Annotation of src/sys/kern/uipc_socket.c, Revision 1.166
1.166 ! ad 1: /* $NetBSD: uipc_socket.c,v 1.165 2008/05/24 18:43:02 christos 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>
1.166 ! ad 66: __KERNEL_RCSID(0, "$NetBSD: uipc_socket.c,v 1.165 2008/05/24 18:43:02 christos 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"
1.1 cgd 72:
1.9 mycroft 73: #include <sys/param.h>
74: #include <sys/systm.h>
75: #include <sys/proc.h>
76: #include <sys/file.h>
1.142 dyoung 77: #include <sys/filedesc.h>
1.9 mycroft 78: #include <sys/malloc.h>
79: #include <sys/mbuf.h>
80: #include <sys/domain.h>
81: #include <sys/kernel.h>
82: #include <sys/protosw.h>
83: #include <sys/socket.h>
84: #include <sys/socketvar.h>
1.21 christos 85: #include <sys/signalvar.h>
1.9 mycroft 86: #include <sys/resourcevar.h>
1.72 jdolecek 87: #include <sys/event.h>
1.89 christos 88: #include <sys/poll.h>
1.118 elad 89: #include <sys/kauth.h>
1.136 ad 90: #include <sys/mutex.h>
91: #include <sys/condvar.h>
1.37 thorpej 92:
1.64 thorpej 93: #include <uvm/uvm.h>
94:
1.77 thorpej 95: MALLOC_DEFINE(M_SOOPTS, "soopts", "socket options");
96: MALLOC_DEFINE(M_SONAME, "soname", "socket name");
1.37 thorpej 97:
1.142 dyoung 98: extern const struct fileops socketops;
99:
1.54 lukem 100: extern int somaxconn; /* patchable (XXX sysctl) */
101: int somaxconn = SOMAXCONN;
1.160 ad 102: kmutex_t *softnet_lock;
1.49 jonathan 103:
1.64 thorpej 104: #ifdef SOSEND_COUNTERS
105: #include <sys/device.h>
106:
1.113 thorpej 107: static struct evcnt sosend_loan_big = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64 thorpej 108: NULL, "sosend", "loan big");
1.113 thorpej 109: static struct evcnt sosend_copy_big = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64 thorpej 110: NULL, "sosend", "copy big");
1.113 thorpej 111: static struct evcnt sosend_copy_small = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64 thorpej 112: NULL, "sosend", "copy small");
1.113 thorpej 113: static struct evcnt sosend_kvalimit = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64 thorpej 114: NULL, "sosend", "kva limit");
115:
116: #define SOSEND_COUNTER_INCR(ev) (ev)->ev_count++
117:
1.101 matt 118: EVCNT_ATTACH_STATIC(sosend_loan_big);
119: EVCNT_ATTACH_STATIC(sosend_copy_big);
120: EVCNT_ATTACH_STATIC(sosend_copy_small);
121: EVCNT_ATTACH_STATIC(sosend_kvalimit);
1.64 thorpej 122: #else
123:
124: #define SOSEND_COUNTER_INCR(ev) /* nothing */
125:
126: #endif /* SOSEND_COUNTERS */
127:
1.119 yamt 128: static struct callback_entry sokva_reclaimerentry;
1.1 cgd 129:
1.71 thorpej 130: #ifdef SOSEND_NO_LOAN
1.121 yamt 131: int sock_loan_thresh = -1;
1.71 thorpej 132: #else
1.121 yamt 133: int sock_loan_thresh = 4096;
1.65 thorpej 134: #endif
1.64 thorpej 135:
1.136 ad 136: static kmutex_t so_pendfree_lock;
1.113 thorpej 137: static struct mbuf *so_pendfree;
1.64 thorpej 138:
1.84 ragge 139: #ifndef SOMAXKVA
140: #define SOMAXKVA (16 * 1024 * 1024)
141: #endif
142: int somaxkva = SOMAXKVA;
1.113 thorpej 143: static int socurkva;
1.136 ad 144: static kcondvar_t socurkva_cv;
1.64 thorpej 145:
146: #define SOCK_LOAN_CHUNK 65536
147:
1.117 yamt 148: static size_t sodopendfree(void);
149: static size_t sodopendfreel(void);
1.93 yamt 150:
1.113 thorpej 151: static vsize_t
1.129 yamt 152: sokvareserve(struct socket *so, vsize_t len)
1.80 yamt 153: {
1.98 christos 154: int error;
1.80 yamt 155:
1.136 ad 156: mutex_enter(&so_pendfree_lock);
1.80 yamt 157: while (socurkva + len > somaxkva) {
1.93 yamt 158: size_t freed;
159:
160: /*
161: * try to do pendfree.
162: */
163:
1.117 yamt 164: freed = sodopendfreel();
1.93 yamt 165:
166: /*
167: * if some kva was freed, try again.
168: */
169:
170: if (freed)
1.80 yamt 171: continue;
1.93 yamt 172:
1.80 yamt 173: SOSEND_COUNTER_INCR(&sosend_kvalimit);
1.136 ad 174: error = cv_wait_sig(&socurkva_cv, &so_pendfree_lock);
1.98 christos 175: if (error) {
176: len = 0;
177: break;
178: }
1.80 yamt 179: }
1.93 yamt 180: socurkva += len;
1.136 ad 181: mutex_exit(&so_pendfree_lock);
1.98 christos 182: return len;
1.95 yamt 183: }
184:
1.113 thorpej 185: static void
1.95 yamt 186: sokvaunreserve(vsize_t len)
187: {
188:
1.136 ad 189: mutex_enter(&so_pendfree_lock);
1.95 yamt 190: socurkva -= len;
1.136 ad 191: cv_broadcast(&socurkva_cv);
192: mutex_exit(&so_pendfree_lock);
1.95 yamt 193: }
194:
195: /*
196: * sokvaalloc: allocate kva for loan.
197: */
198:
199: vaddr_t
200: sokvaalloc(vsize_t len, struct socket *so)
201: {
202: vaddr_t lva;
203:
204: /*
205: * reserve kva.
206: */
207:
1.98 christos 208: if (sokvareserve(so, len) == 0)
209: return 0;
1.93 yamt 210:
211: /*
212: * allocate kva.
213: */
1.80 yamt 214:
1.109 yamt 215: lva = uvm_km_alloc(kernel_map, len, 0, UVM_KMF_VAONLY | UVM_KMF_WAITVA);
1.95 yamt 216: if (lva == 0) {
217: sokvaunreserve(len);
1.80 yamt 218: return (0);
1.95 yamt 219: }
1.80 yamt 220:
221: return lva;
222: }
223:
1.93 yamt 224: /*
225: * sokvafree: free kva for loan.
226: */
227:
1.80 yamt 228: void
229: sokvafree(vaddr_t sva, vsize_t len)
230: {
1.93 yamt 231:
232: /*
233: * free kva.
234: */
1.80 yamt 235:
1.109 yamt 236: uvm_km_free(kernel_map, sva, len, UVM_KMF_VAONLY);
1.93 yamt 237:
238: /*
239: * unreserve kva.
240: */
241:
1.95 yamt 242: sokvaunreserve(len);
1.80 yamt 243: }
244:
1.64 thorpej 245: static void
1.134 christos 246: sodoloanfree(struct vm_page **pgs, void *buf, size_t size)
1.64 thorpej 247: {
1.156 yamt 248: vaddr_t sva, eva;
1.64 thorpej 249: vsize_t len;
1.156 yamt 250: int npgs;
251:
252: KASSERT(pgs != NULL);
1.64 thorpej 253:
254: eva = round_page((vaddr_t) buf + size);
255: sva = trunc_page((vaddr_t) buf);
256: len = eva - sva;
257: npgs = len >> PAGE_SHIFT;
258:
259: pmap_kremove(sva, len);
260: pmap_update(pmap_kernel());
261: uvm_unloan(pgs, npgs, UVM_LOAN_TOPAGE);
1.80 yamt 262: sokvafree(sva, len);
1.64 thorpej 263: }
264:
265: static size_t
1.152 matt 266: sodopendfree(void)
1.64 thorpej 267: {
1.93 yamt 268: size_t rv;
1.64 thorpej 269:
1.160 ad 270: if (__predict_true(so_pendfree == NULL))
271: return 0;
272:
1.136 ad 273: mutex_enter(&so_pendfree_lock);
1.117 yamt 274: rv = sodopendfreel();
1.136 ad 275: mutex_exit(&so_pendfree_lock);
1.93 yamt 276:
277: return rv;
278: }
279:
280: /*
281: * sodopendfreel: free mbufs on "pendfree" list.
1.136 ad 282: * unlock and relock so_pendfree_lock when freeing mbufs.
1.93 yamt 283: *
1.136 ad 284: * => called with so_pendfree_lock held.
1.93 yamt 285: */
286:
287: static size_t
1.152 matt 288: sodopendfreel(void)
1.93 yamt 289: {
1.137 ad 290: struct mbuf *m, *next;
1.93 yamt 291: size_t rv = 0;
292:
1.136 ad 293: KASSERT(mutex_owned(&so_pendfree_lock));
1.64 thorpej 294:
1.137 ad 295: while (so_pendfree != NULL) {
1.64 thorpej 296: m = so_pendfree;
1.93 yamt 297: so_pendfree = NULL;
1.136 ad 298: mutex_exit(&so_pendfree_lock);
1.93 yamt 299:
300: for (; m != NULL; m = next) {
301: next = m->m_next;
1.156 yamt 302: KASSERT((~m->m_flags & (M_EXT|M_EXT_PAGES)) == 0);
303: KASSERT(m->m_ext.ext_refcnt == 0);
1.93 yamt 304:
305: rv += m->m_ext.ext_size;
1.156 yamt 306: sodoloanfree(m->m_ext.ext_pgs, m->m_ext.ext_buf,
1.93 yamt 307: m->m_ext.ext_size);
1.145 ad 308: pool_cache_put(mb_cache, m);
1.93 yamt 309: }
1.64 thorpej 310:
1.136 ad 311: mutex_enter(&so_pendfree_lock);
1.64 thorpej 312: }
313:
314: return (rv);
315: }
316:
1.80 yamt 317: void
1.134 christos 318: soloanfree(struct mbuf *m, void *buf, size_t size, void *arg)
1.64 thorpej 319: {
320:
1.156 yamt 321: KASSERT(m != NULL);
1.64 thorpej 322:
1.93 yamt 323: /*
324: * postpone freeing mbuf.
325: *
326: * we can't do it in interrupt context
327: * because we need to put kva back to kernel_map.
