Annotation of src/sys/kern/uipc_socket.c, Revision 1.44.6.2
1.44.6.2! itojun 1: /* $NetBSD: uipc_socket.c,v 1.44.6.1 1999/06/28 06:36:53 itojun 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:
1.44.6.1 itojun 65: #ifdef KEY
66: #include <netkey/key.h>
67: #endif
68:
1.1 cgd 69: /*
70: * Socket operation routines.
71: * These routines are called by the routines in
72: * sys_socket.c or from a system process, and
73: * implement the semantics of socket operations by
74: * switching out to the protocol specific routines.
75: */
76: /*ARGSUSED*/
1.3 andrew 77: int
1.1 cgd 78: socreate(dom, aso, type, proto)
1.11 mycroft 79: int dom;
1.1 cgd 80: struct socket **aso;
81: register int type;
82: int proto;
83: {
84: struct proc *p = curproc; /* XXX */
85: register struct protosw *prp;
86: register struct socket *so;
87: register int error;
1.39 matt 88: int s;
1.1 cgd 89:
90: if (proto)
91: prp = pffindproto(dom, proto, type);
92: else
93: prp = pffindtype(dom, type);
1.15 mycroft 94: if (prp == 0 || prp->pr_usrreq == 0)
1.1 cgd 95: return (EPROTONOSUPPORT);
96: if (prp->pr_type != type)
97: return (EPROTOTYPE);
1.39 matt 98: s = splsoftnet();
1.37 thorpej 99: so = pool_get(&socket_pool, PR_WAITOK);
1.38 perry 100: memset((caddr_t)so, 0, sizeof(*so));
1.31 thorpej 101: TAILQ_INIT(&so->so_q0);
102: TAILQ_INIT(&so->so_q);
1.1 cgd 103: so->so_type = type;
104: so->so_proto = prp;
1.33 matt 105: so->so_send = sosend;
106: so->so_receive = soreceive;
1.44 lukem 107: if (p != 0)
108: so->so_uid = p->p_ucred->cr_uid;
1.22 mycroft 109: error = (*prp->pr_usrreq)(so, PRU_ATTACH, (struct mbuf *)0,
110: (struct mbuf *)(long)proto, (struct mbuf *)0, p);
1.1 cgd 111: if (error) {
112: so->so_state |= SS_NOFDREF;
113: sofree(so);
1.39 matt 114: splx(s);
1.1 cgd 115: return (error);
116: }
1.10 deraadt 117: #ifdef COMPAT_SUNOS
1.18 christos 118: {
119: extern struct emul emul_sunos;
120: if (p->p_emul == &emul_sunos && type == SOCK_DGRAM)
121: so->so_options |= SO_BROADCAST;
122: }
1.10 deraadt 123: #endif
1.39 matt 124: splx(s);
1.1 cgd 125: *aso = so;
126: return (0);
127: }
128:
1.3 andrew 129: int
1.1 cgd 130: sobind(so, nam)
131: struct socket *so;
132: struct mbuf *nam;
133: {
1.22 mycroft 134: struct proc *p = curproc; /* XXX */
1.20 mycroft 135: int s = splsoftnet();
1.1 cgd 136: int error;
137:
1.22 mycroft 138: error = (*so->so_proto->pr_usrreq)(so, PRU_BIND, (struct mbuf *)0,
139: nam, (struct mbuf *)0, p);
1.1 cgd 140: splx(s);
141: return (error);
142: }
143:
1.3 andrew 144: int
1.1 cgd 145: solisten(so, backlog)
146: register struct socket *so;
147: int backlog;
148: {
1.20 mycroft 149: int s = splsoftnet(), error;
1.1 cgd 150:
1.22 mycroft 151: error = (*so->so_proto->pr_usrreq)(so, PRU_LISTEN, (struct mbuf *)0,
152: (struct mbuf *)0, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 153: if (error) {
154: splx(s);
155: return (error);
156: }
1.31 thorpej 157: if (so->so_q.tqh_first == NULL)
1.1 cgd 158: so->so_options |= SO_ACCEPTCONN;
159: if (backlog < 0)
160: backlog = 0;
161: so->so_qlimit = min(backlog, SOMAXCONN);
162: splx(s);
163: return (0);
164: }
165:
1.21 christos 166: void
1.1 cgd 167: sofree(so)
168: register struct socket *so;
169: {
170:
1.44.6.2! itojun 171: if (so->so_pcb || (so->so_state & SS_NOFDREF) == 0)
1.1 cgd 172: return;
1.43 mycroft 173: if (so->so_head) {
174: /*
175: * We must not decommission a socket that's on the accept(2)
176: * queue. If we do, then accept(2) may hang after select(2)
177: * indicated that the listening socket was ready.
178: */
179: if (!soqremque(so, 0))
180: return;
181: }
1.1 cgd 182: sbrelease(&so->so_snd);
183: sorflush(so);
1.37 thorpej 184: pool_put(&socket_pool, so);
1.1 cgd 185: }
186:
187: /*
188: * Close a socket on last file table reference removal.
189: * Initiate disconnect if connected.
190: * Free socket when disconnect complete.
