Annotation of src/sys/kern/uipc_socket.c, Revision 1.215.4.3
1.215.4.2 rmind 1: /* $NetBSD$ */
1.64 thorpej 2:
3: /*-
1.188 ad 4: * Copyright (c) 2002, 2007, 2008, 2009 The NetBSD Foundation, Inc.
1.64 thorpej 5: * All rights reserved.
6: *
7: * This code is derived from software contributed to The NetBSD Foundation
1.188 ad 8: * by Jason R. Thorpe of Wasabi Systems, Inc, and by Andrew Doran.
1.64 thorpej 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:
1.215.4.1 rmind 65: /*
66: * Socket operation routines.
67: *
68: * These routines are called by the routines in sys_socket.c or from a
69: * system process, and implement the semantics of socket operations by
70: * switching out to the protocol specific routines.
71: */
72:
1.59 lukem 73: #include <sys/cdefs.h>
1.215.4.2 rmind 74: __KERNEL_RCSID(0, "$NetBSD$");
1.64 thorpej 75:
1.179 christos 76: #include "opt_compat_netbsd.h"
1.64 thorpej 77: #include "opt_sock_counters.h"
78: #include "opt_sosend_loan.h"
1.81 martin 79: #include "opt_mbuftrace.h"
1.84 ragge 80: #include "opt_somaxkva.h"
1.167 ad 81: #include "opt_multiprocessor.h" /* XXX */
1.1 cgd 82:
1.9 mycroft 83: #include <sys/param.h>
84: #include <sys/systm.h>
85: #include <sys/proc.h>
86: #include <sys/file.h>
1.142 dyoung 87: #include <sys/filedesc.h>
1.173 plunky 88: #include <sys/kmem.h>
1.9 mycroft 89: #include <sys/mbuf.h>
90: #include <sys/domain.h>
91: #include <sys/kernel.h>
92: #include <sys/protosw.h>
93: #include <sys/socket.h>
94: #include <sys/socketvar.h>
1.21 christos 95: #include <sys/signalvar.h>
1.9 mycroft 96: #include <sys/resourcevar.h>
1.174 pooka 97: #include <sys/uidinfo.h>
1.72 jdolecek 98: #include <sys/event.h>
1.89 christos 99: #include <sys/poll.h>
1.118 elad 100: #include <sys/kauth.h>
1.136 ad 101: #include <sys/mutex.h>
102: #include <sys/condvar.h>
1.205 bouyer 103: #include <sys/kthread.h>
1.37 thorpej 104:
1.179 christos 105: #ifdef COMPAT_50
106: #include <compat/sys/time.h>
1.184 christos 107: #include <compat/sys/socket.h>
1.179 christos 108: #endif
109:
1.202 uebayasi 110: #include <uvm/uvm_extern.h>
111: #include <uvm/uvm_loan.h>
112: #include <uvm/uvm_page.h>
1.64 thorpej 113:
1.77 thorpej 114: MALLOC_DEFINE(M_SONAME, "soname", "socket name");
1.37 thorpej 115:
1.142 dyoung 116: extern const struct fileops socketops;
117:
1.54 lukem 118: extern int somaxconn; /* patchable (XXX sysctl) */
119: int somaxconn = SOMAXCONN;
1.160 ad 120: kmutex_t *softnet_lock;
1.49 jonathan 121:
1.64 thorpej 122: #ifdef SOSEND_COUNTERS
123: #include <sys/device.h>
124:
1.113 thorpej 125: static struct evcnt sosend_loan_big = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64 thorpej 126: NULL, "sosend", "loan big");
1.113 thorpej 127: static struct evcnt sosend_copy_big = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64 thorpej 128: NULL, "sosend", "copy big");
1.113 thorpej 129: static struct evcnt sosend_copy_small = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64 thorpej 130: NULL, "sosend", "copy small");
1.113 thorpej 131: static struct evcnt sosend_kvalimit = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64 thorpej 132: NULL, "sosend", "kva limit");
133:
134: #define SOSEND_COUNTER_INCR(ev) (ev)->ev_count++
135:
1.101 matt 136: EVCNT_ATTACH_STATIC(sosend_loan_big);
137: EVCNT_ATTACH_STATIC(sosend_copy_big);
138: EVCNT_ATTACH_STATIC(sosend_copy_small);
139: EVCNT_ATTACH_STATIC(sosend_kvalimit);
1.64 thorpej 140: #else
141:
142: #define SOSEND_COUNTER_INCR(ev) /* nothing */
143:
144: #endif /* SOSEND_COUNTERS */
145:
1.167 ad 146: #if defined(SOSEND_NO_LOAN) || defined(MULTIPROCESSOR)
1.121 yamt 147: int sock_loan_thresh = -1;
1.71 thorpej 148: #else
1.121 yamt 149: int sock_loan_thresh = 4096;
1.65 thorpej 150: #endif
1.64 thorpej 151:
1.136 ad 152: static kmutex_t so_pendfree_lock;
1.205 bouyer 153: static struct mbuf *so_pendfree = NULL;
1.64 thorpej 154:
1.84 ragge 155: #ifndef SOMAXKVA
156: #define SOMAXKVA (16 * 1024 * 1024)
157: #endif
158: int somaxkva = SOMAXKVA;
1.113 thorpej 159: static int socurkva;
1.136 ad 160: static kcondvar_t socurkva_cv;
1.64 thorpej 161:
1.191 elad 162: static kauth_listener_t socket_listener;
163:
1.64 thorpej 164: #define SOCK_LOAN_CHUNK 65536
165:
1.205 bouyer 166: static void sopendfree_thread(void *);
167: static kcondvar_t pendfree_thread_cv;
168: static lwp_t *sopendfree_lwp;
1.93 yamt 169:
1.212 pooka 170: static void sysctl_kern_socket_setup(void);
1.178 pooka 171: static struct sysctllog *socket_sysctllog;
172:
1.113 thorpej 173: static vsize_t
1.129 yamt 174: sokvareserve(struct socket *so, vsize_t len)
1.80 yamt 175: {
1.98 christos 176: int error;
1.80 yamt 177:
1.136 ad 178: mutex_enter(&so_pendfree_lock);
1.80 yamt 179: while (socurkva + len > somaxkva) {
180: SOSEND_COUNTER_INCR(&sosend_kvalimit);
1.136 ad 181: error = cv_wait_sig(&socurkva_cv, &so_pendfree_lock);
1.98 christos 182: if (error) {
183: len = 0;
184: break;
185: }
1.80 yamt 186: }
1.93 yamt 187: socurkva += len;
1.136 ad 188: mutex_exit(&so_pendfree_lock);
1.98 christos 189: return len;
1.95 yamt 190: }
191:
1.113 thorpej 192: static void
1.95 yamt 193: sokvaunreserve(vsize_t len)
194: {
195:
1.136 ad 196: mutex_enter(&so_pendfree_lock);
1.95 yamt 197: socurkva -= len;
1.136 ad 198: cv_broadcast(&socurkva_cv);
199: mutex_exit(&so_pendfree_lock);
1.95 yamt 200: }
201:
202: /*
203: * sokvaalloc: allocate kva for loan.
204: */
205:
206: vaddr_t
1.209 matt 207: sokvaalloc(vaddr_t sva, vsize_t len, struct socket *so)
1.95 yamt 208: {
209: vaddr_t lva;
210:
211: /*
212: * reserve kva.
213: */
214:
1.98 christos 215: if (sokvareserve(so, len) == 0)
216: return 0;
1.93 yamt 217:
218: /*
219: * allocate kva.
220: */
1.80 yamt 221:
1.209 matt 222: lva = uvm_km_alloc(kernel_map, len, atop(sva) & uvmexp.colormask,
223: UVM_KMF_COLORMATCH | UVM_KMF_VAONLY | UVM_KMF_WAITVA);
1.95 yamt 224: if (lva == 0) {
225: sokvaunreserve(len);
1.80 yamt 226: return (0);
1.95 yamt 227: }
1.80 yamt 228:
229: return lva;
230: }
231:
1.93 yamt 232: /*
233: * sokvafree: free kva for loan.
234: */
235:
1.80 yamt 236: void
237: sokvafree(vaddr_t sva, vsize_t len)
238: {
1.93 yamt 239:
240: /*
241: * free kva.
242: */
1.80 yamt 243:
1.109 yamt 244: uvm_km_free(kernel_map, sva, len, UVM_KMF_VAONLY);
1.93 yamt 245:
246: /*
247: * unreserve kva.
248: */
249:
1.95 yamt 250: sokvaunreserve(len);
1.80 yamt 251: }
252:
1.64 thorpej 253: static void
1.134 christos 254: sodoloanfree(struct vm_page **pgs, void *buf, size_t size)
1.64 thorpej 255: {
1.156 yamt 256: vaddr_t sva, eva;
1.64 thorpej 257: vsize_t len;
1.156 yamt 258: int npgs;
259:
260: KASSERT(pgs != NULL);
1.64 thorpej 261:
262: eva = round_page((vaddr_t) buf + size);
263: sva = trunc_page((vaddr_t) buf);
264: len = eva - sva;
265: npgs = len >> PAGE_SHIFT;
266:
267: pmap_kremove(sva, len);
268: pmap_update(pmap_kernel());
269: uvm_unloan(pgs, npgs, UVM_LOAN_TOPAGE);
1.80 yamt 270: sokvafree(sva, len);
1.64 thorpej 271: }
272:
1.93 yamt 273: /*
1.205 bouyer 274: * sopendfree_thread: free mbufs on "pendfree" list.
1.136 ad 275: * unlock and relock so_pendfree_lock when freeing mbufs.
1.93 yamt 276: */
277:
1.205 bouyer 278: static void
279: sopendfree_thread(void *v)
1.93 yamt 280: {
1.137 ad 281: struct mbuf *m, *next;
1.205 bouyer 282: size_t rv;
1.93 yamt 283:
1.205 bouyer 284: mutex_enter(&so_pendfree_lock);
1.64 thorpej 285:
1.205 bouyer 286: for (;;) {
287: rv = 0;
288: while (so_pendfree != NULL) {
289: m = so_pendfree;
290: so_pendfree = NULL;
291: mutex_exit(&so_pendfree_lock);
292:
293: for (; m != NULL; m = next) {
294: next = m->m_next;
295: KASSERT((~m->m_flags & (M_EXT|M_EXT_PAGES)) == 0);
296: KASSERT(m->m_ext.ext_refcnt == 0);
297:
298: rv += m->m_ext.ext_size;
299: sodoloanfree(m->m_ext.ext_pgs, m->m_ext.ext_buf,
300: m->m_ext.ext_size);
301: pool_cache_put(mb_cache, m);
302: }
1.93 yamt 303:
1.205 bouyer 304: mutex_enter(&so_pendfree_lock);
1.93 yamt 305: }
1.205 bouyer 306: if (rv)
307: cv_broadcast(&socurkva_cv);
308: cv_wait(&pendfree_thread_cv, &so_pendfree_lock);
1.64 thorpej 309: }
1.205 bouyer 310: panic("sopendfree_thread");
311: /* NOTREACHED */
1.64 thorpej 312: }
313:
1.80 yamt 314: void
1.134 christos 315: soloanfree(struct mbuf *m, void *buf, size_t size, void *arg)
1.64 thorpej 316: {
317:
1.156 yamt 318: KASSERT(m != NULL);
1.64 thorpej 319:
1.93 yamt 320: /*
321: * postpone freeing mbuf.
322: *
323: * we can't do it in interrupt context
324: * because we need to put kva back to kernel_map.
