Annotation of src/sys/kern/uipc_socket.c, Revision 1.181
1.181 ! christos 1: /* $NetBSD: uipc_socket.c,v 1.180 2009/01/14 20:33:19 cegger 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>
1.181 ! christos 66: __KERNEL_RCSID(0, "$NetBSD: uipc_socket.c,v 1.180 2009/01/14 20:33:19 cegger Exp $");
1.64 thorpej 67:
1.179 christos 68: #include "opt_compat_netbsd.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>
1.173 plunky 80: #include <sys/kmem.h>
1.9 mycroft 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>
1.174 pooka 89: #include <sys/uidinfo.h>
1.72 jdolecek 90: #include <sys/event.h>
1.89 christos 91: #include <sys/poll.h>
1.118 elad 92: #include <sys/kauth.h>
1.136 ad 93: #include <sys/mutex.h>
94: #include <sys/condvar.h>
1.37 thorpej 95:
1.179 christos 96: #ifdef COMPAT_50
97: #include <compat/sys/time.h>
98: #define SO_OSNDTIMEO 0x1005
99: #define SO_ORCVTIMEO 0x1006
100: #endif
101:
1.64 thorpej 102: #include <uvm/uvm.h>
103:
1.77 thorpej 104: MALLOC_DEFINE(M_SOOPTS, "soopts", "socket options");
105: MALLOC_DEFINE(M_SONAME, "soname", "socket name");
1.37 thorpej 106:
1.142 dyoung 107: extern const struct fileops socketops;
108:
1.54 lukem 109: extern int somaxconn; /* patchable (XXX sysctl) */
110: int somaxconn = SOMAXCONN;
1.160 ad 111: kmutex_t *softnet_lock;
1.49 jonathan 112:
1.64 thorpej 113: #ifdef SOSEND_COUNTERS
114: #include <sys/device.h>
115:
1.113 thorpej 116: static struct evcnt sosend_loan_big = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64 thorpej 117: NULL, "sosend", "loan big");
1.113 thorpej 118: static struct evcnt sosend_copy_big = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64 thorpej 119: NULL, "sosend", "copy big");
1.113 thorpej 120: static struct evcnt sosend_copy_small = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64 thorpej 121: NULL, "sosend", "copy small");
1.113 thorpej 122: static struct evcnt sosend_kvalimit = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64 thorpej 123: NULL, "sosend", "kva limit");
124:
125: #define SOSEND_COUNTER_INCR(ev) (ev)->ev_count++
126:
1.101 matt 127: EVCNT_ATTACH_STATIC(sosend_loan_big);
128: EVCNT_ATTACH_STATIC(sosend_copy_big);
129: EVCNT_ATTACH_STATIC(sosend_copy_small);
130: EVCNT_ATTACH_STATIC(sosend_kvalimit);
1.64 thorpej 131: #else
132:
133: #define SOSEND_COUNTER_INCR(ev) /* nothing */
134:
135: #endif /* SOSEND_COUNTERS */
136:
1.119 yamt 137: static struct callback_entry sokva_reclaimerentry;
1.1 cgd 138:
1.167 ad 139: #if defined(SOSEND_NO_LOAN) || defined(MULTIPROCESSOR)
1.121 yamt 140: int sock_loan_thresh = -1;
1.71 thorpej 141: #else
1.121 yamt 142: int sock_loan_thresh = 4096;
1.65 thorpej 143: #endif
1.64 thorpej 144:
1.136 ad 145: static kmutex_t so_pendfree_lock;
1.113 thorpej 146: static struct mbuf *so_pendfree;
1.64 thorpej 147:
1.84 ragge 148: #ifndef SOMAXKVA
149: #define SOMAXKVA (16 * 1024 * 1024)
150: #endif
151: int somaxkva = SOMAXKVA;
1.113 thorpej 152: static int socurkva;
1.136 ad 153: static kcondvar_t socurkva_cv;
1.64 thorpej 154:
155: #define SOCK_LOAN_CHUNK 65536
156:
1.117 yamt 157: static size_t sodopendfree(void);
158: static size_t sodopendfreel(void);
1.93 yamt 159:
1.178 pooka 160: static void sysctl_kern_somaxkva_setup(void);
161: static struct sysctllog *socket_sysctllog;
162:
1.113 thorpej 163: static vsize_t
1.129 yamt 164: sokvareserve(struct socket *so, vsize_t len)
1.80 yamt 165: {
1.98 christos 166: int error;
1.80 yamt 167:
1.136 ad 168: mutex_enter(&so_pendfree_lock);
1.80 yamt 169: while (socurkva + len > somaxkva) {
1.93 yamt 170: size_t freed;
171:
172: /*
173: * try to do pendfree.
174: */
175:
1.117 yamt 176: freed = sodopendfreel();
1.93 yamt 177:
178: /*
179: * if some kva was freed, try again.
180: */
181:
182: if (freed)
1.80 yamt 183: continue;
1.93 yamt 184:
1.80 yamt 185: SOSEND_COUNTER_INCR(&sosend_kvalimit);
1.136 ad 186: error = cv_wait_sig(&socurkva_cv, &so_pendfree_lock);
1.98 christos 187: if (error) {
188: len = 0;
189: break;
190: }
1.80 yamt 191: }
1.93 yamt 192: socurkva += len;
1.136 ad 193: mutex_exit(&so_pendfree_lock);
1.98 christos 194: return len;
1.95 yamt 195: }
196:
1.113 thorpej 197: static void
1.95 yamt 198: sokvaunreserve(vsize_t len)
199: {
200:
1.136 ad 201: mutex_enter(&so_pendfree_lock);
1.95 yamt 202: socurkva -= len;
1.136 ad 203: cv_broadcast(&socurkva_cv);
204: mutex_exit(&so_pendfree_lock);
1.95 yamt 205: }
206:
207: /*
208: * sokvaalloc: allocate kva for loan.
209: */
210:
211: vaddr_t
212: sokvaalloc(vsize_t len, struct socket *so)
213: {
214: vaddr_t lva;
215:
216: /*
217: * reserve kva.
218: */
219:
1.98 christos 220: if (sokvareserve(so, len) == 0)
221: return 0;
1.93 yamt 222:
223: /*
224: * allocate kva.
225: */
1.80 yamt 226:
1.109 yamt 227: lva = uvm_km_alloc(kernel_map, len, 0, UVM_KMF_VAONLY | UVM_KMF_WAITVA);
1.95 yamt 228: if (lva == 0) {
229: sokvaunreserve(len);
1.80 yamt 230: return (0);
1.95 yamt 231: }
1.80 yamt 232:
233: return lva;
234: }
235:
1.93 yamt 236: /*
237: * sokvafree: free kva for loan.
238: */
239:
1.80 yamt 240: void
241: sokvafree(vaddr_t sva, vsize_t len)
242: {
1.93 yamt 243:
244: /*
245: * free kva.
246: */
1.80 yamt 247:
1.109 yamt 248: uvm_km_free(kernel_map, sva, len, UVM_KMF_VAONLY);
1.93 yamt 249:
250: /*
251: * unreserve kva.
252: */
253:
1.95 yamt 254: sokvaunreserve(len);
1.80 yamt 255: }
256:
1.64 thorpej 257: static void
1.134 christos 258: sodoloanfree(struct vm_page **pgs, void *buf, size_t size)
1.64 thorpej 259: {
1.156 yamt 260: vaddr_t sva, eva;
1.64 thorpej 261: vsize_t len;
1.156 yamt 262: int npgs;
263:
264: KASSERT(pgs != NULL);
1.64 thorpej 265:
266: eva = round_page((vaddr_t) buf + size);
267: sva = trunc_page((vaddr_t) buf);
268: len = eva - sva;
269: npgs = len >> PAGE_SHIFT;
270:
271: pmap_kremove(sva, len);
272: pmap_update(pmap_kernel());
273: uvm_unloan(pgs, npgs, UVM_LOAN_TOPAGE);
1.80 yamt 274: sokvafree(sva, len);
1.64 thorpej 275: }
276:
277: static size_t
1.152 matt 278: sodopendfree(void)
1.64 thorpej 279: {
1.93 yamt 280: size_t rv;
1.64 thorpej 281:
1.160 ad 282: if (__predict_true(so_pendfree == NULL))
283: return 0;
284:
1.136 ad 285: mutex_enter(&so_pendfree_lock);
1.117 yamt 286: rv = sodopendfreel();
1.136 ad 287: mutex_exit(&so_pendfree_lock);
1.93 yamt 288:
289: return rv;
290: }
291:
292: /*
293: * sodopendfreel: free mbufs on "pendfree" list.
1.136 ad 294: * unlock and relock so_pendfree_lock when freeing mbufs.
1.93 yamt 295: *
1.136 ad 296: * => called with so_pendfree_lock held.
1.93 yamt 297: */
298:
299: static size_t
1.152 matt 300: sodopendfreel(void)
1.93 yamt 301: {
1.137 ad 302: struct mbuf *m, *next;
1.93 yamt 303: size_t rv = 0;
304:
1.136 ad 305: KASSERT(mutex_owned(&so_pendfree_lock));
1.64 thorpej 306:
1.137 ad 307: while (so_pendfree != NULL) {
1.64 thorpej 308: m = so_pendfree;
1.93 yamt 309: so_pendfree = NULL;
1.136 ad 310: mutex_exit(&so_pendfree_lock);
1.93 yamt 311:
312: for (; m != NULL; m = next) {
313: next = m->m_next;
1.156 yamt 314: KASSERT((~m->m_flags & (M_EXT|M_EXT_PAGES)) == 0);
315: KASSERT(m->m_ext.ext_refcnt == 0);
1.93 yamt 316:
317: rv += m->m_ext.ext_size;
1.156 yamt 318: sodoloanfree(m->m_ext.ext_pgs, m->m_ext.ext_buf,
1.93 yamt 319: m->m_ext.ext_size);
1.145 ad 320: pool_cache_put(mb_cache, m);
1.93 yamt 321: }
1.64 thorpej 322:
1.136 ad 323: mutex_enter(&so_pendfree_lock);
1.64 thorpej 324: }
325:
326: return (rv);
327: }
328:
1.80 yamt 329: void
1.134 christos 330: soloanfree(struct mbuf *m, void *buf, size_t size, void *arg)
1.64 thorpej 331: {
332:
1.156 yamt 333: KASSERT(m != NULL);
1.64 thorpej 334:
1.93 yamt 335: /*
336: * postpone freeing mbuf.
337: *
338: * we can't do it in interrupt context
339: * because we need to put kva back to kernel_map.
