![[BACK]](/icons/back.gif){kind=icon}
Return to uipc_socket.c CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [cvs.NetBSD.org] / src / sys / kern|
Return to uipc_socket.c CVS log | Up to [cvs.NetBSD.org] / src / sys / kern |
1.34 ! kleink 1: /* $NetBSD: uipc_socket.c,v 1.33 1998/04/25 17:35:18 matt 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: */
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));
1.31 thorpej 82: TAILQ_INIT(&so->so_q0);
83: TAILQ_INIT(&so->so_q);
1.1 cgd 84: so->so_type = type;
85: so->so_proto = prp;
1.33 matt 86: so->so_send = sosend;
87: so->so_receive = soreceive;
1.22 mycroft 88: error = (*prp->pr_usrreq)(so, PRU_ATTACH, (struct mbuf *)0,
89: (struct mbuf *)(long)proto, (struct mbuf *)0, p);
1.1 cgd 90: if (error) {
91: so->so_state |= SS_NOFDREF;
92: sofree(so);
93: return (error);
94: }
1.10 deraadt 95: #ifdef COMPAT_SUNOS
1.18 christos 96: {
97: extern struct emul emul_sunos;
98: if (p->p_emul == &emul_sunos && type == SOCK_DGRAM)
99: so->so_options |= SO_BROADCAST;
100: }
1.10 deraadt 101: #endif
1.1 cgd 102: *aso = so;
103: return (0);
104: }
105:
1.3 andrew 106: int
1.1 cgd 107: sobind(so, nam)
108: struct socket *so;
109: struct mbuf *nam;
110: {
1.22 mycroft 111: struct proc *p = curproc; /* XXX */
1.20 mycroft 112: int s = splsoftnet();
1.1 cgd 113: int error;
114:
1.22 mycroft 115: error = (*so->so_proto->pr_usrreq)(so, PRU_BIND, (struct mbuf *)0,
116: nam, (struct mbuf *)0, p);
1.1 cgd 117: splx(s);
118: return (error);
119: }
120:
1.3 andrew 121: int
1.1 cgd 122: solisten(so, backlog)
123: register struct socket *so;
124: int backlog;
125: {
1.20 mycroft 126: int s = splsoftnet(), error;
1.1 cgd 127:
1.22 mycroft 128: error = (*so->so_proto->pr_usrreq)(so, PRU_LISTEN, (struct mbuf *)0,
129: (struct mbuf *)0, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 130: if (error) {
131: splx(s);
132: return (error);
133: }
1.31 thorpej 134: if (so->so_q.tqh_first == NULL)
1.1 cgd 135: so->so_options |= SO_ACCEPTCONN;
136: if (backlog < 0)
137: backlog = 0;
138: so->so_qlimit = min(backlog, SOMAXCONN);
139: splx(s);
140: return (0);
141: }
142:
1.21 christos 143: void
1.1 cgd 144: sofree(so)
145: register struct socket *so;
146: {
147:
148: if (so->so_pcb || (so->so_state & SS_NOFDREF) == 0)
149: return;
150: if (so->so_head) {
151: if (!soqremque(so, 0) && !soqremque(so, 1))
152: panic("sofree dq");
153: so->so_head = 0;
154: }
155: sbrelease(&so->so_snd);
156: sorflush(so);
157: FREE(so, M_SOCKET);
158: }
159:
160: /*
161: * Close a socket on last file table reference removal.
162: * Initiate disconnect if connected.
163: * Free socket when disconnect complete.
164: */
1.3 andrew 165: int
1.1 cgd 166: soclose(so)
167: register struct socket *so;
168: {
1.20 mycroft 169: int s = splsoftnet(); /* conservative */
1.1 cgd 170: int error = 0;
171:
172: if (so->so_options & SO_ACCEPTCONN) {
1.31 thorpej 173: while (so->so_q0.tqh_first)
174: (void) soabort(so->so_q0.tqh_first);
175: while (so->so_q.tqh_first)
176: (void) soabort(so->so_q.tqh_first);
1.1 cgd 177: }
178: if (so->so_pcb == 0)
179: goto discard;
180: if (so->so_state & SS_ISCONNECTED) {
181: if ((so->so_state & SS_ISDISCONNECTING) == 0) {
182: error = sodisconnect(so);
183: if (error)
184: goto drop;
185: }
186: if (so->so_options & SO_LINGER) {
187: if ((so->so_state & SS_ISDISCONNECTING) &&
188: (so->so_state & SS_NBIO))
189: goto drop;
1.21 christos 190: while (so->so_state & SS_ISCONNECTED) {
191: error = tsleep((caddr_t)&so->so_timeo,
192: PSOCK | PCATCH, netcls,
1.30 thorpej 193: so->so_linger * hz);
1.21 christos 194: if (error)
1.1 cgd 195: break;
1.21 christos 196: }
1.1 cgd 197: }
198: }
199: drop:
200: if (so->so_pcb) {
1.22 mycroft 201: int error2 = (*so->so_proto->pr_usrreq)(so, PRU_DETACH,
202: (struct mbuf *)0, (struct mbuf *)0, (struct mbuf *)0,
203: (struct proc *)0);
1.1 cgd 204: if (error == 0)
205: error = error2;
206: }
207: discard:
208: if (so->so_state & SS_NOFDREF)
209: panic("soclose: NOFDREF");
210: so->so_state |= SS_NOFDREF;
211: sofree(so);
212: splx(s);
213: return (error);
214: }
215:
216: /*
1.20 mycroft 217: * Must be called at splsoftnet...
