|
|
1.28.4.1! thorpej 1: /* $NetBSD: uipc_socket.c,v 1.29 1997/08/27 07:10:01 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: *
1.16 cgd 35: * @(#)uipc_socket.c 8.3 (Berkeley) 4/15/94
1.1 cgd 36: */
37:
1.9 mycroft 38: #include <sys/param.h>
39: #include <sys/systm.h>
40: #include <sys/proc.h>
41: #include <sys/file.h>
42: #include <sys/malloc.h>
43: #include <sys/mbuf.h>
44: #include <sys/domain.h>
45: #include <sys/kernel.h>
46: #include <sys/protosw.h>
47: #include <sys/socket.h>
48: #include <sys/socketvar.h>
1.21 christos 49: #include <sys/signalvar.h>
1.9 mycroft 50: #include <sys/resourcevar.h>
1.1 cgd 51:
52: /*
53: * Socket operation routines.
54: * These routines are called by the routines in
55: * sys_socket.c or from a system process, and
56: * implement the semantics of socket operations by
57: * switching out to the protocol specific routines.
58: */
59: /*ARGSUSED*/
1.3 andrew 60: int
1.1 cgd 61: socreate(dom, aso, type, proto)
1.11 mycroft 62: int dom;
1.1 cgd 63: struct socket **aso;
64: register int type;
65: int proto;
66: {
67: struct proc *p = curproc; /* XXX */
68: register struct protosw *prp;
69: register struct socket *so;
70: register int error;
71:
72: if (proto)
73: prp = pffindproto(dom, proto, type);
74: else
75: prp = pffindtype(dom, type);
1.15 mycroft 76: if (prp == 0 || prp->pr_usrreq == 0)
1.1 cgd 77: return (EPROTONOSUPPORT);
78: if (prp->pr_type != type)
79: return (EPROTOTYPE);
80: MALLOC(so, struct socket *, sizeof(*so), M_SOCKET, M_WAIT);
81: bzero((caddr_t)so, sizeof(*so));
82: so->so_type = type;
83: so->so_proto = prp;
1.22 mycroft 84: error = (*prp->pr_usrreq)(so, PRU_ATTACH, (struct mbuf *)0,
85: (struct mbuf *)(long)proto, (struct mbuf *)0, p);
1.1 cgd 86: if (error) {
87: so->so_state |= SS_NOFDREF;
88: sofree(so);
89: return (error);
90: }
1.10 deraadt 91: #ifdef COMPAT_SUNOS
1.18 christos 92: {
93: extern struct emul emul_sunos;
94: if (p->p_emul == &emul_sunos && type == SOCK_DGRAM)
95: so->so_options |= SO_BROADCAST;
96: }
1.10 deraadt 97: #endif
1.1 cgd 98: *aso = so;
99: return (0);
100: }
101:
1.3 andrew 102: int
1.1 cgd 103: sobind(so, nam)
104: struct socket *so;
105: struct mbuf *nam;
106: {
1.22 mycroft 107: struct proc *p = curproc; /* XXX */
1.20 mycroft 108: int s = splsoftnet();
1.1 cgd 109: int error;
110:
1.22 mycroft 111: error = (*so->so_proto->pr_usrreq)(so, PRU_BIND, (struct mbuf *)0,
112: nam, (struct mbuf *)0, p);
1.1 cgd 113: splx(s);
114: return (error);
115: }
116:
1.3 andrew 117: int
1.1 cgd 118: solisten(so, backlog)
119: register struct socket *so;
120: int backlog;
121: {
1.20 mycroft 122: int s = splsoftnet(), error;
1.1 cgd 123:
1.22 mycroft 124: error = (*so->so_proto->pr_usrreq)(so, PRU_LISTEN, (struct mbuf *)0,
125: (struct mbuf *)0, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 126: if (error) {
127: splx(s);
128: return (error);
129: }
130: if (so->so_q == 0)
131: so->so_options |= SO_ACCEPTCONN;
132: if (backlog < 0)
133: backlog = 0;
134: so->so_qlimit = min(backlog, SOMAXCONN);
135: splx(s);
136: return (0);
137: }
138:
1.21 christos 139: void
1.1 cgd 140: sofree(so)
141: register struct socket *so;
142: {
143:
144: if (so->so_pcb || (so->so_state & SS_NOFDREF) == 0)
145: return;
146: if (so->so_head) {
147: if (!soqremque(so, 0) && !soqremque(so, 1))
148: panic("sofree dq");
149: so->so_head = 0;
150: }
151: sbrelease(&so->so_snd);
152: sorflush(so);
153: FREE(so, M_SOCKET);
154: }
155:
156: /*
157: * Close a socket on last file table reference removal.
158: * Initiate disconnect if connected.
159: * Free socket when disconnect complete.