191: */
1.3 andrew 192: int
1.1 cgd 193: soclose(so)
194: register struct socket *so;
195: {
1.41 mycroft 196: struct socket *so2;
1.20 mycroft 197: int s = splsoftnet(); /* conservative */
1.1 cgd 198: int error = 0;
199:
200: if (so->so_options & SO_ACCEPTCONN) {
1.41 mycroft 201: while ((so2 = so->so_q0.tqh_first) != 0) {
1.42 mycroft 202: (void) soqremque(so2, 0);
1.41 mycroft 203: (void) soabort(so2);
204: }
205: while ((so2 = so->so_q.tqh_first) != 0) {
1.42 mycroft 206: (void) soqremque(so2, 1);
1.41 mycroft 207: (void) soabort(so2);
208: }
1.1 cgd 209: }
1.44.6.2! itojun 210: if (so->so_pcb == 0)
1.1 cgd 211: goto discard;
212: if (so->so_state & SS_ISCONNECTED) {
213: if ((so->so_state & SS_ISDISCONNECTING) == 0) {
214: error = sodisconnect(so);
215: if (error)
216: goto drop;
217: }
218: if (so->so_options & SO_LINGER) {
219: if ((so->so_state & SS_ISDISCONNECTING) &&
220: (so->so_state & SS_NBIO))
221: goto drop;
1.21 christos 222: while (so->so_state & SS_ISCONNECTED) {
223: error = tsleep((caddr_t)&so->so_timeo,
224: PSOCK | PCATCH, netcls,
1.30 thorpej 225: so->so_linger * hz);
1.21 christos 226: if (error)
1.1 cgd 227: break;
1.21 christos 228: }
1.1 cgd 229: }
230: }
231: drop:
1.44.6.2! itojun 232: if (so->so_pcb) {
1.22 mycroft 233: int error2 = (*so->so_proto->pr_usrreq)(so, PRU_DETACH,
234: (struct mbuf *)0, (struct mbuf *)0, (struct mbuf *)0,
235: (struct proc *)0);
1.1 cgd 236: if (error == 0)
237: error = error2;
238: }
239: discard:
240: if (so->so_state & SS_NOFDREF)
241: panic("soclose: NOFDREF");
242: so->so_state |= SS_NOFDREF;
243: sofree(so);
244: splx(s);
245: return (error);
246: }
247:
248: /*
1.20 mycroft 249: * Must be called at splsoftnet...
1.1 cgd 250: */
1.3 andrew 251: int
1.1 cgd 252: soabort(so)
253: struct socket *so;
254: {
255:
1.22 mycroft 256: return (*so->so_proto->pr_usrreq)(so, PRU_ABORT, (struct mbuf *)0,
257: (struct mbuf *)0, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 258: }
259:
1.3 andrew 260: int
1.1 cgd 261: soaccept(so, nam)
262: register struct socket *so;
263: struct mbuf *nam;
264: {
1.20 mycroft 265: int s = splsoftnet();
1.1 cgd 266: int error;
267:
268: if ((so->so_state & SS_NOFDREF) == 0)
269: panic("soaccept: !NOFDREF");
270: so->so_state &= ~SS_NOFDREF;
1.41 mycroft 271: if ((so->so_state & SS_ISDISCONNECTED) == 0)
272: error = (*so->so_proto->pr_usrreq)(so, PRU_ACCEPT,
273: (struct mbuf *)0, nam, (struct mbuf *)0, (struct proc *)0);
274: else
275: error = 0;
1.1 cgd 276: splx(s);
277: return (error);
278: }
279:
1.3 andrew 280: int
1.1 cgd 281: soconnect(so, nam)
282: register struct socket *so;
283: struct mbuf *nam;
284: {
1.23 mycroft 285: struct proc *p = curproc; /* XXX */
1.1 cgd 286: int s;
287: int error;
288:
289: if (so->so_options & SO_ACCEPTCONN)
290: return (EOPNOTSUPP);
1.20 mycroft 291: s = splsoftnet();
1.1 cgd 292: /*
293: * If protocol is connection-based, can only connect once.
294: * Otherwise, if connected, try to disconnect first.
295: * This allows user to disconnect by connecting to, e.g.,
296: * a null address.
297: */
298: if (so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING) &&
299: ((so->so_proto->pr_flags & PR_CONNREQUIRED) ||
300: (error = sodisconnect(so))))
301: error = EISCONN;
302: else
303: error = (*so->so_proto->pr_usrreq)(so, PRU_CONNECT,
1.23 mycroft 304: (struct mbuf *)0, nam, (struct mbuf *)0, p);
1.1 cgd 305: splx(s);
306: return (error);
307: }
308:
1.3 andrew 309: int
1.1 cgd 310: soconnect2(so1, so2)
311: register struct socket *so1;
312: struct socket *so2;
313: {
1.20 mycroft 314: int s = splsoftnet();
1.1 cgd 315: int error;
316:
1.22 mycroft 317: error = (*so1->so_proto->pr_usrreq)(so1, PRU_CONNECT2,
318: (struct mbuf *)0, (struct mbuf *)so2, (struct mbuf *)0,
319: (struct proc *)0);
1.1 cgd 320: splx(s);
321: return (error);
322: }
323:
1.3 andrew 324: int
1.1 cgd 325: sodisconnect(so)
326: register struct socket *so;
327: {
1.20 mycroft 328: int s = splsoftnet();
1.1 cgd 329: int error;
330:
331: if ((so->so_state & SS_ISCONNECTED) == 0) {
332: error = ENOTCONN;
333: goto bad;
334: }
335: if (so->so_state & SS_ISDISCONNECTING) {
336: error = EALREADY;
337: goto bad;
338: }
1.22 mycroft 339: error = (*so->so_proto->pr_usrreq)(so, PRU_DISCONNECT,
340: (struct mbuf *)0, (struct mbuf *)0, (struct mbuf *)0,
341: (struct proc *)0);
1.1 cgd 342: bad:
343: splx(s);
344: return (error);
345: }
346:
1.15 mycroft 347: #define SBLOCKWAIT(f) (((f) & MSG_DONTWAIT) ? M_NOWAIT : M_WAITOK)
1.1 cgd 348: /*
349: * Send on a socket.
