Annotation of src/sys/kern/uipc_socket.c, Revision 1.170
1.170 ! tls 1: /* $NetBSD: uipc_socket.c,v 1.169 2008/07/25 22:45:58 dsl 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>
1.170 ! tls 66: __KERNEL_RCSID(0, "$NetBSD: uipc_socket.c,v 1.169 2008/07/25 22:45:58 dsl Exp $");
1.64 thorpej 67:
1.170 ! tls 68: #include "opt_inet.h"
1.64 thorpej 69: #include "opt_sock_counters.h"
70: #include "opt_sosend_loan.h"
1.81 martin 71: #include "opt_mbuftrace.h"
1.84 ragge 72: #include "opt_somaxkva.h"
1.167 ad 73: #include "opt_multiprocessor.h" /* XXX */
1.1 cgd 74:
1.9 mycroft 75: #include <sys/param.h>
76: #include <sys/systm.h>
77: #include <sys/proc.h>
78: #include <sys/file.h>
1.142 dyoung 79: #include <sys/filedesc.h>
1.9 mycroft 80: #include <sys/malloc.h>
81: #include <sys/mbuf.h>
82: #include <sys/domain.h>
83: #include <sys/kernel.h>
84: #include <sys/protosw.h>
85: #include <sys/socket.h>
86: #include <sys/socketvar.h>
1.21 christos 87: #include <sys/signalvar.h>
1.9 mycroft 88: #include <sys/resourcevar.h>
1.72 jdolecek 89: #include <sys/event.h>
1.89 christos 90: #include <sys/poll.h>
1.118 elad 91: #include <sys/kauth.h>
1.136 ad 92: #include <sys/mutex.h>
93: #include <sys/condvar.h>
1.37 thorpej 94:
1.64 thorpej 95: #include <uvm/uvm.h>
96:
1.77 thorpej 97: MALLOC_DEFINE(M_SOOPTS, "soopts", "socket options");
98: MALLOC_DEFINE(M_SONAME, "soname", "socket name");
1.37 thorpej 99:
1.142 dyoung 100: extern const struct fileops socketops;
101:
1.54 lukem 102: extern int somaxconn; /* patchable (XXX sysctl) */
103: int somaxconn = SOMAXCONN;
1.160 ad 104: kmutex_t *softnet_lock;
1.49 jonathan 105:
1.64 thorpej 106: #ifdef SOSEND_COUNTERS
107: #include <sys/device.h>
108:
1.113 thorpej 109: static struct evcnt sosend_loan_big = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64 thorpej 110: NULL, "sosend", "loan big");
1.113 thorpej 111: static struct evcnt sosend_copy_big = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64 thorpej 112: NULL, "sosend", "copy big");
1.113 thorpej 113: static struct evcnt sosend_copy_small = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64 thorpej 114: NULL, "sosend", "copy small");
1.113 thorpej 115: static struct evcnt sosend_kvalimit = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64 thorpej 116: NULL, "sosend", "kva limit");
117:
118: #define SOSEND_COUNTER_INCR(ev) (ev)->ev_count++
119:
1.101 matt 120: EVCNT_ATTACH_STATIC(sosend_loan_big);
121: EVCNT_ATTACH_STATIC(sosend_copy_big);
122: EVCNT_ATTACH_STATIC(sosend_copy_small);
123: EVCNT_ATTACH_STATIC(sosend_kvalimit);
1.64 thorpej 124: #else
125:
126: #define SOSEND_COUNTER_INCR(ev) /* nothing */
127:
128: #endif /* SOSEND_COUNTERS */
129:
1.119 yamt 130: static struct callback_entry sokva_reclaimerentry;
1.1 cgd 131:
1.167 ad 132: #if defined(SOSEND_NO_LOAN) || defined(MULTIPROCESSOR)
1.121 yamt 133: int sock_loan_thresh = -1;
1.71 thorpej 134: #else
1.121 yamt 135: int sock_loan_thresh = 4096;
1.65 thorpej 136: #endif
1.64 thorpej 137:
1.136 ad 138: static kmutex_t so_pendfree_lock;
1.113 thorpej 139: static struct mbuf *so_pendfree;
1.64 thorpej 140:
1.84 ragge 141: #ifndef SOMAXKVA
142: #define SOMAXKVA (16 * 1024 * 1024)
143: #endif
144: int somaxkva = SOMAXKVA;
1.113 thorpej 145: static int socurkva;
1.136 ad 146: static kcondvar_t socurkva_cv;
1.64 thorpej 147:
148: #define SOCK_LOAN_CHUNK 65536
149:
1.117 yamt 150: static size_t sodopendfree(void);
151: static size_t sodopendfreel(void);
1.93 yamt 152:
1.113 thorpej 153: static vsize_t
1.129 yamt 154: sokvareserve(struct socket *so, vsize_t len)
1.80 yamt 155: {
1.98 christos 156: int error;
1.80 yamt 157:
1.136 ad 158: mutex_enter(&so_pendfree_lock);
1.80 yamt 159: while (socurkva + len > somaxkva) {
1.93 yamt 160: size_t freed;
161:
162: /*
163: * try to do pendfree.
164: */
165:
1.117 yamt 166: freed = sodopendfreel();
1.93 yamt 167:
168: /*
169: * if some kva was freed, try again.
170: */
171:
172: if (freed)
1.80 yamt 173: continue;
1.93 yamt 174:
1.80 yamt 175: SOSEND_COUNTER_INCR(&sosend_kvalimit);
1.136 ad 176: error = cv_wait_sig(&socurkva_cv, &so_pendfree_lock);
1.98 christos 177: if (error) {
178: len = 0;
179: break;
180: }
1.80 yamt 181: }
1.93 yamt 182: socurkva += len;
1.136 ad 183: mutex_exit(&so_pendfree_lock);
1.98 christos 184: return len;
1.95 yamt 185: }
186:
1.113 thorpej 187: static void
1.95 yamt 188: sokvaunreserve(vsize_t len)
189: {
190:
1.136 ad 191: mutex_enter(&so_pendfree_lock);
1.95 yamt 192: socurkva -= len;
1.136 ad 193: cv_broadcast(&socurkva_cv);
194: mutex_exit(&so_pendfree_lock);
1.95 yamt 195: }
196:
197: /*
198: * sokvaalloc: allocate kva for loan.
199: */
200:
201: vaddr_t
202: sokvaalloc(vsize_t len, struct socket *so)
203: {
204: vaddr_t lva;
205:
206: /*
207: * reserve kva.
208: */
209:
1.98 christos 210: if (sokvareserve(so, len) == 0)
211: return 0;
1.93 yamt 212:
213: /*
214: * allocate kva.
215: */
1.80 yamt 216:
1.109 yamt 217: lva = uvm_km_alloc(kernel_map, len, 0, UVM_KMF_VAONLY | UVM_KMF_WAITVA);
1.95 yamt 218: if (lva == 0) {
219: sokvaunreserve(len);
1.80 yamt 220: return (0);
1.95 yamt 221: }
1.80 yamt 222:
223: return lva;
224: }
225:
1.93 yamt 226: /*
227: * sokvafree: free kva for loan.
228: */
229:
1.80 yamt 230: void
231: sokvafree(vaddr_t sva, vsize_t len)
232: {
1.93 yamt 233:
234: /*
235: * free kva.
236: */
1.80 yamt 237:
1.109 yamt 238: uvm_km_free(kernel_map, sva, len, UVM_KMF_VAONLY);
1.93 yamt 239:
240: /*
241: * unreserve kva.
242: */
243:
1.95 yamt 244: sokvaunreserve(len);
1.80 yamt 245: }
246:
1.64 thorpej 247: static void
1.134 christos 248: sodoloanfree(struct vm_page **pgs, void *buf, size_t size)
1.64 thorpej 249: {
1.156 yamt 250: vaddr_t sva, eva;
1.64 thorpej 251: vsize_t len;
1.156 yamt 252: int npgs;
253:
254: KASSERT(pgs != NULL);
1.64 thorpej 255:
256: eva = round_page((vaddr_t) buf + size);
257: sva = trunc_page((vaddr_t) buf);
258: len = eva - sva;
259: npgs = len >> PAGE_SHIFT;
260:
261: pmap_kremove(sva, len);
262: pmap_update(pmap_kernel());
263: uvm_unloan(pgs, npgs, UVM_LOAN_TOPAGE);
1.80 yamt 264: sokvafree(sva, len);
1.64 thorpej 265: }
266:
267: static size_t
1.152 matt 268: sodopendfree(void)
1.64 thorpej 269: {
1.93 yamt 270: size_t rv;
1.64 thorpej 271:
1.160 ad 272: if (__predict_true(so_pendfree == NULL))
273: return 0;
274:
1.136 ad 275: mutex_enter(&so_pendfree_lock);
1.117 yamt 276: rv = sodopendfreel();
1.136 ad 277: mutex_exit(&so_pendfree_lock);
1.93 yamt 278:
279: return rv;
280: }
281:
282: /*
283: * sodopendfreel: free mbufs on "pendfree" list.
1.136 ad 284: * unlock and relock so_pendfree_lock when freeing mbufs.
1.93 yamt 285: *
1.136 ad 286: * => called with so_pendfree_lock held.
1.93 yamt 287: */
288:
289: static size_t
1.152 matt 290: sodopendfreel(void)
1.93 yamt 291: {
1.137 ad 292: struct mbuf *m, *next;
1.93 yamt 293: size_t rv = 0;
294:
1.136 ad 295: KASSERT(mutex_owned(&so_pendfree_lock));
1.64 thorpej 296:
1.137 ad 297: while (so_pendfree != NULL) {
1.64 thorpej 298: m = so_pendfree;
1.93 yamt 299: so_pendfree = NULL;
1.136 ad 300: mutex_exit(&so_pendfree_lock);
1.93 yamt 301:
302: for (; m != NULL; m = next) {
303: next = m->m_next;
1.156 yamt 304: KASSERT((~m->m_flags & (M_EXT|M_EXT_PAGES)) == 0);
305: KASSERT(m->m_ext.ext_refcnt == 0);
1.93 yamt 306:
307: rv += m->m_ext.ext_size;
1.156 yamt 308: sodoloanfree(m->m_ext.ext_pgs, m->m_ext.ext_buf,
1.93 yamt 309: m->m_ext.ext_size);
1.145 ad 310: pool_cache_put(mb_cache, m);
1.93 yamt 311: }
1.64 thorpej 312:
1.136 ad 313: mutex_enter(&so_pendfree_lock);
1.64 thorpej 314: }
315:
316: return (rv);
317: }
318:
1.80 yamt 319: void
1.134 christos 320: soloanfree(struct mbuf *m, void *buf, size_t size, void *arg)
1.64 thorpej 321: {
322:
1.156 yamt 323: KASSERT(m != NULL);
1.64 thorpej 324:
1.93 yamt 325: /*
326: * postpone freeing mbuf.
327: *
328: * we can't do it in interrupt context
329: * because we need to put kva back to kernel_map.