160: */
1.3 andrew 161: int
1.1 cgd 162: soclose(so)
163: register struct socket *so;
164: {
1.20 mycroft 165: int s = splsoftnet(); /* conservative */
1.1 cgd 166: int error = 0;
167:
168: if (so->so_options & SO_ACCEPTCONN) {
169: while (so->so_q0)
170: (void) soabort(so->so_q0);
171: while (so->so_q)
172: (void) soabort(so->so_q);
173: }
174: if (so->so_pcb == 0)
175: goto discard;
176: if (so->so_state & SS_ISCONNECTED) {
177: if ((so->so_state & SS_ISDISCONNECTING) == 0) {
178: error = sodisconnect(so);
179: if (error)
180: goto drop;
181: }
182: if (so->so_options & SO_LINGER) {
183: if ((so->so_state & SS_ISDISCONNECTING) &&
184: (so->so_state & SS_NBIO))
185: goto drop;
1.21 christos 186: while (so->so_state & SS_ISCONNECTED) {
187: error = tsleep((caddr_t)&so->so_timeo,
188: PSOCK | PCATCH, netcls,
189: so->so_linger);
190: if (error)
1.1 cgd 191: break;
1.21 christos 192: }
1.1 cgd 193: }
194: }
195: drop:
196: if (so->so_pcb) {
1.22 mycroft 197: int error2 = (*so->so_proto->pr_usrreq)(so, PRU_DETACH,
198: (struct mbuf *)0, (struct mbuf *)0, (struct mbuf *)0,
199: (struct proc *)0);
1.1 cgd 200: if (error == 0)
201: error = error2;
202: }
203: discard:
204: if (so->so_state & SS_NOFDREF)
205: panic("soclose: NOFDREF");
206: so->so_state |= SS_NOFDREF;
207: sofree(so);
208: splx(s);
209: return (error);
210: }
211:
212: /*
1.20 mycroft 213: * Must be called at splsoftnet...
1.1 cgd 214: */
1.3 andrew 215: int
1.1 cgd 216: soabort(so)
217: struct socket *so;
218: {
219:
1.22 mycroft 220: return (*so->so_proto->pr_usrreq)(so, PRU_ABORT, (struct mbuf *)0,
221: (struct mbuf *)0, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 222: }
223:
1.3 andrew 224: int
1.1 cgd 225: soaccept(so, nam)
226: register struct socket *so;
227: struct mbuf *nam;
228: {
1.20 mycroft 229: int s = splsoftnet();
1.1 cgd 230: int error;
231:
232: if ((so->so_state & SS_NOFDREF) == 0)
233: panic("soaccept: !NOFDREF");
234: so->so_state &= ~SS_NOFDREF;
1.22 mycroft 235: error = (*so->so_proto->pr_usrreq)(so, PRU_ACCEPT, (struct mbuf *)0,
236: nam, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 237: splx(s);
238: return (error);
239: }
240:
1.3 andrew 241: int
1.1 cgd 242: soconnect(so, nam)
243: register struct socket *so;
244: struct mbuf *nam;
245: {
1.23 mycroft 246: struct proc *p = curproc; /* XXX */
1.1 cgd 247: int s;
248: int error;
249:
250: if (so->so_options & SO_ACCEPTCONN)
251: return (EOPNOTSUPP);
1.20 mycroft 252: s = splsoftnet();
1.1 cgd 253: /*
254: * If protocol is connection-based, can only connect once.
255: * Otherwise, if connected, try to disconnect first.
256: * This allows user to disconnect by connecting to, e.g.,
257: * a null address.
258: */
259: if (so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING) &&
260: ((so->so_proto->pr_flags & PR_CONNREQUIRED) ||
261: (error = sodisconnect(so))))
262: error = EISCONN;
263: else
264: error = (*so->so_proto->pr_usrreq)(so, PRU_CONNECT,
1.23 mycroft 265: (struct mbuf *)0, nam, (struct mbuf *)0, p);
1.1 cgd 266: splx(s);
267: return (error);
268: }
269:
1.3 andrew 270: int
1.1 cgd 271: soconnect2(so1, so2)
272: register struct socket *so1;
273: struct socket *so2;
274: {
1.20 mycroft 275: int s = splsoftnet();
1.1 cgd 276: int error;
277:
1.22 mycroft 278: error = (*so1->so_proto->pr_usrreq)(so1, PRU_CONNECT2,
279: (struct mbuf *)0, (struct mbuf *)so2, (struct mbuf *)0,
280: (struct proc *)0);
1.1 cgd 281: splx(s);
282: return (error);
283: }
284:
1.3 andrew 285: int
1.1 cgd 286: sodisconnect(so)
287: register struct socket *so;
288: {
1.20 mycroft 289: int s = splsoftnet();
1.1 cgd 290: int error;
291:
292: if ((so->so_state & SS_ISCONNECTED) == 0) {
293: error = ENOTCONN;
294: goto bad;
295: }
296: if (so->so_state & SS_ISDISCONNECTING) {
297: error = EALREADY;
298: goto bad;
299: }
1.22 mycroft 300: error = (*so->so_proto->pr_usrreq)(so, PRU_DISCONNECT,
301: (struct mbuf *)0, (struct mbuf *)0, (struct mbuf *)0,
302: (struct proc *)0);
1.1 cgd 303: bad:
304: splx(s);
305: return (error);
306: }
307:
1.15 mycroft 308: #define SBLOCKWAIT(f) (((f) & MSG_DONTWAIT) ? M_NOWAIT : M_WAITOK)
1.1 cgd 309: /*
310: * Send on a socket.
