[BACK]Return to uipc_socket.c CVS log [TXT][DIR] Up to [cvs.NetBSD.org] / src / sys / kern

Annotation of src/sys/kern/uipc_socket.c, Revision 1.249

1.249   ! christos    1: /*     $NetBSD: uipc_socket.c,v 1.248 2016/06/10 13:27:15 ozaki-r Exp $        */
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:
1.222     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>
1.249   ! christos   74: __KERNEL_RCSID(0, "$NetBSD: uipc_socket.c,v 1.248 2016/06/10 13:27:15 ozaki-r Exp $");
1.64      thorpej    75:
1.246     pooka      76: #ifdef _KERNEL_OPT
1.179     christos   77: #include "opt_compat_netbsd.h"
1.64      thorpej    78: #include "opt_sock_counters.h"
                     79: #include "opt_sosend_loan.h"
1.81      martin     80: #include "opt_mbuftrace.h"
1.84      ragge      81: #include "opt_somaxkva.h"
1.167     ad         82: #include "opt_multiprocessor.h"        /* XXX */
1.247     rjs        83: #include "opt_sctp.h"
1.246     pooka      84: #endif
1.1       cgd        85:
1.9       mycroft    86: #include <sys/param.h>
                     87: #include <sys/systm.h>
                     88: #include <sys/proc.h>
                     89: #include <sys/file.h>
1.142     dyoung     90: #include <sys/filedesc.h>
1.173     plunky     91: #include <sys/kmem.h>
1.9       mycroft    92: #include <sys/mbuf.h>
                     93: #include <sys/domain.h>
                     94: #include <sys/kernel.h>
                     95: #include <sys/protosw.h>
                     96: #include <sys/socket.h>
                     97: #include <sys/socketvar.h>
1.21      christos   98: #include <sys/signalvar.h>
1.9       mycroft    99: #include <sys/resourcevar.h>
1.174     pooka     100: #include <sys/uidinfo.h>
1.72      jdolecek  101: #include <sys/event.h>
1.89      christos  102: #include <sys/poll.h>
1.118     elad      103: #include <sys/kauth.h>
1.136     ad        104: #include <sys/mutex.h>
                    105: #include <sys/condvar.h>
1.205     bouyer    106: #include <sys/kthread.h>
1.37      thorpej   107:
1.179     christos  108: #ifdef COMPAT_50
                    109: #include <compat/sys/time.h>
1.184     christos  110: #include <compat/sys/socket.h>
1.179     christos  111: #endif
                    112:
1.202     uebayasi  113: #include <uvm/uvm_extern.h>
                    114: #include <uvm/uvm_loan.h>
                    115: #include <uvm/uvm_page.h>
1.64      thorpej   116:
1.77      thorpej   117: MALLOC_DEFINE(M_SONAME, "soname", "socket name");
1.37      thorpej   118:
1.142     dyoung    119: extern const struct fileops socketops;
                    120:
1.54      lukem     121: extern int     somaxconn;                      /* patchable (XXX sysctl) */
                    122: int            somaxconn = SOMAXCONN;
1.160     ad        123: kmutex_t       *softnet_lock;
1.49      jonathan  124:
1.64      thorpej   125: #ifdef SOSEND_COUNTERS
                    126: #include <sys/device.h>
                    127:
1.113     thorpej   128: static struct evcnt sosend_loan_big = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64      thorpej   129:     NULL, "sosend", "loan big");
1.113     thorpej   130: static struct evcnt sosend_copy_big = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64      thorpej   131:     NULL, "sosend", "copy big");
1.113     thorpej   132: static struct evcnt sosend_copy_small = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64      thorpej   133:     NULL, "sosend", "copy small");
1.113     thorpej   134: static struct evcnt sosend_kvalimit = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64      thorpej   135:     NULL, "sosend", "kva limit");
                    136:
                    137: #define        SOSEND_COUNTER_INCR(ev)         (ev)->ev_count++
                    138:
1.101     matt      139: EVCNT_ATTACH_STATIC(sosend_loan_big);
                    140: EVCNT_ATTACH_STATIC(sosend_copy_big);
                    141: EVCNT_ATTACH_STATIC(sosend_copy_small);
                    142: EVCNT_ATTACH_STATIC(sosend_kvalimit);
1.64      thorpej   143: #else
                    144:
                    145: #define        SOSEND_COUNTER_INCR(ev)         /* nothing */
                    146:
                    147: #endif /* SOSEND_COUNTERS */
                    148:
1.167     ad        149: #if defined(SOSEND_NO_LOAN) || defined(MULTIPROCESSOR)
1.121     yamt      150: int sock_loan_thresh = -1;
1.71      thorpej   151: #else
1.121     yamt      152: int sock_loan_thresh = 4096;
1.65      thorpej   153: #endif
1.64      thorpej   154:
1.136     ad        155: static kmutex_t so_pendfree_lock;
1.205     bouyer    156: static struct mbuf *so_pendfree = NULL;
1.64      thorpej   157:
1.84      ragge     158: #ifndef SOMAXKVA
                    159: #define        SOMAXKVA (16 * 1024 * 1024)
                    160: #endif
                    161: int somaxkva = SOMAXKVA;
1.113     thorpej   162: static int socurkva;
1.136     ad        163: static kcondvar_t socurkva_cv;
1.64      thorpej   164:
1.191     elad      165: static kauth_listener_t socket_listener;
                    166:
1.64      thorpej   167: #define        SOCK_LOAN_CHUNK         65536
                    168:
1.205     bouyer    169: static void sopendfree_thread(void *);
                    170: static kcondvar_t pendfree_thread_cv;
                    171: static lwp_t *sopendfree_lwp;
1.93      yamt      172:
1.212     pooka     173: static void sysctl_kern_socket_setup(void);
1.178     pooka     174: static struct sysctllog *socket_sysctllog;
                    175:
1.113     thorpej   176: static vsize_t
1.129     yamt      177: sokvareserve(struct socket *so, vsize_t len)
1.80      yamt      178: {
1.98      christos  179:        int error;
1.80      yamt      180:
1.136     ad        181:        mutex_enter(&so_pendfree_lock);
1.80      yamt      182:        while (socurkva + len > somaxkva) {
                    183:                SOSEND_COUNTER_INCR(&sosend_kvalimit);
1.136     ad        184:                error = cv_wait_sig(&socurkva_cv, &so_pendfree_lock);
1.98      christos  185:                if (error) {
                    186:                        len = 0;
                    187:                        break;
                    188:                }
1.80      yamt      189:        }
1.93      yamt      190:        socurkva += len;
1.136     ad        191:        mutex_exit(&so_pendfree_lock);
1.98      christos  192:        return len;
1.95      yamt      193: }
                    194:
1.113     thorpej   195: static void
1.95      yamt      196: sokvaunreserve(vsize_t len)
                    197: {
                    198:
1.136     ad        199:        mutex_enter(&so_pendfree_lock);
1.95      yamt      200:        socurkva -= len;
1.136     ad        201:        cv_broadcast(&socurkva_cv);
                    202:        mutex_exit(&so_pendfree_lock);
1.95      yamt      203: }
                    204:
                    205: /*
                    206:  * sokvaalloc: allocate kva for loan.