1.1 cgd 218: */
1.3 andrew 219: int
1.1 cgd 220: soabort(so)
221: struct socket *so;
222: {
223:
1.22 mycroft 224: return (*so->so_proto->pr_usrreq)(so, PRU_ABORT, (struct mbuf *)0,
225: (struct mbuf *)0, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 226: }
227:
1.3 andrew 228: int
1.1 cgd 229: soaccept(so, nam)
230: register struct socket *so;
231: struct mbuf *nam;
232: {
1.20 mycroft 233: int s = splsoftnet();
1.1 cgd 234: int error;
235:
236: if ((so->so_state & SS_NOFDREF) == 0)
237: panic("soaccept: !NOFDREF");
238: so->so_state &= ~SS_NOFDREF;
1.22 mycroft 239: error = (*so->so_proto->pr_usrreq)(so, PRU_ACCEPT, (struct mbuf *)0,
240: nam, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 241: splx(s);
242: return (error);
243: }
244:
1.3 andrew 245: int
1.1 cgd 246: soconnect(so, nam)
247: register struct socket *so;
248: struct mbuf *nam;
249: {
1.23 mycroft 250: struct proc *p = curproc; /* XXX */
1.1 cgd 251: int s;
252: int error;
253:
254: if (so->so_options & SO_ACCEPTCONN)
255: return (EOPNOTSUPP);
1.20 mycroft 256: s = splsoftnet();
1.1 cgd 257: /*
258: * If protocol is connection-based, can only connect once.
259: * Otherwise, if connected, try to disconnect first.
260: * This allows user to disconnect by connecting to, e.g.,
261: * a null address.
262: */
263: if (so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING) &&
264: ((so->so_proto->pr_flags & PR_CONNREQUIRED) ||
265: (error = sodisconnect(so))))
266: error = EISCONN;
267: else
268: error = (*so->so_proto->pr_usrreq)(so, PRU_CONNECT,
1.23 mycroft 269: (struct mbuf *)0, nam, (struct mbuf *)0, p);
1.1 cgd 270: splx(s);
271: return (error);
272: }
273:
1.3 andrew 274: int
1.1 cgd 275: soconnect2(so1, so2)
276: register struct socket *so1;
277: struct socket *so2;
278: {
1.20 mycroft 279: int s = splsoftnet();
1.1 cgd 280: int error;
281:
1.22 mycroft 282: error = (*so1->so_proto->pr_usrreq)(so1, PRU_CONNECT2,
283: (struct mbuf *)0, (struct mbuf *)so2, (struct mbuf *)0,
284: (struct proc *)0);
1.1 cgd 285: splx(s);
286: return (error);
287: }
288:
1.3 andrew 289: int
1.1 cgd 290: sodisconnect(so)
291: register struct socket *so;
292: {
1.20 mycroft 293: int s = splsoftnet();
1.1 cgd 294: int error;
295:
296: if ((so->so_state & SS_ISCONNECTED) == 0) {
297: error = ENOTCONN;
298: goto bad;
299: }
300: if (so->so_state & SS_ISDISCONNECTING) {
301: error = EALREADY;
302: goto bad;
303: }
1.22 mycroft 304: error = (*so->so_proto->pr_usrreq)(so, PRU_DISCONNECT,
305: (struct mbuf *)0, (struct mbuf *)0, (struct mbuf *)0,
306: (struct proc *)0);
1.1 cgd 307: bad:
308: splx(s);
309: return (error);
310: }
311:
1.15 mycroft 312: #define SBLOCKWAIT(f) (((f) & MSG_DONTWAIT) ? M_NOWAIT : M_WAITOK)
1.1 cgd 313: /*
314: * Send on a socket.
315: * If send must go all at once and message is larger than
316: * send buffering, then hard error.
317: * Lock against other senders.
318: * If must go all at once and not enough room now, then
319: * inform user that this would block and do nothing.
320: * Otherwise, if nonblocking, send as much as possible.
321: * The data to be sent is described by "uio" if nonzero,
322: * otherwise by the mbuf chain "top" (which must be null
323: * if uio is not). Data provided in mbuf chain must be small
324: * enough to send all at once.
