Annotation of src/sys/kern/uipc_socket.c, Revision 1.35
1.35 ! thorpej 1: /* $NetBSD: uipc_socket.c,v 1.34 1998/04/27 13:31:45 kleink 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>
1.1 cgd 53:
54: /*
55: * Socket operation routines.
56: * These routines are called by the routines in
57: * sys_socket.c or from a system process, and
58: * implement the semantics of socket operations by
59: * switching out to the protocol specific routines.
60: */
61: /*ARGSUSED*/
1.3 andrew 62: int
1.1 cgd 63: socreate(dom, aso, type, proto)
1.11 mycroft 64: int dom;
1.1 cgd 65: struct socket **aso;
66: register int type;
67: int proto;
68: {
69: struct proc *p = curproc; /* XXX */
70: register struct protosw *prp;
71: register struct socket *so;
72: register int error;
73:
74: if (proto)
75: prp = pffindproto(dom, proto, type);
76: else
77: prp = pffindtype(dom, type);
1.15 mycroft 78: if (prp == 0 || prp->pr_usrreq == 0)
1.1 cgd 79: return (EPROTONOSUPPORT);
80: if (prp->pr_type != type)
81: return (EPROTOTYPE);
82: MALLOC(so, struct socket *, sizeof(*so), M_SOCKET, M_WAIT);
83: bzero((caddr_t)so, sizeof(*so));
1.31 thorpej 84: TAILQ_INIT(&so->so_q0);
85: TAILQ_INIT(&so->so_q);
1.1 cgd 86: so->so_type = type;
87: so->so_proto = prp;
1.33 matt 88: so->so_send = sosend;
89: so->so_receive = soreceive;
1.22 mycroft 90: error = (*prp->pr_usrreq)(so, PRU_ATTACH, (struct mbuf *)0,
91: (struct mbuf *)(long)proto, (struct mbuf *)0, p);
1.1 cgd 92: if (error) {
93: so->so_state |= SS_NOFDREF;
94: sofree(so);
95: return (error);
96: }
1.10 deraadt 97: #ifdef COMPAT_SUNOS
1.18 christos 98: {
99: extern struct emul emul_sunos;
100: if (p->p_emul == &emul_sunos && type == SOCK_DGRAM)
101: so->so_options |= SO_BROADCAST;
102: }
1.10 deraadt 103: #endif
1.1 cgd 104: *aso = so;
105: return (0);
106: }
107:
1.3 andrew 108: int
1.1 cgd 109: sobind(so, nam)
110: struct socket *so;
111: struct mbuf *nam;
112: {
1.22 mycroft 113: struct proc *p = curproc; /* XXX */
1.20 mycroft 114: int s = splsoftnet();
1.1 cgd 115: int error;
116:
1.22 mycroft 117: error = (*so->so_proto->pr_usrreq)(so, PRU_BIND, (struct mbuf *)0,
118: nam, (struct mbuf *)0, p);
1.1 cgd 119: splx(s);
120: return (error);
121: }
122:
1.3 andrew 123: int
1.1 cgd 124: solisten(so, backlog)
125: register struct socket *so;
126: int backlog;
127: {
1.20 mycroft 128: int s = splsoftnet(), error;
1.1 cgd 129:
1.22 mycroft 130: error = (*so->so_proto->pr_usrreq)(so, PRU_LISTEN, (struct mbuf *)0,
131: (struct mbuf *)0, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 132: if (error) {
133: splx(s);
134: return (error);
135: }
1.31 thorpej 136: if (so->so_q.tqh_first == NULL)
1.1 cgd 137: so->so_options |= SO_ACCEPTCONN;
138: if (backlog < 0)
139: backlog = 0;
140: so->so_qlimit = min(backlog, SOMAXCONN);
141: splx(s);
142: return (0);
143: }
144:
1.21 christos 145: void
1.1 cgd 146: sofree(so)
147: register struct socket *so;
148: {
149:
150: if (so->so_pcb || (so->so_state & SS_NOFDREF) == 0)
151: return;
152: if (so->so_head) {
153: if (!soqremque(so, 0) && !soqremque(so, 1))
154: panic("sofree dq");
155: so->so_head = 0;
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: {
1.20 mycroft 171: int s = splsoftnet(); /* conservative */
1.1 cgd 172: int error = 0;
173:
174: if (so->so_options & SO_ACCEPTCONN) {
1.31 thorpej 175: while (so->so_q0.tqh_first)
176: (void) soabort(so->so_q0.tqh_first);
177: while (so->so_q.tqh_first)
178: (void) soabort(so->so_q.tqh_first);
1.1 cgd 179: }
180: if (so->so_pcb == 0)
181: goto discard;
182: if (so->so_state & SS_ISCONNECTED) {
183: if ((so->so_state & SS_ISDISCONNECTING) == 0) {
184: error = sodisconnect(so);
185: if (error)
186: goto drop;
187: }
188: if (so->so_options & SO_LINGER) {
189: if ((so->so_state & SS_ISDISCONNECTING) &&
190: (so->so_state & SS_NBIO))
191: goto drop;
1.21 christos 192: while (so->so_state & SS_ISCONNECTED) {
193: error = tsleep((caddr_t)&so->so_timeo,
194: PSOCK | PCATCH, netcls,
1.30 thorpej 195: so->so_linger * hz);
1.21 christos 196: if (error)
1.1 cgd 197: break;
1.21 christos 198: }
1.1 cgd 199: }
200: }
201: drop:
202: if (so->so_pcb) {
1.22 mycroft 203: int error2 = (*so->so_proto->pr_usrreq)(so, PRU_DETACH,
204: (struct mbuf *)0, (struct mbuf *)0, (struct mbuf *)0,
205: (struct proc *)0);
1.1 cgd 206: if (error == 0)
207: error = error2;
208: }
209: discard:
210: if (so->so_state & SS_NOFDREF)
211: panic("soclose: NOFDREF");
212: so->so_state |= SS_NOFDREF;
213: sofree(so);
214: splx(s);
215: return (error);
216: }
217:
218: /*
1.20 mycroft 219: * Must be called at splsoftnet...