330: */
331:
1.136 ad 332: mutex_enter(&so_pendfree_lock);
1.92 yamt 333: m->m_next = so_pendfree;
334: so_pendfree = m;
1.136 ad 335: cv_broadcast(&socurkva_cv);
336: mutex_exit(&so_pendfree_lock);
1.64 thorpej 337: }
338:
339: static long
340: sosend_loan(struct socket *so, struct uio *uio, struct mbuf *m, long space)
341: {
342: struct iovec *iov = uio->uio_iov;
343: vaddr_t sva, eva;
344: vsize_t len;
1.156 yamt 345: vaddr_t lva;
346: int npgs, error;
347: vaddr_t va;
348: int i;
1.64 thorpej 349:
1.116 yamt 350: if (VMSPACE_IS_KERNEL_P(uio->uio_vmspace))
1.64 thorpej 351: return (0);
352:
353: if (iov->iov_len < (size_t) space)
354: space = iov->iov_len;
355: if (space > SOCK_LOAN_CHUNK)
356: space = SOCK_LOAN_CHUNK;
357:
358: eva = round_page((vaddr_t) iov->iov_base + space);
359: sva = trunc_page((vaddr_t) iov->iov_base);
360: len = eva - sva;
361: npgs = len >> PAGE_SHIFT;
362:
1.79 thorpej 363: KASSERT(npgs <= M_EXT_MAXPAGES);
364:
1.80 yamt 365: lva = sokvaalloc(len, so);
1.64 thorpej 366: if (lva == 0)
1.80 yamt 367: return 0;
1.64 thorpej 368:
1.116 yamt 369: error = uvm_loan(&uio->uio_vmspace->vm_map, sva, len,
1.79 thorpej 370: m->m_ext.ext_pgs, UVM_LOAN_TOPAGE);
1.64 thorpej 371: if (error) {
1.80 yamt 372: sokvafree(lva, len);
1.64 thorpej 373: return (0);
374: }
375:
376: for (i = 0, va = lva; i < npgs; i++, va += PAGE_SIZE)
1.79 thorpej 377: pmap_kenter_pa(va, VM_PAGE_TO_PHYS(m->m_ext.ext_pgs[i]),
378: VM_PROT_READ);
1.64 thorpej 379: pmap_update(pmap_kernel());
380:
381: lva += (vaddr_t) iov->iov_base & PAGE_MASK;
382:
1.134 christos 383: MEXTADD(m, (void *) lva, space, M_MBUF, soloanfree, so);
1.79 thorpej 384: m->m_flags |= M_EXT_PAGES | M_EXT_ROMAP;
1.64 thorpej 385:
386: uio->uio_resid -= space;
387: /* uio_offset not updated, not set/used for write(2) */
1.134 christos 388: uio->uio_iov->iov_base = (char *)uio->uio_iov->iov_base + space;
1.64 thorpej 389: uio->uio_iov->iov_len -= space;
390: if (uio->uio_iov->iov_len == 0) {
391: uio->uio_iov++;
392: uio->uio_iovcnt--;
393: }
394:
395: return (space);
396: }
397:
1.119 yamt 398: static int
1.129 yamt 399: sokva_reclaim_callback(struct callback_entry *ce, void *obj, void *arg)
1.119 yamt 400: {
401:
402: KASSERT(ce == &sokva_reclaimerentry);
403: KASSERT(obj == NULL);
404:
405: sodopendfree();
406: if (!vm_map_starved_p(kernel_map)) {
407: return CALLBACK_CHAIN_ABORT;
408: }
409: return CALLBACK_CHAIN_CONTINUE;
410: }
411:
1.142 dyoung 412: struct mbuf *
1.147 dyoung 413: getsombuf(struct socket *so, int type)
1.142 dyoung 414: {
415: struct mbuf *m;
416:
1.147 dyoung 417: m = m_get(M_WAIT, type);
1.142 dyoung 418: MCLAIM(m, so->so_mowner);
419: return m;
420: }
421:
422: struct mbuf *
423: m_intopt(struct socket *so, int val)
424: {
425: struct mbuf *m;
426:
1.147 dyoung 427: m = getsombuf(so, MT_SOOPTS);
1.142 dyoung 428: m->m_len = sizeof(int);
429: *mtod(m, int *) = val;
430: return m;
431: }
432:
1.119 yamt 433: void
434: soinit(void)
435: {
436:
1.148 ad 437: mutex_init(&so_pendfree_lock, MUTEX_DEFAULT, IPL_VM);
1.160 ad 438: softnet_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NONE);
1.136 ad 439: cv_init(&socurkva_cv, "sokva");
1.166 ad 440: soinit2();
1.136 ad 441:
1.119 yamt 442: /* Set the initial adjusted socket buffer size. */
443: if (sb_max_set(sb_max))
444: panic("bad initial sb_max value: %lu", sb_max);
445:
446: callback_register(&vm_map_to_kernel(kernel_map)->vmk_reclaim_callback,
447: &sokva_reclaimerentry, NULL, sokva_reclaim_callback);
448: }
449:
1.1 cgd 450: /*
451: * Socket operation routines.
452: * These routines are called by the routines in
453: * sys_socket.c or from a system process, and
454: * implement the semantics of socket operations by
455: * switching out to the protocol specific routines.
456: */
457: /*ARGSUSED*/
1.3 andrew 458: int
1.160 ad 459: socreate(int dom, struct socket **aso, int type, int proto, struct lwp *l,
460: struct socket *lockso)
1.1 cgd 461: {
1.99 matt 462: const struct protosw *prp;
1.54 lukem 463: struct socket *so;
1.115 yamt 464: uid_t uid;
1.160 ad 465: int error;
466: kmutex_t *lock;
1.1 cgd 467:
1.132 elad 468: error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET,
469: KAUTH_REQ_NETWORK_SOCKET_OPEN, KAUTH_ARG(dom), KAUTH_ARG(type),
470: KAUTH_ARG(proto));
1.140 dyoung 471: if (error != 0)
472: return error;
1.127 elad 473:
1.1 cgd 474: if (proto)
475: prp = pffindproto(dom, proto, type);
476: else
477: prp = pffindtype(dom, type);
1.140 dyoung 478: if (prp == NULL) {
1.120 ginsbach 479: /* no support for domain */
480: if (pffinddomain(dom) == 0)
1.140 dyoung 481: return EAFNOSUPPORT;
1.120 ginsbach 482: /* no support for socket type */
483: if (proto == 0 && type != 0)
1.140 dyoung 484: return EPROTOTYPE;
485: return EPROTONOSUPPORT;
1.120 ginsbach 486: }
1.140 dyoung 487: if (prp->pr_usrreq == NULL)
488: return EPROTONOSUPPORT;
1.1 cgd 489: if (prp->pr_type != type)
1.140 dyoung 490: return EPROTOTYPE;
1.160 ad 491:
492: so = soget(true);
1.1 cgd 493: so->so_type = type;
494: so->so_proto = prp;
1.33 matt 495: so->so_send = sosend;
496: so->so_receive = soreceive;
1.78 matt 497: #ifdef MBUFTRACE
498: so->so_rcv.sb_mowner = &prp->pr_domain->dom_mowner;
499: so->so_snd.sb_mowner = &prp->pr_domain->dom_mowner;
500: so->so_mowner = &prp->pr_domain->dom_mowner;
501: #endif
1.138 rmind 502: uid = kauth_cred_geteuid(l->l_cred);
1.115 yamt 503: so->so_uidinfo = uid_find(uid);
1.168 yamt 504: so->so_egid = kauth_cred_getegid(l->l_cred);
505: so->so_cpid = l->l_proc->p_pid;
1.160 ad 506: if (lockso != NULL) {
507: /* Caller wants us to share a lock. */
508: lock = lockso->so_lock;
509: so->so_lock = lock;
510: mutex_obj_hold(lock);
511: mutex_enter(lock);
512: } else {
513: /* Lock assigned and taken during PRU_ATTACH. */
514: }
1.140 dyoung 515: error = (*prp->pr_usrreq)(so, PRU_ATTACH, NULL,
516: (struct mbuf *)(long)proto, NULL, l);
1.160 ad 517: KASSERT(solocked(so));
1.140 dyoung 518: if (error != 0) {
1.1 cgd 519: so->so_state |= SS_NOFDREF;
520: sofree(so);
1.140 dyoung 521: return error;
1.1 cgd 522: }
1.160 ad 523: sounlock(so);
1.1 cgd 524: *aso = so;
1.140 dyoung 525: return 0;
1.1 cgd 526: }
527:
1.142 dyoung 528: /* On success, write file descriptor to fdout and return zero. On
529: * failure, return non-zero; *fdout will be undefined.
530: */
531: int
532: fsocreate(int domain, struct socket **sop, int type, int protocol,
533: struct lwp *l, int *fdout)
534: {
535: struct socket *so;
536: struct file *fp;
537: int fd, error;
538:
1.155 ad 539: if ((error = fd_allocfile(&fp, &fd)) != 0)
1.142 dyoung 540: return (error);
541: fp->f_flag = FREAD|FWRITE;
542: fp->f_type = DTYPE_SOCKET;
543: fp->f_ops = &socketops;
1.160 ad 544: error = socreate(domain, &so, type, protocol, l, NULL);
1.142 dyoung 545: if (error != 0) {
1.155 ad 546: fd_abort(curproc, fp, fd);
1.142 dyoung 547: } else {
548: if (sop != NULL)
549: *sop = so;
550: fp->f_data = so;
1.155 ad 551: fd_affix(curproc, fp, fd);
1.142 dyoung 552: *fdout = fd;
553: }
554: return error;
555: }
556:
1.3 andrew 557: int
1.114 christos 558: sobind(struct socket *so, struct mbuf *nam, struct lwp *l)
1.1 cgd 559: {
1.160 ad 560: int error;
1.1 cgd 561:
1.160 ad 562: solock(so);
1.140 dyoung 563: error = (*so->so_proto->pr_usrreq)(so, PRU_BIND, NULL, nam, NULL, l);
1.160 ad 564: sounlock(so);
1.140 dyoung 565: return error;
1.1 cgd 566: }
567:
1.3 andrew 568: int
1.150 elad 569: solisten(struct socket *so, int backlog, struct lwp *l)
1.1 cgd 570: {
1.160 ad 571: int error;
1.1 cgd 572:
1.160 ad 573: solock(so);
1.158 ad 574: if ((so->so_state & (SS_ISCONNECTED | SS_ISCONNECTING |
1.163 ad 575: SS_ISDISCONNECTING)) != 0) {
576: sounlock(so);
1.158 ad 577: return (EOPNOTSUPP);
1.163 ad 578: }
1.140 dyoung 579: error = (*so->so_proto->pr_usrreq)(so, PRU_LISTEN, NULL,
1.150 elad 580: NULL, NULL, l);
1.140 dyoung 581: if (error != 0) {
1.160 ad 582: sounlock(so);
1.140 dyoung 583: return error;
1.1 cgd 584: }
1.63 matt 585: if (TAILQ_EMPTY(&so->so_q))
1.1 cgd 586: so->so_options |= SO_ACCEPTCONN;
587: if (backlog < 0)
588: backlog = 0;
1.49 jonathan 589: so->so_qlimit = min(backlog, somaxconn);
1.160 ad 590: sounlock(so);
1.140 dyoung 591: return 0;
1.1 cgd 592: }
593:
1.21 christos 594: void
1.54 lukem 595: sofree(struct socket *so)
1.1 cgd 596: {
1.161 ad 597: u_int refs;
1.1 cgd 598:
1.160 ad 599: KASSERT(solocked(so));
600:
601: if (so->so_pcb || (so->so_state & SS_NOFDREF) == 0) {
602: sounlock(so);
1.1 cgd 603: return;
1.160 ad 604: }
1.43 mycroft 605: if (so->so_head) {
606: /*
607: * We must not decommission a socket that's on the accept(2)
608: * queue. If we do, then accept(2) may hang after select(2)
609: * indicated that the listening socket was ready.
