Annotation of src/sys/kern/uipc_socket.c, Revision 1.29.4.2
1.29.4.2! cgd 1: /* $NetBSD: uipc_socket.c,v 1.29.4.1 1998/01/30 19:24:12 mellon 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));
1.29.4.1 mellon 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;
1.22 mycroft 86: error = (*prp->pr_usrreq)(so, PRU_ATTACH, (struct mbuf *)0,
87: (struct mbuf *)(long)proto, (struct mbuf *)0, p);
1.1 cgd 88: if (error) {
89: so->so_state |= SS_NOFDREF;
90: sofree(so);
91: return (error);
92: }
1.10 deraadt 93: #ifdef COMPAT_SUNOS
1.18 christos 94: {
95: extern struct emul emul_sunos;
96: if (p->p_emul == &emul_sunos && type == SOCK_DGRAM)
97: so->so_options |= SO_BROADCAST;
98: }
1.10 deraadt 99: #endif
1.1 cgd 100: *aso = so;
101: return (0);
102: }
103:
1.3 andrew 104: int
1.1 cgd 105: sobind(so, nam)
106: struct socket *so;
107: struct mbuf *nam;
108: {
1.22 mycroft 109: struct proc *p = curproc; /* XXX */
1.20 mycroft 110: int s = splsoftnet();
1.1 cgd 111: int error;
112:
1.22 mycroft 113: error = (*so->so_proto->pr_usrreq)(so, PRU_BIND, (struct mbuf *)0,
114: nam, (struct mbuf *)0, p);
1.1 cgd 115: splx(s);
116: return (error);
117: }
118:
1.3 andrew 119: int
1.1 cgd 120: solisten(so, backlog)
121: register struct socket *so;
122: int backlog;
123: {
1.20 mycroft 124: int s = splsoftnet(), error;
1.1 cgd 125:
1.22 mycroft 126: error = (*so->so_proto->pr_usrreq)(so, PRU_LISTEN, (struct mbuf *)0,
127: (struct mbuf *)0, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 128: if (error) {
129: splx(s);
130: return (error);
131: }
1.29.4.1 mellon 132: if (so->so_q.tqh_first == NULL)
1.1 cgd 133: so->so_options |= SO_ACCEPTCONN;
134: if (backlog < 0)
135: backlog = 0;
136: so->so_qlimit = min(backlog, SOMAXCONN);
137: splx(s);
138: return (0);
139: }
140:
1.21 christos 141: void
1.1 cgd 142: sofree(so)
143: register struct socket *so;
144: {
145:
146: if (so->so_pcb || (so->so_state & SS_NOFDREF) == 0)
147: return;
148: if (so->so_head) {
1.29.4.2! cgd 149: /*
! 150: * We must not decommission a socket that's on the accept(2)
! 151: * queue. If we do, then accept(2) may hang after select(2)
! 152: * indicated that the listening socket was ready.
! 153: */
! 154: if (!soqremque(so, 0))
! 155: return;
1.1 cgd 156: }
157: sbrelease(&so->so_snd);
158: sorflush(so);
159: FREE(so, M_SOCKET);
160: }
161:
162: /*
163: * Close a socket on last file table reference removal.
164: * Initiate disconnect if connected.
165: * Free socket when disconnect complete.
166: */
1.3 andrew 167: int
1.1 cgd 168: soclose(so)
169: register struct socket *so;
170: {
1.29.4.2! cgd 171: struct socket *so2;
1.20 mycroft 172: int s = splsoftnet(); /* conservative */
1.1 cgd 173: int error = 0;
174:
175: if (so->so_options & SO_ACCEPTCONN) {
1.29.4.2! cgd 176: while ((so2 = so->so_q0.tqh_first) != 0) {
! 177: (void) soqremque(so2, 0);
! 178: (void) soabort(so2);
! 179: }
! 180: while ((so2 = so->so_q.tqh_first) != 0) {
! 181: (void) soqremque(so2, 1);
! 182: (void) soabort(so2);
! 183: }
1.1 cgd 184: }
185: if (so->so_pcb == 0)
186: goto discard;
187: if (so->so_state & SS_ISCONNECTED) {
188: if ((so->so_state & SS_ISDISCONNECTING) == 0) {
189: error = sodisconnect(so);
190: if (error)
191: goto drop;
192: }
193: if (so->so_options & SO_LINGER) {
194: if ((so->so_state & SS_ISDISCONNECTING) &&
195: (so->so_state & SS_NBIO))
196: goto drop;
1.21 christos 197: while (so->so_state & SS_ISCONNECTED) {
198: error = tsleep((caddr_t)&so->so_timeo,
199: PSOCK | PCATCH, netcls,
1.29.4.1 mellon 200: so->so_linger * hz);
1.21 christos 201: if (error)
1.1 cgd 202: break;
1.21 christos 203: }
1.1 cgd 204: }
205: }
206: drop:
207: if (so->so_pcb) {
1.22 mycroft 208: int error2 = (*so->so_proto->pr_usrreq)(so, PRU_DETACH,
209: (struct mbuf *)0, (struct mbuf *)0, (struct mbuf *)0,
210: (struct proc *)0);
1.1 cgd 211: if (error == 0)
212: error = error2;
213: }
214: discard:
215: if (so->so_state & SS_NOFDREF)
216: panic("soclose: NOFDREF");
217: so->so_state |= SS_NOFDREF;
218: sofree(so);
219: splx(s);
220: return (error);
221: }
222:
223: /*
1.20 mycroft 224: * Must be called at splsoftnet...