1.1 cgd 220: */
1.3 andrew 221: int
1.1 cgd 222: soabort(so)
223: struct socket *so;
224: {
225:
1.22 mycroft 226: return (*so->so_proto->pr_usrreq)(so, PRU_ABORT, (struct mbuf *)0,
227: (struct mbuf *)0, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 228: }
229:
1.3 andrew 230: int
1.1 cgd 231: soaccept(so, nam)
232: register struct socket *so;
233: struct mbuf *nam;
234: {
1.20 mycroft 235: int s = splsoftnet();
1.1 cgd 236: int error;
237:
238: if ((so->so_state & SS_NOFDREF) == 0)
239: panic("soaccept: !NOFDREF");
240: so->so_state &= ~SS_NOFDREF;
1.22 mycroft 241: error = (*so->so_proto->pr_usrreq)(so, PRU_ACCEPT, (struct mbuf *)0,
242: nam, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 243: splx(s);
244: return (error);
245: }
246:
1.3 andrew 247: int
1.1 cgd 248: soconnect(so, nam)
249: register struct socket *so;
250: struct mbuf *nam;
251: {
1.23 mycroft 252: struct proc *p = curproc; /* XXX */
1.1 cgd 253: int s;
254: int error;
255:
256: if (so->so_options & SO_ACCEPTCONN)
257: return (EOPNOTSUPP);
1.20 mycroft 258: s = splsoftnet();
1.1 cgd 259: /*
260: * If protocol is connection-based, can only connect once.
261: * Otherwise, if connected, try to disconnect first.
262: * This allows user to disconnect by connecting to, e.g.,
263: * a null address.
264: */
265: if (so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING) &&
266: ((so->so_proto->pr_flags & PR_CONNREQUIRED) ||
267: (error = sodisconnect(so))))
268: error = EISCONN;
269: else
270: error = (*so->so_proto->pr_usrreq)(so, PRU_CONNECT,
1.23 mycroft 271: (struct mbuf *)0, nam, (struct mbuf *)0, p);
1.1 cgd 272: splx(s);
273: return (error);
274: }
275:
1.3 andrew 276: int
1.1 cgd 277: soconnect2(so1, so2)
278: register struct socket *so1;
279: struct socket *so2;
280: {
1.20 mycroft 281: int s = splsoftnet();
1.1 cgd 282: int error;
283:
1.22 mycroft 284: error = (*so1->so_proto->pr_usrreq)(so1, PRU_CONNECT2,
285: (struct mbuf *)0, (struct mbuf *)so2, (struct mbuf *)0,
286: (struct proc *)0);
1.1 cgd 287: splx(s);
288: return (error);
289: }
290:
1.3 andrew 291: int
1.1 cgd 292: sodisconnect(so)
293: register struct socket *so;
294: {
1.20 mycroft 295: int s = splsoftnet();
1.1 cgd 296: int error;
297:
298: if ((so->so_state & SS_ISCONNECTED) == 0) {
299: error = ENOTCONN;
300: goto bad;
301: }
302: if (so->so_state & SS_ISDISCONNECTING) {
303: error = EALREADY;
304: goto bad;
305: }
1.22 mycroft 306: error = (*so->so_proto->pr_usrreq)(so, PRU_DISCONNECT,
307: (struct mbuf *)0, (struct mbuf *)0, (struct mbuf *)0,
308: (struct proc *)0);
1.1 cgd 309: bad:
310: splx(s);
311: return (error);
312: }
313:
1.15 mycroft 314: #define SBLOCKWAIT(f) (((f) & MSG_DONTWAIT) ? M_NOWAIT : M_WAITOK)
1.1 cgd 315: /*
316: * Send on a socket.
317: * If send must go all at once and message is larger than
318: * send buffering, then hard error.