610: */
1.160 ad 611: if (!soqremque(so, 0)) {
612: sounlock(so);
1.43 mycroft 613: return;
1.160 ad 614: }
1.43 mycroft 615: }
1.98 christos 616: if (so->so_rcv.sb_hiwat)
1.110 christos 617: (void)chgsbsize(so->so_uidinfo, &so->so_rcv.sb_hiwat, 0,
1.98 christos 618: RLIM_INFINITY);
619: if (so->so_snd.sb_hiwat)
1.110 christos 620: (void)chgsbsize(so->so_uidinfo, &so->so_snd.sb_hiwat, 0,
1.98 christos 621: RLIM_INFINITY);
622: sbrelease(&so->so_snd, so);
1.160 ad 623: KASSERT(!cv_has_waiters(&so->so_cv));
624: KASSERT(!cv_has_waiters(&so->so_rcv.sb_cv));
625: KASSERT(!cv_has_waiters(&so->so_snd.sb_cv));
1.1 cgd 626: sorflush(so);
1.161 ad 627: refs = so->so_aborting; /* XXX */
1.170 ! tls 628: #ifdef INET
! 629: /* remove acccept filter if one is present. */
! 630: if (so->so_accf != NULL)
! 631: do_setopt_accept_filter(so, NULL);
! 632: #endif
1.160 ad 633: sounlock(so);
1.161 ad 634: if (refs == 0) /* XXX */
635: soput(so);
1.1 cgd 636: }
637:
638: /*
639: * Close a socket on last file table reference removal.
640: * Initiate disconnect if connected.
641: * Free socket when disconnect complete.
642: */
1.3 andrew 643: int
1.54 lukem 644: soclose(struct socket *so)
1.1 cgd 645: {
1.54 lukem 646: struct socket *so2;
1.160 ad 647: int error;
648: int error2;
1.1 cgd 649:
1.54 lukem 650: error = 0;
1.160 ad 651: solock(so);
1.1 cgd 652: if (so->so_options & SO_ACCEPTCONN) {
1.160 ad 653: do {
1.169 dsl 654: while ((so2 = TAILQ_FIRST(&so->so_q0)) != 0) {
1.160 ad 655: KASSERT(solocked2(so, so2));
656: (void) soqremque(so2, 0);
657: /* soabort drops the lock. */
658: (void) soabort(so2);
659: solock(so);
660: }
1.169 dsl 661: while ((so2 = TAILQ_FIRST(&so->so_q)) != 0) {
1.160 ad 662: KASSERT(solocked2(so, so2));
663: (void) soqremque(so2, 1);
664: /* soabort drops the lock. */
665: (void) soabort(so2);
666: solock(so);
667: }
1.169 dsl 668: } while (!TAILQ_EMPTY(&so->so_q0));
1.1 cgd 669: }
670: if (so->so_pcb == 0)
671: goto discard;
672: if (so->so_state & SS_ISCONNECTED) {
673: if ((so->so_state & SS_ISDISCONNECTING) == 0) {
674: error = sodisconnect(so);
675: if (error)
676: goto drop;
677: }
678: if (so->so_options & SO_LINGER) {
1.151 ad 679: if ((so->so_state & SS_ISDISCONNECTING) && so->so_nbio)
1.1 cgd 680: goto drop;
1.21 christos 681: while (so->so_state & SS_ISCONNECTED) {
1.160 ad 682: error = sowait(so, so->so_linger * hz);
1.21 christos 683: if (error)
1.1 cgd 684: break;
1.21 christos 685: }
1.1 cgd 686: }
687: }
1.54 lukem 688: drop:
1.1 cgd 689: if (so->so_pcb) {
1.160 ad 690: error2 = (*so->so_proto->pr_usrreq)(so, PRU_DETACH,
1.140 dyoung 691: NULL, NULL, NULL, NULL);
1.1 cgd 692: if (error == 0)
693: error = error2;
694: }
1.54 lukem 695: discard:
1.1 cgd 696: if (so->so_state & SS_NOFDREF)
697: panic("soclose: NOFDREF");
698: so->so_state |= SS_NOFDREF;
699: sofree(so);
700: return (error);
701: }
702:
703: /*
1.160 ad 704: * Must be called with the socket locked.. Will return with it unlocked.
1.1 cgd 705: */
1.3 andrew 706: int
1.54 lukem 707: soabort(struct socket *so)
1.1 cgd 708: {
1.161 ad 709: u_int refs;
1.139 yamt 710: int error;
1.160 ad 711:
712: KASSERT(solocked(so));
713: KASSERT(so->so_head == NULL);
1.1 cgd 714:
1.161 ad 715: so->so_aborting++; /* XXX */
1.140 dyoung 716: error = (*so->so_proto->pr_usrreq)(so, PRU_ABORT, NULL,
717: NULL, NULL, NULL);
1.161 ad 718: refs = --so->so_aborting; /* XXX */
1.164 drochner 719: if (error || (refs == 0)) {
1.139 yamt 720: sofree(so);
1.160 ad 721: } else {
722: sounlock(so);
1.139 yamt 723: }
724: return error;
1.1 cgd 725: }
726:
1.3 andrew 727: int
1.54 lukem 728: soaccept(struct socket *so, struct mbuf *nam)
1.1 cgd 729: {
1.160 ad 730: int error;
731:
732: KASSERT(solocked(so));
1.1 cgd 733:
1.54 lukem 734: error = 0;
1.1 cgd 735: if ((so->so_state & SS_NOFDREF) == 0)
736: panic("soaccept: !NOFDREF");
737: so->so_state &= ~SS_NOFDREF;
1.55 thorpej 738: if ((so->so_state & SS_ISDISCONNECTED) == 0 ||
739: (so->so_proto->pr_flags & PR_ABRTACPTDIS) == 0)
1.41 mycroft 740: error = (*so->so_proto->pr_usrreq)(so, PRU_ACCEPT,
1.140 dyoung 741: NULL, nam, NULL, NULL);
1.41 mycroft 742: else
1.53 itojun 743: error = ECONNABORTED;
1.52 itojun 744:
1.1 cgd 745: return (error);
746: }
747:
1.3 andrew 748: int
1.114 christos 749: soconnect(struct socket *so, struct mbuf *nam, struct lwp *l)
1.1 cgd 750: {
1.160 ad 751: int error;
752:
753: KASSERT(solocked(so));
1.1 cgd 754:
755: if (so->so_options & SO_ACCEPTCONN)
756: return (EOPNOTSUPP);
757: /*
758: * If protocol is connection-based, can only connect once.
759: * Otherwise, if connected, try to disconnect first.
760: * This allows user to disconnect by connecting to, e.g.,
761: * a null address.
762: */
763: if (so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING) &&
764: ((so->so_proto->pr_flags & PR_CONNREQUIRED) ||
765: (error = sodisconnect(so))))
766: error = EISCONN;
767: else
768: error = (*so->so_proto->pr_usrreq)(so, PRU_CONNECT,
1.140 dyoung 769: NULL, nam, NULL, l);
1.1 cgd 770: return (error);
771: }
772:
1.3 andrew 773: int
1.54 lukem 774: soconnect2(struct socket *so1, struct socket *so2)
1.1 cgd 775: {
1.160 ad 776: int error;
777:
778: KASSERT(solocked2(so1, so2));
1.1 cgd 779:
1.22 mycroft 780: error = (*so1->so_proto->pr_usrreq)(so1, PRU_CONNECT2,
1.140 dyoung 781: NULL, (struct mbuf *)so2, NULL, NULL);
1.1 cgd 782: return (error);
783: }
784:
1.3 andrew 785: int
1.54 lukem 786: sodisconnect(struct socket *so)
1.1 cgd 787: {
1.160 ad 788: int error;
789:
790: KASSERT(solocked(so));
1.1 cgd 791:
792: if ((so->so_state & SS_ISCONNECTED) == 0) {
793: error = ENOTCONN;
1.160 ad 794: } else if (so->so_state & SS_ISDISCONNECTING) {
1.1 cgd 795: error = EALREADY;
1.160 ad 796: } else {
797: error = (*so->so_proto->pr_usrreq)(so, PRU_DISCONNECT,
798: NULL, NULL, NULL, NULL);
1.1 cgd 799: }
1.117 yamt 800: sodopendfree();
1.1 cgd 801: return (error);
802: }
803:
1.15 mycroft 804: #define SBLOCKWAIT(f) (((f) & MSG_DONTWAIT) ? M_NOWAIT : M_WAITOK)
1.1 cgd 805: /*
806: * Send on a socket.
807: * If send must go all at once and message is larger than
808: * send buffering, then hard error.
809: * Lock against other senders.
810: * If must go all at once and not enough room now, then
811: * inform user that this would block and do nothing.
812: * Otherwise, if nonblocking, send as much as possible.
813: * The data to be sent is described by "uio" if nonzero,
814: * otherwise by the mbuf chain "top" (which must be null
815: * if uio is not). Data provided in mbuf chain must be small
816: * enough to send all at once.
817: *
818: * Returns nonzero on error, timeout or signal; callers
819: * must check for short counts if EINTR/ERESTART are returned.
820: * Data and control buffers are freed on return.
821: */
1.3 andrew 822: int
1.54 lukem 823: sosend(struct socket *so, struct mbuf *addr, struct uio *uio, struct mbuf *top,
1.114 christos 824: struct mbuf *control, int flags, struct lwp *l)
1.1 cgd 825: {
1.54 lukem 826: struct mbuf **mp, *m;
1.114 christos 827: struct proc *p;
1.58 jdolecek 828: long space, len, resid, clen, mlen;
829: int error, s, dontroute, atomic;
1.54 lukem 830:
1.114 christos 831: p = l->l_proc;
1.117 yamt 832: sodopendfree();
1.160 ad 833: clen = 0;
1.64 thorpej 834:
1.160 ad 835: /*
836: * solock() provides atomicity of access. splsoftnet() prevents
837: * protocol processing soft interrupts from interrupting us and
838: * blocking (expensive).