325: */
326:
1.136 ad 327: mutex_enter(&so_pendfree_lock);
1.92 yamt 328: m->m_next = so_pendfree;
329: so_pendfree = m;
1.205 bouyer 330: cv_signal(&pendfree_thread_cv);
1.136 ad 331: mutex_exit(&so_pendfree_lock);
1.64 thorpej 332: }
333:
334: static long
335: sosend_loan(struct socket *so, struct uio *uio, struct mbuf *m, long space)
336: {
337: struct iovec *iov = uio->uio_iov;
338: vaddr_t sva, eva;
339: vsize_t len;
1.156 yamt 340: vaddr_t lva;
341: int npgs, error;
342: vaddr_t va;
343: int i;
1.64 thorpej 344:
1.116 yamt 345: if (VMSPACE_IS_KERNEL_P(uio->uio_vmspace))
1.64 thorpej 346: return (0);
347:
348: if (iov->iov_len < (size_t) space)
349: space = iov->iov_len;
350: if (space > SOCK_LOAN_CHUNK)
351: space = SOCK_LOAN_CHUNK;
352:
353: eva = round_page((vaddr_t) iov->iov_base + space);
354: sva = trunc_page((vaddr_t) iov->iov_base);
355: len = eva - sva;
356: npgs = len >> PAGE_SHIFT;
357:
1.79 thorpej 358: KASSERT(npgs <= M_EXT_MAXPAGES);
359:
1.209 matt 360: lva = sokvaalloc(sva, len, so);
1.64 thorpej 361: if (lva == 0)
1.80 yamt 362: return 0;
1.64 thorpej 363:
1.116 yamt 364: error = uvm_loan(&uio->uio_vmspace->vm_map, sva, len,
1.79 thorpej 365: m->m_ext.ext_pgs, UVM_LOAN_TOPAGE);
1.64 thorpej 366: if (error) {
1.80 yamt 367: sokvafree(lva, len);
1.64 thorpej 368: return (0);
369: }
370:
371: for (i = 0, va = lva; i < npgs; i++, va += PAGE_SIZE)
1.79 thorpej 372: pmap_kenter_pa(va, VM_PAGE_TO_PHYS(m->m_ext.ext_pgs[i]),
1.194 cegger 373: VM_PROT_READ, 0);
1.64 thorpej 374: pmap_update(pmap_kernel());
375:
376: lva += (vaddr_t) iov->iov_base & PAGE_MASK;
377:
1.134 christos 378: MEXTADD(m, (void *) lva, space, M_MBUF, soloanfree, so);
1.79 thorpej 379: m->m_flags |= M_EXT_PAGES | M_EXT_ROMAP;
1.64 thorpej 380:
381: uio->uio_resid -= space;
382: /* uio_offset not updated, not set/used for write(2) */
1.134 christos 383: uio->uio_iov->iov_base = (char *)uio->uio_iov->iov_base + space;
1.64 thorpej 384: uio->uio_iov->iov_len -= space;
385: if (uio->uio_iov->iov_len == 0) {
386: uio->uio_iov++;
387: uio->uio_iovcnt--;
388: }
389:
390: return (space);
391: }
392:
1.142 dyoung 393: struct mbuf *
1.147 dyoung 394: getsombuf(struct socket *so, int type)
1.142 dyoung 395: {
396: struct mbuf *m;
397:
1.147 dyoung 398: m = m_get(M_WAIT, type);
1.142 dyoung 399: MCLAIM(m, so->so_mowner);
400: return m;
401: }
402:
1.191 elad 403: static int
404: socket_listener_cb(kauth_cred_t cred, kauth_action_t action, void *cookie,
405: void *arg0, void *arg1, void *arg2, void *arg3)
406: {
407: int result;
408: enum kauth_network_req req;
409:
410: result = KAUTH_RESULT_DEFER;
411: req = (enum kauth_network_req)arg0;
412:
1.193 elad 413: if ((action != KAUTH_NETWORK_SOCKET) &&
414: (action != KAUTH_NETWORK_BIND))
1.191 elad 415: return result;
416:
417: switch (req) {
1.193 elad 418: case KAUTH_REQ_NETWORK_BIND_PORT:
419: result = KAUTH_RESULT_ALLOW;
420: break;
421:
1.191 elad 422: case KAUTH_REQ_NETWORK_SOCKET_DROP: {
423: /* Normal users can only drop their own connections. */
424: struct socket *so = (struct socket *)arg1;
425:
1.215.4.2 rmind 426: if (proc_uidmatch(cred, so->so_cred) == 0)
1.191 elad 427: result = KAUTH_RESULT_ALLOW;
428:
429: break;
430: }
431:
432: case KAUTH_REQ_NETWORK_SOCKET_OPEN:
433: /* We allow "raw" routing/bluetooth sockets to anyone. */
1.203 matt 434: if ((u_long)arg1 == PF_ROUTE || (u_long)arg1 == PF_OROUTE
435: || (u_long)arg1 == PF_BLUETOOTH) {
1.191 elad 436: result = KAUTH_RESULT_ALLOW;
1.203 matt 437: } else {
1.191 elad 438: /* Privileged, let secmodel handle this. */
439: if ((u_long)arg2 == SOCK_RAW)
440: break;
441: }
442:
443: result = KAUTH_RESULT_ALLOW;
444:
445: break;
446:
1.192 elad 447: case KAUTH_REQ_NETWORK_SOCKET_CANSEE:
448: result = KAUTH_RESULT_ALLOW;
449:
450: break;
451:
1.191 elad 452: default:
453: break;
454: }
455:
456: return result;
457: }
458:
1.119 yamt 459: void
460: soinit(void)
461: {
462:
1.212 pooka 463: sysctl_kern_socket_setup();
1.178 pooka 464:
1.148 ad 465: mutex_init(&so_pendfree_lock, MUTEX_DEFAULT, IPL_VM);
1.160 ad 466: softnet_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NONE);
1.136 ad 467: cv_init(&socurkva_cv, "sokva");
1.205 bouyer 468: cv_init(&pendfree_thread_cv, "sopendfr");
1.166 ad 469: soinit2();
1.136 ad 470:
1.119 yamt 471: /* Set the initial adjusted socket buffer size. */
472: if (sb_max_set(sb_max))
473: panic("bad initial sb_max value: %lu", sb_max);
474:
1.191 elad 475: socket_listener = kauth_listen_scope(KAUTH_SCOPE_NETWORK,
476: socket_listener_cb, NULL);
1.119 yamt 477: }
478:
1.205 bouyer 479: void
480: soinit1(void)
481: {
482: int error = kthread_create(PRI_NONE, KTHREAD_MPSAFE, NULL,
483: sopendfree_thread, NULL, &sopendfree_lwp, "sopendfree");
484: if (error)
485: panic("soinit1 %d", error);
486: }
487:
1.1 cgd 488: /*
1.215.4.1 rmind 489: * socreate: create a new socket of the specified type and the protocol.
490: *
491: * => Caller may specify another socket for lock sharing (must not be held).
492: * => Returns the new socket without lock held.
1.1 cgd 493: */
1.3 andrew 494: int
1.160 ad 495: socreate(int dom, struct socket **aso, int type, int proto, struct lwp *l,
496: struct socket *lockso)
1.1 cgd 497: {
1.99 matt 498: const struct protosw *prp;
1.54 lukem 499: struct socket *so;
1.115 yamt 500: uid_t uid;
1.160 ad 501: int error;
502: kmutex_t *lock;
1.1 cgd 503:
1.132 elad 504: error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET,
505: KAUTH_REQ_NETWORK_SOCKET_OPEN, KAUTH_ARG(dom), KAUTH_ARG(type),
506: KAUTH_ARG(proto));
1.140 dyoung 507: if (error != 0)
508: return error;
1.127 elad 509:
1.1 cgd 510: if (proto)
511: prp = pffindproto(dom, proto, type);
512: else
513: prp = pffindtype(dom, type);
1.140 dyoung 514: if (prp == NULL) {
1.120 ginsbach 515: /* no support for domain */
516: if (pffinddomain(dom) == 0)
1.140 dyoung 517: return EAFNOSUPPORT;
1.120 ginsbach 518: /* no support for socket type */
519: if (proto == 0 && type != 0)
1.140 dyoung 520: return EPROTOTYPE;
521: return EPROTONOSUPPORT;
1.120 ginsbach 522: }
1.215.4.1 rmind 523: if (prp->pr_usrreqs == NULL)
1.140 dyoung 524: return EPROTONOSUPPORT;
1.1 cgd 525: if (prp->pr_type != type)
1.140 dyoung 526: return EPROTOTYPE;
1.160 ad 527:
528: so = soget(true);
1.1 cgd 529: so->so_type = type;
530: so->so_proto = prp;
1.215.4.3! rmind 531: so->so_refcnt = 1;
! 532:
1.33 matt 533: so->so_send = sosend;
534: so->so_receive = soreceive;
1.78 matt 535: #ifdef MBUFTRACE
536: so->so_rcv.sb_mowner = &prp->pr_domain->dom_mowner;
537: so->so_snd.sb_mowner = &prp->pr_domain->dom_mowner;
538: so->so_mowner = &prp->pr_domain->dom_mowner;
539: #endif
1.138 rmind 540: uid = kauth_cred_geteuid(l->l_cred);
1.115 yamt 541: so->so_uidinfo = uid_find(uid);
1.168 yamt 542: so->so_cpid = l->l_proc->p_pid;
1.215.4.1 rmind 543:
544: /*
545: * Lock assigned and taken during pr_attach, unless we share
546: * the lock with another socket, e.g. socketpair(2) case.
547: */
548: if (lockso) {
1.160 ad 549: lock = lockso->so_lock;
550: so->so_lock = lock;
551: mutex_obj_hold(lock);
552: }
1.215.4.1 rmind 553:
554: error = (*prp->pr_usrreqs->pr_attach)(so, proto);
555: if (error) {
556: solock(so);
557: KASSERT(so->so_pcb == NULL);
1.1 cgd 558: so->so_state |= SS_NOFDREF;
559: sofree(so);
1.140 dyoung 560: return error;
1.1 cgd 561: }
1.198 elad 562: so->so_cred = kauth_cred_dup(l->l_cred);
1.215.4.1 rmind 563:
1.1 cgd 564: *aso = so;
1.140 dyoung 565: return 0;
1.1 cgd 566: }
567:
1.215.4.1 rmind 568: /*
569: * fsocreate: create a socket and a file descriptor associated with it.
570: *
571: * => On success, write file descriptor to fdout and return zero.
572: * => On failure, return non-zero; *fdout will be undefined.
1.142 dyoung 573: */
574: int
1.215.4.1 rmind 575: fsocreate(int domain, struct socket **sop, int type, int protocol, int *fdout)
1.142 dyoung 576: {
1.215.4.1 rmind 577: lwp_t *l = curlwp;
578: int error, fd, flags;
579: struct socket *so;
580: struct file *fp;
1.142 dyoung 581:
1.215.4.1 rmind 582: if ((error = fd_allocfile(&fp, &fd)) != 0) {
1.204 christos 583: return error;
1.215.4.1 rmind 584: }
585:
586: flags = type & SOCK_FLAGS_MASK;
1.204 christos 587: fd_set_exclose(l, fd, (flags & SOCK_CLOEXEC) != 0);
1.207 christos 588: fp->f_flag = FREAD|FWRITE|((flags & SOCK_NONBLOCK) ? FNONBLOCK : 0)|
589: ((flags & SOCK_NOSIGPIPE) ? FNOSIGPIPE : 0);
1.142 dyoung 590: fp->f_type = DTYPE_SOCKET;
591: fp->f_ops = &socketops;
1.215.4.1 rmind 592:
593: type &= ~SOCK_FLAGS_MASK;
1.160 ad 594: error = socreate(domain, &so, type, protocol, l, NULL);
1.142 dyoung 595: if (error != 0) {
1.155 ad 596: fd_abort(curproc, fp, fd);
1.215.4.1 rmind 597: return error;
1.142 dyoung 598: }
1.215.4.1 rmind 599: if (flags & SOCK_NONBLOCK) {
600: so->so_state |= SS_NBIO;
601: }
602:
603: fp->f_data = so;
604: fd_affix(curproc, fp, fd);
605: if (sop != NULL) {
606: *sop = so;
607: }
608: *fdout = fd;
1.142 dyoung 609: return error;
610: }
611:
1.215.4.3! rmind 612: void
! 613: soref(struct socket *so)
! 614: {
! 615: atomic_inc_uint(&so->so_refcnt);
! 616: }
! 617:
! 618: void
! 619: sounref(struct socket *so)
! 620: {
! 621: if (atomic_dec_uint_nv(&so->so_refcnt) > 0) {
! 622: return;
! 623: }
! 624: soput(so);
! 625: }
! 626:
1.3 andrew 627: int
1.190 dyoung 628: sofamily(const struct socket *so)
629: {
630: const struct protosw *pr;
631: const struct domain *dom;
632:
633: if ((pr = so->so_proto) == NULL)
634: return AF_UNSPEC;
635: if ((dom = pr->pr_domain) == NULL)
636: return AF_UNSPEC;
637: return dom->dom_family;
638: }
639:
640: int
1.114 christos 641: sobind(struct socket *so, struct mbuf *nam, struct lwp *l)
1.1 cgd 642: {
1.215.4.1 rmind 643: int error;
1.1 cgd 644:
1.160 ad 645: solock(so);
1.215.4.1 rmind 646: error = (*so->so_proto->pr_usrreqs->pr_generic)(so,
647: PRU_BIND, NULL, nam, NULL, l);
1.160 ad 648: sounlock(so);
1.140 dyoung 649: return error;
1.1 cgd 650: }
651:
1.3 andrew 652: int
1.150 elad 653: solisten(struct socket *so, int backlog, struct lwp *l)
1.1 cgd 654: {
1.215.4.1 rmind 655: int error;
1.1 cgd 656:
1.160 ad 657: solock(so);
1.158 ad 658: if ((so->so_state & (SS_ISCONNECTED | SS_ISCONNECTING |
1.163 ad 659: SS_ISDISCONNECTING)) != 0) {
1.215.4.1 rmind 660: sounlock(so);
1.210 matt 661: return (EINVAL);
1.163 ad 662: }
1.215.4.1 rmind 663: error = (*so->so_proto->pr_usrreqs->pr_generic)(so, PRU_LISTEN, NULL,
1.150 elad 664: NULL, NULL, l);
1.140 dyoung 665: if (error != 0) {
1.160 ad 666: sounlock(so);
1.140 dyoung 667: return error;
1.1 cgd 668: }
1.63 matt 669: if (TAILQ_EMPTY(&so->so_q))
1.1 cgd 670: so->so_options |= SO_ACCEPTCONN;
671: if (backlog < 0)
672: backlog = 0;
1.49 jonathan 673: so->so_qlimit = min(backlog, somaxconn);
1.160 ad 674: sounlock(so);
1.140 dyoung 675: return 0;
1.1 cgd 676: }
677:
1.21 christos 678: void
1.54 lukem 679: sofree(struct socket *so)
1.1 cgd 680: {
681:
1.160 ad 682: KASSERT(solocked(so));
683:
684: if (so->so_pcb || (so->so_state & SS_NOFDREF) == 0) {
685: sounlock(so);
1.1 cgd 686: return;
1.160 ad 687: }
1.43 mycroft 688: if (so->so_head) {
689: /*
690: * We must not decommission a socket that's on the accept(2)
691: * queue. If we do, then accept(2) may hang after select(2)
692: * indicated that the listening socket was ready.