340: */
341:
1.136 ad 342: mutex_enter(&so_pendfree_lock);
1.92 yamt 343: m->m_next = so_pendfree;
344: so_pendfree = m;
1.136 ad 345: cv_broadcast(&socurkva_cv);
346: mutex_exit(&so_pendfree_lock);
1.64 thorpej 347: }
348:
349: static long
350: sosend_loan(struct socket *so, struct uio *uio, struct mbuf *m, long space)
351: {
352: struct iovec *iov = uio->uio_iov;
353: vaddr_t sva, eva;
354: vsize_t len;
1.156 yamt 355: vaddr_t lva;
356: int npgs, error;
357: vaddr_t va;
358: int i;
1.64 thorpej 359:
1.116 yamt 360: if (VMSPACE_IS_KERNEL_P(uio->uio_vmspace))
1.64 thorpej 361: return (0);
362:
363: if (iov->iov_len < (size_t) space)
364: space = iov->iov_len;
365: if (space > SOCK_LOAN_CHUNK)
366: space = SOCK_LOAN_CHUNK;
367:
368: eva = round_page((vaddr_t) iov->iov_base + space);
369: sva = trunc_page((vaddr_t) iov->iov_base);
370: len = eva - sva;
371: npgs = len >> PAGE_SHIFT;
372:
1.79 thorpej 373: KASSERT(npgs <= M_EXT_MAXPAGES);
374:
1.80 yamt 375: lva = sokvaalloc(len, so);
1.64 thorpej 376: if (lva == 0)
1.80 yamt 377: return 0;
1.64 thorpej 378:
1.116 yamt 379: error = uvm_loan(&uio->uio_vmspace->vm_map, sva, len,
1.79 thorpej 380: m->m_ext.ext_pgs, UVM_LOAN_TOPAGE);
1.64 thorpej 381: if (error) {
1.80 yamt 382: sokvafree(lva, len);
1.64 thorpej 383: return (0);
384: }
385:
386: for (i = 0, va = lva; i < npgs; i++, va += PAGE_SIZE)
1.79 thorpej 387: pmap_kenter_pa(va, VM_PAGE_TO_PHYS(m->m_ext.ext_pgs[i]),
388: VM_PROT_READ);
1.64 thorpej 389: pmap_update(pmap_kernel());
390:
391: lva += (vaddr_t) iov->iov_base & PAGE_MASK;
392:
1.134 christos 393: MEXTADD(m, (void *) lva, space, M_MBUF, soloanfree, so);
1.79 thorpej 394: m->m_flags |= M_EXT_PAGES | M_EXT_ROMAP;
1.64 thorpej 395:
396: uio->uio_resid -= space;
397: /* uio_offset not updated, not set/used for write(2) */
1.134 christos 398: uio->uio_iov->iov_base = (char *)uio->uio_iov->iov_base + space;
1.64 thorpej 399: uio->uio_iov->iov_len -= space;
400: if (uio->uio_iov->iov_len == 0) {
401: uio->uio_iov++;
402: uio->uio_iovcnt--;
403: }
404:
405: return (space);
406: }
407:
1.119 yamt 408: static int
1.129 yamt 409: sokva_reclaim_callback(struct callback_entry *ce, void *obj, void *arg)
1.119 yamt 410: {
411:
412: KASSERT(ce == &sokva_reclaimerentry);
413: KASSERT(obj == NULL);
414:
415: sodopendfree();
416: if (!vm_map_starved_p(kernel_map)) {
417: return CALLBACK_CHAIN_ABORT;
418: }
419: return CALLBACK_CHAIN_CONTINUE;
420: }
421:
1.142 dyoung 422: struct mbuf *
1.147 dyoung 423: getsombuf(struct socket *so, int type)
1.142 dyoung 424: {
425: struct mbuf *m;
426:
1.147 dyoung 427: m = m_get(M_WAIT, type);
1.142 dyoung 428: MCLAIM(m, so->so_mowner);
429: return m;
430: }
431:
1.119 yamt 432: void
433: soinit(void)
434: {
435:
1.178 pooka 436: sysctl_kern_somaxkva_setup();
437:
1.148 ad 438: mutex_init(&so_pendfree_lock, MUTEX_DEFAULT, IPL_VM);
1.160 ad 439: softnet_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NONE);
1.136 ad 440: cv_init(&socurkva_cv, "sokva");
1.166 ad 441: soinit2();
1.136 ad 442:
1.119 yamt 443: /* Set the initial adjusted socket buffer size. */
444: if (sb_max_set(sb_max))
445: panic("bad initial sb_max value: %lu", sb_max);
446:
447: callback_register(&vm_map_to_kernel(kernel_map)->vmk_reclaim_callback,
448: &sokva_reclaimerentry, NULL, sokva_reclaim_callback);
449: }
450:
1.1 cgd 451: /*
452: * Socket operation routines.
453: * These routines are called by the routines in
454: * sys_socket.c or from a system process, and
455: * implement the semantics of socket operations by
456: * switching out to the protocol specific routines.
457: */
458: /*ARGSUSED*/
1.3 andrew 459: int
1.160 ad 460: socreate(int dom, struct socket **aso, int type, int proto, struct lwp *l,
461: struct socket *lockso)
1.1 cgd 462: {
1.99 matt 463: const struct protosw *prp;
1.54 lukem 464: struct socket *so;
1.115 yamt 465: uid_t uid;
1.160 ad 466: int error;
467: kmutex_t *lock;
1.1 cgd 468:
1.132 elad 469: error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET,
470: KAUTH_REQ_NETWORK_SOCKET_OPEN, KAUTH_ARG(dom), KAUTH_ARG(type),
471: KAUTH_ARG(proto));
1.140 dyoung 472: if (error != 0)
473: return error;
1.127 elad 474:
1.1 cgd 475: if (proto)
476: prp = pffindproto(dom, proto, type);
477: else
478: prp = pffindtype(dom, type);
1.140 dyoung 479: if (prp == NULL) {
1.120 ginsbach 480: /* no support for domain */
481: if (pffinddomain(dom) == 0)
1.140 dyoung 482: return EAFNOSUPPORT;
1.120 ginsbach 483: /* no support for socket type */
484: if (proto == 0 && type != 0)
1.140 dyoung 485: return EPROTOTYPE;
486: return EPROTONOSUPPORT;
1.120 ginsbach 487: }
1.140 dyoung 488: if (prp->pr_usrreq == NULL)
489: return EPROTONOSUPPORT;
1.1 cgd 490: if (prp->pr_type != type)
1.140 dyoung 491: return EPROTOTYPE;
1.160 ad 492:
493: so = soget(true);
1.1 cgd 494: so->so_type = type;
495: so->so_proto = prp;
1.33 matt 496: so->so_send = sosend;
497: so->so_receive = soreceive;
1.78 matt 498: #ifdef MBUFTRACE
499: so->so_rcv.sb_mowner = &prp->pr_domain->dom_mowner;
500: so->so_snd.sb_mowner = &prp->pr_domain->dom_mowner;
501: so->so_mowner = &prp->pr_domain->dom_mowner;
502: #endif
1.138 rmind 503: uid = kauth_cred_geteuid(l->l_cred);
1.115 yamt 504: so->so_uidinfo = uid_find(uid);
1.168 yamt 505: so->so_egid = kauth_cred_getegid(l->l_cred);
506: so->so_cpid = l->l_proc->p_pid;
1.160 ad 507: if (lockso != NULL) {
508: /* Caller wants us to share a lock. */
509: lock = lockso->so_lock;
510: so->so_lock = lock;
511: mutex_obj_hold(lock);
512: mutex_enter(lock);
513: } else {
514: /* Lock assigned and taken during PRU_ATTACH. */
515: }
1.140 dyoung 516: error = (*prp->pr_usrreq)(so, PRU_ATTACH, NULL,
517: (struct mbuf *)(long)proto, NULL, l);
1.160 ad 518: KASSERT(solocked(so));
1.140 dyoung 519: if (error != 0) {
1.1 cgd 520: so->so_state |= SS_NOFDREF;
521: sofree(so);
1.140 dyoung 522: return error;
1.1 cgd 523: }
1.160 ad 524: sounlock(so);
1.1 cgd 525: *aso = so;
1.140 dyoung 526: return 0;
1.1 cgd 527: }
528:
1.142 dyoung 529: /* On success, write file descriptor to fdout and return zero. On
530: * failure, return non-zero; *fdout will be undefined.
531: */
532: int
533: fsocreate(int domain, struct socket **sop, int type, int protocol,
534: struct lwp *l, int *fdout)
535: {
536: struct socket *so;
537: struct file *fp;
538: int fd, error;
539:
1.155 ad 540: if ((error = fd_allocfile(&fp, &fd)) != 0)
1.142 dyoung 541: return (error);
542: fp->f_flag = FREAD|FWRITE;
543: fp->f_type = DTYPE_SOCKET;
544: fp->f_ops = &socketops;
1.160 ad 545: error = socreate(domain, &so, type, protocol, l, NULL);
1.142 dyoung 546: if (error != 0) {
1.155 ad 547: fd_abort(curproc, fp, fd);
1.142 dyoung 548: } else {
549: if (sop != NULL)
550: *sop = so;
551: fp->f_data = so;
1.155 ad 552: fd_affix(curproc, fp, fd);
1.142 dyoung 553: *fdout = fd;
554: }
555: return error;
556: }
557:
1.3 andrew 558: int
1.114 christos 559: sobind(struct socket *so, struct mbuf *nam, struct lwp *l)
1.1 cgd 560: {
1.160 ad 561: int error;
1.1 cgd 562:
1.160 ad 563: solock(so);
1.140 dyoung 564: error = (*so->so_proto->pr_usrreq)(so, PRU_BIND, NULL, nam, NULL, l);
1.160 ad 565: sounlock(so);
1.140 dyoung 566: return error;
1.1 cgd 567: }
568:
1.3 andrew 569: int
1.150 elad 570: solisten(struct socket *so, int backlog, struct lwp *l)
1.1 cgd 571: {
1.160 ad 572: int error;
1.1 cgd 573:
1.160 ad 574: solock(so);
1.158 ad 575: if ((so->so_state & (SS_ISCONNECTED | SS_ISCONNECTING |
1.163 ad 576: SS_ISDISCONNECTING)) != 0) {
577: sounlock(so);
1.158 ad 578: return (EOPNOTSUPP);
1.163 ad 579: }
1.140 dyoung 580: error = (*so->so_proto->pr_usrreq)(so, PRU_LISTEN, NULL,
1.150 elad 581: NULL, NULL, l);
1.140 dyoung 582: if (error != 0) {
1.160 ad 583: sounlock(so);
1.140 dyoung 584: return error;
1.1 cgd 585: }
1.63 matt 586: if (TAILQ_EMPTY(&so->so_q))
1.1 cgd 587: so->so_options |= SO_ACCEPTCONN;
588: if (backlog < 0)
589: backlog = 0;
1.49 jonathan 590: so->so_qlimit = min(backlog, somaxconn);
1.160 ad 591: sounlock(so);
1.140 dyoung 592: return 0;
1.1 cgd 593: }
594:
1.21 christos 595: void
1.54 lukem 596: sofree(struct socket *so)
1.1 cgd 597: {
1.161 ad 598: u_int refs;
1.1 cgd 599:
1.160 ad 600: KASSERT(solocked(so));
601:
602: if (so->so_pcb || (so->so_state & SS_NOFDREF) == 0) {
603: sounlock(so);
1.1 cgd 604: return;
1.160 ad 605: }
1.43 mycroft 606: if (so->so_head) {
607: /*
608: * We must not decommission a socket that's on the accept(2)
609: * queue. If we do, then accept(2) may hang after select(2)
610: * indicated that the listening socket was ready.
611: */
1.160 ad 612: if (!soqremque(so, 0)) {
613: sounlock(so);
1.43 mycroft 614: return;
1.160 ad 615: }
1.43 mycroft 616: }
1.98 christos 617: if (so->so_rcv.sb_hiwat)
1.110 christos 618: (void)chgsbsize(so->so_uidinfo, &so->so_rcv.sb_hiwat, 0,
1.98 christos 619: RLIM_INFINITY);
620: if (so->so_snd.sb_hiwat)
1.110 christos 621: (void)chgsbsize(so->so_uidinfo, &so->so_snd.sb_hiwat, 0,
1.98 christos 622: RLIM_INFINITY);
623: sbrelease(&so->so_snd, so);
1.160 ad 624: KASSERT(!cv_has_waiters(&so->so_cv));
625: KASSERT(!cv_has_waiters(&so->so_rcv.sb_cv));
626: KASSERT(!cv_has_waiters(&so->so_snd.sb_cv));
1.1 cgd 627: sorflush(so);
1.161 ad 628: refs = so->so_aborting; /* XXX */
1.177 ad 629: /* Remove acccept filter if one is present. */
1.170 tls 630: if (so->so_accf != NULL)
1.177 ad 631: (void)accept_filt_clear(so);
1.160 ad 632: sounlock(so);
1.161 ad 633: if (refs == 0) /* XXX */
634: soput(so);
1.1 cgd 635: }
636:
637: /*
638: * Close a socket on last file table reference removal.
639: * Initiate disconnect if connected.
640: * Free socket when disconnect complete.