328: */
329:
1.136 ad 330: mutex_enter(&so_pendfree_lock);
1.92 yamt 331: m->m_next = so_pendfree;
332: so_pendfree = m;
1.136 ad 333: cv_broadcast(&socurkva_cv);
334: mutex_exit(&so_pendfree_lock);
1.64 thorpej 335: }
336:
337: static long
338: sosend_loan(struct socket *so, struct uio *uio, struct mbuf *m, long space)
339: {
340: struct iovec *iov = uio->uio_iov;
341: vaddr_t sva, eva;
342: vsize_t len;
1.156 yamt 343: vaddr_t lva;
344: int npgs, error;
345: vaddr_t va;
346: int i;
1.64 thorpej 347:
1.116 yamt 348: if (VMSPACE_IS_KERNEL_P(uio->uio_vmspace))
1.64 thorpej 349: return (0);
350:
351: if (iov->iov_len < (size_t) space)
352: space = iov->iov_len;
353: if (space > SOCK_LOAN_CHUNK)
354: space = SOCK_LOAN_CHUNK;
355:
356: eva = round_page((vaddr_t) iov->iov_base + space);
357: sva = trunc_page((vaddr_t) iov->iov_base);
358: len = eva - sva;
359: npgs = len >> PAGE_SHIFT;
360:
1.79 thorpej 361: KASSERT(npgs <= M_EXT_MAXPAGES);
362:
1.80 yamt 363: lva = sokvaalloc(len, so);
1.64 thorpej 364: if (lva == 0)
1.80 yamt 365: return 0;
1.64 thorpej 366:
1.116 yamt 367: error = uvm_loan(&uio->uio_vmspace->vm_map, sva, len,
1.79 thorpej 368: m->m_ext.ext_pgs, UVM_LOAN_TOPAGE);
1.64 thorpej 369: if (error) {
1.80 yamt 370: sokvafree(lva, len);
1.64 thorpej 371: return (0);
372: }
373:
374: for (i = 0, va = lva; i < npgs; i++, va += PAGE_SIZE)
1.79 thorpej 375: pmap_kenter_pa(va, VM_PAGE_TO_PHYS(m->m_ext.ext_pgs[i]),
376: VM_PROT_READ);
1.64 thorpej 377: pmap_update(pmap_kernel());
378:
379: lva += (vaddr_t) iov->iov_base & PAGE_MASK;
380:
1.134 christos 381: MEXTADD(m, (void *) lva, space, M_MBUF, soloanfree, so);
1.79 thorpej 382: m->m_flags |= M_EXT_PAGES | M_EXT_ROMAP;
1.64 thorpej 383:
384: uio->uio_resid -= space;
385: /* uio_offset not updated, not set/used for write(2) */
1.134 christos 386: uio->uio_iov->iov_base = (char *)uio->uio_iov->iov_base + space;
1.64 thorpej 387: uio->uio_iov->iov_len -= space;
388: if (uio->uio_iov->iov_len == 0) {
389: uio->uio_iov++;
390: uio->uio_iovcnt--;
391: }
392:
393: return (space);
394: }
395:
1.119 yamt 396: static int
1.129 yamt 397: sokva_reclaim_callback(struct callback_entry *ce, void *obj, void *arg)
1.119 yamt 398: {
399:
400: KASSERT(ce == &sokva_reclaimerentry);
401: KASSERT(obj == NULL);
402:
403: sodopendfree();
404: if (!vm_map_starved_p(kernel_map)) {
405: return CALLBACK_CHAIN_ABORT;
406: }
407: return CALLBACK_CHAIN_CONTINUE;
408: }
409:
1.142 dyoung 410: struct mbuf *
1.147 dyoung 411: getsombuf(struct socket *so, int type)
1.142 dyoung 412: {
413: struct mbuf *m;
414:
1.147 dyoung 415: m = m_get(M_WAIT, type);
1.142 dyoung 416: MCLAIM(m, so->so_mowner);
417: return m;
418: }
419:
420: struct mbuf *
421: m_intopt(struct socket *so, int val)
422: {
423: struct mbuf *m;
424:
1.147 dyoung 425: m = getsombuf(so, MT_SOOPTS);
1.142 dyoung 426: m->m_len = sizeof(int);
427: *mtod(m, int *) = val;
428: return m;
429: }
430:
1.119 yamt 431: void
432: soinit(void)
433: {
434:
1.148 ad 435: mutex_init(&so_pendfree_lock, MUTEX_DEFAULT, IPL_VM);
1.160 ad 436: softnet_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NONE);
1.136 ad 437: cv_init(&socurkva_cv, "sokva");
1.166 ! ad 438: soinit2();
1.136 ad 439:
1.119 yamt 440: /* Set the initial adjusted socket buffer size. */
441: if (sb_max_set(sb_max))
442: panic("bad initial sb_max value: %lu", sb_max);
443:
444: callback_register(&vm_map_to_kernel(kernel_map)->vmk_reclaim_callback,
445: &sokva_reclaimerentry, NULL, sokva_reclaim_callback);
446: }
447:
1.1 cgd 448: /*
449: * Socket operation routines.
450: * These routines are called by the routines in
451: * sys_socket.c or from a system process, and
452: * implement the semantics of socket operations by
453: * switching out to the protocol specific routines.
454: */
455: /*ARGSUSED*/
1.3 andrew 456: int
1.160 ad 457: socreate(int dom, struct socket **aso, int type, int proto, struct lwp *l,
458: struct socket *lockso)
1.1 cgd 459: {
1.99 matt 460: const struct protosw *prp;
1.54 lukem 461: struct socket *so;
1.115 yamt 462: uid_t uid;
1.160 ad 463: int error;
464: kmutex_t *lock;
1.1 cgd 465:
1.132 elad 466: error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET,
467: KAUTH_REQ_NETWORK_SOCKET_OPEN, KAUTH_ARG(dom), KAUTH_ARG(type),
468: KAUTH_ARG(proto));
1.140 dyoung 469: if (error != 0)
470: return error;
1.127 elad 471:
1.1 cgd 472: if (proto)
473: prp = pffindproto(dom, proto, type);
474: else
475: prp = pffindtype(dom, type);
1.140 dyoung 476: if (prp == NULL) {
1.120 ginsbach 477: /* no support for domain */
478: if (pffinddomain(dom) == 0)
1.140 dyoung 479: return EAFNOSUPPORT;
1.120 ginsbach 480: /* no support for socket type */
481: if (proto == 0 && type != 0)
1.140 dyoung 482: return EPROTOTYPE;
483: return EPROTONOSUPPORT;
1.120 ginsbach 484: }
1.140 dyoung 485: if (prp->pr_usrreq == NULL)
486: return EPROTONOSUPPORT;
1.1 cgd 487: if (prp->pr_type != type)
1.140 dyoung 488: return EPROTOTYPE;
1.160 ad 489:
490: so = soget(true);
1.1 cgd 491: so->so_type = type;
492: so->so_proto = prp;
1.33 matt 493: so->so_send = sosend;
494: so->so_receive = soreceive;
1.78 matt 495: #ifdef MBUFTRACE
496: so->so_rcv.sb_mowner = &prp->pr_domain->dom_mowner;
497: so->so_snd.sb_mowner = &prp->pr_domain->dom_mowner;
498: so->so_mowner = &prp->pr_domain->dom_mowner;
499: #endif
1.138 rmind 500: uid = kauth_cred_geteuid(l->l_cred);
1.115 yamt 501: so->so_uidinfo = uid_find(uid);
1.160 ad 502: if (lockso != NULL) {
503: /* Caller wants us to share a lock. */
504: lock = lockso->so_lock;
505: so->so_lock = lock;
506: mutex_obj_hold(lock);
507: mutex_enter(lock);
508: } else {
509: /* Lock assigned and taken during PRU_ATTACH. */
510: }
1.140 dyoung 511: error = (*prp->pr_usrreq)(so, PRU_ATTACH, NULL,
512: (struct mbuf *)(long)proto, NULL, l);
1.160 ad 513: KASSERT(solocked(so));
1.140 dyoung 514: if (error != 0) {
1.1 cgd 515: so->so_state |= SS_NOFDREF;
516: sofree(so);
1.140 dyoung 517: return error;
1.1 cgd 518: }
1.160 ad 519: sounlock(so);
1.1 cgd 520: *aso = so;
1.140 dyoung 521: return 0;
1.1 cgd 522: }
523:
1.142 dyoung 524: /* On success, write file descriptor to fdout and return zero. On
525: * failure, return non-zero; *fdout will be undefined.
526: */
527: int
528: fsocreate(int domain, struct socket **sop, int type, int protocol,
529: struct lwp *l, int *fdout)
530: {
531: struct socket *so;
532: struct file *fp;
533: int fd, error;
534:
1.155 ad 535: if ((error = fd_allocfile(&fp, &fd)) != 0)
1.142 dyoung 536: return (error);
537: fp->f_flag = FREAD|FWRITE;
538: fp->f_type = DTYPE_SOCKET;
539: fp->f_ops = &socketops;
1.160 ad 540: error = socreate(domain, &so, type, protocol, l, NULL);
1.142 dyoung 541: if (error != 0) {
1.155 ad 542: fd_abort(curproc, fp, fd);
1.142 dyoung 543: } else {
544: if (sop != NULL)
545: *sop = so;
546: fp->f_data = so;
1.155 ad 547: fd_affix(curproc, fp, fd);
1.142 dyoung 548: *fdout = fd;
549: }
550: return error;
551: }
552:
1.3 andrew 553: int
1.114 christos 554: sobind(struct socket *so, struct mbuf *nam, struct lwp *l)
1.1 cgd 555: {
1.160 ad 556: int error;
1.1 cgd 557:
1.160 ad 558: solock(so);
1.140 dyoung 559: error = (*so->so_proto->pr_usrreq)(so, PRU_BIND, NULL, nam, NULL, l);
1.160 ad 560: sounlock(so);
1.140 dyoung 561: return error;
1.1 cgd 562: }
563:
1.3 andrew 564: int
1.150 elad 565: solisten(struct socket *so, int backlog, struct lwp *l)
1.1 cgd 566: {
1.160 ad 567: int error;
1.1 cgd 568:
1.160 ad 569: solock(so);
1.158 ad 570: if ((so->so_state & (SS_ISCONNECTED | SS_ISCONNECTING |
1.163 ad 571: SS_ISDISCONNECTING)) != 0) {
572: sounlock(so);
1.158 ad 573: return (EOPNOTSUPP);
1.163 ad 574: }
1.140 dyoung 575: error = (*so->so_proto->pr_usrreq)(so, PRU_LISTEN, NULL,
1.150 elad 576: NULL, NULL, l);
1.140 dyoung 577: if (error != 0) {
1.160 ad 578: sounlock(so);
1.140 dyoung 579: return error;
1.1 cgd 580: }
1.63 matt 581: if (TAILQ_EMPTY(&so->so_q))
1.1 cgd 582: so->so_options |= SO_ACCEPTCONN;
583: if (backlog < 0)
584: backlog = 0;
1.49 jonathan 585: so->so_qlimit = min(backlog, somaxconn);
1.160 ad 586: sounlock(so);
1.140 dyoung 587: return 0;
1.1 cgd 588: }
589:
1.21 christos 590: void
1.54 lukem 591: sofree(struct socket *so)
1.1 cgd 592: {
1.161 ad 593: u_int refs;
1.1 cgd 594:
1.160 ad 595: KASSERT(solocked(so));
596:
597: if (so->so_pcb || (so->so_state & SS_NOFDREF) == 0) {
598: sounlock(so);
1.1 cgd 599: return;
1.160 ad 600: }
1.43 mycroft 601: if (so->so_head) {
602: /*
603: * We must not decommission a socket that's on the accept(2)
604: * queue. If we do, then accept(2) may hang after select(2)
605: * indicated that the listening socket was ready.