1.1 cgd 225: */
1.3 andrew 226: int
1.1 cgd 227: soabort(so)
228: struct socket *so;
229: {
230:
1.22 mycroft 231: return (*so->so_proto->pr_usrreq)(so, PRU_ABORT, (struct mbuf *)0,
232: (struct mbuf *)0, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 233: }
234:
1.3 andrew 235: int
1.1 cgd 236: soaccept(so, nam)
237: register struct socket *so;
238: struct mbuf *nam;
239: {
1.20 mycroft 240: int s = splsoftnet();
1.1 cgd 241: int error;
242:
243: if ((so->so_state & SS_NOFDREF) == 0)
244: panic("soaccept: !NOFDREF");
245: so->so_state &= ~SS_NOFDREF;
1.29.4.2! cgd 246: if ((so->so_state & SS_ISDISCONNECTED) == 0)
! 247: error = (*so->so_proto->pr_usrreq)(so, PRU_ACCEPT,
! 248: (struct mbuf *)0, nam, (struct mbuf *)0, (struct proc *)0);
! 249: else
! 250: error = 0;
1.1 cgd 251: splx(s);
252: return (error);
253: }
254:
1.3 andrew 255: int
1.1 cgd 256: soconnect(so, nam)
257: register struct socket *so;
258: struct mbuf *nam;
259: {
1.23 mycroft 260: struct proc *p = curproc; /* XXX */
1.1 cgd 261: int s;
262: int error;
263:
264: if (so->so_options & SO_ACCEPTCONN)
265: return (EOPNOTSUPP);
1.20 mycroft 266: s = splsoftnet();
1.1 cgd 267: /*
268: * If protocol is connection-based, can only connect once.
269: * Otherwise, if connected, try to disconnect first.
270: * This allows user to disconnect by connecting to, e.g.,
271: * a null address.
272: */
273: if (so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING) &&
274: ((so->so_proto->pr_flags & PR_CONNREQUIRED) ||
275: (error = sodisconnect(so))))
276: error = EISCONN;
277: else
278: error = (*so->so_proto->pr_usrreq)(so, PRU_CONNECT,
1.23 mycroft 279: (struct mbuf *)0, nam, (struct mbuf *)0, p);
1.1 cgd 280: splx(s);
281: return (error);
282: }
283:
1.3 andrew 284: int
1.1 cgd 285: soconnect2(so1, so2)
286: register struct socket *so1;
287: struct socket *so2;
288: {
1.20 mycroft 289: int s = splsoftnet();
1.1 cgd 290: int error;
291:
1.22 mycroft 292: error = (*so1->so_proto->pr_usrreq)(so1, PRU_CONNECT2,
293: (struct mbuf *)0, (struct mbuf *)so2, (struct mbuf *)0,
294: (struct proc *)0);
1.1 cgd 295: splx(s);
296: return (error);
297: }
298:
1.3 andrew 299: int
1.1 cgd 300: sodisconnect(so)
301: register struct socket *so;
302: {
1.20 mycroft 303: int s = splsoftnet();
1.1 cgd 304: int error;
305:
306: if ((so->so_state & SS_ISCONNECTED) == 0) {
307: error = ENOTCONN;
308: goto bad;
309: }
310: if (so->so_state & SS_ISDISCONNECTING) {
311: error = EALREADY;
312: goto bad;
313: }
1.22 mycroft 314: error = (*so->so_proto->pr_usrreq)(so, PRU_DISCONNECT,
315: (struct mbuf *)0, (struct mbuf *)0, (struct mbuf *)0,
316: (struct proc *)0);
1.1 cgd 317: bad:
318: splx(s);
319: return (error);
320: }
321:
1.15 mycroft 322: #define SBLOCKWAIT(f) (((f) & MSG_DONTWAIT) ? M_NOWAIT : M_WAITOK)
1.1 cgd 323: /*
324: * Send on a socket.
325: * If send must go all at once and message is larger than
326: * send buffering, then hard error.
327: * Lock against other senders.
328: * If must go all at once and not enough room now, then
329: * inform user that this would block and do nothing.
330: * Otherwise, if nonblocking, send as much as possible.