641: */
1.3 andrew 642: int
1.54 lukem 643: soclose(struct socket *so)
1.1 cgd 644: {
1.54 lukem 645: struct socket *so2;
1.160 ad 646: int error;
647: int error2;
1.1 cgd 648:
1.54 lukem 649: error = 0;
1.160 ad 650: solock(so);
1.1 cgd 651: if (so->so_options & SO_ACCEPTCONN) {
1.172 ad 652: for (;;) {
653: if ((so2 = TAILQ_FIRST(&so->so_q0)) != 0) {
1.160 ad 654: KASSERT(solocked2(so, so2));
655: (void) soqremque(so2, 0);
656: /* soabort drops the lock. */
657: (void) soabort(so2);
658: solock(so);
1.172 ad 659: continue;
1.160 ad 660: }
1.172 ad 661: if ((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);
1.172 ad 667: continue;
1.160 ad 668: }
1.172 ad 669: break;
670: }
1.1 cgd 671: }
672: if (so->so_pcb == 0)
673: goto discard;
674: if (so->so_state & SS_ISCONNECTED) {
675: if ((so->so_state & SS_ISDISCONNECTING) == 0) {
676: error = sodisconnect(so);
677: if (error)
678: goto drop;
679: }
680: if (so->so_options & SO_LINGER) {
1.151 ad 681: if ((so->so_state & SS_ISDISCONNECTING) && so->so_nbio)
1.1 cgd 682: goto drop;
1.21 christos 683: while (so->so_state & SS_ISCONNECTED) {
1.160 ad 684: error = sowait(so, so->so_linger * hz);
1.21 christos 685: if (error)
1.1 cgd 686: break;
1.21 christos 687: }
1.1 cgd 688: }
689: }
1.54 lukem 690: drop:
1.1 cgd 691: if (so->so_pcb) {
1.160 ad 692: error2 = (*so->so_proto->pr_usrreq)(so, PRU_DETACH,
1.140 dyoung 693: NULL, NULL, NULL, NULL);
1.1 cgd 694: if (error == 0)
695: error = error2;
696: }
1.54 lukem 697: discard:
1.1 cgd 698: if (so->so_state & SS_NOFDREF)
699: panic("soclose: NOFDREF");
700: so->so_state |= SS_NOFDREF;
701: sofree(so);
702: return (error);
703: }
704:
705: /*
1.160 ad 706: * Must be called with the socket locked.. Will return with it unlocked.
1.1 cgd 707: */
1.3 andrew 708: int
1.54 lukem 709: soabort(struct socket *so)
1.1 cgd 710: {
1.161 ad 711: u_int refs;
1.139 yamt 712: int error;
1.160 ad 713:
714: KASSERT(solocked(so));
715: KASSERT(so->so_head == NULL);
1.1 cgd 716:
1.161 ad 717: so->so_aborting++; /* XXX */
1.140 dyoung 718: error = (*so->so_proto->pr_usrreq)(so, PRU_ABORT, NULL,
719: NULL, NULL, NULL);
1.161 ad 720: refs = --so->so_aborting; /* XXX */
1.164 drochner 721: if (error || (refs == 0)) {
1.139 yamt 722: sofree(so);
1.160 ad 723: } else {
724: sounlock(so);
1.139 yamt 725: }
726: return error;
1.1 cgd 727: }
728:
1.3 andrew 729: int
1.54 lukem 730: soaccept(struct socket *so, struct mbuf *nam)
1.1 cgd 731: {
1.160 ad 732: int error;
733:
734: KASSERT(solocked(so));
1.1 cgd 735:
1.54 lukem 736: error = 0;
1.1 cgd 737: if ((so->so_state & SS_NOFDREF) == 0)
738: panic("soaccept: !NOFDREF");
739: so->so_state &= ~SS_NOFDREF;
1.55 thorpej 740: if ((so->so_state & SS_ISDISCONNECTED) == 0 ||
741: (so->so_proto->pr_flags & PR_ABRTACPTDIS) == 0)
1.41 mycroft 742: error = (*so->so_proto->pr_usrreq)(so, PRU_ACCEPT,
1.140 dyoung 743: NULL, nam, NULL, NULL);
1.41 mycroft 744: else
1.53 itojun 745: error = ECONNABORTED;
1.52 itojun 746:
1.1 cgd 747: return (error);
748: }
749:
1.3 andrew 750: int
1.114 christos 751: soconnect(struct socket *so, struct mbuf *nam, struct lwp *l)
1.1 cgd 752: {
1.160 ad 753: int error;
754:
755: KASSERT(solocked(so));
1.1 cgd 756:
757: if (so->so_options & SO_ACCEPTCONN)
758: return (EOPNOTSUPP);
759: /*
760: * If protocol is connection-based, can only connect once.
761: * Otherwise, if connected, try to disconnect first.
762: * This allows user to disconnect by connecting to, e.g.,
763: * a null address.
764: */
765: if (so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING) &&
766: ((so->so_proto->pr_flags & PR_CONNREQUIRED) ||
767: (error = sodisconnect(so))))
768: error = EISCONN;
769: else
770: error = (*so->so_proto->pr_usrreq)(so, PRU_CONNECT,
1.140 dyoung 771: NULL, nam, NULL, l);
1.1 cgd 772: return (error);
773: }
774:
1.3 andrew 775: int
1.54 lukem 776: soconnect2(struct socket *so1, struct socket *so2)
1.1 cgd 777: {
1.160 ad 778: int error;
779:
780: KASSERT(solocked2(so1, so2));
1.1 cgd 781:
1.22 mycroft 782: error = (*so1->so_proto->pr_usrreq)(so1, PRU_CONNECT2,
1.140 dyoung 783: NULL, (struct mbuf *)so2, NULL, NULL);
1.1 cgd 784: return (error);
785: }
786:
1.3 andrew 787: int
1.54 lukem 788: sodisconnect(struct socket *so)
1.1 cgd 789: {
1.160 ad 790: int error;
791:
792: KASSERT(solocked(so));
1.1 cgd 793:
794: if ((so->so_state & SS_ISCONNECTED) == 0) {
795: error = ENOTCONN;
1.160 ad 796: } else if (so->so_state & SS_ISDISCONNECTING) {
1.1 cgd 797: error = EALREADY;
1.160 ad 798: } else {
799: error = (*so->so_proto->pr_usrreq)(so, PRU_DISCONNECT,
800: NULL, NULL, NULL, NULL);
1.1 cgd 801: }
1.117 yamt 802: sodopendfree();
1.1 cgd 803: return (error);
804: }
805:
1.15 mycroft 806: #define SBLOCKWAIT(f) (((f) & MSG_DONTWAIT) ? M_NOWAIT : M_WAITOK)
1.1 cgd 807: /*
808: * Send on a socket.
809: * If send must go all at once and message is larger than
810: * send buffering, then hard error.
811: * Lock against other senders.
812: * If must go all at once and not enough room now, then
813: * inform user that this would block and do nothing.
814: * Otherwise, if nonblocking, send as much as possible.
815: * The data to be sent is described by "uio" if nonzero,
816: * otherwise by the mbuf chain "top" (which must be null
817: * if uio is not). Data provided in mbuf chain must be small
818: * enough to send all at once.
819: *
820: * Returns nonzero on error, timeout or signal; callers
821: * must check for short counts if EINTR/ERESTART are returned.
822: * Data and control buffers are freed on return.
823: */
1.3 andrew 824: int
1.54 lukem 825: sosend(struct socket *so, struct mbuf *addr, struct uio *uio, struct mbuf *top,
1.114 christos 826: struct mbuf *control, int flags, struct lwp *l)
1.1 cgd 827: {
1.54 lukem 828: struct mbuf **mp, *m;
1.114 christos 829: struct proc *p;
1.58 jdolecek 830: long space, len, resid, clen, mlen;
831: int error, s, dontroute, atomic;
1.54 lukem 832:
1.114 christos 833: p = l->l_proc;
1.117 yamt 834: sodopendfree();
1.160 ad 835: clen = 0;
1.64 thorpej 836:
1.160 ad 837: /*
838: * solock() provides atomicity of access. splsoftnet() prevents
839: * protocol processing soft interrupts from interrupting us and
840: * blocking (expensive).
841: */
842: s = splsoftnet();
843: solock(so);
1.54 lukem 844: atomic = sosendallatonce(so) || top;
1.1 cgd 845: if (uio)
846: resid = uio->uio_resid;
847: else
848: resid = top->m_pkthdr.len;
1.7 cgd 849: /*
850: * In theory resid should be unsigned.
851: * However, space must be signed, as it might be less than 0
852: * if we over-committed, and we must use a signed comparison
853: * of space and resid. On the other hand, a negative resid
854: * causes us to loop sending 0-length segments to the protocol.
855: */
1.29 mycroft 856: if (resid < 0) {
857: error = EINVAL;
858: goto out;
859: }
1.1 cgd 860: dontroute =
861: (flags & MSG_DONTROUTE) && (so->so_options & SO_DONTROUTE) == 0 &&
862: (so->so_proto->pr_flags & PR_ATOMIC);
1.165 christos 863: l->l_ru.ru_msgsnd++;
1.1 cgd 864: if (control)
865: clen = control->m_len;
1.54 lukem 866: restart:
1.21 christos 867: if ((error = sblock(&so->so_snd, SBLOCKWAIT(flags))) != 0)
1.1 cgd 868: goto out;
869: do {
1.160 ad 870: if (so->so_state & SS_CANTSENDMORE) {
871: error = EPIPE;
872: goto release;
873: }
1.48 thorpej 874: if (so->so_error) {
875: error = so->so_error;
876: so->so_error = 0;
877: goto release;
878: }
1.1 cgd 879: if ((so->so_state & SS_ISCONNECTED) == 0) {
880: if (so->so_proto->pr_flags & PR_CONNREQUIRED) {
881: if ((so->so_state & SS_ISCONFIRMING) == 0 &&
1.160 ad 882: !(resid == 0 && clen != 0)) {
883: error = ENOTCONN;
884: goto release;
885: }
886: } else if (addr == 0) {
887: error = EDESTADDRREQ;
888: goto release;
889: }
1.1 cgd 890: }
891: space = sbspace(&so->so_snd);
892: if (flags & MSG_OOB)
893: space += 1024;
1.21 christos 894: if ((atomic && resid > so->so_snd.sb_hiwat) ||
1.160 ad 895: clen > so->so_snd.sb_hiwat) {
896: error = EMSGSIZE;
897: goto release;
898: }
1.96 mycroft 899: if (space < resid + clen &&
1.1 cgd 900: (atomic || space < so->so_snd.sb_lowat || space < clen)) {
1.160 ad 901: if (so->so_nbio) {
902: error = EWOULDBLOCK;
903: goto release;
904: }
1.1 cgd 905: sbunlock(&so->so_snd);
906: error = sbwait(&so->so_snd);
907: if (error)
908: goto out;
909: goto restart;
910: }
911: mp = ⊤
912: space -= clen;
913: do {
1.45 tv 914: if (uio == NULL) {
915: /*
916: * Data is prepackaged in "top".