350: * If send must go all at once and message is larger than
351: * send buffering, then hard error.
352: * Lock against other senders.
353: * If must go all at once and not enough room now, then
354: * inform user that this would block and do nothing.
355: * Otherwise, if nonblocking, send as much as possible.
356: * The data to be sent is described by "uio" if nonzero,
357: * otherwise by the mbuf chain "top" (which must be null
358: * if uio is not). Data provided in mbuf chain must be small
359: * enough to send all at once.
360: *
361: * Returns nonzero on error, timeout or signal; callers
362: * must check for short counts if EINTR/ERESTART are returned.
363: * Data and control buffers are freed on return.
364: */
1.3 andrew 365: int
1.1 cgd 366: sosend(so, addr, uio, top, control, flags)
367: register struct socket *so;
368: struct mbuf *addr;
369: struct uio *uio;
370: struct mbuf *top;
371: struct mbuf *control;
372: int flags;
373: {
1.15 mycroft 374: struct proc *p = curproc; /* XXX */
1.1 cgd 375: struct mbuf **mp;
376: register struct mbuf *m;
377: register long space, len, resid;
378: int clen = 0, error, s, dontroute, mlen;
379: int atomic = sosendallatonce(so) || top;
380:
381: if (uio)
382: resid = uio->uio_resid;
383: else
384: resid = top->m_pkthdr.len;
1.7 cgd 385: /*
386: * In theory resid should be unsigned.
387: * However, space must be signed, as it might be less than 0
388: * if we over-committed, and we must use a signed comparison
389: * of space and resid. On the other hand, a negative resid
390: * causes us to loop sending 0-length segments to the protocol.
391: */
1.29 mycroft 392: if (resid < 0) {
393: error = EINVAL;
394: goto out;
395: }
1.1 cgd 396: dontroute =
397: (flags & MSG_DONTROUTE) && (so->so_options & SO_DONTROUTE) == 0 &&
398: (so->so_proto->pr_flags & PR_ATOMIC);
1.12 mycroft 399: p->p_stats->p_ru.ru_msgsnd++;
1.1 cgd 400: if (control)
401: clen = control->m_len;
402: #define snderr(errno) { error = errno; splx(s); goto release; }
403:
404: restart:
1.21 christos 405: if ((error = sblock(&so->so_snd, SBLOCKWAIT(flags))) != 0)
1.1 cgd 406: goto out;
407: do {
1.20 mycroft 408: s = splsoftnet();
1.1 cgd 409: if (so->so_state & SS_CANTSENDMORE)
410: snderr(EPIPE);
411: if (so->so_error)
412: snderr(so->so_error);
413: if ((so->so_state & SS_ISCONNECTED) == 0) {
414: if (so->so_proto->pr_flags & PR_CONNREQUIRED) {
415: if ((so->so_state & SS_ISCONFIRMING) == 0 &&
416: !(resid == 0 && clen != 0))
417: snderr(ENOTCONN);
418: } else if (addr == 0)
419: snderr(EDESTADDRREQ);
420: }
421: space = sbspace(&so->so_snd);
422: if (flags & MSG_OOB)
423: space += 1024;
1.21 christos 424: if ((atomic && resid > so->so_snd.sb_hiwat) ||
1.11 mycroft 425: clen > so->so_snd.sb_hiwat)
426: snderr(EMSGSIZE);
427: if (space < resid + clen && uio &&
1.1 cgd 428: (atomic || space < so->so_snd.sb_lowat || space < clen)) {
429: if (so->so_state & SS_NBIO)
430: snderr(EWOULDBLOCK);
431: sbunlock(&so->so_snd);
432: error = sbwait(&so->so_snd);
433: splx(s);
434: if (error)
435: goto out;
436: goto restart;
437: }
438: splx(s);
439: mp = ⊤
440: space -= clen;
441: do {
442: if (uio == NULL) {
443: /*
444: * Data is prepackaged in "top".
