Annotation of src/sys/kern/uipc_socket.c, Revision 1.167
1.167 ! ad 1: /* $NetBSD: uipc_socket.c,v 1.166 2008/05/26 17:21:18 ad Exp $ */
1.64 thorpej 2:
3: /*-
1.154 ad 4: * Copyright (c) 2002, 2007, 2008 The NetBSD Foundation, Inc.
1.64 thorpej 5: * All rights reserved.
6: *
7: * This code is derived from software contributed to The NetBSD Foundation
8: * by Jason R. Thorpe of Wasabi Systems, Inc.
9: *
10: * Redistribution and use in source and binary forms, with or without
11: * modification, are permitted provided that the following conditions
12: * are met:
13: * 1. Redistributions of source code must retain the above copyright
14: * notice, this list of conditions and the following disclaimer.
15: * 2. Redistributions in binary form must reproduce the above copyright
16: * notice, this list of conditions and the following disclaimer in the
17: * documentation and/or other materials provided with the distribution.
18: *
19: * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
20: * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
21: * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
22: * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
23: * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
24: * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
25: * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
26: * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
27: * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
28: * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
29: * POSSIBILITY OF SUCH DAMAGE.
30: */
1.16 cgd 31:
1.1 cgd 32: /*
1.159 ad 33: * Copyright (c) 2004 The FreeBSD Foundation
34: * Copyright (c) 2004 Robert Watson
1.15 mycroft 35: * Copyright (c) 1982, 1986, 1988, 1990, 1993
36: * The Regents of the University of California. All rights reserved.
1.1 cgd 37: *
38: * Redistribution and use in source and binary forms, with or without
39: * modification, are permitted provided that the following conditions
40: * are met:
41: * 1. Redistributions of source code must retain the above copyright
42: * notice, this list of conditions and the following disclaimer.
43: * 2. Redistributions in binary form must reproduce the above copyright
44: * notice, this list of conditions and the following disclaimer in the
45: * documentation and/or other materials provided with the distribution.
1.85 agc 46: * 3. Neither the name of the University nor the names of its contributors
1.1 cgd 47: * may be used to endorse or promote products derived from this software
48: * without specific prior written permission.
49: *
50: * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
51: * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
52: * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
53: * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
54: * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
55: * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
56: * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
57: * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
58: * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
59: * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
60: * SUCH DAMAGE.
61: *
1.32 fvdl 62: * @(#)uipc_socket.c 8.6 (Berkeley) 5/2/95
1.1 cgd 63: */
1.59 lukem 64:
65: #include <sys/cdefs.h>
1.167 ! ad 66: __KERNEL_RCSID(0, "$NetBSD: uipc_socket.c,v 1.166 2008/05/26 17:21:18 ad Exp $");
1.64 thorpej 67:
68: #include "opt_sock_counters.h"
69: #include "opt_sosend_loan.h"
1.81 martin 70: #include "opt_mbuftrace.h"
1.84 ragge 71: #include "opt_somaxkva.h"
1.167 ! ad 72: #include "opt_multiprocessor.h" /* XXX */
1.1 cgd 73:
1.9 mycroft 74: #include <sys/param.h>
75: #include <sys/systm.h>
76: #include <sys/proc.h>
77: #include <sys/file.h>
1.142 dyoung 78: #include <sys/filedesc.h>
1.9 mycroft 79: #include <sys/malloc.h>
80: #include <sys/mbuf.h>
81: #include <sys/domain.h>
82: #include <sys/kernel.h>
83: #include <sys/protosw.h>
84: #include <sys/socket.h>
85: #include <sys/socketvar.h>
1.21 christos 86: #include <sys/signalvar.h>
1.9 mycroft 87: #include <sys/resourcevar.h>
1.72 jdolecek 88: #include <sys/event.h>
1.89 christos 89: #include <sys/poll.h>
1.118 elad 90: #include <sys/kauth.h>
1.136 ad 91: #include <sys/mutex.h>
92: #include <sys/condvar.h>
1.37 thorpej 93:
1.64 thorpej 94: #include <uvm/uvm.h>
95:
1.77 thorpej 96: MALLOC_DEFINE(M_SOOPTS, "soopts", "socket options");
97: MALLOC_DEFINE(M_SONAME, "soname", "socket name");
1.37 thorpej 98:
1.142 dyoung 99: extern const struct fileops socketops;
100:
1.54 lukem 101: extern int somaxconn; /* patchable (XXX sysctl) */
102: int somaxconn = SOMAXCONN;
1.160 ad 103: kmutex_t *softnet_lock;
1.49 jonathan 104:
1.64 thorpej 105: #ifdef SOSEND_COUNTERS
106: #include <sys/device.h>
107:
1.113 thorpej 108: static struct evcnt sosend_loan_big = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64 thorpej 109: NULL, "sosend", "loan big");
1.113 thorpej 110: static struct evcnt sosend_copy_big = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64 thorpej 111: NULL, "sosend", "copy big");
1.113 thorpej 112: static struct evcnt sosend_copy_small = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64 thorpej 113: NULL, "sosend", "copy small");
1.113 thorpej 114: static struct evcnt sosend_kvalimit = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64 thorpej 115: NULL, "sosend", "kva limit");
116:
117: #define SOSEND_COUNTER_INCR(ev) (ev)->ev_count++
118:
1.101 matt 119: EVCNT_ATTACH_STATIC(sosend_loan_big);
120: EVCNT_ATTACH_STATIC(sosend_copy_big);
121: EVCNT_ATTACH_STATIC(sosend_copy_small);
122: EVCNT_ATTACH_STATIC(sosend_kvalimit);
1.64 thorpej 123: #else
124:
125: #define SOSEND_COUNTER_INCR(ev) /* nothing */
126:
127: #endif /* SOSEND_COUNTERS */
128:
1.119 yamt 129: static struct callback_entry sokva_reclaimerentry;
1.1 cgd 130:
1.167 ! ad 131: #if defined(SOSEND_NO_LOAN) || defined(MULTIPROCESSOR)
1.121 yamt 132: int sock_loan_thresh = -1;
1.71 thorpej 133: #else
1.121 yamt 134: int sock_loan_thresh = 4096;
1.65 thorpej 135: #endif
1.64 thorpej 136:
1.136 ad 137: static kmutex_t so_pendfree_lock;
1.113 thorpej 138: static struct mbuf *so_pendfree;
1.64 thorpej 139:
1.84 ragge 140: #ifndef SOMAXKVA
141: #define SOMAXKVA (16 * 1024 * 1024)
142: #endif
143: int somaxkva = SOMAXKVA;
1.113 thorpej 144: static int socurkva;
1.136 ad 145: static kcondvar_t socurkva_cv;
1.64 thorpej 146:
147: #define SOCK_LOAN_CHUNK 65536
148:
1.117 yamt 149: static size_t sodopendfree(void);
150: static size_t sodopendfreel(void);
1.93 yamt 151:
1.113 thorpej 152: static vsize_t
1.129 yamt 153: sokvareserve(struct socket *so, vsize_t len)
1.80 yamt 154: {
1.98 christos 155: int error;
1.80 yamt 156:
1.136 ad 157: mutex_enter(&so_pendfree_lock);
1.80 yamt 158: while (socurkva + len > somaxkva) {
1.93 yamt 159: size_t freed;
160:
161: /*
162: * try to do pendfree.
163: */
164:
1.117 yamt 165: freed = sodopendfreel();
1.93 yamt 166:
167: /*
168: * if some kva was freed, try again.
169: */
170:
171: if (freed)
1.80 yamt 172: continue;
1.93 yamt 173:
1.80 yamt 174: SOSEND_COUNTER_INCR(&sosend_kvalimit);
1.136 ad 175: error = cv_wait_sig(&socurkva_cv, &so_pendfree_lock);
1.98 christos 176: if (error) {
177: len = 0;
178: break;
179: }
1.80 yamt 180: }
1.93 yamt 181: socurkva += len;
1.136 ad 182: mutex_exit(&so_pendfree_lock);
1.98 christos 183: return len;
1.95 yamt 184: }
185:
1.113 thorpej 186: static void
1.95 yamt 187: sokvaunreserve(vsize_t len)
188: {
189:
1.136 ad 190: mutex_enter(&so_pendfree_lock);
1.95 yamt 191: socurkva -= len;
1.136 ad 192: cv_broadcast(&socurkva_cv);
193: mutex_exit(&so_pendfree_lock);
1.95 yamt 194: }
195:
196: /*
197: * sokvaalloc: allocate kva for loan.
198: */
199:
200: vaddr_t
201: sokvaalloc(vsize_t len, struct socket *so)
202: {
203: vaddr_t lva;
204:
205: /*
206: * reserve kva.
207: */
208:
1.98 christos 209: if (sokvareserve(so, len) == 0)
210: return 0;
1.93 yamt 211:
212: /*
213: * allocate kva.
214: */
1.80 yamt 215:
1.109 yamt 216: lva = uvm_km_alloc(kernel_map, len, 0, UVM_KMF_VAONLY | UVM_KMF_WAITVA);
1.95 yamt 217: if (lva == 0) {
218: sokvaunreserve(len);
1.80 yamt 219: return (0);
1.95 yamt 220: }
1.80 yamt 221:
222: return lva;
223: }
224:
1.93 yamt 225: /*
226: * sokvafree: free kva for loan.
227: */
228:
1.80 yamt 229: void
230: sokvafree(vaddr_t sva, vsize_t len)
231: {
1.93 yamt 232:
233: /*
234: * free kva.
235: */
1.80 yamt 236:
1.109 yamt 237: uvm_km_free(kernel_map, sva, len, UVM_KMF_VAONLY);
1.93 yamt 238:
239: /*
240: * unreserve kva.
241: */
242:
1.95 yamt 243: sokvaunreserve(len);
1.80 yamt 244: }
245:
1.64 thorpej 246: static void
1.134 christos 247: sodoloanfree(struct vm_page **pgs, void *buf, size_t size)
1.64 thorpej 248: {
1.156 yamt 249: vaddr_t sva, eva;
1.64 thorpej 250: vsize_t len;
1.156 yamt 251: int npgs;
252:
253: KASSERT(pgs != NULL);
1.64 thorpej 254:
255: eva = round_page((vaddr_t) buf + size);
256: sva = trunc_page((vaddr_t) buf);
257: len = eva - sva;
258: npgs = len >> PAGE_SHIFT;
259:
260: pmap_kremove(sva, len);
261: pmap_update(pmap_kernel());
262: uvm_unloan(pgs, npgs, UVM_LOAN_TOPAGE);
1.80 yamt 263: sokvafree(sva, len);
1.64 thorpej 264: }
265:
266: static size_t
1.152 matt 267: sodopendfree(void)
1.64 thorpej 268: {
1.93 yamt 269: size_t rv;
1.64 thorpej 270:
1.160 ad 271: if (__predict_true(so_pendfree == NULL))
272: return 0;
273:
1.136 ad 274: mutex_enter(&so_pendfree_lock);
1.117 yamt 275: rv = sodopendfreel();
1.136 ad 276: mutex_exit(&so_pendfree_lock);
1.93 yamt 277:
278: return rv;
279: }
280:
281: /*
282: * sodopendfreel: free mbufs on "pendfree" list.