                    207:  */
                    208:
                    209: vaddr_t
1.209     matt      210: sokvaalloc(vaddr_t sva, vsize_t len, struct socket *so)
1.95      yamt      211: {
                    212:        vaddr_t lva;
                    213:
                    214:        /*
                    215:         * reserve kva.
                    216:         */
                    217:
1.98      christos  218:        if (sokvareserve(so, len) == 0)
                    219:                return 0;
1.93      yamt      220:
                    221:        /*
                    222:         * allocate kva.
                    223:         */
1.80      yamt      224:
1.209     matt      225:        lva = uvm_km_alloc(kernel_map, len, atop(sva) & uvmexp.colormask,
                    226:            UVM_KMF_COLORMATCH | UVM_KMF_VAONLY | UVM_KMF_WAITVA);
1.95      yamt      227:        if (lva == 0) {
                    228:                sokvaunreserve(len);
1.80      yamt      229:                return (0);
1.95      yamt      230:        }
1.80      yamt      231:
                    232:        return lva;
                    233: }
                    234:
1.93      yamt      235: /*
                    236:  * sokvafree: free kva for loan.
                    237:  */
                    238:
1.80      yamt      239: void
                    240: sokvafree(vaddr_t sva, vsize_t len)
                    241: {
1.93      yamt      242:
                    243:        /*
                    244:         * free kva.
                    245:         */
1.80      yamt      246:
1.109     yamt      247:        uvm_km_free(kernel_map, sva, len, UVM_KMF_VAONLY);
1.93      yamt      248:
                    249:        /*
                    250:         * unreserve kva.
                    251:         */
                    252:
1.95      yamt      253:        sokvaunreserve(len);
1.80      yamt      254: }
                    255:
1.64      thorpej   256: static void
1.134     christos  257: sodoloanfree(struct vm_page **pgs, void *buf, size_t size)
1.64      thorpej   258: {
1.156     yamt      259:        vaddr_t sva, eva;
1.64      thorpej   260:        vsize_t len;
1.156     yamt      261:        int npgs;
                    262:
                    263:        KASSERT(pgs != NULL);
1.64      thorpej   264:
                    265:        eva = round_page((vaddr_t) buf + size);
                    266:        sva = trunc_page((vaddr_t) buf);
                    267:        len = eva - sva;
                    268:        npgs = len >> PAGE_SHIFT;
                    269:
                    270:        pmap_kremove(sva, len);
                    271:        pmap_update(pmap_kernel());
                    272:        uvm_unloan(pgs, npgs, UVM_LOAN_TOPAGE);
1.80      yamt      273:        sokvafree(sva, len);
1.64      thorpej   274: }
                    275:
1.93      yamt      276: /*
1.205     bouyer    277:  * sopendfree_thread: free mbufs on "pendfree" list.
1.136     ad        278:  * unlock and relock so_pendfree_lock when freeing mbufs.
1.93      yamt      279:  */
                    280:
1.205     bouyer    281: static void
                    282: sopendfree_thread(void *v)
1.93      yamt      283: {
1.137     ad        284:        struct mbuf *m, *next;
1.205     bouyer    285:        size_t rv;
1.93      yamt      286:
1.205     bouyer    287:        mutex_enter(&so_pendfree_lock);
1.64      thorpej   288:
1.205     bouyer    289:        for (;;) {
                    290:                rv = 0;
                    291:                while (so_pendfree != NULL) {
                    292:                        m = so_pendfree;
                    293:                        so_pendfree = NULL;
                    294:                        mutex_exit(&so_pendfree_lock);
                    295:
                    296:                        for (; m != NULL; m = next) {
                    297:                                next = m->m_next;
                    298:                                KASSERT((~m->m_flags & (M_EXT|M_EXT_PAGES)) == 0);
                    299:                                KASSERT(m->m_ext.ext_refcnt == 0);
                    300:
                    301:                                rv += m->m_ext.ext_size;
                    302:                                sodoloanfree(m->m_ext.ext_pgs, m->m_ext.ext_buf,
                    303:                                    m->m_ext.ext_size);
                    304:                                pool_cache_put(mb_cache, m);
                    305:                        }
1.93      yamt      306:
1.205     bouyer    307:                        mutex_enter(&so_pendfree_lock);
1.93      yamt      308:                }
1.205     bouyer    309:                if (rv)
                    310:                        cv_broadcast(&socurkva_cv);
                    311:                cv_wait(&pendfree_thread_cv, &so_pendfree_lock);
1.64      thorpej   312:        }
1.205     bouyer    313:        panic("sopendfree_thread");
                    314:        /* NOTREACHED */
1.64      thorpej   315: }
                    316:
1.80      yamt      317: void
1.134     christos  318: soloanfree(struct mbuf *m, void *buf, size_t size, void *arg)
1.64      thorpej   319: {
                    320:
1.156     yamt      321:        KASSERT(m != NULL);
1.64      thorpej   322:
1.93      yamt      323:        /*
                    324:         * postpone freeing mbuf.