325: *
326: * Returns nonzero on error, timeout or signal; callers
327: * must check for short counts if EINTR/ERESTART are returned.
328: * Data and control buffers are freed on return.
329: */
1.3 andrew 330: int
1.1 cgd 331: sosend(so, addr, uio, top, control, flags)
332: register struct socket *so;
333: struct mbuf *addr;
334: struct uio *uio;
335: struct mbuf *top;
336: struct mbuf *control;
337: int flags;
338: {
1.15 mycroft 339: struct proc *p = curproc; /* XXX */
1.1 cgd 340: struct mbuf **mp;
341: register struct mbuf *m;
342: register long space, len, resid;
343: int clen = 0, error, s, dontroute, mlen;
344: int atomic = sosendallatonce(so) || top;
345:
346: if (uio)
347: resid = uio->uio_resid;
348: else
349: resid = top->m_pkthdr.len;
1.7 cgd 350: /*
351: * In theory resid should be unsigned.
352: * However, space must be signed, as it might be less than 0
353: * if we over-committed, and we must use a signed comparison
354: * of space and resid. On the other hand, a negative resid
355: * causes us to loop sending 0-length segments to the protocol.
356: */
1.29 mycroft 357: if (resid < 0) {
358: error = EINVAL;
359: goto out;
360: }
1.1 cgd 361: dontroute =
362: (flags & MSG_DONTROUTE) && (so->so_options & SO_DONTROUTE) == 0 &&
363: (so->so_proto->pr_flags & PR_ATOMIC);
1.12 mycroft 364: p->p_stats->p_ru.ru_msgsnd++;
1.1 cgd 365: if (control)
366: clen = control->m_len;
367: #define snderr(errno) { error = errno; splx(s); goto release; }
368:
369: restart:
1.21 christos 370: if ((error = sblock(&so->so_snd, SBLOCKWAIT(flags))) != 0)
1.1 cgd 371: goto out;
372: do {
1.20 mycroft 373: s = splsoftnet();
1.1 cgd 374: if (so->so_state & SS_CANTSENDMORE)
375: snderr(EPIPE);
376: if (so->so_error)
377: snderr(so->so_error);
378: if ((so->so_state & SS_ISCONNECTED) == 0) {
379: if (so->so_proto->pr_flags & PR_CONNREQUIRED) {
380: if ((so->so_state & SS_ISCONFIRMING) == 0 &&
381: !(resid == 0 && clen != 0))
382: snderr(ENOTCONN);
383: } else if (addr == 0)
384: snderr(EDESTADDRREQ);
385: }
386: space = sbspace(&so->so_snd);
387: if (flags & MSG_OOB)
388: space += 1024;
1.21 christos 389: if ((atomic && resid > so->so_snd.sb_hiwat) ||
1.11 mycroft 390: clen > so->so_snd.sb_hiwat)
391: snderr(EMSGSIZE);
392: if (space < resid + clen && uio &&
1.1 cgd 393: (atomic || space < so->so_snd.sb_lowat || space < clen)) {
394: if (so->so_state & SS_NBIO)
395: snderr(EWOULDBLOCK);
396: sbunlock(&so->so_snd);
397: error = sbwait(&so->so_snd);
398: splx(s);
399: if (error)
400: goto out;
401: goto restart;
402: }
403: splx(s);
404: mp = ⊤
405: space -= clen;
406: do {
407: if (uio == NULL) {
408: /*
409: * Data is prepackaged in "top".
410: */
411: resid = 0;
412: if (flags & MSG_EOR)
413: top->m_flags |= M_EOR;
414: } else do {
415: if (top == 0) {
416: MGETHDR(m, M_WAIT, MT_DATA);
417: mlen = MHLEN;
418: m->m_pkthdr.len = 0;
419: m->m_pkthdr.rcvif = (struct ifnet *)0;
420: } else {
421: MGET(m, M_WAIT, MT_DATA);
422: mlen = MLEN;
423: }
1.15 mycroft 424: if (resid >= MINCLSIZE && space >= MCLBYTES) {
1.1 cgd 425: MCLGET(m, M_WAIT);
426: if ((m->m_flags & M_EXT) == 0)
427: goto nopages;
428: mlen = MCLBYTES;
1.15 mycroft 429: #ifdef MAPPED_MBUFS
430: len = min(MCLBYTES, resid);
431: #else
432: if (atomic && top == 0) {
433: len = min(MCLBYTES - max_hdr, resid);
434: m->m_data += max_hdr;
435: } else
436: len = min(MCLBYTES, resid);
437: #endif
1.25 explorer 438: space -= len;
1.1 cgd 439: } else {
440: nopages:
441: len = min(min(mlen, resid), space);
1.15 mycroft 442: space -= len;
1.1 cgd 443: /*
444: * For datagram protocols, leave room
445: * for protocol headers in first mbuf.