311: * If send must go all at once and message is larger than
312: * send buffering, then hard error.
313: * Lock against other senders.
314: * If must go all at once and not enough room now, then
315: * inform user that this would block and do nothing.
316: * Otherwise, if nonblocking, send as much as possible.
317: * The data to be sent is described by "uio" if nonzero,
318: * otherwise by the mbuf chain "top" (which must be null
319: * if uio is not). Data provided in mbuf chain must be small
320: * enough to send all at once.
321: *
322: * Returns nonzero on error, timeout or signal; callers
323: * must check for short counts if EINTR/ERESTART are returned.
324: * Data and control buffers are freed on return.
325: */
1.3 andrew 326: int
1.1 cgd 327: sosend(so, addr, uio, top, control, flags)
328: register struct socket *so;
329: struct mbuf *addr;
330: struct uio *uio;
331: struct mbuf *top;
332: struct mbuf *control;
333: int flags;
334: {
1.15 mycroft 335: struct proc *p = curproc; /* XXX */
1.1 cgd 336: struct mbuf **mp;
337: register struct mbuf *m;
338: register long space, len, resid;
339: int clen = 0, error, s, dontroute, mlen;
340: int atomic = sosendallatonce(so) || top;
341:
342: if (uio)
343: resid = uio->uio_resid;
344: else
345: resid = top->m_pkthdr.len;
1.7 cgd 346: /*
347: * In theory resid should be unsigned.
348: * However, space must be signed, as it might be less than 0
349: * if we over-committed, and we must use a signed comparison
350: * of space and resid. On the other hand, a negative resid
351: * causes us to loop sending 0-length segments to the protocol.
352: */
1.28.4.1! thorpej 353: if (resid < 0) {
! 354: error = EINVAL;
! 355: goto out;
! 356: }
1.1 cgd 357: dontroute =
358: (flags & MSG_DONTROUTE) && (so->so_options & SO_DONTROUTE) == 0 &&
359: (so->so_proto->pr_flags & PR_ATOMIC);
1.12 mycroft 360: p->p_stats->p_ru.ru_msgsnd++;
1.1 cgd 361: if (control)
362: clen = control->m_len;
363: #define snderr(errno) { error = errno; splx(s); goto release; }
364:
365: restart:
1.21 christos 366: if ((error = sblock(&so->so_snd, SBLOCKWAIT(flags))) != 0)
1.1 cgd 367: goto out;
368: do {
1.20 mycroft 369: s = splsoftnet();
1.1 cgd 370: if (so->so_state & SS_CANTSENDMORE)
371: snderr(EPIPE);
372: if (so->so_error)
373: snderr(so->so_error);
374: if ((so->so_state & SS_ISCONNECTED) == 0) {
375: if (so->so_proto->pr_flags & PR_CONNREQUIRED) {
376: if ((so->so_state & SS_ISCONFIRMING) == 0 &&
377: !(resid == 0 && clen != 0))
378: snderr(ENOTCONN);
379: } else if (addr == 0)
380: snderr(EDESTADDRREQ);
381: }
382: space = sbspace(&so->so_snd);
383: if (flags & MSG_OOB)
384: space += 1024;
1.21 christos 385: if ((atomic && resid > so->so_snd.sb_hiwat) ||
1.11 mycroft 386: clen > so->so_snd.sb_hiwat)
387: snderr(EMSGSIZE);
388: if (space < resid + clen && uio &&
1.1 cgd 389: (atomic || space < so->so_snd.sb_lowat || space < clen)) {
390: if (so->so_state & SS_NBIO)
391: snderr(EWOULDBLOCK);
392: sbunlock(&so->so_snd);
393: error = sbwait(&so->so_snd);
394: splx(s);
395: if (error)
396: goto out;
397: goto restart;
398: }
399: splx(s);
400: mp = ⊤
401: space -= clen;
402: do {
403: if (uio == NULL) {
404: /*
405: * Data is prepackaged in "top".