                    325:         *
                    326:         * we can't do it in interrupt context
                    327:         * because we need to put kva back to kernel_map.
                    328:         */
                    329:
1.136     ad        330:        mutex_enter(&so_pendfree_lock);
1.92      yamt      331:        m->m_next = so_pendfree;
                    332:        so_pendfree = m;
1.205     bouyer    333:        cv_signal(&pendfree_thread_cv);
1.136     ad        334:        mutex_exit(&so_pendfree_lock);
1.64      thorpej   335: }
                    336:
                    337: static long
                    338: sosend_loan(struct socket *so, struct uio *uio, struct mbuf *m, long space)
                    339: {
                    340:        struct iovec *iov = uio->uio_iov;
                    341:        vaddr_t sva, eva;
                    342:        vsize_t len;
1.156     yamt      343:        vaddr_t lva;
                    344:        int npgs, error;
                    345:        vaddr_t va;
                    346:        int i;
1.64      thorpej   347:
1.116     yamt      348:        if (VMSPACE_IS_KERNEL_P(uio->uio_vmspace))
1.64      thorpej   349:                return (0);
                    350:
                    351:        if (iov->iov_len < (size_t) space)
                    352:                space = iov->iov_len;
                    353:        if (space > SOCK_LOAN_CHUNK)
                    354:                space = SOCK_LOAN_CHUNK;
                    355:
                    356:        eva = round_page((vaddr_t) iov->iov_base + space);
                    357:        sva = trunc_page((vaddr_t) iov->iov_base);
                    358:        len = eva - sva;
                    359:        npgs = len >> PAGE_SHIFT;
                    360:
1.79      thorpej   361:        KASSERT(npgs <= M_EXT_MAXPAGES);
                    362:
1.209     matt      363:        lva = sokvaalloc(sva, len, so);
1.64      thorpej   364:        if (lva == 0)
1.80      yamt      365:                return 0;
1.64      thorpej   366:
1.116     yamt      367:        error = uvm_loan(&uio->uio_vmspace->vm_map, sva, len,
1.79      thorpej   368:            m->m_ext.ext_pgs, UVM_LOAN_TOPAGE);
1.64      thorpej   369:        if (error) {
1.80      yamt      370:                sokvafree(lva, len);
1.64      thorpej   371:                return (0);
                    372:        }
                    373:
                    374:        for (i = 0, va = lva; i < npgs; i++, va += PAGE_SIZE)
1.79      thorpej   375:                pmap_kenter_pa(va, VM_PAGE_TO_PHYS(m->m_ext.ext_pgs[i]),
1.194     cegger    376:                    VM_PROT_READ, 0);
1.64      thorpej   377:        pmap_update(pmap_kernel());
                    378:
                    379:        lva += (vaddr_t) iov->iov_base & PAGE_MASK;
                    380:
1.134     christos  381:        MEXTADD(m, (void *) lva, space, M_MBUF, soloanfree, so);
1.79      thorpej   382:        m->m_flags |= M_EXT_PAGES | M_EXT_ROMAP;
1.64      thorpej   383:
                    384:        uio->uio_resid -= space;
                    385:        /* uio_offset not updated, not set/used for write(2) */
1.134     christos  386:        uio->uio_iov->iov_base = (char *)uio->uio_iov->iov_base + space;
1.64      thorpej   387:        uio->uio_iov->iov_len -= space;
                    388:        if (uio->uio_iov->iov_len == 0) {
                    389:                uio->uio_iov++;
                    390:                uio->uio_iovcnt--;
                    391:        }
                    392:
                    393:        return (space);
                    394: }
                    395:
1.142     dyoung    396: struct mbuf *
1.147     dyoung    397: getsombuf(struct socket *so, int type)
1.142     dyoung    398: {
                    399:        struct mbuf *m;
                    400:
1.147     dyoung    401:        m = m_get(M_WAIT, type);
1.142     dyoung    402:        MCLAIM(m, so->so_mowner);
                    403:        return m;
                    404: }
                    405:
1.191     elad      406: static int
                    407: socket_listener_cb(kauth_cred_t cred, kauth_action_t action, void *cookie,
                    408:     void *arg0, void *arg1, void *arg2, void *arg3)
                    409: {
                    410:        int result;
                    411:        enum kauth_network_req req;
                    412:
                    413:        result = KAUTH_RESULT_DEFER;
                    414:        req = (enum kauth_network_req)arg0;
                    415:
1.193     elad      416:        if ((action != KAUTH_NETWORK_SOCKET) &&
                    417:            (action != KAUTH_NETWORK_BIND))
1.191     elad      418:                return result;
                    419:
                    420:        switch (req) {
1.193     elad      421:        case KAUTH_REQ_NETWORK_BIND_PORT:
                    422:                result = KAUTH_RESULT_ALLOW;
                    423:                break;
                    424:
1.191     elad      425:        case KAUTH_REQ_NETWORK_SOCKET_DROP: {
                    426:                /* Normal users can only drop their own connections. */
                    427:                struct socket *so = (struct socket *)arg1;
                    428:
1.220     christos  429:                if (so->so_cred && proc_uidmatch(cred, so->so_cred) == 0)
1.191     elad      430:                        result = KAUTH_RESULT_ALLOW;
                    431:
                    432:                break;
                    433:                }
                    434:
                    435:        case KAUTH_REQ_NETWORK_SOCKET_OPEN:
                    436:                /* We allow "raw" routing/bluetooth sockets to anyone. */
1.203     matt      437:                if ((u_long)arg1 == PF_ROUTE || (u_long)arg1 == PF_OROUTE