446: */
447: if (atomic && top == 0 && len < mlen)
448: MH_ALIGN(m, len);
449: }
450: error = uiomove(mtod(m, caddr_t), (int)len, uio);
451: resid = uio->uio_resid;
452: m->m_len = len;
453: *mp = m;
454: top->m_pkthdr.len += len;
455: if (error)
456: goto release;
457: mp = &m->m_next;
458: if (resid <= 0) {
459: if (flags & MSG_EOR)
460: top->m_flags |= M_EOR;
461: break;
462: }
463: } while (space > 0 && atomic);
464: if (dontroute)
465: so->so_options |= SO_DONTROUTE;
1.20 mycroft 466: s = splsoftnet(); /* XXX */
1.22 mycroft 467: error = (*so->so_proto->pr_usrreq)(so,
1.24 mycroft 468: (flags & MSG_OOB) ? PRU_SENDOOB : PRU_SEND,
469: top, addr, control, p);
1.1 cgd 470: splx(s);
471: if (dontroute)
472: so->so_options &= ~SO_DONTROUTE;
473: clen = 0;
474: control = 0;
475: top = 0;
476: mp = ⊤
477: if (error)
478: goto release;
479: } while (resid && space > 0);
480: } while (resid);
481:
482: release:
483: sbunlock(&so->so_snd);
484: out:
485: if (top)
486: m_freem(top);
487: if (control)
488: m_freem(control);
489: return (error);
490: }
491:
492: /*
493: * Implement receive operations on a socket.
494: * We depend on the way that records are added to the sockbuf
495: * by sbappend*. In particular, each record (mbufs linked through m_next)
496: * must begin with an address if the protocol so specifies,
497: * followed by an optional mbuf or mbufs containing ancillary data,
498: * and then zero or more mbufs of data.
499: * In order to avoid blocking network interrupts for the entire time here,
500: * we splx() while doing the actual copy to user space.
501: * Although the sockbuf is locked, new data may still be appended,
502: * and thus we must maintain consistency of the sockbuf during that time.
503: *
504: * The caller may receive the data as a single mbuf chain by supplying
505: * an mbuf **mp0 for use in returning the chain. The uio is then used
506: * only for the count in uio_resid.
507: */
1.3 andrew 508: int
1.1 cgd 509: soreceive(so, paddr, uio, mp0, controlp, flagsp)
510: register struct socket *so;
511: struct mbuf **paddr;
512: struct uio *uio;
513: struct mbuf **mp0;
514: struct mbuf **controlp;
515: int *flagsp;
516: {
517: register struct mbuf *m, **mp;
518: register int flags, len, error, s, offset;
519: struct protosw *pr = so->so_proto;
520: struct mbuf *nextrecord;
1.21 christos 521: int moff, type = 0;
1.3 andrew 522: int orig_resid = uio->uio_resid;
1.1 cgd 523:
524: mp = mp0;
525: if (paddr)
526: *paddr = 0;
527: if (controlp)
528: *controlp = 0;
529: if (flagsp)
530: flags = *flagsp &~ MSG_EOR;
531: else
532: flags = 0;
533: if (flags & MSG_OOB) {
534: m = m_get(M_WAIT, MT_DATA);
1.17 cgd 535: error = (*pr->pr_usrreq)(so, PRU_RCVOOB, m,
1.22 mycroft 536: (struct mbuf *)(long)(flags & MSG_PEEK), (struct mbuf *)0,
537: (struct proc *)0);
1.1 cgd 538: if (error)
539: goto bad;
540: do {
541: error = uiomove(mtod(m, caddr_t),
542: (int) min(uio->uio_resid, m->m_len), uio);
543: m = m_free(m);
544: } while (uio->uio_resid && error == 0 && m);
545: bad:
546: if (m)
547: m_freem(m);
548: return (error);
549: }
550: if (mp)
551: *mp = (struct mbuf *)0;
552: if (so->so_state & SS_ISCONFIRMING && uio->uio_resid)
1.22 mycroft 553: (*pr->pr_usrreq)(so, PRU_RCVD, (struct mbuf *)0,
554: (struct mbuf *)0, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 555:
556: restart:
1.21 christos 557: if ((error = sblock(&so->so_rcv, SBLOCKWAIT(flags))) != 0)
1.1 cgd 558: return (error);
1.20 mycroft 559: s = splsoftnet();
1.1 cgd 560:
561: m = so->so_rcv.sb_mb;
562: /*
563: * If we have less data than requested, block awaiting more
564: * (subject to any timeout) if:
1.15 mycroft 565: * 1. the current count is less than the low water mark,
1.1 cgd 566: * 2. MSG_WAITALL is set, and it is possible to do the entire
1.15 mycroft 567: * receive operation at once if we block (resid <= hiwat), or
568: * 3. MSG_DONTWAIT is not set.
1.1 cgd 569: * If MSG_WAITALL is set but resid is larger than the receive buffer,
570: * we have to do the receive in sections, and thus risk returning
571: * a short count if a timeout or signal occurs after we start.