917: */
918: resid = 0;
919: if (flags & MSG_EOR)
920: top->m_flags |= M_EOR;
921: } else do {
1.160 ad 922: sounlock(so);
923: splx(s);
1.144 dyoung 924: if (top == NULL) {
1.78 matt 925: m = m_gethdr(M_WAIT, MT_DATA);
1.45 tv 926: mlen = MHLEN;
927: m->m_pkthdr.len = 0;
1.140 dyoung 928: m->m_pkthdr.rcvif = NULL;
1.45 tv 929: } else {
1.78 matt 930: m = m_get(M_WAIT, MT_DATA);
1.45 tv 931: mlen = MLEN;
932: }
1.78 matt 933: MCLAIM(m, so->so_snd.sb_mowner);
1.121 yamt 934: if (sock_loan_thresh >= 0 &&
935: uio->uio_iov->iov_len >= sock_loan_thresh &&
936: space >= sock_loan_thresh &&
1.64 thorpej 937: (len = sosend_loan(so, uio, m,
938: space)) != 0) {
939: SOSEND_COUNTER_INCR(&sosend_loan_big);
940: space -= len;
941: goto have_data;
942: }
1.45 tv 943: if (resid >= MINCLSIZE && space >= MCLBYTES) {
1.64 thorpej 944: SOSEND_COUNTER_INCR(&sosend_copy_big);
1.78 matt 945: m_clget(m, M_WAIT);
1.45 tv 946: if ((m->m_flags & M_EXT) == 0)
947: goto nopages;
948: mlen = MCLBYTES;
949: if (atomic && top == 0) {
1.58 jdolecek 950: len = lmin(MCLBYTES - max_hdr,
1.54 lukem 951: resid);
1.45 tv 952: m->m_data += max_hdr;
953: } else
1.58 jdolecek 954: len = lmin(MCLBYTES, resid);
1.45 tv 955: space -= len;
956: } else {
1.64 thorpej 957: nopages:
958: SOSEND_COUNTER_INCR(&sosend_copy_small);
1.58 jdolecek 959: len = lmin(lmin(mlen, resid), space);
1.45 tv 960: space -= len;
961: /*
962: * For datagram protocols, leave room
963: * for protocol headers in first mbuf.
964: */
965: if (atomic && top == 0 && len < mlen)
966: MH_ALIGN(m, len);
967: }
1.144 dyoung 968: error = uiomove(mtod(m, void *), (int)len, uio);
1.64 thorpej 969: have_data:
1.45 tv 970: resid = uio->uio_resid;
971: m->m_len = len;
972: *mp = m;
973: top->m_pkthdr.len += len;
1.160 ad 974: s = splsoftnet();
975: solock(so);
1.144 dyoung 976: if (error != 0)
1.45 tv 977: goto release;
978: mp = &m->m_next;
979: if (resid <= 0) {
980: if (flags & MSG_EOR)
981: top->m_flags |= M_EOR;
982: break;
983: }
984: } while (space > 0 && atomic);
1.108 perry 985:
1.160 ad 986: if (so->so_state & SS_CANTSENDMORE) {
987: error = EPIPE;
988: goto release;
989: }
1.45 tv 990: if (dontroute)
991: so->so_options |= SO_DONTROUTE;
992: if (resid > 0)
993: so->so_state |= SS_MORETOCOME;
1.46 sommerfe 994: error = (*so->so_proto->pr_usrreq)(so,
995: (flags & MSG_OOB) ? PRU_SENDOOB : PRU_SEND,
1.160 ad 996: top, addr, control, curlwp);
1.45 tv 997: if (dontroute)
998: so->so_options &= ~SO_DONTROUTE;
999: if (resid > 0)
1000: so->so_state &= ~SS_MORETOCOME;
1001: clen = 0;
1.144 dyoung 1002: control = NULL;
1003: top = NULL;
1.45 tv 1004: mp = ⊤
1.144 dyoung 1005: if (error != 0)
1.1 cgd 1006: goto release;
1007: } while (resid && space > 0);
1008: } while (resid);
1009:
1.54 lukem 1010: release:
1.1 cgd 1011: sbunlock(&so->so_snd);
1.54 lukem 1012: out:
1.160 ad 1013: sounlock(so);
1014: splx(s);
1.1 cgd 1015: if (top)
1016: m_freem(top);
1017: if (control)
1018: m_freem(control);
1019: return (error);
1020: }
1021:
1022: /*
1.159 ad 1023: * Following replacement or removal of the first mbuf on the first
1024: * mbuf chain of a socket buffer, push necessary state changes back
1025: * into the socket buffer so that other consumers see the values
1026: * consistently. 'nextrecord' is the callers locally stored value of
1027: * the original value of sb->sb_mb->m_nextpkt which must be restored
1028: * when the lead mbuf changes. NOTE: 'nextrecord' may be NULL.
1029: */
1030: static void
1031: sbsync(struct sockbuf *sb, struct mbuf *nextrecord)
1032: {
1033:
1.160 ad 1034: KASSERT(solocked(sb->sb_so));
1035:
1.159 ad 1036: /*
1037: * First, update for the new value of nextrecord. If necessary,
1038: * make it the first record.
1039: */
1040: if (sb->sb_mb != NULL)
1041: sb->sb_mb->m_nextpkt = nextrecord;
1042: else
1043: sb->sb_mb = nextrecord;
1044:
1045: /*
1046: * Now update any dependent socket buffer fields to reflect
1047: * the new state. This is an inline of SB_EMPTY_FIXUP, with
1048: * the addition of a second clause that takes care of the
1049: * case where sb_mb has been updated, but remains the last
1050: * record.
1051: */
1052: if (sb->sb_mb == NULL) {
1053: sb->sb_mbtail = NULL;
1054: sb->sb_lastrecord = NULL;
1055: } else if (sb->sb_mb->m_nextpkt == NULL)
1056: sb->sb_lastrecord = sb->sb_mb;
1057: }
1058:
1059: /*
1.1 cgd 1060: * Implement receive operations on a socket.
1061: * We depend on the way that records are added to the sockbuf
1062: * by sbappend*. In particular, each record (mbufs linked through m_next)
1063: * must begin with an address if the protocol so specifies,
1064: * followed by an optional mbuf or mbufs containing ancillary data,
1065: * and then zero or more mbufs of data.
1066: * In order to avoid blocking network interrupts for the entire time here,
1067: * we splx() while doing the actual copy to user space.
1068: * Although the sockbuf is locked, new data may still be appended,
1069: * and thus we must maintain consistency of the sockbuf during that time.
1070: *
1071: * The caller may receive the data as a single mbuf chain by supplying
1072: * an mbuf **mp0 for use in returning the chain. The uio is then used
1073: * only for the count in uio_resid.
1074: */
1.3 andrew 1075: int
1.54 lukem 1076: soreceive(struct socket *so, struct mbuf **paddr, struct uio *uio,
1077: struct mbuf **mp0, struct mbuf **controlp, int *flagsp)
1.1 cgd 1078: {
1.116 yamt 1079: struct lwp *l = curlwp;
1.160 ad 1080: struct mbuf *m, **mp, *mt;
1.146 dyoung 1081: int atomic, flags, len, error, s, offset, moff, type, orig_resid;
1.99 matt 1082: const struct protosw *pr;
1.54 lukem 1083: struct mbuf *nextrecord;
1.67 he 1084: int mbuf_removed = 0;
1.146 dyoung 1085: const struct domain *dom;
1.64 thorpej 1086:
1.54 lukem 1087: pr = so->so_proto;
1.146 dyoung 1088: atomic = pr->pr_flags & PR_ATOMIC;
1089: dom = pr->pr_domain;
1.1 cgd 1090: mp = mp0;
1.54 lukem 1091: type = 0;
1092: orig_resid = uio->uio_resid;
1.102 jonathan 1093:
1.144 dyoung 1094: if (paddr != NULL)
1095: *paddr = NULL;
1096: if (controlp != NULL)
1097: *controlp = NULL;
1098: if (flagsp != NULL)
1.1 cgd 1099: flags = *flagsp &~ MSG_EOR;
1100: else
1101: flags = 0;
1.66 enami 1102:
1103: if ((flags & MSG_DONTWAIT) == 0)
1.117 yamt 1104: sodopendfree();
1.66 enami 1105:
1.1 cgd 1106: if (flags & MSG_OOB) {
1107: m = m_get(M_WAIT, MT_DATA);
1.160 ad 1108: solock(so);
1.17 cgd 1109: error = (*pr->pr_usrreq)(so, PRU_RCVOOB, m,
1.140 dyoung 1110: (struct mbuf *)(long)(flags & MSG_PEEK), NULL, l);
1.160 ad 1111: sounlock(so);
1.1 cgd 1112: if (error)
1113: goto bad;
1114: do {
1.134 christos 1115: error = uiomove(mtod(m, void *),
1.1 cgd 1116: (int) min(uio->uio_resid, m->m_len), uio);
1117: m = m_free(m);
1.144 dyoung 1118: } while (uio->uio_resid > 0 && error == 0 && m);
1.54 lukem 1119: bad:
1.144 dyoung 1120: if (m != NULL)
1.1 cgd 1121: m_freem(m);
1.144 dyoung 1122: return error;
1.1 cgd 1123: }
1.144 dyoung 1124: if (mp != NULL)
1.140 dyoung 1125: *mp = NULL;
1.160 ad 1126:
1127: /*
1128: * solock() provides atomicity of access. splsoftnet() prevents
1129: * protocol processing soft interrupts from interrupting us and
1130: * blocking (expensive).
1131: */
1132: s = splsoftnet();
1133: solock(so);
1.1 cgd 1134: if (so->so_state & SS_ISCONFIRMING && uio->uio_resid)
1.140 dyoung 1135: (*pr->pr_usrreq)(so, PRU_RCVD, NULL, NULL, NULL, l);
1.1 cgd 1136:
1.54 lukem 1137: restart:
1.160 ad 1138: if ((error = sblock(&so->so_rcv, SBLOCKWAIT(flags))) != 0) {
1139: sounlock(so);
1140: splx(s);
1.144 dyoung 1141: return error;
1.160 ad 1142: }
1.1 cgd 1143:
1144: m = so->so_rcv.sb_mb;
1145: /*
1146: * If we have less data than requested, block awaiting more
1147: * (subject to any timeout) if:
1.15 mycroft 1148: * 1. the current count is less than the low water mark,
1.1 cgd 1149: * 2. MSG_WAITALL is set, and it is possible to do the entire
1.15 mycroft 1150: * receive operation at once if we block (resid <= hiwat), or
1151: * 3. MSG_DONTWAIT is not set.
1.1 cgd 1152: * If MSG_WAITALL is set but resid is larger than the receive buffer,
1153: * we have to do the receive in sections, and thus risk returning
1154: * a short count if a timeout or signal occurs after we start.
1155: */
1.144 dyoung 1156: if (m == NULL ||
1157: ((flags & MSG_DONTWAIT) == 0 &&
1158: so->so_rcv.sb_cc < uio->uio_resid &&
1159: (so->so_rcv.sb_cc < so->so_rcv.sb_lowat ||
1160: ((flags & MSG_WAITALL) &&
1161: uio->uio_resid <= so->so_rcv.sb_hiwat)) &&
1.146 dyoung 1162: m->m_nextpkt == NULL && !atomic)) {
1.1 cgd 1163: #ifdef DIAGNOSTIC
1.144 dyoung 1164: if (m == NULL && so->so_rcv.sb_cc)
1.1 cgd 1165: panic("receive 1");
1166: #endif
1167: if (so->so_error) {
1.144 dyoung 1168: if (m != NULL)
1.15 mycroft 1169: goto dontblock;
1.1 cgd 1170: error = so->so_error;
1171: if ((flags & MSG_PEEK) == 0)
1172: so->so_error = 0;
1173: goto release;
1174: }
1175: if (so->so_state & SS_CANTRCVMORE) {
1.144 dyoung 1176: if (m != NULL)
1.15 mycroft 1177: goto dontblock;
1.1 cgd 1178: else
1179: goto release;
1180: }
1.144 dyoung 1181: for (; m != NULL; m = m->m_next)
1.1 cgd 1182: if (m->m_type == MT_OOBDATA || (m->m_flags & M_EOR)) {
1183: m = so->so_rcv.sb_mb;
1184: goto dontblock;
1185: }
1186: if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) == 0 &&
1187: (so->so_proto->pr_flags & PR_CONNREQUIRED)) {
1188: error = ENOTCONN;
1189: goto release;
1190: }
1191: if (uio->uio_resid == 0)
1192: goto release;
1.151 ad 1193: if (so->so_nbio || (flags & MSG_DONTWAIT)) {
1.1 cgd 1194: error = EWOULDBLOCK;
1195: goto release;
1196: }
1.69 thorpej 1197: SBLASTRECORDCHK(&so->so_rcv, "soreceive sbwait 1");
1198: SBLASTMBUFCHK(&so->so_rcv, "soreceive sbwait 1");
1.1 cgd 1199: sbunlock(&so->so_rcv);
1200: error = sbwait(&so->so_rcv);
1.160 ad 1201: if (error != 0) {
1202: sounlock(so);
1203: splx(s);
1.144 dyoung 1204: return error;
1.160 ad 1205: }
1.1 cgd 1206: goto restart;
1207: }
1.54 lukem 1208: dontblock:
1.69 thorpej 1209: /*
1210: * On entry here, m points to the first record of the socket buffer.