693: */
1.160 ad 694: if (!soqremque(so, 0)) {
695: sounlock(so);
1.43 mycroft 696: return;
1.160 ad 697: }
1.43 mycroft 698: }
1.98 christos 699: if (so->so_rcv.sb_hiwat)
1.110 christos 700: (void)chgsbsize(so->so_uidinfo, &so->so_rcv.sb_hiwat, 0,
1.98 christos 701: RLIM_INFINITY);
702: if (so->so_snd.sb_hiwat)
1.110 christos 703: (void)chgsbsize(so->so_uidinfo, &so->so_snd.sb_hiwat, 0,
1.98 christos 704: RLIM_INFINITY);
705: sbrelease(&so->so_snd, so);
1.160 ad 706: KASSERT(!cv_has_waiters(&so->so_cv));
707: KASSERT(!cv_has_waiters(&so->so_rcv.sb_cv));
708: KASSERT(!cv_has_waiters(&so->so_snd.sb_cv));
1.1 cgd 709: sorflush(so);
1.177 ad 710: /* Remove acccept filter if one is present. */
1.170 tls 711: if (so->so_accf != NULL)
1.177 ad 712: (void)accept_filt_clear(so);
1.160 ad 713: sounlock(so);
1.215.4.3! rmind 714:
! 715: /* Will soput() if the last reference. */
! 716: sounref(so);
1.1 cgd 717: }
718:
719: /*
1.215.4.1 rmind 720: * soclose: close a socket on last file table reference removal.
721: * Initiate disconnect if connected. Free socket when disconnect complete.
1.1 cgd 722: */
1.3 andrew 723: int
1.54 lukem 724: soclose(struct socket *so)
1.1 cgd 725: {
1.215.4.1 rmind 726: struct socket *so2;
727: int error;
1.1 cgd 728:
1.54 lukem 729: error = 0;
1.160 ad 730: solock(so);
1.1 cgd 731: if (so->so_options & SO_ACCEPTCONN) {
1.172 ad 732: for (;;) {
733: if ((so2 = TAILQ_FIRST(&so->so_q0)) != 0) {
1.160 ad 734: KASSERT(solocked2(so, so2));
735: (void) soqremque(so2, 0);
736: /* soabort drops the lock. */
737: (void) soabort(so2);
738: solock(so);
1.172 ad 739: continue;
1.160 ad 740: }
1.172 ad 741: if ((so2 = TAILQ_FIRST(&so->so_q)) != 0) {
1.160 ad 742: KASSERT(solocked2(so, so2));
743: (void) soqremque(so2, 1);
744: /* soabort drops the lock. */
745: (void) soabort(so2);
746: solock(so);
1.172 ad 747: continue;
1.160 ad 748: }
1.172 ad 749: break;
750: }
1.1 cgd 751: }
1.215.4.1 rmind 752: if (so->so_pcb == NULL)
1.1 cgd 753: goto discard;
754: if (so->so_state & SS_ISCONNECTED) {
755: if ((so->so_state & SS_ISDISCONNECTING) == 0) {
756: error = sodisconnect(so);
757: if (error)
758: goto drop;
759: }
760: if (so->so_options & SO_LINGER) {
1.206 christos 761: if ((so->so_state & (SS_ISDISCONNECTING|SS_NBIO)) ==
762: (SS_ISDISCONNECTING|SS_NBIO))
1.1 cgd 763: goto drop;
1.21 christos 764: while (so->so_state & SS_ISCONNECTED) {
1.185 yamt 765: error = sowait(so, true, so->so_linger * hz);
1.21 christos 766: if (error)
1.1 cgd 767: break;
1.21 christos 768: }
1.1 cgd 769: }
770: }
1.54 lukem 771: drop:
1.1 cgd 772: if (so->so_pcb) {
1.215.4.1 rmind 773: (*so->so_proto->pr_usrreqs->pr_detach)(so);
1.1 cgd 774: }
1.54 lukem 775: discard:
1.215.4.1 rmind 776: KASSERT((so->so_state & SS_NOFDREF) == 0);
1.198 elad 777: kauth_cred_free(so->so_cred);
1.1 cgd 778: so->so_state |= SS_NOFDREF;
779: sofree(so);
1.215.4.1 rmind 780: return error;
1.1 cgd 781: }
782:
783: /*
1.160 ad 784: * Must be called with the socket locked.. Will return with it unlocked.
1.1 cgd 785: */
1.3 andrew 786: int
1.54 lukem 787: soabort(struct socket *so)
1.1 cgd 788: {
1.161 ad 789: u_int refs;
1.139 yamt 790: int error;
1.215.4.3! rmind 791:
1.160 ad 792: KASSERT(solocked(so));
793: KASSERT(so->so_head == NULL);
1.1 cgd 794:
1.215.4.3! rmind 795: soref(so);
1.215.4.1 rmind 796: error = (*so->so_proto->pr_usrreqs->pr_generic)(so,
797: PRU_ABORT, NULL, NULL, NULL, NULL);
1.215.4.3! rmind 798: refs = so->so_refcnt;
! 799: sounref(so);
! 800:
! 801: /* XXX: Fix PRU_ABORT to behave consistently. */
! 802: if (error || refs == 1) {
1.139 yamt 803: sofree(so);
1.160 ad 804: } else {
805: sounlock(so);
1.139 yamt 806: }
1.215.4.3! rmind 807: sounref(so);
1.139 yamt 808: return error;
1.1 cgd 809: }
810:
1.3 andrew 811: int
1.54 lukem 812: soaccept(struct socket *so, struct mbuf *nam)
1.1 cgd 813: {
1.160 ad 814: int error;
815:
816: KASSERT(solocked(so));
1.1 cgd 817:
1.54 lukem 818: error = 0;
1.215.4.1 rmind 819: KASSERT((so->so_state & SS_NOFDREF) != 0);
1.1 cgd 820: so->so_state &= ~SS_NOFDREF;
1.215.4.1 rmind 821:
1.55 thorpej 822: if ((so->so_state & SS_ISDISCONNECTED) == 0 ||
823: (so->so_proto->pr_flags & PR_ABRTACPTDIS) == 0)
1.215.4.1 rmind 824: error = (*so->so_proto->pr_usrreqs->pr_generic)(so,
825: PRU_ACCEPT, NULL, nam, NULL, NULL);
1.41 mycroft 826: else
1.53 itojun 827: error = ECONNABORTED;
1.52 itojun 828:
1.1 cgd 829: return (error);
830: }
831:
1.3 andrew 832: int
1.114 christos 833: soconnect(struct socket *so, struct mbuf *nam, struct lwp *l)
1.1 cgd 834: {
1.215.4.1 rmind 835: int error;
1.160 ad 836:
837: KASSERT(solocked(so));
1.1 cgd 838:
839: if (so->so_options & SO_ACCEPTCONN)
840: return (EOPNOTSUPP);
841: /*
842: * If protocol is connection-based, can only connect once.
843: * Otherwise, if connected, try to disconnect first.
844: * This allows user to disconnect by connecting to, e.g.,
845: * a null address.
846: */
847: if (so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING) &&
848: ((so->so_proto->pr_flags & PR_CONNREQUIRED) ||
849: (error = sodisconnect(so))))
850: error = EISCONN;
851: else
1.215.4.1 rmind 852: error = (*so->so_proto->pr_usrreqs->pr_generic)(so,
853: PRU_CONNECT, NULL, nam, NULL, l);
854:
855: return error;
1.1 cgd 856: }
857:
1.3 andrew 858: int
1.54 lukem 859: soconnect2(struct socket *so1, struct socket *so2)
1.1 cgd 860: {
1.160 ad 861: int error;
862:
863: KASSERT(solocked2(so1, so2));
1.1 cgd 864:
1.215.4.1 rmind 865: error = (*so1->so_proto->pr_usrreqs->pr_generic)(so1, PRU_CONNECT2,
1.140 dyoung 866: NULL, (struct mbuf *)so2, NULL, NULL);
1.1 cgd 867: return (error);
868: }
869:
1.3 andrew 870: int
1.54 lukem 871: sodisconnect(struct socket *so)
1.1 cgd 872: {
1.160 ad 873: int error;
874:
875: KASSERT(solocked(so));
1.1 cgd 876:
877: if ((so->so_state & SS_ISCONNECTED) == 0) {
878: error = ENOTCONN;
1.160 ad 879: } else if (so->so_state & SS_ISDISCONNECTING) {
1.1 cgd 880: error = EALREADY;
1.160 ad 881: } else {
1.215.4.1 rmind 882: error = (*so->so_proto->pr_usrreqs->pr_generic)(so,
883: PRU_DISCONNECT, NULL, NULL, NULL, NULL);
1.1 cgd 884: }
885: return (error);
886: }
887:
1.15 mycroft 888: #define SBLOCKWAIT(f) (((f) & MSG_DONTWAIT) ? M_NOWAIT : M_WAITOK)
1.1 cgd 889: /*
890: * Send on a socket.
891: * If send must go all at once and message is larger than
892: * send buffering, then hard error.
893: * Lock against other senders.
894: * If must go all at once and not enough room now, then
895: * inform user that this would block and do nothing.
896: * Otherwise, if nonblocking, send as much as possible.
897: * The data to be sent is described by "uio" if nonzero,
898: * otherwise by the mbuf chain "top" (which must be null
899: * if uio is not). Data provided in mbuf chain must be small
900: * enough to send all at once.
901: *
902: * Returns nonzero on error, timeout or signal; callers
903: * must check for short counts if EINTR/ERESTART are returned.
904: * Data and control buffers are freed on return.
905: */
1.3 andrew 906: int
1.54 lukem 907: sosend(struct socket *so, struct mbuf *addr, struct uio *uio, struct mbuf *top,
1.114 christos 908: struct mbuf *control, int flags, struct lwp *l)
1.1 cgd 909: {
1.54 lukem 910: struct mbuf **mp, *m;
1.58 jdolecek 911: long space, len, resid, clen, mlen;
912: int error, s, dontroute, atomic;
1.196 dsl 913: short wakeup_state = 0;
1.54 lukem 914:
1.160 ad 915: clen = 0;
1.64 thorpej 916:
1.160 ad 917: /*
918: * solock() provides atomicity of access. splsoftnet() prevents
919: * protocol processing soft interrupts from interrupting us and
920: * blocking (expensive).
921: */
922: s = splsoftnet();
923: solock(so);
1.54 lukem 924: atomic = sosendallatonce(so) || top;
1.1 cgd 925: if (uio)
926: resid = uio->uio_resid;
927: else
928: resid = top->m_pkthdr.len;
1.7 cgd 929: /*
930: * In theory resid should be unsigned.
931: * However, space must be signed, as it might be less than 0
932: * if we over-committed, and we must use a signed comparison
933: * of space and resid. On the other hand, a negative resid
934: * causes us to loop sending 0-length segments to the protocol.
935: */
1.29 mycroft 936: if (resid < 0) {
937: error = EINVAL;
938: goto out;
939: }
1.1 cgd 940: dontroute =
941: (flags & MSG_DONTROUTE) && (so->so_options & SO_DONTROUTE) == 0 &&
942: (so->so_proto->pr_flags & PR_ATOMIC);
1.165 christos 943: l->l_ru.ru_msgsnd++;
1.1 cgd 944: if (control)
945: clen = control->m_len;
1.54 lukem 946: restart:
1.21 christos 947: if ((error = sblock(&so->so_snd, SBLOCKWAIT(flags))) != 0)
1.1 cgd 948: goto out;
949: do {
1.160 ad 950: if (so->so_state & SS_CANTSENDMORE) {
951: error = EPIPE;
952: goto release;
953: }
1.48 thorpej 954: if (so->so_error) {
955: error = so->so_error;
956: so->so_error = 0;
957: goto release;
958: }
1.1 cgd 959: if ((so->so_state & SS_ISCONNECTED) == 0) {
960: if (so->so_proto->pr_flags & PR_CONNREQUIRED) {
961: if ((so->so_state & SS_ISCONFIRMING) == 0 &&
1.160 ad 962: !(resid == 0 && clen != 0)) {
963: error = ENOTCONN;
964: goto release;
965: }
966: } else if (addr == 0) {
967: error = EDESTADDRREQ;
968: goto release;
969: }
1.1 cgd 970: }
971: space = sbspace(&so->so_snd);
972: if (flags & MSG_OOB)
973: space += 1024;
1.21 christos 974: if ((atomic && resid > so->so_snd.sb_hiwat) ||
1.160 ad 975: clen > so->so_snd.sb_hiwat) {
976: error = EMSGSIZE;
977: goto release;
978: }
1.96 mycroft 979: if (space < resid + clen &&
1.1 cgd 980: (atomic || space < so->so_snd.sb_lowat || space < clen)) {
1.206 christos 981: if ((so->so_state & SS_NBIO) || (flags & MSG_NBIO)) {
1.160 ad 982: error = EWOULDBLOCK;
983: goto release;
984: }
1.1 cgd 985: sbunlock(&so->so_snd);
1.196 dsl 986: if (wakeup_state & SS_RESTARTSYS) {
987: error = ERESTART;
988: goto out;
989: }
1.1 cgd 990: error = sbwait(&so->so_snd);
991: if (error)
992: goto out;
1.196 dsl 993: wakeup_state = so->so_state;
1.1 cgd 994: goto restart;
995: }
1.196 dsl 996: wakeup_state = 0;
1.1 cgd 997: mp = ⊤
998: space -= clen;
999: do {
1.45 tv 1000: if (uio == NULL) {
1001: /*
1002: * Data is prepackaged in "top".