                    438:                    || (u_long)arg1 == PF_BLUETOOTH) {
1.191     elad      439:                        result = KAUTH_RESULT_ALLOW;
1.203     matt      440:                } else {
1.191     elad      441:                        /* Privileged, let secmodel handle this. */
                    442:                        if ((u_long)arg2 == SOCK_RAW)
                    443:                                break;
                    444:                }
                    445:
                    446:                result = KAUTH_RESULT_ALLOW;
                    447:
                    448:                break;
                    449:
1.192     elad      450:        case KAUTH_REQ_NETWORK_SOCKET_CANSEE:
                    451:                result = KAUTH_RESULT_ALLOW;
                    452:
                    453:                break;
                    454:
1.191     elad      455:        default:
                    456:                break;
                    457:        }
                    458:
                    459:        return result;
                    460: }
                    461:
1.119     yamt      462: void
                    463: soinit(void)
                    464: {
                    465:
1.212     pooka     466:        sysctl_kern_socket_setup();
1.178     pooka     467:
1.148     ad        468:        mutex_init(&so_pendfree_lock, MUTEX_DEFAULT, IPL_VM);
1.160     ad        469:        softnet_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NONE);
1.136     ad        470:        cv_init(&socurkva_cv, "sokva");
1.205     bouyer    471:        cv_init(&pendfree_thread_cv, "sopendfr");
1.166     ad        472:        soinit2();
1.136     ad        473:
1.119     yamt      474:        /* Set the initial adjusted socket buffer size. */
                    475:        if (sb_max_set(sb_max))
                    476:                panic("bad initial sb_max value: %lu", sb_max);
                    477:
1.191     elad      478:        socket_listener = kauth_listen_scope(KAUTH_SCOPE_NETWORK,
                    479:            socket_listener_cb, NULL);
1.119     yamt      480: }
                    481:
1.205     bouyer    482: void
                    483: soinit1(void)
                    484: {
                    485:        int error = kthread_create(PRI_NONE, KTHREAD_MPSAFE, NULL,
                    486:            sopendfree_thread, NULL, &sopendfree_lwp, "sopendfree");
                    487:        if (error)
                    488:                panic("soinit1 %d", error);
                    489: }
                    490:
1.1       cgd       491: /*
1.222     rmind     492:  * socreate: create a new socket of the specified type and the protocol.
                    493:  *
                    494:  * => Caller may specify another socket for lock sharing (must not be held).
                    495:  * => Returns the new socket without lock held.
1.224     rmind     496:  */
1.3       andrew    497: int
1.160     ad        498: socreate(int dom, struct socket **aso, int type, int proto, struct lwp *l,
                    499:         struct socket *lockso)
1.1       cgd       500: {
1.99      matt      501:        const struct protosw    *prp;
1.54      lukem     502:        struct socket   *so;
1.115     yamt      503:        uid_t           uid;
1.160     ad        504:        int             error;
                    505:        kmutex_t        *lock;
1.1       cgd       506:
1.132     elad      507:        error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET,
                    508:            KAUTH_REQ_NETWORK_SOCKET_OPEN, KAUTH_ARG(dom), KAUTH_ARG(type),
                    509:            KAUTH_ARG(proto));
1.140     dyoung    510:        if (error != 0)
                    511:                return error;
1.127     elad      512:
1.1       cgd       513:        if (proto)
                    514:                prp = pffindproto(dom, proto, type);
                    515:        else
                    516:                prp = pffindtype(dom, type);
1.140     dyoung    517:        if (prp == NULL) {
1.120     ginsbach  518:                /* no support for domain */
                    519:                if (pffinddomain(dom) == 0)
1.140     dyoung    520:                        return EAFNOSUPPORT;
1.120     ginsbach  521:                /* no support for socket type */
                    522:                if (proto == 0 && type != 0)
1.140     dyoung    523:                        return EPROTOTYPE;
                    524:                return EPROTONOSUPPORT;
1.120     ginsbach  525:        }
1.223     rmind     526:        if (prp->pr_usrreqs == NULL)
1.140     dyoung    527:                return EPROTONOSUPPORT;
1.1       cgd       528:        if (prp->pr_type != type)
1.140     dyoung    529:                return EPROTOTYPE;
1.160     ad        530:
                    531:        so = soget(true);
1.1       cgd       532:        so->so_type = type;
                    533:        so->so_proto = prp;
1.33      matt      534:        so->so_send = sosend;
                    535:        so->so_receive = soreceive;
1.78      matt      536: #ifdef MBUFTRACE
                    537:        so->so_rcv.sb_mowner = &prp->pr_domain->dom_mowner;
                    538:        so->so_snd.sb_mowner = &prp->pr_domain->dom_mowner;
                    539:        so->so_mowner = &prp->pr_domain->dom_mowner;
                    540: #endif
1.138     rmind     541:        uid = kauth_cred_geteuid(l->l_cred);
1.115     yamt      542:        so->so_uidinfo = uid_find(uid);
1.168     yamt      543:        so->so_cpid = l->l_proc->p_pid;
1.224     rmind     544:
                    545:        /*
                    546:         * Lock assigned and taken during PCB attach, unless we share
                    547:         * the lock with another socket, e.g. socketpair(2) case.