1.159 ad 1211: * From this point onward, we maintain 'nextrecord' as a cache of the
1212: * pointer to the next record in the socket buffer. We must keep the
1213: * various socket buffer pointers and local stack versions of the
1214: * pointers in sync, pushing out modifications before dropping the
1.160 ad 1215: * socket lock, and re-reading them when picking it up.
1.159 ad 1216: *
1217: * Otherwise, we will race with the network stack appending new data
1218: * or records onto the socket buffer by using inconsistent/stale
1219: * versions of the field, possibly resulting in socket buffer
1220: * corruption.
1221: *
1222: * By holding the high-level sblock(), we prevent simultaneous
1223: * readers from pulling off the front of the socket buffer.
1.69 thorpej 1224: */
1.144 dyoung 1225: if (l != NULL)
1.157 ad 1226: l->l_ru.ru_msgrcv++;
1.69 thorpej 1227: KASSERT(m == so->so_rcv.sb_mb);
1228: SBLASTRECORDCHK(&so->so_rcv, "soreceive 1");
1229: SBLASTMBUFCHK(&so->so_rcv, "soreceive 1");
1.1 cgd 1230: nextrecord = m->m_nextpkt;
1231: if (pr->pr_flags & PR_ADDR) {
1232: #ifdef DIAGNOSTIC
1233: if (m->m_type != MT_SONAME)
1234: panic("receive 1a");
1235: #endif
1.3 andrew 1236: orig_resid = 0;
1.1 cgd 1237: if (flags & MSG_PEEK) {
1238: if (paddr)
1239: *paddr = m_copy(m, 0, m->m_len);
1240: m = m->m_next;
1241: } else {
1242: sbfree(&so->so_rcv, m);
1.67 he 1243: mbuf_removed = 1;
1.144 dyoung 1244: if (paddr != NULL) {
1.1 cgd 1245: *paddr = m;
1246: so->so_rcv.sb_mb = m->m_next;
1.144 dyoung 1247: m->m_next = NULL;
1.1 cgd 1248: m = so->so_rcv.sb_mb;
1249: } else {
1250: MFREE(m, so->so_rcv.sb_mb);
1251: m = so->so_rcv.sb_mb;
1252: }
1.159 ad 1253: sbsync(&so->so_rcv, nextrecord);
1.1 cgd 1254: }
1255: }
1.159 ad 1256:
1257: /*
1258: * Process one or more MT_CONTROL mbufs present before any data mbufs
1259: * in the first mbuf chain on the socket buffer. If MSG_PEEK, we
1260: * just copy the data; if !MSG_PEEK, we call into the protocol to
1261: * perform externalization (or freeing if controlp == NULL).
1262: */
1263: if (__predict_false(m != NULL && m->m_type == MT_CONTROL)) {
1264: struct mbuf *cm = NULL, *cmn;
1265: struct mbuf **cme = &cm;
1266:
1267: do {
1268: if (flags & MSG_PEEK) {
1269: if (controlp != NULL) {
1270: *controlp = m_copy(m, 0, m->m_len);
1271: controlp = &(*controlp)->m_next;
1272: }
1273: m = m->m_next;
1274: } else {
1275: sbfree(&so->so_rcv, m);
1.1 cgd 1276: so->so_rcv.sb_mb = m->m_next;
1.144 dyoung 1277: m->m_next = NULL;
1.159 ad 1278: *cme = m;
1279: cme = &(*cme)->m_next;
1.1 cgd 1280: m = so->so_rcv.sb_mb;
1.159 ad 1281: }
1282: } while (m != NULL && m->m_type == MT_CONTROL);
1283: if ((flags & MSG_PEEK) == 0)
1284: sbsync(&so->so_rcv, nextrecord);
1285: for (; cm != NULL; cm = cmn) {
1286: cmn = cm->m_next;
1287: cm->m_next = NULL;
1288: type = mtod(cm, struct cmsghdr *)->cmsg_type;
1289: if (controlp != NULL) {
1290: if (dom->dom_externalize != NULL &&
1291: type == SCM_RIGHTS) {
1.160 ad 1292: sounlock(so);
1.159 ad 1293: splx(s);
1294: error = (*dom->dom_externalize)(cm, l);
1295: s = splsoftnet();
1.160 ad 1296: solock(so);
1.159 ad 1297: }
1298: *controlp = cm;
1299: while (*controlp != NULL)
1300: controlp = &(*controlp)->m_next;
1.1 cgd 1301: } else {
1.106 itojun 1302: /*
1303: * Dispose of any SCM_RIGHTS message that went
1304: * through the read path rather than recv.
1305: */
1.159 ad 1306: if (dom->dom_dispose != NULL &&
1307: type == SCM_RIGHTS) {
1.160 ad 1308: sounlock(so);
1.159 ad 1309: (*dom->dom_dispose)(cm);
1.160 ad 1310: solock(so);
1.159 ad 1311: }
1312: m_freem(cm);
1.1 cgd 1313: }
1314: }
1.159 ad 1315: if (m != NULL)
1316: nextrecord = so->so_rcv.sb_mb->m_nextpkt;
1317: else
1318: nextrecord = so->so_rcv.sb_mb;
1319: orig_resid = 0;
1.1 cgd 1320: }
1.69 thorpej 1321:
1.159 ad 1322: /* If m is non-NULL, we have some data to read. */
1323: if (__predict_true(m != NULL)) {
1.1 cgd 1324: type = m->m_type;
1325: if (type == MT_OOBDATA)
1326: flags |= MSG_OOB;
1327: }
1.69 thorpej 1328: SBLASTRECORDCHK(&so->so_rcv, "soreceive 2");
1329: SBLASTMBUFCHK(&so->so_rcv, "soreceive 2");
1330:
1.1 cgd 1331: moff = 0;
1332: offset = 0;
1.144 dyoung 1333: while (m != NULL && uio->uio_resid > 0 && error == 0) {
1.1 cgd 1334: if (m->m_type == MT_OOBDATA) {
1335: if (type != MT_OOBDATA)
1336: break;
1337: } else if (type == MT_OOBDATA)
1338: break;
1339: #ifdef DIAGNOSTIC
1340: else if (m->m_type != MT_DATA && m->m_type != MT_HEADER)
1341: panic("receive 3");
1342: #endif
1343: so->so_state &= ~SS_RCVATMARK;
1344: len = uio->uio_resid;
1345: if (so->so_oobmark && len > so->so_oobmark - offset)
1346: len = so->so_oobmark - offset;
1347: if (len > m->m_len - moff)
1348: len = m->m_len - moff;
1349: /*
1350: * If mp is set, just pass back the mbufs.
1351: * Otherwise copy them out via the uio, then free.
1352: * Sockbuf must be consistent here (points to current mbuf,
1353: * it points to next record) when we drop priority;
1354: * we must note any additions to the sockbuf when we
1355: * block interrupts again.
1356: */
1.144 dyoung 1357: if (mp == NULL) {
1.69 thorpej 1358: SBLASTRECORDCHK(&so->so_rcv, "soreceive uiomove");
1359: SBLASTMBUFCHK(&so->so_rcv, "soreceive uiomove");
1.160 ad 1360: sounlock(so);
1.1 cgd 1361: splx(s);
1.134 christos 1362: error = uiomove(mtod(m, char *) + moff, (int)len, uio);
1.20 mycroft 1363: s = splsoftnet();
1.160 ad 1364: solock(so);
1.144 dyoung 1365: if (error != 0) {
1.67 he 1366: /*
1367: * If any part of the record has been removed
1368: * (such as the MT_SONAME mbuf, which will
1369: * happen when PR_ADDR, and thus also
1370: * PR_ATOMIC, is set), then drop the entire
1371: * record to maintain the atomicity of the
1372: * receive operation.
1373: *
1374: * This avoids a later panic("receive 1a")
1375: * when compiled with DIAGNOSTIC.
1376: */
1.146 dyoung 1377: if (m && mbuf_removed && atomic)
1.67 he 1378: (void) sbdroprecord(&so->so_rcv);
1379:
1.57 jdolecek 1380: goto release;
1.67 he 1381: }
1.1 cgd 1382: } else
1383: uio->uio_resid -= len;
1384: if (len == m->m_len - moff) {
1385: if (m->m_flags & M_EOR)
1386: flags |= MSG_EOR;
1387: if (flags & MSG_PEEK) {
1388: m = m->m_next;
1389: moff = 0;
1390: } else {
1391: nextrecord = m->m_nextpkt;
1392: sbfree(&so->so_rcv, m);
1393: if (mp) {
1394: *mp = m;
1395: mp = &m->m_next;
1396: so->so_rcv.sb_mb = m = m->m_next;
1.140 dyoung 1397: *mp = NULL;
1.1 cgd 1398: } else {
1399: MFREE(m, so->so_rcv.sb_mb);
1400: m = so->so_rcv.sb_mb;
1401: }
1.69 thorpej 1402: /*
1403: * If m != NULL, we also know that
1404: * so->so_rcv.sb_mb != NULL.
1405: */
1406: KASSERT(so->so_rcv.sb_mb == m);
1407: if (m) {
1.1 cgd 1408: m->m_nextpkt = nextrecord;
1.69 thorpej 1409: if (nextrecord == NULL)
1410: so->so_rcv.sb_lastrecord = m;
1411: } else {
1412: so->so_rcv.sb_mb = nextrecord;
1.70 thorpej 1413: SB_EMPTY_FIXUP(&so->so_rcv);
1.69 thorpej 1414: }
1415: SBLASTRECORDCHK(&so->so_rcv, "soreceive 3");
1416: SBLASTMBUFCHK(&so->so_rcv, "soreceive 3");
1.1 cgd 1417: }
1.144 dyoung 1418: } else if (flags & MSG_PEEK)
1419: moff += len;
1420: else {
1.160 ad 1421: if (mp != NULL) {
1422: mt = m_copym(m, 0, len, M_NOWAIT);
1423: if (__predict_false(mt == NULL)) {
1424: sounlock(so);
1425: mt = m_copym(m, 0, len, M_WAIT);
1426: solock(so);
1427: }
1428: *mp = mt;
1429: }
1.144 dyoung 1430: m->m_data += len;
1431: m->m_len -= len;
1432: so->so_rcv.sb_cc -= len;
1.1 cgd 1433: }
1434: if (so->so_oobmark) {
1435: if ((flags & MSG_PEEK) == 0) {
1436: so->so_oobmark -= len;
1437: if (so->so_oobmark == 0) {
1438: so->so_state |= SS_RCVATMARK;
1439: break;
1440: }
1.7 cgd 1441: } else {
1.1 cgd 1442: offset += len;
1.7 cgd 1443: if (offset == so->so_oobmark)
1444: break;
1445: }
1.1 cgd 1446: }
1447: if (flags & MSG_EOR)
1448: break;
1449: /*
1450: * If the MSG_WAITALL flag is set (for non-atomic socket),
1451: * we must not quit until "uio->uio_resid == 0" or an error
1452: * termination. If a signal/timeout occurs, return
1453: * with a short count but without error.
1454: * Keep sockbuf locked against other readers.
1455: */
1.144 dyoung 1456: while (flags & MSG_WAITALL && m == NULL && uio->uio_resid > 0 &&
1.3 andrew 1457: !sosendallatonce(so) && !nextrecord) {
1.1 cgd 1458: if (so->so_error || so->so_state & SS_CANTRCVMORE)
1459: break;
1.68 matt 1460: /*
1461: * If we are peeking and the socket receive buffer is
1462: * full, stop since we can't get more data to peek at.