445: */
446: resid = 0;
447: if (flags & MSG_EOR)
448: top->m_flags |= M_EOR;
449: } else do {
450: if (top == 0) {
451: MGETHDR(m, M_WAIT, MT_DATA);
452: mlen = MHLEN;
453: m->m_pkthdr.len = 0;
454: m->m_pkthdr.rcvif = (struct ifnet *)0;
455: } else {
456: MGET(m, M_WAIT, MT_DATA);
457: mlen = MLEN;
458: }
1.15 mycroft 459: if (resid >= MINCLSIZE && space >= MCLBYTES) {
1.1 cgd 460: MCLGET(m, M_WAIT);
461: if ((m->m_flags & M_EXT) == 0)
462: goto nopages;
463: mlen = MCLBYTES;
1.15 mycroft 464: #ifdef MAPPED_MBUFS
465: len = min(MCLBYTES, resid);
466: #else
467: if (atomic && top == 0) {
468: len = min(MCLBYTES - max_hdr, resid);
469: m->m_data += max_hdr;
470: } else
471: len = min(MCLBYTES, resid);
472: #endif
1.25 explorer 473: space -= len;
1.1 cgd 474: } else {
475: nopages:
476: len = min(min(mlen, resid), space);
1.15 mycroft 477: space -= len;
1.1 cgd 478: /*
479: * For datagram protocols, leave room
480: * for protocol headers in first mbuf.
481: */
482: if (atomic && top == 0 && len < mlen)
483: MH_ALIGN(m, len);
484: }
485: error = uiomove(mtod(m, caddr_t), (int)len, uio);
486: resid = uio->uio_resid;
487: m->m_len = len;
488: *mp = m;
489: top->m_pkthdr.len += len;
490: if (error)
491: goto release;
492: mp = &m->m_next;
493: if (resid <= 0) {
494: if (flags & MSG_EOR)
495: top->m_flags |= M_EOR;
496: break;
497: }
498: } while (space > 0 && atomic);
499: if (dontroute)
500: so->so_options |= SO_DONTROUTE;
1.40 thorpej 501: if (resid > 0)
502: so->so_state |= SS_MORETOCOME;
1.20 mycroft 503: s = splsoftnet(); /* XXX */
1.22 mycroft 504: error = (*so->so_proto->pr_usrreq)(so,
1.24 mycroft 505: (flags & MSG_OOB) ? PRU_SENDOOB : PRU_SEND,
506: top, addr, control, p);
1.1 cgd 507: splx(s);
508: if (dontroute)
509: so->so_options &= ~SO_DONTROUTE;
1.40 thorpej 510: if (resid > 0)
511: so->so_state &= ~SS_MORETOCOME;
1.1 cgd 512: clen = 0;
513: control = 0;
514: top = 0;
515: mp = ⊤
516: if (error)
517: goto release;
518: } while (resid && space > 0);
519: } while (resid);
520:
521: release:
522: sbunlock(&so->so_snd);
523: out:
524: if (top)
525: m_freem(top);
526: if (control)
527: m_freem(control);
528: return (error);
529: }
530:
531: /*
532: * Implement receive operations on a socket.
533: * We depend on the way that records are added to the sockbuf
534: * by sbappend*. In particular, each record (mbufs linked through m_next)
535: * must begin with an address if the protocol so specifies,
536: * followed by an optional mbuf or mbufs containing ancillary data,
537: * and then zero or more mbufs of data.
538: * In order to avoid blocking network interrupts for the entire time here,
539: * we splx() while doing the actual copy to user space.
540: * Although the sockbuf is locked, new data may still be appended,
541: * and thus we must maintain consistency of the sockbuf during that time.
542: *
543: * The caller may receive the data as a single mbuf chain by supplying
544: * an mbuf **mp0 for use in returning the chain. The uio is then used
545: * only for the count in uio_resid.
546: */
1.3 andrew 547: int
1.1 cgd 548: soreceive(so, paddr, uio, mp0, controlp, flagsp)
549: register struct socket *so;
550: struct mbuf **paddr;
551: struct uio *uio;
552: struct mbuf **mp0;
553: struct mbuf **controlp;
554: int *flagsp;
555: {
556: register struct mbuf *m, **mp;
557: register int flags, len, error, s, offset;
558: struct protosw *pr = so->so_proto;
559: struct mbuf *nextrecord;
1.21 christos 560: int moff, type = 0;
1.3 andrew 561: int orig_resid = uio->uio_resid;
1.1 cgd 562:
563: mp = mp0;
564: if (paddr)
565: *paddr = 0;
566: if (controlp)
567: *controlp = 0;
568: if (flagsp)
569: flags = *flagsp &~ MSG_EOR;
570: else
571: flags = 0;
572: if (flags & MSG_OOB) {
573: m = m_get(M_WAIT, MT_DATA);
1.17 cgd 574: error = (*pr->pr_usrreq)(so, PRU_RCVOOB, m,
1.22 mycroft 575: (struct mbuf *)(long)(flags & MSG_PEEK), (struct mbuf *)0,
576: (struct proc *)0);
1.1 cgd 577: if (error)
578: goto bad;
579: do {
580: error = uiomove(mtod(m, caddr_t),
581: (int) min(uio->uio_resid, m->m_len), uio);
582: m = m_free(m);
583: } while (uio->uio_resid && error == 0 && m);
584: bad:
585: if (m)
586: m_freem(m);
587: return (error);
588: }
589: if (mp)
590: *mp = (struct mbuf *)0;
591: if (so->so_state & SS_ISCONFIRMING && uio->uio_resid)
1.22 mycroft 592: (*pr->pr_usrreq)(so, PRU_RCVD, (struct mbuf *)0,
593: (struct mbuf *)0, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 594:
595: restart:
1.21 christos 596: if ((error = sblock(&so->so_rcv, SBLOCKWAIT(flags))) != 0)
1.1 cgd 597: return (error);
1.20 mycroft 598: s = splsoftnet();
1.1 cgd 599:
600: m = so->so_rcv.sb_mb;
601: /*
602: * If we have less data than requested, block awaiting more
603: * (subject to any timeout) if:
1.15 mycroft 604: * 1. the current count is less than the low water mark,
1.1 cgd 605: * 2. MSG_WAITALL is set, and it is possible to do the entire
1.15 mycroft 606: * receive operation at once if we block (resid <= hiwat), or
607: * 3. MSG_DONTWAIT is not set.