1.136 ad 283: * unlock and relock so_pendfree_lock when freeing mbufs.
1.93 yamt 284: *
1.136 ad 285: * => called with so_pendfree_lock held.
1.93 yamt 286: */
287:
288: static size_t
1.152 matt 289: sodopendfreel(void)
1.93 yamt 290: {
1.137 ad 291: struct mbuf *m, *next;
1.93 yamt 292: size_t rv = 0;
293:
1.136 ad 294: KASSERT(mutex_owned(&so_pendfree_lock));
1.64 thorpej 295:
1.137 ad 296: while (so_pendfree != NULL) {
1.64 thorpej 297: m = so_pendfree;
1.93 yamt 298: so_pendfree = NULL;
1.136 ad 299: mutex_exit(&so_pendfree_lock);
1.93 yamt 300:
301: for (; m != NULL; m = next) {
302: next = m->m_next;
1.156 yamt 303: KASSERT((~m->m_flags & (M_EXT|M_EXT_PAGES)) == 0);
304: KASSERT(m->m_ext.ext_refcnt == 0);
1.93 yamt 305:
306: rv += m->m_ext.ext_size;
1.156 yamt 307: sodoloanfree(m->m_ext.ext_pgs, m->m_ext.ext_buf,
1.93 yamt 308: m->m_ext.ext_size);
1.145 ad 309: pool_cache_put(mb_cache, m);
1.93 yamt 310: }
1.64 thorpej 311:
1.136 ad 312: mutex_enter(&so_pendfree_lock);
1.64 thorpej 313: }
314:
315: return (rv);
316: }
317:
1.80 yamt 318: void
1.134 christos 319: soloanfree(struct mbuf *m, void *buf, size_t size, void *arg)
1.64 thorpej 320: {
321:
1.156 yamt 322: KASSERT(m != NULL);
1.64 thorpej 323:
1.93 yamt 324: /*
325: * postpone freeing mbuf.
326: *
327: * we can't do it in interrupt context
328: * because we need to put kva back to kernel_map.
329: */
330:
1.136 ad 331: mutex_enter(&so_pendfree_lock);
1.92 yamt 332: m->m_next = so_pendfree;
333: so_pendfree = m;
1.136 ad 334: cv_broadcast(&socurkva_cv);
335: mutex_exit(&so_pendfree_lock);
1.64 thorpej 336: }
337:
338: static long
339: sosend_loan(struct socket *so, struct uio *uio, struct mbuf *m, long space)
340: {
341: struct iovec *iov = uio->uio_iov;
342: vaddr_t sva, eva;
343: vsize_t len;
1.156 yamt 344: vaddr_t lva;
345: int npgs, error;
346: vaddr_t va;
347: int i;
1.64 thorpej 348:
1.116 yamt 349: if (VMSPACE_IS_KERNEL_P(uio->uio_vmspace))
1.64 thorpej 350: return (0);
351:
352: if (iov->iov_len < (size_t) space)
353: space = iov->iov_len;
354: if (space > SOCK_LOAN_CHUNK)
355: space = SOCK_LOAN_CHUNK;
356:
357: eva = round_page((vaddr_t) iov->iov_base + space);
358: sva = trunc_page((vaddr_t) iov->iov_base);
359: len = eva - sva;
360: npgs = len >> PAGE_SHIFT;
361:
1.79 thorpej 362: KASSERT(npgs <= M_EXT_MAXPAGES);
363:
1.80 yamt 364: lva = sokvaalloc(len, so);
1.64 thorpej 365: if (lva == 0)
1.80 yamt 366: return 0;
1.64 thorpej 367:
1.116 yamt 368: error = uvm_loan(&uio->uio_vmspace->vm_map, sva, len,
1.79 thorpej 369: m->m_ext.ext_pgs, UVM_LOAN_TOPAGE);
1.64 thorpej 370: if (error) {
1.80 yamt 371: sokvafree(lva, len);
1.64 thorpej 372: return (0);
373: }
374:
375: for (i = 0, va = lva; i < npgs; i++, va += PAGE_SIZE)
1.79 thorpej 376: pmap_kenter_pa(va, VM_PAGE_TO_PHYS(m->m_ext.ext_pgs[i]),
377: VM_PROT_READ);
1.64 thorpej 378: pmap_update(pmap_kernel());
379:
380: lva += (vaddr_t) iov->iov_base & PAGE_MASK;
381:
1.134 christos 382: MEXTADD(m, (void *) lva, space, M_MBUF, soloanfree, so);
1.79 thorpej 383: m->m_flags |= M_EXT_PAGES | M_EXT_ROMAP;
1.64 thorpej 384:
385: uio->uio_resid -= space;
386: /* uio_offset not updated, not set/used for write(2) */
1.134 christos 387: uio->uio_iov->iov_base = (char *)uio->uio_iov->iov_base + space;
1.64 thorpej 388: uio->uio_iov->iov_len -= space;
389: if (uio->uio_iov->iov_len == 0) {
390: uio->uio_iov++;
391: uio->uio_iovcnt--;
392: }
393:
394: return (space);
395: }
396:
1.119 yamt 397: static int
1.129 yamt 398: sokva_reclaim_callback(struct callback_entry *ce, void *obj, void *arg)
1.119 yamt 399: {
400:
401: KASSERT(ce == &sokva_reclaimerentry);
402: KASSERT(obj == NULL);
403:
404: sodopendfree();
405: if (!vm_map_starved_p(kernel_map)) {
406: return CALLBACK_CHAIN_ABORT;
407: }
408: return CALLBACK_CHAIN_CONTINUE;
409: }
410:
1.142 dyoung 411: struct mbuf *
1.147 dyoung 412: getsombuf(struct socket *so, int type)
1.142 dyoung 413: {
414: struct mbuf *m;
415:
1.147 dyoung 416: m = m_get(M_WAIT, type);
1.142 dyoung 417: MCLAIM(m, so->so_mowner);
418: return m;
419: }
420:
421: struct mbuf *
422: m_intopt(struct socket *so, int val)
423: {
424: struct mbuf *m;
425:
1.147 dyoung 426: m = getsombuf(so, MT_SOOPTS);
1.142 dyoung 427: m->m_len = sizeof(int);
428: *mtod(m, int *) = val;
429: return m;
430: }
431:
1.119 yamt 432: void
433: soinit(void)
434: {
435:
1.148 ad 436: mutex_init(&so_pendfree_lock, MUTEX_DEFAULT, IPL_VM);
1.160 ad 437: softnet_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NONE);
1.136 ad 438: cv_init(&socurkva_cv, "sokva");
1.166 ad 439: soinit2();
1.136 ad 440:
1.119 yamt 441: /* Set the initial adjusted socket buffer size. */
442: if (sb_max_set(sb_max))
443: panic("bad initial sb_max value: %lu", sb_max);
444:
445: callback_register(&vm_map_to_kernel(kernel_map)->vmk_reclaim_callback,
446: &sokva_reclaimerentry, NULL, sokva_reclaim_callback);
447: }
448:
1.1 cgd 449: /*
450: * Socket operation routines.
451: * These routines are called by the routines in
452: * sys_socket.c or from a system process, and
453: * implement the semantics of socket operations by
454: * switching out to the protocol specific routines.
455: */
456: /*ARGSUSED*/
1.3 andrew 457: int
1.160 ad 458: socreate(int dom, struct socket **aso, int type, int proto, struct lwp *l,
459: struct socket *lockso)
1.1 cgd 460: {
1.99 matt 461: const struct protosw *prp;
1.54 lukem 462: struct socket *so;
1.115 yamt 463: uid_t uid;
1.160 ad 464: int error;
465: kmutex_t *lock;
1.1 cgd 466:
1.132 elad 467: error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET,
468: KAUTH_REQ_NETWORK_SOCKET_OPEN, KAUTH_ARG(dom), KAUTH_ARG(type),
469: KAUTH_ARG(proto));
1.140 dyoung 470: if (error != 0)
471: return error;
1.127 elad 472:
1.1 cgd 473: if (proto)
474: prp = pffindproto(dom, proto, type);
475: else
476: prp = pffindtype(dom, type);
1.140 dyoung 477: if (prp == NULL) {
1.120 ginsbach 478: /* no support for domain */
479: if (pffinddomain(dom) == 0)
1.140 dyoung 480: return EAFNOSUPPORT;
1.120 ginsbach 481: /* no support for socket type */
482: if (proto == 0 && type != 0)
1.140 dyoung 483: return EPROTOTYPE;
484: return EPROTONOSUPPORT;
1.120 ginsbach 485: }
1.140 dyoung 486: if (prp->pr_usrreq == NULL)
487: return EPROTONOSUPPORT;
1.1 cgd 488: if (prp->pr_type != type)
1.140 dyoung 489: return EPROTOTYPE;
1.160 ad 490:
491: so = soget(true);
1.1 cgd 492: so->so_type = type;
493: so->so_proto = prp;
1.33 matt 494: so->so_send = sosend;
495: so->so_receive = soreceive;
1.78 matt 496: #ifdef MBUFTRACE
497: so->so_rcv.sb_mowner = &prp->pr_domain->dom_mowner;
498: so->so_snd.sb_mowner = &prp->pr_domain->dom_mowner;
499: so->so_mowner = &prp->pr_domain->dom_mowner;
500: #endif
1.138 rmind 501: uid = kauth_cred_geteuid(l->l_cred);
1.115 yamt 502: so->so_uidinfo = uid_find(uid);
1.160 ad 503: if (lockso != NULL) {
504: /* Caller wants us to share a lock. */
505: lock = lockso->so_lock;
506: so->so_lock = lock;
507: mutex_obj_hold(lock);
508: mutex_enter(lock);
509: } else {
510: /* Lock assigned and taken during PRU_ATTACH. */
511: }
1.140 dyoung 512: error = (*prp->pr_usrreq)(so, PRU_ATTACH, NULL,
513: (struct mbuf *)(long)proto, NULL, l);
1.160 ad 514: KASSERT(solocked(so));
1.140 dyoung 515: if (error != 0) {
1.1 cgd 516: so->so_state |= SS_NOFDREF;
517: sofree(so);
1.140 dyoung 518: return error;
1.1 cgd 519: }
1.160 ad 520: sounlock(so);
1.1 cgd 521: *aso = so;
1.140 dyoung 522: return 0;
1.1 cgd 523: }
524:
1.142 dyoung 525: /* On success, write file descriptor to fdout and return zero. On
526: * failure, return non-zero; *fdout will be undefined.