606: */
1.160 ad 607: if (!soqremque(so, 0)) {
608: sounlock(so);
1.43 mycroft 609: return;
1.160 ad 610: }
1.43 mycroft 611: }
1.98 christos 612: if (so->so_rcv.sb_hiwat)
1.110 christos 613: (void)chgsbsize(so->so_uidinfo, &so->so_rcv.sb_hiwat, 0,
1.98 christos 614: RLIM_INFINITY);
615: if (so->so_snd.sb_hiwat)
1.110 christos 616: (void)chgsbsize(so->so_uidinfo, &so->so_snd.sb_hiwat, 0,
1.98 christos 617: RLIM_INFINITY);
618: sbrelease(&so->so_snd, so);
1.160 ad 619: KASSERT(!cv_has_waiters(&so->so_cv));
620: KASSERT(!cv_has_waiters(&so->so_rcv.sb_cv));
621: KASSERT(!cv_has_waiters(&so->so_snd.sb_cv));
1.1 cgd 622: sorflush(so);
1.161 ad 623: refs = so->so_aborting; /* XXX */
1.160 ad 624: sounlock(so);
1.161 ad 625: if (refs == 0) /* XXX */
626: soput(so);
1.1 cgd 627: }
628:
629: /*
630: * Close a socket on last file table reference removal.
631: * Initiate disconnect if connected.
632: * Free socket when disconnect complete.
633: */
1.3 andrew 634: int
1.54 lukem 635: soclose(struct socket *so)
1.1 cgd 636: {
1.54 lukem 637: struct socket *so2;
1.160 ad 638: int error;
639: int error2;
1.1 cgd 640:
1.54 lukem 641: error = 0;
1.160 ad 642: solock(so);
1.1 cgd 643: if (so->so_options & SO_ACCEPTCONN) {
1.160 ad 644: do {
645: if ((so2 = TAILQ_FIRST(&so->so_q0)) != 0) {
646: KASSERT(solocked2(so, so2));
647: (void) soqremque(so2, 0);
648: /* soabort drops the lock. */
649: (void) soabort(so2);
650: solock(so);
651: continue;
652: }
653: if ((so2 = TAILQ_FIRST(&so->so_q)) != 0) {
654: KASSERT(solocked2(so, so2));
655: (void) soqremque(so2, 1);
656: /* soabort drops the lock. */
657: (void) soabort(so2);
658: solock(so);
659: continue;
660: }
661: } while (0);
1.1 cgd 662: }
663: if (so->so_pcb == 0)
664: goto discard;
665: if (so->so_state & SS_ISCONNECTED) {
666: if ((so->so_state & SS_ISDISCONNECTING) == 0) {
667: error = sodisconnect(so);
668: if (error)
669: goto drop;
670: }
671: if (so->so_options & SO_LINGER) {
1.151 ad 672: if ((so->so_state & SS_ISDISCONNECTING) && so->so_nbio)
1.1 cgd 673: goto drop;
1.21 christos 674: while (so->so_state & SS_ISCONNECTED) {
1.160 ad 675: error = sowait(so, so->so_linger * hz);
1.21 christos 676: if (error)
1.1 cgd 677: break;
1.21 christos 678: }
1.1 cgd 679: }
680: }
1.54 lukem 681: drop:
1.1 cgd 682: if (so->so_pcb) {
1.160 ad 683: error2 = (*so->so_proto->pr_usrreq)(so, PRU_DETACH,
1.140 dyoung 684: NULL, NULL, NULL, NULL);
1.1 cgd 685: if (error == 0)
686: error = error2;
687: }
1.54 lukem 688: discard:
1.1 cgd 689: if (so->so_state & SS_NOFDREF)
690: panic("soclose: NOFDREF");
691: so->so_state |= SS_NOFDREF;
692: sofree(so);
693: return (error);
694: }
695:
696: /*
1.160 ad 697: * Must be called with the socket locked.. Will return with it unlocked.
1.1 cgd 698: */
1.3 andrew 699: int
1.54 lukem 700: soabort(struct socket *so)
1.1 cgd 701: {
1.161 ad 702: u_int refs;
1.139 yamt 703: int error;
1.160 ad 704:
705: KASSERT(solocked(so));
706: KASSERT(so->so_head == NULL);
1.1 cgd 707:
1.161 ad 708: so->so_aborting++; /* XXX */
1.140 dyoung 709: error = (*so->so_proto->pr_usrreq)(so, PRU_ABORT, NULL,
710: NULL, NULL, NULL);
1.161 ad 711: refs = --so->so_aborting; /* XXX */
1.164 drochner 712: if (error || (refs == 0)) {
1.139 yamt 713: sofree(so);
1.160 ad 714: } else {
715: sounlock(so);
1.139 yamt 716: }
717: return error;
1.1 cgd 718: }
719:
1.3 andrew 720: int
1.54 lukem 721: soaccept(struct socket *so, struct mbuf *nam)
1.1 cgd 722: {
1.160 ad 723: int error;
724:
725: KASSERT(solocked(so));
1.1 cgd 726:
1.54 lukem 727: error = 0;
1.1 cgd 728: if ((so->so_state & SS_NOFDREF) == 0)
729: panic("soaccept: !NOFDREF");
730: so->so_state &= ~SS_NOFDREF;
1.55 thorpej 731: if ((so->so_state & SS_ISDISCONNECTED) == 0 ||
732: (so->so_proto->pr_flags & PR_ABRTACPTDIS) == 0)
1.41 mycroft 733: error = (*so->so_proto->pr_usrreq)(so, PRU_ACCEPT,
1.140 dyoung 734: NULL, nam, NULL, NULL);
1.41 mycroft 735: else
1.53 itojun 736: error = ECONNABORTED;
1.52 itojun 737:
1.1 cgd 738: return (error);
739: }
740:
1.3 andrew 741: int
1.114 christos 742: soconnect(struct socket *so, struct mbuf *nam, struct lwp *l)
1.1 cgd 743: {
1.160 ad 744: int error;
745:
746: KASSERT(solocked(so));
1.1 cgd 747:
748: if (so->so_options & SO_ACCEPTCONN)
749: return (EOPNOTSUPP);
750: /*
751: * If protocol is connection-based, can only connect once.
752: * Otherwise, if connected, try to disconnect first.
753: * This allows user to disconnect by connecting to, e.g.,
754: * a null address.
755: */
756: if (so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING) &&
757: ((so->so_proto->pr_flags & PR_CONNREQUIRED) ||
758: (error = sodisconnect(so))))
759: error = EISCONN;
760: else
761: error = (*so->so_proto->pr_usrreq)(so, PRU_CONNECT,
1.140 dyoung 762: NULL, nam, NULL, l);
1.1 cgd 763: return (error);
764: }
765:
1.3 andrew 766: int
1.54 lukem 767: soconnect2(struct socket *so1, struct socket *so2)
1.1 cgd 768: {
1.160 ad 769: int error;
770:
771: KASSERT(solocked2(so1, so2));
1.1 cgd 772:
1.22 mycroft 773: error = (*so1->so_proto->pr_usrreq)(so1, PRU_CONNECT2,
1.140 dyoung 774: NULL, (struct mbuf *)so2, NULL, NULL);
1.1 cgd 775: return (error);
776: }
777:
1.3 andrew 778: int
1.54 lukem 779: sodisconnect(struct socket *so)
1.1 cgd 780: {
1.160 ad 781: int error;
782:
783: KASSERT(solocked(so));
1.1 cgd 784:
785: if ((so->so_state & SS_ISCONNECTED) == 0) {
786: error = ENOTCONN;
1.160 ad 787: } else if (so->so_state & SS_ISDISCONNECTING) {
1.1 cgd 788: error = EALREADY;
1.160 ad 789: } else {
790: error = (*so->so_proto->pr_usrreq)(so, PRU_DISCONNECT,
791: NULL, NULL, NULL, NULL);
1.1 cgd 792: }
1.117 yamt 793: sodopendfree();
1.1 cgd 794: return (error);
795: }
796:
1.15 mycroft 797: #define SBLOCKWAIT(f) (((f) & MSG_DONTWAIT) ? M_NOWAIT : M_WAITOK)
1.1 cgd 798: /*
799: * Send on a socket.
800: * If send must go all at once and message is larger than
801: * send buffering, then hard error.
802: * Lock against other senders.
803: * If must go all at once and not enough room now, then
804: * inform user that this would block and do nothing.
805: * Otherwise, if nonblocking, send as much as possible.
806: * The data to be sent is described by "uio" if nonzero,
807: * otherwise by the mbuf chain "top" (which must be null
808: * if uio is not). Data provided in mbuf chain must be small
809: * enough to send all at once.
810: *
811: * Returns nonzero on error, timeout or signal; callers
812: * must check for short counts if EINTR/ERESTART are returned.
813: * Data and control buffers are freed on return.
814: */
1.3 andrew 815: int
1.54 lukem 816: sosend(struct socket *so, struct mbuf *addr, struct uio *uio, struct mbuf *top,
1.114 christos 817: struct mbuf *control, int flags, struct lwp *l)
1.1 cgd 818: {
1.54 lukem 819: struct mbuf **mp, *m;
1.114 christos 820: struct proc *p;
1.58 jdolecek 821: long space, len, resid, clen, mlen;
822: int error, s, dontroute, atomic;
1.54 lukem 823:
1.114 christos 824: p = l->l_proc;
1.117 yamt 825: sodopendfree();
1.160 ad 826: clen = 0;
1.64 thorpej 827:
1.160 ad 828: /*
829: * solock() provides atomicity of access. splsoftnet() prevents
830: * protocol processing soft interrupts from interrupting us and
831: * blocking (expensive).
832: */
833: s = splsoftnet();
834: solock(so);
1.54 lukem 835: atomic = sosendallatonce(so) || top;
1.1 cgd 836: if (uio)
837: resid = uio->uio_resid;
838: else
839: resid = top->m_pkthdr.len;
1.7 cgd 840: /*
841: * In theory resid should be unsigned.