319: * Lock against other senders.
320: * If must go all at once and not enough room now, then
321: * inform user that this would block and do nothing.
322: * Otherwise, if nonblocking, send as much as possible.
323: * The data to be sent is described by "uio" if nonzero,
324: * otherwise by the mbuf chain "top" (which must be null
325: * if uio is not). Data provided in mbuf chain must be small
326: * enough to send all at once.
327: *
328: * Returns nonzero on error, timeout or signal; callers
329: * must check for short counts if EINTR/ERESTART are returned.
330: * Data and control buffers are freed on return.
331: */
1.3 andrew 332: int
1.1 cgd 333: sosend(so, addr, uio, top, control, flags)
334: register struct socket *so;
335: struct mbuf *addr;
336: struct uio *uio;
337: struct mbuf *top;
338: struct mbuf *control;
339: int flags;
340: {
1.15 mycroft 341: struct proc *p = curproc; /* XXX */
1.1 cgd 342: struct mbuf **mp;
343: register struct mbuf *m;
344: register long space, len, resid;
345: int clen = 0, error, s, dontroute, mlen;
346: int atomic = sosendallatonce(so) || top;
347:
348: if (uio)
349: resid = uio->uio_resid;
350: else
351: resid = top->m_pkthdr.len;
1.7 cgd 352: /*
353: * In theory resid should be unsigned.
354: * However, space must be signed, as it might be less than 0
355: * if we over-committed, and we must use a signed comparison
356: * of space and resid. On the other hand, a negative resid
357: * causes us to loop sending 0-length segments to the protocol.
358: */
1.29 mycroft 359: if (resid < 0) {
360: error = EINVAL;
361: goto out;
362: }
1.1 cgd 363: dontroute =
364: (flags & MSG_DONTROUTE) && (so->so_options & SO_DONTROUTE) == 0 &&
365: (so->so_proto->pr_flags & PR_ATOMIC);
1.12 mycroft 366: p->p_stats->p_ru.ru_msgsnd++;
1.1 cgd 367: if (control)
368: clen = control->m_len;
369: #define snderr(errno) { error = errno; splx(s); goto release; }
370:
371: restart:
1.21 christos 372: if ((error = sblock(&so->so_snd, SBLOCKWAIT(flags))) != 0)
1.1 cgd 373: goto out;
374: do {
1.20 mycroft 375: s = splsoftnet();
1.1 cgd 376: if (so->so_state & SS_CANTSENDMORE)
377: snderr(EPIPE);
378: if (so->so_error)
379: snderr(so->so_error);
380: if ((so->so_state & SS_ISCONNECTED) == 0) {
381: if (so->so_proto->pr_flags & PR_CONNREQUIRED) {
382: if ((so->so_state & SS_ISCONFIRMING) == 0 &&
383: !(resid == 0 && clen != 0))
384: snderr(ENOTCONN);
385: } else if (addr == 0)
386: snderr(EDESTADDRREQ);
387: }
388: space = sbspace(&so->so_snd);
389: if (flags & MSG_OOB)
390: space += 1024;
1.21 christos 391: if ((atomic && resid > so->so_snd.sb_hiwat) ||
1.11 mycroft 392: clen > so->so_snd.sb_hiwat)
393: snderr(EMSGSIZE);
394: if (space < resid + clen && uio &&
1.1 cgd 395: (atomic || space < so->so_snd.sb_lowat || space < clen)) {
396: if (so->so_state & SS_NBIO)
397: snderr(EWOULDBLOCK);
398: sbunlock(&so->so_snd);
399: error = sbwait(&so->so_snd);
400: splx(s);
401: if (error)
402: goto out;
403: goto restart;
404: }
405: splx(s);
406: mp = ⊤
407: space -= clen;
408: do {
409: if (uio == NULL) {
410: /*
411: * Data is prepackaged in "top".
412: */
413: resid = 0;
414: if (flags & MSG_EOR)
415: top->m_flags |= M_EOR;
416: } else do {
417: if (top == 0) {
418: MGETHDR(m, M_WAIT, MT_DATA);
419: mlen = MHLEN;
420: m->m_pkthdr.len = 0;
421: m->m_pkthdr.rcvif = (struct ifnet *)0;
422: } else {
423: MGET(m, M_WAIT, MT_DATA);
424: mlen = MLEN;
425: }
1.15 mycroft 426: if (resid >= MINCLSIZE && space >= MCLBYTES) {
1.1 cgd 427: MCLGET(m, M_WAIT);
428: if ((m->m_flags & M_EXT) == 0)
429: goto nopages;
430: mlen = MCLBYTES;
1.15 mycroft 431: #ifdef MAPPED_MBUFS
432: len = min(MCLBYTES, resid);
433: #else
434: if (atomic && top == 0) {
435: len = min(MCLBYTES - max_hdr, resid);
436: m->m_data += max_hdr;
437: } else
438: len = min(MCLBYTES, resid);
439: #endif
1.25 explorer 440: space -= len;
1.1 cgd 441: } else {
442: nopages:
443: len = min(min(mlen, resid), space);
1.15 mycroft 444: space -= len;
1.1 cgd 445: /*
446: * For datagram protocols, leave room
447: * for protocol headers in first mbuf.
