */
939: optval = *mtod(m, int *);
940: if (optval < 1) {
941: error = EINVAL;
942: goto bad;
943: }
944:
1.1 cgd 945: switch (optname) {
946:
947: case SO_SNDBUF:
948: case SO_RCVBUF:
949: if (sbreserve(optname == SO_SNDBUF ?
950: &so->so_snd : &so->so_rcv,
1.28 thorpej 951: (u_long) optval) == 0) {
1.1 cgd 952: error = ENOBUFS;
953: goto bad;
954: }
955: break;
956:
1.28 thorpej 957: /*
958: * Make sure the low-water is never greater than
959: * the high-water.
960: */
1.1 cgd 961: case SO_SNDLOWAT:
1.28 thorpej 962: so->so_snd.sb_lowat =
963: (optval > so->so_snd.sb_hiwat) ?
964: so->so_snd.sb_hiwat : optval;
1.1 cgd 965: break;
966: case SO_RCVLOWAT:
1.28 thorpej 967: so->so_rcv.sb_lowat =
968: (optval > so->so_rcv.sb_hiwat) ?
969: so->so_rcv.sb_hiwat : optval;
1.1 cgd 970: break;
971: }
972: break;
1.28 thorpej 973: }
1.1 cgd 974:
975: case SO_SNDTIMEO:
976: case SO_RCVTIMEO:
977: {
978: struct timeval *tv;
979: short val;
980:
1.36 perry 981: if (m == NULL || m->m_len < sizeof(*tv)) {
1.1 cgd 982: error = EINVAL;
983: goto bad;
984: }
985: tv = mtod(m, struct timeval *);
1.19 cgd 986: if (tv->tv_sec * hz + tv->tv_usec / tick > SHRT_MAX) {
1.1 cgd 987: error = EDOM;
988: goto bad;
989: }
990: val = tv->tv_sec * hz + tv->tv_usec / tick;
991:
992: switch (optname) {
993:
994: case SO_SNDTIMEO:
995: so->so_snd.sb_timeo = val;
996: break;
997: case SO_RCVTIMEO:
998: so->so_rcv.sb_timeo = val;
999: break;
1000: }
1001: break;
1002: }
1003:
1004: default:
1005: error = ENOPROTOOPT;
1006: break;
1007: }
1.15 mycroft 1008: if (error == 0 && so->so_proto && so->so_proto->pr_ctloutput) {
1009: (void) ((*so->so_proto->pr_ctloutput)
1010: (PRCO_SETOPT, so, level, optname, &m0));
1011: m = NULL; /* freed by protocol */
1012: }
1.1 cgd 1013: }
1014: bad:
1015: if (m)
1016: (void) m_free(m);
1017: return (error);
1018: }
1019:
1.14 mycroft 1020: int
1.1 cgd 1021: sogetopt(so, level, optname, mp)
1022: register struct socket *so;
1023: int level, optname;
1024: struct mbuf **mp;
1025: {
1026: register struct mbuf *m;
1027:
1028: if (level != SOL_SOCKET) {
1029: if (so->so_proto && so->so_proto->pr_ctloutput) {
1030: return ((*so->so_proto->pr_ctloutput)
1031: (PRCO_GETOPT, so, level, optname, mp));
1032: } else
1033: return (ENOPROTOOPT);
1034: } else {
1035: m = m_get(M_WAIT, MT_SOOPTS);
1.36 perry 1036: m->m_len = sizeof(int);
1.1 cgd 1037:
1038: switch (optname) {
1039:
1040: case SO_LINGER:
1.36 perry 1041: m->m_len = sizeof(struct linger);
1.1 cgd 1042: mtod(m, struct linger *)->l_onoff =
1043: so->so_options & SO_LINGER;
1044: mtod(m, struct linger *)->l_linger = so->so_linger;
1045: break;
1046:
1047: case SO_USELOOPBACK:
1048: case SO_DONTROUTE:
1049: case SO_DEBUG:
1050: case SO_KEEPALIVE:
1051: case SO_REUSEADDR:
1.15 mycroft 1052: case SO_REUSEPORT:
1.1 cgd 1053: case SO_BROADCAST:
1054: case SO_OOBINLINE:
1.26 thorpej 1055: case SO_TIMESTAMP:
1.1 cgd 1056: *mtod(m, int *) = so->so_options & optname;
1057: break;
1058:
1059: case SO_TYPE:
1060: *mtod(m, int *) = so->so_type;
1061: break;
1062:
1063: case SO_ERROR:
1064: *mtod(m, int *) = so->so_error;
1065: so->so_error = 0;
1066: break;
1067:
1068: case SO_SNDBUF:
1069: *mtod(m, int *) = so->so_snd.sb_hiwat;
1070: break;
1071:
1072: case SO_RCVBUF:
1073: *mtod(m, int *) = so->so_rcv.sb_hiwat;
1074: break;
1075:
1076: case SO_SNDLOWAT:
1077: *mtod(m, int *) = so->so_snd.sb_lowat;
1078: break;
1079:
1080: case SO_RCVLOWAT:
1081: *mtod(m, int *) = so->so_rcv.sb_lowat;
1082: break;
1083:
1084: case SO_SNDTIMEO:
1085: case SO_RCVTIMEO:
1086: {
1087: int val = (optname == SO_SNDTIMEO ?
1088: so->so_snd.sb_timeo : so->so_rcv.sb_timeo);
1089:
1090: m->m_len = sizeof(struct timeval);
1091: mtod(m, struct timeval *)->tv_sec = val / hz;
1092: mtod(m, struct timeval *)->tv_usec =
1.27 kleink 1093: (val % hz) * tick;
1.1 cgd 1094: break;
1095: }
1096:
1097: default:
1098: (void)m_free(m);
1099: return (ENOPROTOOPT);
1100: }
1101: *mp = m;
1102: return (0);
1103: }
1104: }
1105:
1.14 mycroft 1106: void
1.1 cgd 1107: sohasoutofband(so)
1108: register struct socket *so;
1109: {
1110: struct proc *p;
1111:
1112: if (so->so_pgid < 0)
1113: gsignal(-so->so_pgid, SIGURG);
1114: else if (so->so_pgid > 0 && (p = pfind(so->so_pgid)) != 0)
1115: psignal(p, SIGURG);
1.2 cgd 1116: selwakeup(&so->so_rcv.sb_sel);
1.1 cgd 1117: }
CVSweb <webmaster@jp.NetBSD.org>