839: */
840: s = splsoftnet();
841: solock(so);
1.54 lukem 842: atomic = sosendallatonce(so) || top;
1.1 cgd 843: if (uio)
844: resid = uio->uio_resid;
845: else
846: resid = top->m_pkthdr.len;
1.7 cgd 847: /*
848: * In theory resid should be unsigned.
849: * However, space must be signed, as it might be less than 0
850: * if we over-committed, and we must use a signed comparison
851: * of space and resid. On the other hand, a negative resid
852: * causes us to loop sending 0-length segments to the protocol.
853: */
1.29 mycroft 854: if (resid < 0) {
855: error = EINVAL;
856: goto out;
857: }
1.1 cgd 858: dontroute =
859: (flags & MSG_DONTROUTE) && (so->so_options & SO_DONTROUTE) == 0 &&
860: (so->so_proto->pr_flags & PR_ATOMIC);
1.165 christos 861: l->l_ru.ru_msgsnd++;
1.1 cgd 862: if (control)
863: clen = control->m_len;
1.54 lukem 864: restart:
1.21 christos 865: if ((error = sblock(&so->so_snd, SBLOCKWAIT(flags))) != 0)
1.1 cgd 866: goto out;
867: do {
1.160 ad 868: if (so->so_state & SS_CANTSENDMORE) {
869: error = EPIPE;
870: goto release;
871: }
1.48 thorpej 872: if (so->so_error) {
873: error = so->so_error;
874: so->so_error = 0;
875: goto release;
876: }
1.1 cgd 877: if ((so->so_state & SS_ISCONNECTED) == 0) {
878: if (so->so_proto->pr_flags & PR_CONNREQUIRED) {
879: if ((so->so_state & SS_ISCONFIRMING) == 0 &&
1.160 ad 880: !(resid == 0 && clen != 0)) {
881: error = ENOTCONN;
882: goto release;
883: }
884: } else if (addr == 0) {
885: error = EDESTADDRREQ;
886: goto release;
887: }
1.1 cgd 888: }
889: space = sbspace(&so->so_snd);
890: if (flags & MSG_OOB)
891: space += 1024;
1.21 christos 892: if ((atomic && resid > so->so_snd.sb_hiwat) ||
1.160 ad 893: clen > so->so_snd.sb_hiwat) {
894: error = EMSGSIZE;
895: goto release;
896: }
1.96 mycroft 897: if (space < resid + clen &&
1.1 cgd 898: (atomic || space < so->so_snd.sb_lowat || space < clen)) {
1.160 ad 899: if (so->so_nbio) {
900: error = EWOULDBLOCK;
901: goto release;
902: }
1.1 cgd 903: sbunlock(&so->so_snd);
904: error = sbwait(&so->so_snd);
905: if (error)
906: goto out;
907: goto restart;
908: }
909: mp = ⊤
910: space -= clen;
911: do {
1.45 tv 912: if (uio == NULL) {
913: /*
914: * Data is prepackaged in "top".
915: */
916: resid = 0;
917: if (flags & MSG_EOR)
918: top->m_flags |= M_EOR;
919: } else do {
1.160 ad 920: sounlock(so);
921: splx(s);
1.144 dyoung 922: if (top == NULL) {
1.78 matt 923: m = m_gethdr(M_WAIT, MT_DATA);
1.45 tv 924: mlen = MHLEN;
925: m->m_pkthdr.len = 0;
1.140 dyoung 926: m->m_pkthdr.rcvif = NULL;
1.45 tv 927: } else {
1.78 matt 928: m = m_get(M_WAIT, MT_DATA);
1.45 tv 929: mlen = MLEN;
930: }
1.78 matt 931: MCLAIM(m, so->so_snd.sb_mowner);
1.121 yamt 932: if (sock_loan_thresh >= 0 &&
933: uio->uio_iov->iov_len >= sock_loan_thresh &&
934: space >= sock_loan_thresh &&
1.64 thorpej 935: (len = sosend_loan(so, uio, m,
936: space)) != 0) {
937: SOSEND_COUNTER_INCR(&sosend_loan_big);
938: space -= len;
939: goto have_data;
940: }
1.45 tv 941: if (resid >= MINCLSIZE && space >= MCLBYTES) {
1.64 thorpej 942: SOSEND_COUNTER_INCR(&sosend_copy_big);
1.78 matt 943: m_clget(m, M_WAIT);
1.45 tv 944: if ((m->m_flags & M_EXT) == 0)
945: goto nopages;
946: mlen = MCLBYTES;
947: if (atomic && top == 0) {
1.58 jdolecek 948: len = lmin(MCLBYTES - max_hdr,
1.54 lukem 949: resid);
1.45 tv 950: m->m_data += max_hdr;
951: } else
1.58 jdolecek 952: len = lmin(MCLBYTES, resid);
1.45 tv 953: space -= len;
954: } else {
1.64 thorpej 955: nopages:
956: SOSEND_COUNTER_INCR(&sosend_copy_small);
1.58 jdolecek 957: len = lmin(lmin(mlen, resid), space);
1.45 tv 958: space -= len;
959: /*
960: * For datagram protocols, leave room
961: * for protocol headers in first mbuf.
962: */
963: if (atomic && top == 0 && len < mlen)
964: MH_ALIGN(m, len);
965: }
1.144 dyoung 966: error = uiomove(mtod(m, void *), (int)len, uio);
1.64 thorpej 967: have_data:
1.45 tv 968: resid = uio->uio_resid;
969: m->m_len = len;
970: *mp = m;
971: top->m_pkthdr.len += len;
1.160 ad 972: s = splsoftnet();
973: solock(so);
1.144 dyoung 974: if (error != 0)
1.45 tv 975: goto release;
976: mp = &m->m_next;
977: if (resid <= 0) {
978: if (flags & MSG_EOR)
979: top->m_flags |= M_EOR;
980: break;
981: }
982: } while (space > 0 && atomic);
1.108 perry 983:
1.160 ad 984: if (so->so_state & SS_CANTSENDMORE) {
985: error = EPIPE;
986: goto release;
987: }
1.45 tv 988: if (dontroute)
989: so->so_options |= SO_DONTROUTE;
990: if (resid > 0)
991: so->so_state |= SS_MORETOCOME;
1.46 sommerfe 992: error = (*so->so_proto->pr_usrreq)(so,
993: (flags & MSG_OOB) ? PRU_SENDOOB : PRU_SEND,
1.160 ad 994: top, addr, control, curlwp);
1.45 tv 995: if (dontroute)
996: so->so_options &= ~SO_DONTROUTE;
997: if (resid > 0)
998: so->so_state &= ~SS_MORETOCOME;
999: clen = 0;
1.144 dyoung 1000: control = NULL;
1001: top = NULL;
1.45 tv 1002: mp = ⊤
1.144 dyoung 1003: if (error != 0)
1.1 cgd 1004: goto release;
1005: } while (resid && space > 0);
1006: } while (resid);
1007:
1.54 lukem 1008: release:
1.1 cgd 1009: sbunlock(&so->so_snd);
1.54 lukem 1010: out:
1.160 ad 1011: sounlock(so);
1012: splx(s);
1.1 cgd 1013: if (top)
1014: m_freem(top);
1015: if (control)
1016: m_freem(control);
1017: return (error);
1018: }
1019:
1020: /*
1.159 ad 1021: * Following replacement or removal of the first mbuf on the first
1022: * mbuf chain of a socket buffer, push necessary state changes back
1023: * into the socket buffer so that other consumers see the values
1024: * consistently. 'nextrecord' is the callers locally stored value of
1025: * the original value of sb->sb_mb->m_nextpkt which must be restored
1026: * when the lead mbuf changes. NOTE: 'nextrecord' may be NULL.
1027: */
1028: static void
1029: sbsync(struct sockbuf *sb, struct mbuf *nextrecord)
1030: {
1031:
1.160 ad 1032: KASSERT(solocked(sb->sb_so));
1033:
1.159 ad 1034: /*
1035: * First, update for the new value of nextrecord. If necessary,
1036: * make it the first record.
1037: */
1038: if (sb->sb_mb != NULL)
1039: sb->sb_mb->m_nextpkt = nextrecord;
1040: else
1041: sb->sb_mb = nextrecord;
1042:
1043: /*
1044: * Now update any dependent socket buffer fields to reflect
1045: * the new state. This is an inline of SB_EMPTY_FIXUP, with
1046: * the addition of a second clause that takes care of the
1047: * case where sb_mb has been updated, but remains the last
1048: * record.
1049: */
1050: if (sb->sb_mb == NULL) {
1051: sb->sb_mbtail = NULL;
1052: sb->sb_lastrecord = NULL;
1053: } else if (sb->sb_mb->m_nextpkt == NULL)
1054: sb->sb_lastrecord = sb->sb_mb;
1055: }
1056:
1057: /*
1.1 cgd 1058: * Implement receive operations on a socket.
1059: * We depend on the way that records are added to the sockbuf
1060: * by sbappend*. In particular, each record (mbufs linked through m_next)
1061: * must begin with an address if the protocol so specifies,
1062: * followed by an optional mbuf or mbufs containing ancillary data,
1063: * and then zero or more mbufs of data.
1064: * In order to avoid blocking network interrupts for the entire time here,
1065: * we splx() while doing the actual copy to user space.
1066: * Although the sockbuf is locked, new data may still be appended,
1067: * and thus we must maintain consistency of the sockbuf during that time.
1068: *
1069: * The caller may receive the data as a single mbuf chain by supplying
1070: * an mbuf **mp0 for use in returning the chain. The uio is then used
1071: * only for the count in uio_resid.
1072: */
1.3 andrew 1073: int
1.54 lukem 1074: soreceive(struct socket *so, struct mbuf **paddr, struct uio *uio,
1075: struct mbuf **mp0, struct mbuf **controlp, int *flagsp)
1.1 cgd 1076: {
1.116 yamt 1077: struct lwp *l = curlwp;
1.160 ad 1078: struct mbuf *m, **mp, *mt;
1.146 dyoung 1079: int atomic, flags, len, error, s, offset, moff, type, orig_resid;
1.99 matt 1080: const struct protosw *pr;
1.54 lukem 1081: struct mbuf *nextrecord;
1.67 he 1082: int mbuf_removed = 0;
1.146 dyoung 1083: const struct domain *dom;
1.64 thorpej 1084:
1.54 lukem 1085: pr = so->so_proto;
1.146 dyoung 1086: atomic = pr->pr_flags & PR_ATOMIC;
1087: dom = pr->pr_domain;
1.1 cgd 1088: mp = mp0;
1.54 lukem 1089: type = 0;
1090: orig_resid = uio->uio_resid;
1.102 jonathan 1091:
1.144 dyoung 1092: if (paddr != NULL)
1093: *paddr = NULL;
1094: if (controlp != NULL)
1095: *controlp = NULL;
1096: if (flagsp != NULL)
1.1 cgd 1097: flags = *flagsp &~ MSG_EOR;
1098: else
1099: flags = 0;
1.66 enami 1100:
1101: if ((flags & MSG_DONTWAIT) == 0)
1.117 yamt 1102: sodopendfree();
1.66 enami 1103:
1.1 cgd 1104: if (flags & MSG_OOB) {
1105: m = m_get(M_WAIT, MT_DATA);
1.160 ad 1106: solock(so);
1.17 cgd 1107: error = (*pr->pr_usrreq)(so, PRU_RCVOOB, m,
1.140 dyoung 1108: (struct mbuf *)(long)(flags & MSG_PEEK), NULL, l);
1.160 ad 1109: sounlock(so);
1.1 cgd 1110: if (error)
1111: goto bad;
1112: do {
1.134 christos 1113: error = uiomove(mtod(m, void *),
1.1 cgd 1114: (int) min(uio->uio_resid, m->m_len), uio);
1115: m = m_free(m);
1.144 dyoung 1116: } while (uio->uio_resid > 0 && error == 0 && m);
1.54 lukem 1117: bad:
1.144 dyoung 1118: if (m != NULL)
1.1 cgd 1119: m_freem(m);
1.144 dyoung 1120: return error;
1.1 cgd 1121: }
1.144 dyoung 1122: if (mp != NULL)
1.140 dyoung 1123: *mp = NULL;
1.160 ad 1124:
1125: /*
1126: * solock() provides atomicity of access. splsoftnet() prevents
1127: * protocol processing soft interrupts from interrupting us and
1128: * blocking (expensive).