448: */
449: if (atomic && top == 0 && len < mlen)
450: MH_ALIGN(m, len);
451: }
452: error = uiomove(mtod(m, caddr_t), (int)len, uio);
453: resid = uio->uio_resid;
454: m->m_len = len;
455: *mp = m;
456: top->m_pkthdr.len += len;
457: if (error)
458: goto release;
459: mp = &m->m_next;
460: if (resid <= 0) {
461: if (flags & MSG_EOR)
462: top->m_flags |= M_EOR;
463: break;
464: }
465: } while (space > 0 && atomic);
466: if (dontroute)
467: so->so_options |= SO_DONTROUTE;
1.20 mycroft 468: s = splsoftnet(); /* XXX */
1.22 mycroft 469: error = (*so->so_proto->pr_usrreq)(so,
1.24 mycroft 470: (flags & MSG_OOB) ? PRU_SENDOOB : PRU_SEND,
471: top, addr, control, p);
1.1 cgd 472: splx(s);
473: if (dontroute)
474: so->so_options &= ~SO_DONTROUTE;
475: clen = 0;
476: control = 0;
477: top = 0;
478: mp = ⊤
479: if (error)
480: goto release;
481: } while (resid && space > 0);
482: } while (resid);
483:
484: release:
485: sbunlock(&so->so_snd);
486: out:
487: if (top)
488: m_freem(top);
489: if (control)
490: m_freem(control);
491: return (error);
492: }
493:
494: /*
495: * Implement receive operations on a socket.
496: * We depend on the way that records are added to the sockbuf
497: * by sbappend*. In particular, each record (mbufs linked through m_next)
498: * must begin with an address if the protocol so specifies,
499: * followed by an optional mbuf or mbufs containing ancillary data,
500: * and then zero or more mbufs of data.
501: * In order to avoid blocking network interrupts for the entire time here,
502: * we splx() while doing the actual copy to user space.
503: * Although the sockbuf is locked, new data may still be appended,
504: * and thus we must maintain consistency of the sockbuf during that time.
505: *
506: * The caller may receive the data as a single mbuf chain by supplying
507: * an mbuf **mp0 for use in returning the chain. The uio is then used
508: * only for the count in uio_resid.
509: */
1.3 andrew 510: int
1.1 cgd 511: soreceive(so, paddr, uio, mp0, controlp, flagsp)
512: register struct socket *so;
513: struct mbuf **paddr;
514: struct uio *uio;
515: struct mbuf **mp0;
516: struct mbuf **controlp;
517: int *flagsp;
518: {
519: register struct mbuf *m, **mp;
520: register int flags, len, error, s, offset;
521: struct protosw *pr = so->so_proto;
522: struct mbuf *nextrecord;
1.21 christos 523: int moff, type = 0;
1.3 andrew 524: int orig_resid = uio->uio_resid;
1.1 cgd 525:
526: mp = mp0;
527: if (paddr)
528: *paddr = 0;
529: if (controlp)
530: *controlp = 0;
531: if (flagsp)
532: flags = *flagsp &~ MSG_EOR;
533: else
534: flags = 0;
535: if (flags & MSG_OOB) {
536: m = m_get(M_WAIT, MT_DATA);
1.17 cgd 537: error = (*pr->pr_usrreq)(so, PRU_RCVOOB, m,
1.22 mycroft 538: (struct mbuf *)(long)(flags & MSG_PEEK), (struct mbuf *)0,
539: (struct proc *)0);
1.1 cgd 540: if (error)
541: goto bad;
542: do {
543: error = uiomove(mtod(m, caddr_t),
544: (int) min(uio->uio_resid, m->m_len), uio);
545: m = m_free(m);
546: } while (uio->uio_resid && error == 0 && m);
547: bad:
548: if (m)
549: m_freem(m);
550: return (error);
551: }
552: if (mp)
553: *mp = (struct mbuf *)0;
554: if (so->so_state & SS_ISCONFIRMING && uio->uio_resid)
1.22 mycroft 555: (*pr->pr_usrreq)(so, PRU_RCVD, (struct mbuf *)0,
556: (struct mbuf *)0, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 557:
558: restart:
1.21 christos 559: if ((error = sblock(&so->so_rcv, SBLOCKWAIT(flags))) != 0)
1.1 cgd 560: return (error);
1.20 mycroft 561: s = splsoftnet();
1.1 cgd 562:
563: m = so->so_rcv.sb_mb;
564: /*
565: * If we have less data than requested, block awaiting more
566: * (subject to any timeout) if:
1.15 mycroft 567: * 1. the current count is less than the low water mark,
1.1 cgd 568: * 2. MSG_WAITALL is set, and it is possible to do the entire
1.15 mycroft 569: * receive operation at once if we block (resid <= hiwat), or
570: * 3. MSG_DONTWAIT is not set.