1003: */
1004: resid = 0;
1005: if (flags & MSG_EOR)
1006: top->m_flags |= M_EOR;
1007: } else do {
1.160 ad 1008: sounlock(so);
1009: splx(s);
1.144 dyoung 1010: if (top == NULL) {
1.78 matt 1011: m = m_gethdr(M_WAIT, MT_DATA);
1.45 tv 1012: mlen = MHLEN;
1013: m->m_pkthdr.len = 0;
1.140 dyoung 1014: m->m_pkthdr.rcvif = NULL;
1.45 tv 1015: } else {
1.78 matt 1016: m = m_get(M_WAIT, MT_DATA);
1.45 tv 1017: mlen = MLEN;
1018: }
1.78 matt 1019: MCLAIM(m, so->so_snd.sb_mowner);
1.121 yamt 1020: if (sock_loan_thresh >= 0 &&
1021: uio->uio_iov->iov_len >= sock_loan_thresh &&
1022: space >= sock_loan_thresh &&
1.64 thorpej 1023: (len = sosend_loan(so, uio, m,
1024: space)) != 0) {
1025: SOSEND_COUNTER_INCR(&sosend_loan_big);
1026: space -= len;
1027: goto have_data;
1028: }
1.45 tv 1029: if (resid >= MINCLSIZE && space >= MCLBYTES) {
1.64 thorpej 1030: SOSEND_COUNTER_INCR(&sosend_copy_big);
1.201 oki 1031: m_clget(m, M_DONTWAIT);
1.45 tv 1032: if ((m->m_flags & M_EXT) == 0)
1033: goto nopages;
1034: mlen = MCLBYTES;
1035: if (atomic && top == 0) {
1.58 jdolecek 1036: len = lmin(MCLBYTES - max_hdr,
1.54 lukem 1037: resid);
1.45 tv 1038: m->m_data += max_hdr;
1039: } else
1.58 jdolecek 1040: len = lmin(MCLBYTES, resid);
1.45 tv 1041: space -= len;
1042: } else {
1.64 thorpej 1043: nopages:
1044: SOSEND_COUNTER_INCR(&sosend_copy_small);
1.58 jdolecek 1045: len = lmin(lmin(mlen, resid), space);
1.45 tv 1046: space -= len;
1047: /*
1048: * For datagram protocols, leave room
1049: * for protocol headers in first mbuf.
1050: */
1051: if (atomic && top == 0 && len < mlen)
1052: MH_ALIGN(m, len);
1053: }
1.144 dyoung 1054: error = uiomove(mtod(m, void *), (int)len, uio);
1.64 thorpej 1055: have_data:
1.45 tv 1056: resid = uio->uio_resid;
1057: m->m_len = len;
1058: *mp = m;
1059: top->m_pkthdr.len += len;
1.160 ad 1060: s = splsoftnet();
1061: solock(so);
1.144 dyoung 1062: if (error != 0)
1.45 tv 1063: goto release;
1064: mp = &m->m_next;
1065: if (resid <= 0) {
1066: if (flags & MSG_EOR)
1067: top->m_flags |= M_EOR;
1068: break;
1069: }
1070: } while (space > 0 && atomic);
1.108 perry 1071:
1.160 ad 1072: if (so->so_state & SS_CANTSENDMORE) {
1073: error = EPIPE;
1074: goto release;
1075: }
1.45 tv 1076: if (dontroute)
1077: so->so_options |= SO_DONTROUTE;
1078: if (resid > 0)
1079: so->so_state |= SS_MORETOCOME;
1.215.4.1 rmind 1080: error = (*so->so_proto->pr_usrreqs->pr_generic)(so,
1.46 sommerfe 1081: (flags & MSG_OOB) ? PRU_SENDOOB : PRU_SEND,
1.160 ad 1082: top, addr, control, curlwp);
1.45 tv 1083: if (dontroute)
1084: so->so_options &= ~SO_DONTROUTE;
1085: if (resid > 0)
1086: so->so_state &= ~SS_MORETOCOME;
1087: clen = 0;
1.144 dyoung 1088: control = NULL;
1089: top = NULL;
1.45 tv 1090: mp = ⊤
1.144 dyoung 1091: if (error != 0)
1.1 cgd 1092: goto release;
1093: } while (resid && space > 0);
1094: } while (resid);
1095:
1.54 lukem 1096: release:
1.1 cgd 1097: sbunlock(&so->so_snd);
1.54 lukem 1098: out:
1.160 ad 1099: sounlock(so);
1100: splx(s);
1.1 cgd 1101: if (top)
1102: m_freem(top);
1103: if (control)
1104: m_freem(control);
1105: return (error);
1106: }
1107:
1108: /*
1.159 ad 1109: * Following replacement or removal of the first mbuf on the first
1110: * mbuf chain of a socket buffer, push necessary state changes back
1111: * into the socket buffer so that other consumers see the values
1112: * consistently. 'nextrecord' is the callers locally stored value of
1113: * the original value of sb->sb_mb->m_nextpkt which must be restored
1114: * when the lead mbuf changes. NOTE: 'nextrecord' may be NULL.
1115: */
1116: static void
1117: sbsync(struct sockbuf *sb, struct mbuf *nextrecord)
1118: {
1119:
1.160 ad 1120: KASSERT(solocked(sb->sb_so));
1121:
1.159 ad 1122: /*
1123: * First, update for the new value of nextrecord. If necessary,
1124: * make it the first record.
1125: */
1126: if (sb->sb_mb != NULL)
1127: sb->sb_mb->m_nextpkt = nextrecord;
1128: else
1129: sb->sb_mb = nextrecord;
1130:
1131: /*
1132: * Now update any dependent socket buffer fields to reflect
1133: * the new state. This is an inline of SB_EMPTY_FIXUP, with
1134: * the addition of a second clause that takes care of the
1135: * case where sb_mb has been updated, but remains the last
1136: * record.
1137: */
1138: if (sb->sb_mb == NULL) {
1139: sb->sb_mbtail = NULL;
1140: sb->sb_lastrecord = NULL;
1141: } else if (sb->sb_mb->m_nextpkt == NULL)
1142: sb->sb_lastrecord = sb->sb_mb;
1143: }
1144:
1145: /*
1.1 cgd 1146: * Implement receive operations on a socket.
1147: * We depend on the way that records are added to the sockbuf
1148: * by sbappend*. In particular, each record (mbufs linked through m_next)
1149: * must begin with an address if the protocol so specifies,
1150: * followed by an optional mbuf or mbufs containing ancillary data,
1151: * and then zero or more mbufs of data.
1152: * In order to avoid blocking network interrupts for the entire time here,
1153: * we splx() while doing the actual copy to user space.
1154: * Although the sockbuf is locked, new data may still be appended,
1155: * and thus we must maintain consistency of the sockbuf during that time.
1156: *
1157: * The caller may receive the data as a single mbuf chain by supplying
1158: * an mbuf **mp0 for use in returning the chain. The uio is then used
1159: * only for the count in uio_resid.
1160: */
1.3 andrew 1161: int
1.54 lukem 1162: soreceive(struct socket *so, struct mbuf **paddr, struct uio *uio,
1163: struct mbuf **mp0, struct mbuf **controlp, int *flagsp)
1.1 cgd 1164: {
1.116 yamt 1165: struct lwp *l = curlwp;
1.160 ad 1166: struct mbuf *m, **mp, *mt;
1.211 chs 1167: size_t len, offset, moff, orig_resid;
1168: int atomic, flags, error, s, type;
1.99 matt 1169: const struct protosw *pr;
1.54 lukem 1170: struct mbuf *nextrecord;
1.67 he 1171: int mbuf_removed = 0;
1.146 dyoung 1172: const struct domain *dom;
1.196 dsl 1173: short wakeup_state = 0;
1.64 thorpej 1174:
1.54 lukem 1175: pr = so->so_proto;
1.146 dyoung 1176: atomic = pr->pr_flags & PR_ATOMIC;
1177: dom = pr->pr_domain;
1.1 cgd 1178: mp = mp0;
1.54 lukem 1179: type = 0;
1180: orig_resid = uio->uio_resid;
1.102 jonathan 1181:
1.144 dyoung 1182: if (paddr != NULL)
1183: *paddr = NULL;
1184: if (controlp != NULL)
1185: *controlp = NULL;
1186: if (flagsp != NULL)
1.1 cgd 1187: flags = *flagsp &~ MSG_EOR;
1188: else
1189: flags = 0;
1.66 enami 1190:
1.1 cgd 1191: if (flags & MSG_OOB) {
1192: m = m_get(M_WAIT, MT_DATA);
1.160 ad 1193: solock(so);
1.215.4.1 rmind 1194: error = (*pr->pr_usrreqs->pr_generic)(so, PRU_RCVOOB, m,
1.140 dyoung 1195: (struct mbuf *)(long)(flags & MSG_PEEK), NULL, l);
1.160 ad 1196: sounlock(so);
1.1 cgd 1197: if (error)
1198: goto bad;
1199: do {
1.134 christos 1200: error = uiomove(mtod(m, void *),
1.211 chs 1201: MIN(uio->uio_resid, m->m_len), uio);
1.1 cgd 1202: m = m_free(m);
1.144 dyoung 1203: } while (uio->uio_resid > 0 && error == 0 && m);
1.54 lukem 1204: bad:
1.144 dyoung 1205: if (m != NULL)
1.1 cgd 1206: m_freem(m);
1.144 dyoung 1207: return error;
1.1 cgd 1208: }
1.144 dyoung 1209: if (mp != NULL)
1.140 dyoung 1210: *mp = NULL;
1.160 ad 1211:
1212: /*
1213: * solock() provides atomicity of access. splsoftnet() prevents
1214: * protocol processing soft interrupts from interrupting us and
1215: * blocking (expensive).
1216: */
1217: s = splsoftnet();
1218: solock(so);
1.1 cgd 1219: if (so->so_state & SS_ISCONFIRMING && uio->uio_resid)
1.215.4.1 rmind 1220: (*pr->pr_usrreqs->pr_generic)(so, PRU_RCVD, NULL, NULL, NULL, l);
1.1 cgd 1221:
1.54 lukem 1222: restart:
1.160 ad 1223: if ((error = sblock(&so->so_rcv, SBLOCKWAIT(flags))) != 0) {
1224: sounlock(so);
1225: splx(s);
1.144 dyoung 1226: return error;
1.160 ad 1227: }
1.1 cgd 1228:
1229: m = so->so_rcv.sb_mb;
1230: /*
1231: * If we have less data than requested, block awaiting more
1232: * (subject to any timeout) if:
1.15 mycroft 1233: * 1. the current count is less than the low water mark,
1.1 cgd 1234: * 2. MSG_WAITALL is set, and it is possible to do the entire
1.15 mycroft 1235: * receive operation at once if we block (resid <= hiwat), or
1236: * 3. MSG_DONTWAIT is not set.
1.1 cgd 1237: * If MSG_WAITALL is set but resid is larger than the receive buffer,
1238: * we have to do the receive in sections, and thus risk returning
1239: * a short count if a timeout or signal occurs after we start.