527: */
528: int
529: fsocreate(int domain, struct socket **sop, int type, int protocol,
530: struct lwp *l, int *fdout)
531: {
532: struct socket *so;
533: struct file *fp;
534: int fd, error;
535:
1.155 ad 536: if ((error = fd_allocfile(&fp, &fd)) != 0)
1.142 dyoung 537: return (error);
538: fp->f_flag = FREAD|FWRITE;
539: fp->f_type = DTYPE_SOCKET;
540: fp->f_ops = &socketops;
1.160 ad 541: error = socreate(domain, &so, type, protocol, l, NULL);
1.142 dyoung 542: if (error != 0) {
1.155 ad 543: fd_abort(curproc, fp, fd);
1.142 dyoung 544: } else {
545: if (sop != NULL)
546: *sop = so;
547: fp->f_data = so;
1.155 ad 548: fd_affix(curproc, fp, fd);
1.142 dyoung 549: *fdout = fd;
550: }
551: return error;
552: }
553:
1.3 andrew 554: int
1.114 christos 555: sobind(struct socket *so, struct mbuf *nam, struct lwp *l)
1.1 cgd 556: {
1.160 ad 557: int error;
1.1 cgd 558:
1.160 ad 559: solock(so);
1.140 dyoung 560: error = (*so->so_proto->pr_usrreq)(so, PRU_BIND, NULL, nam, NULL, l);
1.160 ad 561: sounlock(so);
1.140 dyoung 562: return error;
1.1 cgd 563: }
564:
1.3 andrew 565: int
1.150 elad 566: solisten(struct socket *so, int backlog, struct lwp *l)
1.1 cgd 567: {
1.160 ad 568: int error;
1.1 cgd 569:
1.160 ad 570: solock(so);
1.158 ad 571: if ((so->so_state & (SS_ISCONNECTED | SS_ISCONNECTING |
1.163 ad 572: SS_ISDISCONNECTING)) != 0) {
573: sounlock(so);
1.158 ad 574: return (EOPNOTSUPP);
1.163 ad 575: }
1.140 dyoung 576: error = (*so->so_proto->pr_usrreq)(so, PRU_LISTEN, NULL,
1.150 elad 577: NULL, NULL, l);
1.140 dyoung 578: if (error != 0) {
1.160 ad 579: sounlock(so);
1.140 dyoung 580: return error;
1.1 cgd 581: }
1.63 matt 582: if (TAILQ_EMPTY(&so->so_q))
1.1 cgd 583: so->so_options |= SO_ACCEPTCONN;
584: if (backlog < 0)
585: backlog = 0;
1.49 jonathan 586: so->so_qlimit = min(backlog, somaxconn);
1.160 ad 587: sounlock(so);
1.140 dyoung 588: return 0;
1.1 cgd 589: }
590:
1.21 christos 591: void
1.54 lukem 592: sofree(struct socket *so)
1.1 cgd 593: {
1.161 ad 594: u_int refs;
1.1 cgd 595:
1.160 ad 596: KASSERT(solocked(so));
597:
598: if (so->so_pcb || (so->so_state & SS_NOFDREF) == 0) {
599: sounlock(so);
1.1 cgd 600: return;
1.160 ad 601: }
1.43 mycroft 602: if (so->so_head) {
603: /*
604: * We must not decommission a socket that's on the accept(2)
605: * queue. If we do, then accept(2) may hang after select(2)
606: * indicated that the listening socket was ready.
607: */
1.160 ad 608: if (!soqremque(so, 0)) {
609: sounlock(so);
1.43 mycroft 610: return;
1.160 ad 611: }
1.43 mycroft 612: }
1.98 christos 613: if (so->so_rcv.sb_hiwat)
1.110 christos 614: (void)chgsbsize(so->so_uidinfo, &so->so_rcv.sb_hiwat, 0,
1.98 christos 615: RLIM_INFINITY);
616: if (so->so_snd.sb_hiwat)
1.110 christos 617: (void)chgsbsize(so->so_uidinfo, &so->so_snd.sb_hiwat, 0,
1.98 christos 618: RLIM_INFINITY);
619: sbrelease(&so->so_snd, so);
1.160 ad 620: KASSERT(!cv_has_waiters(&so->so_cv));
621: KASSERT(!cv_has_waiters(&so->so_rcv.sb_cv));
622: KASSERT(!cv_has_waiters(&so->so_snd.sb_cv));
1.1 cgd 623: sorflush(so);
1.161 ad 624: refs = so->so_aborting; /* XXX */
1.160 ad 625: sounlock(so);
1.161 ad 626: if (refs == 0) /* XXX */
627: soput(so);
1.1 cgd 628: }
629:
630: /*
631: * Close a socket on last file table reference removal.
632: * Initiate disconnect if connected.
633: * Free socket when disconnect complete.
634: */
1.3 andrew 635: int
1.54 lukem 636: soclose(struct socket *so)
1.1 cgd 637: {
1.54 lukem 638: struct socket *so2;
1.160 ad 639: int error;
640: int error2;
1.1 cgd 641:
1.54 lukem 642: error = 0;
1.160 ad 643: solock(so);
1.1 cgd 644: if (so->so_options & SO_ACCEPTCONN) {
1.160 ad 645: do {
646: if ((so2 = TAILQ_FIRST(&so->so_q0)) != 0) {
647: KASSERT(solocked2(so, so2));
648: (void) soqremque(so2, 0);
649: /* soabort drops the lock. */
650: (void) soabort(so2);
651: solock(so);
652: continue;
653: }
654: if ((so2 = TAILQ_FIRST(&so->so_q)) != 0) {
655: KASSERT(solocked2(so, so2));
656: (void) soqremque(so2, 1);
657: /* soabort drops the lock. */
658: (void) soabort(so2);
659: solock(so);
660: continue;
661: }
662: } while (0);
1.1 cgd 663: }
664: if (so->so_pcb == 0)
665: goto discard;
666: if (so->so_state & SS_ISCONNECTED) {
667: if ((so->so_state & SS_ISDISCONNECTING) == 0) {
668: error = sodisconnect(so);
669: if (error)
670: goto drop;
671: }
672: if (so->so_options & SO_LINGER) {
1.151 ad 673: if ((so->so_state & SS_ISDISCONNECTING) && so->so_nbio)
1.1 cgd 674: goto drop;
1.21 christos 675: while (so->so_state & SS_ISCONNECTED) {
1.160 ad 676: error = sowait(so, so->so_linger * hz);
1.21 christos 677: if (error)
1.1 cgd 678: break;
1.21 christos 679: }
1.1 cgd 680: }
681: }
1.54 lukem 682: drop:
1.1 cgd 683: if (so->so_pcb) {
1.160 ad 684: error2 = (*so->so_proto->pr_usrreq)(so, PRU_DETACH,
1.140 dyoung 685: NULL, NULL, NULL, NULL);
1.1 cgd 686: if (error == 0)
687: error = error2;
688: }
1.54 lukem 689: discard:
1.1 cgd 690: if (so->so_state & SS_NOFDREF)
691: panic("soclose: NOFDREF");
692: so->so_state |= SS_NOFDREF;
693: sofree(so);
694: return (error);
695: }
696:
697: /*
1.160 ad 698: * Must be called with the socket locked.. Will return with it unlocked.
1.1 cgd 699: */
1.3 andrew 700: int
1.54 lukem 701: soabort(struct socket *so)
1.1 cgd 702: {
1.161 ad 703: u_int refs;
1.139 yamt 704: int error;
1.160 ad 705:
706: KASSERT(solocked(so));
707: KASSERT(so->so_head == NULL);
1.1 cgd 708:
1.161 ad 709: so->so_aborting++; /* XXX */
1.140 dyoung 710: error = (*so->so_proto->pr_usrreq)(so, PRU_ABORT, NULL,
711: NULL, NULL, NULL);
1.161 ad 712: refs = --so->so_aborting; /* XXX */
1.164 drochner 713: if (error || (refs == 0)) {
1.139 yamt 714: sofree(so);
1.160 ad 715: } else {
716: sounlock(so);
1.139 yamt 717: }
718: return error;
1.1 cgd 719: }
720:
1.3 andrew 721: int
1.54 lukem 722: soaccept(struct socket *so, struct mbuf *nam)
1.1 cgd 723: {
1.160 ad 724: int error;
725:
726: KASSERT(solocked(so));
1.1 cgd 727:
1.54 lukem 728: error = 0;
1.1 cgd 729: if ((so->so_state & SS_NOFDREF) == 0)
730: panic("soaccept: !NOFDREF");
731: so->so_state &= ~SS_NOFDREF;
1.55 thorpej 732: if ((so->so_state & SS_ISDISCONNECTED) == 0 ||
733: (so->so_proto->pr_flags & PR_ABRTACPTDIS) == 0)
1.41 mycroft 734: error = (*so->so_proto->pr_usrreq)(so, PRU_ACCEPT,
1.140 dyoung 735: NULL, nam, NULL, NULL);
1.41 mycroft 736: else
1.53 itojun 737: error = ECONNABORTED;
1.52 itojun 738:
1.1 cgd 739: return (error);
740: }
741:
1.3 andrew 742: int
1.114 christos 743: soconnect(struct socket *so, struct mbuf *nam, struct lwp *l)
1.1 cgd 744: {
1.160 ad 745: int error;
746:
747: KASSERT(solocked(so));
1.1 cgd 748:
749: if (so->so_options & SO_ACCEPTCONN)
750: return (EOPNOTSUPP);
751: /*
752: * If protocol is connection-based, can only connect once.
753: * Otherwise, if connected, try to disconnect first.
754: * This allows user to disconnect by connecting to, e.g.,
755: * a null address.
756: */
757: if (so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING) &&
758: ((so->so_proto->pr_flags & PR_CONNREQUIRED) ||
759: (error = sodisconnect(so))))
760: error = EISCONN;
761: else
762: error = (*so->so_proto->pr_usrreq)(so, PRU_CONNECT,
1.140 dyoung 763: NULL, nam, NULL, l);
1.1 cgd 764: return (error);
765: }
766:
1.3 andrew 767: int
1.54 lukem 768: soconnect2(struct socket *so1, struct socket *so2)
1.1 cgd 769: {
1.160 ad 770: int error;
771:
772: KASSERT(solocked2(so1, so2));
1.1 cgd 773:
1.22 mycroft 774: error = (*so1->so_proto->pr_usrreq)(so1, PRU_CONNECT2,
1.140 dyoung 775: NULL, (struct mbuf *)so2, NULL, NULL);
1.1 cgd 776: return (error);
777: }
778:
1.3 andrew 779: int
1.54 lukem 780: sodisconnect(struct socket *so)
1.1 cgd 781: {
1.160 ad 782: int error;
783:
784: KASSERT(solocked(so));
1.1 cgd 785:
786: if ((so->so_state & SS_ISCONNECTED) == 0) {
787: error = ENOTCONN;
1.160 ad 788: } else if (so->so_state & SS_ISDISCONNECTING) {
1.1 cgd 789: error = EALREADY;
1.160 ad 790: } else {
791: error = (*so->so_proto->pr_usrreq)(so, PRU_DISCONNECT,
792: NULL, NULL, NULL, NULL);
1.1 cgd 793: }
1.117 yamt 794: sodopendfree();
1.1 cgd 795: return (error);
796: }
797:
1.15 mycroft 798: #define SBLOCKWAIT(f) (((f) & MSG_DONTWAIT) ? M_NOWAIT : M_WAITOK)
1.1 cgd 799: /*
800: * Send on a socket.
801: * If send must go all at once and message is larger than
802: * send buffering, then hard error.
803: * Lock against other senders.
804: * If must go all at once and not enough room now, then
805: * inform user that this would block and do nothing.