1.1 cgd 608: * If MSG_WAITALL is set but resid is larger than the receive buffer,
609: * we have to do the receive in sections, and thus risk returning
610: * a short count if a timeout or signal occurs after we start.
611: */
1.21 christos 612: if (m == 0 || (((flags & MSG_DONTWAIT) == 0 &&
1.15 mycroft 613: so->so_rcv.sb_cc < uio->uio_resid) &&
1.1 cgd 614: (so->so_rcv.sb_cc < so->so_rcv.sb_lowat ||
615: ((flags & MSG_WAITALL) && uio->uio_resid <= so->so_rcv.sb_hiwat)) &&
1.21 christos 616: m->m_nextpkt == 0 && (pr->pr_flags & PR_ATOMIC) == 0)) {
1.1 cgd 617: #ifdef DIAGNOSTIC
618: if (m == 0 && so->so_rcv.sb_cc)
619: panic("receive 1");
620: #endif
621: if (so->so_error) {
622: if (m)
1.15 mycroft 623: goto dontblock;
1.1 cgd 624: error = so->so_error;
625: if ((flags & MSG_PEEK) == 0)
626: so->so_error = 0;
627: goto release;
628: }
629: if (so->so_state & SS_CANTRCVMORE) {
630: if (m)
1.15 mycroft 631: goto dontblock;
1.1 cgd 632: else
633: goto release;
634: }
635: for (; m; m = m->m_next)
636: if (m->m_type == MT_OOBDATA || (m->m_flags & M_EOR)) {
637: m = so->so_rcv.sb_mb;
638: goto dontblock;
639: }
640: if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) == 0 &&
641: (so->so_proto->pr_flags & PR_CONNREQUIRED)) {
642: error = ENOTCONN;
643: goto release;
644: }
645: if (uio->uio_resid == 0)
646: goto release;
1.15 mycroft 647: if ((so->so_state & SS_NBIO) || (flags & MSG_DONTWAIT)) {
1.1 cgd 648: error = EWOULDBLOCK;
649: goto release;
650: }
651: sbunlock(&so->so_rcv);
652: error = sbwait(&so->so_rcv);
653: splx(s);
654: if (error)
655: return (error);
656: goto restart;
657: }
658: dontblock:
1.15 mycroft 659: #ifdef notyet /* XXXX */
660: if (uio->uio_procp)
661: uio->uio_procp->p_stats->p_ru.ru_msgrcv++;
662: #endif
1.1 cgd 663: nextrecord = m->m_nextpkt;
664: if (pr->pr_flags & PR_ADDR) {
665: #ifdef DIAGNOSTIC
666: if (m->m_type != MT_SONAME)
667: panic("receive 1a");
668: #endif
1.3 andrew 669: orig_resid = 0;
1.1 cgd 670: if (flags & MSG_PEEK) {
671: if (paddr)
672: *paddr = m_copy(m, 0, m->m_len);
673: m = m->m_next;
674: } else {
675: sbfree(&so->so_rcv, m);
676: if (paddr) {
677: *paddr = m;
678: so->so_rcv.sb_mb = m->m_next;
679: m->m_next = 0;
680: m = so->so_rcv.sb_mb;
681: } else {
682: MFREE(m, so->so_rcv.sb_mb);
683: m = so->so_rcv.sb_mb;
684: }
685: }
686: }
687: while (m && m->m_type == MT_CONTROL && error == 0) {
688: if (flags & MSG_PEEK) {
689: if (controlp)
690: *controlp = m_copy(m, 0, m->m_len);
691: m = m->m_next;
692: } else {
693: sbfree(&so->so_rcv, m);
694: if (controlp) {
695: if (pr->pr_domain->dom_externalize &&
696: mtod(m, struct cmsghdr *)->cmsg_type ==
697: SCM_RIGHTS)
698: error = (*pr->pr_domain->dom_externalize)(m);
699: *controlp = m;
700: so->so_rcv.sb_mb = m->m_next;
701: m->m_next = 0;
702: m = so->so_rcv.sb_mb;
703: } else {
704: MFREE(m, so->so_rcv.sb_mb);
705: m = so->so_rcv.sb_mb;
706: }
707: }
1.3 andrew 708: if (controlp) {
709: orig_resid = 0;
1.1 cgd 710: controlp = &(*controlp)->m_next;
1.3 andrew 711: }
1.1 cgd 712: }
713: if (m) {
714: if ((flags & MSG_PEEK) == 0)
715: m->m_nextpkt = nextrecord;
716: type = m->m_type;
717: if (type == MT_OOBDATA)
718: flags |= MSG_OOB;
719: }
720: moff = 0;
721: offset = 0;
722: while (m && uio->uio_resid > 0 && error == 0) {
723: if (m->m_type == MT_OOBDATA) {
724: if (type != MT_OOBDATA)
725: break;
726: } else if (type == MT_OOBDATA)
727: break;
728: #ifdef DIAGNOSTIC
729: else if (m->m_type != MT_DATA && m->m_type != MT_HEADER)
730: panic("receive 3");
731: #endif
732: so->so_state &= ~SS_RCVATMARK;
733: len = uio->uio_resid;
734: if (so->so_oobmark && len > so->so_oobmark - offset)
735: len = so->so_oobmark - offset;
736: if (len > m->m_len - moff)
737: len = m->m_len - moff;
738: /*
739: * If mp is set, just pass back the mbufs.