1240: */
1.144 dyoung 1241: if (m == NULL ||
1242: ((flags & MSG_DONTWAIT) == 0 &&
1243: so->so_rcv.sb_cc < uio->uio_resid &&
1244: (so->so_rcv.sb_cc < so->so_rcv.sb_lowat ||
1245: ((flags & MSG_WAITALL) &&
1246: uio->uio_resid <= so->so_rcv.sb_hiwat)) &&
1.146 dyoung 1247: m->m_nextpkt == NULL && !atomic)) {
1.1 cgd 1248: #ifdef DIAGNOSTIC
1.144 dyoung 1249: if (m == NULL && so->so_rcv.sb_cc)
1.1 cgd 1250: panic("receive 1");
1251: #endif
1252: if (so->so_error) {
1.144 dyoung 1253: if (m != NULL)
1.15 mycroft 1254: goto dontblock;
1.1 cgd 1255: error = so->so_error;
1256: if ((flags & MSG_PEEK) == 0)
1257: so->so_error = 0;
1258: goto release;
1259: }
1260: if (so->so_state & SS_CANTRCVMORE) {
1.144 dyoung 1261: if (m != NULL)
1.15 mycroft 1262: goto dontblock;
1.1 cgd 1263: else
1264: goto release;
1265: }
1.144 dyoung 1266: for (; m != NULL; m = m->m_next)
1.1 cgd 1267: if (m->m_type == MT_OOBDATA || (m->m_flags & M_EOR)) {
1268: m = so->so_rcv.sb_mb;
1269: goto dontblock;
1270: }
1271: if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) == 0 &&
1272: (so->so_proto->pr_flags & PR_CONNREQUIRED)) {
1273: error = ENOTCONN;
1274: goto release;
1275: }
1276: if (uio->uio_resid == 0)
1277: goto release;
1.206 christos 1278: if ((so->so_state & SS_NBIO) ||
1279: (flags & (MSG_DONTWAIT|MSG_NBIO))) {
1.1 cgd 1280: error = EWOULDBLOCK;
1281: goto release;
1282: }
1.69 thorpej 1283: SBLASTRECORDCHK(&so->so_rcv, "soreceive sbwait 1");
1284: SBLASTMBUFCHK(&so->so_rcv, "soreceive sbwait 1");
1.1 cgd 1285: sbunlock(&so->so_rcv);
1.196 dsl 1286: if (wakeup_state & SS_RESTARTSYS)
1287: error = ERESTART;
1288: else
1289: error = sbwait(&so->so_rcv);
1.160 ad 1290: if (error != 0) {
1291: sounlock(so);
1292: splx(s);
1.144 dyoung 1293: return error;
1.160 ad 1294: }
1.196 dsl 1295: wakeup_state = so->so_state;
1.1 cgd 1296: goto restart;
1297: }
1.54 lukem 1298: dontblock:
1.69 thorpej 1299: /*
1300: * On entry here, m points to the first record of the socket buffer.
1.159 ad 1301: * From this point onward, we maintain 'nextrecord' as a cache of the
1302: * pointer to the next record in the socket buffer. We must keep the
1303: * various socket buffer pointers and local stack versions of the
1304: * pointers in sync, pushing out modifications before dropping the
1.160 ad 1305: * socket lock, and re-reading them when picking it up.
1.159 ad 1306: *
1307: * Otherwise, we will race with the network stack appending new data
1308: * or records onto the socket buffer by using inconsistent/stale
1309: * versions of the field, possibly resulting in socket buffer
1310: * corruption.
1311: *
1312: * By holding the high-level sblock(), we prevent simultaneous
1313: * readers from pulling off the front of the socket buffer.
1.69 thorpej 1314: */
1.144 dyoung 1315: if (l != NULL)
1.157 ad 1316: l->l_ru.ru_msgrcv++;
1.69 thorpej 1317: KASSERT(m == so->so_rcv.sb_mb);
1318: SBLASTRECORDCHK(&so->so_rcv, "soreceive 1");
1319: SBLASTMBUFCHK(&so->so_rcv, "soreceive 1");
1.1 cgd 1320: nextrecord = m->m_nextpkt;
1321: if (pr->pr_flags & PR_ADDR) {
1322: #ifdef DIAGNOSTIC
1323: if (m->m_type != MT_SONAME)
1324: panic("receive 1a");
1325: #endif
1.3 andrew 1326: orig_resid = 0;
1.1 cgd 1327: if (flags & MSG_PEEK) {
1328: if (paddr)
1329: *paddr = m_copy(m, 0, m->m_len);
1330: m = m->m_next;
1331: } else {
1332: sbfree(&so->so_rcv, m);
1.67 he 1333: mbuf_removed = 1;
1.144 dyoung 1334: if (paddr != NULL) {
1.1 cgd 1335: *paddr = m;
1336: so->so_rcv.sb_mb = m->m_next;
1.144 dyoung 1337: m->m_next = NULL;
1.1 cgd 1338: m = so->so_rcv.sb_mb;
1339: } else {
1340: MFREE(m, so->so_rcv.sb_mb);
1341: m = so->so_rcv.sb_mb;
1342: }
1.159 ad 1343: sbsync(&so->so_rcv, nextrecord);
1.1 cgd 1344: }
1345: }
1.159 ad 1346:
1347: /*
1348: * Process one or more MT_CONTROL mbufs present before any data mbufs
1349: * in the first mbuf chain on the socket buffer. If MSG_PEEK, we
1350: * just copy the data; if !MSG_PEEK, we call into the protocol to
1351: * perform externalization (or freeing if controlp == NULL).
1352: */
1353: if (__predict_false(m != NULL && m->m_type == MT_CONTROL)) {
1354: struct mbuf *cm = NULL, *cmn;
1355: struct mbuf **cme = &cm;
1356:
1357: do {
1358: if (flags & MSG_PEEK) {
1359: if (controlp != NULL) {
1360: *controlp = m_copy(m, 0, m->m_len);
1361: controlp = &(*controlp)->m_next;
1362: }
1363: m = m->m_next;
1364: } else {
1365: sbfree(&so->so_rcv, m);
1.1 cgd 1366: so->so_rcv.sb_mb = m->m_next;
1.144 dyoung 1367: m->m_next = NULL;
1.159 ad 1368: *cme = m;
1369: cme = &(*cme)->m_next;
1.1 cgd 1370: m = so->so_rcv.sb_mb;
1.159 ad 1371: }
1372: } while (m != NULL && m->m_type == MT_CONTROL);
1373: if ((flags & MSG_PEEK) == 0)
1374: sbsync(&so->so_rcv, nextrecord);
1375: for (; cm != NULL; cm = cmn) {
1376: cmn = cm->m_next;
1377: cm->m_next = NULL;
1378: type = mtod(cm, struct cmsghdr *)->cmsg_type;
1379: if (controlp != NULL) {
1380: if (dom->dom_externalize != NULL &&
1381: type == SCM_RIGHTS) {
1.160 ad 1382: sounlock(so);
1.159 ad 1383: splx(s);
1.204 christos 1384: error = (*dom->dom_externalize)(cm, l,
1385: (flags & MSG_CMSG_CLOEXEC) ?
1386: O_CLOEXEC : 0);
1.159 ad 1387: s = splsoftnet();
1.160 ad 1388: solock(so);
1.159 ad 1389: }
1390: *controlp = cm;
1391: while (*controlp != NULL)
1392: controlp = &(*controlp)->m_next;
1.1 cgd 1393: } else {
1.106 itojun 1394: /*
1395: * Dispose of any SCM_RIGHTS message that went
1396: * through the read path rather than recv.
1397: */
1.159 ad 1398: if (dom->dom_dispose != NULL &&
1399: type == SCM_RIGHTS) {
1.160 ad 1400: sounlock(so);
1.159 ad 1401: (*dom->dom_dispose)(cm);
1.160 ad 1402: solock(so);
1.159 ad 1403: }
1404: m_freem(cm);
1.1 cgd 1405: }
1406: }
1.159 ad 1407: if (m != NULL)
1408: nextrecord = so->so_rcv.sb_mb->m_nextpkt;
1409: else
1410: nextrecord = so->so_rcv.sb_mb;
1411: orig_resid = 0;
1.1 cgd 1412: }
1.69 thorpej 1413:
1.159 ad 1414: /* If m is non-NULL, we have some data to read. */
1415: if (__predict_true(m != NULL)) {
1.1 cgd 1416: type = m->m_type;
1417: if (type == MT_OOBDATA)
1418: flags |= MSG_OOB;
1419: }
1.69 thorpej 1420: SBLASTRECORDCHK(&so->so_rcv, "soreceive 2");
1421: SBLASTMBUFCHK(&so->so_rcv, "soreceive 2");
1422:
1.1 cgd 1423: moff = 0;
1424: offset = 0;
1.144 dyoung 1425: while (m != NULL && uio->uio_resid > 0 && error == 0) {
1.1 cgd 1426: if (m->m_type == MT_OOBDATA) {
1427: if (type != MT_OOBDATA)
1428: break;
1429: } else if (type == MT_OOBDATA)
1430: break;
1431: #ifdef DIAGNOSTIC
1432: else if (m->m_type != MT_DATA && m->m_type != MT_HEADER)
1433: panic("receive 3");
1434: #endif
1435: so->so_state &= ~SS_RCVATMARK;
1.196 dsl 1436: wakeup_state = 0;
1.1 cgd 1437: len = uio->uio_resid;
1438: if (so->so_oobmark && len > so->so_oobmark - offset)
1439: len = so->so_oobmark - offset;
1440: if (len > m->m_len - moff)
1441: len = m->m_len - moff;
1442: /*
1443: * If mp is set, just pass back the mbufs.
1444: * Otherwise copy them out via the uio, then free.
1445: * Sockbuf must be consistent here (points to current mbuf,
1446: * it points to next record) when we drop priority;
1447: * we must note any additions to the sockbuf when we
1448: * block interrupts again.
1449: */
1.144 dyoung 1450: if (mp == NULL) {
1.69 thorpej 1451: SBLASTRECORDCHK(&so->so_rcv, "soreceive uiomove");
1452: SBLASTMBUFCHK(&so->so_rcv, "soreceive uiomove");
1.160 ad 1453: sounlock(so);
1.1 cgd 1454: splx(s);
1.211 chs 1455: error = uiomove(mtod(m, char *) + moff, len, uio);
1.20 mycroft 1456: s = splsoftnet();
1.160 ad 1457: solock(so);
1.144 dyoung 1458: if (error != 0) {
1.67 he 1459: /*
1460: * If any part of the record has been removed
1461: * (such as the MT_SONAME mbuf, which will
1462: * happen when PR_ADDR, and thus also
1463: * PR_ATOMIC, is set), then drop the entire
1464: * record to maintain the atomicity of the
1465: * receive operation.
1466: *
1467: * This avoids a later panic("receive 1a")
1468: * when compiled with DIAGNOSTIC.
1469: */
1.146 dyoung 1470: if (m && mbuf_removed && atomic)
1.67 he 1471: (void) sbdroprecord(&so->so_rcv);
1472:
1.57 jdolecek 1473: goto release;
1.67 he 1474: }
1.1 cgd 1475: } else
1476: uio->uio_resid -= len;
1477: if (len == m->m_len - moff) {
1478: if (m->m_flags & M_EOR)
1479: flags |= MSG_EOR;
1480: if (flags & MSG_PEEK) {
1481: m = m->m_next;
1482: moff = 0;
1483: } else {
1484: nextrecord = m->m_nextpkt;
1485: sbfree(&so->so_rcv, m);
1486: if (mp) {
1487: *mp = m;
1488: mp = &m->m_next;
1489: so->so_rcv.sb_mb = m = m->m_next;
1.140 dyoung 1490: *mp = NULL;
1.1 cgd 1491: } else {
1492: MFREE(m, so->so_rcv.sb_mb);
1493: m = so->so_rcv.sb_mb;
1494: }
1.69 thorpej 1495: /*
1496: * If m != NULL, we also know that
1497: * so->so_rcv.sb_mb != NULL.
1498: */
1499: KASSERT(so->so_rcv.sb_mb == m);
1500: if (m) {
1.1 cgd 1501: m->m_nextpkt = nextrecord;
1.69 thorpej 1502: if (nextrecord == NULL)
1503: so->so_rcv.sb_lastrecord = m;
1504: } else {
1505: so->so_rcv.sb_mb = nextrecord;
1.70 thorpej 1506: SB_EMPTY_FIXUP(&so->so_rcv);
1.69 thorpej 1507: }
1508: SBLASTRECORDCHK(&so->so_rcv, "soreceive 3");
1509: SBLASTMBUFCHK(&so->so_rcv, "soreceive 3");
1.1 cgd 1510: }
1.144 dyoung 1511: } else if (flags & MSG_PEEK)
1512: moff += len;
1513: else {
1.160 ad 1514: if (mp != NULL) {
1515: mt = m_copym(m, 0, len, M_NOWAIT);
1516: if (__predict_false(mt == NULL)) {
1517: sounlock(so);
1518: mt = m_copym(m, 0, len, M_WAIT);
1519: solock(so);
1520: }
1521: *mp = mt;
1522: }
1.144 dyoung 1523: m->m_data += len;
1524: m->m_len -= len;
1525: so->so_rcv.sb_cc -= len;
1.1 cgd 1526: }
1527: if (so->so_oobmark) {
1528: if ((flags & MSG_PEEK) == 0) {
1529: so->so_oobmark -= len;
1530: if (so->so_oobmark == 0) {
1531: so->so_state |= SS_RCVATMARK;
1532: break;
1533: }
1.7 cgd 1534: } else {
1.1 cgd 1535: offset += len;
1.7 cgd 1536: if (offset == so->so_oobmark)
1537: break;
1538: }
1.1 cgd 1539: }
1540: if (flags & MSG_EOR)
1541: break;
1542: /*
1543: * If the MSG_WAITALL flag is set (for non-atomic socket),
1544: * we must not quit until "uio->uio_resid == 0" or an error
1545: * termination. If a signal/timeout occurs, return
1546: * with a short count but without error.
1547: * Keep sockbuf locked against other readers.