331: * The data to be sent is described by "uio" if nonzero,
332: * otherwise by the mbuf chain "top" (which must be null
333: * if uio is not). Data provided in mbuf chain must be small
334: * enough to send all at once.
335: *
336: * Returns nonzero on error, timeout or signal; callers
337: * must check for short counts if EINTR/ERESTART are returned.
338: * Data and control buffers are freed on return.
339: */
1.3 andrew 340: int
1.1 cgd 341: sosend(so, addr, uio, top, control, flags)
342: register struct socket *so;
343: struct mbuf *addr;
344: struct uio *uio;
345: struct mbuf *top;
346: struct mbuf *control;
347: int flags;
348: {
1.15 mycroft 349: struct proc *p = curproc; /* XXX */
1.1 cgd 350: struct mbuf **mp;
351: register struct mbuf *m;
352: register long space, len, resid;
353: int clen = 0, error, s, dontroute, mlen;
354: int atomic = sosendallatonce(so) || top;
355:
356: if (uio)
357: resid = uio->uio_resid;
358: else
359: resid = top->m_pkthdr.len;
1.7 cgd 360: /*
361: * In theory resid should be unsigned.
362: * However, space must be signed, as it might be less than 0
363: * if we over-committed, and we must use a signed comparison
364: * of space and resid. On the other hand, a negative resid
365: * causes us to loop sending 0-length segments to the protocol.
366: */
1.29 mycroft 367: if (resid < 0) {
368: error = EINVAL;
369: goto out;
370: }
1.1 cgd 371: dontroute =
372: (flags & MSG_DONTROUTE) && (so->so_options & SO_DONTROUTE) == 0 &&
373: (so->so_proto->pr_flags & PR_ATOMIC);
1.12 mycroft 374: p->p_stats->p_ru.ru_msgsnd++;
1.1 cgd 375: if (control)
376: clen = control->m_len;
377: #define snderr(errno) { error = errno; splx(s); goto release; }
378:
379: restart:
1.21 christos 380: if ((error = sblock(&so->so_snd, SBLOCKWAIT(flags))) != 0)
1.1 cgd 381: goto out;
382: do {
1.20 mycroft 383: s = splsoftnet();
1.1 cgd 384: if (so->so_state & SS_CANTSENDMORE)
385: snderr(EPIPE);
386: if (so->so_error)
387: snderr(so->so_error);
388: if ((so->so_state & SS_ISCONNECTED) == 0) {
389: if (so->so_proto->pr_flags & PR_CONNREQUIRED) {
390: if ((so->so_state & SS_ISCONFIRMING) == 0 &&
391: !(resid == 0 && clen != 0))
392: snderr(ENOTCONN);
393: } else if (addr == 0)
394: snderr(EDESTADDRREQ);
395: }
396: space = sbspace(&so->so_snd);
397: if (flags & MSG_OOB)
398: space += 1024;
1.21 christos 399: if ((atomic && resid > so->so_snd.sb_hiwat) ||
1.11 mycroft 400: clen > so->so_snd.sb_hiwat)
401: snderr(EMSGSIZE);
402: if (space < resid + clen && uio &&
1.1 cgd 403: (atomic || space < so->so_snd.sb_lowat || space < clen)) {
404: if (so->so_state & SS_NBIO)
405: snderr(EWOULDBLOCK);
406: sbunlock(&so->so_snd);
407: error = sbwait(&so->so_snd);
408: splx(s);
409: if (error)
410: goto out;
411: goto restart;
412: }
413: splx(s);
414: mp = ⊤
415: space -= clen;
416: do {
417: if (uio == NULL) {
418: /*
419: * Data is prepackaged in "top".
420: */
421: resid = 0;
422: if (flags & MSG_EOR)
423: top->m_flags |= M_EOR;
424: } else do {
425: if (top == 0) {
426: MGETHDR(m, M_WAIT, MT_DATA);
427: mlen = MHLEN;
428: m->m_pkthdr.len = 0;
429: m->m_pkthdr.rcvif = (struct ifnet *)0;
430: } else {
431: MGET(m, M_WAIT, MT_DATA);
432: mlen = MLEN;
433: }
1.15 mycroft 434: if (resid >= MINCLSIZE && space >= MCLBYTES) {
1.1 cgd 435: MCLGET(m, M_WAIT);
436: if ((m->m_flags & M_EXT) == 0)
437: goto nopages;
438: mlen = MCLBYTES;
1.15 mycroft 439: #ifdef MAPPED_MBUFS
440: len = min(MCLBYTES, resid);
441: #else
442: if (atomic && top == 0) {
443: len = min(MCLBYTES - max_hdr, resid);
444: m->m_data += max_hdr;
445: } else
446: len = min(MCLBYTES, resid);
447: #endif
1.25 explorer 448: space -= len;
1.1 cgd 449: } else {
450: nopages:
451: len = min(min(mlen, resid), space);
1.15 mycroft 452: space -= len;
1.1 cgd 453: /*
454: * For datagram protocols, leave room
455: * for protocol headers in first mbuf.