572: */
1.21 christos 573: if (m == 0 || (((flags & MSG_DONTWAIT) == 0 &&
1.15 mycroft 574: so->so_rcv.sb_cc < uio->uio_resid) &&
1.1 cgd 575: (so->so_rcv.sb_cc < so->so_rcv.sb_lowat ||
576: ((flags & MSG_WAITALL) && uio->uio_resid <= so->so_rcv.sb_hiwat)) &&
1.21 christos 577: m->m_nextpkt == 0 && (pr->pr_flags & PR_ATOMIC) == 0)) {
1.1 cgd 578: #ifdef DIAGNOSTIC
579: if (m == 0 && so->so_rcv.sb_cc)
580: panic("receive 1");
581: #endif
582: if (so->so_error) {
583: if (m)
1.15 mycroft 584: goto dontblock;
1.1 cgd 585: error = so->so_error;
586: if ((flags & MSG_PEEK) == 0)
587: so->so_error = 0;
588: goto release;
589: }
590: if (so->so_state & SS_CANTRCVMORE) {
591: if (m)
1.15 mycroft 592: goto dontblock;
1.1 cgd 593: else
594: goto release;
595: }
596: for (; m; m = m->m_next)
597: if (m->m_type == MT_OOBDATA || (m->m_flags & M_EOR)) {
598: m = so->so_rcv.sb_mb;
599: goto dontblock;
600: }
601: if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) == 0 &&
602: (so->so_proto->pr_flags & PR_CONNREQUIRED)) {
603: error = ENOTCONN;
604: goto release;
605: }
606: if (uio->uio_resid == 0)
607: goto release;
1.15 mycroft 608: if ((so->so_state & SS_NBIO) || (flags & MSG_DONTWAIT)) {
1.1 cgd 609: error = EWOULDBLOCK;
610: goto release;
611: }
612: sbunlock(&so->so_rcv);
613: error = sbwait(&so->so_rcv);
614: splx(s);
615: if (error)
616: return (error);
617: goto restart;
618: }
619: dontblock:
1.15 mycroft 620: #ifdef notyet /* XXXX */
621: if (uio->uio_procp)
622: uio->uio_procp->p_stats->p_ru.ru_msgrcv++;
623: #endif
1.1 cgd 624: nextrecord = m->m_nextpkt;
625: if (pr->pr_flags & PR_ADDR) {
626: #ifdef DIAGNOSTIC
627: if (m->m_type != MT_SONAME)
628: panic("receive 1a");
629: #endif
1.3 andrew 630: orig_resid = 0;
1.1 cgd 631: if (flags & MSG_PEEK) {
632: if (paddr)
633: *paddr = m_copy(m, 0, m->m_len);
634: m = m->m_next;
635: } else {
636: sbfree(&so->so_rcv, m);
637: if (paddr) {
638: *paddr = m;
639: so->so_rcv.sb_mb = m->m_next;
640: m->m_next = 0;
641: m = so->so_rcv.sb_mb;
642: } else {
643: MFREE(m, so->so_rcv.sb_mb);
644: m = so->so_rcv.sb_mb;
645: }
646: }
647: }
648: while (m && m->m_type == MT_CONTROL && error == 0) {
649: if (flags & MSG_PEEK) {
650: if (controlp)
651: *controlp = m_copy(m, 0, m->m_len);
652: m = m->m_next;
653: } else {
654: sbfree(&so->so_rcv, m);
655: if (controlp) {
656: if (pr->pr_domain->dom_externalize &&
657: mtod(m, struct cmsghdr *)->cmsg_type ==
658: SCM_RIGHTS)
659: error = (*pr->pr_domain->dom_externalize)(m);
660: *controlp = m;
661: so->so_rcv.sb_mb = m->m_next;
662: m->m_next = 0;
663: m = so->so_rcv.sb_mb;
664: } else {
665: MFREE(m, so->so_rcv.sb_mb);
666: m = so->so_rcv.sb_mb;
667: }
668: }
1.3 andrew 669: if (controlp) {
670: orig_resid = 0;
1.1 cgd 671: controlp = &(*controlp)->m_next;
1.3 andrew 672: }
1.1 cgd 673: }
674: if (m) {
675: if ((flags & MSG_PEEK) == 0)
676: m->m_nextpkt = nextrecord;
677: type = m->m_type;
678: if (type == MT_OOBDATA)
679: flags |= MSG_OOB;
680: }
681: moff = 0;
682: offset = 0;
683: while (m && uio->uio_resid > 0 && error == 0) {
684: if (m->m_type == MT_OOBDATA) {
685: if (type != MT_OOBDATA)
686: break;
687: } else if (type == MT_OOBDATA)
688: break;
689: #ifdef DIAGNOSTIC
690: else if (m->m_type != MT_DATA && m->m_type != MT_HEADER)
691: panic("receive 3");
692: #endif
693: so->so_state &= ~SS_RCVATMARK;
694: len = uio->uio_resid;
695: if (so->so_oobmark && len > so->so_oobmark - offset)
696: len = so->so_oobmark - offset;
697: if (len > m->m_len - moff)
698: len = m->m_len - moff;
699: /*
700: * If mp is set, just pass back the mbufs.