1.1 cgd 571: * If MSG_WAITALL is set but resid is larger than the receive buffer,
572: * we have to do the receive in sections, and thus risk returning
573: * a short count if a timeout or signal occurs after we start.
574: */
1.21 christos 575: if (m == 0 || (((flags & MSG_DONTWAIT) == 0 &&
1.15 mycroft 576: so->so_rcv.sb_cc < uio->uio_resid) &&
1.1 cgd 577: (so->so_rcv.sb_cc < so->so_rcv.sb_lowat ||
578: ((flags & MSG_WAITALL) && uio->uio_resid <= so->so_rcv.sb_hiwat)) &&
1.21 christos 579: m->m_nextpkt == 0 && (pr->pr_flags & PR_ATOMIC) == 0)) {
1.1 cgd 580: #ifdef DIAGNOSTIC
581: if (m == 0 && so->so_rcv.sb_cc)
582: panic("receive 1");
583: #endif
584: if (so->so_error) {
585: if (m)
1.15 mycroft 586: goto dontblock;
1.1 cgd 587: error = so->so_error;
588: if ((flags & MSG_PEEK) == 0)
589: so->so_error = 0;
590: goto release;
591: }
592: if (so->so_state & SS_CANTRCVMORE) {
593: if (m)
1.15 mycroft 594: goto dontblock;
1.1 cgd 595: else
596: goto release;
597: }
598: for (; m; m = m->m_next)
599: if (m->m_type == MT_OOBDATA || (m->m_flags & M_EOR)) {
600: m = so->so_rcv.sb_mb;
601: goto dontblock;
602: }
603: if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) == 0 &&
604: (so->so_proto->pr_flags & PR_CONNREQUIRED)) {
605: error = ENOTCONN;
606: goto release;
607: }
608: if (uio->uio_resid == 0)
609: goto release;
1.15 mycroft 610: if ((so->so_state & SS_NBIO) || (flags & MSG_DONTWAIT)) {
1.1 cgd 611: error = EWOULDBLOCK;
612: goto release;
613: }
614: sbunlock(&so->so_rcv);
615: error = sbwait(&so->so_rcv);
616: splx(s);
617: if (error)
618: return (error);
619: goto restart;
620: }
621: dontblock:
1.15 mycroft 622: #ifdef notyet /* XXXX */
623: if (uio->uio_procp)
624: uio->uio_procp->p_stats->p_ru.ru_msgrcv++;
625: #endif
1.1 cgd 626: nextrecord = m->m_nextpkt;
627: if (pr->pr_flags & PR_ADDR) {
628: #ifdef DIAGNOSTIC
629: if (m->m_type != MT_SONAME)
630: panic("receive 1a");
631: #endif
1.3 andrew 632: orig_resid = 0;
1.1 cgd 633: if (flags & MSG_PEEK) {
634: if (paddr)
635: *paddr = m_copy(m, 0, m->m_len);
636: m = m->m_next;
637: } else {
638: sbfree(&so->so_rcv, m);
639: if (paddr) {
640: *paddr = m;
641: so->so_rcv.sb_mb = m->m_next;
642: m->m_next = 0;
643: m = so->so_rcv.sb_mb;
644: } else {
645: MFREE(m, so->so_rcv.sb_mb);
646: m = so->so_rcv.sb_mb;
647: }
648: }
649: }
650: while (m && m->m_type == MT_CONTROL && error == 0) {
651: if (flags & MSG_PEEK) {
652: if (controlp)
653: *controlp = m_copy(m, 0, m->m_len);
654: m = m->m_next;
655: } else {
656: sbfree(&so->so_rcv, m);
657: if (controlp) {
658: if (pr->pr_domain->dom_externalize &&
659: mtod(m, struct cmsghdr *)->cmsg_type ==
660: SCM_RIGHTS)
661: error = (*pr->pr_domain->dom_externalize)(m);
662: *controlp = m;
663: so->so_rcv.sb_mb = m->m_next;
664: m->m_next = 0;
665: m = so->so_rcv.sb_mb;
666: } else {
667: MFREE(m, so->so_rcv.sb_mb);
668: m = so->so_rcv.sb_mb;
669: }
670: }
1.3 andrew 671: if (controlp) {
672: orig_resid = 0;
1.1 cgd 673: controlp = &(*controlp)->m_next;
1.3 andrew 674: }
1.1 cgd 675: }
676: if (m) {
677: if ((flags & MSG_PEEK) == 0)
678: m->m_nextpkt = nextrecord;
679: type = m->m_type;
680: if (type == MT_OOBDATA)
681: flags |= MSG_OOB;
682: }
683: moff = 0;
684: offset = 0;
685: while (m && uio->uio_resid > 0 && error == 0) {
686: if (m->m_type == MT_OOBDATA) {
687: if (type != MT_OOBDATA)
688: break;
689: } else if (type == MT_OOBDATA)
690: break;
691: #ifdef DIAGNOSTIC
692: else if (m->m_type != MT_DATA && m->m_type != MT_HEADER)
693: panic("receive 3");
694: #endif
695: so->so_state &= ~SS_RCVATMARK;
696: len = uio->uio_resid;
697: if (so->so_oobmark && len > so->so_oobmark - offset)
698: len = so->so_oobmark - offset;
699: if (len > m->m_len - moff)
700: len = m->m_len - moff;
701: /*
702: * If mp is set, just pass back the mbufs.