406: */
407: resid = 0;
408: if (flags & MSG_EOR)
409: top->m_flags |= M_EOR;
410: } else do {
411: if (top == 0) {
412: MGETHDR(m, M_WAIT, MT_DATA);
413: mlen = MHLEN;
414: m->m_pkthdr.len = 0;
415: m->m_pkthdr.rcvif = (struct ifnet *)0;
416: } else {
417: MGET(m, M_WAIT, MT_DATA);
418: mlen = MLEN;
419: }
1.15 mycroft 420: if (resid >= MINCLSIZE && space >= MCLBYTES) {
1.1 cgd 421: MCLGET(m, M_WAIT);
422: if ((m->m_flags & M_EXT) == 0)
423: goto nopages;
424: mlen = MCLBYTES;
1.15 mycroft 425: #ifdef MAPPED_MBUFS
426: len = min(MCLBYTES, resid);
427: #else
428: if (atomic && top == 0) {
429: len = min(MCLBYTES - max_hdr, resid);
430: m->m_data += max_hdr;
431: } else
432: len = min(MCLBYTES, resid);
433: #endif
1.25 explorer 434: space -= len;
1.1 cgd 435: } else {
436: nopages:
437: len = min(min(mlen, resid), space);
1.15 mycroft 438: space -= len;
1.1 cgd 439: /*
440: * For datagram protocols, leave room
441: * for protocol headers in first mbuf.
442: */
443: if (atomic && top == 0 && len < mlen)
444: MH_ALIGN(m, len);
445: }
446: error = uiomove(mtod(m, caddr_t), (int)len, uio);
447: resid = uio->uio_resid;
448: m->m_len = len;
449: *mp = m;
450: top->m_pkthdr.len += len;
451: if (error)
452: goto release;
453: mp = &m->m_next;
454: if (resid <= 0) {
455: if (flags & MSG_EOR)
456: top->m_flags |= M_EOR;
457: break;
458: }
459: } while (space > 0 && atomic);
460: if (dontroute)
461: so->so_options |= SO_DONTROUTE;
1.20 mycroft 462: s = splsoftnet(); /* XXX */
1.22 mycroft 463: error = (*so->so_proto->pr_usrreq)(so,
1.24 mycroft 464: (flags & MSG_OOB) ? PRU_SENDOOB : PRU_SEND,
465: top, addr, control, p);
1.1 cgd 466: splx(s);
467: if (dontroute)
468: so->so_options &= ~SO_DONTROUTE;
469: clen = 0;
470: control = 0;
471: top = 0;
472: mp = ⊤
473: if (error)
474: goto release;
475: } while (resid && space > 0);
476: } while (resid);
477:
478: release:
479: sbunlock(&so->so_snd);
480: out:
481: if (top)
482: m_freem(top);
483: if (control)
484: m_freem(control);
485: return (error);
486: }
487:
488: /*
489: * Implement receive operations on a socket.
490: * We depend on the way that records are added to the sockbuf
491: * by sbappend*. In particular, each record (mbufs linked through m_next)
492: * must begin with an address if the protocol so specifies,
493: * followed by an optional mbuf or mbufs containing ancillary data,
494: * and then zero or more mbufs of data.
495: * In order to avoid blocking network interrupts for the entire time here,
496: * we splx() while doing the actual copy to user space.
497: * Although the sockbuf is locked, new data may still be appended,
498: * and thus we must maintain consistency of the sockbuf during that time.
499: *
500: * The caller may receive the data as a single mbuf chain by supplying
501: * an mbuf **mp0 for use in returning the chain. The uio is then used
502: * only for the count in uio_resid.
503: */
1.3 andrew 504: int
1.1 cgd 505: soreceive(so, paddr, uio, mp0, controlp, flagsp)
506: register struct socket *so;
507: struct mbuf **paddr;
508: struct uio *uio;
509: struct mbuf **mp0;
510: struct mbuf **controlp;
511: int *flagsp;
512: {
513: register struct mbuf *m, **mp;
514: register int flags, len, error, s, offset;
515: struct protosw *pr = so->so_proto;
516: struct mbuf *nextrecord;
1.21 christos 517: int moff, type = 0;
1.3 andrew 518: int orig_resid = uio->uio_resid;
1.1 cgd 519:
520: mp = mp0;
521: if (paddr)
522: *paddr = 0;
523: if (controlp)
524: *controlp = 0;
525: if (flagsp)
526: flags = *flagsp &~ MSG_EOR;
527: else
528: flags = 0;
529: if (flags & MSG_OOB) {
530: m = m_get(M_WAIT, MT_DATA);
1.17 cgd 531: error = (*pr->pr_usrreq)(so, PRU_RCVOOB, m,
1.22 mycroft 532: (struct mbuf *)(long)(flags & MSG_PEEK), (struct mbuf *)0,
533: (struct proc *)0);
1.1 cgd 534: if (error)
535: goto bad;
536: do {
537: error = uiomove(mtod(m, caddr_t),
538: (int) min(uio->uio_resid, m->m_len), uio);
539: m = m_free(m);
540: } while (uio->uio_resid && error == 0 && m);
541: bad:
542: if (m)
543: m_freem(m);
544: return (error);
545: }
546: if (mp)
547: *mp = (struct mbuf *)0;
548: if (so->so_state & SS_ISCONFIRMING && uio->uio_resid)
1.22 mycroft 549: (*pr->pr_usrreq)(so, PRU_RCVD, (struct mbuf *)0,
550: (struct mbuf *)0, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 551:
552: restart:
1.21 christos 553: if ((error = sblock(&so->so_rcv, SBLOCKWAIT(flags))) != 0)
1.1 cgd 554: return (error);
1.20 mycroft 555: s = splsoftnet();
1.1 cgd 556:
557: m = so->so_rcv.sb_mb;
558: /*
559: * If we have less data than requested, block awaiting more
560: * (subject to any timeout) if:
1.15 mycroft 561: * 1. the current count is less than the low water mark,
1.1 cgd 562: * 2. MSG_WAITALL is set, and it is possible to do the entire
1.15 mycroft 563: * receive operation at once if we block (resid <= hiwat), or
564: * 3. MSG_DONTWAIT is not set.