1129: */
1130: s = splsoftnet();
1131: solock(so);
1.1 cgd 1132: if (so->so_state & SS_ISCONFIRMING && uio->uio_resid)
1.140 dyoung 1133: (*pr->pr_usrreq)(so, PRU_RCVD, NULL, NULL, NULL, l);
1.1 cgd 1134:
1.54 lukem 1135: restart:
1.160 ad 1136: if ((error = sblock(&so->so_rcv, SBLOCKWAIT(flags))) != 0) {
1137: sounlock(so);
1138: splx(s);
1.144 dyoung 1139: return error;
1.160 ad 1140: }
1.1 cgd 1141:
1142: m = so->so_rcv.sb_mb;
1143: /*
1144: * If we have less data than requested, block awaiting more
1145: * (subject to any timeout) if:
1.15 mycroft 1146: * 1. the current count is less than the low water mark,
1.1 cgd 1147: * 2. MSG_WAITALL is set, and it is possible to do the entire
1.15 mycroft 1148: * receive operation at once if we block (resid <= hiwat), or
1149: * 3. MSG_DONTWAIT is not set.
1.1 cgd 1150: * If MSG_WAITALL is set but resid is larger than the receive buffer,
1151: * we have to do the receive in sections, and thus risk returning
1152: * a short count if a timeout or signal occurs after we start.
1153: */
1.144 dyoung 1154: if (m == NULL ||
1155: ((flags & MSG_DONTWAIT) == 0 &&
1156: so->so_rcv.sb_cc < uio->uio_resid &&
1157: (so->so_rcv.sb_cc < so->so_rcv.sb_lowat ||
1158: ((flags & MSG_WAITALL) &&
1159: uio->uio_resid <= so->so_rcv.sb_hiwat)) &&
1.146 dyoung 1160: m->m_nextpkt == NULL && !atomic)) {
1.1 cgd 1161: #ifdef DIAGNOSTIC
1.144 dyoung 1162: if (m == NULL && so->so_rcv.sb_cc)
1.1 cgd 1163: panic("receive 1");
1164: #endif
1165: if (so->so_error) {
1.144 dyoung 1166: if (m != NULL)
1.15 mycroft 1167: goto dontblock;
1.1 cgd 1168: error = so->so_error;
1169: if ((flags & MSG_PEEK) == 0)
1170: so->so_error = 0;
1171: goto release;
1172: }
1173: if (so->so_state & SS_CANTRCVMORE) {
1.144 dyoung 1174: if (m != NULL)
1.15 mycroft 1175: goto dontblock;
1.1 cgd 1176: else
1177: goto release;
1178: }
1.144 dyoung 1179: for (; m != NULL; m = m->m_next)
1.1 cgd 1180: if (m->m_type == MT_OOBDATA || (m->m_flags & M_EOR)) {
1181: m = so->so_rcv.sb_mb;
1182: goto dontblock;
1183: }
1184: if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) == 0 &&
1185: (so->so_proto->pr_flags & PR_CONNREQUIRED)) {
1186: error = ENOTCONN;
1187: goto release;
1188: }
1189: if (uio->uio_resid == 0)
1190: goto release;
1.151 ad 1191: if (so->so_nbio || (flags & MSG_DONTWAIT)) {
1.1 cgd 1192: error = EWOULDBLOCK;
1193: goto release;
1194: }
1.69 thorpej 1195: SBLASTRECORDCHK(&so->so_rcv, "soreceive sbwait 1");
1196: SBLASTMBUFCHK(&so->so_rcv, "soreceive sbwait 1");
1.1 cgd 1197: sbunlock(&so->so_rcv);
1198: error = sbwait(&so->so_rcv);
1.160 ad 1199: if (error != 0) {
1200: sounlock(so);
1201: splx(s);
1.144 dyoung 1202: return error;
1.160 ad 1203: }
1.1 cgd 1204: goto restart;
1205: }
1.54 lukem 1206: dontblock:
1.69 thorpej 1207: /*
1208: * On entry here, m points to the first record of the socket buffer.
1.159 ad 1209: * From this point onward, we maintain 'nextrecord' as a cache of the
1210: * pointer to the next record in the socket buffer. We must keep the
1211: * various socket buffer pointers and local stack versions of the
1212: * pointers in sync, pushing out modifications before dropping the
1.160 ad 1213: * socket lock, and re-reading them when picking it up.
1.159 ad 1214: *
1215: * Otherwise, we will race with the network stack appending new data
1216: * or records onto the socket buffer by using inconsistent/stale
1217: * versions of the field, possibly resulting in socket buffer
1218: * corruption.
1219: *
1220: * By holding the high-level sblock(), we prevent simultaneous
1221: * readers from pulling off the front of the socket buffer.
1.69 thorpej 1222: */
1.144 dyoung 1223: if (l != NULL)
1.157 ad 1224: l->l_ru.ru_msgrcv++;
1.69 thorpej 1225: KASSERT(m == so->so_rcv.sb_mb);
1226: SBLASTRECORDCHK(&so->so_rcv, "soreceive 1");
1227: SBLASTMBUFCHK(&so->so_rcv, "soreceive 1");
1.1 cgd 1228: nextrecord = m->m_nextpkt;
1229: if (pr->pr_flags & PR_ADDR) {
1230: #ifdef DIAGNOSTIC
1231: if (m->m_type != MT_SONAME)
1232: panic("receive 1a");
1233: #endif
1.3 andrew 1234: orig_resid = 0;
1.1 cgd 1235: if (flags & MSG_PEEK) {
1236: if (paddr)
1237: *paddr = m_copy(m, 0, m->m_len);
1238: m = m->m_next;
1239: } else {
1240: sbfree(&so->so_rcv, m);
1.67 he 1241: mbuf_removed = 1;
1.144 dyoung 1242: if (paddr != NULL) {
1.1 cgd 1243: *paddr = m;
1244: so->so_rcv.sb_mb = m->m_next;
1.144 dyoung 1245: m->m_next = NULL;
1.1 cgd 1246: m = so->so_rcv.sb_mb;
1247: } else {
1248: MFREE(m, so->so_rcv.sb_mb);
1249: m = so->so_rcv.sb_mb;
1250: }
1.159 ad 1251: sbsync(&so->so_rcv, nextrecord);
1.1 cgd 1252: }
1253: }
1.159 ad 1254:
1255: /*
1256: * Process one or more MT_CONTROL mbufs present before any data mbufs
1257: * in the first mbuf chain on the socket buffer. If MSG_PEEK, we
1258: * just copy the data; if !MSG_PEEK, we call into the protocol to
1259: * perform externalization (or freeing if controlp == NULL).
1260: */
1261: if (__predict_false(m != NULL && m->m_type == MT_CONTROL)) {
1262: struct mbuf *cm = NULL, *cmn;
1263: struct mbuf **cme = &cm;
1264:
1265: do {
1266: if (flags & MSG_PEEK) {
1267: if (controlp != NULL) {
1268: *controlp = m_copy(m, 0, m->m_len);
1269: controlp = &(*controlp)->m_next;
1270: }
1271: m = m->m_next;
1272: } else {
1273: sbfree(&so->so_rcv, m);
1.1 cgd 1274: so->so_rcv.sb_mb = m->m_next;
1.144 dyoung 1275: m->m_next = NULL;
1.159 ad 1276: *cme = m;
1277: cme = &(*cme)->m_next;
1.1 cgd 1278: m = so->so_rcv.sb_mb;
1.159 ad 1279: }
1280: } while (m != NULL && m->m_type == MT_CONTROL);
1281: if ((flags & MSG_PEEK) == 0)
1282: sbsync(&so->so_rcv, nextrecord);
1283: for (; cm != NULL; cm = cmn) {
1284: cmn = cm->m_next;
1285: cm->m_next = NULL;
1286: type = mtod(cm, struct cmsghdr *)->cmsg_type;
1287: if (controlp != NULL) {
1288: if (dom->dom_externalize != NULL &&
1289: type == SCM_RIGHTS) {
1.160 ad 1290: sounlock(so);
1.159 ad 1291: splx(s);
1292: error = (*dom->dom_externalize)(cm, l);
1293: s = splsoftnet();
1.160 ad 1294: solock(so);
1.159 ad 1295: }
1296: *controlp = cm;
1297: while (*controlp != NULL)
1298: controlp = &(*controlp)->m_next;
1.1 cgd 1299: } else {
1.106 itojun 1300: /*
1301: * Dispose of any SCM_RIGHTS message that went
1302: * through the read path rather than recv.
1303: */
1.159 ad 1304: if (dom->dom_dispose != NULL &&
1305: type == SCM_RIGHTS) {
1.160 ad 1306: sounlock(so);
1.159 ad 1307: (*dom->dom_dispose)(cm);
1.160 ad 1308: solock(so);
1.159 ad 1309: }
1310: m_freem(cm);
1.1 cgd 1311: }
1312: }
1.159 ad 1313: if (m != NULL)
1314: nextrecord = so->so_rcv.sb_mb->m_nextpkt;
1315: else
1316: nextrecord = so->so_rcv.sb_mb;
1317: orig_resid = 0;
1.1 cgd 1318: }
1.69 thorpej 1319:
1.159 ad 1320: /* If m is non-NULL, we have some data to read. */
1321: if (__predict_true(m != NULL)) {
1.1 cgd 1322: type = m->m_type;
1323: if (type == MT_OOBDATA)
1324: flags |= MSG_OOB;
1325: }
1.69 thorpej 1326: SBLASTRECORDCHK(&so->so_rcv, "soreceive 2");
1327: SBLASTMBUFCHK(&so->so_rcv, "soreceive 2");
1328:
1.1 cgd 1329: moff = 0;
1330: offset = 0;
1.144 dyoung 1331: while (m != NULL && uio->uio_resid > 0 && error == 0) {
1.1 cgd 1332: if (m->m_type == MT_OOBDATA) {
1333: if (type != MT_OOBDATA)
1334: break;
1335: } else if (type == MT_OOBDATA)
1336: break;
1337: #ifdef DIAGNOSTIC
1338: else if (m->m_type != MT_DATA && m->m_type != MT_HEADER)
1339: panic("receive 3");
1340: #endif
1341: so->so_state &= ~SS_RCVATMARK;
1342: len = uio->uio_resid;
1343: if (so->so_oobmark && len > so->so_oobmark - offset)
1344: len = so->so_oobmark - offset;
1345: if (len > m->m_len - moff)
1346: len = m->m_len - moff;
1347: /*
1348: * If mp is set, just pass back the mbufs.