                    548:         */
                    549:        if (lockso) {
1.160     ad        550:                lock = lockso->so_lock;
                    551:                so->so_lock = lock;
                    552:                mutex_obj_hold(lock);
                    553:                mutex_enter(lock);
                    554:        }
1.224     rmind     555:
                    556:        /* Attach the PCB (returns with the socket lock held). */
                    557:        error = (*prp->pr_usrreqs->pr_attach)(so, proto);
1.160     ad        558:        KASSERT(solocked(so));
1.224     rmind     559:
                    560:        if (error) {
1.222     rmind     561:                KASSERT(so->so_pcb == NULL);
1.1       cgd       562:                so->so_state |= SS_NOFDREF;
                    563:                sofree(so);
1.140     dyoung    564:                return error;
1.1       cgd       565:        }
1.198     elad      566:        so->so_cred = kauth_cred_dup(l->l_cred);
1.160     ad        567:        sounlock(so);
1.224     rmind     568:
1.1       cgd       569:        *aso = so;
1.140     dyoung    570:        return 0;
1.1       cgd       571: }
                    572:
1.222     rmind     573: /*
                    574:  * fsocreate: create a socket and a file descriptor associated with it.
                    575:  *
                    576:  * => On success, write file descriptor to fdout and return zero.
                    577:  * => On failure, return non-zero; *fdout will be undefined.
1.142     dyoung    578:  */
                    579: int
1.222     rmind     580: fsocreate(int domain, struct socket **sop, int type, int proto, int *fdout)
1.142     dyoung    581: {
1.222     rmind     582:        lwp_t *l = curlwp;
                    583:        int error, fd, flags;
                    584:        struct socket *so;
                    585:        struct file *fp;
1.142     dyoung    586:
1.222     rmind     587:        if ((error = fd_allocfile(&fp, &fd)) != 0) {
1.204     christos  588:                return error;
1.222     rmind     589:        }
                    590:        flags = type & SOCK_FLAGS_MASK;
1.204     christos  591:        fd_set_exclose(l, fd, (flags & SOCK_CLOEXEC) != 0);
1.207     christos  592:        fp->f_flag = FREAD|FWRITE|((flags & SOCK_NONBLOCK) ? FNONBLOCK : 0)|
                    593:            ((flags & SOCK_NOSIGPIPE) ? FNOSIGPIPE : 0);
1.142     dyoung    594:        fp->f_type = DTYPE_SOCKET;
                    595:        fp->f_ops = &socketops;
1.222     rmind     596:
                    597:        type &= ~SOCK_FLAGS_MASK;
                    598:        error = socreate(domain, &so, type, proto, l, NULL);
                    599:        if (error) {
1.155     ad        600:                fd_abort(curproc, fp, fd);
1.222     rmind     601:                return error;
                    602:        }
                    603:        if (flags & SOCK_NONBLOCK) {
                    604:                so->so_state |= SS_NBIO;
                    605:        }
1.235     matt      606:        fp->f_socket = so;
1.222     rmind     607:        fd_affix(curproc, fp, fd);
                    608:
                    609:        if (sop != NULL) {
                    610:                *sop = so;
1.142     dyoung    611:        }
1.222     rmind     612:        *fdout = fd;
1.142     dyoung    613:        return error;
                    614: }
                    615:
1.3       andrew    616: int
1.190     dyoung    617: sofamily(const struct socket *so)
                    618: {
                    619:        const struct protosw *pr;
                    620:        const struct domain *dom;
                    621:
                    622:        if ((pr = so->so_proto) == NULL)
                    623:                return AF_UNSPEC;
                    624:        if ((dom = pr->pr_domain) == NULL)
                    625:                return AF_UNSPEC;
                    626:        return dom->dom_family;
                    627: }
                    628:
                    629: int
1.236     rtr       630: sobind(struct socket *so, struct sockaddr *nam, struct lwp *l)
1.1       cgd       631: {
1.160     ad        632:        int     error;
1.1       cgd       633:
1.160     ad        634:        solock(so);
1.237     rtr       635:        if (nam->sa_family != so->so_proto->pr_domain->dom_family) {
                    636:                sounlock(so);
1.238     rtr       637:                return EAFNOSUPPORT;
1.237     rtr       638:        }
1.231     rtr       639:        error = (*so->so_proto->pr_usrreqs->pr_bind)(so, nam, l);
1.160     ad        640:        sounlock(so);
1.140     dyoung    641:        return error;
1.1       cgd       642: }
                    643:
1.3       andrew    644: int
1.150     elad      645: solisten(struct socket *so, int backlog, struct lwp *l)
1.1       cgd       646: {
1.160     ad        647:        int     error;
1.247     rjs       648:        short   oldopt, oldqlimit;
1.1       cgd       649:
1.160     ad        650:        solock(so);
1.158     ad        651:        if ((so->so_state & (SS_ISCONNECTED | SS_ISCONNECTING |
1.163     ad        652:            SS_ISDISCONNECTING)) != 0) {
1.222     rmind     653:                sounlock(so);
                    654:                return EINVAL;
1.163     ad        655:        }
1.247     rjs       656:        oldopt = so->so_options;
                    657:        oldqlimit = so->so_qlimit;
                    658:        if (TAILQ_EMPTY(&so->so_q))
                    659:                so->so_options |= SO_ACCEPTCONN;
                    660:        if (backlog < 0)
                    661:                backlog = 0;
                    662:        so->so_qlimit = min(backlog, somaxconn);
                    663:
1.231     rtr       664:        error = (*so->so_proto->pr_usrreqs->pr_listen)(so, l);
1.140     dyoung    665:        if (error != 0) {
1.247     rjs       666:                so->so_options = oldopt;
                    667:                so->so_qlimit = oldqlimit;
1.160     ad        668:                sounlock(so);
1.140     dyoung    669:                return error;
1.1       cgd       670:        }
1.160     ad        671:        sounlock(so);
1.140     dyoung    672:        return 0;
1.1       cgd       673: }
                    674:
1.21      christos  675: void
1.54      lukem     676: sofree(struct socket *so)
1.1       cgd       677: {
1.161     ad        678:        u_int refs;
1.1       cgd       679:
1.160     ad        680:        KASSERT(solocked(so));
                    681:
                    682:        if (so->so_pcb || (so->so_state & SS_NOFDREF) == 0) {
                    683:                sounlock(so);
1.1       cgd       684:                return;
1.160     ad        685:        }
1.43      mycroft   686:        if (so->so_head) {
                    687:                /*
                    688:                 * We must not decommission a socket that's on the accept(2)
                    689:                 * queue.  If we do, then accept(2) may hang after select(2)
                    690:                 * indicated that the listening socket was ready.