806: * Otherwise, if nonblocking, send as much as possible.
807: * The data to be sent is described by "uio" if nonzero,
808: * otherwise by the mbuf chain "top" (which must be null
809: * if uio is not). Data provided in mbuf chain must be small
810: * enough to send all at once.
811: *
812: * Returns nonzero on error, timeout or signal; callers
813: * must check for short counts if EINTR/ERESTART are returned.
814: * Data and control buffers are freed on return.
815: */
1.3 andrew 816: int
1.54 lukem 817: sosend(struct socket *so, struct mbuf *addr, struct uio *uio, struct mbuf *top,
1.114 christos 818: struct mbuf *control, int flags, struct lwp *l)
1.1 cgd 819: {
1.54 lukem 820: struct mbuf **mp, *m;
1.114 christos 821: struct proc *p;
1.58 jdolecek 822: long space, len, resid, clen, mlen;
823: int error, s, dontroute, atomic;
1.54 lukem 824:
1.114 christos 825: p = l->l_proc;
1.117 yamt 826: sodopendfree();
1.160 ad 827: clen = 0;
1.64 thorpej 828:
1.160 ad 829: /*
830: * solock() provides atomicity of access. splsoftnet() prevents
831: * protocol processing soft interrupts from interrupting us and
832: * blocking (expensive).
833: */
834: s = splsoftnet();
835: solock(so);
1.54 lukem 836: atomic = sosendallatonce(so) || top;
1.1 cgd 837: if (uio)
838: resid = uio->uio_resid;
839: else
840: resid = top->m_pkthdr.len;
1.7 cgd 841: /*
842: * In theory resid should be unsigned.
843: * However, space must be signed, as it might be less than 0
844: * if we over-committed, and we must use a signed comparison
845: * of space and resid. On the other hand, a negative resid
846: * causes us to loop sending 0-length segments to the protocol.
847: */
1.29 mycroft 848: if (resid < 0) {
849: error = EINVAL;
850: goto out;
851: }
1.1 cgd 852: dontroute =
853: (flags & MSG_DONTROUTE) && (so->so_options & SO_DONTROUTE) == 0 &&
854: (so->so_proto->pr_flags & PR_ATOMIC);
1.165 christos 855: l->l_ru.ru_msgsnd++;
1.1 cgd 856: if (control)
857: clen = control->m_len;
1.54 lukem 858: restart:
1.21 christos 859: if ((error = sblock(&so->so_snd, SBLOCKWAIT(flags))) != 0)
1.1 cgd 860: goto out;
861: do {
1.160 ad 862: if (so->so_state & SS_CANTSENDMORE) {
863: error = EPIPE;
864: goto release;
865: }
1.48 thorpej 866: if (so->so_error) {
867: error = so->so_error;
868: so->so_error = 0;
869: goto release;
870: }
1.1 cgd 871: if ((so->so_state & SS_ISCONNECTED) == 0) {
872: if (so->so_proto->pr_flags & PR_CONNREQUIRED) {
873: if ((so->so_state & SS_ISCONFIRMING) == 0 &&
1.160 ad 874: !(resid == 0 && clen != 0)) {
875: error = ENOTCONN;
876: goto release;
877: }
878: } else if (addr == 0) {
879: error = EDESTADDRREQ;
880: goto release;
881: }
1.1 cgd 882: }
883: space = sbspace(&so->so_snd);
884: if (flags & MSG_OOB)
885: space += 1024;
1.21 christos 886: if ((atomic && resid > so->so_snd.sb_hiwat) ||
1.160 ad 887: clen > so->so_snd.sb_hiwat) {
888: error = EMSGSIZE;
889: goto release;
890: }
1.96 mycroft 891: if (space < resid + clen &&
1.1 cgd 892: (atomic || space < so->so_snd.sb_lowat || space < clen)) {
1.160 ad 893: if (so->so_nbio) {
894: error = EWOULDBLOCK;
895: goto release;
896: }
1.1 cgd 897: sbunlock(&so->so_snd);
898: error = sbwait(&so->so_snd);
899: if (error)
900: goto out;
901: goto restart;
902: }
903: mp = ⊤
904: space -= clen;
905: do {
1.45 tv 906: if (uio == NULL) {
907: /*
908: * Data is prepackaged in "top".
909: */
910: resid = 0;
911: if (flags & MSG_EOR)
912: top->m_flags |= M_EOR;
913: } else do {
1.160 ad 914: sounlock(so);
915: splx(s);
1.144 dyoung 916: if (top == NULL) {
1.78 matt 917: m = m_gethdr(M_WAIT, MT_DATA);
1.45 tv 918: mlen = MHLEN;
919: m->m_pkthdr.len = 0;
1.140 dyoung 920: m->m_pkthdr.rcvif = NULL;
1.45 tv 921: } else {
1.78 matt 922: m = m_get(M_WAIT, MT_DATA);
1.45 tv 923: mlen = MLEN;
924: }
1.78 matt 925: MCLAIM(m, so->so_snd.sb_mowner);
1.121 yamt 926: if (sock_loan_thresh >= 0 &&
927: uio->uio_iov->iov_len >= sock_loan_thresh &&
928: space >= sock_loan_thresh &&
1.64 thorpej 929: (len = sosend_loan(so, uio, m,
930: space)) != 0) {
931: SOSEND_COUNTER_INCR(&sosend_loan_big);
932: space -= len;
933: goto have_data;
934: }
1.45 tv 935: if (resid >= MINCLSIZE && space >= MCLBYTES) {
1.64 thorpej 936: SOSEND_COUNTER_INCR(&sosend_copy_big);
1.78 matt 937: m_clget(m, M_WAIT);
1.45 tv 938: if ((m->m_flags & M_EXT) == 0)
939: goto nopages;
940: mlen = MCLBYTES;
941: if (atomic && top == 0) {
1.58 jdolecek 942: len = lmin(MCLBYTES - max_hdr,
1.54 lukem 943: resid);
1.45 tv 944: m->m_data += max_hdr;
945: } else
1.58 jdolecek 946: len = lmin(MCLBYTES, resid);
1.45 tv 947: space -= len;
948: } else {
1.64 thorpej 949: nopages:
950: SOSEND_COUNTER_INCR(&sosend_copy_small);
1.58 jdolecek 951: len = lmin(lmin(mlen, resid), space);
1.45 tv 952: space -= len;
953: /*
954: * For datagram protocols, leave room
955: * for protocol headers in first mbuf.
956: */
957: if (atomic && top == 0 && len < mlen)
958: MH_ALIGN(m, len);
959: }
1.144 dyoung 960: error = uiomove(mtod(m, void *), (int)len, uio);
1.64 thorpej 961: have_data:
1.45 tv 962: resid = uio->uio_resid;
963: m->m_len = len;
964: *mp = m;
965: top->m_pkthdr.len += len;
1.160 ad 966: s = splsoftnet();
967: solock(so);
1.144 dyoung 968: if (error != 0)
1.45 tv 969: goto release;
970: mp = &m->m_next;
971: if (resid <= 0) {
972: if (flags & MSG_EOR)
973: top->m_flags |= M_EOR;
974: break;
975: }
976: } while (space > 0 && atomic);
1.108 perry 977:
1.160 ad 978: if (so->so_state & SS_CANTSENDMORE) {
979: error = EPIPE;
980: goto release;
981: }
1.45 tv 982: if (dontroute)
983: so->so_options |= SO_DONTROUTE;
984: if (resid > 0)
985: so->so_state |= SS_MORETOCOME;
1.46 sommerfe 986: error = (*so->so_proto->pr_usrreq)(so,
987: (flags & MSG_OOB) ? PRU_SENDOOB : PRU_SEND,
1.160 ad 988: top, addr, control, curlwp);
1.45 tv 989: if (dontroute)
990: so->so_options &= ~SO_DONTROUTE;
991: if (resid > 0)
992: so->so_state &= ~SS_MORETOCOME;
993: clen = 0;
1.144 dyoung 994: control = NULL;
995: top = NULL;
1.45 tv 996: mp = ⊤
1.144 dyoung 997: if (error != 0)
1.1 cgd 998: goto release;
999: } while (resid && space > 0);
1000: } while (resid);
1001:
1.54 lukem 1002: release:
1.1 cgd 1003: sbunlock(&so->so_snd);
1.54 lukem 1004: out:
1.160 ad 1005: sounlock(so);
1006: splx(s);
1.1 cgd 1007: if (top)
1008: m_freem(top);
1009: if (control)
1010: m_freem(control);
1011: return (error);
1012: }
1013:
1014: /*
1.159 ad 1015: * Following replacement or removal of the first mbuf on the first
1016: * mbuf chain of a socket buffer, push necessary state changes back
1017: * into the socket buffer so that other consumers see the values
1018: * consistently. 'nextrecord' is the callers locally stored value of
1019: * the original value of sb->sb_mb->m_nextpkt which must be restored
1020: * when the lead mbuf changes. NOTE: 'nextrecord' may be NULL.
1021: */
1022: static void
1023: sbsync(struct sockbuf *sb, struct mbuf *nextrecord)
1024: {
1025:
1.160 ad 1026: KASSERT(solocked(sb->sb_so));
1027:
1.159 ad 1028: /*
1029: * First, update for the new value of nextrecord. If necessary,
1030: * make it the first record.
1031: */
1032: if (sb->sb_mb != NULL)
1033: sb->sb_mb->m_nextpkt = nextrecord;
1034: else
1035: sb->sb_mb = nextrecord;
1036:
1037: /*
1038: * Now update any dependent socket buffer fields to reflect
1039: * the new state. This is an inline of SB_EMPTY_FIXUP, with
1040: * the addition of a second clause that takes care of the
1041: * case where sb_mb has been updated, but remains the last
1042: * record.
1043: */
1044: if (sb->sb_mb == NULL) {
1045: sb->sb_mbtail = NULL;
1046: sb->sb_lastrecord = NULL;
1047: } else if (sb->sb_mb->m_nextpkt == NULL)
1048: sb->sb_lastrecord = sb->sb_mb;
1049: }
1050:
1051: /*
1.1 cgd 1052: * Implement receive operations on a socket.
1053: * We depend on the way that records are added to the sockbuf
1054: * by sbappend*. In particular, each record (mbufs linked through m_next)
1055: * must begin with an address if the protocol so specifies,
1056: * followed by an optional mbuf or mbufs containing ancillary data,
1057: * and then zero or more mbufs of data.
1058: * In order to avoid blocking network interrupts for the entire time here,
1059: * we splx() while doing the actual copy to user space.
1060: * Although the sockbuf is locked, new data may still be appended,
1061: * and thus we must maintain consistency of the sockbuf during that time.
1062: *
1063: * The caller may receive the data as a single mbuf chain by supplying
1064: * an mbuf **mp0 for use in returning the chain. The uio is then used
1065: * only for the count in uio_resid.