1349: * Otherwise copy them out via the uio, then free.
1350: * Sockbuf must be consistent here (points to current mbuf,
1351: * it points to next record) when we drop priority;
1352: * we must note any additions to the sockbuf when we
1353: * block interrupts again.
1354: */
1.144 dyoung 1355: if (mp == NULL) {
1.69 thorpej 1356: SBLASTRECORDCHK(&so->so_rcv, "soreceive uiomove");
1357: SBLASTMBUFCHK(&so->so_rcv, "soreceive uiomove");
1.160 ad 1358: sounlock(so);
1.1 cgd 1359: splx(s);
1.134 christos 1360: error = uiomove(mtod(m, char *) + moff, (int)len, uio);
1.20 mycroft 1361: s = splsoftnet();
1.160 ad 1362: solock(so);
1.144 dyoung 1363: if (error != 0) {
1.67 he 1364: /*
1365: * If any part of the record has been removed
1366: * (such as the MT_SONAME mbuf, which will
1367: * happen when PR_ADDR, and thus also
1368: * PR_ATOMIC, is set), then drop the entire
1369: * record to maintain the atomicity of the
1370: * receive operation.
1371: *
1372: * This avoids a later panic("receive 1a")
1373: * when compiled with DIAGNOSTIC.
1374: */
1.146 dyoung 1375: if (m && mbuf_removed && atomic)
1.67 he 1376: (void) sbdroprecord(&so->so_rcv);
1377:
1.57 jdolecek 1378: goto release;
1.67 he 1379: }
1.1 cgd 1380: } else
1381: uio->uio_resid -= len;
1382: if (len == m->m_len - moff) {
1383: if (m->m_flags & M_EOR)
1384: flags |= MSG_EOR;
1385: if (flags & MSG_PEEK) {
1386: m = m->m_next;
1387: moff = 0;
1388: } else {
1389: nextrecord = m->m_nextpkt;
1390: sbfree(&so->so_rcv, m);
1391: if (mp) {
1392: *mp = m;
1393: mp = &m->m_next;
1394: so->so_rcv.sb_mb = m = m->m_next;
1.140 dyoung 1395: *mp = NULL;
1.1 cgd 1396: } else {
1397: MFREE(m, so->so_rcv.sb_mb);
1398: m = so->so_rcv.sb_mb;
1399: }
1.69 thorpej 1400: /*
1401: * If m != NULL, we also know that
1402: * so->so_rcv.sb_mb != NULL.
1403: */
1404: KASSERT(so->so_rcv.sb_mb == m);
1405: if (m) {
1.1 cgd 1406: m->m_nextpkt = nextrecord;
1.69 thorpej 1407: if (nextrecord == NULL)
1408: so->so_rcv.sb_lastrecord = m;
1409: } else {
1410: so->so_rcv.sb_mb = nextrecord;
1.70 thorpej 1411: SB_EMPTY_FIXUP(&so->so_rcv);
1.69 thorpej 1412: }
1413: SBLASTRECORDCHK(&so->so_rcv, "soreceive 3");
1414: SBLASTMBUFCHK(&so->so_rcv, "soreceive 3");
1.1 cgd 1415: }
1.144 dyoung 1416: } else if (flags & MSG_PEEK)
1417: moff += len;
1418: else {
1.160 ad 1419: if (mp != NULL) {
1420: mt = m_copym(m, 0, len, M_NOWAIT);
1421: if (__predict_false(mt == NULL)) {
1422: sounlock(so);
1423: mt = m_copym(m, 0, len, M_WAIT);
1424: solock(so);
1425: }
1426: *mp = mt;
1427: }
1.144 dyoung 1428: m->m_data += len;
1429: m->m_len -= len;
1430: so->so_rcv.sb_cc -= len;
1.1 cgd 1431: }
1432: if (so->so_oobmark) {
1433: if ((flags & MSG_PEEK) == 0) {
1434: so->so_oobmark -= len;
1435: if (so->so_oobmark == 0) {
1436: so->so_state |= SS_RCVATMARK;
1437: break;
1438: }
1.7 cgd 1439: } else {
1.1 cgd 1440: offset += len;
1.7 cgd 1441: if (offset == so->so_oobmark)
1442: break;
1443: }
1.1 cgd 1444: }
1445: if (flags & MSG_EOR)
1446: break;
1447: /*
1448: * If the MSG_WAITALL flag is set (for non-atomic socket),
1449: * we must not quit until "uio->uio_resid == 0" or an error
1450: * termination. If a signal/timeout occurs, return
1451: * with a short count but without error.
1452: * Keep sockbuf locked against other readers.
1453: */
1.144 dyoung 1454: while (flags & MSG_WAITALL && m == NULL && uio->uio_resid > 0 &&
1.3 andrew 1455: !sosendallatonce(so) && !nextrecord) {
1.1 cgd 1456: if (so->so_error || so->so_state & SS_CANTRCVMORE)
1457: break;
1.68 matt 1458: /*
1459: * If we are peeking and the socket receive buffer is
1460: * full, stop since we can't get more data to peek at.
1461: */
1462: if ((flags & MSG_PEEK) && sbspace(&so->so_rcv) <= 0)
1463: break;
1464: /*
1465: * If we've drained the socket buffer, tell the
1466: * protocol in case it needs to do something to
1467: * get it filled again.
1468: */
1469: if ((pr->pr_flags & PR_WANTRCVD) && so->so_pcb)
1470: (*pr->pr_usrreq)(so, PRU_RCVD,
1.140 dyoung 1471: NULL, (struct mbuf *)(long)flags, NULL, l);
1.69 thorpej 1472: SBLASTRECORDCHK(&so->so_rcv, "soreceive sbwait 2");
1473: SBLASTMBUFCHK(&so->so_rcv, "soreceive sbwait 2");
1.1 cgd 1474: error = sbwait(&so->so_rcv);
1.144 dyoung 1475: if (error != 0) {
1.1 cgd 1476: sbunlock(&so->so_rcv);
1.160 ad 1477: sounlock(so);
1.1 cgd 1478: splx(s);
1.144 dyoung 1479: return 0;
1.1 cgd 1480: }
1.21 christos 1481: if ((m = so->so_rcv.sb_mb) != NULL)
1.1 cgd 1482: nextrecord = m->m_nextpkt;
1483: }
1484: }
1.3 andrew 1485:
1.146 dyoung 1486: if (m && atomic) {
1.3 andrew 1487: flags |= MSG_TRUNC;
1488: if ((flags & MSG_PEEK) == 0)
1489: (void) sbdroprecord(&so->so_rcv);
1490: }
1.1 cgd 1491: if ((flags & MSG_PEEK) == 0) {
1.144 dyoung 1492: if (m == NULL) {
1.69 thorpej 1493: /*
1.70 thorpej 1494: * First part is an inline SB_EMPTY_FIXUP(). Second
1.69 thorpej 1495: * part makes sure sb_lastrecord is up-to-date if
1496: * there is still data in the socket buffer.
1497: */
1.1 cgd 1498: so->so_rcv.sb_mb = nextrecord;
1.69 thorpej 1499: if (so->so_rcv.sb_mb == NULL) {
1500: so->so_rcv.sb_mbtail = NULL;
1501: so->so_rcv.sb_lastrecord = NULL;
1502: } else if (nextrecord->m_nextpkt == NULL)
1503: so->so_rcv.sb_lastrecord = nextrecord;
1504: }
1505: SBLASTRECORDCHK(&so->so_rcv, "soreceive 4");
1506: SBLASTMBUFCHK(&so->so_rcv, "soreceive 4");
1.1 cgd 1507: if (pr->pr_flags & PR_WANTRCVD && so->so_pcb)
1.140 dyoung 1508: (*pr->pr_usrreq)(so, PRU_RCVD, NULL,
1509: (struct mbuf *)(long)flags, NULL, l);
1.1 cgd 1510: }
1.3 andrew 1511: if (orig_resid == uio->uio_resid && orig_resid &&
1512: (flags & MSG_EOR) == 0 && (so->so_state & SS_CANTRCVMORE) == 0) {
1513: sbunlock(&so->so_rcv);
1514: goto restart;
1515: }
1.108 perry 1516:
1.144 dyoung 1517: if (flagsp != NULL)
1.1 cgd 1518: *flagsp |= flags;
1.54 lukem 1519: release:
1.1 cgd 1520: sbunlock(&so->so_rcv);
1.160 ad 1521: sounlock(so);
1.1 cgd 1522: splx(s);
1.144 dyoung 1523: return error;
1.1 cgd 1524: }
1525:
1.14 mycroft 1526: int
1.54 lukem 1527: soshutdown(struct socket *so, int how)
1.1 cgd 1528: {
1.99 matt 1529: const struct protosw *pr;
1.160 ad 1530: int error;
1531:
1532: KASSERT(solocked(so));
1.34 kleink 1533:
1.54 lukem 1534: pr = so->so_proto;
1.34 kleink 1535: if (!(how == SHUT_RD || how == SHUT_WR || how == SHUT_RDWR))
1536: return (EINVAL);
1.1 cgd 1537:
1.160 ad 1538: if (how == SHUT_RD || how == SHUT_RDWR) {
1.1 cgd 1539: sorflush(so);
1.160 ad 1540: error = 0;
1541: }
1.34 kleink 1542: if (how == SHUT_WR || how == SHUT_RDWR)
1.160 ad 1543: error = (*pr->pr_usrreq)(so, PRU_SHUTDOWN, NULL,
1.140 dyoung 1544: NULL, NULL, NULL);
1.160 ad 1545:
1546: return error;
1.1 cgd 1547: }
1548:
1.14 mycroft 1549: void
1.54 lukem 1550: sorflush(struct socket *so)
1.1 cgd 1551: {
1.54 lukem 1552: struct sockbuf *sb, asb;
1.99 matt 1553: const struct protosw *pr;
1.160 ad 1554:
1555: KASSERT(solocked(so));
1.1 cgd 1556:
1.54 lukem 1557: sb = &so->so_rcv;
1558: pr = so->so_proto;
1.160 ad 1559: socantrcvmore(so);
1.1 cgd 1560: sb->sb_flags |= SB_NOINTR;
1.160 ad 1561: (void )sblock(sb, M_WAITOK);
1.1 cgd 1562: sbunlock(sb);
1563: asb = *sb;
1.86 wrstuden 1564: /*
1565: * Clear most of the sockbuf structure, but leave some of the
1566: * fields valid.