1463: */
1464: if ((flags & MSG_PEEK) && sbspace(&so->so_rcv) <= 0)
1465: break;
1466: /*
1467: * If we've drained the socket buffer, tell the
1468: * protocol in case it needs to do something to
1469: * get it filled again.
1470: */
1471: if ((pr->pr_flags & PR_WANTRCVD) && so->so_pcb)
1472: (*pr->pr_usrreq)(so, PRU_RCVD,
1.140 dyoung 1473: NULL, (struct mbuf *)(long)flags, NULL, l);
1.69 thorpej 1474: SBLASTRECORDCHK(&so->so_rcv, "soreceive sbwait 2");
1475: SBLASTMBUFCHK(&so->so_rcv, "soreceive sbwait 2");
1.1 cgd 1476: error = sbwait(&so->so_rcv);
1.144 dyoung 1477: if (error != 0) {
1.1 cgd 1478: sbunlock(&so->so_rcv);
1.160 ad 1479: sounlock(so);
1.1 cgd 1480: splx(s);
1.144 dyoung 1481: return 0;
1.1 cgd 1482: }
1.21 christos 1483: if ((m = so->so_rcv.sb_mb) != NULL)
1.1 cgd 1484: nextrecord = m->m_nextpkt;
1485: }
1486: }
1.3 andrew 1487:
1.146 dyoung 1488: if (m && atomic) {
1.3 andrew 1489: flags |= MSG_TRUNC;
1490: if ((flags & MSG_PEEK) == 0)
1491: (void) sbdroprecord(&so->so_rcv);
1492: }
1.1 cgd 1493: if ((flags & MSG_PEEK) == 0) {
1.144 dyoung 1494: if (m == NULL) {
1.69 thorpej 1495: /*
1.70 thorpej 1496: * First part is an inline SB_EMPTY_FIXUP(). Second
1.69 thorpej 1497: * part makes sure sb_lastrecord is up-to-date if
1498: * there is still data in the socket buffer.
1499: */
1.1 cgd 1500: so->so_rcv.sb_mb = nextrecord;
1.69 thorpej 1501: if (so->so_rcv.sb_mb == NULL) {
1502: so->so_rcv.sb_mbtail = NULL;
1503: so->so_rcv.sb_lastrecord = NULL;
1504: } else if (nextrecord->m_nextpkt == NULL)
1505: so->so_rcv.sb_lastrecord = nextrecord;
1506: }
1507: SBLASTRECORDCHK(&so->so_rcv, "soreceive 4");
1508: SBLASTMBUFCHK(&so->so_rcv, "soreceive 4");
1.1 cgd 1509: if (pr->pr_flags & PR_WANTRCVD && so->so_pcb)
1.140 dyoung 1510: (*pr->pr_usrreq)(so, PRU_RCVD, NULL,
1511: (struct mbuf *)(long)flags, NULL, l);
1.1 cgd 1512: }
1.3 andrew 1513: if (orig_resid == uio->uio_resid && orig_resid &&
1514: (flags & MSG_EOR) == 0 && (so->so_state & SS_CANTRCVMORE) == 0) {
1515: sbunlock(&so->so_rcv);
1516: goto restart;
1517: }
1.108 perry 1518:
1.144 dyoung 1519: if (flagsp != NULL)
1.1 cgd 1520: *flagsp |= flags;
1.54 lukem 1521: release:
1.1 cgd 1522: sbunlock(&so->so_rcv);
1.160 ad 1523: sounlock(so);
1.1 cgd 1524: splx(s);
1.144 dyoung 1525: return error;
1.1 cgd 1526: }
1527:
1.14 mycroft 1528: int
1.54 lukem 1529: soshutdown(struct socket *so, int how)
1.1 cgd 1530: {
1.99 matt 1531: const struct protosw *pr;
1.160 ad 1532: int error;
1533:
1534: KASSERT(solocked(so));
1.34 kleink 1535:
1.54 lukem 1536: pr = so->so_proto;
1.34 kleink 1537: if (!(how == SHUT_RD || how == SHUT_WR || how == SHUT_RDWR))
1538: return (EINVAL);
1.1 cgd 1539:
1.160 ad 1540: if (how == SHUT_RD || how == SHUT_RDWR) {
1.1 cgd 1541: sorflush(so);
1.160 ad 1542: error = 0;
1543: }
1.34 kleink 1544: if (how == SHUT_WR || how == SHUT_RDWR)
1.160 ad 1545: error = (*pr->pr_usrreq)(so, PRU_SHUTDOWN, NULL,
1.140 dyoung 1546: NULL, NULL, NULL);
1.160 ad 1547:
1548: return error;
1.1 cgd 1549: }
1550:
1.14 mycroft 1551: void
1.54 lukem 1552: sorflush(struct socket *so)
1.1 cgd 1553: {
1.54 lukem 1554: struct sockbuf *sb, asb;
1.99 matt 1555: const struct protosw *pr;
1.160 ad 1556:
1557: KASSERT(solocked(so));
1.1 cgd 1558:
1.54 lukem 1559: sb = &so->so_rcv;
1560: pr = so->so_proto;
1.160 ad 1561: socantrcvmore(so);
1.1 cgd 1562: sb->sb_flags |= SB_NOINTR;
1.160 ad 1563: (void )sblock(sb, M_WAITOK);
1.1 cgd 1564: sbunlock(sb);
1565: asb = *sb;
1.86 wrstuden 1566: /*
1567: * Clear most of the sockbuf structure, but leave some of the
1568: * fields valid.
1569: */
1570: memset(&sb->sb_startzero, 0,
1571: sizeof(*sb) - offsetof(struct sockbuf, sb_startzero));
1.160 ad 1572: if (pr->pr_flags & PR_RIGHTS && pr->pr_domain->dom_dispose) {
1573: sounlock(so);
1.1 cgd 1574: (*pr->pr_domain->dom_dispose)(asb.sb_mb);
1.160 ad 1575: solock(so);
1576: }
1.98 christos 1577: sbrelease(&asb, so);
1.1 cgd 1578: }
1579:
1.171 plunky 1580: /*
1581: * internal set SOL_SOCKET options
1582: */
1.142 dyoung 1583: static int
1.171 plunky 1584: sosetopt1(struct socket *so, const struct sockopt *sopt)
1.1 cgd 1585: {
1.179 christos 1586: int error = 0, optval, opt;
1.171 plunky 1587: struct linger l;
1588: struct timeval tv;
1.142 dyoung 1589:
1.179 christos 1590: switch ((opt = sopt->sopt_name)) {
1.142 dyoung 1591:
1.170 tls 1592: case SO_ACCEPTFILTER:
1.177 ad 1593: error = accept_filt_setopt(so, sopt);
1594: KASSERT(solocked(so));
1.170 tls 1595: break;
1596:
1.171 plunky 1597: case SO_LINGER:
1598: error = sockopt_get(sopt, &l, sizeof(l));
1.177 ad 1599: solock(so);
1.171 plunky 1600: if (error)
1.177 ad 1601: break;
1.171 plunky 1602: if (l.l_linger < 0 || l.l_linger > USHRT_MAX ||
1.177 ad 1603: l.l_linger > (INT_MAX / hz)) {
1604: error = EDOM;
1605: break;
1606: }
1.171 plunky 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;
1.177 ad 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:
1.171 plunky 1623: error = sockopt_getint(sopt, &optval);
1.177 ad 1624: solock(so);
1.171 plunky 1625: if (error)
1.177 ad 1626: break;
1.171 plunky 1627: if (optval)
1.179 christos 1628: so->so_options |= opt;
1.142 dyoung 1629: else
1.179 christos 1630: so->so_options &= ~opt;
1.142 dyoung 1631: break;
1632:
1633: case SO_SNDBUF:
1634: case SO_RCVBUF:
1635: case SO_SNDLOWAT:
1636: case SO_RCVLOWAT:
1.171 plunky 1637: error = sockopt_getint(sopt, &optval);
1.177 ad 1638: solock(so);
1.171 plunky 1639: if (error)
1.177 ad 1640: break;
1.1 cgd 1641:
1.142 dyoung 1642: /*
1643: * Values < 1 make no sense for any of these
1644: * options, so disallow them.
1645: */
1.177 ad 1646: if (optval < 1) {
1647: error = EINVAL;
1648: break;
1649: }
1.1 cgd 1650:
1.179 christos 1651: switch (opt) {
1.171 plunky 1652: case SO_SNDBUF:
1.177 ad 1653: if (sbreserve(&so->so_snd, (u_long)optval, so) == 0) {
1654: error = ENOBUFS;
1655: break;
1656: }
1.171 plunky 1657: so->so_snd.sb_flags &= ~SB_AUTOSIZE;
1658: break;
1.1 cgd 1659:
1660: case SO_RCVBUF:
1.177 ad 1661: if (sbreserve(&so->so_rcv, (u_long)optval, so) == 0) {
1662: error = ENOBUFS;
1663: break;
1664: }
1.171 plunky 1665: so->so_rcv.sb_flags &= ~SB_AUTOSIZE;
1.142 dyoung 1666: break;
1667:
1668: /*
1669: * Make sure the low-water is never greater than
1670: * the high-water.