701: * Otherwise copy them out via the uio, then free.
702: * Sockbuf must be consistent here (points to current mbuf,
703: * it points to next record) when we drop priority;
704: * we must note any additions to the sockbuf when we
705: * block interrupts again.
706: */
707: if (mp == 0) {
708: splx(s);
709: error = uiomove(mtod(m, caddr_t) + moff, (int)len, uio);
1.20 mycroft 710: s = splsoftnet();
1.1 cgd 711: } else
712: uio->uio_resid -= len;
713: if (len == m->m_len - moff) {
714: if (m->m_flags & M_EOR)
715: flags |= MSG_EOR;
716: if (flags & MSG_PEEK) {
717: m = m->m_next;
718: moff = 0;
719: } else {
720: nextrecord = m->m_nextpkt;
721: sbfree(&so->so_rcv, m);
722: if (mp) {
723: *mp = m;
724: mp = &m->m_next;
725: so->so_rcv.sb_mb = m = m->m_next;
726: *mp = (struct mbuf *)0;
727: } else {
728: MFREE(m, so->so_rcv.sb_mb);
729: m = so->so_rcv.sb_mb;
730: }
731: if (m)
732: m->m_nextpkt = nextrecord;
733: }
734: } else {
735: if (flags & MSG_PEEK)
736: moff += len;
737: else {
738: if (mp)
739: *mp = m_copym(m, 0, len, M_WAIT);
740: m->m_data += len;
741: m->m_len -= len;
742: so->so_rcv.sb_cc -= len;
743: }
744: }
745: if (so->so_oobmark) {
746: if ((flags & MSG_PEEK) == 0) {
747: so->so_oobmark -= len;
748: if (so->so_oobmark == 0) {
749: so->so_state |= SS_RCVATMARK;
750: break;
751: }
1.7 cgd 752: } else {
1.1 cgd 753: offset += len;
1.7 cgd 754: if (offset == so->so_oobmark)
755: break;
756: }
1.1 cgd 757: }
758: if (flags & MSG_EOR)
759: break;
760: /*
761: * If the MSG_WAITALL flag is set (for non-atomic socket),
762: * we must not quit until "uio->uio_resid == 0" or an error
763: * termination. If a signal/timeout occurs, return
764: * with a short count but without error.
765: * Keep sockbuf locked against other readers.
766: */
767: while (flags & MSG_WAITALL && m == 0 && uio->uio_resid > 0 &&
1.3 andrew 768: !sosendallatonce(so) && !nextrecord) {
1.1 cgd 769: if (so->so_error || so->so_state & SS_CANTRCVMORE)
770: break;
771: error = sbwait(&so->so_rcv);
772: if (error) {
773: sbunlock(&so->so_rcv);
774: splx(s);
775: return (0);
776: }
1.21 christos 777: if ((m = so->so_rcv.sb_mb) != NULL)
1.1 cgd 778: nextrecord = m->m_nextpkt;
779: }
780: }
1.3 andrew 781:
782: if (m && pr->pr_flags & PR_ATOMIC) {
783: flags |= MSG_TRUNC;
784: if ((flags & MSG_PEEK) == 0)
785: (void) sbdroprecord(&so->so_rcv);
786: }
1.1 cgd 787: if ((flags & MSG_PEEK) == 0) {
788: if (m == 0)
789: so->so_rcv.sb_mb = nextrecord;
790: if (pr->pr_flags & PR_WANTRCVD && so->so_pcb)
1.22 mycroft 791: (*pr->pr_usrreq)(so, PRU_RCVD, (struct mbuf *)0,
792: (struct mbuf *)(long)flags, (struct mbuf *)0,
793: (struct proc *)0);
1.1 cgd 794: }
1.3 andrew 795: if (orig_resid == uio->uio_resid && orig_resid &&
796: (flags & MSG_EOR) == 0 && (so->so_state & SS_CANTRCVMORE) == 0) {
797: sbunlock(&so->so_rcv);
798: splx(s);
799: goto restart;
800: }
801:
1.1 cgd 802: if (flagsp)
803: *flagsp |= flags;
804: release:
805: sbunlock(&so->so_rcv);
806: splx(s);
807: return (error);
808: }
809:
1.14 mycroft 810: int
1.1 cgd 811: soshutdown(so, how)
1.34 ! kleink 812: struct socket *so;
! 813: int how;
1.1 cgd 814: {
1.34 ! kleink 815: struct protosw *pr = so->so_proto;
! 816:
! 817: if (!(how == SHUT_RD || how == SHUT_WR || how == SHUT_RDWR))
! 818: return (EINVAL);
1.1 cgd 819:
1.34 ! kleink 820: if (how == SHUT_RD || how == SHUT_RDWR)
1.1 cgd 821: sorflush(so);
1.34 ! kleink 822: if (how == SHUT_WR || how == SHUT_RDWR)
1.22 mycroft 823: return (*pr->pr_usrreq)(so, PRU_SHUTDOWN, (struct mbuf *)0,
824: (struct mbuf *)0, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 825: return (0);
826: }
827:
1.14 mycroft 828: void
1.1 cgd 829: sorflush(so)
830: register struct socket *so;
831: {
832: register struct sockbuf *sb = &so->so_rcv;
833: register struct protosw *pr = so->so_proto;
834: register int s;
835: struct sockbuf asb;
836:
837: sb->sb_flags |= SB_NOINTR;
1.15 mycroft 838: (void) sblock(sb, M_WAITOK);
1.1 cgd 839: s = splimp();
840: socantrcvmore(so);
841: sbunlock(sb);
842: asb = *sb;
843: bzero((caddr_t)sb, sizeof (*sb));
844: splx(s);
845: if (pr->pr_flags & PR_RIGHTS && pr->pr_domain->dom_dispose)
846: (*pr->pr_domain->dom_dispose)(asb.sb_mb);
847: sbrelease(&asb);
848: }
849:
1.14 mycroft 850: int
1.1 cgd 851: sosetopt(so, level, optname, m0)
852: register struct socket *so;
853: int level, optname;
854: struct mbuf *m0;
855: {
856: int error = 0;
857: register struct mbuf *m = m0;
858:
859: if (level != SOL_SOCKET) {
860: if (so->so_proto && so->so_proto->pr_ctloutput)
861: return ((*so->so_proto->pr_ctloutput)
862: (PRCO_SETOPT, so, level, optname, &m0));
863: error = ENOPROTOOPT;
864: } else {
865: switch (optname) {
866:
867: case SO_LINGER:
868: if (m == NULL || m->m_len != sizeof (struct linger)) {
869: error = EINVAL;
870: goto bad;
871: }
872: so->so_linger = mtod(m, struct linger *)->l_linger;
873: /* fall thru... */
874:
875: case SO_DEBUG:
876: case SO_KEEPALIVE:
877: case SO_DONTROUTE:
878: case SO_USELOOPBACK:
879: case SO_BROADCAST:
880: case SO_REUSEADDR:
1.15 mycroft 881: case SO_REUSEPORT:
1.1 cgd 882: case SO_OOBINLINE:
1.26 thorpej 883: case SO_TIMESTAMP:
1.1 cgd 884: if (m == NULL || m->m_len < sizeof (int)) {
885: error = EINVAL;
886: goto bad;
887: }
888: if (*mtod(m, int *))
889: so->so_options |= optname;
890: else
891: so->so_options &= ~optname;
892: break;
893:
894: case SO_SNDBUF:
895: case SO_RCVBUF:
896: case SO_SNDLOWAT:
897: case SO_RCVLOWAT:
1.28 thorpej 898: {
899: int optval;
900:
1.1 cgd 901: if (m == NULL || m->m_len < sizeof (int)) {
902: error = EINVAL;
903: goto bad;
904: }
1.28 thorpej 905:
906: /*
907: * Values < 1 make no sense for any of these
908: * options, so disallow them.
909: */
910: optval = *mtod(m, int *);
911: if (optval < 1) {
912: error = EINVAL;
913: goto bad;
914: }
915:
1.1 cgd 916: switch (optname) {
917:
918: case SO_SNDBUF:
919: case SO_RCVBUF:
920: if (sbreserve(optname == SO_SNDBUF ?
921: &so->so_snd : &so->so_rcv,
1.28 thorpej 922: (u_long) optval) == 0) {
1.1 cgd 923: error = ENOBUFS;
924: goto bad;
925: }
926: break;
927:
1.28 thorpej 928: /*
929: * Make sure the low-water is never greater than
930: * the high-water.
931: */
1.1 cgd 932: case SO_SNDLOWAT:
1.28 thorpej 933: so->so_snd.sb_lowat =
934: (optval > so->so_snd.sb_hiwat) ?
935: so->so_snd.sb_hiwat : optval;
1.1 cgd 936: break;
937: case SO_RCVLOWAT:
1.28 thorpej 938: so->so_rcv.sb_lowat =
939: (optval > so->so_rcv.sb_hiwat) ?