1066: */
1.3 andrew 1067: int
1.54 lukem 1068: soreceive(struct socket *so, struct mbuf **paddr, struct uio *uio,
1069: struct mbuf **mp0, struct mbuf **controlp, int *flagsp)
1.1 cgd 1070: {
1.116 yamt 1071: struct lwp *l = curlwp;
1.160 ad 1072: struct mbuf *m, **mp, *mt;
1.146 dyoung 1073: int atomic, flags, len, error, s, offset, moff, type, orig_resid;
1.99 matt 1074: const struct protosw *pr;
1.54 lukem 1075: struct mbuf *nextrecord;
1.67 he 1076: int mbuf_removed = 0;
1.146 dyoung 1077: const struct domain *dom;
1.64 thorpej 1078:
1.54 lukem 1079: pr = so->so_proto;
1.146 dyoung 1080: atomic = pr->pr_flags & PR_ATOMIC;
1081: dom = pr->pr_domain;
1.1 cgd 1082: mp = mp0;
1.54 lukem 1083: type = 0;
1084: orig_resid = uio->uio_resid;
1.102 jonathan 1085:
1.144 dyoung 1086: if (paddr != NULL)
1087: *paddr = NULL;
1088: if (controlp != NULL)
1089: *controlp = NULL;
1090: if (flagsp != NULL)
1.1 cgd 1091: flags = *flagsp &~ MSG_EOR;
1092: else
1093: flags = 0;
1.66 enami 1094:
1095: if ((flags & MSG_DONTWAIT) == 0)
1.117 yamt 1096: sodopendfree();
1.66 enami 1097:
1.1 cgd 1098: if (flags & MSG_OOB) {
1099: m = m_get(M_WAIT, MT_DATA);
1.160 ad 1100: solock(so);
1.17 cgd 1101: error = (*pr->pr_usrreq)(so, PRU_RCVOOB, m,
1.140 dyoung 1102: (struct mbuf *)(long)(flags & MSG_PEEK), NULL, l);
1.160 ad 1103: sounlock(so);
1.1 cgd 1104: if (error)
1105: goto bad;
1106: do {
1.134 christos 1107: error = uiomove(mtod(m, void *),
1.1 cgd 1108: (int) min(uio->uio_resid, m->m_len), uio);
1109: m = m_free(m);
1.144 dyoung 1110: } while (uio->uio_resid > 0 && error == 0 && m);
1.54 lukem 1111: bad:
1.144 dyoung 1112: if (m != NULL)
1.1 cgd 1113: m_freem(m);
1.144 dyoung 1114: return error;
1.1 cgd 1115: }
1.144 dyoung 1116: if (mp != NULL)
1.140 dyoung 1117: *mp = NULL;
1.160 ad 1118:
1119: /*
1120: * solock() provides atomicity of access. splsoftnet() prevents
1121: * protocol processing soft interrupts from interrupting us and
1122: * blocking (expensive).
1123: */
1124: s = splsoftnet();
1125: solock(so);
1.1 cgd 1126: if (so->so_state & SS_ISCONFIRMING && uio->uio_resid)
1.140 dyoung 1127: (*pr->pr_usrreq)(so, PRU_RCVD, NULL, NULL, NULL, l);
1.1 cgd 1128:
1.54 lukem 1129: restart:
1.160 ad 1130: if ((error = sblock(&so->so_rcv, SBLOCKWAIT(flags))) != 0) {
1131: sounlock(so);
1132: splx(s);
1.144 dyoung 1133: return error;
1.160 ad 1134: }
1.1 cgd 1135:
1136: m = so->so_rcv.sb_mb;
1137: /*
1138: * If we have less data than requested, block awaiting more
1139: * (subject to any timeout) if:
1.15 mycroft 1140: * 1. the current count is less than the low water mark,
1.1 cgd 1141: * 2. MSG_WAITALL is set, and it is possible to do the entire
1.15 mycroft 1142: * receive operation at once if we block (resid <= hiwat), or
1143: * 3. MSG_DONTWAIT is not set.
1.1 cgd 1144: * If MSG_WAITALL is set but resid is larger than the receive buffer,
1145: * we have to do the receive in sections, and thus risk returning
1146: * a short count if a timeout or signal occurs after we start.
1147: */
1.144 dyoung 1148: if (m == NULL ||
1149: ((flags & MSG_DONTWAIT) == 0 &&
1150: so->so_rcv.sb_cc < uio->uio_resid &&
1151: (so->so_rcv.sb_cc < so->so_rcv.sb_lowat ||
1152: ((flags & MSG_WAITALL) &&
1153: uio->uio_resid <= so->so_rcv.sb_hiwat)) &&
1.146 dyoung 1154: m->m_nextpkt == NULL && !atomic)) {
1.1 cgd 1155: #ifdef DIAGNOSTIC
1.144 dyoung 1156: if (m == NULL && so->so_rcv.sb_cc)
1.1 cgd 1157: panic("receive 1");
1158: #endif
1159: if (so->so_error) {
1.144 dyoung 1160: if (m != NULL)
1.15 mycroft 1161: goto dontblock;
1.1 cgd 1162: error = so->so_error;
1163: if ((flags & MSG_PEEK) == 0)
1164: so->so_error = 0;
1165: goto release;
1166: }
1167: if (so->so_state & SS_CANTRCVMORE) {
1.144 dyoung 1168: if (m != NULL)
1.15 mycroft 1169: goto dontblock;
1.1 cgd 1170: else
1171: goto release;
1172: }
1.144 dyoung 1173: for (; m != NULL; m = m->m_next)
1.1 cgd 1174: if (m->m_type == MT_OOBDATA || (m->m_flags & M_EOR)) {
1175: m = so->so_rcv.sb_mb;
1176: goto dontblock;
1177: }
1178: if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) == 0 &&
1179: (so->so_proto->pr_flags & PR_CONNREQUIRED)) {
1180: error = ENOTCONN;
1181: goto release;
1182: }
1183: if (uio->uio_resid == 0)
1184: goto release;
1.151 ad 1185: if (so->so_nbio || (flags & MSG_DONTWAIT)) {
1.1 cgd 1186: error = EWOULDBLOCK;
1187: goto release;
1188: }
1.69 thorpej 1189: SBLASTRECORDCHK(&so->so_rcv, "soreceive sbwait 1");
1190: SBLASTMBUFCHK(&so->so_rcv, "soreceive sbwait 1");
1.1 cgd 1191: sbunlock(&so->so_rcv);
1192: error = sbwait(&so->so_rcv);
1.160 ad 1193: if (error != 0) {
1194: sounlock(so);
1195: splx(s);
1.144 dyoung 1196: return error;
1.160 ad 1197: }
1.1 cgd 1198: goto restart;
1199: }
1.54 lukem 1200: dontblock:
1.69 thorpej 1201: /*
1202: * On entry here, m points to the first record of the socket buffer.
1.159 ad 1203: * From this point onward, we maintain 'nextrecord' as a cache of the
1204: * pointer to the next record in the socket buffer. We must keep the
1205: * various socket buffer pointers and local stack versions of the
1206: * pointers in sync, pushing out modifications before dropping the
1.160 ad 1207: * socket lock, and re-reading them when picking it up.
1.159 ad 1208: *
1209: * Otherwise, we will race with the network stack appending new data
1210: * or records onto the socket buffer by using inconsistent/stale
1211: * versions of the field, possibly resulting in socket buffer
1212: * corruption.
1213: *
1214: * By holding the high-level sblock(), we prevent simultaneous
1215: * readers from pulling off the front of the socket buffer.
1.69 thorpej 1216: */
1.144 dyoung 1217: if (l != NULL)
1.157 ad 1218: l->l_ru.ru_msgrcv++;
1.69 thorpej 1219: KASSERT(m == so->so_rcv.sb_mb);
1220: SBLASTRECORDCHK(&so->so_rcv, "soreceive 1");
1221: SBLASTMBUFCHK(&so->so_rcv, "soreceive 1");
1.1 cgd 1222: nextrecord = m->m_nextpkt;
1223: if (pr->pr_flags & PR_ADDR) {
1224: #ifdef DIAGNOSTIC
1225: if (m->m_type != MT_SONAME)
1226: panic("receive 1a");
1227: #endif
1.3 andrew 1228: orig_resid = 0;
1.1 cgd 1229: if (flags & MSG_PEEK) {
1230: if (paddr)
1231: *paddr = m_copy(m, 0, m->m_len);
1232: m = m->m_next;
1233: } else {
1234: sbfree(&so->so_rcv, m);
1.67 he 1235: mbuf_removed = 1;
1.144 dyoung 1236: if (paddr != NULL) {
1.1 cgd 1237: *paddr = m;
1238: so->so_rcv.sb_mb = m->m_next;
1.144 dyoung 1239: m->m_next = NULL;
1.1 cgd 1240: m = so->so_rcv.sb_mb;
1241: } else {
1242: MFREE(m, so->so_rcv.sb_mb);
1243: m = so->so_rcv.sb_mb;
1244: }
1.159 ad 1245: sbsync(&so->so_rcv, nextrecord);
1.1 cgd 1246: }
1247: }
1.159 ad 1248:
1249: /*
1250: * Process one or more MT_CONTROL mbufs present before any data mbufs
1251: * in the first mbuf chain on the socket buffer. If MSG_PEEK, we
1252: * just copy the data; if !MSG_PEEK, we call into the protocol to
1253: * perform externalization (or freeing if controlp == NULL).
1254: */
1255: if (__predict_false(m != NULL && m->m_type == MT_CONTROL)) {
1256: struct mbuf *cm = NULL, *cmn;
1257: struct mbuf **cme = &cm;
1258:
1259: do {
1260: if (flags & MSG_PEEK) {
1261: if (controlp != NULL) {
1262: *controlp = m_copy(m, 0, m->m_len);
1263: controlp = &(*controlp)->m_next;
1264: }
1265: m = m->m_next;
1266: } else {
1267: sbfree(&so->so_rcv, m);
1.1 cgd 1268: so->so_rcv.sb_mb = m->m_next;
1.144 dyoung 1269: m->m_next = NULL;
1.159 ad 1270: *cme = m;
1271: cme = &(*cme)->m_next;
1.1 cgd 1272: m = so->so_rcv.sb_mb;
1.159 ad 1273: }
1274: } while (m != NULL && m->m_type == MT_CONTROL);
1275: if ((flags & MSG_PEEK) == 0)
1276: sbsync(&so->so_rcv, nextrecord);
1277: for (; cm != NULL; cm = cmn) {
1278: cmn = cm->m_next;
1279: cm->m_next = NULL;
1280: type = mtod(cm, struct cmsghdr *)->cmsg_type;
1281: if (controlp != NULL) {
1282: if (dom->dom_externalize != NULL &&
1283: type == SCM_RIGHTS) {
1.160 ad 1284: sounlock(so);
1.159 ad 1285: splx(s);
1286: error = (*dom->dom_externalize)(cm, l);
1287: s = splsoftnet();
1.160 ad 1288: solock(so);
1.159 ad 1289: }
1290: *controlp = cm;
1291: while (*controlp != NULL)
1292: controlp = &(*controlp)->m_next;
1.1 cgd 1293: } else {
1.106 itojun 1294: /*
1295: * Dispose of any SCM_RIGHTS message that went
1296: * through the read path rather than recv.