456: */
457: if (atomic && top == 0 && len < mlen)
458: MH_ALIGN(m, len);
459: }
460: error = uiomove(mtod(m, caddr_t), (int)len, uio);
461: resid = uio->uio_resid;
462: m->m_len = len;
463: *mp = m;
464: top->m_pkthdr.len += len;
465: if (error)
466: goto release;
467: mp = &m->m_next;
468: if (resid <= 0) {
469: if (flags & MSG_EOR)
470: top->m_flags |= M_EOR;
471: break;
472: }
473: } while (space > 0 && atomic);
474: if (dontroute)
475: so->so_options |= SO_DONTROUTE;
1.20 mycroft 476: s = splsoftnet(); /* XXX */
1.22 mycroft 477: error = (*so->so_proto->pr_usrreq)(so,
1.24 mycroft 478: (flags & MSG_OOB) ? PRU_SENDOOB : PRU_SEND,
479: top, addr, control, p);
1.1 cgd 480: splx(s);
481: if (dontroute)
482: so->so_options &= ~SO_DONTROUTE;
483: clen = 0;
484: control = 0;
485: top = 0;
486: mp = ⊤
487: if (error)
488: goto release;
489: } while (resid && space > 0);
490: } while (resid);
491:
492: release:
493: sbunlock(&so->so_snd);
494: out:
495: if (top)
496: m_freem(top);
497: if (control)
498: m_freem(control);
499: return (error);
500: }
501:
502: /*
503: * Implement receive operations on a socket.
504: * We depend on the way that records are added to the sockbuf
505: * by sbappend*. In particular, each record (mbufs linked through m_next)
506: * must begin with an address if the protocol so specifies,
507: * followed by an optional mbuf or mbufs containing ancillary data,
508: * and then zero or more mbufs of data.
509: * In order to avoid blocking network interrupts for the entire time here,
510: * we splx() while doing the actual copy to user space.
511: * Although the sockbuf is locked, new data may still be appended,
512: * and thus we must maintain consistency of the sockbuf during that time.
513: *
514: * The caller may receive the data as a single mbuf chain by supplying
515: * an mbuf **mp0 for use in returning the chain. The uio is then used
516: * only for the count in uio_resid.
517: */
1.3 andrew 518: int
1.1 cgd 519: soreceive(so, paddr, uio, mp0, controlp, flagsp)
520: register struct socket *so;
521: struct mbuf **paddr;
522: struct uio *uio;
523: struct mbuf **mp0;
524: struct mbuf **controlp;
525: int *flagsp;
526: {
527: register struct mbuf *m, **mp;
528: register int flags, len, error, s, offset;
529: struct protosw *pr = so->so_proto;
530: struct mbuf *nextrecord;
1.21 christos 531: int moff, type = 0;
1.3 andrew 532: int orig_resid = uio->uio_resid;
1.1 cgd 533:
534: mp = mp0;
535: if (paddr)
536: *paddr = 0;
537: if (controlp)
538: *controlp = 0;
539: if (flagsp)
540: flags = *flagsp &~ MSG_EOR;
541: else
542: flags = 0;
543: if (flags & MSG_OOB) {
544: m = m_get(M_WAIT, MT_DATA);
1.17 cgd 545: error = (*pr->pr_usrreq)(so, PRU_RCVOOB, m,
1.22 mycroft 546: (struct mbuf *)(long)(flags & MSG_PEEK), (struct mbuf *)0,
547: (struct proc *)0);
1.1 cgd 548: if (error)
549: goto bad;
550: do {
551: error = uiomove(mtod(m, caddr_t),
552: (int) min(uio->uio_resid, m->m_len), uio);
553: m = m_free(m);
554: } while (uio->uio_resid && error == 0 && m);
555: bad:
556: if (m)
557: m_freem(m);
558: return (error);
559: }
560: if (mp)
561: *mp = (struct mbuf *)0;
562: if (so->so_state & SS_ISCONFIRMING && uio->uio_resid)
1.22 mycroft 563: (*pr->pr_usrreq)(so, PRU_RCVD, (struct mbuf *)0,
564: (struct mbuf *)0, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 565:
566: restart:
1.21 christos 567: if ((error = sblock(&so->so_rcv, SBLOCKWAIT(flags))) != 0)
1.1 cgd 568: return (error);
1.20 mycroft 569: s = splsoftnet();
1.1 cgd 570:
571: m = so->so_rcv.sb_mb;
572: /*
573: * If we have less data than requested, block awaiting more
574: * (subject to any timeout) if:
1.15 mycroft 575: * 1. the current count is less than the low water mark,
1.1 cgd 576: * 2. MSG_WAITALL is set, and it is possible to do the entire
1.15 mycroft 577: * receive operation at once if we block (resid <= hiwat), or
578: * 3. MSG_DONTWAIT is not set.