842: * However, space must be signed, as it might be less than 0
843: * if we over-committed, and we must use a signed comparison
844: * of space and resid. On the other hand, a negative resid
845: * causes us to loop sending 0-length segments to the protocol.
846: */
1.29 mycroft 847: if (resid < 0) {
848: error = EINVAL;
849: goto out;
850: }
1.1 cgd 851: dontroute =
852: (flags & MSG_DONTROUTE) && (so->so_options & SO_DONTROUTE) == 0 &&
853: (so->so_proto->pr_flags & PR_ATOMIC);
1.165 christos 854: l->l_ru.ru_msgsnd++;
1.1 cgd 855: if (control)
856: clen = control->m_len;
1.54 lukem 857: restart:
1.21 christos 858: if ((error = sblock(&so->so_snd, SBLOCKWAIT(flags))) != 0)
1.1 cgd 859: goto out;
860: do {
1.160 ad 861: if (so->so_state & SS_CANTSENDMORE) {
862: error = EPIPE;
863: goto release;
864: }
1.48 thorpej 865: if (so->so_error) {
866: error = so->so_error;
867: so->so_error = 0;
868: goto release;
869: }
1.1 cgd 870: if ((so->so_state & SS_ISCONNECTED) == 0) {
871: if (so->so_proto->pr_flags & PR_CONNREQUIRED) {
872: if ((so->so_state & SS_ISCONFIRMING) == 0 &&
1.160 ad 873: !(resid == 0 && clen != 0)) {
874: error = ENOTCONN;
875: goto release;
876: }
877: } else if (addr == 0) {
878: error = EDESTADDRREQ;
879: goto release;
880: }
1.1 cgd 881: }
882: space = sbspace(&so->so_snd);
883: if (flags & MSG_OOB)
884: space += 1024;
1.21 christos 885: if ((atomic && resid > so->so_snd.sb_hiwat) ||
1.160 ad 886: clen > so->so_snd.sb_hiwat) {
887: error = EMSGSIZE;
888: goto release;
889: }
1.96 mycroft 890: if (space < resid + clen &&
1.1 cgd 891: (atomic || space < so->so_snd.sb_lowat || space < clen)) {
1.160 ad 892: if (so->so_nbio) {
893: error = EWOULDBLOCK;
894: goto release;
895: }
1.1 cgd 896: sbunlock(&so->so_snd);
897: error = sbwait(&so->so_snd);
898: if (error)
899: goto out;
900: goto restart;
901: }
902: mp = ⊤
903: space -= clen;
904: do {
1.45 tv 905: if (uio == NULL) {
906: /*
907: * Data is prepackaged in "top".
908: */
909: resid = 0;
910: if (flags & MSG_EOR)
911: top->m_flags |= M_EOR;
912: } else do {
1.160 ad 913: sounlock(so);
914: splx(s);
1.144 dyoung 915: if (top == NULL) {
1.78 matt 916: m = m_gethdr(M_WAIT, MT_DATA);
1.45 tv 917: mlen = MHLEN;
918: m->m_pkthdr.len = 0;
1.140 dyoung 919: m->m_pkthdr.rcvif = NULL;
1.45 tv 920: } else {
1.78 matt 921: m = m_get(M_WAIT, MT_DATA);
1.45 tv 922: mlen = MLEN;
923: }
1.78 matt 924: MCLAIM(m, so->so_snd.sb_mowner);
1.121 yamt 925: if (sock_loan_thresh >= 0 &&
926: uio->uio_iov->iov_len >= sock_loan_thresh &&
927: space >= sock_loan_thresh &&
1.64 thorpej 928: (len = sosend_loan(so, uio, m,
929: space)) != 0) {
930: SOSEND_COUNTER_INCR(&sosend_loan_big);
931: space -= len;
932: goto have_data;
933: }
1.45 tv 934: if (resid >= MINCLSIZE && space >= MCLBYTES) {
1.64 thorpej 935: SOSEND_COUNTER_INCR(&sosend_copy_big);
1.78 matt 936: m_clget(m, M_WAIT);
1.45 tv 937: if ((m->m_flags & M_EXT) == 0)
938: goto nopages;
939: mlen = MCLBYTES;
940: if (atomic && top == 0) {
1.58 jdolecek 941: len = lmin(MCLBYTES - max_hdr,
1.54 lukem 942: resid);
1.45 tv 943: m->m_data += max_hdr;
944: } else
1.58 jdolecek 945: len = lmin(MCLBYTES, resid);
1.45 tv 946: space -= len;
947: } else {
1.64 thorpej 948: nopages:
949: SOSEND_COUNTER_INCR(&sosend_copy_small);
1.58 jdolecek 950: len = lmin(lmin(mlen, resid), space);
1.45 tv 951: space -= len;
952: /*
953: * For datagram protocols, leave room
954: * for protocol headers in first mbuf.
955: */
956: if (atomic && top == 0 && len < mlen)
957: MH_ALIGN(m, len);
958: }
1.144 dyoung 959: error = uiomove(mtod(m, void *), (int)len, uio);
1.64 thorpej 960: have_data:
1.45 tv 961: resid = uio->uio_resid;
962: m->m_len = len;
963: *mp = m;
964: top->m_pkthdr.len += len;
1.160 ad 965: s = splsoftnet();
966: solock(so);
1.144 dyoung 967: if (error != 0)
1.45 tv 968: goto release;
969: mp = &m->m_next;
970: if (resid <= 0) {
971: if (flags & MSG_EOR)
972: top->m_flags |= M_EOR;
973: break;
974: }
975: } while (space > 0 && atomic);
1.108 perry 976:
1.160 ad 977: if (so->so_state & SS_CANTSENDMORE) {
978: error = EPIPE;
979: goto release;
980: }
1.45 tv 981: if (dontroute)
982: so->so_options |= SO_DONTROUTE;
983: if (resid > 0)
984: so->so_state |= SS_MORETOCOME;
1.46 sommerfe 985: error = (*so->so_proto->pr_usrreq)(so,
986: (flags & MSG_OOB) ? PRU_SENDOOB : PRU_SEND,
1.160 ad 987: top, addr, control, curlwp);
1.45 tv 988: if (dontroute)
989: so->so_options &= ~SO_DONTROUTE;
990: if (resid > 0)
991: so->so_state &= ~SS_MORETOCOME;
992: clen = 0;
1.144 dyoung 993: control = NULL;
994: top = NULL;
1.45 tv 995: mp = ⊤
1.144 dyoung 996: if (error != 0)
1.1 cgd 997: goto release;
998: } while (resid && space > 0);
999: } while (resid);
1000:
1.54 lukem 1001: release:
1.1 cgd 1002: sbunlock(&so->so_snd);
1.54 lukem 1003: out:
1.160 ad 1004: sounlock(so);
1005: splx(s);
1.1 cgd 1006: if (top)
1007: m_freem(top);
1008: if (control)
1009: m_freem(control);
1010: return (error);
1011: }
1012:
1013: /*
1.159 ad 1014: * Following replacement or removal of the first mbuf on the first
1015: * mbuf chain of a socket buffer, push necessary state changes back
1016: * into the socket buffer so that other consumers see the values
1017: * consistently. 'nextrecord' is the callers locally stored value of
1018: * the original value of sb->sb_mb->m_nextpkt which must be restored
1019: * when the lead mbuf changes. NOTE: 'nextrecord' may be NULL.
1020: */
1021: static void
1022: sbsync(struct sockbuf *sb, struct mbuf *nextrecord)
1023: {
1024:
1.160 ad 1025: KASSERT(solocked(sb->sb_so));
1026:
1.159 ad 1027: /*
1028: * First, update for the new value of nextrecord. If necessary,
1029: * make it the first record.
1030: */
1031: if (sb->sb_mb != NULL)
1032: sb->sb_mb->m_nextpkt = nextrecord;
1033: else
1034: sb->sb_mb = nextrecord;
1035:
1036: /*
1037: * Now update any dependent socket buffer fields to reflect
1038: * the new state. This is an inline of SB_EMPTY_FIXUP, with
1039: * the addition of a second clause that takes care of the
1040: * case where sb_mb has been updated, but remains the last
1041: * record.
1042: */
1043: if (sb->sb_mb == NULL) {
1044: sb->sb_mbtail = NULL;
1045: sb->sb_lastrecord = NULL;
1046: } else if (sb->sb_mb->m_nextpkt == NULL)
1047: sb->sb_lastrecord = sb->sb_mb;
1048: }
1049:
1050: /*
1.1 cgd 1051: * Implement receive operations on a socket.
1052: * We depend on the way that records are added to the sockbuf
1053: * by sbappend*. In particular, each record (mbufs linked through m_next)
1054: * must begin with an address if the protocol so specifies,
1055: * followed by an optional mbuf or mbufs containing ancillary data,
1056: * and then zero or more mbufs of data.
1057: * In order to avoid blocking network interrupts for the entire time here,
1058: * we splx() while doing the actual copy to user space.
1059: * Although the sockbuf is locked, new data may still be appended,
1060: * and thus we must maintain consistency of the sockbuf during that time.
1061: *
1062: * The caller may receive the data as a single mbuf chain by supplying
1063: * an mbuf **mp0 for use in returning the chain. The uio is then used
1064: * only for the count in uio_resid.
1065: */
1.3 andrew 1066: int
1.54 lukem 1067: soreceive(struct socket *so, struct mbuf **paddr, struct uio *uio,
1068: struct mbuf **mp0, struct mbuf **controlp, int *flagsp)
1.1 cgd 1069: {
1.116 yamt 1070: struct lwp *l = curlwp;
1.160 ad 1071: struct mbuf *m, **mp, *mt;
1.146 dyoung 1072: int atomic, flags, len, error, s, offset, moff, type, orig_resid;
1.99 matt 1073: const struct protosw *pr;
1.54 lukem 1074: struct mbuf *nextrecord;
1.67 he 1075: int mbuf_removed = 0;
1.146 dyoung 1076: const struct domain *dom;
1.64 thorpej 1077:
1.54 lukem 1078: pr = so->so_proto;
1.146 dyoung 1079: atomic = pr->pr_flags & PR_ATOMIC;
1080: dom = pr->pr_domain;
1.1 cgd 1081: mp = mp0;
1.54 lukem 1082: type = 0;
1083: orig_resid = uio->uio_resid;
1.102 jonathan 1084:
1.144 dyoung 1085: if (paddr != NULL)
1086: *paddr = NULL;
1087: if (controlp != NULL)
1088: *controlp = NULL;
1089: if (flagsp != NULL)
1.1 cgd 1090: flags = *flagsp &~ MSG_EOR;
1091: else
1092: flags = 0;
1.66 enami 1093:
1094: if ((flags & MSG_DONTWAIT) == 0)
1.117 yamt 1095: sodopendfree();
1.66 enami 1096:
1.1 cgd 1097: if (flags & MSG_OOB) {
1098: m = m_get(M_WAIT, MT_DATA);
1.160 ad 1099: solock(so);
1.17 cgd 1100: error = (*pr->pr_usrreq)(so, PRU_RCVOOB, m,
1.140 dyoung 1101: (struct mbuf *)(long)(flags & MSG_PEEK), NULL, l);
1.160 ad 1102: sounlock(so);
1.1 cgd 1103: if (error)
1104: goto bad;
1105: do {
1.134 christos 1106: error = uiomove(mtod(m, void *),
1.1 cgd 1107: (int) min(uio->uio_resid, m->m_len), uio);
1108: m = m_free(m);
1.144 dyoung 1109: } while (uio->uio_resid > 0 && error == 0 && m);
1.54 lukem 1110: bad:
1.144 dyoung 1111: if (m != NULL)
1.1 cgd 1112: m_freem(m);
1.144 dyoung 1113: return error;
1.1 cgd 1114: }
1.144 dyoung 1115: if (mp != NULL)
1.140 dyoung 1116: *mp = NULL;
1.160 ad 1117:
1118: /*
1119: * solock() provides atomicity of access. splsoftnet() prevents
1120: * protocol processing soft interrupts from interrupting us and
1121: * blocking (expensive).