1297: */
1.159 ad 1298: if (dom->dom_dispose != NULL &&
1299: type == SCM_RIGHTS) {
1.160 ad 1300: sounlock(so);
1.159 ad 1301: (*dom->dom_dispose)(cm);
1.160 ad 1302: solock(so);
1.159 ad 1303: }
1304: m_freem(cm);
1.1 cgd 1305: }
1306: }
1.159 ad 1307: if (m != NULL)
1308: nextrecord = so->so_rcv.sb_mb->m_nextpkt;
1309: else
1310: nextrecord = so->so_rcv.sb_mb;
1311: orig_resid = 0;
1.1 cgd 1312: }
1.69 thorpej 1313:
1.159 ad 1314: /* If m is non-NULL, we have some data to read. */
1315: if (__predict_true(m != NULL)) {
1.1 cgd 1316: type = m->m_type;
1317: if (type == MT_OOBDATA)
1318: flags |= MSG_OOB;
1319: }
1.69 thorpej 1320: SBLASTRECORDCHK(&so->so_rcv, "soreceive 2");
1321: SBLASTMBUFCHK(&so->so_rcv, "soreceive 2");
1322:
1.1 cgd 1323: moff = 0;
1324: offset = 0;
1.144 dyoung 1325: while (m != NULL && uio->uio_resid > 0 && error == 0) {
1.1 cgd 1326: if (m->m_type == MT_OOBDATA) {
1327: if (type != MT_OOBDATA)
1328: break;
1329: } else if (type == MT_OOBDATA)
1330: break;
1331: #ifdef DIAGNOSTIC
1332: else if (m->m_type != MT_DATA && m->m_type != MT_HEADER)
1333: panic("receive 3");
1334: #endif
1335: so->so_state &= ~SS_RCVATMARK;
1336: len = uio->uio_resid;
1337: if (so->so_oobmark && len > so->so_oobmark - offset)
1338: len = so->so_oobmark - offset;
1339: if (len > m->m_len - moff)
1340: len = m->m_len - moff;
1341: /*
1342: * If mp is set, just pass back the mbufs.
1343: * Otherwise copy them out via the uio, then free.
1344: * Sockbuf must be consistent here (points to current mbuf,
1345: * it points to next record) when we drop priority;
1346: * we must note any additions to the sockbuf when we
1347: * block interrupts again.
1348: */
1.144 dyoung 1349: if (mp == NULL) {
1.69 thorpej 1350: SBLASTRECORDCHK(&so->so_rcv, "soreceive uiomove");
1351: SBLASTMBUFCHK(&so->so_rcv, "soreceive uiomove");
1.160 ad 1352: sounlock(so);
1.1 cgd 1353: splx(s);
1.134 christos 1354: error = uiomove(mtod(m, char *) + moff, (int)len, uio);
1.20 mycroft 1355: s = splsoftnet();
1.160 ad 1356: solock(so);
1.144 dyoung 1357: if (error != 0) {
1.67 he 1358: /*
1359: * If any part of the record has been removed
1360: * (such as the MT_SONAME mbuf, which will
1361: * happen when PR_ADDR, and thus also
1362: * PR_ATOMIC, is set), then drop the entire
1363: * record to maintain the atomicity of the
1364: * receive operation.
1365: *
1366: * This avoids a later panic("receive 1a")
1367: * when compiled with DIAGNOSTIC.
1368: */
1.146 dyoung 1369: if (m && mbuf_removed && atomic)
1.67 he 1370: (void) sbdroprecord(&so->so_rcv);
1371:
1.57 jdolecek 1372: goto release;
1.67 he 1373: }
1.1 cgd 1374: } else
1375: uio->uio_resid -= len;
1376: if (len == m->m_len - moff) {
1377: if (m->m_flags & M_EOR)
1378: flags |= MSG_EOR;
1379: if (flags & MSG_PEEK) {
1380: m = m->m_next;
1381: moff = 0;
1382: } else {
1383: nextrecord = m->m_nextpkt;
1384: sbfree(&so->so_rcv, m);
1385: if (mp) {
1386: *mp = m;
1387: mp = &m->m_next;
1388: so->so_rcv.sb_mb = m = m->m_next;
1.140 dyoung 1389: *mp = NULL;
1.1 cgd 1390: } else {
1391: MFREE(m, so->so_rcv.sb_mb);
1392: m = so->so_rcv.sb_mb;
1393: }
1.69 thorpej 1394: /*
1395: * If m != NULL, we also know that
1396: * so->so_rcv.sb_mb != NULL.
1397: */
1398: KASSERT(so->so_rcv.sb_mb == m);
1399: if (m) {
1.1 cgd 1400: m->m_nextpkt = nextrecord;
1.69 thorpej 1401: if (nextrecord == NULL)
1402: so->so_rcv.sb_lastrecord = m;
1403: } else {
1404: so->so_rcv.sb_mb = nextrecord;
1.70 thorpej 1405: SB_EMPTY_FIXUP(&so->so_rcv);
1.69 thorpej 1406: }
1407: SBLASTRECORDCHK(&so->so_rcv, "soreceive 3");
1408: SBLASTMBUFCHK(&so->so_rcv, "soreceive 3");
1.1 cgd 1409: }
1.144 dyoung 1410: } else if (flags & MSG_PEEK)
1411: moff += len;
1412: else {
1.160 ad 1413: if (mp != NULL) {
1414: mt = m_copym(m, 0, len, M_NOWAIT);
1415: if (__predict_false(mt == NULL)) {
1416: sounlock(so);
1417: mt = m_copym(m, 0, len, M_WAIT);
1418: solock(so);
1419: }
1420: *mp = mt;
1421: }
1.144 dyoung 1422: m->m_data += len;
1423: m->m_len -= len;
1424: so->so_rcv.sb_cc -= len;
1.1 cgd 1425: }
1426: if (so->so_oobmark) {
1427: if ((flags & MSG_PEEK) == 0) {
1428: so->so_oobmark -= len;
1429: if (so->so_oobmark == 0) {
1430: so->so_state |= SS_RCVATMARK;
1431: break;
1432: }
1.7 cgd 1433: } else {
1.1 cgd 1434: offset += len;
1.7 cgd 1435: if (offset == so->so_oobmark)
1436: break;
1437: }
1.1 cgd 1438: }
1439: if (flags & MSG_EOR)
1440: break;
1441: /*
1442: * If the MSG_WAITALL flag is set (for non-atomic socket),
1443: * we must not quit until "uio->uio_resid == 0" or an error
1444: * termination. If a signal/timeout occurs, return
1445: * with a short count but without error.
1446: * Keep sockbuf locked against other readers.
1447: */
1.144 dyoung 1448: while (flags & MSG_WAITALL && m == NULL && uio->uio_resid > 0 &&
1.3 andrew 1449: !sosendallatonce(so) && !nextrecord) {
1.1 cgd 1450: if (so->so_error || so->so_state & SS_CANTRCVMORE)
1451: break;
1.68 matt 1452: /*
1453: * If we are peeking and the socket receive buffer is
1454: * full, stop since we can't get more data to peek at.
1455: */
1456: if ((flags & MSG_PEEK) && sbspace(&so->so_rcv) <= 0)
1457: break;
1458: /*
1459: * If we've drained the socket buffer, tell the
1460: * protocol in case it needs to do something to
1461: * get it filled again.
1462: */
1463: if ((pr->pr_flags & PR_WANTRCVD) && so->so_pcb)
1464: (*pr->pr_usrreq)(so, PRU_RCVD,
1.140 dyoung 1465: NULL, (struct mbuf *)(long)flags, NULL, l);
1.69 thorpej 1466: SBLASTRECORDCHK(&so->so_rcv, "soreceive sbwait 2");
1467: SBLASTMBUFCHK(&so->so_rcv, "soreceive sbwait 2");
1.1 cgd 1468: error = sbwait(&so->so_rcv);
1.144 dyoung 1469: if (error != 0) {
1.1 cgd 1470: sbunlock(&so->so_rcv);
1.160 ad 1471: sounlock(so);
1.1 cgd 1472: splx(s);
1.144 dyoung 1473: return 0;
1.1 cgd 1474: }
1.21 christos 1475: if ((m = so->so_rcv.sb_mb) != NULL)
1.1 cgd 1476: nextrecord = m->m_nextpkt;
1477: }
1478: }
1.3 andrew 1479:
1.146 dyoung 1480: if (m && atomic) {
1.3 andrew 1481: flags |= MSG_TRUNC;
1482: if ((flags & MSG_PEEK) == 0)
1483: (void) sbdroprecord(&so->so_rcv);
1484: }
1.1 cgd 1485: if ((flags & MSG_PEEK) == 0) {
1.144 dyoung 1486: if (m == NULL) {
1.69 thorpej 1487: /*
1.70 thorpej 1488: * First part is an inline SB_EMPTY_FIXUP(). Second
1.69 thorpej 1489: * part makes sure sb_lastrecord is up-to-date if
1490: * there is still data in the socket buffer.
1491: */
1.1 cgd 1492: so->so_rcv.sb_mb = nextrecord;
1.69 thorpej 1493: if (so->so_rcv.sb_mb == NULL) {
1494: so->so_rcv.sb_mbtail = NULL;
1495: so->so_rcv.sb_lastrecord = NULL;
1496: } else if (nextrecord->m_nextpkt == NULL)
1497: so->so_rcv.sb_lastrecord = nextrecord;
1498: }
1499: SBLASTRECORDCHK(&so->so_rcv, "soreceive 4");
1500: SBLASTMBUFCHK(&so->so_rcv, "soreceive 4");
1.1 cgd 1501: if (pr->pr_flags & PR_WANTRCVD && so->so_pcb)
1.140 dyoung 1502: (*pr->pr_usrreq)(so, PRU_RCVD, NULL,
1503: (struct mbuf *)(long)flags, NULL, l);
1.1 cgd 1504: }
1.3 andrew 1505: if (orig_resid == uio->uio_resid && orig_resid &&
1506: (flags & MSG_EOR) == 0 && (so->so_state & SS_CANTRCVMORE) == 0) {
1507: sbunlock(&so->so_rcv);
1508: goto restart;
1509: }
1.108 perry 1510:
1.144 dyoung 1511: if (flagsp != NULL)
1.1 cgd 1512: *flagsp |= flags;
1.54 lukem 1513: release:
1.1 cgd 1514: sbunlock(&so->so_rcv);
1.160 ad 1515: sounlock(so);
1.1 cgd 1516: splx(s);
1.144 dyoung 1517: return error;
1.1 cgd 1518: }
1519:
1.14 mycroft 1520: int
1.54 lukem 1521: soshutdown(struct socket *so, int how)
1.1 cgd 1522: {
1.99 matt 1523: const struct protosw *pr;
1.160 ad 1524: int error;
1525:
1526: KASSERT(solocked(so));
1.34 kleink 1527:
1.54 lukem 1528: pr = so->so_proto;
1.34 kleink 1529: if (!(how == SHUT_RD || how == SHUT_WR || how == SHUT_RDWR))
1530: return (EINVAL);
1.1 cgd 1531:
1.160 ad 1532: if (how == SHUT_RD || how == SHUT_RDWR) {
1.1 cgd 1533: sorflush(so);
1.160 ad 1534: error = 0;
1535: }
1.34 kleink 1536: if (how == SHUT_WR || how == SHUT_RDWR)
1.160 ad 1537: error = (*pr->pr_usrreq)(so, PRU_SHUTDOWN, NULL,
1.140 dyoung 1538: NULL, NULL, NULL);
1.160 ad 1539:
1540: return error;
1.1 cgd 1541: }
1542:
1.14 mycroft 1543: void
1.54 lukem 1544: sorflush(struct socket *so)
1.1 cgd 1545: {
1.54 lukem 1546: struct sockbuf *sb, asb;
1.99 matt 1547: const struct protosw *pr;
1.160 ad 1548:
1549: KASSERT(solocked(so));
1.1 cgd 1550:
1.54 lukem 1551: sb = &so->so_rcv;
1552: pr = so->so_proto;
1.160 ad 1553: socantrcvmore(so);
1.1 cgd 1554: sb->sb_flags |= SB_NOINTR;
1.160 ad 1555: (void )sblock(sb, M_WAITOK);
1.1 cgd 1556: sbunlock(sb);
1557: asb = *sb;
1.86 wrstuden 1558: /*
1559: * Clear most of the sockbuf structure, but leave some of the
1560: * fields valid.