740: * Otherwise copy them out via the uio, then free.
741: * Sockbuf must be consistent here (points to current mbuf,
742: * it points to next record) when we drop priority;
743: * we must note any additions to the sockbuf when we
744: * block interrupts again.
745: */
746: if (mp == 0) {
747: splx(s);
748: error = uiomove(mtod(m, caddr_t) + moff, (int)len, uio);
1.20 mycroft 749: s = splsoftnet();
1.1 cgd 750: } else
751: uio->uio_resid -= len;
752: if (len == m->m_len - moff) {
753: if (m->m_flags & M_EOR)
754: flags |= MSG_EOR;
755: if (flags & MSG_PEEK) {
756: m = m->m_next;
757: moff = 0;
758: } else {
759: nextrecord = m->m_nextpkt;
760: sbfree(&so->so_rcv, m);
761: if (mp) {
762: *mp = m;
763: mp = &m->m_next;
764: so->so_rcv.sb_mb = m = m->m_next;
765: *mp = (struct mbuf *)0;
766: } else {
767: MFREE(m, so->so_rcv.sb_mb);
768: m = so->so_rcv.sb_mb;
769: }
770: if (m)
771: m->m_nextpkt = nextrecord;
772: }
773: } else {
774: if (flags & MSG_PEEK)
775: moff += len;
776: else {
777: if (mp)
778: *mp = m_copym(m, 0, len, M_WAIT);
779: m->m_data += len;
780: m->m_len -= len;
781: so->so_rcv.sb_cc -= len;
782: }
783: }
784: if (so->so_oobmark) {
785: if ((flags & MSG_PEEK) == 0) {
786: so->so_oobmark -= len;
787: if (so->so_oobmark == 0) {
788: so->so_state |= SS_RCVATMARK;
789: break;
790: }
1.7 cgd 791: } else {
1.1 cgd 792: offset += len;
1.7 cgd 793: if (offset == so->so_oobmark)
794: break;
795: }
1.1 cgd 796: }
797: if (flags & MSG_EOR)
798: break;
799: /*
800: * If the MSG_WAITALL flag is set (for non-atomic socket),
801: * we must not quit until "uio->uio_resid == 0" or an error
802: * termination. If a signal/timeout occurs, return
803: * with a short count but without error.
804: * Keep sockbuf locked against other readers.
805: */
806: while (flags & MSG_WAITALL && m == 0 && uio->uio_resid > 0 &&
1.3 andrew 807: !sosendallatonce(so) && !nextrecord) {
1.1 cgd 808: if (so->so_error || so->so_state & SS_CANTRCVMORE)
809: break;
810: error = sbwait(&so->so_rcv);
811: if (error) {
812: sbunlock(&so->so_rcv);
813: splx(s);
814: return (0);
815: }
1.21 christos 816: if ((m = so->so_rcv.sb_mb) != NULL)
1.1 cgd 817: nextrecord = m->m_nextpkt;
818: }
819: }
1.3 andrew 820:
821: if (m && pr->pr_flags & PR_ATOMIC) {
822: flags |= MSG_TRUNC;
823: if ((flags & MSG_PEEK) == 0)
824: (void) sbdroprecord(&so->so_rcv);
825: }
1.1 cgd 826: if ((flags & MSG_PEEK) == 0) {
827: if (m == 0)
828: so->so_rcv.sb_mb = nextrecord;
1.44.6.2! itojun 829: if (pr->pr_flags & PR_WANTRCVD && so->so_pcb)
1.22 mycroft 830: (*pr->pr_usrreq)(so, PRU_RCVD, (struct mbuf *)0,
831: (struct mbuf *)(long)flags, (struct mbuf *)0,
832: (struct proc *)0);
1.1 cgd 833: }
1.3 andrew 834: if (orig_resid == uio->uio_resid && orig_resid &&
835: (flags & MSG_EOR) == 0 && (so->so_state & SS_CANTRCVMORE) == 0) {
836: sbunlock(&so->so_rcv);
837: splx(s);
838: goto restart;
839: }
840:
1.1 cgd 841: if (flagsp)
842: *flagsp |= flags;
843: release:
844: sbunlock(&so->so_rcv);
845: splx(s);
846: return (error);
847: }
848:
1.14 mycroft 849: int
1.1 cgd 850: soshutdown(so, how)
1.34 kleink 851: struct socket *so;
852: int how;
1.1 cgd 853: {
1.34 kleink 854: struct protosw *pr = so->so_proto;
855:
856: if (!(how == SHUT_RD || how == SHUT_WR || how == SHUT_RDWR))