703: * Otherwise copy them out via the uio, then free.
704: * Sockbuf must be consistent here (points to current mbuf,
705: * it points to next record) when we drop priority;
706: * we must note any additions to the sockbuf when we
707: * block interrupts again.
708: */
709: if (mp == 0) {
710: splx(s);
711: error = uiomove(mtod(m, caddr_t) + moff, (int)len, uio);
1.20 mycroft 712: s = splsoftnet();
1.1 cgd 713: } else
714: uio->uio_resid -= len;
715: if (len == m->m_len - moff) {
716: if (m->m_flags & M_EOR)
717: flags |= MSG_EOR;
718: if (flags & MSG_PEEK) {
719: m = m->m_next;
720: moff = 0;
721: } else {
722: nextrecord = m->m_nextpkt;
723: sbfree(&so->so_rcv, m);
724: if (mp) {
725: *mp = m;
726: mp = &m->m_next;
727: so->so_rcv.sb_mb = m = m->m_next;
728: *mp = (struct mbuf *)0;
729: } else {
730: MFREE(m, so->so_rcv.sb_mb);
731: m = so->so_rcv.sb_mb;
732: }
733: if (m)
734: m->m_nextpkt = nextrecord;
735: }
736: } else {
737: if (flags & MSG_PEEK)
738: moff += len;
739: else {
740: if (mp)
741: *mp = m_copym(m, 0, len, M_WAIT);
742: m->m_data += len;
743: m->m_len -= len;
744: so->so_rcv.sb_cc -= len;
745: }
746: }
747: if (so->so_oobmark) {
748: if ((flags & MSG_PEEK) == 0) {
749: so->so_oobmark -= len;
750: if (so->so_oobmark == 0) {
751: so->so_state |= SS_RCVATMARK;
752: break;
753: }
1.7 cgd 754: } else {
1.1 cgd 755: offset += len;
1.7 cgd 756: if (offset == so->so_oobmark)
757: break;
758: }
1.1 cgd 759: }
760: if (flags & MSG_EOR)
761: break;
762: /*
763: * If the MSG_WAITALL flag is set (for non-atomic socket),
764: * we must not quit until "uio->uio_resid == 0" or an error
765: * termination. If a signal/timeout occurs, return
766: * with a short count but without error.
767: * Keep sockbuf locked against other readers.
768: */
769: while (flags & MSG_WAITALL && m == 0 && uio->uio_resid > 0 &&
1.3 andrew 770: !sosendallatonce(so) && !nextrecord) {
1.1 cgd 771: if (so->so_error || so->so_state & SS_CANTRCVMORE)
772: break;
773: error = sbwait(&so->so_rcv);
774: if (error) {
775: sbunlock(&so->so_rcv);
776: splx(s);
777: return (0);
778: }
1.21 christos 779: if ((m = so->so_rcv.sb_mb) != NULL)
1.1 cgd 780: nextrecord = m->m_nextpkt;
781: }
782: }
1.3 andrew 783:
784: if (m && pr->pr_flags & PR_ATOMIC) {
785: flags |= MSG_TRUNC;
786: if ((flags & MSG_PEEK) == 0)
787: (void) sbdroprecord(&so->so_rcv);
788: }
1.1 cgd 789: if ((flags & MSG_PEEK) == 0) {
790: if (m == 0)
791: so->so_rcv.sb_mb = nextrecord;
792: if (pr->pr_flags & PR_WANTRCVD && so->so_pcb)
1.22 mycroft 793: (*pr->pr_usrreq)(so, PRU_RCVD, (struct mbuf *)0,
794: (struct mbuf *)(long)flags, (struct mbuf *)0,
795: (struct proc *)0);
1.1 cgd 796: }
1.3 andrew 797: if (orig_resid == uio->uio_resid && orig_resid &&
798: (flags & MSG_EOR) == 0 && (so->so_state & SS_CANTRCVMORE) == 0) {
799: sbunlock(&so->so_rcv);
800: splx(s);
801: goto restart;
802: }
803:
1.1 cgd 804: if (flagsp)
805: *flagsp |= flags;
806: release:
807: sbunlock(&so->so_rcv);
808: splx(s);
809: return (error);
810: }
811:
1.14 mycroft 812: int
1.1 cgd 813: soshutdown(so, how)
1.34 kleink 814: struct socket *so;
815: int how;
1.1 cgd 816: {
1.34 kleink 817: struct protosw *pr = so->so_proto;
818:
819: if (!(how == SHUT_RD || how == SHUT_WR || how == SHUT_RDWR))
820: return (EINVAL);
1.1 cgd 821:
1.34 kleink 822: if (how == SHUT_RD || how == SHUT_RDWR)
1.1 cgd 823: sorflush(so);
1.34 kleink 824: if (how == SHUT_WR || how == SHUT_RDWR)
1.22 mycroft 825: return (*pr->pr_usrreq)(so, PRU_SHUTDOWN, (struct mbuf *)0,
826: (struct mbuf *)0, (struct mbuf *)0, (struct proc *)0);
1.1 cgd 827: return (0);
828: }
829:
1.14 mycroft 830: void
1.1 cgd 831: sorflush(so)
832: register struct socket *so;
833: {
834: register struct sockbuf *sb = &so->so_rcv;
835: register struct protosw *pr = so->so_proto;
836: register int s;
837: struct sockbuf asb;
838:
839: sb->sb_flags |= SB_NOINTR;
1.15 mycroft 840: (void) sblock(sb, M_WAITOK);
1.1 cgd 841: s = splimp();
842: socantrcvmore(so);
843: sbunlock(sb);
844: asb = *sb;
845: bzero((caddr_t)sb, sizeof (*sb));
846: splx(s);
847: if (pr->pr_flags & PR_RIGHTS && pr->pr_domain->dom_dispose)
848: (*pr->pr_domain->dom_dispose)(asb.sb_mb);
849: sbrelease(&asb);
850: }
851:
1.14 mycroft 852: int
1.1 cgd 853: sosetopt(so, level, optname, m0)
854: register struct socket *so;
855: int level, optname;
856: struct mbuf *m0;
857: {
858: int error = 0;
859: register struct mbuf *m = m0;
860:
861: if (level != SOL_SOCKET) {
862: if (so->so_proto && so->so_proto->pr_ctloutput)
863: return ((*so->so_proto->pr_ctloutput)
864: (PRCO_SETOPT, so, level, optname, &m0));
865: error = ENOPROTOOPT;
866: } else {
867: switch (optname) {
868:
869: case SO_LINGER:
870: if (m == NULL || m->m_len != sizeof (struct linger)) {
871: error = EINVAL;
872: goto bad;
873: }
874: so->so_linger = mtod(m, struct linger *)->l_linger;
875: /* fall thru... */
876:
877: case SO_DEBUG:
878: case SO_KEEPALIVE:
879: case SO_DONTROUTE:
880: case SO_USELOOPBACK:
881: case SO_BROADCAST:
882: case SO_REUSEADDR:
1.15 mycroft 883: case SO_REUSEPORT:
1.1 cgd 884: case SO_OOBINLINE:
1.26 thorpej 885: case SO_TIMESTAMP:
1.1 cgd 886: if (m == NULL || m->m_len < sizeof (int)) {
887: error = EINVAL;
888: goto bad;
889: }
890: if (*mtod(m, int *))
891: so->so_options |= optname;
892: else
893: so->so_options &= ~optname;
894: break;
895:
896: case SO_SNDBUF:
897: case SO_RCVBUF:
898: case SO_SNDLOWAT:
899: case SO_RCVLOWAT:
1.28 thorpej 900: {
901: int optval;
902:
1.1 cgd 903: if (m == NULL || m->m_len < sizeof (int)) {
904: error = EINVAL;
905: goto bad;
906: }
1.28 thorpej 907:
908: /*
909: * Values < 1 make no sense for any of these
910: * options, so disallow them.
911: */
912: optval = *mtod(m, int *);
913: if (optval < 1) {
914: error = EINVAL;
915: goto bad;
916: }
917:
1.1 cgd 918: switch (optname) {
919:
920: case SO_SNDBUF:
921: case SO_RCVBUF:
922: if (sbreserve(optname == SO_SNDBUF ?
923: &so->so_snd : &so->so_rcv,
1.28 thorpej 924: (u_long) optval) == 0) {
1.1 cgd 925: error = ENOBUFS;
926: goto bad;
927: }
928: break;
929:
1.28 thorpej 930: /*
931: * Make sure the low-water is never greater than
932: * the high-water.
933: */
1.1 cgd 934: case SO_SNDLOWAT:
1.28 thorpej 935: so->so_snd.sb_lowat =
936: (optval > so->so_snd.sb_hiwat) ?
937: so->so_snd.sb_hiwat : optval;
1.1 cgd 938: break;
939: case SO_RCVLOWAT:
1.28 thorpej 940: so->so_rcv.sb_lowat =
941: (optval > so->so_rcv.sb_hiwat) ?