                    691:                 */
1.160     ad        692:                if (!soqremque(so, 0)) {
                    693:                        sounlock(so);
1.43      mycroft   694:                        return;
1.160     ad        695:                }
1.43      mycroft   696:        }
1.98      christos  697:        if (so->so_rcv.sb_hiwat)
1.110     christos  698:                (void)chgsbsize(so->so_uidinfo, &so->so_rcv.sb_hiwat, 0,
1.98      christos  699:                    RLIM_INFINITY);
                    700:        if (so->so_snd.sb_hiwat)
1.110     christos  701:                (void)chgsbsize(so->so_uidinfo, &so->so_snd.sb_hiwat, 0,
1.98      christos  702:                    RLIM_INFINITY);
                    703:        sbrelease(&so->so_snd, so);
1.160     ad        704:        KASSERT(!cv_has_waiters(&so->so_cv));
                    705:        KASSERT(!cv_has_waiters(&so->so_rcv.sb_cv));
                    706:        KASSERT(!cv_has_waiters(&so->so_snd.sb_cv));
1.1       cgd       707:        sorflush(so);
1.161     ad        708:        refs = so->so_aborting; /* XXX */
1.177     ad        709:        /* Remove acccept filter if one is present. */
1.170     tls       710:        if (so->so_accf != NULL)
1.177     ad        711:                (void)accept_filt_clear(so);
1.160     ad        712:        sounlock(so);
1.161     ad        713:        if (refs == 0)          /* XXX */
                    714:                soput(so);
1.1       cgd       715: }
                    716:
                    717: /*
1.222     rmind     718:  * soclose: close a socket on last file table reference removal.
                    719:  * Initiate disconnect if connected.  Free socket when disconnect complete.
1.1       cgd       720:  */
1.3       andrew    721: int
1.54      lukem     722: soclose(struct socket *so)
1.1       cgd       723: {
1.222     rmind     724:        struct socket *so2;
                    725:        int error = 0;
1.1       cgd       726:
1.160     ad        727:        solock(so);
1.1       cgd       728:        if (so->so_options & SO_ACCEPTCONN) {
1.172     ad        729:                for (;;) {
                    730:                        if ((so2 = TAILQ_FIRST(&so->so_q0)) != 0) {
1.160     ad        731:                                KASSERT(solocked2(so, so2));
                    732:                                (void) soqremque(so2, 0);
                    733:                                /* soabort drops the lock. */
                    734:                                (void) soabort(so2);
                    735:                                solock(so);
1.172     ad        736:                                continue;
1.160     ad        737:                        }
1.172     ad        738:                        if ((so2 = TAILQ_FIRST(&so->so_q)) != 0) {
1.160     ad        739:                                KASSERT(solocked2(so, so2));
                    740:                                (void) soqremque(so2, 1);
                    741:                                /* soabort drops the lock. */
                    742:                                (void) soabort(so2);
                    743:                                solock(so);
1.172     ad        744:                                continue;
1.160     ad        745:                        }
1.172     ad        746:                        break;
                    747:                }
1.1       cgd       748:        }
1.222     rmind     749:        if (so->so_pcb == NULL)
1.1       cgd       750:                goto discard;
                    751:        if (so->so_state & SS_ISCONNECTED) {
                    752:                if ((so->so_state & SS_ISDISCONNECTING) == 0) {
                    753:                        error = sodisconnect(so);
                    754:                        if (error)
                    755:                                goto drop;
                    756:                }
                    757:                if (so->so_options & SO_LINGER) {
1.206     christos  758:                        if ((so->so_state & (SS_ISDISCONNECTING|SS_NBIO)) ==
                    759:                            (SS_ISDISCONNECTING|SS_NBIO))
1.1       cgd       760:                                goto drop;
1.21      christos  761:                        while (so->so_state & SS_ISCONNECTED) {
1.185     yamt      762:                                error = sowait(so, true, so->so_linger * hz);
1.21      christos  763:                                if (error)
1.1       cgd       764:                                        break;
1.21      christos  765:                        }
1.1       cgd       766:                }
                    767:        }
1.54      lukem     768:  drop:
1.1       cgd       769:        if (so->so_pcb) {
1.224     rmind     770:                KASSERT(solocked(so));
                    771:                (*so->so_proto->pr_usrreqs->pr_detach)(so);
1.1       cgd       772:        }
1.54      lukem     773:  discard:
1.222     rmind     774:        KASSERT((so->so_state & SS_NOFDREF) == 0);
1.198     elad      775:        kauth_cred_free(so->so_cred);
1.1       cgd       776:        so->so_state |= SS_NOFDREF;
                    777:        sofree(so);
1.222     rmind     778:        return error;
1.1       cgd       779: }
                    780:
                    781: /*
1.160     ad        782:  * Must be called with the socket locked..  Will return with it unlocked.