1.1 cgd 579: * If MSG_WAITALL is set but resid is larger than the receive buffer,
580: * we have to do the receive in sections, and thus risk returning
581: * a short count if a timeout or signal occurs after we start.
582: */
1.21 christos 583: if (m == 0 || (((flags & MSG_DONTWAIT) == 0 &&
1.15 mycroft 584: so->so_rcv.sb_cc < uio->uio_resid) &&
1.1 cgd 585: (so->so_rcv.sb_cc < so->so_rcv.sb_lowat ||
586: ((flags & MSG_WAITALL) && uio->uio_resid <= so->so_rcv.sb_hiwat)) &&
1.21 christos 587: m->m_nextpkt == 0 && (pr->pr_flags & PR_ATOMIC) == 0)) {
1.1 cgd 588: #ifdef DIAGNOSTIC
589: if (m == 0 && so->so_rcv.sb_cc)
590: panic("receive 1");
591: #endif
592: if (so->so_error) {
593: if (m)
1.15 mycroft 594: goto dontblock;
1.1 cgd 595: error = so->so_error;
596: if ((flags & MSG_PEEK) == 0)
597: so->so_error = 0;
598: goto release;
599: }
600: if (so->so_state & SS_CANTRCVMORE) {
601: if (m)
1.15 mycroft 602: goto dontblock;
1.1 cgd 603: else
604: goto release;
605: }
606: for (; m; m = m->m_next)
607: if (m->m_type == MT_OOBDATA || (m->m_flags & M_EOR)) {
608: m = so->so_rcv.sb_mb;
609: goto dontblock;
610: }
611: if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) == 0 &&
612: (so->so_proto->pr_flags & PR_CONNREQUIRED)) {
613: error = ENOTCONN;
614: goto release;
615: }
616: if (uio->uio_resid == 0)
617: goto release;
1.15 mycroft 618: if ((so->so_state & SS_NBIO) || (flags & MSG_DONTWAIT)) {
1.1 cgd 619: error = EWOULDBLOCK;
620: goto release;
621: }
622: sbunlock(&so->so_rcv);
623: error = sbwait(&so->so_rcv);
624: splx(s);
625: if (error)
626: return (error);
627: goto restart;
628: }
629: dontblock:
1.15 mycroft 630: #ifdef notyet /* XXXX */
631: if (uio->uio_procp)
632: uio->uio_procp->p_stats->p_ru.ru_msgrcv++;
633: #endif
1.1 cgd 634: nextrecord = m->m_nextpkt;
635: if (pr->pr_flags & PR_ADDR) {
636: #ifdef DIAGNOSTIC
637: if (m->m_type != MT_SONAME)
638: panic("receive 1a");
639: #endif
1.3 andrew 640: orig_resid = 0;
1.1 cgd 641: if (flags & MSG_PEEK) {
642: if (paddr)
643: *paddr = m_copy(m, 0, m->m_len);
644: m = m->m_next;
645: } else {
646: sbfree(&so->so_rcv, m);
647: if (paddr) {
648: *paddr = m;
649: so->so_rcv.sb_mb = m->m_next;
650: m->m_next = 0;
651: m = so->so_rcv.sb_mb;
652: } else {
653: MFREE(m, so->so_rcv.sb_mb);
654: m = so->so_rcv.sb_mb;
655: }
656: }
657: }
658: while (m && m->m_type == MT_CONTROL && error == 0) {
659: if (flags & MSG_PEEK) {
660: if (controlp)
661: *controlp = m_copy(m, 0, m->m_len);
662: m = m->m_next;
663: } else {
664: sbfree(&so->so_rcv, m);
665: if (controlp) {
666: if (pr->pr_domain->dom_externalize &&
667: mtod(m, struct cmsghdr *)->cmsg_type ==
668: SCM_RIGHTS)
669: error = (*pr->pr_domain->dom_externalize)(m);
670: *controlp = m;
671: so->so_rcv.sb_mb = m->m_next;
672: m->m_next = 0;
673: m = so->so_rcv.sb_mb;
674: } else {
675: MFREE(m, so->so_rcv.sb_mb);
676: m = so->so_rcv.sb_mb;
677: }
678: }
1.3 andrew 679: if (controlp) {
680: orig_resid = 0;
1.1 cgd 681: controlp = &(*controlp)->m_next;
1.3 andrew 682: }
1.1 cgd 683: }
684: if (m) {
685: if ((flags & MSG_PEEK) == 0)
686: m->m_nextpkt = nextrecord;
687: type = m->m_type;
688: if (type == MT_OOBDATA)
689: flags |= MSG_OOB;
690: }
691: moff = 0;
692: offset = 0;
693: while (m && uio->uio_resid > 0 && error == 0) {
694: if (m->m_type == MT_OOBDATA) {
695: if (type != MT_OOBDATA)
696: break;
697: } else if (type == MT_OOBDATA)
698: break;
699: #ifdef DIAGNOSTIC
700: else if (m->m_type != MT_DATA && m->m_type != MT_HEADER)
701: panic("receive 3");
702: #endif
703: so->so_state &= ~SS_RCVATMARK;
704: len = uio->uio_resid;
705: if (so->so_oobmark && len > so->so_oobmark - offset)
706: len = so->so_oobmark - offset;
707: if (len > m->m_len - moff)
708: len = m->m_len - moff;
709: /*
710: * If mp is set, just pass back the mbufs.