1.1 cgd 565: * If MSG_WAITALL is set but resid is larger than the receive buffer,
566: * we have to do the receive in sections, and thus risk returning
567: * a short count if a timeout or signal occurs after we start.
568: */
1.21 christos 569: if (m == 0 || (((flags & MSG_DONTWAIT) == 0 &&
1.15 mycroft 570: so->so_rcv.sb_cc < uio->uio_resid) &&
1.1 cgd 571: (so->so_rcv.sb_cc < so->so_rcv.sb_lowat ||
572: ((flags & MSG_WAITALL) && uio->uio_resid <= so->so_rcv.sb_hiwat)) &&
1.21 christos 573: m->m_nextpkt == 0 && (pr->pr_flags & PR_ATOMIC) == 0)) {
1.1 cgd 574: #ifdef DIAGNOSTIC
575: if (m == 0 && so->so_rcv.sb_cc)
576: panic("receive 1");
577: #endif
578: if (so->so_error) {
579: if (m)
1.15 mycroft 580: goto dontblock;
1.1 cgd 581: error = so->so_error;
582: if ((flags & MSG_PEEK) == 0)
583: so->so_error = 0;
584: goto release;
585: }
586: if (so->so_state & SS_CANTRCVMORE) {
587: if (m)
1.15 mycroft 588: goto dontblock;
1.1 cgd 589: else
590: goto release;
591: }
592: for (; m; m = m->m_next)
593: if (m->m_type == MT_OOBDATA || (m->m_flags & M_EOR)) {
594: m = so->so_rcv.sb_mb;
595: goto dontblock;
596: }
597: if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) == 0 &&
598: (so->so_proto->pr_flags & PR_CONNREQUIRED)) {
599: error = ENOTCONN;
600: goto release;
601: }
602: if (uio->uio_resid == 0)
603: goto release;
1.15 mycroft 604: if ((so->so_state & SS_NBIO) || (flags & MSG_DONTWAIT)) {
1.1 cgd 605: error = EWOULDBLOCK;
606: goto release;
607: }
608: sbunlock(&so->so_rcv);
609: error = sbwait(&so->so_rcv);
610: splx(s);
611: if (error)
612: return (error);
613: goto restart;
614: }
615: dontblock:
1.15 mycroft 616: #ifdef notyet /* XXXX */
617: if (uio->uio_procp)
618: uio->uio_procp->p_stats->p_ru.ru_msgrcv++;
619: #endif
1.1 cgd 620: nextrecord = m->m_nextpkt;
621: if (pr->pr_flags & PR_ADDR) {
622: #ifdef DIAGNOSTIC
623: if (m->m_type != MT_SONAME)
624: panic("receive 1a");
625: #endif
1.3 andrew 626: orig_resid = 0;
1.1 cgd 627: if (flags & MSG_PEEK) {
628: if (paddr)
629: *paddr = m_copy(m, 0, m->m_len);
630: m = m->m_next;
631: } else {
632: sbfree(&so->so_rcv, m);
633: if (paddr) {
634: *paddr = m;
635: so->so_rcv.sb_mb = m->m_next;
636: m->m_next = 0;
637: m = so->so_rcv.sb_mb;
638: } else {
639: MFREE(m, so->so_rcv.sb_mb);
640: m = so->so_rcv.sb_mb;
641: }
642: }
643: }
644: while (m && m->m_type == MT_CONTROL && error == 0) {
645: if (flags & MSG_PEEK) {
646: if (controlp)
647: *controlp = m_copy(m, 0, m->m_len);
648: m = m->m_next;
649: } else {
650: sbfree(&so->so_rcv, m);
651: if (controlp) {
652: if (pr->pr_domain->dom_externalize &&
653: mtod(m, struct cmsghdr *)->cmsg_type ==
654: SCM_RIGHTS)
655: error = (*pr->pr_domain->dom_externalize)(m);
656: *controlp = m;
657: so->so_rcv.sb_mb = m->m_next;
658: m->m_next = 0;
659: m = so->so_rcv.sb_mb;
660: } else {
661: MFREE(m, so->so_rcv.sb_mb);
662: m = so->so_rcv.sb_mb;
663: }
664: }
1.3 andrew 665: if (controlp) {
666: orig_resid = 0;
1.1 cgd 667: controlp = &(*controlp)->m_next;
1.3 andrew 668: }
1.1 cgd 669: }
670: if (m) {
671: if ((flags & MSG_PEEK) == 0)
672: m->m_nextpkt = nextrecord;
673: type = m->m_type;
674: if (type == MT_OOBDATA)
675: flags |= MSG_OOB;
676: }
677: moff = 0;
678: offset = 0;
679: while (m && uio->uio_resid > 0 && error == 0) {
680: if (m->m_type == MT_OOBDATA) {
681: if (type != MT_OOBDATA)
682: break;
683: } else if (type == MT_OOBDATA)
684: break;
685: #ifdef DIAGNOSTIC
686: else if (m->m_type != MT_DATA && m->m_type != MT_HEADER)
687: panic("receive 3");
688: #endif
689: so->so_state &= ~SS_RCVATMARK;
690: len = uio->uio_resid;
691: if (so->so_oobmark && len > so->so_oobmark - offset)
692: len = so->so_oobmark - offset;
693: if (len > m->m_len - moff)
694: len = m->m_len - moff;
695: /*
696: * If mp is set, just pass back the mbufs.