1.1       cgd       783:  */
1.3       andrew    784: int
1.54      lukem     785: soabort(struct socket *so)
1.1       cgd       786: {
1.161     ad        787:        u_int refs;
1.139     yamt      788:        int error;
1.160     ad        789:
                    790:        KASSERT(solocked(so));
                    791:        KASSERT(so->so_head == NULL);
1.1       cgd       792:
1.161     ad        793:        so->so_aborting++;              /* XXX */
1.230     mrg       794:        error = (*so->so_proto->pr_usrreqs->pr_abort)(so);
1.161     ad        795:        refs = --so->so_aborting;       /* XXX */
1.164     drochner  796:        if (error || (refs == 0)) {
1.139     yamt      797:                sofree(so);
1.160     ad        798:        } else {
                    799:                sounlock(so);
1.139     yamt      800:        }
                    801:        return error;
1.1       cgd       802: }
                    803:
1.3       andrew    804: int
1.239     rtr       805: soaccept(struct socket *so, struct sockaddr *nam)
1.1       cgd       806: {
1.222     rmind     807:        int error;
1.160     ad        808:
                    809:        KASSERT(solocked(so));
1.222     rmind     810:        KASSERT((so->so_state & SS_NOFDREF) != 0);
1.1       cgd       811:
                    812:        so->so_state &= ~SS_NOFDREF;
1.55      thorpej   813:        if ((so->so_state & SS_ISDISCONNECTED) == 0 ||
                    814:            (so->so_proto->pr_flags & PR_ABRTACPTDIS) == 0)
1.225     rtr       815:                error = (*so->so_proto->pr_usrreqs->pr_accept)(so, nam);
1.41      mycroft   816:        else
1.53      itojun    817:                error = ECONNABORTED;
1.52      itojun    818:
1.222     rmind     819:        return error;
1.1       cgd       820: }
                    821:
1.3       andrew    822: int
1.240     rtr       823: soconnect(struct socket *so, struct sockaddr *nam, struct lwp *l)
1.1       cgd       824: {
1.222     rmind     825:        int error;
1.160     ad        826:
                    827:        KASSERT(solocked(so));
1.1       cgd       828:
                    829:        if (so->so_options & SO_ACCEPTCONN)
1.222     rmind     830:                return EOPNOTSUPP;
1.1       cgd       831:        /*
                    832:         * If protocol is connection-based, can only connect once.
                    833:         * Otherwise, if connected, try to disconnect first.
                    834:         * This allows user to disconnect by connecting to, e.g.,
                    835:         * a null address.
                    836:         */
                    837:        if (so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING) &&
                    838:            ((so->so_proto->pr_flags & PR_CONNREQUIRED) ||
1.241     rtr       839:            (error = sodisconnect(so)))) {
1.1       cgd       840:                error = EISCONN;
1.241     rtr       841:        } else {
1.242     rtr       842:                if (nam->sa_family != so->so_proto->pr_domain->dom_family) {
1.241     rtr       843:                        return EAFNOSUPPORT;
                    844:                }
1.231     rtr       845:                error = (*so->so_proto->pr_usrreqs->pr_connect)(so, nam, l);
1.241     rtr       846:        }
1.222     rmind     847:
                    848:        return error;
1.1       cgd       849: }
                    850:
1.3       andrew    851: int
1.54      lukem     852: soconnect2(struct socket *so1, struct socket *so2)
1.1       cgd       853: {
1.160     ad        854:        KASSERT(solocked2(so1, so2));
1.1       cgd       855:
1.234     rtr       856:        return (*so1->so_proto->pr_usrreqs->pr_connect2)(so1, so2);
1.1       cgd       857: }
                    858:
1.3       andrew    859: int
1.54      lukem     860: sodisconnect(struct socket *so)
1.1       cgd       861: {
1.160     ad        862:        int     error;
                    863:
                    864:        KASSERT(solocked(so));
1.1       cgd       865:
                    866:        if ((so->so_state & SS_ISCONNECTED) == 0) {
                    867:                error = ENOTCONN;
1.160     ad        868:        } else if (so->so_state & SS_ISDISCONNECTING) {
1.1       cgd       869:                error = EALREADY;
1.160     ad        870:        } else {
1.229     rtr       871:                error = (*so->so_proto->pr_usrreqs->pr_disconnect)(so);
1.1       cgd       872:        }
                    873:        return (error);
                    874: }
                    875:
1.15      mycroft   876: #define        SBLOCKWAIT(f)   (((f) & MSG_DONTWAIT) ? M_NOWAIT : M_WAITOK)
1.1       cgd       877: /*
                    878:  * Send on a socket.
                    879:  * If send must go all at once and message is larger than
                    880:  * send buffering, then hard error.
                    881:  * Lock against other senders.
                    882:  * If must go all at once and not enough room now, then
                    883:  * inform user that this would block and do nothing.
                    884:  * Otherwise, if nonblocking, send as much as possible.
                    885:  * The data to be sent is described by "uio" if nonzero,
                    886:  * otherwise by the mbuf chain "top" (which must be null
                    887:  * if uio is not).  Data provided in mbuf chain must be small
                    888:  * enough to send all at once.
                    889:  *
                    890:  * Returns nonzero on error, timeout or signal; callers
                    891:  * must check for short counts if EINTR/ERESTART are returned.
                    892:  * Data and control buffers are freed on return.
                    893:  */
1.3       andrew    894: int
1.245     rtr       895: sosend(struct socket *so, struct sockaddr *addr, struct uio *uio,
                    896:        struct mbuf *top, struct mbuf *control, int flags, struct lwp *l)
1.1       cgd       897: {
1.54      lukem     898:        struct mbuf     **mp, *m;
1.58      jdolecek  899:        long            space, len, resid, clen, mlen;
                    900:        int             error, s, dontroute, atomic;
1.196     dsl       901:        short           wakeup_state = 0;
1.54      lukem     902:
1.160     ad        903:        clen = 0;
1.64      thorpej   904:
1.160     ad        905:        /*
                    906:         * solock() provides atomicity of access.  splsoftnet() prevents
                    907:         * protocol processing soft interrupts from interrupting us and
                    908:         * blocking (expensive).