1548: */
1.144 dyoung 1549: while (flags & MSG_WAITALL && m == NULL && uio->uio_resid > 0 &&
1.3 andrew 1550: !sosendallatonce(so) && !nextrecord) {
1.1 cgd 1551: if (so->so_error || so->so_state & SS_CANTRCVMORE)
1552: break;
1.68 matt 1553: /*
1554: * If we are peeking and the socket receive buffer is
1555: * full, stop since we can't get more data to peek at.
1556: */
1557: if ((flags & MSG_PEEK) && sbspace(&so->so_rcv) <= 0)
1558: break;
1559: /*
1560: * If we've drained the socket buffer, tell the
1561: * protocol in case it needs to do something to
1562: * get it filled again.
1563: */
1564: if ((pr->pr_flags & PR_WANTRCVD) && so->so_pcb)
1.215.4.1 rmind 1565: (*pr->pr_usrreqs->pr_generic)(so, PRU_RCVD,
1.140 dyoung 1566: NULL, (struct mbuf *)(long)flags, NULL, l);
1.69 thorpej 1567: SBLASTRECORDCHK(&so->so_rcv, "soreceive sbwait 2");
1568: SBLASTMBUFCHK(&so->so_rcv, "soreceive sbwait 2");
1.196 dsl 1569: if (wakeup_state & SS_RESTARTSYS)
1570: error = ERESTART;
1571: else
1572: error = sbwait(&so->so_rcv);
1.144 dyoung 1573: if (error != 0) {
1.1 cgd 1574: sbunlock(&so->so_rcv);
1.160 ad 1575: sounlock(so);
1.1 cgd 1576: splx(s);
1.144 dyoung 1577: return 0;
1.1 cgd 1578: }
1.21 christos 1579: if ((m = so->so_rcv.sb_mb) != NULL)
1.1 cgd 1580: nextrecord = m->m_nextpkt;
1.196 dsl 1581: wakeup_state = so->so_state;
1.1 cgd 1582: }
1583: }
1.3 andrew 1584:
1.146 dyoung 1585: if (m && atomic) {
1.3 andrew 1586: flags |= MSG_TRUNC;
1587: if ((flags & MSG_PEEK) == 0)
1588: (void) sbdroprecord(&so->so_rcv);
1589: }
1.1 cgd 1590: if ((flags & MSG_PEEK) == 0) {
1.144 dyoung 1591: if (m == NULL) {
1.69 thorpej 1592: /*
1.70 thorpej 1593: * First part is an inline SB_EMPTY_FIXUP(). Second
1.69 thorpej 1594: * part makes sure sb_lastrecord is up-to-date if
1595: * there is still data in the socket buffer.
1596: */
1.1 cgd 1597: so->so_rcv.sb_mb = nextrecord;
1.69 thorpej 1598: if (so->so_rcv.sb_mb == NULL) {
1599: so->so_rcv.sb_mbtail = NULL;
1600: so->so_rcv.sb_lastrecord = NULL;
1601: } else if (nextrecord->m_nextpkt == NULL)
1602: so->so_rcv.sb_lastrecord = nextrecord;
1603: }
1604: SBLASTRECORDCHK(&so->so_rcv, "soreceive 4");
1605: SBLASTMBUFCHK(&so->so_rcv, "soreceive 4");
1.1 cgd 1606: if (pr->pr_flags & PR_WANTRCVD && so->so_pcb)
1.215.4.1 rmind 1607: (*pr->pr_usrreqs->pr_generic)(so, PRU_RCVD, NULL,
1.140 dyoung 1608: (struct mbuf *)(long)flags, NULL, l);
1.1 cgd 1609: }
1.3 andrew 1610: if (orig_resid == uio->uio_resid && orig_resid &&
1611: (flags & MSG_EOR) == 0 && (so->so_state & SS_CANTRCVMORE) == 0) {
1612: sbunlock(&so->so_rcv);
1613: goto restart;
1614: }
1.108 perry 1615:
1.144 dyoung 1616: if (flagsp != NULL)
1.1 cgd 1617: *flagsp |= flags;
1.54 lukem 1618: release:
1.1 cgd 1619: sbunlock(&so->so_rcv);
1.160 ad 1620: sounlock(so);
1.1 cgd 1621: splx(s);
1.144 dyoung 1622: return error;
1.1 cgd 1623: }
1624:
1.14 mycroft 1625: int
1.54 lukem 1626: soshutdown(struct socket *so, int how)
1.1 cgd 1627: {
1.99 matt 1628: const struct protosw *pr;
1.160 ad 1629: int error;
1630:
1631: KASSERT(solocked(so));
1.34 kleink 1632:
1.54 lukem 1633: pr = so->so_proto;
1.34 kleink 1634: if (!(how == SHUT_RD || how == SHUT_WR || how == SHUT_RDWR))
1635: return (EINVAL);
1.1 cgd 1636:
1.160 ad 1637: if (how == SHUT_RD || how == SHUT_RDWR) {
1.1 cgd 1638: sorflush(so);
1.160 ad 1639: error = 0;
1640: }
1.34 kleink 1641: if (how == SHUT_WR || how == SHUT_RDWR)
1.215.4.1 rmind 1642: error = (*pr->pr_usrreqs->pr_generic)(so, PRU_SHUTDOWN, NULL,
1.140 dyoung 1643: NULL, NULL, NULL);
1.160 ad 1644:
1645: return error;
1.1 cgd 1646: }
1647:
1.195 dsl 1648: void
1.196 dsl 1649: sorestart(struct socket *so)
1.188 ad 1650: {
1.196 dsl 1651: /*
1652: * An application has called close() on an fd on which another
1653: * of its threads has called a socket system call.
1654: * Mark this and wake everyone up, and code that would block again
1655: * instead returns ERESTART.
1656: * On system call re-entry the fd is validated and EBADF returned.
1657: * Any other fd will block again on the 2nd syscall.
1658: */
1.188 ad 1659: solock(so);
1.196 dsl 1660: so->so_state |= SS_RESTARTSYS;
1.188 ad 1661: cv_broadcast(&so->so_cv);
1.196 dsl 1662: cv_broadcast(&so->so_snd.sb_cv);
1663: cv_broadcast(&so->so_rcv.sb_cv);
1.188 ad 1664: sounlock(so);
1665: }
1666:
1.14 mycroft 1667: void
1.54 lukem 1668: sorflush(struct socket *so)
1.1 cgd 1669: {
1.54 lukem 1670: struct sockbuf *sb, asb;
1.99 matt 1671: const struct protosw *pr;
1.160 ad 1672:
1673: KASSERT(solocked(so));
1.1 cgd 1674:
1.54 lukem 1675: sb = &so->so_rcv;
1676: pr = so->so_proto;
1.160 ad 1677: socantrcvmore(so);
1.1 cgd 1678: sb->sb_flags |= SB_NOINTR;
1.160 ad 1679: (void )sblock(sb, M_WAITOK);
1.1 cgd 1680: sbunlock(sb);
1681: asb = *sb;
1.86 wrstuden 1682: /*
1683: * Clear most of the sockbuf structure, but leave some of the
1684: * fields valid.
1685: */
1686: memset(&sb->sb_startzero, 0,
1687: sizeof(*sb) - offsetof(struct sockbuf, sb_startzero));
1.160 ad 1688: if (pr->pr_flags & PR_RIGHTS && pr->pr_domain->dom_dispose) {
1689: sounlock(so);
1.1 cgd 1690: (*pr->pr_domain->dom_dispose)(asb.sb_mb);
1.160 ad 1691: solock(so);
1692: }
1.98 christos 1693: sbrelease(&asb, so);
1.1 cgd 1694: }
1695:
1.171 plunky 1696: /*
1697: * internal set SOL_SOCKET options
1698: */
1.142 dyoung 1699: static int
1.171 plunky 1700: sosetopt1(struct socket *so, const struct sockopt *sopt)
1.1 cgd 1701: {
1.182 christos 1702: int error = EINVAL, optval, opt;
1.171 plunky 1703: struct linger l;
1704: struct timeval tv;
1.142 dyoung 1705:
1.179 christos 1706: switch ((opt = sopt->sopt_name)) {
1.142 dyoung 1707:
1.170 tls 1708: case SO_ACCEPTFILTER:
1.177 ad 1709: error = accept_filt_setopt(so, sopt);
1710: KASSERT(solocked(so));
1.170 tls 1711: break;
1712:
1.171 plunky 1713: case SO_LINGER:
1714: error = sockopt_get(sopt, &l, sizeof(l));
1.177 ad 1715: solock(so);
1.171 plunky 1716: if (error)
1.177 ad 1717: break;
1.171 plunky 1718: if (l.l_linger < 0 || l.l_linger > USHRT_MAX ||
1.177 ad 1719: l.l_linger > (INT_MAX / hz)) {
1720: error = EDOM;
1721: break;
1722: }
1.171 plunky 1723: so->so_linger = l.l_linger;
1724: if (l.l_onoff)
1725: so->so_options |= SO_LINGER;
1726: else
1727: so->so_options &= ~SO_LINGER;
1.177 ad 1728: break;
1.1 cgd 1729:
1.142 dyoung 1730: case SO_DEBUG:
1731: case SO_KEEPALIVE:
1732: case SO_DONTROUTE:
1733: case SO_USELOOPBACK:
1734: case SO_BROADCAST:
1735: case SO_REUSEADDR:
1736: case SO_REUSEPORT:
1737: case SO_OOBINLINE:
1738: case SO_TIMESTAMP:
1.207 christos 1739: case SO_NOSIGPIPE:
1.184 christos 1740: #ifdef SO_OTIMESTAMP
1741: case SO_OTIMESTAMP:
1742: #endif
1.171 plunky 1743: error = sockopt_getint(sopt, &optval);
1.177 ad 1744: solock(so);
1.171 plunky 1745: if (error)
1.177 ad 1746: break;
1.171 plunky 1747: if (optval)
1.179 christos 1748: so->so_options |= opt;
1.142 dyoung 1749: else
1.179 christos 1750: so->so_options &= ~opt;
1.142 dyoung 1751: break;
1752:
1753: case SO_SNDBUF:
1754: case SO_RCVBUF:
1755: case SO_SNDLOWAT:
1756: case SO_RCVLOWAT:
1.171 plunky 1757: error = sockopt_getint(sopt, &optval);
1.177 ad 1758: solock(so);
1.171 plunky 1759: if (error)
1.177 ad 1760: break;
1.1 cgd 1761:
1.142 dyoung 1762: /*
1763: * Values < 1 make no sense for any of these
1764: * options, so disallow them.
1765: */
1.177 ad 1766: if (optval < 1) {
1767: error = EINVAL;
1768: break;
1769: }
1.1 cgd 1770:
1.179 christos 1771: switch (opt) {
1.171 plunky 1772: case SO_SNDBUF:
1.177 ad 1773: if (sbreserve(&so->so_snd, (u_long)optval, so) == 0) {
1774: error = ENOBUFS;
1775: break;
1776: }
1.171 plunky 1777: so->so_snd.sb_flags &= ~SB_AUTOSIZE;
1778: break;
1.1 cgd 1779:
1780: case SO_RCVBUF:
1.177 ad 1781: if (sbreserve(&so->so_rcv, (u_long)optval, so) == 0) {
1782: error = ENOBUFS;
1783: break;
1784: }
1.171 plunky 1785: so->so_rcv.sb_flags &= ~SB_AUTOSIZE;
1.142 dyoung 1786: break;
1787:
1788: /*
1789: * Make sure the low-water is never greater than
1790: * the high-water.