697: * Otherwise copy them out via the uio, then free.
698: * Sockbuf must be consistent here (points to current mbuf,
699: * it points to next record) when we drop priority;
700: * we must note any additions to the sockbuf when we
701: * block interrupts again.
702: */
703: if (mp == 0) {
704: splx(s);
705: error = uiomove(mtod(m, caddr_t) + moff, (int)len, uio);
1.20 mycroft 706: s = splsoftnet();
1.1 cgd 707: } else
708: uio->uio_resid -= len;
709: if (len == m->m_len - moff) {
710: if (m->m_flags & M_EOR)
711: flags |= MSG_EOR;
712: if (flags & MSG_PEEK) {
713: m = m->m_next;
714: moff = 0;
715: } else {
716: nextrecord = m->m_nextpkt;
717: sbfree(&so->so_rcv, m);
718: if (mp) {
719: *mp = m;
720: mp = &m->m_next;
721: so->so_rcv.sb_mb = m = m->m_next;
722: *mp = (struct mbuf *)0;
723: } else {
724: MFREE(m, so->so_rcv.sb_mb);
725: m = so->so_rcv.sb_mb;
726: }
727: if (m)
728: m->m_nextpkt = nextrecord;
729: }
730: } else {
731: if (flags & MSG_PEEK)
732: moff += len;
733: else {
734: if (mp)
735: *mp = m_copym(m, 0, len, M_WAIT);
736: m->m_data += len;
737: m->m_len -= len;
738: so->so_rcv.sb_cc -= len;
739: }
740: }
741: if (so->so_oobmark) {
742: if ((flags & MSG_PEEK) == 0) {
743: so->so_oobmark -= len;
744: if (so->so_oobmark == 0) {
745: so->so_state |= SS_RCVATMARK;
746: break;
747: }
1.7 cgd 748: } else {
1.1 cgd 749: offset += len;
1.7 cgd 750: if (offset == so->so_oobmark)
751: break;
752: }
1.1 cgd 753: }
754: if (flags & MSG_EOR)
755: break;
756: /*
757: * If the MSG_WAITALL flag is set (for non-atomic socket),
758: * we must not quit until "uio->uio_resid == 0" or an error
759: * termination. If a signal/timeout occurs, return
760: * with a short count but without error.
761: * Keep sockbuf locked against other readers.
762: */
763: while (flags & MSG_WAITALL && m == 0 && uio->uio_resid > 0 &&
1.3 andrew 764: !sosendallatonce(so) && !nextrecord) {
1.1 cgd 765: if (so->so_error || so->so_state & SS_CANTRCVMORE)
766: break;
767: error = sbwait(&so->so_rcv);
768: if (error) {
769: sbunlock(&so->so_rcv);
770: splx(s);
771: return (0);
772: }
1.21 christos 773: if ((m = so->so_rcv.sb_mb) != NULL)
1.1 cgd 774: nextrecord = m->m_nextpkt;
775: }
776: }
1.3 andrew 777:
778: if (m && pr->pr_flags & PR_ATOMIC) {
779: flags |= MSG_TRUNC;
780: if ((flags & MSG_PEEK) == 0)
781: (void) sbdroprecord(&so->so_rcv);
782: }
1.1 cgd 783: if ((flags & MSG_PEEK) == 0) {
784: if (m == 0)
785: so->so_rcv.sb_mb = nextrecord;
786: if (pr->pr_flags & PR_WANTRCVD && so->so_pcb)
1.22 mycroft 787: (*pr->pr_usrreq)(so, PRU_RCVD, (struct mbuf *)0,
788: (struct mbuf *)(long)flags, (struct mbuf *)0,
789: (struct proc *)0);
1.1 cgd 790: }
1.3 andrew 791: if (orig_resid == uio->uio_resid && orig_resid &&
792: (flags & MSG_EOR) == 0 && (so->so_state & SS_CANTRCVMORE) == 0) {
793: sbunlock(&so->so_rcv);
794: splx(s);
795: goto restart;
796: }
797:
1.1 cgd 798: if (flagsp)
799: *flagsp |= flags;
800: release:
801: sbunlock(&so->so_rcv);
802: splx(s);
803: return (error);
804: }
805:
1.14 mycroft 806: int
1.1 cgd 807: soshutdown(so, how)
808: register struct socket *so;
809: register int how;
810: {