1122: */
1123: s = splsoftnet();
1124: solock(so);
1.1 cgd 1125: if (so->so_state & SS_ISCONFIRMING && uio->uio_resid)
1.140 dyoung 1126: (*pr->pr_usrreq)(so, PRU_RCVD, NULL, NULL, NULL, l);
1.1 cgd 1127:
1.54 lukem 1128: restart:
1.160 ad 1129: if ((error = sblock(&so->so_rcv, SBLOCKWAIT(flags))) != 0) {
1130: sounlock(so);
1131: splx(s);
1.144 dyoung 1132: return error;
1.160 ad 1133: }
1.1 cgd 1134:
1135: m = so->so_rcv.sb_mb;
1136: /*
1137: * If we have less data than requested, block awaiting more
1138: * (subject to any timeout) if:
1.15 mycroft 1139: * 1. the current count is less than the low water mark,
1.1 cgd 1140: * 2. MSG_WAITALL is set, and it is possible to do the entire
1.15 mycroft 1141: * receive operation at once if we block (resid <= hiwat), or
1142: * 3. MSG_DONTWAIT is not set.
1.1 cgd 1143: * If MSG_WAITALL is set but resid is larger than the receive buffer,
1144: * we have to do the receive in sections, and thus risk returning
1145: * a short count if a timeout or signal occurs after we start.
1146: */
1.144 dyoung 1147: if (m == NULL ||
1148: ((flags & MSG_DONTWAIT) == 0 &&
1149: so->so_rcv.sb_cc < uio->uio_resid &&
1150: (so->so_rcv.sb_cc < so->so_rcv.sb_lowat ||
1151: ((flags & MSG_WAITALL) &&
1152: uio->uio_resid <= so->so_rcv.sb_hiwat)) &&
1.146 dyoung 1153: m->m_nextpkt == NULL && !atomic)) {
1.1 cgd 1154: #ifdef DIAGNOSTIC
1.144 dyoung 1155: if (m == NULL && so->so_rcv.sb_cc)
1.1 cgd 1156: panic("receive 1");
1157: #endif
1158: if (so->so_error) {
1.144 dyoung 1159: if (m != NULL)
1.15 mycroft 1160: goto dontblock;
1.1 cgd 1161: error = so->so_error;
1162: if ((flags & MSG_PEEK) == 0)
1163: so->so_error = 0;
1164: goto release;
1165: }
1166: if (so->so_state & SS_CANTRCVMORE) {
1.144 dyoung 1167: if (m != NULL)
1.15 mycroft 1168: goto dontblock;
1.1 cgd 1169: else
1170: goto release;
1171: }
1.144 dyoung 1172: for (; m != NULL; m = m->m_next)
1.1 cgd 1173: if (m->m_type == MT_OOBDATA || (m->m_flags & M_EOR)) {
1174: m = so->so_rcv.sb_mb;
1175: goto dontblock;
1176: }
1177: if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) == 0 &&
1178: (so->so_proto->pr_flags & PR_CONNREQUIRED)) {
1179: error = ENOTCONN;
1180: goto release;
1181: }
1182: if (uio->uio_resid == 0)
1183: goto release;
1.151 ad 1184: if (so->so_nbio || (flags & MSG_DONTWAIT)) {
1.1 cgd 1185: error = EWOULDBLOCK;
1186: goto release;
1187: }
1.69 thorpej 1188: SBLASTRECORDCHK(&so->so_rcv, "soreceive sbwait 1");
1189: SBLASTMBUFCHK(&so->so_rcv, "soreceive sbwait 1");
1.1 cgd 1190: sbunlock(&so->so_rcv);
1191: error = sbwait(&so->so_rcv);
1.160 ad 1192: if (error != 0) {
1193: sounlock(so);
1194: splx(s);
1.144 dyoung 1195: return error;
1.160 ad 1196: }
1.1 cgd 1197: goto restart;
1198: }
1.54 lukem 1199: dontblock:
1.69 thorpej 1200: /*
1201: * On entry here, m points to the first record of the socket buffer.
1.159 ad 1202: * From this point onward, we maintain 'nextrecord' as a cache of the
1203: * pointer to the next record in the socket buffer. We must keep the
1204: * various socket buffer pointers and local stack versions of the
1205: * pointers in sync, pushing out modifications before dropping the
1.160 ad 1206: * socket lock, and re-reading them when picking it up.
1.159 ad 1207: *
1208: * Otherwise, we will race with the network stack appending new data
1209: * or records onto the socket buffer by using inconsistent/stale
1210: * versions of the field, possibly resulting in socket buffer
1211: * corruption.
1212: *
1213: * By holding the high-level sblock(), we prevent simultaneous
1214: * readers from pulling off the front of the socket buffer.
1.69 thorpej 1215: */
1.144 dyoung 1216: if (l != NULL)
1.157 ad 1217: l->l_ru.ru_msgrcv++;
1.69 thorpej 1218: KASSERT(m == so->so_rcv.sb_mb);
1219: SBLASTRECORDCHK(&so->so_rcv, "soreceive 1");
1220: SBLASTMBUFCHK(&so->so_rcv, "soreceive 1");
1.1 cgd 1221: nextrecord = m->m_nextpkt;
1222: if (pr->pr_flags & PR_ADDR) {
1223: #ifdef DIAGNOSTIC
1224: if (m->m_type != MT_SONAME)
1225: panic("receive 1a");
1226: #endif
1.3 andrew 1227: orig_resid = 0;
1.1 cgd 1228: if (flags & MSG_PEEK) {
1229: if (paddr)
1230: *paddr = m_copy(m, 0, m->m_len);
1231: m = m->m_next;
1232: } else {
1233: sbfree(&so->so_rcv, m);
1.67 he 1234: mbuf_removed = 1;
1.144 dyoung 1235: if (paddr != NULL) {
1.1 cgd 1236: *paddr = m;
1237: so->so_rcv.sb_mb = m->m_next;
1.144 dyoung 1238: m->m_next = NULL;
1.1 cgd 1239: m = so->so_rcv.sb_mb;
1240: } else {
1241: MFREE(m, so->so_rcv.sb_mb);
1242: m = so->so_rcv.sb_mb;
1243: }
1.159 ad 1244: sbsync(&so->so_rcv, nextrecord);
1.1 cgd 1245: }
1246: }
1.159 ad 1247:
1248: /*
1249: * Process one or more MT_CONTROL mbufs present before any data mbufs
1250: * in the first mbuf chain on the socket buffer. If MSG_PEEK, we
1251: * just copy the data; if !MSG_PEEK, we call into the protocol to
1252: * perform externalization (or freeing if controlp == NULL).
1253: */
1254: if (__predict_false(m != NULL && m->m_type == MT_CONTROL)) {
1255: struct mbuf *cm = NULL, *cmn;
1256: struct mbuf **cme = &cm;
1257:
1258: do {
1259: if (flags & MSG_PEEK) {
1260: if (controlp != NULL) {
1261: *controlp = m_copy(m, 0, m->m_len);
1262: controlp = &(*controlp)->m_next;
1263: }
1264: m = m->m_next;
1265: } else {
1266: sbfree(&so->so_rcv, m);
1.1 cgd 1267: so->so_rcv.sb_mb = m->m_next;
1.144 dyoung 1268: m->m_next = NULL;
1.159 ad 1269: *cme = m;
1270: cme = &(*cme)->m_next;
1.1 cgd 1271: m = so->so_rcv.sb_mb;
1.159 ad 1272: }
1273: } while (m != NULL && m->m_type == MT_CONTROL);
1274: if ((flags & MSG_PEEK) == 0)
1275: sbsync(&so->so_rcv, nextrecord);
1276: for (; cm != NULL; cm = cmn) {
1277: cmn = cm->m_next;
1278: cm->m_next = NULL;
1279: type = mtod(cm, struct cmsghdr *)->cmsg_type;
1280: if (controlp != NULL) {
1281: if (dom->dom_externalize != NULL &&
1282: type == SCM_RIGHTS) {
1.160 ad 1283: sounlock(so);
1.159 ad 1284: splx(s);
1285: error = (*dom->dom_externalize)(cm, l);
1286: s = splsoftnet();
1.160 ad 1287: solock(so);
1.159 ad 1288: }
1289: *controlp = cm;
1290: while (*controlp != NULL)
1291: controlp = &(*controlp)->m_next;
1.1 cgd 1292: } else {
1.106 itojun 1293: /*
1294: * Dispose of any SCM_RIGHTS message that went
1295: * through the read path rather than recv.
1296: */
1.159 ad 1297: if (dom->dom_dispose != NULL &&
1298: type == SCM_RIGHTS) {
1.160 ad 1299: sounlock(so);
1.159 ad 1300: (*dom->dom_dispose)(cm);
1.160 ad 1301: solock(so);
1.159 ad 1302: }
1303: m_freem(cm);
1.1 cgd 1304: }
1305: }
1.159 ad 1306: if (m != NULL)
1307: nextrecord = so->so_rcv.sb_mb->m_nextpkt;
1308: else
1309: nextrecord = so->so_rcv.sb_mb;
1310: orig_resid = 0;
1.1 cgd 1311: }
1.69 thorpej 1312:
1.159 ad 1313: /* If m is non-NULL, we have some data to read. */
1314: if (__predict_true(m != NULL)) {
1.1 cgd 1315: type = m->m_type;
1316: if (type == MT_OOBDATA)
1317: flags |= MSG_OOB;
1318: }
1.69 thorpej 1319: SBLASTRECORDCHK(&so->so_rcv, "soreceive 2");
1320: SBLASTMBUFCHK(&so->so_rcv, "soreceive 2");
1321:
1.1 cgd 1322: moff = 0;
1323: offset = 0;
1.144 dyoung 1324: while (m != NULL && uio->uio_resid > 0 && error == 0) {
1.1 cgd 1325: if (m->m_type == MT_OOBDATA) {
1326: if (type != MT_OOBDATA)
1327: break;
1328: } else if (type == MT_OOBDATA)
1329: break;
1330: #ifdef DIAGNOSTIC
1331: else if (m->m_type != MT_DATA && m->m_type != MT_HEADER)
1332: panic("receive 3");
1333: #endif
1334: so->so_state &= ~SS_RCVATMARK;
1335: len = uio->uio_resid;
1336: if (so->so_oobmark && len > so->so_oobmark - offset)
1337: len = so->so_oobmark - offset;
1338: if (len > m->m_len - moff)
1339: len = m->m_len - moff;
1340: /*
1341: * If mp is set, just pass back the mbufs.