1567: */
1568: memset(&sb->sb_startzero, 0,
1569: sizeof(*sb) - offsetof(struct sockbuf, sb_startzero));
1.160 ad 1570: if (pr->pr_flags & PR_RIGHTS && pr->pr_domain->dom_dispose) {
1571: sounlock(so);
1.1 cgd 1572: (*pr->pr_domain->dom_dispose)(asb.sb_mb);
1.160 ad 1573: solock(so);
1574: }
1.98 christos 1575: sbrelease(&asb, so);
1.1 cgd 1576: }
1577:
1.142 dyoung 1578: static int
1579: sosetopt1(struct socket *so, int level, int optname, struct mbuf *m)
1.1 cgd 1580: {
1.170 ! tls 1581: #ifdef INET
! 1582: int error, optval, val;
! 1583: #else
1.142 dyoung 1584: int optval, val;
1.170 ! tls 1585: #endif
1.130 christos 1586: struct linger *l;
1.141 yamt 1587: struct sockbuf *sb;
1.142 dyoung 1588: struct timeval *tv;
1589:
1590: switch (optname) {
1591:
1.170 ! tls 1592: #ifdef INET
! 1593: case SO_ACCEPTFILTER:
! 1594: error = do_setopt_accept_filter(so, m);
! 1595: if (error)
! 1596: return error;
! 1597: break;
! 1598: #endif
! 1599:
1.142 dyoung 1600: case SO_LINGER:
1601: if (m == NULL || m->m_len != sizeof(struct linger))
1602: return EINVAL;
1603: l = mtod(m, struct linger *);
1604: if (l->l_linger < 0 || l->l_linger > USHRT_MAX ||
1605: l->l_linger > (INT_MAX / hz))
1606: return EDOM;
1607: so->so_linger = l->l_linger;
1608: if (l->l_onoff)
1609: so->so_options |= SO_LINGER;
1610: else
1611: so->so_options &= ~SO_LINGER;
1612: break;
1.1 cgd 1613:
1.142 dyoung 1614: case SO_DEBUG:
1615: case SO_KEEPALIVE:
1616: case SO_DONTROUTE:
1617: case SO_USELOOPBACK:
1618: case SO_BROADCAST:
1619: case SO_REUSEADDR:
1620: case SO_REUSEPORT:
1621: case SO_OOBINLINE:
1622: case SO_TIMESTAMP:
1623: if (m == NULL || m->m_len < sizeof(int))
1624: return EINVAL;
1625: if (*mtod(m, int *))
1626: so->so_options |= optname;
1627: else
1628: so->so_options &= ~optname;
1629: break;
1630:
1631: case SO_SNDBUF:
1632: case SO_RCVBUF:
1633: case SO_SNDLOWAT:
1634: case SO_RCVLOWAT:
1635: if (m == NULL || m->m_len < sizeof(int))
1636: return EINVAL;
1.1 cgd 1637:
1.142 dyoung 1638: /*
1639: * Values < 1 make no sense for any of these
1640: * options, so disallow them.
1641: */
1642: optval = *mtod(m, int *);
1643: if (optval < 1)
1644: return EINVAL;
1.1 cgd 1645:
1.142 dyoung 1646: switch (optname) {
1.1 cgd 1647:
1648: case SO_SNDBUF:
1649: case SO_RCVBUF:
1.142 dyoung 1650: sb = (optname == SO_SNDBUF) ?
1651: &so->so_snd : &so->so_rcv;
1652: if (sbreserve(sb, (u_long)optval, so) == 0)
1653: return ENOBUFS;
1654: sb->sb_flags &= ~SB_AUTOSIZE;
1655: break;
1656:
1657: /*
1658: * Make sure the low-water is never greater than
1659: * the high-water.
1660: */
1.1 cgd 1661: case SO_SNDLOWAT:
1.142 dyoung 1662: so->so_snd.sb_lowat =
1663: (optval > so->so_snd.sb_hiwat) ?
1664: so->so_snd.sb_hiwat : optval;
1665: break;
1.1 cgd 1666: case SO_RCVLOWAT:
1.142 dyoung 1667: so->so_rcv.sb_lowat =
1668: (optval > so->so_rcv.sb_hiwat) ?
1669: so->so_rcv.sb_hiwat : optval;
1670: break;
1671: }
1672: break;
1.28 thorpej 1673:
1.142 dyoung 1674: case SO_SNDTIMEO:
1675: case SO_RCVTIMEO:
1676: if (m == NULL || m->m_len < sizeof(*tv))
1677: return EINVAL;
1678: tv = mtod(m, struct timeval *);
1679: if (tv->tv_sec > (INT_MAX - tv->tv_usec / tick) / hz)
1680: return EDOM;
1681: val = tv->tv_sec * hz + tv->tv_usec / tick;
1682: if (val == 0 && tv->tv_usec != 0)
1683: val = 1;
1.28 thorpej 1684:
1.142 dyoung 1685: switch (optname) {
1.28 thorpej 1686:
1.142 dyoung 1687: case SO_SNDTIMEO:
1688: so->so_snd.sb_timeo = val;
1.1 cgd 1689: break;
1690: case SO_RCVTIMEO:
1.142 dyoung 1691: so->so_rcv.sb_timeo = val;
1692: break;
1693: }
1694: break;
1.1 cgd 1695:
1.142 dyoung 1696: default:
1697: return ENOPROTOOPT;
1698: }
1699: return 0;
1700: }
1.1 cgd 1701:
1.142 dyoung 1702: int
1703: sosetopt(struct socket *so, int level, int optname, struct mbuf *m)
1704: {
1705: int error, prerr;
1.1 cgd 1706:
1.160 ad 1707: solock(so);
1.142 dyoung 1708: if (level == SOL_SOCKET)
1709: error = sosetopt1(so, level, optname, m);
1710: else
1711: error = ENOPROTOOPT;
1.1 cgd 1712:
1.142 dyoung 1713: if ((error == 0 || error == ENOPROTOOPT) &&
1714: so->so_proto != NULL && so->so_proto->pr_ctloutput != NULL) {
1715: /* give the protocol stack a shot */
1716: prerr = (*so->so_proto->pr_ctloutput)(PRCO_SETOPT, so, level,
1717: optname, &m);
1718: if (prerr == 0)
1719: error = 0;
1720: else if (prerr != ENOPROTOOPT)
1721: error = prerr;
1722: } else if (m != NULL)
1723: (void)m_free(m);
1.160 ad 1724: sounlock(so);
1.142 dyoung 1725: return error;
1.1 cgd 1726: }
1727:
1.14 mycroft 1728: int
1.54 lukem 1729: sogetopt(struct socket *so, int level, int optname, struct mbuf **mp)
1.1 cgd 1730: {
1.54 lukem 1731: struct mbuf *m;
1.160 ad 1732: int error;
1.1 cgd 1733:
1.160 ad 1734: solock(so);
1.1 cgd 1735: if (level != SOL_SOCKET) {
1736: if (so->so_proto && so->so_proto->pr_ctloutput) {
1.160 ad 1737: error = ((*so->so_proto->pr_ctloutput)
1.1 cgd 1738: (PRCO_GETOPT, so, level, optname, mp));
1739: } else
1.160 ad 1740: error = (ENOPROTOOPT);
1.1 cgd 1741: } else {
1742: m = m_get(M_WAIT, MT_SOOPTS);
1.36 perry 1743: m->m_len = sizeof(int);
1.1 cgd 1744:
1745: switch (optname) {
1746:
1.170 ! tls 1747: #ifdef INET
! 1748: case SO_ACCEPTFILTER:
! 1749: error = do_getopt_accept_filter(so, m);
! 1750: break;
! 1751: #endif
! 1752:
1.1 cgd 1753: case SO_LINGER:
1.36 perry 1754: m->m_len = sizeof(struct linger);
1.1 cgd 1755: mtod(m, struct linger *)->l_onoff =
1.131 christos 1756: (so->so_options & SO_LINGER) ? 1 : 0;
1.1 cgd 1757: mtod(m, struct linger *)->l_linger = so->so_linger;
1758: break;
1759:
1760: case SO_USELOOPBACK:
1761: case SO_DONTROUTE:
1762: case SO_DEBUG:
1763: case SO_KEEPALIVE:
1764: case SO_REUSEADDR:
1.15 mycroft 1765: case SO_REUSEPORT:
1.1 cgd 1766: case SO_BROADCAST:
1767: case SO_OOBINLINE:
1.26 thorpej 1768: case SO_TIMESTAMP:
1.131 christos 1769: *mtod(m, int *) = (so->so_options & optname) ? 1 : 0;
1.1 cgd 1770: break;
1771:
1772: case SO_TYPE:
1773: *mtod(m, int *) = so->so_type;
1774: break;
1775:
1776: case SO_ERROR:
1777: *mtod(m, int *) = so->so_error;
1778: so->so_error = 0;
1779: break;
1780:
1781: case SO_SNDBUF:
1782: *mtod(m, int *) = so->so_snd.sb_hiwat;
1783: break;
1784:
1785: case SO_RCVBUF:
1786: *mtod(m, int *) = so->so_rcv.sb_hiwat;
1787: break;
1788:
1789: case SO_SNDLOWAT:
1790: *mtod(m, int *) = so->so_snd.sb_lowat;
1791: break;
1792:
1793: case SO_RCVLOWAT:
1794: *mtod(m, int *) = so->so_rcv.sb_lowat;
1795: break;
1796:
1797: case SO_SNDTIMEO:
1798: case SO_RCVTIMEO:
1799: {
1800: int val = (optname == SO_SNDTIMEO ?