811: register struct protosw *pr = so->so_proto;
812:
813: how++;
814: if (how & FREAD)
815: sorflush(so);
816: if (how & FWRITE)
1.22 mycroft 817: return (*pr->pr_usrreq)(so, PRU_SHUTDOWN, (struct mbuf *)0,
818: (struct mbuf *)0, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 819: return (0);
820: }
821:
1.14 mycroft 822: void
1.1 cgd 823: sorflush(so)
824: register struct socket *so;
825: {
826: register struct sockbuf *sb = &so->so_rcv;
827: register struct protosw *pr = so->so_proto;
828: register int s;
829: struct sockbuf asb;
830:
831: sb->sb_flags |= SB_NOINTR;
1.15 mycroft 832: (void) sblock(sb, M_WAITOK);
1.1 cgd 833: s = splimp();
834: socantrcvmore(so);
835: sbunlock(sb);
836: asb = *sb;
837: bzero((caddr_t)sb, sizeof (*sb));
838: splx(s);
839: if (pr->pr_flags & PR_RIGHTS && pr->pr_domain->dom_dispose)
840: (*pr->pr_domain->dom_dispose)(asb.sb_mb);
841: sbrelease(&asb);
842: }
843:
1.14 mycroft 844: int
1.1 cgd 845: sosetopt(so, level, optname, m0)
846: register struct socket *so;
847: int level, optname;
848: struct mbuf *m0;
849: {
850: int error = 0;
851: register struct mbuf *m = m0;
852:
853: if (level != SOL_SOCKET) {
854: if (so->so_proto && so->so_proto->pr_ctloutput)
855: return ((*so->so_proto->pr_ctloutput)
856: (PRCO_SETOPT, so, level, optname, &m0));
857: error = ENOPROTOOPT;
858: } else {
859: switch (optname) {
860:
861: case SO_LINGER:
862: if (m == NULL || m->m_len != sizeof (struct linger)) {
863: error = EINVAL;
864: goto bad;
865: }
866: so->so_linger = mtod(m, struct linger *)->l_linger;
867: /* fall thru... */
868:
869: case SO_DEBUG:
870: case SO_KEEPALIVE:
871: case SO_DONTROUTE:
872: case SO_USELOOPBACK:
873: case SO_BROADCAST:
874: case SO_REUSEADDR:
1.15 mycroft 875: case SO_REUSEPORT:
1.1 cgd 876: case SO_OOBINLINE:
1.26 thorpej 877: case SO_TIMESTAMP:
1.1 cgd 878: if (m == NULL || m->m_len < sizeof (int)) {
879: error = EINVAL;
880: goto bad;
881: }
882: if (*mtod(m, int *))
883: so->so_options |= optname;
884: else
885: so->so_options &= ~optname;
886: break;
887:
888: case SO_SNDBUF:
889: case SO_RCVBUF:
890: case SO_SNDLOWAT:
891: case SO_RCVLOWAT:
1.28 thorpej 892: {
893: int optval;
894:
1.1 cgd 895: if (m == NULL || m->m_len < sizeof (int)) {
896: error = EINVAL;
897: goto bad;
898: }
1.28 thorpej 899:
900: /*
901: * Values < 1 make no sense for any of these
902: * options, so disallow them.
903: */
904: optval = *mtod(m, int *);
905: if (optval < 1) {
906: error = EINVAL;
907: goto bad;
908: }
909:
1.1 cgd 910: switch (optname) {
911:
912: case SO_SNDBUF:
913: case SO_RCVBUF:
914: if (sbreserve(optname == SO_SNDBUF ?
915: &so->so_snd : &so->so_rcv,
1.28 thorpej 916: (u_long) optval) == 0) {
1.1 cgd 917: error = ENOBUFS;
918: goto bad;
919: }
920: break;
921:
1.28 thorpej 922: /*
923: * Make sure the low-water is never greater than
924: * the high-water.
925: */
1.1 cgd 926: case SO_SNDLOWAT:
1.28 thorpej 927: so->so_snd.sb_lowat =
928: (optval > so->so_snd.sb_hiwat) ?
929: so->so_snd.sb_hiwat : optval;
1.1 cgd 930: break;
931: case SO_RCVLOWAT:
1.28 thorpej 932: so->so_rcv.sb_lowat =
933: (optval > so->so_rcv.sb_hiwat) ?