711: * Otherwise copy them out via the uio, then free.
712: * Sockbuf must be consistent here (points to current mbuf,
713: * it points to next record) when we drop priority;
714: * we must note any additions to the sockbuf when we
715: * block interrupts again.
716: */
717: if (mp == 0) {
718: splx(s);
719: error = uiomove(mtod(m, caddr_t) + moff, (int)len, uio);
1.20 mycroft 720: s = splsoftnet();
1.1 cgd 721: } else
722: uio->uio_resid -= len;
723: if (len == m->m_len - moff) {
724: if (m->m_flags & M_EOR)
725: flags |= MSG_EOR;
726: if (flags & MSG_PEEK) {
727: m = m->m_next;
728: moff = 0;
729: } else {
730: nextrecord = m->m_nextpkt;
731: sbfree(&so->so_rcv, m);
732: if (mp) {
733: *mp = m;
734: mp = &m->m_next;
735: so->so_rcv.sb_mb = m = m->m_next;
736: *mp = (struct mbuf *)0;
737: } else {
738: MFREE(m, so->so_rcv.sb_mb);
739: m = so->so_rcv.sb_mb;
740: }
741: if (m)
742: m->m_nextpkt = nextrecord;
743: }
744: } else {
745: if (flags & MSG_PEEK)
746: moff += len;
747: else {
748: if (mp)
749: *mp = m_copym(m, 0, len, M_WAIT);
750: m->m_data += len;
751: m->m_len -= len;
752: so->so_rcv.sb_cc -= len;
753: }
754: }
755: if (so->so_oobmark) {
756: if ((flags & MSG_PEEK) == 0) {
757: so->so_oobmark -= len;
758: if (so->so_oobmark == 0) {
759: so->so_state |= SS_RCVATMARK;
760: break;
761: }
1.7 cgd 762: } else {
1.1 cgd 763: offset += len;
1.7 cgd 764: if (offset == so->so_oobmark)
765: break;
766: }
1.1 cgd 767: }
768: if (flags & MSG_EOR)
769: break;
770: /*
771: * If the MSG_WAITALL flag is set (for non-atomic socket),
772: * we must not quit until "uio->uio_resid == 0" or an error
773: * termination. If a signal/timeout occurs, return
774: * with a short count but without error.
775: * Keep sockbuf locked against other readers.
776: */
777: while (flags & MSG_WAITALL && m == 0 && uio->uio_resid > 0 &&
1.3 andrew 778: !sosendallatonce(so) && !nextrecord) {
1.1 cgd 779: if (so->so_error || so->so_state & SS_CANTRCVMORE)
780: break;
781: error = sbwait(&so->so_rcv);
782: if (error) {
783: sbunlock(&so->so_rcv);
784: splx(s);
785: return (0);
786: }
1.21 christos 787: if ((m = so->so_rcv.sb_mb) != NULL)
1.1 cgd 788: nextrecord = m->m_nextpkt;
789: }
790: }
1.3 andrew 791:
792: if (m && pr->pr_flags & PR_ATOMIC) {
793: flags |= MSG_TRUNC;
794: if ((flags & MSG_PEEK) == 0)
795: (void) sbdroprecord(&so->so_rcv);
796: }
1.1 cgd 797: if ((flags & MSG_PEEK) == 0) {
798: if (m == 0)
799: so->so_rcv.sb_mb = nextrecord;
800: if (pr->pr_flags & PR_WANTRCVD && so->so_pcb)
1.22 mycroft 801: (*pr->pr_usrreq)(so, PRU_RCVD, (struct mbuf *)0,
802: (struct mbuf *)(long)flags, (struct mbuf *)0,
803: (struct proc *)0);
1.1 cgd 804: }
1.3 andrew 805: if (orig_resid == uio->uio_resid && orig_resid &&
806: (flags & MSG_EOR) == 0 && (so->so_state & SS_CANTRCVMORE) == 0) {
807: sbunlock(&so->so_rcv);
808: splx(s);
809: goto restart;
810: }
811:
1.1 cgd 812: if (flagsp)
813: *flagsp |= flags;
814: release:
815: sbunlock(&so->so_rcv);
816: splx(s);
817: return (error);
818: }
819:
1.14 mycroft 820: int
1.1 cgd 821: soshutdown(so, how)
822: register struct socket *so;
823: register int how;
824: {