1342: * Otherwise copy them out via the uio, then free.
1343: * Sockbuf must be consistent here (points to current mbuf,
1344: * it points to next record) when we drop priority;
1345: * we must note any additions to the sockbuf when we
1346: * block interrupts again.
1347: */
1.144 dyoung 1348: if (mp == NULL) {
1.69 thorpej 1349: SBLASTRECORDCHK(&so->so_rcv, "soreceive uiomove");
1350: SBLASTMBUFCHK(&so->so_rcv, "soreceive uiomove");
1.160 ad 1351: sounlock(so);
1.1 cgd 1352: splx(s);
1.134 christos 1353: error = uiomove(mtod(m, char *) + moff, (int)len, uio);
1.20 mycroft 1354: s = splsoftnet();
1.160 ad 1355: solock(so);
1.144 dyoung 1356: if (error != 0) {
1.67 he 1357: /*
1358: * If any part of the record has been removed
1359: * (such as the MT_SONAME mbuf, which will
1360: * happen when PR_ADDR, and thus also
1361: * PR_ATOMIC, is set), then drop the entire
1362: * record to maintain the atomicity of the
1363: * receive operation.
1364: *
1365: * This avoids a later panic("receive 1a")
1366: * when compiled with DIAGNOSTIC.
1367: */
1.146 dyoung 1368: if (m && mbuf_removed && atomic)
1.67 he 1369: (void) sbdroprecord(&so->so_rcv);
1370:
1.57 jdolecek 1371: goto release;
1.67 he 1372: }
1.1 cgd 1373: } else
1374: uio->uio_resid -= len;
1375: if (len == m->m_len - moff) {
1376: if (m->m_flags & M_EOR)
1377: flags |= MSG_EOR;
1378: if (flags & MSG_PEEK) {
1379: m = m->m_next;
1380: moff = 0;
1381: } else {
1382: nextrecord = m->m_nextpkt;
1383: sbfree(&so->so_rcv, m);
1384: if (mp) {
1385: *mp = m;
1386: mp = &m->m_next;
1387: so->so_rcv.sb_mb = m = m->m_next;
1.140 dyoung 1388: *mp = NULL;
1.1 cgd 1389: } else {
1390: MFREE(m, so->so_rcv.sb_mb);
1391: m = so->so_rcv.sb_mb;
1392: }
1.69 thorpej 1393: /*
1394: * If m != NULL, we also know that
1395: * so->so_rcv.sb_mb != NULL.
1396: */
1397: KASSERT(so->so_rcv.sb_mb == m);
1398: if (m) {
1.1 cgd 1399: m->m_nextpkt = nextrecord;
1.69 thorpej 1400: if (nextrecord == NULL)
1401: so->so_rcv.sb_lastrecord = m;
1402: } else {
1403: so->so_rcv.sb_mb = nextrecord;
1.70 thorpej 1404: SB_EMPTY_FIXUP(&so->so_rcv);
1.69 thorpej 1405: }
1406: SBLASTRECORDCHK(&so->so_rcv, "soreceive 3");
1407: SBLASTMBUFCHK(&so->so_rcv, "soreceive 3");
1.1 cgd 1408: }
1.144 dyoung 1409: } else if (flags & MSG_PEEK)
1410: moff += len;
1411: else {
1.160 ad 1412: if (mp != NULL) {
1413: mt = m_copym(m, 0, len, M_NOWAIT);
1414: if (__predict_false(mt == NULL)) {
1415: sounlock(so);
1416: mt = m_copym(m, 0, len, M_WAIT);
1417: solock(so);
1418: }
1419: *mp = mt;
1420: }
1.144 dyoung 1421: m->m_data += len;
1422: m->m_len -= len;
1423: so->so_rcv.sb_cc -= len;
1.1 cgd 1424: }
1425: if (so->so_oobmark) {
1426: if ((flags & MSG_PEEK) == 0) {
1427: so->so_oobmark -= len;
1428: if (so->so_oobmark == 0) {
1429: so->so_state |= SS_RCVATMARK;
1430: break;
1431: }
1.7 cgd 1432: } else {
1.1 cgd 1433: offset += len;
1.7 cgd 1434: if (offset == so->so_oobmark)
1435: break;
1436: }
1.1 cgd 1437: }
1438: if (flags & MSG_EOR)
1439: break;
1440: /*
1441: * If the MSG_WAITALL flag is set (for non-atomic socket),
1442: * we must not quit until "uio->uio_resid == 0" or an error
1443: * termination. If a signal/timeout occurs, return
1444: * with a short count but without error.
1445: * Keep sockbuf locked against other readers.
1446: */
1.144 dyoung 1447: while (flags & MSG_WAITALL && m == NULL && uio->uio_resid > 0 &&
1.3 andrew 1448: !sosendallatonce(so) && !nextrecord) {
1.1 cgd 1449: if (so->so_error || so->so_state & SS_CANTRCVMORE)
1450: break;
1.68 matt 1451: /*
1452: * If we are peeking and the socket receive buffer is
1453: * full, stop since we can't get more data to peek at.
1454: */
1455: if ((flags & MSG_PEEK) && sbspace(&so->so_rcv) <= 0)
1456: break;
1457: /*
1458: * If we've drained the socket buffer, tell the
1459: * protocol in case it needs to do something to
1460: * get it filled again.
1461: */
1462: if ((pr->pr_flags & PR_WANTRCVD) && so->so_pcb)
1463: (*pr->pr_usrreq)(so, PRU_RCVD,
1.140 dyoung 1464: NULL, (struct mbuf *)(long)flags, NULL, l);
1.69 thorpej 1465: SBLASTRECORDCHK(&so->so_rcv, "soreceive sbwait 2");
1466: SBLASTMBUFCHK(&so->so_rcv, "soreceive sbwait 2");
1.1 cgd 1467: error = sbwait(&so->so_rcv);
1.144 dyoung 1468: if (error != 0) {
1.1 cgd 1469: sbunlock(&so->so_rcv);
1.160 ad 1470: sounlock(so);
1.1 cgd 1471: splx(s);
1.144 dyoung 1472: return 0;
1.1 cgd 1473: }
1.21 christos 1474: if ((m = so->so_rcv.sb_mb) != NULL)
1.1 cgd 1475: nextrecord = m->m_nextpkt;
1476: }
1477: }
1.3 andrew 1478:
1.146 dyoung 1479: if (m && atomic) {
1.3 andrew 1480: flags |= MSG_TRUNC;
1481: if ((flags & MSG_PEEK) == 0)
1482: (void) sbdroprecord(&so->so_rcv);
1483: }
1.1 cgd 1484: if ((flags & MSG_PEEK) == 0) {
1.144 dyoung 1485: if (m == NULL) {
1.69 thorpej 1486: /*
1.70 thorpej 1487: * First part is an inline SB_EMPTY_FIXUP(). Second
1.69 thorpej 1488: * part makes sure sb_lastrecord is up-to-date if
1489: * there is still data in the socket buffer.
1490: */
1.1 cgd 1491: so->so_rcv.sb_mb = nextrecord;
1.69 thorpej 1492: if (so->so_rcv.sb_mb == NULL) {
1493: so->so_rcv.sb_mbtail = NULL;
1494: so->so_rcv.sb_lastrecord = NULL;
1495: } else if (nextrecord->m_nextpkt == NULL)
1496: so->so_rcv.sb_lastrecord = nextrecord;
1497: }
1498: SBLASTRECORDCHK(&so->so_rcv, "soreceive 4");
1499: SBLASTMBUFCHK(&so->so_rcv, "soreceive 4");
1.1 cgd 1500: if (pr->pr_flags & PR_WANTRCVD && so->so_pcb)
1.140 dyoung 1501: (*pr->pr_usrreq)(so, PRU_RCVD, NULL,
1502: (struct mbuf *)(long)flags, NULL, l);
1.1 cgd 1503: }
1.3 andrew 1504: if (orig_resid == uio->uio_resid && orig_resid &&
1505: (flags & MSG_EOR) == 0 && (so->so_state & SS_CANTRCVMORE) == 0) {
1506: sbunlock(&so->so_rcv);
1507: goto restart;
1508: }
1.108 perry 1509:
1.144 dyoung 1510: if (flagsp != NULL)
1.1 cgd 1511: *flagsp |= flags;
1.54 lukem 1512: release:
1.1 cgd 1513: sbunlock(&so->so_rcv);
1.160 ad 1514: sounlock(so);
1.1 cgd 1515: splx(s);
1.144 dyoung 1516: return error;
1.1 cgd 1517: }
1518:
1.14 mycroft 1519: int
1.54 lukem 1520: soshutdown(struct socket *so, int how)
1.1 cgd 1521: {
1.99 matt 1522: const struct protosw *pr;
1.160 ad 1523: int error;
1524:
1525: KASSERT(solocked(so));
1.34 kleink 1526:
1.54 lukem 1527: pr = so->so_proto;
1.34 kleink 1528: if (!(how == SHUT_RD || how == SHUT_WR || how == SHUT_RDWR))
1529: return (EINVAL);
1.1 cgd 1530:
1.160 ad 1531: if (how == SHUT_RD || how == SHUT_RDWR) {
1.1 cgd 1532: sorflush(so);
1.160 ad 1533: error = 0;
1534: }
1.34 kleink 1535: if (how == SHUT_WR || how == SHUT_RDWR)
1.160 ad 1536: error = (*pr->pr_usrreq)(so, PRU_SHUTDOWN, NULL,
1.140 dyoung 1537: NULL, NULL, NULL);
1.160 ad 1538:
1539: return error;
1.1 cgd 1540: }
1541:
1.14 mycroft 1542: void
1.54 lukem 1543: sorflush(struct socket *so)
1.1 cgd 1544: {
1.54 lukem 1545: struct sockbuf *sb, asb;
1.99 matt 1546: const struct protosw *pr;
1.160 ad 1547:
1548: KASSERT(solocked(so));
1.1 cgd 1549:
1.54 lukem 1550: sb = &so->so_rcv;
1551: pr = so->so_proto;
1.160 ad 1552: socantrcvmore(so);
1.1 cgd 1553: sb->sb_flags |= SB_NOINTR;
1.160 ad 1554: (void )sblock(sb, M_WAITOK);
1.1 cgd 1555: sbunlock(sb);
1556: asb = *sb;
1.86 wrstuden 1557: /*
1558: * Clear most of the sockbuf structure, but leave some of the
1559: * fields valid.