857: return (EINVAL);
1.1 cgd 858:
1.34 kleink 859: if (how == SHUT_RD || how == SHUT_RDWR)
1.1 cgd 860: sorflush(so);
1.34 kleink 861: if (how == SHUT_WR || how == SHUT_RDWR)
1.22 mycroft 862: return (*pr->pr_usrreq)(so, PRU_SHUTDOWN, (struct mbuf *)0,
863: (struct mbuf *)0, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 864: return (0);
865: }
866:
1.14 mycroft 867: void
1.1 cgd 868: sorflush(so)
869: register struct socket *so;
870: {
871: register struct sockbuf *sb = &so->so_rcv;
872: register struct protosw *pr = so->so_proto;
873: register int s;
874: struct sockbuf asb;
875:
876: sb->sb_flags |= SB_NOINTR;
1.15 mycroft 877: (void) sblock(sb, M_WAITOK);
1.1 cgd 878: s = splimp();
879: socantrcvmore(so);
880: sbunlock(sb);
881: asb = *sb;
1.38 perry 882: memset((caddr_t)sb, 0, sizeof(*sb));
1.1 cgd 883: splx(s);
884: if (pr->pr_flags & PR_RIGHTS && pr->pr_domain->dom_dispose)
885: (*pr->pr_domain->dom_dispose)(asb.sb_mb);
886: sbrelease(&asb);
887: }
888:
1.14 mycroft 889: int
1.1 cgd 890: sosetopt(so, level, optname, m0)
891: register struct socket *so;
892: int level, optname;
893: struct mbuf *m0;
894: {
895: int error = 0;
896: register struct mbuf *m = m0;
897:
898: if (level != SOL_SOCKET) {
899: if (so->so_proto && so->so_proto->pr_ctloutput)
900: return ((*so->so_proto->pr_ctloutput)
901: (PRCO_SETOPT, so, level, optname, &m0));
902: error = ENOPROTOOPT;
903: } else {
904: switch (optname) {
905:
906: case SO_LINGER:
1.36 perry 907: if (m == NULL || m->m_len != sizeof(struct linger)) {
1.1 cgd 908: error = EINVAL;
909: goto bad;
910: }
911: so->so_linger = mtod(m, struct linger *)->l_linger;
912: /* fall thru... */
913:
914: case SO_DEBUG:
915: case SO_KEEPALIVE:
916: case SO_DONTROUTE:
917: case SO_USELOOPBACK:
918: case SO_BROADCAST:
919: case SO_REUSEADDR:
1.15 mycroft 920: case SO_REUSEPORT:
1.1 cgd 921: case SO_OOBINLINE:
1.26 thorpej 922: case SO_TIMESTAMP:
1.36 perry 923: if (m == NULL || m->m_len < sizeof(int)) {
1.1 cgd 924: error = EINVAL;
925: goto bad;
926: }
927: if (*mtod(m, int *))
928: so->so_options |= optname;
929: else
930: so->so_options &= ~optname;
931: break;
932:
933: case SO_SNDBUF:
934: case SO_RCVBUF:
935: case SO_SNDLOWAT:
936: case SO_RCVLOWAT:
1.28 thorpej 937: {
938: int optval;
939:
1.36 perry 940: if (m == NULL || m->m_len < sizeof(int)) {
1.1 cgd 941: error = EINVAL;
942: goto bad;
943: }
1.28 thorpej 944:
945: /*
946: * Values < 1 make no sense for any of these
947: * options, so disallow them.
948: */
949: optval = *mtod(m, int *);
950: if (optval < 1) {
951: error = EINVAL;
952: goto bad;
953: }
954:
1.1 cgd 955: switch (optname) {
956:
957: case SO_SNDBUF:
958: case SO_RCVBUF:
959: if (sbreserve(optname == SO_SNDBUF ?
960: &so->so_snd : &so->so_rcv,
1.28 thorpej 961: (u_long) optval) == 0) {
1.1 cgd 962: error = ENOBUFS;
963: goto bad;
964: }
965: break;
966:
1.28 thorpej 967: /*
968: * Make sure the low-water is never greater than
969: * the high-water.
970: */
1.1 cgd 971: case SO_SNDLOWAT:
1.28 thorpej 972: so->so_snd.sb_lowat =
973: (optval > so->so_snd.sb_hiwat) ?
974: so->so_snd.sb_hiwat : optval;
1.1 cgd 975: break;
976: case SO_RCVLOWAT:
1.28 thorpej 977: so->so_rcv.sb_lowat =
978: (optval > so->so_rcv.sb_hiwat) ?