825: register struct protosw *pr = so->so_proto;
826:
827: how++;
828: if (how & FREAD)
829: sorflush(so);
830: if (how & FWRITE)
1.22 mycroft 831: return (*pr->pr_usrreq)(so, PRU_SHUTDOWN, (struct mbuf *)0,
832: (struct mbuf *)0, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 833: return (0);
834: }
835:
1.14 mycroft 836: void
1.1 cgd 837: sorflush(so)
838: register struct socket *so;
839: {
840: register struct sockbuf *sb = &so->so_rcv;
841: register struct protosw *pr = so->so_proto;
842: register int s;
843: struct sockbuf asb;
844:
845: sb->sb_flags |= SB_NOINTR;
1.15 mycroft 846: (void) sblock(sb, M_WAITOK);
1.1 cgd 847: s = splimp();
848: socantrcvmore(so);
849: sbunlock(sb);
850: asb = *sb;
851: bzero((caddr_t)sb, sizeof (*sb));
852: splx(s);
853: if (pr->pr_flags & PR_RIGHTS && pr->pr_domain->dom_dispose)
854: (*pr->pr_domain->dom_dispose)(asb.sb_mb);
855: sbrelease(&asb);
856: }
857:
1.14 mycroft 858: int
1.1 cgd 859: sosetopt(so, level, optname, m0)
860: register struct socket *so;
861: int level, optname;
862: struct mbuf *m0;
863: {
864: int error = 0;
865: register struct mbuf *m = m0;
866:
867: if (level != SOL_SOCKET) {
868: if (so->so_proto && so->so_proto->pr_ctloutput)
869: return ((*so->so_proto->pr_ctloutput)
870: (PRCO_SETOPT, so, level, optname, &m0));
871: error = ENOPROTOOPT;
872: } else {
873: switch (optname) {
874:
875: case SO_LINGER:
876: if (m == NULL || m->m_len != sizeof (struct linger)) {
877: error = EINVAL;
878: goto bad;
879: }
880: so->so_linger = mtod(m, struct linger *)->l_linger;
881: /* fall thru... */
882:
883: case SO_DEBUG:
884: case SO_KEEPALIVE:
885: case SO_DONTROUTE:
886: case SO_USELOOPBACK:
887: case SO_BROADCAST:
888: case SO_REUSEADDR:
1.15 mycroft 889: case SO_REUSEPORT:
1.1 cgd 890: case SO_OOBINLINE:
1.26 thorpej 891: case SO_TIMESTAMP:
1.1 cgd 892: if (m == NULL || m->m_len < sizeof (int)) {
893: error = EINVAL;
894: goto bad;
895: }
896: if (*mtod(m, int *))
897: so->so_options |= optname;
898: else
899: so->so_options &= ~optname;
900: break;
901:
902: case SO_SNDBUF:
903: case SO_RCVBUF:
904: case SO_SNDLOWAT:
905: case SO_RCVLOWAT:
1.28 thorpej 906: {
907: int optval;
908:
1.1 cgd 909: if (m == NULL || m->m_len < sizeof (int)) {
910: error = EINVAL;
911: goto bad;
912: }
1.28 thorpej 913:
914: /*
915: * Values < 1 make no sense for any of these
916: * options, so disallow them.
917: */
918: optval = *mtod(m, int *);
919: if (optval < 1) {
920: error = EINVAL;
921: goto bad;
922: }
923:
1.1 cgd 924: switch (optname) {
925:
926: case SO_SNDBUF:
927: case SO_RCVBUF:
928: if (sbreserve(optname == SO_SNDBUF ?
929: &so->so_snd : &so->so_rcv,
1.28 thorpej 930: (u_long) optval) == 0) {
1.1 cgd 931: error = ENOBUFS;
932: goto bad;
933: }
934: break;
935:
1.28 thorpej 936: /*
937: * Make sure the low-water is never greater than
938: * the high-water.
939: */
1.1 cgd 940: case SO_SNDLOWAT:
1.28 thorpej 941: so->so_snd.sb_lowat =
942: (optval > so->so_snd.sb_hiwat) ?
943: so->so_snd.sb_hiwat : optval;
1.1 cgd 944: break;
945: case SO_RCVLOWAT:
1.28 thorpej 946: so->so_rcv.sb_lowat =
947: (optval > so->so_rcv.sb_hiwat) ?