1791: */
1.1 cgd 1792: case SO_SNDLOWAT:
1.171 plunky 1793: if (optval > so->so_snd.sb_hiwat)
1794: optval = so->so_snd.sb_hiwat;
1795:
1796: so->so_snd.sb_lowat = optval;
1.142 dyoung 1797: break;
1.171 plunky 1798:
1.1 cgd 1799: case SO_RCVLOWAT:
1.171 plunky 1800: if (optval > so->so_rcv.sb_hiwat)
1801: optval = so->so_rcv.sb_hiwat;
1802:
1803: so->so_rcv.sb_lowat = optval;
1.142 dyoung 1804: break;
1805: }
1806: break;
1.28 thorpej 1807:
1.179 christos 1808: #ifdef COMPAT_50
1809: case SO_OSNDTIMEO:
1810: case SO_ORCVTIMEO: {
1811: struct timeval50 otv;
1812: error = sockopt_get(sopt, &otv, sizeof(otv));
1.186 pooka 1813: if (error) {
1814: solock(so);
1.183 christos 1815: break;
1.186 pooka 1816: }
1.179 christos 1817: timeval50_to_timeval(&otv, &tv);
1818: opt = opt == SO_OSNDTIMEO ? SO_SNDTIMEO : SO_RCVTIMEO;
1.182 christos 1819: error = 0;
1.179 christos 1820: /*FALLTHROUGH*/
1821: }
1822: #endif /* COMPAT_50 */
1823:
1.142 dyoung 1824: case SO_SNDTIMEO:
1825: case SO_RCVTIMEO:
1.182 christos 1826: if (error)
1.179 christos 1827: error = sockopt_get(sopt, &tv, sizeof(tv));
1.177 ad 1828: solock(so);
1.171 plunky 1829: if (error)
1.177 ad 1830: break;
1.171 plunky 1831:
1.177 ad 1832: if (tv.tv_sec > (INT_MAX - tv.tv_usec / tick) / hz) {
1833: error = EDOM;
1834: break;
1835: }
1.28 thorpej 1836:
1.171 plunky 1837: optval = tv.tv_sec * hz + tv.tv_usec / tick;
1838: if (optval == 0 && tv.tv_usec != 0)
1839: optval = 1;
1.28 thorpej 1840:
1.179 christos 1841: switch (opt) {
1.142 dyoung 1842: case SO_SNDTIMEO:
1.171 plunky 1843: so->so_snd.sb_timeo = optval;
1.1 cgd 1844: break;
1845: case SO_RCVTIMEO:
1.171 plunky 1846: so->so_rcv.sb_timeo = optval;
1.142 dyoung 1847: break;
1848: }
1849: break;
1.1 cgd 1850:
1.142 dyoung 1851: default:
1.177 ad 1852: solock(so);
1853: error = ENOPROTOOPT;
1854: break;
1.142 dyoung 1855: }
1.177 ad 1856: KASSERT(solocked(so));
1857: return error;
1.142 dyoung 1858: }
1.1 cgd 1859:
1.142 dyoung 1860: int
1.171 plunky 1861: sosetopt(struct socket *so, struct sockopt *sopt)
1.142 dyoung 1862: {
1863: int error, prerr;
1.1 cgd 1864:
1.177 ad 1865: if (sopt->sopt_level == SOL_SOCKET) {
1.171 plunky 1866: error = sosetopt1(so, sopt);
1.177 ad 1867: KASSERT(solocked(so));
1868: } else {
1.142 dyoung 1869: error = ENOPROTOOPT;
1.177 ad 1870: solock(so);
1871: }
1.1 cgd 1872:
1.142 dyoung 1873: if ((error == 0 || error == ENOPROTOOPT) &&
1874: so->so_proto != NULL && so->so_proto->pr_ctloutput != NULL) {
1875: /* give the protocol stack a shot */
1.171 plunky 1876: prerr = (*so->so_proto->pr_ctloutput)(PRCO_SETOPT, so, sopt);
1.142 dyoung 1877: if (prerr == 0)
1878: error = 0;
1879: else if (prerr != ENOPROTOOPT)
1880: error = prerr;
1.171 plunky 1881: }
1.160 ad 1882: sounlock(so);
1.142 dyoung 1883: return error;
1.1 cgd 1884: }
1885:
1.171 plunky 1886: /*
1887: * so_setsockopt() is a wrapper providing a sockopt structure for sosetopt()
1888: */
1889: int
1890: so_setsockopt(struct lwp *l, struct socket *so, int level, int name,
1891: const void *val, size_t valsize)
1892: {
1893: struct sockopt sopt;
1894: int error;
1895:
1896: KASSERT(valsize == 0 || val != NULL);
1897:
1898: sockopt_init(&sopt, level, name, valsize);
1899: sockopt_set(&sopt, val, valsize);
1900:
1901: error = sosetopt(so, &sopt);
1902:
1903: sockopt_destroy(&sopt);
1904:
1905: return error;
1906: }
1907:
1908: /*
1909: * internal get SOL_SOCKET options
1910: */
1911: static int
1912: sogetopt1(struct socket *so, struct sockopt *sopt)
1913: {
1.179 christos 1914: int error, optval, opt;
1.171 plunky 1915: struct linger l;
1916: struct timeval tv;
1917:
1.179 christos 1918: switch ((opt = sopt->sopt_name)) {
1.171 plunky 1919:
1920: case SO_ACCEPTFILTER:
1.177 ad 1921: error = accept_filt_getopt(so, sopt);
1.171 plunky 1922: break;
1923:
1924: case SO_LINGER:
1925: l.l_onoff = (so->so_options & SO_LINGER) ? 1 : 0;
1926: l.l_linger = so->so_linger;
1927:
1928: error = sockopt_set(sopt, &l, sizeof(l));
1929: break;
1930:
1931: case SO_USELOOPBACK:
1932: case SO_DONTROUTE:
1933: case SO_DEBUG:
1934: case SO_KEEPALIVE:
1935: case SO_REUSEADDR:
1936: case SO_REUSEPORT:
1937: case SO_BROADCAST:
1938: case SO_OOBINLINE:
1939: case SO_TIMESTAMP:
1.207 christos 1940: case SO_NOSIGPIPE:
1.184 christos 1941: #ifdef SO_OTIMESTAMP
1942: case SO_OTIMESTAMP:
1943: #endif
1.179 christos 1944: error = sockopt_setint(sopt, (so->so_options & opt) ? 1 : 0);
1.171 plunky 1945: break;
1946:
1947: case SO_TYPE:
1948: error = sockopt_setint(sopt, so->so_type);
1949: break;
1950:
1951: case SO_ERROR:
1952: error = sockopt_setint(sopt, so->so_error);
1953: so->so_error = 0;
1954: break;
1955:
1956: case SO_SNDBUF:
1957: error = sockopt_setint(sopt, so->so_snd.sb_hiwat);
1958: break;
1959:
1960: case SO_RCVBUF:
1961: error = sockopt_setint(sopt, so->so_rcv.sb_hiwat);
1962: break;
1963:
1964: case SO_SNDLOWAT:
1965: error = sockopt_setint(sopt, so->so_snd.sb_lowat);
1966: break;
1967:
1968: case SO_RCVLOWAT:
1969: error = sockopt_setint(sopt, so->so_rcv.sb_lowat);
1970: break;
1971:
1.179 christos 1972: #ifdef COMPAT_50
1973: case SO_OSNDTIMEO:
1974: case SO_ORCVTIMEO: {
1975: struct timeval50 otv;
1976:
1977: optval = (opt == SO_OSNDTIMEO ?
1978: so->so_snd.sb_timeo : so->so_rcv.sb_timeo);
1979:
1980: otv.tv_sec = optval / hz;
1981: otv.tv_usec = (optval % hz) * tick;
1982:
1983: error = sockopt_set(sopt, &otv, sizeof(otv));
1984: break;
1985: }
1986: #endif /* COMPAT_50 */
1987:
1.171 plunky 1988: case SO_SNDTIMEO:
1989: case SO_RCVTIMEO:
1.179 christos 1990: optval = (opt == SO_SNDTIMEO ?
1.171 plunky 1991: so->so_snd.sb_timeo : so->so_rcv.sb_timeo);
1992:
1993: tv.tv_sec = optval / hz;
1994: tv.tv_usec = (optval % hz) * tick;
1995:
1996: error = sockopt_set(sopt, &tv, sizeof(tv));
1997: break;
1998:
1999: case SO_OVERFLOWED:
2000: error = sockopt_setint(sopt, so->so_rcv.sb_overflowed);
2001: break;
2002:
2003: default:
2004: error = ENOPROTOOPT;
2005: break;
2006: }
2007:
2008: return (error);
2009: }
2010:
1.14 mycroft 2011: int
1.171 plunky 2012: sogetopt(struct socket *so, struct sockopt *sopt)
1.1 cgd 2013: {
1.160 ad 2014: int error;
1.1 cgd 2015:
1.160 ad 2016: solock(so);
1.171 plunky 2017: if (sopt->sopt_level != SOL_SOCKET) {
1.1 cgd 2018: if (so->so_proto && so->so_proto->pr_ctloutput) {
1.160 ad 2019: error = ((*so->so_proto->pr_ctloutput)
1.171 plunky 2020: (PRCO_GETOPT, so, sopt));
1.1 cgd 2021: } else
1.160 ad 2022: error = (ENOPROTOOPT);
1.1 cgd 2023: } else {
1.171 plunky 2024: error = sogetopt1(so, sopt);
2025: }
2026: sounlock(so);
2027: return (error);
2028: }
2029:
2030: /*
2031: * alloc sockopt data buffer buffer
2032: * - will be released at destroy
2033: */
1.176 plunky 2034: static int
2035: sockopt_alloc(struct sockopt *sopt, size_t len, km_flag_t kmflag)
1.171 plunky 2036: {
2037:
2038: KASSERT(sopt->sopt_size == 0);
2039:
1.176 plunky 2040: if (len > sizeof(sopt->sopt_buf)) {
2041: sopt->sopt_data = kmem_zalloc(len, kmflag);
2042: if (sopt->sopt_data == NULL)
2043: return ENOMEM;
2044: } else
1.171 plunky 2045: sopt->sopt_data = sopt->sopt_buf;
2046:
2047: sopt->sopt_size = len;
1.176 plunky 2048: return 0;
1.171 plunky 2049: }
2050:
2051: /*
2052: * initialise sockopt storage
1.176 plunky 2053: * - MAY sleep during allocation
1.171 plunky 2054: */
2055: void
2056: sockopt_init(struct sockopt *sopt, int level, int name, size_t size)
2057: {
1.1 cgd 2058:
1.171 plunky 2059: memset(sopt, 0, sizeof(*sopt));
1.1 cgd 2060:
1.171 plunky 2061: sopt->sopt_level = level;
2062: sopt->sopt_name = name;
1.176 plunky 2063: (void)sockopt_alloc(sopt, size, KM_SLEEP);
1.171 plunky 2064: }
2065:
2066: /*
2067: * destroy sockopt storage
2068: * - will release any held memory references
2069: */
2070: void
2071: sockopt_destroy(struct sockopt *sopt)
2072: {
2073:
2074: if (sopt->sopt_data != sopt->sopt_buf)
1.173 plunky 2075: kmem_free(sopt->sopt_data, sopt->sopt_size);
1.171 plunky 2076:
2077: memset(sopt, 0, sizeof(*sopt));
2078: }
2079:
2080: /*
2081: * set sockopt value
2082: * - value is copied into sockopt
1.176 plunky 2083: * - memory is allocated when necessary, will not sleep
1.171 plunky 2084: */
2085: int
2086: sockopt_set(struct sockopt *sopt, const void *buf, size_t len)
2087: {
1.176 plunky 2088: int error;
1.171 plunky 2089:
1.176 plunky 2090: if (sopt->sopt_size == 0) {
2091: error = sockopt_alloc(sopt, len, KM_NOSLEEP);
2092: if (error)
2093: return error;
2094: }
1.171 plunky 2095:
2096: KASSERT(sopt->sopt_size == len);
2097: memcpy(sopt->sopt_data, buf, len);
2098: return 0;
2099: }
2100:
2101: /*
2102: * common case of set sockopt integer value
2103: */
2104: int
2105: sockopt_setint(struct sockopt *sopt, int val)
2106: {
2107:
2108: return sockopt_set(sopt, &val, sizeof(int));
2109: }
2110:
2111: /*
2112: * get sockopt value
2113: * - correct size must be given
2114: */
2115: int
2116: sockopt_get(const struct sockopt *sopt, void *buf, size_t len)
2117: {
1.170 tls 2118:
1.171 plunky 2119: if (sopt->sopt_size != len)
2120: return EINVAL;
1.1 cgd 2121:
1.171 plunky 2122: memcpy(buf, sopt->sopt_data, len);
2123: return 0;
2124: }
1.1 cgd 2125:
1.171 plunky 2126: /*
2127: * common case of get sockopt integer value
2128: */
2129: int
2130: sockopt_getint(const struct sockopt *sopt, int *valp)
2131: {
1.1 cgd 2132:
1.171 plunky 2133: return sockopt_get(sopt, valp, sizeof(int));
2134: }
1.1 cgd 2135:
1.171 plunky 2136: /*
2137: * set sockopt value from mbuf
2138: * - ONLY for legacy code
2139: * - mbuf is released by sockopt
1.176 plunky 2140: * - will not sleep
1.171 plunky 2141: */
2142: int
2143: sockopt_setmbuf(struct sockopt *sopt, struct mbuf *m)
2144: {
2145: size_t len;
1.176 plunky 2146: int error;
1.1 cgd 2147:
1.171 plunky 2148: len = m_length(m);
1.1 cgd 2149:
1.176 plunky 2150: if (sopt->sopt_size == 0) {
2151: error = sockopt_alloc(sopt, len, KM_NOSLEEP);
2152: if (error)
2153: return error;
2154: }
1.1 cgd 2155:
1.171 plunky 2156: KASSERT(sopt->sopt_size == len);