1801: so->so_snd.sb_timeo : so->so_rcv.sb_timeo);
1802:
1803: m->m_len = sizeof(struct timeval);
1804: mtod(m, struct timeval *)->tv_sec = val / hz;
1805: mtod(m, struct timeval *)->tv_usec =
1.27 kleink 1806: (val % hz) * tick;
1.1 cgd 1807: break;
1808: }
1809:
1.107 darrenr 1810: case SO_OVERFLOWED:
1811: *mtod(m, int *) = so->so_rcv.sb_overflowed;
1812: break;
1813:
1.1 cgd 1814: default:
1.160 ad 1815: sounlock(so);
1.1 cgd 1816: (void)m_free(m);
1817: return (ENOPROTOOPT);
1818: }
1819: *mp = m;
1.160 ad 1820: error = 0;
1.1 cgd 1821: }
1.160 ad 1822:
1823: sounlock(so);
1824: return (error);
1.1 cgd 1825: }
1826:
1.14 mycroft 1827: void
1.54 lukem 1828: sohasoutofband(struct socket *so)
1.1 cgd 1829: {
1.153 rmind 1830:
1.90 christos 1831: fownsignal(so->so_pgid, SIGURG, POLL_PRI, POLLPRI|POLLRDBAND, so);
1.153 rmind 1832: selnotify(&so->so_rcv.sb_sel, POLLPRI | POLLRDBAND, 0);
1.1 cgd 1833: }
1.72 jdolecek 1834:
1835: static void
1836: filt_sordetach(struct knote *kn)
1837: {
1838: struct socket *so;
1839:
1.155 ad 1840: so = ((file_t *)kn->kn_obj)->f_data;
1.160 ad 1841: solock(so);
1.73 christos 1842: SLIST_REMOVE(&so->so_rcv.sb_sel.sel_klist, kn, knote, kn_selnext);
1843: if (SLIST_EMPTY(&so->so_rcv.sb_sel.sel_klist))
1.72 jdolecek 1844: so->so_rcv.sb_flags &= ~SB_KNOTE;
1.160 ad 1845: sounlock(so);
1.72 jdolecek 1846: }
1847:
1848: /*ARGSUSED*/
1849: static int
1.129 yamt 1850: filt_soread(struct knote *kn, long hint)
1.72 jdolecek 1851: {
1852: struct socket *so;
1.160 ad 1853: int rv;
1.72 jdolecek 1854:
1.155 ad 1855: so = ((file_t *)kn->kn_obj)->f_data;
1.160 ad 1856: if (hint != NOTE_SUBMIT)
1857: solock(so);
1.72 jdolecek 1858: kn->kn_data = so->so_rcv.sb_cc;
1859: if (so->so_state & SS_CANTRCVMORE) {
1.108 perry 1860: kn->kn_flags |= EV_EOF;
1.72 jdolecek 1861: kn->kn_fflags = so->so_error;
1.160 ad 1862: rv = 1;
1863: } else if (so->so_error) /* temporary udp error */
1864: rv = 1;
1865: else if (kn->kn_sfflags & NOTE_LOWAT)
1866: rv = (kn->kn_data >= kn->kn_sdata);
1867: else
1868: rv = (kn->kn_data >= so->so_rcv.sb_lowat);
1869: if (hint != NOTE_SUBMIT)
1870: sounlock(so);
1871: return rv;
1.72 jdolecek 1872: }
1873:
1874: static void
1875: filt_sowdetach(struct knote *kn)
1876: {
1877: struct socket *so;
1878:
1.155 ad 1879: so = ((file_t *)kn->kn_obj)->f_data;
1.160 ad 1880: solock(so);
1.73 christos 1881: SLIST_REMOVE(&so->so_snd.sb_sel.sel_klist, kn, knote, kn_selnext);
1882: if (SLIST_EMPTY(&so->so_snd.sb_sel.sel_klist))
1.72 jdolecek 1883: so->so_snd.sb_flags &= ~SB_KNOTE;
1.160 ad 1884: sounlock(so);
1.72 jdolecek 1885: }
1886:
1887: /*ARGSUSED*/
1888: static int
1.129 yamt 1889: filt_sowrite(struct knote *kn, long hint)
1.72 jdolecek 1890: {
1891: struct socket *so;
1.160 ad 1892: int rv;
1.72 jdolecek 1893:
1.155 ad 1894: so = ((file_t *)kn->kn_obj)->f_data;
1.160 ad 1895: if (hint != NOTE_SUBMIT)
1896: solock(so);
1.72 jdolecek 1897: kn->kn_data = sbspace(&so->so_snd);
1898: if (so->so_state & SS_CANTSENDMORE) {
1.108 perry 1899: kn->kn_flags |= EV_EOF;
1.72 jdolecek 1900: kn->kn_fflags = so->so_error;
1.160 ad 1901: rv = 1;
1902: } else if (so->so_error) /* temporary udp error */
1903: rv = 1;
1904: else if (((so->so_state & SS_ISCONNECTED) == 0) &&
1.72 jdolecek 1905: (so->so_proto->pr_flags & PR_CONNREQUIRED))
1.160 ad 1906: rv = 0;
1907: else if (kn->kn_sfflags & NOTE_LOWAT)
1908: rv = (kn->kn_data >= kn->kn_sdata);
1909: else
1910: rv = (kn->kn_data >= so->so_snd.sb_lowat);
1911: if (hint != NOTE_SUBMIT)
1912: sounlock(so);
1913: return rv;
1.72 jdolecek 1914: }
1915:
1916: /*ARGSUSED*/
1917: static int
1.129 yamt 1918: filt_solisten(struct knote *kn, long hint)
1.72 jdolecek 1919: {
1920: struct socket *so;
1.160 ad 1921: int rv;
1.72 jdolecek 1922:
1.155 ad 1923: so = ((file_t *)kn->kn_obj)->f_data;
1.72 jdolecek 1924:
1925: /*
1926: * Set kn_data to number of incoming connections, not
1927: * counting partial (incomplete) connections.
1.108 perry 1928: */
1.160 ad 1929: if (hint != NOTE_SUBMIT)
1930: solock(so);
1.72 jdolecek 1931: kn->kn_data = so->so_qlen;
1.160 ad 1932: rv = (kn->kn_data > 0);
1933: if (hint != NOTE_SUBMIT)
1934: sounlock(so);
1935: return rv;
1.72 jdolecek 1936: }
1937:
1938: static const struct filterops solisten_filtops =
1939: { 1, NULL, filt_sordetach, filt_solisten };
1940: static const struct filterops soread_filtops =
1941: { 1, NULL, filt_sordetach, filt_soread };
1942: static const struct filterops sowrite_filtops =
1943: { 1, NULL, filt_sowdetach, filt_sowrite };
1944:
1945: int
1.129 yamt 1946: soo_kqfilter(struct file *fp, struct knote *kn)
1.72 jdolecek 1947: {
1948: struct socket *so;
1949: struct sockbuf *sb;
1950:
1.155 ad 1951: so = ((file_t *)kn->kn_obj)->f_data;
1.160 ad 1952: solock(so);
1.72 jdolecek 1953: switch (kn->kn_filter) {
1954: case EVFILT_READ:
1955: if (so->so_options & SO_ACCEPTCONN)
1956: kn->kn_fop = &solisten_filtops;
1957: else
1958: kn->kn_fop = &soread_filtops;
1959: sb = &so->so_rcv;
1960: break;
1961: case EVFILT_WRITE:
1962: kn->kn_fop = &sowrite_filtops;
1963: sb = &so->so_snd;
1964: break;
1965: default:
1.160 ad 1966: sounlock(so);
1.149 pooka 1967: return (EINVAL);
1.72 jdolecek 1968: }
1.73 christos 1969: SLIST_INSERT_HEAD(&sb->sb_sel.sel_klist, kn, kn_selnext);
1.72 jdolecek 1970: sb->sb_flags |= SB_KNOTE;
1.160 ad 1971: sounlock(so);
1.72 jdolecek 1972: return (0);
1973: }
1974:
1.154 ad 1975: static int
1976: sodopoll(struct socket *so, int events)
1977: {
1978: int revents;
1979:
1980: revents = 0;
1981:
1982: if (events & (POLLIN | POLLRDNORM))
1983: if (soreadable(so))
1984: revents |= events & (POLLIN | POLLRDNORM);
1985:
1986: if (events & (POLLOUT | POLLWRNORM))
1987: if (sowritable(so))
1988: revents |= events & (POLLOUT | POLLWRNORM);
1989:
1990: if (events & (POLLPRI | POLLRDBAND))
1991: if (so->so_oobmark || (so->so_state & SS_RCVATMARK))
1992: revents |= events & (POLLPRI | POLLRDBAND);
1993:
1994: return revents;
1995: }
1996:
1997: int
1998: sopoll(struct socket *so, int events)
1999: {
2000: int revents = 0;
2001:
1.160 ad 2002: #ifndef DIAGNOSTIC
2003: /*
2004: * Do a quick, unlocked check in expectation that the socket
2005: * will be ready for I/O. Don't do this check if DIAGNOSTIC,
2006: * as the solocked() assertions will fail.
2007: */
1.154 ad 2008: if ((revents = sodopoll(so, events)) != 0)
2009: return revents;
1.160 ad 2010: #endif
1.154 ad 2011:
1.160 ad 2012: solock(so);
1.154 ad 2013: if ((revents = sodopoll(so, events)) == 0) {
2014: if (events & (POLLIN | POLLPRI | POLLRDNORM | POLLRDBAND)) {
2015: selrecord(curlwp, &so->so_rcv.sb_sel);
1.160 ad 2016: so->so_rcv.sb_flags |= SB_NOTIFY;
1.154 ad 2017: }
2018:
2019: if (events & (POLLOUT | POLLWRNORM)) {
2020: selrecord(curlwp, &so->so_snd.sb_sel);
1.160 ad 2021: so->so_snd.sb_flags |= SB_NOTIFY;
1.154 ad 2022: }
2023: }
1.160 ad 2024: sounlock(so);
1.154 ad 2025:
2026: return revents;
2027: }
2028:
2029:
1.94 yamt 2030: #include <sys/sysctl.h>
2031:
2032: static int sysctl_kern_somaxkva(SYSCTLFN_PROTO);
2033:
2034: /*
2035: * sysctl helper routine for kern.somaxkva. ensures that the given
2036: * value is not too small.
2037: * (XXX should we maybe make sure it's not too large as well?)
2038: */
2039: static int
2040: sysctl_kern_somaxkva(SYSCTLFN_ARGS)
2041: {
2042: int error, new_somaxkva;
2043: struct sysctlnode node;
2044:
2045: new_somaxkva = somaxkva;
2046: node = *rnode;
2047: node.sysctl_data = &new_somaxkva;
2048: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2049: if (error || newp == NULL)
2050: return (error);
2051:
2052: if (new_somaxkva < (16 * 1024 * 1024)) /* sanity */
2053: return (EINVAL);
2054:
1.136 ad 2055: mutex_enter(&so_pendfree_lock);
1.94 yamt 2056: somaxkva = new_somaxkva;
1.136 ad 2057: cv_broadcast(&socurkva_cv);
2058: mutex_exit(&so_pendfree_lock);
1.94 yamt 2059:
2060: return (error);
2061: }
2062:
2063: SYSCTL_SETUP(sysctl_kern_somaxkva_setup, "sysctl kern.somaxkva setup")
2064: {
2065:
1.97 atatat 2066: sysctl_createv(clog, 0, NULL, NULL,
2067: CTLFLAG_PERMANENT,
2068: CTLTYPE_NODE, "kern", NULL,
2069: NULL, 0, NULL, 0,
2070: CTL_KERN, CTL_EOL);
2071:
2072: sysctl_createv(clog, 0, NULL, NULL,
2073: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.103 atatat 2074: CTLTYPE_INT, "somaxkva",
2075: SYSCTL_DESCR("Maximum amount of kernel memory to be "
2076: "used for socket buffers"),
1.94 yamt 2077: sysctl_kern_somaxkva, 0, NULL, 0,
2078: CTL_KERN, KERN_SOMAXKVA, CTL_EOL);
2079: }
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