979: so->so_rcv.sb_hiwat : optval;
1.1 cgd 980: break;
981: }
982: break;
1.28 thorpej 983: }
1.1 cgd 984:
985: case SO_SNDTIMEO:
986: case SO_RCVTIMEO:
987: {
988: struct timeval *tv;
989: short val;
990:
1.36 perry 991: if (m == NULL || m->m_len < sizeof(*tv)) {
1.1 cgd 992: error = EINVAL;
993: goto bad;
994: }
995: tv = mtod(m, struct timeval *);
1.19 cgd 996: if (tv->tv_sec * hz + tv->tv_usec / tick > SHRT_MAX) {
1.1 cgd 997: error = EDOM;
998: goto bad;
999: }
1000: val = tv->tv_sec * hz + tv->tv_usec / tick;
1001:
1002: switch (optname) {
1003:
1004: case SO_SNDTIMEO:
1005: so->so_snd.sb_timeo = val;
1006: break;
1007: case SO_RCVTIMEO:
1008: so->so_rcv.sb_timeo = val;
1009: break;
1010: }
1011: break;
1012: }
1013:
1014: default:
1015: error = ENOPROTOOPT;
1016: break;
1017: }
1.15 mycroft 1018: if (error == 0 && so->so_proto && so->so_proto->pr_ctloutput) {
1019: (void) ((*so->so_proto->pr_ctloutput)
1020: (PRCO_SETOPT, so, level, optname, &m0));
1021: m = NULL; /* freed by protocol */
1022: }
1.1 cgd 1023: }
1024: bad:
1025: if (m)
1026: (void) m_free(m);
1027: return (error);
1028: }
1029:
1.14 mycroft 1030: int
1.1 cgd 1031: sogetopt(so, level, optname, mp)
1032: register struct socket *so;
1033: int level, optname;
1034: struct mbuf **mp;
1035: {
1036: register struct mbuf *m;
1037:
1038: if (level != SOL_SOCKET) {
1039: if (so->so_proto && so->so_proto->pr_ctloutput) {
1040: return ((*so->so_proto->pr_ctloutput)
1041: (PRCO_GETOPT, so, level, optname, mp));
1042: } else
1043: return (ENOPROTOOPT);
1044: } else {
1045: m = m_get(M_WAIT, MT_SOOPTS);
1.36 perry 1046: m->m_len = sizeof(int);
1.1 cgd 1047:
1048: switch (optname) {
1049:
1050: case SO_LINGER:
1.36 perry 1051: m->m_len = sizeof(struct linger);
1.1 cgd 1052: mtod(m, struct linger *)->l_onoff =
1053: so->so_options & SO_LINGER;
1054: mtod(m, struct linger *)->l_linger = so->so_linger;
1055: break;
1056:
1057: case SO_USELOOPBACK:
1058: case SO_DONTROUTE:
1059: case SO_DEBUG:
1060: case SO_KEEPALIVE:
1061: case SO_REUSEADDR:
1.15 mycroft 1062: case SO_REUSEPORT:
1.1 cgd 1063: case SO_BROADCAST:
1064: case SO_OOBINLINE:
1.26 thorpej 1065: case SO_TIMESTAMP:
1.1 cgd 1066: *mtod(m, int *) = so->so_options & optname;
1067: break;
1068:
1069: case SO_TYPE:
1070: *mtod(m, int *) = so->so_type;
1071: break;
1072:
1073: case SO_ERROR:
1074: *mtod(m, int *) = so->so_error;
1075: so->so_error = 0;
1076: break;
1077:
1078: case SO_SNDBUF:
1079: *mtod(m, int *) = so->so_snd.sb_hiwat;
1080: break;
1081:
1082: case SO_RCVBUF:
1083: *mtod(m, int *) = so->so_rcv.sb_hiwat;
1084: break;
1085:
1086: case SO_SNDLOWAT:
1087: *mtod(m, int *) = so->so_snd.sb_lowat;
1088: break;
1089:
1090: case SO_RCVLOWAT:
1091: *mtod(m, int *) = so->so_rcv.sb_lowat;
1092: break;
1093:
1094: case SO_SNDTIMEO:
1095: case SO_RCVTIMEO:
1096: {
1097: int val = (optname == SO_SNDTIMEO ?
1098: so->so_snd.sb_timeo : so->so_rcv.sb_timeo);
1099:
1100: m->m_len = sizeof(struct timeval);
1101: mtod(m, struct timeval *)->tv_sec = val / hz;
1102: mtod(m, struct timeval *)->tv_usec =
1.27 kleink 1103: (val % hz) * tick;
1.1 cgd 1104: break;
1105: }
1106:
1107: default:
1108: (void)m_free(m);
1109: return (ENOPROTOOPT);
1110: }
1111: *mp = m;
1112: return (0);
1113: }
1114: }
1115:
1.14 mycroft 1116: void
1.1 cgd 1117: sohasoutofband(so)
1118: register struct socket *so;
1119: {
1120: struct proc *p;
1121:
1122: if (so->so_pgid < 0)
1123: gsignal(-so->so_pgid, SIGURG);
1124: else if (so->so_pgid > 0 && (p = pfind(so->so_pgid)) != 0)
1125: psignal(p, SIGURG);
1.2 cgd 1126: selwakeup(&so->so_rcv.sb_sel);
1.1 cgd 1127: }
CVSweb <webmaster@jp.NetBSD.org>