948: so->so_rcv.sb_hiwat : optval;
1.1 cgd 949: break;
950: }
951: break;
1.28 thorpej 952: }
1.1 cgd 953:
954: case SO_SNDTIMEO:
955: case SO_RCVTIMEO:
956: {
957: struct timeval *tv;
958: short val;
959:
960: if (m == NULL || m->m_len < sizeof (*tv)) {
961: error = EINVAL;
962: goto bad;
963: }
964: tv = mtod(m, struct timeval *);
1.19 cgd 965: if (tv->tv_sec * hz + tv->tv_usec / tick > SHRT_MAX) {
1.1 cgd 966: error = EDOM;
967: goto bad;
968: }
969: val = tv->tv_sec * hz + tv->tv_usec / tick;
970:
971: switch (optname) {
972:
973: case SO_SNDTIMEO:
974: so->so_snd.sb_timeo = val;
975: break;
976: case SO_RCVTIMEO:
977: so->so_rcv.sb_timeo = val;
978: break;
979: }
980: break;
981: }
982:
983: default:
984: error = ENOPROTOOPT;
985: break;
986: }
1.15 mycroft 987: if (error == 0 && so->so_proto && so->so_proto->pr_ctloutput) {
988: (void) ((*so->so_proto->pr_ctloutput)
989: (PRCO_SETOPT, so, level, optname, &m0));
990: m = NULL; /* freed by protocol */
991: }
1.1 cgd 992: }
993: bad:
994: if (m)
995: (void) m_free(m);
996: return (error);
997: }
998:
1.14 mycroft 999: int
1.1 cgd 1000: sogetopt(so, level, optname, mp)
1001: register struct socket *so;
1002: int level, optname;
1003: struct mbuf **mp;
1004: {
1005: register struct mbuf *m;
1006:
1007: if (level != SOL_SOCKET) {
1008: if (so->so_proto && so->so_proto->pr_ctloutput) {
1009: return ((*so->so_proto->pr_ctloutput)
1010: (PRCO_GETOPT, so, level, optname, mp));
1011: } else
1012: return (ENOPROTOOPT);
1013: } else {
1014: m = m_get(M_WAIT, MT_SOOPTS);
1015: m->m_len = sizeof (int);
1016:
1017: switch (optname) {
1018:
1019: case SO_LINGER:
1020: m->m_len = sizeof (struct linger);
1021: mtod(m, struct linger *)->l_onoff =
1022: so->so_options & SO_LINGER;
1023: mtod(m, struct linger *)->l_linger = so->so_linger;
1024: break;
1025:
1026: case SO_USELOOPBACK:
1027: case SO_DONTROUTE:
1028: case SO_DEBUG:
1029: case SO_KEEPALIVE:
1030: case SO_REUSEADDR:
1.15 mycroft 1031: case SO_REUSEPORT:
1.1 cgd 1032: case SO_BROADCAST:
1033: case SO_OOBINLINE:
1.26 thorpej 1034: case SO_TIMESTAMP:
1.1 cgd 1035: *mtod(m, int *) = so->so_options & optname;
1036: break;
1037:
1038: case SO_TYPE:
1039: *mtod(m, int *) = so->so_type;
1040: break;
1041:
1042: case SO_ERROR:
1043: *mtod(m, int *) = so->so_error;
1044: so->so_error = 0;
1045: break;
1046:
1047: case SO_SNDBUF:
1048: *mtod(m, int *) = so->so_snd.sb_hiwat;
1049: break;
1050:
1051: case SO_RCVBUF:
1052: *mtod(m, int *) = so->so_rcv.sb_hiwat;
1053: break;
1054:
1055: case SO_SNDLOWAT:
1056: *mtod(m, int *) = so->so_snd.sb_lowat;
1057: break;
1058:
1059: case SO_RCVLOWAT:
1060: *mtod(m, int *) = so->so_rcv.sb_lowat;
1061: break;
1062:
1063: case SO_SNDTIMEO:
1064: case SO_RCVTIMEO:
1065: {
1066: int val = (optname == SO_SNDTIMEO ?
1067: so->so_snd.sb_timeo : so->so_rcv.sb_timeo);
1068:
1069: m->m_len = sizeof(struct timeval);
1070: mtod(m, struct timeval *)->tv_sec = val / hz;
1071: mtod(m, struct timeval *)->tv_usec =
1.27 kleink 1072: (val % hz) * tick;
1.1 cgd 1073: break;
1074: }
1075:
1076: default:
1077: (void)m_free(m);
1078: return (ENOPROTOOPT);
1079: }
1080: *mp = m;
1081: return (0);
1082: }
1083: }
1084:
1.14 mycroft 1085: void
1.1 cgd 1086: sohasoutofband(so)
1087: register struct socket *so;
1088: {
1089: struct proc *p;
1090:
1091: if (so->so_pgid < 0)
1092: gsignal(-so->so_pgid, SIGURG);
1093: else if (so->so_pgid > 0 && (p = pfind(so->so_pgid)) != 0)
1094: psignal(p, SIGURG);
1.2 cgd 1095: selwakeup(&so->so_rcv.sb_sel);
1.1 cgd 1096: }
CVSweb <webmaster@jp.NetBSD.org>