942: so->so_rcv.sb_hiwat : optval;
1.1 cgd 943: break;
944: }
945: break;
1.28 thorpej 946: }
1.1 cgd 947:
948: case SO_SNDTIMEO:
949: case SO_RCVTIMEO:
950: {
951: struct timeval *tv;
952: short val;
953:
954: if (m == NULL || m->m_len < sizeof (*tv)) {
955: error = EINVAL;
956: goto bad;
957: }
958: tv = mtod(m, struct timeval *);
1.19 cgd 959: if (tv->tv_sec * hz + tv->tv_usec / tick > SHRT_MAX) {
1.1 cgd 960: error = EDOM;
961: goto bad;
962: }
963: val = tv->tv_sec * hz + tv->tv_usec / tick;
964:
965: switch (optname) {
966:
967: case SO_SNDTIMEO:
968: so->so_snd.sb_timeo = val;
969: break;
970: case SO_RCVTIMEO:
971: so->so_rcv.sb_timeo = val;
972: break;
973: }
974: break;
975: }
976:
977: default:
978: error = ENOPROTOOPT;
979: break;
980: }
1.15 mycroft 981: if (error == 0 && so->so_proto && so->so_proto->pr_ctloutput) {
982: (void) ((*so->so_proto->pr_ctloutput)
983: (PRCO_SETOPT, so, level, optname, &m0));
984: m = NULL; /* freed by protocol */
985: }
1.1 cgd 986: }
987: bad:
988: if (m)
989: (void) m_free(m);
990: return (error);
991: }
992:
1.14 mycroft 993: int
1.1 cgd 994: sogetopt(so, level, optname, mp)
995: register struct socket *so;
996: int level, optname;
997: struct mbuf **mp;
998: {
999: register struct mbuf *m;
1000:
1001: if (level != SOL_SOCKET) {
1002: if (so->so_proto && so->so_proto->pr_ctloutput) {
1003: return ((*so->so_proto->pr_ctloutput)
1004: (PRCO_GETOPT, so, level, optname, mp));
1005: } else
1006: return (ENOPROTOOPT);
1007: } else {
1008: m = m_get(M_WAIT, MT_SOOPTS);
1009: m->m_len = sizeof (int);
1010:
1011: switch (optname) {
1012:
1013: case SO_LINGER:
1014: m->m_len = sizeof (struct linger);
1015: mtod(m, struct linger *)->l_onoff =
1016: so->so_options & SO_LINGER;
1017: mtod(m, struct linger *)->l_linger = so->so_linger;
1018: break;
1019:
1020: case SO_USELOOPBACK:
1021: case SO_DONTROUTE:
1022: case SO_DEBUG:
1023: case SO_KEEPALIVE:
1024: case SO_REUSEADDR:
1.15 mycroft 1025: case SO_REUSEPORT:
1.1 cgd 1026: case SO_BROADCAST:
1027: case SO_OOBINLINE:
1.26 thorpej 1028: case SO_TIMESTAMP:
1.1 cgd 1029: *mtod(m, int *) = so->so_options & optname;
1030: break;
1031:
1032: case SO_TYPE:
1033: *mtod(m, int *) = so->so_type;
1034: break;
1035:
1036: case SO_ERROR:
1037: *mtod(m, int *) = so->so_error;
1038: so->so_error = 0;
1039: break;
1040:
1041: case SO_SNDBUF:
1042: *mtod(m, int *) = so->so_snd.sb_hiwat;
1043: break;
1044:
1045: case SO_RCVBUF:
1046: *mtod(m, int *) = so->so_rcv.sb_hiwat;
1047: break;
1048:
1049: case SO_SNDLOWAT:
1050: *mtod(m, int *) = so->so_snd.sb_lowat;
1051: break;
1052:
1053: case SO_RCVLOWAT:
1054: *mtod(m, int *) = so->so_rcv.sb_lowat;
1055: break;
1056:
1057: case SO_SNDTIMEO:
1058: case SO_RCVTIMEO:
1059: {
1060: int val = (optname == SO_SNDTIMEO ?
1061: so->so_snd.sb_timeo : so->so_rcv.sb_timeo);
1062:
1063: m->m_len = sizeof(struct timeval);
1064: mtod(m, struct timeval *)->tv_sec = val / hz;
1065: mtod(m, struct timeval *)->tv_usec =
1.27 kleink 1066: (val % hz) * tick;
1.1 cgd 1067: break;
1068: }
1069:
1070: default:
1071: (void)m_free(m);
1072: return (ENOPROTOOPT);
1073: }
1074: *mp = m;
1075: return (0);
1076: }
1077: }
1078:
1.14 mycroft 1079: void
1.1 cgd 1080: sohasoutofband(so)
1081: register struct socket *so;
1082: {
1083: struct proc *p;
1084:
1085: if (so->so_pgid < 0)
1086: gsignal(-so->so_pgid, SIGURG);
1087: else if (so->so_pgid > 0 && (p = pfind(so->so_pgid)) != 0)
1088: psignal(p, SIGURG);
1.2 cgd 1089: selwakeup(&so->so_rcv.sb_sel);
1.1 cgd 1090: }
CVSweb <webmaster@jp.NetBSD.org>