1561: */
1562: memset(&sb->sb_startzero, 0,
1563: sizeof(*sb) - offsetof(struct sockbuf, sb_startzero));
1.160 ad 1564: if (pr->pr_flags & PR_RIGHTS && pr->pr_domain->dom_dispose) {
1565: sounlock(so);
1.1 cgd 1566: (*pr->pr_domain->dom_dispose)(asb.sb_mb);
1.160 ad 1567: solock(so);
1568: }
1.98 christos 1569: sbrelease(&asb, so);
1.1 cgd 1570: }
1571:
1.142 dyoung 1572: static int
1573: sosetopt1(struct socket *so, int level, int optname, struct mbuf *m)
1.1 cgd 1574: {
1.142 dyoung 1575: int optval, val;
1.130 christos 1576: struct linger *l;
1.141 yamt 1577: struct sockbuf *sb;
1.142 dyoung 1578: struct timeval *tv;
1579:
1580: switch (optname) {
1581:
1582: case SO_LINGER:
1583: if (m == NULL || m->m_len != sizeof(struct linger))
1584: return EINVAL;
1585: l = mtod(m, struct linger *);
1586: if (l->l_linger < 0 || l->l_linger > USHRT_MAX ||
1587: l->l_linger > (INT_MAX / hz))
1588: return EDOM;
1589: so->so_linger = l->l_linger;
1590: if (l->l_onoff)
1591: so->so_options |= SO_LINGER;
1592: else
1593: so->so_options &= ~SO_LINGER;
1594: break;
1.1 cgd 1595:
1.142 dyoung 1596: case SO_DEBUG:
1597: case SO_KEEPALIVE:
1598: case SO_DONTROUTE:
1599: case SO_USELOOPBACK:
1600: case SO_BROADCAST:
1601: case SO_REUSEADDR:
1602: case SO_REUSEPORT:
1603: case SO_OOBINLINE:
1604: case SO_TIMESTAMP:
1605: if (m == NULL || m->m_len < sizeof(int))
1606: return EINVAL;
1607: if (*mtod(m, int *))
1608: so->so_options |= optname;
1609: else
1610: so->so_options &= ~optname;
1611: break;
1612:
1613: case SO_SNDBUF:
1614: case SO_RCVBUF:
1615: case SO_SNDLOWAT:
1616: case SO_RCVLOWAT:
1617: if (m == NULL || m->m_len < sizeof(int))
1618: return EINVAL;
1.1 cgd 1619:
1.142 dyoung 1620: /*
1621: * Values < 1 make no sense for any of these
1622: * options, so disallow them.
1623: */
1624: optval = *mtod(m, int *);
1625: if (optval < 1)
1626: return EINVAL;
1.1 cgd 1627:
1.142 dyoung 1628: switch (optname) {
1.1 cgd 1629:
1630: case SO_SNDBUF:
1631: case SO_RCVBUF:
1.142 dyoung 1632: sb = (optname == SO_SNDBUF) ?
1633: &so->so_snd : &so->so_rcv;
1634: if (sbreserve(sb, (u_long)optval, so) == 0)
1635: return ENOBUFS;
1636: sb->sb_flags &= ~SB_AUTOSIZE;
1637: break;
1638:
1639: /*
1640: * Make sure the low-water is never greater than
1641: * the high-water.
1642: */
1.1 cgd 1643: case SO_SNDLOWAT:
1.142 dyoung 1644: so->so_snd.sb_lowat =
1645: (optval > so->so_snd.sb_hiwat) ?
1646: so->so_snd.sb_hiwat : optval;
1647: break;
1.1 cgd 1648: case SO_RCVLOWAT:
1.142 dyoung 1649: so->so_rcv.sb_lowat =
1650: (optval > so->so_rcv.sb_hiwat) ?
1651: so->so_rcv.sb_hiwat : optval;
1652: break;
1653: }
1654: break;
1.28 thorpej 1655:
1.142 dyoung 1656: case SO_SNDTIMEO:
1657: case SO_RCVTIMEO:
1658: if (m == NULL || m->m_len < sizeof(*tv))
1659: return EINVAL;
1660: tv = mtod(m, struct timeval *);
1661: if (tv->tv_sec > (INT_MAX - tv->tv_usec / tick) / hz)
1662: return EDOM;
1663: val = tv->tv_sec * hz + tv->tv_usec / tick;
1664: if (val == 0 && tv->tv_usec != 0)
1665: val = 1;
1.28 thorpej 1666:
1.142 dyoung 1667: switch (optname) {
1.28 thorpej 1668:
1.142 dyoung 1669: case SO_SNDTIMEO:
1670: so->so_snd.sb_timeo = val;
1.1 cgd 1671: break;
1672: case SO_RCVTIMEO:
1.142 dyoung 1673: so->so_rcv.sb_timeo = val;
1674: break;
1675: }
1676: break;
1.1 cgd 1677:
1.142 dyoung 1678: default:
1679: return ENOPROTOOPT;
1680: }
1681: return 0;
1682: }
1.1 cgd 1683:
1.142 dyoung 1684: int
1685: sosetopt(struct socket *so, int level, int optname, struct mbuf *m)
1686: {
1687: int error, prerr;
1.1 cgd 1688:
1.160 ad 1689: solock(so);
1.142 dyoung 1690: if (level == SOL_SOCKET)
1691: error = sosetopt1(so, level, optname, m);
1692: else
1693: error = ENOPROTOOPT;
1.1 cgd 1694:
1.142 dyoung 1695: if ((error == 0 || error == ENOPROTOOPT) &&
1696: so->so_proto != NULL && so->so_proto->pr_ctloutput != NULL) {
1697: /* give the protocol stack a shot */
1698: prerr = (*so->so_proto->pr_ctloutput)(PRCO_SETOPT, so, level,
1699: optname, &m);
1700: if (prerr == 0)
1701: error = 0;
1702: else if (prerr != ENOPROTOOPT)
1703: error = prerr;
1704: } else if (m != NULL)
1705: (void)m_free(m);
1.160 ad 1706: sounlock(so);
1.142 dyoung 1707: return error;
1.1 cgd 1708: }
1709:
1.14 mycroft 1710: int
1.54 lukem 1711: sogetopt(struct socket *so, int level, int optname, struct mbuf **mp)
1.1 cgd 1712: {
1.54 lukem 1713: struct mbuf *m;
1.160 ad 1714: int error;
1.1 cgd 1715:
1.160 ad 1716: solock(so);
1.1 cgd 1717: if (level != SOL_SOCKET) {
1718: if (so->so_proto && so->so_proto->pr_ctloutput) {
1.160 ad 1719: error = ((*so->so_proto->pr_ctloutput)
1.1 cgd 1720: (PRCO_GETOPT, so, level, optname, mp));
1721: } else
1.160 ad 1722: error = (ENOPROTOOPT);
1.1 cgd 1723: } else {
1724: m = m_get(M_WAIT, MT_SOOPTS);
1.36 perry 1725: m->m_len = sizeof(int);
1.1 cgd 1726:
1727: switch (optname) {
1728:
1729: case SO_LINGER:
1.36 perry 1730: m->m_len = sizeof(struct linger);
1.1 cgd 1731: mtod(m, struct linger *)->l_onoff =
1.131 christos 1732: (so->so_options & SO_LINGER) ? 1 : 0;
1.1 cgd 1733: mtod(m, struct linger *)->l_linger = so->so_linger;
1734: break;
1735:
1736: case SO_USELOOPBACK:
1737: case SO_DONTROUTE:
1738: case SO_DEBUG:
1739: case SO_KEEPALIVE:
1740: case SO_REUSEADDR:
1.15 mycroft 1741: case SO_REUSEPORT:
1.1 cgd 1742: case SO_BROADCAST:
1743: case SO_OOBINLINE:
1.26 thorpej 1744: case SO_TIMESTAMP:
1.131 christos 1745: *mtod(m, int *) = (so->so_options & optname) ? 1 : 0;
1.1 cgd 1746: break;
1747:
1748: case SO_TYPE:
1749: *mtod(m, int *) = so->so_type;
1750: break;
1751:
1752: case SO_ERROR:
1753: *mtod(m, int *) = so->so_error;
1754: so->so_error = 0;
1755: break;
1756:
1757: case SO_SNDBUF:
1758: *mtod(m, int *) = so->so_snd.sb_hiwat;
1759: break;
1760:
1761: case SO_RCVBUF:
1762: *mtod(m, int *) = so->so_rcv.sb_hiwat;
1763: break;
1764:
1765: case SO_SNDLOWAT:
1766: *mtod(m, int *) = so->so_snd.sb_lowat;
1767: break;
1768:
1769: case SO_RCVLOWAT:
1770: *mtod(m, int *) = so->so_rcv.sb_lowat;
1771: break;
1772:
1773: case SO_SNDTIMEO:
1774: case SO_RCVTIMEO:
1775: {
1776: int val = (optname == SO_SNDTIMEO ?