1560: */
1561: memset(&sb->sb_startzero, 0,
1562: sizeof(*sb) - offsetof(struct sockbuf, sb_startzero));
1.160 ad 1563: if (pr->pr_flags & PR_RIGHTS && pr->pr_domain->dom_dispose) {
1564: sounlock(so);
1.1 cgd 1565: (*pr->pr_domain->dom_dispose)(asb.sb_mb);
1.160 ad 1566: solock(so);
1567: }
1.98 christos 1568: sbrelease(&asb, so);
1.1 cgd 1569: }
1570:
1.142 dyoung 1571: static int
1572: sosetopt1(struct socket *so, int level, int optname, struct mbuf *m)
1.1 cgd 1573: {
1.142 dyoung 1574: int optval, val;
1.130 christos 1575: struct linger *l;
1.141 yamt 1576: struct sockbuf *sb;
1.142 dyoung 1577: struct timeval *tv;
1578:
1579: switch (optname) {
1580:
1581: case SO_LINGER:
1582: if (m == NULL || m->m_len != sizeof(struct linger))
1583: return EINVAL;
1584: l = mtod(m, struct linger *);
1585: if (l->l_linger < 0 || l->l_linger > USHRT_MAX ||
1586: l->l_linger > (INT_MAX / hz))
1587: return EDOM;
1588: so->so_linger = l->l_linger;
1589: if (l->l_onoff)
1590: so->so_options |= SO_LINGER;
1591: else
1592: so->so_options &= ~SO_LINGER;
1593: break;
1.1 cgd 1594:
1.142 dyoung 1595: case SO_DEBUG:
1596: case SO_KEEPALIVE:
1597: case SO_DONTROUTE:
1598: case SO_USELOOPBACK:
1599: case SO_BROADCAST:
1600: case SO_REUSEADDR:
1601: case SO_REUSEPORT:
1602: case SO_OOBINLINE:
1603: case SO_TIMESTAMP:
1604: if (m == NULL || m->m_len < sizeof(int))
1605: return EINVAL;
1606: if (*mtod(m, int *))
1607: so->so_options |= optname;
1608: else
1609: so->so_options &= ~optname;
1610: break;
1611:
1612: case SO_SNDBUF:
1613: case SO_RCVBUF:
1614: case SO_SNDLOWAT:
1615: case SO_RCVLOWAT:
1616: if (m == NULL || m->m_len < sizeof(int))
1617: return EINVAL;
1.1 cgd 1618:
1.142 dyoung 1619: /*
1620: * Values < 1 make no sense for any of these
1621: * options, so disallow them.
1622: */
1623: optval = *mtod(m, int *);
1624: if (optval < 1)
1625: return EINVAL;
1.1 cgd 1626:
1.142 dyoung 1627: switch (optname) {
1.1 cgd 1628:
1629: case SO_SNDBUF:
1630: case SO_RCVBUF:
1.142 dyoung 1631: sb = (optname == SO_SNDBUF) ?
1632: &so->so_snd : &so->so_rcv;
1633: if (sbreserve(sb, (u_long)optval, so) == 0)
1634: return ENOBUFS;
1635: sb->sb_flags &= ~SB_AUTOSIZE;
1636: break;
1637:
1638: /*
1639: * Make sure the low-water is never greater than
1640: * the high-water.
1641: */
1.1 cgd 1642: case SO_SNDLOWAT:
1.142 dyoung 1643: so->so_snd.sb_lowat =
1644: (optval > so->so_snd.sb_hiwat) ?
1645: so->so_snd.sb_hiwat : optval;
1646: break;
1.1 cgd 1647: case SO_RCVLOWAT:
1.142 dyoung 1648: so->so_rcv.sb_lowat =
1649: (optval > so->so_rcv.sb_hiwat) ?
1650: so->so_rcv.sb_hiwat : optval;
1651: break;
1652: }
1653: break;
1.28 thorpej 1654:
1.142 dyoung 1655: case SO_SNDTIMEO:
1656: case SO_RCVTIMEO:
1657: if (m == NULL || m->m_len < sizeof(*tv))
1658: return EINVAL;
1659: tv = mtod(m, struct timeval *);
1660: if (tv->tv_sec > (INT_MAX - tv->tv_usec / tick) / hz)
1661: return EDOM;
1662: val = tv->tv_sec * hz + tv->tv_usec / tick;
1663: if (val == 0 && tv->tv_usec != 0)
1664: val = 1;
1.28 thorpej 1665:
1.142 dyoung 1666: switch (optname) {
1.28 thorpej 1667:
1.142 dyoung 1668: case SO_SNDTIMEO:
1669: so->so_snd.sb_timeo = val;
1.1 cgd 1670: break;
1671: case SO_RCVTIMEO:
1.142 dyoung 1672: so->so_rcv.sb_timeo = val;
1673: break;
1674: }
1675: break;
1.1 cgd 1676:
1.142 dyoung 1677: default:
1678: return ENOPROTOOPT;
1679: }
1680: return 0;
1681: }
1.1 cgd 1682:
1.142 dyoung 1683: int
1684: sosetopt(struct socket *so, int level, int optname, struct mbuf *m)
1685: {
1686: int error, prerr;
1.1 cgd 1687:
1.160 ad 1688: solock(so);
1.142 dyoung 1689: if (level == SOL_SOCKET)
1690: error = sosetopt1(so, level, optname, m);
1691: else
1692: error = ENOPROTOOPT;
1.1 cgd 1693:
1.142 dyoung 1694: if ((error == 0 || error == ENOPROTOOPT) &&
1695: so->so_proto != NULL && so->so_proto->pr_ctloutput != NULL) {
1696: /* give the protocol stack a shot */
1697: prerr = (*so->so_proto->pr_ctloutput)(PRCO_SETOPT, so, level,
1698: optname, &m);
1699: if (prerr == 0)
1700: error = 0;
1701: else if (prerr != ENOPROTOOPT)
1702: error = prerr;
1703: } else if (m != NULL)
1704: (void)m_free(m);
1.160 ad 1705: sounlock(so);
1.142 dyoung 1706: return error;
1.1 cgd 1707: }
1708:
1.14 mycroft 1709: int
1.54 lukem 1710: sogetopt(struct socket *so, int level, int optname, struct mbuf **mp)
1.1 cgd 1711: {
1.54 lukem 1712: struct mbuf *m;
1.160 ad 1713: int error;
1.1 cgd 1714:
1.160 ad 1715: solock(so);
1.1 cgd 1716: if (level != SOL_SOCKET) {
1717: if (so->so_proto && so->so_proto->pr_ctloutput) {
1.160 ad 1718: error = ((*so->so_proto->pr_ctloutput)
1.1 cgd 1719: (PRCO_GETOPT, so, level, optname, mp));
1720: } else
1.160 ad 1721: error = (ENOPROTOOPT);
1.1 cgd 1722: } else {
1723: m = m_get(M_WAIT, MT_SOOPTS);
1.36 perry 1724: m->m_len = sizeof(int);
1.1 cgd 1725:
1726: switch (optname) {
1727:
1728: case SO_LINGER:
1.36 perry 1729: m->m_len = sizeof(struct linger);
1.1 cgd 1730: mtod(m, struct linger *)->l_onoff =
1.131 christos 1731: (so->so_options & SO_LINGER) ? 1 : 0;
1.1 cgd 1732: mtod(m, struct linger *)->l_linger = so->so_linger;
1733: break;
1734:
1735: case SO_USELOOPBACK:
1736: case SO_DONTROUTE:
1737: case SO_DEBUG:
1738: case SO_KEEPALIVE:
1739: case SO_REUSEADDR:
1.15 mycroft 1740: case SO_REUSEPORT:
1.1 cgd 1741: case SO_BROADCAST:
1742: case SO_OOBINLINE:
1.26 thorpej 1743: case SO_TIMESTAMP:
1.131 christos 1744: *mtod(m, int *) = (so->so_options & optname) ? 1 : 0;
1.1 cgd 1745: break;
1746:
1747: case SO_TYPE:
1748: *mtod(m, int *) = so->so_type;
1749: break;
1750:
1751: case SO_ERROR:
1752: *mtod(m, int *) = so->so_error;
1753: so->so_error = 0;
1754: break;
1755:
1756: case SO_SNDBUF:
1757: *mtod(m, int *) = so->so_snd.sb_hiwat;
1758: break;
1759:
1760: case SO_RCVBUF:
1761: *mtod(m, int *) = so->so_rcv.sb_hiwat;
1762: break;
1763:
1764: case SO_SNDLOWAT:
1765: *mtod(m, int *) = so->so_snd.sb_lowat;
1766: break;
1767:
1768: case SO_RCVLOWAT:
1769: *mtod(m, int *) = so->so_rcv.sb_lowat;
1770: break;
1771:
1772: case SO_SNDTIMEO:
1773: case SO_RCVTIMEO:
1774: {
1775: int val = (optname == SO_SNDTIMEO ?