2157: m_copydata(m, 0, len, sopt->sopt_data);
2158: m_freem(m);
1.1 cgd 2159:
1.171 plunky 2160: return 0;
2161: }
1.1 cgd 2162:
1.171 plunky 2163: /*
2164: * get sockopt value into mbuf
2165: * - ONLY for legacy code
2166: * - mbuf to be released by the caller
1.176 plunky 2167: * - will not sleep
1.171 plunky 2168: */
2169: struct mbuf *
2170: sockopt_getmbuf(const struct sockopt *sopt)
2171: {
2172: struct mbuf *m;
1.107 darrenr 2173:
1.176 plunky 2174: if (sopt->sopt_size > MCLBYTES)
2175: return NULL;
2176:
2177: m = m_get(M_DONTWAIT, MT_SOOPTS);
1.171 plunky 2178: if (m == NULL)
2179: return NULL;
2180:
1.176 plunky 2181: if (sopt->sopt_size > MLEN) {
2182: MCLGET(m, M_DONTWAIT);
2183: if ((m->m_flags & M_EXT) == 0) {
2184: m_free(m);
2185: return NULL;
2186: }
1.1 cgd 2187: }
1.176 plunky 2188:
2189: memcpy(mtod(m, void *), sopt->sopt_data, sopt->sopt_size);
2190: m->m_len = sopt->sopt_size;
1.160 ad 2191:
1.171 plunky 2192: return m;
1.1 cgd 2193: }
2194:
1.14 mycroft 2195: void
1.54 lukem 2196: sohasoutofband(struct socket *so)
1.1 cgd 2197: {
1.153 rmind 2198:
1.90 christos 2199: fownsignal(so->so_pgid, SIGURG, POLL_PRI, POLLPRI|POLLRDBAND, so);
1.189 ad 2200: selnotify(&so->so_rcv.sb_sel, POLLPRI | POLLRDBAND, NOTE_SUBMIT);
1.1 cgd 2201: }
1.72 jdolecek 2202:
2203: static void
2204: filt_sordetach(struct knote *kn)
2205: {
2206: struct socket *so;
2207:
1.155 ad 2208: so = ((file_t *)kn->kn_obj)->f_data;
1.160 ad 2209: solock(so);
1.73 christos 2210: SLIST_REMOVE(&so->so_rcv.sb_sel.sel_klist, kn, knote, kn_selnext);
2211: if (SLIST_EMPTY(&so->so_rcv.sb_sel.sel_klist))
1.72 jdolecek 2212: so->so_rcv.sb_flags &= ~SB_KNOTE;
1.160 ad 2213: sounlock(so);
1.72 jdolecek 2214: }
2215:
2216: /*ARGSUSED*/
2217: static int
1.129 yamt 2218: filt_soread(struct knote *kn, long hint)
1.72 jdolecek 2219: {
2220: struct socket *so;
1.160 ad 2221: int rv;
1.72 jdolecek 2222:
1.155 ad 2223: so = ((file_t *)kn->kn_obj)->f_data;
1.160 ad 2224: if (hint != NOTE_SUBMIT)
2225: solock(so);
1.72 jdolecek 2226: kn->kn_data = so->so_rcv.sb_cc;
2227: if (so->so_state & SS_CANTRCVMORE) {
1.108 perry 2228: kn->kn_flags |= EV_EOF;
1.72 jdolecek 2229: kn->kn_fflags = so->so_error;
1.160 ad 2230: rv = 1;
2231: } else if (so->so_error) /* temporary udp error */
2232: rv = 1;
2233: else if (kn->kn_sfflags & NOTE_LOWAT)
2234: rv = (kn->kn_data >= kn->kn_sdata);
2235: else
2236: rv = (kn->kn_data >= so->so_rcv.sb_lowat);
2237: if (hint != NOTE_SUBMIT)
2238: sounlock(so);
2239: return rv;
1.72 jdolecek 2240: }
2241:
2242: static void
2243: filt_sowdetach(struct knote *kn)
2244: {
2245: struct socket *so;
2246:
1.155 ad 2247: so = ((file_t *)kn->kn_obj)->f_data;
1.160 ad 2248: solock(so);
1.73 christos 2249: SLIST_REMOVE(&so->so_snd.sb_sel.sel_klist, kn, knote, kn_selnext);
2250: if (SLIST_EMPTY(&so->so_snd.sb_sel.sel_klist))
1.72 jdolecek 2251: so->so_snd.sb_flags &= ~SB_KNOTE;
1.160 ad 2252: sounlock(so);
1.72 jdolecek 2253: }
2254:
2255: /*ARGSUSED*/
2256: static int
1.129 yamt 2257: filt_sowrite(struct knote *kn, long hint)
1.72 jdolecek 2258: {
2259: struct socket *so;
1.160 ad 2260: int rv;
1.72 jdolecek 2261:
1.155 ad 2262: so = ((file_t *)kn->kn_obj)->f_data;
1.160 ad 2263: if (hint != NOTE_SUBMIT)
2264: solock(so);
1.72 jdolecek 2265: kn->kn_data = sbspace(&so->so_snd);
2266: if (so->so_state & SS_CANTSENDMORE) {
1.108 perry 2267: kn->kn_flags |= EV_EOF;
1.72 jdolecek 2268: kn->kn_fflags = so->so_error;
1.160 ad 2269: rv = 1;
2270: } else if (so->so_error) /* temporary udp error */
2271: rv = 1;
2272: else if (((so->so_state & SS_ISCONNECTED) == 0) &&
1.72 jdolecek 2273: (so->so_proto->pr_flags & PR_CONNREQUIRED))
1.160 ad 2274: rv = 0;
2275: else if (kn->kn_sfflags & NOTE_LOWAT)
2276: rv = (kn->kn_data >= kn->kn_sdata);
2277: else
2278: rv = (kn->kn_data >= so->so_snd.sb_lowat);
2279: if (hint != NOTE_SUBMIT)
2280: sounlock(so);
2281: return rv;
1.72 jdolecek 2282: }
2283:
2284: /*ARGSUSED*/
2285: static int
1.129 yamt 2286: filt_solisten(struct knote *kn, long hint)
1.72 jdolecek 2287: {
2288: struct socket *so;
1.160 ad 2289: int rv;
1.72 jdolecek 2290:
1.155 ad 2291: so = ((file_t *)kn->kn_obj)->f_data;
1.72 jdolecek 2292:
2293: /*
2294: * Set kn_data to number of incoming connections, not
2295: * counting partial (incomplete) connections.
1.108 perry 2296: */
1.160 ad 2297: if (hint != NOTE_SUBMIT)
2298: solock(so);
1.72 jdolecek 2299: kn->kn_data = so->so_qlen;
1.160 ad 2300: rv = (kn->kn_data > 0);
2301: if (hint != NOTE_SUBMIT)
2302: sounlock(so);
2303: return rv;
1.72 jdolecek 2304: }
2305:
2306: static const struct filterops solisten_filtops =
2307: { 1, NULL, filt_sordetach, filt_solisten };
2308: static const struct filterops soread_filtops =
2309: { 1, NULL, filt_sordetach, filt_soread };
2310: static const struct filterops sowrite_filtops =
2311: { 1, NULL, filt_sowdetach, filt_sowrite };
2312:
2313: int
1.129 yamt 2314: soo_kqfilter(struct file *fp, struct knote *kn)
1.72 jdolecek 2315: {
2316: struct socket *so;
2317: struct sockbuf *sb;
2318:
1.155 ad 2319: so = ((file_t *)kn->kn_obj)->f_data;
1.160 ad 2320: solock(so);
1.72 jdolecek 2321: switch (kn->kn_filter) {
2322: case EVFILT_READ:
2323: if (so->so_options & SO_ACCEPTCONN)
2324: kn->kn_fop = &solisten_filtops;
2325: else
2326: kn->kn_fop = &soread_filtops;
2327: sb = &so->so_rcv;
2328: break;
2329: case EVFILT_WRITE:
2330: kn->kn_fop = &sowrite_filtops;
2331: sb = &so->so_snd;
2332: break;
2333: default:
1.160 ad 2334: sounlock(so);
1.149 pooka 2335: return (EINVAL);
1.72 jdolecek 2336: }
1.73 christos 2337: SLIST_INSERT_HEAD(&sb->sb_sel.sel_klist, kn, kn_selnext);
1.72 jdolecek 2338: sb->sb_flags |= SB_KNOTE;
1.160 ad 2339: sounlock(so);
1.72 jdolecek 2340: return (0);
2341: }
2342:
1.154 ad 2343: static int
2344: sodopoll(struct socket *so, int events)
2345: {
2346: int revents;
2347:
2348: revents = 0;
2349:
2350: if (events & (POLLIN | POLLRDNORM))
2351: if (soreadable(so))
2352: revents |= events & (POLLIN | POLLRDNORM);
2353:
2354: if (events & (POLLOUT | POLLWRNORM))
2355: if (sowritable(so))
2356: revents |= events & (POLLOUT | POLLWRNORM);
2357:
2358: if (events & (POLLPRI | POLLRDBAND))
2359: if (so->so_oobmark || (so->so_state & SS_RCVATMARK))
2360: revents |= events & (POLLPRI | POLLRDBAND);
2361:
2362: return revents;
2363: }
2364:
2365: int
2366: sopoll(struct socket *so, int events)
2367: {
2368: int revents = 0;
2369:
1.160 ad 2370: #ifndef DIAGNOSTIC
2371: /*
2372: * Do a quick, unlocked check in expectation that the socket
2373: * will be ready for I/O. Don't do this check if DIAGNOSTIC,
2374: * as the solocked() assertions will fail.
2375: */
1.154 ad 2376: if ((revents = sodopoll(so, events)) != 0)
2377: return revents;
1.160 ad 2378: #endif
1.154 ad 2379:
1.160 ad 2380: solock(so);
1.154 ad 2381: if ((revents = sodopoll(so, events)) == 0) {
2382: if (events & (POLLIN | POLLPRI | POLLRDNORM | POLLRDBAND)) {
2383: selrecord(curlwp, &so->so_rcv.sb_sel);
1.160 ad 2384: so->so_rcv.sb_flags |= SB_NOTIFY;
1.154 ad 2385: }
2386:
2387: if (events & (POLLOUT | POLLWRNORM)) {
2388: selrecord(curlwp, &so->so_snd.sb_sel);
1.160 ad 2389: so->so_snd.sb_flags |= SB_NOTIFY;
1.154 ad 2390: }
2391: }
1.160 ad 2392: sounlock(so);
1.154 ad 2393:
2394: return revents;
2395: }
2396:
2397:
1.94 yamt 2398: #include <sys/sysctl.h>
2399:
2400: static int sysctl_kern_somaxkva(SYSCTLFN_PROTO);
1.212 pooka 2401: static int sysctl_kern_sbmax(SYSCTLFN_PROTO);
1.94 yamt 2402:
2403: /*
2404: * sysctl helper routine for kern.somaxkva. ensures that the given
2405: * value is not too small.
2406: * (XXX should we maybe make sure it's not too large as well?)
2407: */
2408: static int
2409: sysctl_kern_somaxkva(SYSCTLFN_ARGS)
2410: {
2411: int error, new_somaxkva;
2412: struct sysctlnode node;
2413:
2414: new_somaxkva = somaxkva;
2415: node = *rnode;
2416: node.sysctl_data = &new_somaxkva;
2417: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2418: if (error || newp == NULL)
2419: return (error);
2420:
2421: if (new_somaxkva < (16 * 1024 * 1024)) /* sanity */
2422: return (EINVAL);
2423:
1.136 ad 2424: mutex_enter(&so_pendfree_lock);
1.94 yamt 2425: somaxkva = new_somaxkva;
1.136 ad 2426: cv_broadcast(&socurkva_cv);
2427: mutex_exit(&so_pendfree_lock);
1.94 yamt 2428:
2429: return (error);
2430: }
2431:
1.212 pooka 2432: /*
2433: * sysctl helper routine for kern.sbmax. Basically just ensures that
2434: * any new value is not too small.
2435: */
2436: static int
2437: sysctl_kern_sbmax(SYSCTLFN_ARGS)
2438: {
2439: int error, new_sbmax;
2440: struct sysctlnode node;
2441:
2442: new_sbmax = sb_max;
2443: node = *rnode;
2444: node.sysctl_data = &new_sbmax;
2445: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2446: if (error || newp == NULL)
2447: return (error);
2448:
2449: KERNEL_LOCK(1, NULL);
2450: error = sb_max_set(new_sbmax);
2451: KERNEL_UNLOCK_ONE(NULL);
2452:
2453: return (error);
2454: }
2455:
1.178 pooka 2456: static void
1.212 pooka 2457: sysctl_kern_socket_setup(void)
1.94 yamt 2458: {
2459:
1.178 pooka 2460: KASSERT(socket_sysctllog == NULL);
2461: sysctl_createv(&socket_sysctllog, 0, NULL, NULL,
1.97 atatat 2462: CTLFLAG_PERMANENT,
2463: CTLTYPE_NODE, "kern", NULL,
2464: NULL, 0, NULL, 0,
2465: CTL_KERN, CTL_EOL);
2466:
1.178 pooka 2467: sysctl_createv(&socket_sysctllog, 0, NULL, NULL,
1.97 atatat 2468: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.103 atatat 2469: CTLTYPE_INT, "somaxkva",
2470: SYSCTL_DESCR("Maximum amount of kernel memory to be "
2471: "used for socket buffers"),
1.94 yamt 2472: sysctl_kern_somaxkva, 0, NULL, 0,
2473: CTL_KERN, KERN_SOMAXKVA, CTL_EOL);
1.212 pooka 2474:
2475: sysctl_createv(&socket_sysctllog, 0, NULL, NULL,
2476: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2477: CTLTYPE_INT, "sbmax",
2478: SYSCTL_DESCR("Maximum socket buffer size"),
2479: sysctl_kern_sbmax, 0, NULL, 0,
2480: CTL_KERN, KERN_SBMAX, CTL_EOL);
1.94 yamt 2481: }
CVSweb <webmaster@jp.NetBSD.org>