940: so->so_rcv.sb_hiwat : optval;
1.1 cgd 941: break;
942: }
943: break;
1.28 thorpej 944: }
1.1 cgd 945:
946: case SO_SNDTIMEO:
947: case SO_RCVTIMEO:
948: {
949: struct timeval *tv;
950: short val;
951:
952: if (m == NULL || m->m_len < sizeof (*tv)) {
953: error = EINVAL;
954: goto bad;
955: }
956: tv = mtod(m, struct timeval *);
1.19 cgd 957: if (tv->tv_sec * hz + tv->tv_usec / tick > SHRT_MAX) {
1.1 cgd 958: error = EDOM;
959: goto bad;
960: }
961: val = tv->tv_sec * hz + tv->tv_usec / tick;
962:
963: switch (optname) {
964:
965: case SO_SNDTIMEO:
966: so->so_snd.sb_timeo = val;
967: break;
968: case SO_RCVTIMEO:
969: so->so_rcv.sb_timeo = val;
970: break;
971: }
972: break;
973: }
974:
975: default:
976: error = ENOPROTOOPT;
977: break;
978: }
1.15 mycroft 979: if (error == 0 && so->so_proto && so->so_proto->pr_ctloutput) {
980: (void) ((*so->so_proto->pr_ctloutput)
981: (PRCO_SETOPT, so, level, optname, &m0));
982: m = NULL; /* freed by protocol */
983: }
1.1 cgd 984: }
985: bad:
986: if (m)
987: (void) m_free(m);
988: return (error);
989: }
990:
1.14 mycroft 991: int
1.1 cgd 992: sogetopt(so, level, optname, mp)
993: register struct socket *so;
994: int level, optname;
995: struct mbuf **mp;
996: {
997: register struct mbuf *m;
998:
999: if (level != SOL_SOCKET) {
1000: if (so->so_proto && so->so_proto->pr_ctloutput) {
1001: return ((*so->so_proto->pr_ctloutput)
1002: (PRCO_GETOPT, so, level, optname, mp));
1003: } else
1004: return (ENOPROTOOPT);
1005: } else {
1006: m = m_get(M_WAIT, MT_SOOPTS);
1007: m->m_len = sizeof (int);
1008:
1009: switch (optname) {
1010:
1011: case SO_LINGER:
1012: m->m_len = sizeof (struct linger);
1013: mtod(m, struct linger *)->l_onoff =
1014: so->so_options & SO_LINGER;
1015: mtod(m, struct linger *)->l_linger = so->so_linger;
1016: break;
1017:
1018: case SO_USELOOPBACK:
1019: case SO_DONTROUTE:
1020: case SO_DEBUG:
1021: case SO_KEEPALIVE:
1022: case SO_REUSEADDR:
1.15 mycroft 1023: case SO_REUSEPORT:
1.1 cgd 1024: case SO_BROADCAST:
1025: case SO_OOBINLINE:
1.26 thorpej 1026: case SO_TIMESTAMP:
1.1 cgd 1027: *mtod(m, int *) = so->so_options & optname;
1028: break;
1029:
1030: case SO_TYPE:
1031: *mtod(m, int *) = so->so_type;
1032: break;
1033:
1034: case SO_ERROR:
1035: *mtod(m, int *) = so->so_error;
1036: so->so_error = 0;
1037: break;
1038:
1039: case SO_SNDBUF:
1040: *mtod(m, int *) = so->so_snd.sb_hiwat;
1041: break;
1042:
1043: case SO_RCVBUF:
1044: *mtod(m, int *) = so->so_rcv.sb_hiwat;
1045: break;
1046:
1047: case SO_SNDLOWAT:
1048: *mtod(m, int *) = so->so_snd.sb_lowat;
1049: break;
1050:
1051: case SO_RCVLOWAT:
1052: *mtod(m, int *) = so->so_rcv.sb_lowat;
1053: break;
1054:
1055: case SO_SNDTIMEO:
1056: case SO_RCVTIMEO:
1057: {
1058: int val = (optname == SO_SNDTIMEO ?
1059: so->so_snd.sb_timeo : so->so_rcv.sb_timeo);
1060:
1061: m->m_len = sizeof(struct timeval);
1062: mtod(m, struct timeval *)->tv_sec = val / hz;
1063: mtod(m, struct timeval *)->tv_usec =
1.27 kleink 1064: (val % hz) * tick;
1.1 cgd 1065: break;
1066: }
1067:
1068: default:
1069: (void)m_free(m);
1070: return (ENOPROTOOPT);
1071: }
1072: *mp = m;
1073: return (0);
1074: }
1075: }
1076:
1.14 mycroft 1077: void
1.1 cgd 1078: sohasoutofband(so)
1079: register struct socket *so;
1080: {
1081: struct proc *p;
1082:
1083: if (so->so_pgid < 0)
1084: gsignal(-so->so_pgid, SIGURG);
1085: else if (so->so_pgid > 0 && (p = pfind(so->so_pgid)) != 0)
1086: psignal(p, SIGURG);
1.2 cgd 1087: selwakeup(&so->so_rcv.sb_sel);
1.1 cgd 1088: }