1776: so->so_snd.sb_timeo : so->so_rcv.sb_timeo);
1777:
1778: m->m_len = sizeof(struct timeval);
1779: mtod(m, struct timeval *)->tv_sec = val / hz;
1780: mtod(m, struct timeval *)->tv_usec =
1.27 kleink 1781: (val % hz) * tick;
1.1 cgd 1782: break;
1783: }
1784:
1.107 darrenr 1785: case SO_OVERFLOWED:
1786: *mtod(m, int *) = so->so_rcv.sb_overflowed;
1787: break;
1788:
1.1 cgd 1789: default:
1.160 ad 1790: sounlock(so);
1.1 cgd 1791: (void)m_free(m);
1792: return (ENOPROTOOPT);
1793: }
1794: *mp = m;
1.160 ad 1795: error = 0;
1.1 cgd 1796: }
1.160 ad 1797:
1798: sounlock(so);
1799: return (error);
1.1 cgd 1800: }
1801:
1.14 mycroft 1802: void
1.54 lukem 1803: sohasoutofband(struct socket *so)
1.1 cgd 1804: {
1.153 rmind 1805:
1.90 christos 1806: fownsignal(so->so_pgid, SIGURG, POLL_PRI, POLLPRI|POLLRDBAND, so);
1.153 rmind 1807: selnotify(&so->so_rcv.sb_sel, POLLPRI | POLLRDBAND, 0);
1.1 cgd 1808: }
1.72 jdolecek 1809:
1810: static void
1811: filt_sordetach(struct knote *kn)
1812: {
1813: struct socket *so;
1814:
1.155 ad 1815: so = ((file_t *)kn->kn_obj)->f_data;
1.160 ad 1816: solock(so);
1.73 christos 1817: SLIST_REMOVE(&so->so_rcv.sb_sel.sel_klist, kn, knote, kn_selnext);
1818: if (SLIST_EMPTY(&so->so_rcv.sb_sel.sel_klist))
1.72 jdolecek 1819: so->so_rcv.sb_flags &= ~SB_KNOTE;
1.160 ad 1820: sounlock(so);
1.72 jdolecek 1821: }
1822:
1823: /*ARGSUSED*/
1824: static int
1.129 yamt 1825: filt_soread(struct knote *kn, long hint)
1.72 jdolecek 1826: {
1827: struct socket *so;
1.160 ad 1828: int rv;
1.72 jdolecek 1829:
1.155 ad 1830: so = ((file_t *)kn->kn_obj)->f_data;
1.160 ad 1831: if (hint != NOTE_SUBMIT)
1832: solock(so);
1.72 jdolecek 1833: kn->kn_data = so->so_rcv.sb_cc;
1834: if (so->so_state & SS_CANTRCVMORE) {
1.108 perry 1835: kn->kn_flags |= EV_EOF;
1.72 jdolecek 1836: kn->kn_fflags = so->so_error;
1.160 ad 1837: rv = 1;
1838: } else if (so->so_error) /* temporary udp error */
1839: rv = 1;
1840: else if (kn->kn_sfflags & NOTE_LOWAT)
1841: rv = (kn->kn_data >= kn->kn_sdata);
1842: else
1843: rv = (kn->kn_data >= so->so_rcv.sb_lowat);
1844: if (hint != NOTE_SUBMIT)
1845: sounlock(so);
1846: return rv;
1.72 jdolecek 1847: }
1848:
1849: static void
1850: filt_sowdetach(struct knote *kn)
1851: {
1852: struct socket *so;
1853:
1.155 ad 1854: so = ((file_t *)kn->kn_obj)->f_data;
1.160 ad 1855: solock(so);
1.73 christos 1856: SLIST_REMOVE(&so->so_snd.sb_sel.sel_klist, kn, knote, kn_selnext);
1857: if (SLIST_EMPTY(&so->so_snd.sb_sel.sel_klist))
1.72 jdolecek 1858: so->so_snd.sb_flags &= ~SB_KNOTE;
1.160 ad 1859: sounlock(so);
1.72 jdolecek 1860: }
1861:
1862: /*ARGSUSED*/
1863: static int
1.129 yamt 1864: filt_sowrite(struct knote *kn, long hint)
1.72 jdolecek 1865: {
1866: struct socket *so;
1.160 ad 1867: int rv;
1.72 jdolecek 1868:
1.155 ad 1869: so = ((file_t *)kn->kn_obj)->f_data;
1.160 ad 1870: if (hint != NOTE_SUBMIT)
1871: solock(so);
1.72 jdolecek 1872: kn->kn_data = sbspace(&so->so_snd);
1873: if (so->so_state & SS_CANTSENDMORE) {
1.108 perry 1874: kn->kn_flags |= EV_EOF;
1.72 jdolecek 1875: kn->kn_fflags = so->so_error;
1.160 ad 1876: rv = 1;
1877: } else if (so->so_error) /* temporary udp error */
1878: rv = 1;
1879: else if (((so->so_state & SS_ISCONNECTED) == 0) &&
1.72 jdolecek 1880: (so->so_proto->pr_flags & PR_CONNREQUIRED))
1.160 ad 1881: rv = 0;
1882: else if (kn->kn_sfflags & NOTE_LOWAT)
1883: rv = (kn->kn_data >= kn->kn_sdata);
1884: else
1885: rv = (kn->kn_data >= so->so_snd.sb_lowat);
1886: if (hint != NOTE_SUBMIT)
1887: sounlock(so);
1888: return rv;
1.72 jdolecek 1889: }
1890:
1891: /*ARGSUSED*/
1892: static int
1.129 yamt 1893: filt_solisten(struct knote *kn, long hint)
1.72 jdolecek 1894: {
1895: struct socket *so;
1.160 ad 1896: int rv;
1.72 jdolecek 1897:
1.155 ad 1898: so = ((file_t *)kn->kn_obj)->f_data;
1.72 jdolecek 1899:
1900: /*
1901: * Set kn_data to number of incoming connections, not
1902: * counting partial (incomplete) connections.
1.108 perry 1903: */
1.160 ad 1904: if (hint != NOTE_SUBMIT)
1905: solock(so);
1.72 jdolecek 1906: kn->kn_data = so->so_qlen;
1.160 ad 1907: rv = (kn->kn_data > 0);
1908: if (hint != NOTE_SUBMIT)
1909: sounlock(so);
1910: return rv;
1.72 jdolecek 1911: }
1912:
1913: static const struct filterops solisten_filtops =
1914: { 1, NULL, filt_sordetach, filt_solisten };
1915: static const struct filterops soread_filtops =
1916: { 1, NULL, filt_sordetach, filt_soread };
1917: static const struct filterops sowrite_filtops =
1918: { 1, NULL, filt_sowdetach, filt_sowrite };
1919:
1920: int
1.129 yamt 1921: soo_kqfilter(struct file *fp, struct knote *kn)
1.72 jdolecek 1922: {
1923: struct socket *so;
1924: struct sockbuf *sb;
1925:
1.155 ad 1926: so = ((file_t *)kn->kn_obj)->f_data;
1.160 ad 1927: solock(so);
1.72 jdolecek 1928: switch (kn->kn_filter) {
1929: case EVFILT_READ:
1930: if (so->so_options & SO_ACCEPTCONN)
1931: kn->kn_fop = &solisten_filtops;
1932: else
1933: kn->kn_fop = &soread_filtops;
1934: sb = &so->so_rcv;
1935: break;
1936: case EVFILT_WRITE:
1937: kn->kn_fop = &sowrite_filtops;
1938: sb = &so->so_snd;
1939: break;
1940: default:
1.160 ad 1941: sounlock(so);
1.149 pooka 1942: return (EINVAL);
1.72 jdolecek 1943: }
1.73 christos 1944: SLIST_INSERT_HEAD(&sb->sb_sel.sel_klist, kn, kn_selnext);
1.72 jdolecek 1945: sb->sb_flags |= SB_KNOTE;
1.160 ad 1946: sounlock(so);
1.72 jdolecek 1947: return (0);
1948: }
1949:
1.154 ad 1950: static int
1951: sodopoll(struct socket *so, int events)
1952: {
1953: int revents;
1954:
1955: revents = 0;
1956:
1957: if (events & (POLLIN | POLLRDNORM))
1958: if (soreadable(so))
1959: revents |= events & (POLLIN | POLLRDNORM);
1960:
1961: if (events & (POLLOUT | POLLWRNORM))
1962: if (sowritable(so))
1963: revents |= events & (POLLOUT | POLLWRNORM);
1964:
1965: if (events & (POLLPRI | POLLRDBAND))
1966: if (so->so_oobmark || (so->so_state & SS_RCVATMARK))
1967: revents |= events & (POLLPRI | POLLRDBAND);
1968:
1969: return revents;
1970: }
1971:
1972: int
1973: sopoll(struct socket *so, int events)
1974: {
1975: int revents = 0;
1976:
1.160 ad 1977: #ifndef DIAGNOSTIC
1978: /*
1979: * Do a quick, unlocked check in expectation that the socket
1980: * will be ready for I/O. Don't do this check if DIAGNOSTIC,
1981: * as the solocked() assertions will fail.
1982: */
1.154 ad 1983: if ((revents = sodopoll(so, events)) != 0)
1984: return revents;
1.160 ad 1985: #endif
1.154 ad 1986:
1.160 ad 1987: solock(so);
1.154 ad 1988: if ((revents = sodopoll(so, events)) == 0) {
1989: if (events & (POLLIN | POLLPRI | POLLRDNORM | POLLRDBAND)) {
1990: selrecord(curlwp, &so->so_rcv.sb_sel);
1.160 ad 1991: so->so_rcv.sb_flags |= SB_NOTIFY;
1.154 ad 1992: }
1993:
1994: if (events & (POLLOUT | POLLWRNORM)) {
1995: selrecord(curlwp, &so->so_snd.sb_sel);
1.160 ad 1996: so->so_snd.sb_flags |= SB_NOTIFY;
1.154 ad 1997: }
1998: }
1.160 ad 1999: sounlock(so);
1.154 ad 2000:
2001: return revents;
2002: }
2003:
2004:
1.94 yamt 2005: #include <sys/sysctl.h>
2006:
2007: static int sysctl_kern_somaxkva(SYSCTLFN_PROTO);
2008:
2009: /*
2010: * sysctl helper routine for kern.somaxkva. ensures that the given
2011: * value is not too small.
2012: * (XXX should we maybe make sure it's not too large as well?)
2013: */
2014: static int
2015: sysctl_kern_somaxkva(SYSCTLFN_ARGS)
2016: {
2017: int error, new_somaxkva;
2018: struct sysctlnode node;
2019:
2020: new_somaxkva = somaxkva;
2021: node = *rnode;
2022: node.sysctl_data = &new_somaxkva;
2023: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2024: if (error || newp == NULL)
2025: return (error);
2026:
2027: if (new_somaxkva < (16 * 1024 * 1024)) /* sanity */
2028: return (EINVAL);
2029:
1.136 ad 2030: mutex_enter(&so_pendfree_lock);
1.94 yamt 2031: somaxkva = new_somaxkva;
1.136 ad 2032: cv_broadcast(&socurkva_cv);
2033: mutex_exit(&so_pendfree_lock);
1.94 yamt 2034:
2035: return (error);
2036: }
2037:
2038: SYSCTL_SETUP(sysctl_kern_somaxkva_setup, "sysctl kern.somaxkva setup")
2039: {
2040:
1.97 atatat 2041: sysctl_createv(clog, 0, NULL, NULL,
2042: CTLFLAG_PERMANENT,
2043: CTLTYPE_NODE, "kern", NULL,
2044: NULL, 0, NULL, 0,
2045: CTL_KERN, CTL_EOL);
2046:
2047: sysctl_createv(clog, 0, NULL, NULL,
2048: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.103 atatat 2049: CTLTYPE_INT, "somaxkva",
2050: SYSCTL_DESCR("Maximum amount of kernel memory to be "
2051: "used for socket buffers"),
1.94 yamt 2052: sysctl_kern_somaxkva, 0, NULL, 0,
2053: CTL_KERN, KERN_SOMAXKVA, CTL_EOL);
2054: }
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