1777: so->so_snd.sb_timeo : so->so_rcv.sb_timeo);
1778:
1779: m->m_len = sizeof(struct timeval);
1780: mtod(m, struct timeval *)->tv_sec = val / hz;
1781: mtod(m, struct timeval *)->tv_usec =
1.27 kleink 1782: (val % hz) * tick;
1.1 cgd 1783: break;
1784: }
1785:
1.107 darrenr 1786: case SO_OVERFLOWED:
1787: *mtod(m, int *) = so->so_rcv.sb_overflowed;
1788: break;
1789:
1.1 cgd 1790: default:
1.160 ad 1791: sounlock(so);
1.1 cgd 1792: (void)m_free(m);
1793: return (ENOPROTOOPT);
1794: }
1795: *mp = m;
1.160 ad 1796: error = 0;
1.1 cgd 1797: }
1.160 ad 1798:
1799: sounlock(so);
1800: return (error);
1.1 cgd 1801: }
1802:
1.14 mycroft 1803: void
1.54 lukem 1804: sohasoutofband(struct socket *so)
1.1 cgd 1805: {
1.153 rmind 1806:
1.90 christos 1807: fownsignal(so->so_pgid, SIGURG, POLL_PRI, POLLPRI|POLLRDBAND, so);
1.153 rmind 1808: selnotify(&so->so_rcv.sb_sel, POLLPRI | POLLRDBAND, 0);
1.1 cgd 1809: }
1.72 jdolecek 1810:
1811: static void
1812: filt_sordetach(struct knote *kn)
1813: {
1814: struct socket *so;
1815:
1.155 ad 1816: so = ((file_t *)kn->kn_obj)->f_data;
1.160 ad 1817: solock(so);
1.73 christos 1818: SLIST_REMOVE(&so->so_rcv.sb_sel.sel_klist, kn, knote, kn_selnext);
1819: if (SLIST_EMPTY(&so->so_rcv.sb_sel.sel_klist))
1.72 jdolecek 1820: so->so_rcv.sb_flags &= ~SB_KNOTE;
1.160 ad 1821: sounlock(so);
1.72 jdolecek 1822: }
1823:
1824: /*ARGSUSED*/
1825: static int
1.129 yamt 1826: filt_soread(struct knote *kn, long hint)
1.72 jdolecek 1827: {
1828: struct socket *so;
1.160 ad 1829: int rv;
1.72 jdolecek 1830:
1.155 ad 1831: so = ((file_t *)kn->kn_obj)->f_data;
1.160 ad 1832: if (hint != NOTE_SUBMIT)
1833: solock(so);
1.72 jdolecek 1834: kn->kn_data = so->so_rcv.sb_cc;
1835: if (so->so_state & SS_CANTRCVMORE) {
1.108 perry 1836: kn->kn_flags |= EV_EOF;
1.72 jdolecek 1837: kn->kn_fflags = so->so_error;
1.160 ad 1838: rv = 1;
1839: } else if (so->so_error) /* temporary udp error */
1840: rv = 1;
1841: else if (kn->kn_sfflags & NOTE_LOWAT)
1842: rv = (kn->kn_data >= kn->kn_sdata);
1843: else
1844: rv = (kn->kn_data >= so->so_rcv.sb_lowat);
1845: if (hint != NOTE_SUBMIT)
1846: sounlock(so);
1847: return rv;
1.72 jdolecek 1848: }
1849:
1850: static void
1851: filt_sowdetach(struct knote *kn)
1852: {
1853: struct socket *so;
1854:
1.155 ad 1855: so = ((file_t *)kn->kn_obj)->f_data;
1.160 ad 1856: solock(so);
1.73 christos 1857: SLIST_REMOVE(&so->so_snd.sb_sel.sel_klist, kn, knote, kn_selnext);
1858: if (SLIST_EMPTY(&so->so_snd.sb_sel.sel_klist))
1.72 jdolecek 1859: so->so_snd.sb_flags &= ~SB_KNOTE;
1.160 ad 1860: sounlock(so);
1.72 jdolecek 1861: }
1862:
1863: /*ARGSUSED*/
1864: static int
1.129 yamt 1865: filt_sowrite(struct knote *kn, long hint)
1.72 jdolecek 1866: {
1867: struct socket *so;
1.160 ad 1868: int rv;
1.72 jdolecek 1869:
1.155 ad 1870: so = ((file_t *)kn->kn_obj)->f_data;
1.160 ad 1871: if (hint != NOTE_SUBMIT)
1872: solock(so);
1.72 jdolecek 1873: kn->kn_data = sbspace(&so->so_snd);
1874: if (so->so_state & SS_CANTSENDMORE) {
1.108 perry 1875: kn->kn_flags |= EV_EOF;
1.72 jdolecek 1876: kn->kn_fflags = so->so_error;
1.160 ad 1877: rv = 1;
1878: } else if (so->so_error) /* temporary udp error */
1879: rv = 1;
1880: else if (((so->so_state & SS_ISCONNECTED) == 0) &&
1.72 jdolecek 1881: (so->so_proto->pr_flags & PR_CONNREQUIRED))
1.160 ad 1882: rv = 0;
1883: else if (kn->kn_sfflags & NOTE_LOWAT)
1884: rv = (kn->kn_data >= kn->kn_sdata);
1885: else
1886: rv = (kn->kn_data >= so->so_snd.sb_lowat);
1887: if (hint != NOTE_SUBMIT)
1888: sounlock(so);
1889: return rv;
1.72 jdolecek 1890: }
1891:
1892: /*ARGSUSED*/
1893: static int
1.129 yamt 1894: filt_solisten(struct knote *kn, long hint)
1.72 jdolecek 1895: {
1896: struct socket *so;
1.160 ad 1897: int rv;
1.72 jdolecek 1898:
1.155 ad 1899: so = ((file_t *)kn->kn_obj)->f_data;
1.72 jdolecek 1900:
1901: /*
1902: * Set kn_data to number of incoming connections, not
1903: * counting partial (incomplete) connections.
1.108 perry 1904: */
1.160 ad 1905: if (hint != NOTE_SUBMIT)
1906: solock(so);
1.72 jdolecek 1907: kn->kn_data = so->so_qlen;
1.160 ad 1908: rv = (kn->kn_data > 0);
1909: if (hint != NOTE_SUBMIT)
1910: sounlock(so);
1911: return rv;
1.72 jdolecek 1912: }
1913:
1914: static const struct filterops solisten_filtops =
1915: { 1, NULL, filt_sordetach, filt_solisten };
1916: static const struct filterops soread_filtops =
1917: { 1, NULL, filt_sordetach, filt_soread };
1918: static const struct filterops sowrite_filtops =
1919: { 1, NULL, filt_sowdetach, filt_sowrite };
1920:
1921: int
1.129 yamt 1922: soo_kqfilter(struct file *fp, struct knote *kn)
1.72 jdolecek 1923: {
1924: struct socket *so;
1925: struct sockbuf *sb;
1926:
1.155 ad 1927: so = ((file_t *)kn->kn_obj)->f_data;
1.160 ad 1928: solock(so);
1.72 jdolecek 1929: switch (kn->kn_filter) {
1930: case EVFILT_READ:
1931: if (so->so_options & SO_ACCEPTCONN)
1932: kn->kn_fop = &solisten_filtops;
1933: else
1934: kn->kn_fop = &soread_filtops;
1935: sb = &so->so_rcv;
1936: break;
1937: case EVFILT_WRITE:
1938: kn->kn_fop = &sowrite_filtops;
1939: sb = &so->so_snd;
1940: break;
1941: default:
1.160 ad 1942: sounlock(so);
1.149 pooka 1943: return (EINVAL);
1.72 jdolecek 1944: }
1.73 christos 1945: SLIST_INSERT_HEAD(&sb->sb_sel.sel_klist, kn, kn_selnext);
1.72 jdolecek 1946: sb->sb_flags |= SB_KNOTE;
1.160 ad 1947: sounlock(so);
1.72 jdolecek 1948: return (0);
1949: }
1950:
1.154 ad 1951: static int
1952: sodopoll(struct socket *so, int events)
1953: {
1954: int revents;
1955:
1956: revents = 0;
1957:
1958: if (events & (POLLIN | POLLRDNORM))
1959: if (soreadable(so))
1960: revents |= events & (POLLIN | POLLRDNORM);
1961:
1962: if (events & (POLLOUT | POLLWRNORM))
1963: if (sowritable(so))
1964: revents |= events & (POLLOUT | POLLWRNORM);
1965:
1966: if (events & (POLLPRI | POLLRDBAND))
1967: if (so->so_oobmark || (so->so_state & SS_RCVATMARK))
1968: revents |= events & (POLLPRI | POLLRDBAND);
1969:
1970: return revents;
1971: }
1972:
1973: int
1974: sopoll(struct socket *so, int events)
1975: {
1976: int revents = 0;
1977:
1.160 ad 1978: #ifndef DIAGNOSTIC
1979: /*
1980: * Do a quick, unlocked check in expectation that the socket
1981: * will be ready for I/O. Don't do this check if DIAGNOSTIC,
1982: * as the solocked() assertions will fail.
1983: */
1.154 ad 1984: if ((revents = sodopoll(so, events)) != 0)
1985: return revents;
1.160 ad 1986: #endif
1.154 ad 1987:
1.160 ad 1988: solock(so);
1.154 ad 1989: if ((revents = sodopoll(so, events)) == 0) {
1990: if (events & (POLLIN | POLLPRI | POLLRDNORM | POLLRDBAND)) {
1991: selrecord(curlwp, &so->so_rcv.sb_sel);
1.160 ad 1992: so->so_rcv.sb_flags |= SB_NOTIFY;
1.154 ad 1993: }
1994:
1995: if (events & (POLLOUT | POLLWRNORM)) {
1996: selrecord(curlwp, &so->so_snd.sb_sel);
1.160 ad 1997: so->so_snd.sb_flags |= SB_NOTIFY;
1.154 ad 1998: }
1999: }
1.160 ad 2000: sounlock(so);
1.154 ad 2001:
2002: return revents;
2003: }
2004:
2005:
1.94 yamt 2006: #include <sys/sysctl.h>
2007:
2008: static int sysctl_kern_somaxkva(SYSCTLFN_PROTO);
2009:
2010: /*
2011: * sysctl helper routine for kern.somaxkva. ensures that the given
2012: * value is not too small.
2013: * (XXX should we maybe make sure it's not too large as well?)
2014: */
2015: static int
2016: sysctl_kern_somaxkva(SYSCTLFN_ARGS)
2017: {
2018: int error, new_somaxkva;
2019: struct sysctlnode node;
2020:
2021: new_somaxkva = somaxkva;
2022: node = *rnode;
2023: node.sysctl_data = &new_somaxkva;
2024: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2025: if (error || newp == NULL)
2026: return (error);
2027:
2028: if (new_somaxkva < (16 * 1024 * 1024)) /* sanity */
2029: return (EINVAL);
2030:
1.136 ad 2031: mutex_enter(&so_pendfree_lock);
1.94 yamt 2032: somaxkva = new_somaxkva;
1.136 ad 2033: cv_broadcast(&socurkva_cv);
2034: mutex_exit(&so_pendfree_lock);
1.94 yamt 2035:
2036: return (error);
2037: }
2038:
2039: SYSCTL_SETUP(sysctl_kern_somaxkva_setup, "sysctl kern.somaxkva setup")
2040: {
2041:
1.97 atatat 2042: sysctl_createv(clog, 0, NULL, NULL,
2043: CTLFLAG_PERMANENT,
2044: CTLTYPE_NODE, "kern", NULL,
2045: NULL, 0, NULL, 0,
2046: CTL_KERN, CTL_EOL);
2047:
2048: sysctl_createv(clog, 0, NULL, NULL,
2049: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.103 atatat 2050: CTLTYPE_INT, "somaxkva",
2051: SYSCTL_DESCR("Maximum amount of kernel memory to be "
2052: "used for socket buffers"),
1.94 yamt 2053: sysctl_kern_somaxkva, 0, NULL, 0,
2054: CTL_KERN, KERN_SOMAXKVA, CTL_EOL);
2055: }
CVSweb <webmaster@jp.NetBSD.org>
![[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