1671: */
1.1 cgd 1672: case SO_SNDLOWAT:
1.171 plunky 1673: if (optval > so->so_snd.sb_hiwat)
1674: optval = so->so_snd.sb_hiwat;
1675:
1676: so->so_snd.sb_lowat = optval;
1.142 dyoung 1677: break;
1.171 plunky 1678:
1.1 cgd 1679: case SO_RCVLOWAT:
1.171 plunky 1680: if (optval > so->so_rcv.sb_hiwat)
1681: optval = so->so_rcv.sb_hiwat;
1682:
1683: so->so_rcv.sb_lowat = optval;
1.142 dyoung 1684: break;
1685: }
1686: break;
1.28 thorpej 1687:
1.179 christos 1688: #ifdef COMPAT_50
1689: case SO_OSNDTIMEO:
1690: case SO_ORCVTIMEO: {
1691: struct timeval50 otv;
1692: error = sockopt_get(sopt, &otv, sizeof(otv));
1693: timeval50_to_timeval(&otv, &tv);
1694: opt = opt == SO_OSNDTIMEO ? SO_SNDTIMEO : SO_RCVTIMEO;
1695: error = 1;
1696: /*FALLTHROUGH*/
1697: }
1698: #endif /* COMPAT_50 */
1699:
1.142 dyoung 1700: case SO_SNDTIMEO:
1701: case SO_RCVTIMEO:
1.181 ! christos 1702: if (error == 0)
1.179 christos 1703: error = sockopt_get(sopt, &tv, sizeof(tv));
1.177 ad 1704: solock(so);
1.171 plunky 1705: if (error)
1.177 ad 1706: break;
1.171 plunky 1707:
1.177 ad 1708: if (tv.tv_sec > (INT_MAX - tv.tv_usec / tick) / hz) {
1709: error = EDOM;
1710: break;
1711: }
1.28 thorpej 1712:
1.171 plunky 1713: optval = tv.tv_sec * hz + tv.tv_usec / tick;
1714: if (optval == 0 && tv.tv_usec != 0)
1715: optval = 1;
1.28 thorpej 1716:
1.179 christos 1717: switch (opt) {
1.142 dyoung 1718: case SO_SNDTIMEO:
1.171 plunky 1719: so->so_snd.sb_timeo = optval;
1.1 cgd 1720: break;
1721: case SO_RCVTIMEO:
1.171 plunky 1722: so->so_rcv.sb_timeo = optval;
1.142 dyoung 1723: break;
1724: }
1725: break;
1.1 cgd 1726:
1.142 dyoung 1727: default:
1.177 ad 1728: solock(so);
1729: error = ENOPROTOOPT;
1730: break;
1.142 dyoung 1731: }
1.177 ad 1732: KASSERT(solocked(so));
1733: return error;
1.142 dyoung 1734: }
1.1 cgd 1735:
1.142 dyoung 1736: int
1.171 plunky 1737: sosetopt(struct socket *so, struct sockopt *sopt)
1.142 dyoung 1738: {
1739: int error, prerr;
1.1 cgd 1740:
1.177 ad 1741: if (sopt->sopt_level == SOL_SOCKET) {
1.171 plunky 1742: error = sosetopt1(so, sopt);
1.177 ad 1743: KASSERT(solocked(so));
1744: } else {
1.142 dyoung 1745: error = ENOPROTOOPT;
1.177 ad 1746: solock(so);
1747: }
1.1 cgd 1748:
1.142 dyoung 1749: if ((error == 0 || error == ENOPROTOOPT) &&
1750: so->so_proto != NULL && so->so_proto->pr_ctloutput != NULL) {
1751: /* give the protocol stack a shot */
1.171 plunky 1752: prerr = (*so->so_proto->pr_ctloutput)(PRCO_SETOPT, so, sopt);
1.142 dyoung 1753: if (prerr == 0)
1754: error = 0;
1755: else if (prerr != ENOPROTOOPT)
1756: error = prerr;
1.171 plunky 1757: }
1.160 ad 1758: sounlock(so);
1.142 dyoung 1759: return error;
1.1 cgd 1760: }
1761:
1.171 plunky 1762: /*
1763: * so_setsockopt() is a wrapper providing a sockopt structure for sosetopt()
1764: */
1765: int
1766: so_setsockopt(struct lwp *l, struct socket *so, int level, int name,
1767: const void *val, size_t valsize)
1768: {
1769: struct sockopt sopt;
1770: int error;
1771:
1772: KASSERT(valsize == 0 || val != NULL);
1773:
1774: sockopt_init(&sopt, level, name, valsize);
1775: sockopt_set(&sopt, val, valsize);
1776:
1777: error = sosetopt(so, &sopt);
1778:
1779: sockopt_destroy(&sopt);
1780:
1781: return error;
1782: }
1783:
1784: /*
1785: * internal get SOL_SOCKET options
1786: */
1787: static int
1788: sogetopt1(struct socket *so, struct sockopt *sopt)
1789: {
1.179 christos 1790: int error, optval, opt;
1.171 plunky 1791: struct linger l;
1792: struct timeval tv;
1793:
1.179 christos 1794: switch ((opt = sopt->sopt_name)) {
1.171 plunky 1795:
1796: case SO_ACCEPTFILTER:
1.177 ad 1797: error = accept_filt_getopt(so, sopt);
1.171 plunky 1798: break;
1799:
1800: case SO_LINGER:
1801: l.l_onoff = (so->so_options & SO_LINGER) ? 1 : 0;
1802: l.l_linger = so->so_linger;
1803:
1804: error = sockopt_set(sopt, &l, sizeof(l));
1805: break;
1806:
1807: case SO_USELOOPBACK:
1808: case SO_DONTROUTE:
1809: case SO_DEBUG:
1810: case SO_KEEPALIVE:
1811: case SO_REUSEADDR:
1812: case SO_REUSEPORT:
1813: case SO_BROADCAST:
1814: case SO_OOBINLINE:
1815: case SO_TIMESTAMP:
1.179 christos 1816: error = sockopt_setint(sopt, (so->so_options & opt) ? 1 : 0);
1.171 plunky 1817: break;
1818:
1819: case SO_TYPE:
1820: error = sockopt_setint(sopt, so->so_type);
1821: break;
1822:
1823: case SO_ERROR:
1824: error = sockopt_setint(sopt, so->so_error);
1825: so->so_error = 0;
1826: break;
1827:
1828: case SO_SNDBUF:
1829: error = sockopt_setint(sopt, so->so_snd.sb_hiwat);
1830: break;
1831:
1832: case SO_RCVBUF:
1833: error = sockopt_setint(sopt, so->so_rcv.sb_hiwat);
1834: break;
1835:
1836: case SO_SNDLOWAT:
1837: error = sockopt_setint(sopt, so->so_snd.sb_lowat);
1838: break;
1839:
1840: case SO_RCVLOWAT:
1841: error = sockopt_setint(sopt, so->so_rcv.sb_lowat);
1842: break;
1843:
1.179 christos 1844: #ifdef COMPAT_50
1845: case SO_OSNDTIMEO:
1846: case SO_ORCVTIMEO: {
1847: struct timeval50 otv;
1848:
1849: optval = (opt == SO_OSNDTIMEO ?
1850: so->so_snd.sb_timeo : so->so_rcv.sb_timeo);
1851:
1852: otv.tv_sec = optval / hz;
1853: otv.tv_usec = (optval % hz) * tick;
1854:
1855: error = sockopt_set(sopt, &otv, sizeof(otv));
1856: break;
1857: }
1858: #endif /* COMPAT_50 */
1859:
1.171 plunky 1860: case SO_SNDTIMEO:
1861: case SO_RCVTIMEO:
1.179 christos 1862: optval = (opt == SO_SNDTIMEO ?
1.171 plunky 1863: so->so_snd.sb_timeo : so->so_rcv.sb_timeo);
1864:
1865: tv.tv_sec = optval / hz;
1866: tv.tv_usec = (optval % hz) * tick;
1867:
1868: error = sockopt_set(sopt, &tv, sizeof(tv));
1869: break;
1870:
1871: case SO_OVERFLOWED:
1872: error = sockopt_setint(sopt, so->so_rcv.sb_overflowed);
1873: break;
1874:
1875: default:
1876: error = ENOPROTOOPT;
1877: break;
1878: }
1879:
1880: return (error);
1881: }
1882:
1.14 mycroft 1883: int
1.171 plunky 1884: sogetopt(struct socket *so, struct sockopt *sopt)
1.1 cgd 1885: {
1.160 ad 1886: int error;
1.1 cgd 1887:
1.160 ad 1888: solock(so);
1.171 plunky 1889: if (sopt->sopt_level != SOL_SOCKET) {
1.1 cgd 1890: if (so->so_proto && so->so_proto->pr_ctloutput) {
1.160 ad 1891: error = ((*so->so_proto->pr_ctloutput)
1.171 plunky 1892: (PRCO_GETOPT, so, sopt));
1.1 cgd 1893: } else
1.160 ad 1894: error = (ENOPROTOOPT);
1.1 cgd 1895: } else {
1.171 plunky 1896: error = sogetopt1(so, sopt);
1897: }
1898: sounlock(so);
1899: return (error);
1900: }
1901:
1902: /*
1903: * alloc sockopt data buffer buffer
1904: * - will be released at destroy
1905: */
1.176 plunky 1906: static int
1907: sockopt_alloc(struct sockopt *sopt, size_t len, km_flag_t kmflag)
1.171 plunky 1908: {
1909:
1910: KASSERT(sopt->sopt_size == 0);
1911:
1.176 plunky 1912: if (len > sizeof(sopt->sopt_buf)) {
1913: sopt->sopt_data = kmem_zalloc(len, kmflag);
1914: if (sopt->sopt_data == NULL)
1915: return ENOMEM;
1916: } else
1.171 plunky 1917: sopt->sopt_data = sopt->sopt_buf;
1918:
1919: sopt->sopt_size = len;
1.176 plunky 1920: return 0;
1.171 plunky 1921: }
1922:
1923: /*
1924: * initialise sockopt storage
1.176 plunky 1925: * - MAY sleep during allocation
1.171 plunky 1926: */
1927: void
1928: sockopt_init(struct sockopt *sopt, int level, int name, size_t size)
1929: {
1.1 cgd 1930:
1.171 plunky 1931: memset(sopt, 0, sizeof(*sopt));
1.1 cgd 1932:
1.171 plunky 1933: sopt->sopt_level = level;
1934: sopt->sopt_name = name;
1.176 plunky 1935: (void)sockopt_alloc(sopt, size, KM_SLEEP);
1.171 plunky 1936: }
1937:
1938: /*
1939: * destroy sockopt storage
1940: * - will release any held memory references
1941: */
1942: void
1943: sockopt_destroy(struct sockopt *sopt)
1944: {
1945:
1946: if (sopt->sopt_data != sopt->sopt_buf)
1.173 plunky 1947: kmem_free(sopt->sopt_data, sopt->sopt_size);
1.171 plunky 1948:
1949: memset(sopt, 0, sizeof(*sopt));
1950: }
1951:
1952: /*
1953: * set sockopt value
1954: * - value is copied into sockopt
1.176 plunky 1955: * - memory is allocated when necessary, will not sleep
1.171 plunky 1956: */
1957: int
1958: sockopt_set(struct sockopt *sopt, const void *buf, size_t len)
1959: {
1.176 plunky 1960: int error;
1.171 plunky 1961:
1.176 plunky 1962: if (sopt->sopt_size == 0) {
1963: error = sockopt_alloc(sopt, len, KM_NOSLEEP);
1964: if (error)
1965: return error;
1966: }
1.171 plunky 1967:
1968: KASSERT(sopt->sopt_size == len);
1969: memcpy(sopt->sopt_data, buf, len);
1970: return 0;
1971: }
1972:
1973: /*
1974: * common case of set sockopt integer value
1975: */
1976: int
1977: sockopt_setint(struct sockopt *sopt, int val)
1978: {
1979:
1980: return sockopt_set(sopt, &val, sizeof(int));
1981: }
1982:
1983: /*
1984: * get sockopt value
1985: * - correct size must be given
1986: */
1987: int
1988: sockopt_get(const struct sockopt *sopt, void *buf, size_t len)
1989: {
1.170 tls 1990:
1.171 plunky 1991: if (sopt->sopt_size != len)
1992: return EINVAL;
1.1 cgd 1993:
1.171 plunky 1994: memcpy(buf, sopt->sopt_data, len);
1995: return 0;
1996: }
1.1 cgd 1997:
1.171 plunky 1998: /*
1999: * common case of get sockopt integer value
2000: */
2001: int
2002: sockopt_getint(const struct sockopt *sopt, int *valp)
2003: {
1.1 cgd 2004:
1.171 plunky 2005: return sockopt_get(sopt, valp, sizeof(int));
2006: }
1.1 cgd 2007:
1.171 plunky 2008: /*
2009: * set sockopt value from mbuf
2010: * - ONLY for legacy code
2011: * - mbuf is released by sockopt
1.176 plunky 2012: * - will not sleep
1.171 plunky 2013: */
2014: int