934: so->so_rcv.sb_hiwat : optval;
1.1 cgd 935: break;
936: }
937: break;
1.28 thorpej 938: }
1.1 cgd 939:
940: case SO_SNDTIMEO:
941: case SO_RCVTIMEO:
942: {
943: struct timeval *tv;
944: short val;
945:
946: if (m == NULL || m->m_len < sizeof (*tv)) {
947: error = EINVAL;
948: goto bad;
949: }
950: tv = mtod(m, struct timeval *);
1.19 cgd 951: if (tv->tv_sec * hz + tv->tv_usec / tick > SHRT_MAX) {
1.1 cgd 952: error = EDOM;
953: goto bad;
954: }
955: val = tv->tv_sec * hz + tv->tv_usec / tick;
956:
957: switch (optname) {
958:
959: case SO_SNDTIMEO:
960: so->so_snd.sb_timeo = val;
961: break;
962: case SO_RCVTIMEO:
963: so->so_rcv.sb_timeo = val;
964: break;
965: }
966: break;
967: }
968:
969: default:
970: error = ENOPROTOOPT;
971: break;
972: }
1.15 mycroft 973: if (error == 0 && so->so_proto && so->so_proto->pr_ctloutput) {
974: (void) ((*so->so_proto->pr_ctloutput)
975: (PRCO_SETOPT, so, level, optname, &m0));
976: m = NULL; /* freed by protocol */
977: }
1.1 cgd 978: }
979: bad:
980: if (m)
981: (void) m_free(m);
982: return (error);
983: }
984:
1.14 mycroft 985: int
1.1 cgd 986: sogetopt(so, level, optname, mp)
987: register struct socket *so;
988: int level, optname;
989: struct mbuf **mp;
990: {
991: register struct mbuf *m;
992:
993: if (level != SOL_SOCKET) {
994: if (so->so_proto && so->so_proto->pr_ctloutput) {
995: return ((*so->so_proto->pr_ctloutput)
996: (PRCO_GETOPT, so, level, optname, mp));
997: } else
998: return (ENOPROTOOPT);
999: } else {
1000: m = m_get(M_WAIT, MT_SOOPTS);
1001: m->m_len = sizeof (int);
1002:
1003: switch (optname) {
1004:
1005: case SO_LINGER:
1006: m->m_len = sizeof (struct linger);
1007: mtod(m, struct linger *)->l_onoff =
1008: so->so_options & SO_LINGER;
1009: mtod(m, struct linger *)->l_linger = so->so_linger;
1010: break;
1011:
1012: case SO_USELOOPBACK:
1013: case SO_DONTROUTE:
1014: case SO_DEBUG:
1015: case SO_KEEPALIVE:
1016: case SO_REUSEADDR:
1.15 mycroft 1017: case SO_REUSEPORT:
1.1 cgd 1018: case SO_BROADCAST:
1019: case SO_OOBINLINE:
1.26 thorpej 1020: case SO_TIMESTAMP:
1.1 cgd 1021: *mtod(m, int *) = so->so_options & optname;
1022: break;
1023:
1024: case SO_TYPE:
1025: *mtod(m, int *) = so->so_type;
1026: break;
1027:
1028: case SO_ERROR:
1029: *mtod(m, int *) = so->so_error;
1030: so->so_error = 0;
1031: break;
1032:
1033: case SO_SNDBUF:
1034: *mtod(m, int *) = so->so_snd.sb_hiwat;
1035: break;
1036:
1037: case SO_RCVBUF:
1038: *mtod(m, int *) = so->so_rcv.sb_hiwat;
1039: break;
1040:
1041: case SO_SNDLOWAT:
1042: *mtod(m, int *) = so->so_snd.sb_lowat;
1043: break;
1044:
1045: case SO_RCVLOWAT:
1046: *mtod(m, int *) = so->so_rcv.sb_lowat;
1047: break;
1048:
1049: case SO_SNDTIMEO:
1050: case SO_RCVTIMEO:
1051: {
1052: int val = (optname == SO_SNDTIMEO ?
1053: so->so_snd.sb_timeo : so->so_rcv.sb_timeo);
1054:
1055: m->m_len = sizeof(struct timeval);
1056: mtod(m, struct timeval *)->tv_sec = val / hz;
1057: mtod(m, struct timeval *)->tv_usec =
1.27 kleink 1058: (val % hz) * tick;
1.1 cgd 1059: break;
1060: }
1061:
1062: default:
1063: (void)m_free(m);
1064: return (ENOPROTOOPT);
1065: }
1066: *mp = m;
1067: return (0);
1068: }
1069: }
1070:
1.14 mycroft 1071: void
1.1 cgd 1072: sohasoutofband(so)
1073: register struct socket *so;
1074: {
1075: struct proc *p;
1076:
1077: if (so->so_pgid < 0)
1078: gsignal(-so->so_pgid, SIGURG);
1079: else if (so->so_pgid > 0 && (p = pfind(so->so_pgid)) != 0)
1080: psignal(p, SIGURG);
1.2 cgd 1081: selwakeup(&so->so_rcv.sb_sel);
1.1 cgd 1082: }