2015: sockopt_setmbuf(struct sockopt *sopt, struct mbuf *m)
2016: {
2017: size_t len;
1.176 plunky 2018: int error;
1.1 cgd 2019:
1.171 plunky 2020: len = m_length(m);
1.1 cgd 2021:
1.176 plunky 2022: if (sopt->sopt_size == 0) {
2023: error = sockopt_alloc(sopt, len, KM_NOSLEEP);
2024: if (error)
2025: return error;
2026: }
1.1 cgd 2027:
1.171 plunky 2028: KASSERT(sopt->sopt_size == len);
2029: m_copydata(m, 0, len, sopt->sopt_data);
2030: m_freem(m);
1.1 cgd 2031:
1.171 plunky 2032: return 0;
2033: }
1.1 cgd 2034:
1.171 plunky 2035: /*
2036: * get sockopt value into mbuf
2037: * - ONLY for legacy code
2038: * - mbuf to be released by the caller
1.176 plunky 2039: * - will not sleep
1.171 plunky 2040: */
2041: struct mbuf *
2042: sockopt_getmbuf(const struct sockopt *sopt)
2043: {
2044: struct mbuf *m;
1.107 darrenr 2045:
1.176 plunky 2046: if (sopt->sopt_size > MCLBYTES)
2047: return NULL;
2048:
2049: m = m_get(M_DONTWAIT, MT_SOOPTS);
1.171 plunky 2050: if (m == NULL)
2051: return NULL;
2052:
1.176 plunky 2053: if (sopt->sopt_size > MLEN) {
2054: MCLGET(m, M_DONTWAIT);
2055: if ((m->m_flags & M_EXT) == 0) {
2056: m_free(m);
2057: return NULL;
2058: }
1.1 cgd 2059: }
1.176 plunky 2060:
2061: memcpy(mtod(m, void *), sopt->sopt_data, sopt->sopt_size);
2062: m->m_len = sopt->sopt_size;
1.160 ad 2063:
1.171 plunky 2064: return m;
1.1 cgd 2065: }
2066:
1.14 mycroft 2067: void
1.54 lukem 2068: sohasoutofband(struct socket *so)
1.1 cgd 2069: {
1.153 rmind 2070:
1.90 christos 2071: fownsignal(so->so_pgid, SIGURG, POLL_PRI, POLLPRI|POLLRDBAND, so);
1.153 rmind 2072: selnotify(&so->so_rcv.sb_sel, POLLPRI | POLLRDBAND, 0);
1.1 cgd 2073: }
1.72 jdolecek 2074:
2075: static void
2076: filt_sordetach(struct knote *kn)
2077: {
2078: struct socket *so;
2079:
1.155 ad 2080: so = ((file_t *)kn->kn_obj)->f_data;
1.160 ad 2081: solock(so);
1.73 christos 2082: SLIST_REMOVE(&so->so_rcv.sb_sel.sel_klist, kn, knote, kn_selnext);
2083: if (SLIST_EMPTY(&so->so_rcv.sb_sel.sel_klist))
1.72 jdolecek 2084: so->so_rcv.sb_flags &= ~SB_KNOTE;
1.160 ad 2085: sounlock(so);
1.72 jdolecek 2086: }
2087:
2088: /*ARGSUSED*/
2089: static int
1.129 yamt 2090: filt_soread(struct knote *kn, long hint)
1.72 jdolecek 2091: {
2092: struct socket *so;
1.160 ad 2093: int rv;
1.72 jdolecek 2094:
1.155 ad 2095: so = ((file_t *)kn->kn_obj)->f_data;
1.160 ad 2096: if (hint != NOTE_SUBMIT)
2097: solock(so);
1.72 jdolecek 2098: kn->kn_data = so->so_rcv.sb_cc;
2099: if (so->so_state & SS_CANTRCVMORE) {
1.108 perry 2100: kn->kn_flags |= EV_EOF;
1.72 jdolecek 2101: kn->kn_fflags = so->so_error;
1.160 ad 2102: rv = 1;
2103: } else if (so->so_error) /* temporary udp error */
2104: rv = 1;
2105: else if (kn->kn_sfflags & NOTE_LOWAT)
2106: rv = (kn->kn_data >= kn->kn_sdata);
2107: else
2108: rv = (kn->kn_data >= so->so_rcv.sb_lowat);
2109: if (hint != NOTE_SUBMIT)
2110: sounlock(so);
2111: return rv;
1.72 jdolecek 2112: }
2113:
2114: static void
2115: filt_sowdetach(struct knote *kn)
2116: {
2117: struct socket *so;
2118:
1.155 ad 2119: so = ((file_t *)kn->kn_obj)->f_data;
1.160 ad 2120: solock(so);
1.73 christos 2121: SLIST_REMOVE(&so->so_snd.sb_sel.sel_klist, kn, knote, kn_selnext);
2122: if (SLIST_EMPTY(&so->so_snd.sb_sel.sel_klist))
1.72 jdolecek 2123: so->so_snd.sb_flags &= ~SB_KNOTE;
1.160 ad 2124: sounlock(so);
1.72 jdolecek 2125: }
2126:
2127: /*ARGSUSED*/
2128: static int
1.129 yamt 2129: filt_sowrite(struct knote *kn, long hint)
1.72 jdolecek 2130: {
2131: struct socket *so;
1.160 ad 2132: int rv;
1.72 jdolecek 2133:
1.155 ad 2134: so = ((file_t *)kn->kn_obj)->f_data;
1.160 ad 2135: if (hint != NOTE_SUBMIT)
2136: solock(so);
1.72 jdolecek 2137: kn->kn_data = sbspace(&so->so_snd);
2138: if (so->so_state & SS_CANTSENDMORE) {
1.108 perry 2139: kn->kn_flags |= EV_EOF;
1.72 jdolecek 2140: kn->kn_fflags = so->so_error;
1.160 ad 2141: rv = 1;
2142: } else if (so->so_error) /* temporary udp error */
2143: rv = 1;
2144: else if (((so->so_state & SS_ISCONNECTED) == 0) &&
1.72 jdolecek 2145: (so->so_proto->pr_flags & PR_CONNREQUIRED))
1.160 ad 2146: rv = 0;
2147: else if (kn->kn_sfflags & NOTE_LOWAT)
2148: rv = (kn->kn_data >= kn->kn_sdata);
2149: else
2150: rv = (kn->kn_data >= so->so_snd.sb_lowat);
2151: if (hint != NOTE_SUBMIT)
2152: sounlock(so);
2153: return rv;
1.72 jdolecek 2154: }
2155:
2156: /*ARGSUSED*/
2157: static int
1.129 yamt 2158: filt_solisten(struct knote *kn, long hint)
1.72 jdolecek 2159: {
2160: struct socket *so;
1.160 ad 2161: int rv;
1.72 jdolecek 2162:
1.155 ad 2163: so = ((file_t *)kn->kn_obj)->f_data;
1.72 jdolecek 2164:
2165: /*
2166: * Set kn_data to number of incoming connections, not
2167: * counting partial (incomplete) connections.
1.108 perry 2168: */
1.160 ad 2169: if (hint != NOTE_SUBMIT)
2170: solock(so);
1.72 jdolecek 2171: kn->kn_data = so->so_qlen;
1.160 ad 2172: rv = (kn->kn_data > 0);
2173: if (hint != NOTE_SUBMIT)
2174: sounlock(so);
2175: return rv;
1.72 jdolecek 2176: }
2177:
2178: static const struct filterops solisten_filtops =
2179: { 1, NULL, filt_sordetach, filt_solisten };
2180: static const struct filterops soread_filtops =
2181: { 1, NULL, filt_sordetach, filt_soread };
2182: static const struct filterops sowrite_filtops =
2183: { 1, NULL, filt_sowdetach, filt_sowrite };
2184:
2185: int
1.129 yamt 2186: soo_kqfilter(struct file *fp, struct knote *kn)
1.72 jdolecek 2187: {
2188: struct socket *so;
2189: struct sockbuf *sb;
2190:
1.155 ad 2191: so = ((file_t *)kn->kn_obj)->f_data;
1.160 ad 2192: solock(so);
1.72 jdolecek 2193: switch (kn->kn_filter) {
2194: case EVFILT_READ:
2195: if (so->so_options & SO_ACCEPTCONN)
2196: kn->kn_fop = &solisten_filtops;
2197: else
2198: kn->kn_fop = &soread_filtops;
2199: sb = &so->so_rcv;
2200: break;
2201: case EVFILT_WRITE:
2202: kn->kn_fop = &sowrite_filtops;
2203: sb = &so->so_snd;
2204: break;
2205: default:
1.160 ad 2206: sounlock(so);
1.149 pooka 2207: return (EINVAL);
1.72 jdolecek 2208: }
1.73 christos 2209: SLIST_INSERT_HEAD(&sb->sb_sel.sel_klist, kn, kn_selnext);
1.72 jdolecek 2210: sb->sb_flags |= SB_KNOTE;
1.160 ad 2211: sounlock(so);
1.72 jdolecek 2212: return (0);
2213: }
2214:
1.154 ad 2215: static int
2216: sodopoll(struct socket *so, int events)
2217: {
2218: int revents;
2219:
2220: revents = 0;
2221:
2222: if (events & (POLLIN | POLLRDNORM))
2223: if (soreadable(so))
2224: revents |= events & (POLLIN | POLLRDNORM);
2225:
2226: if (events & (POLLOUT | POLLWRNORM))
2227: if (sowritable(so))
2228: revents |= events & (POLLOUT | POLLWRNORM);
2229:
2230: if (events & (POLLPRI | POLLRDBAND))
2231: if (so->so_oobmark || (so->so_state & SS_RCVATMARK))
2232: revents |= events & (POLLPRI | POLLRDBAND);
2233:
2234: return revents;
2235: }
2236:
2237: int
2238: sopoll(struct socket *so, int events)
2239: {
2240: int revents = 0;
2241:
1.160 ad 2242: #ifndef DIAGNOSTIC
2243: /*
2244: * Do a quick, unlocked check in expectation that the socket
2245: * will be ready for I/O. Don't do this check if DIAGNOSTIC,
2246: * as the solocked() assertions will fail.
2247: */
1.154 ad 2248: if ((revents = sodopoll(so, events)) != 0)
2249: return revents;
1.160 ad 2250: #endif
1.154 ad 2251:
1.160 ad 2252: solock(so);
1.154 ad 2253: if ((revents = sodopoll(so, events)) == 0) {
2254: if (events & (POLLIN | POLLPRI | POLLRDNORM | POLLRDBAND)) {
2255: selrecord(curlwp, &so->so_rcv.sb_sel);
1.160 ad 2256: so->so_rcv.sb_flags |= SB_NOTIFY;
1.154 ad 2257: }
2258:
2259: if (events & (POLLOUT | POLLWRNORM)) {
2260: selrecord(curlwp, &so->so_snd.sb_sel);
1.160 ad 2261: so->so_snd.sb_flags |= SB_NOTIFY;
1.154 ad 2262: }
2263: }
1.160 ad 2264: sounlock(so);
1.154 ad 2265:
2266: return revents;
2267: }
2268:
2269:
1.94 yamt 2270: #include <sys/sysctl.h>
2271:
2272: static int sysctl_kern_somaxkva(SYSCTLFN_PROTO);
2273:
2274: /*
2275: * sysctl helper routine for kern.somaxkva. ensures that the given
2276: * value is not too small.
2277: * (XXX should we maybe make sure it's not too large as well?)
2278: */
2279: static int
2280: sysctl_kern_somaxkva(SYSCTLFN_ARGS)
2281: {
2282: int error, new_somaxkva;
2283: struct sysctlnode node;
2284:
2285: new_somaxkva = somaxkva;
2286: node = *rnode;
2287: node.sysctl_data = &new_somaxkva;
2288: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2289: if (error || newp == NULL)
2290: return (error);
2291:
2292: if (new_somaxkva < (16 * 1024 * 1024)) /* sanity */
2293: return (EINVAL);
2294:
1.136 ad 2295: mutex_enter(&so_pendfree_lock);
1.94 yamt 2296: somaxkva = new_somaxkva;
1.136 ad 2297: cv_broadcast(&socurkva_cv);
2298: mutex_exit(&so_pendfree_lock);
1.94 yamt 2299:
2300: return (error);
2301: }
2302:
1.178 pooka 2303: static void
2304: sysctl_kern_somaxkva_setup()
1.94 yamt 2305: {
2306:
1.178 pooka 2307: KASSERT(socket_sysctllog == NULL);
2308: sysctl_createv(&socket_sysctllog, 0, NULL, NULL,
1.97 atatat 2309: CTLFLAG_PERMANENT,
2310: CTLTYPE_NODE, "kern", NULL,
2311: NULL, 0, NULL, 0,
2312: CTL_KERN, CTL_EOL);
2313:
1.178 pooka 2314: sysctl_createv(&socket_sysctllog, 0, NULL, NULL,
1.97 atatat 2315: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.103 atatat 2316: CTLTYPE_INT, "somaxkva",
2317: SYSCTL_DESCR("Maximum amount of kernel memory to be "
2318: "used for socket buffers"),
1.94 yamt 2319: sysctl_kern_somaxkva, 0, NULL, 0,
2320: CTL_KERN, KERN_SOMAXKVA, CTL_EOL);
2321: }
CVSweb <webmaster@jp.NetBSD.org>