                    909:         */
                    910:        s = splsoftnet();
                    911:        solock(so);
1.54      lukem     912:        atomic = sosendallatonce(so) || top;
1.1       cgd       913:        if (uio)
                    914:                resid = uio->uio_resid;
                    915:        else
                    916:                resid = top->m_pkthdr.len;
1.7       cgd       917:        /*
                    918:         * In theory resid should be unsigned.
                    919:         * However, space must be signed, as it might be less than 0
                    920:         * if we over-committed, and we must use a signed comparison
                    921:         * of space and resid.  On the other hand, a negative resid
                    922:         * causes us to loop sending 0-length segments to the protocol.
                    923:         */
1.29      mycroft   924:        if (resid < 0) {
                    925:                error = EINVAL;
                    926:                goto out;
                    927:        }
1.1       cgd       928:        dontroute =
                    929:            (flags & MSG_DONTROUTE) && (so->so_options & SO_DONTROUTE) == 0 &&
                    930:            (so->so_proto->pr_flags & PR_ATOMIC);
1.165     christos  931:        l->l_ru.ru_msgsnd++;
1.1       cgd       932:        if (control)
                    933:                clen = control->m_len;
1.54      lukem     934:  restart:
1.21      christos  935:        if ((error = sblock(&so->so_snd, SBLOCKWAIT(flags))) != 0)
1.1       cgd       936:                goto out;
                    937:        do {
1.160     ad        938:                if (so->so_state & SS_CANTSENDMORE) {
                    939:                        error = EPIPE;
                    940:                        goto release;
                    941:                }
1.48      thorpej   942:                if (so->so_error) {
                    943:                        error = so->so_error;
                    944:                        so->so_error = 0;
                    945:                        goto release;
                    946:                }
1.1       cgd       947:                if ((so->so_state & SS_ISCONNECTED) == 0) {
                    948:                        if (so->so_proto->pr_flags & PR_CONNREQUIRED) {
1.217     rmind     949:                                if (resid || clen == 0) {
1.160     ad        950:                                        error = ENOTCONN;
                    951:                                        goto release;
                    952:                                }
1.244     rtr       953:                        } else if (addr == NULL) {
1.160     ad        954:                                error = EDESTADDRREQ;
                    955:                                goto release;
                    956:                        }
1.1       cgd       957:                }
                    958:                space = sbspace(&so->so_snd);
                    959:                if (flags & MSG_OOB)
                    960:                        space += 1024;
1.21      christos  961:                if ((atomic && resid > so->so_snd.sb_hiwat) ||
1.160     ad        962:                    clen > so->so_snd.sb_hiwat) {
                    963:                        error = EMSGSIZE;
                    964:                        goto release;
                    965:                }
1.96      mycroft   966:                if (space < resid + clen &&
1.1       cgd       967:                    (atomic || space < so->so_snd.sb_lowat || space < clen)) {
1.206     christos  968:                        if ((so->so_state & SS_NBIO) || (flags & MSG_NBIO)) {
1.160     ad        969:                                error = EWOULDBLOCK;
                    970:                                goto release;
                    971:                        }
1.1       cgd       972:                        sbunlock(&so->so_snd);
1.196     dsl       973:                        if (wakeup_state & SS_RESTARTSYS) {
                    974:                                error = ERESTART;
                    975:                                goto out;
                    976:                        }
1.1       cgd       977:                        error = sbwait(&so->so_snd);
                    978:                        if (error)
                    979:                                goto out;
1.196     dsl       980:                        wakeup_state = so->so_state;
1.1       cgd       981:                        goto restart;
                    982:                }
1.196     dsl       983:                wakeup_state = 0;
1.1       cgd       984:                mp = &top;
                    985:                space -= clen;
                    986:                do {
1.45      tv        987:                        if (uio == NULL) {
                    988:                                /*
                    989:                                 * Data is prepackaged in "top".
                    990:                                 */
                    991:                                resid = 0;
                    992:                                if (flags & MSG_EOR)
                    993:                                        top->m_flags |= M_EOR;
                    994:                        } else do {
1.160     ad        995:                                sounlock(so);
                    996:                                splx(s);
1.144     dyoung    997:                                if (top == NULL) {
1.78      matt      998:                                        m = m_gethdr(M_WAIT, MT_DATA);
1.45      tv        999:                                        mlen = MHLEN;
                   1000:                                        m->m_pkthdr.len = 0;
1.248     ozaki-r  1001:                                        m_reset_rcvif(m);
1.45      tv       1002:                                } else {
1.78      matt     1003:                                        m = m_get(M_WAIT, MT_DATA);
1.45      tv       1004:                                        mlen = MLEN;
                   1005:                                }
1.78      matt     1006:                                MCLAIM(m, so->so_snd.sb_mowner);
1.121     yamt     1007:                                if (sock_loan_thresh >= 0 &&
                   1008:                                    uio->uio_iov->iov_len >= sock_loan_thresh &&
                   1009:                                    space >= sock_loan_thresh &&
1.64      thorpej  1010:                                    (len = sosend_loan(so, uio, m,
                   1011:                                                       space)) != 0) {
                   1012:                                        SOSEND_COUNTER_INCR(&sosend_loan_big);
                   1013:                                        space -= len;
                   1014:                                        goto have_data;
                   1015:                                }
1.45      tv       1016:                                if (resid >= MINCLSIZE && space >= MCLBYTES) {
1.64      thorpej  1017:                                        SOSEND_COUNTER_INCR(&sosend_copy_big);
1.201     oki      1018:                                        m_clget(m, M_DONTWAIT);
1.45      tv       1019:                                        if ((m->m_flags & M_EXT) == 0)
                   1020:                                                goto nopages;
                   1021:                                        mlen = MCLBYTES;
                   1022:                                        if (atomic && top == 0) {
1.58      jdolecek 1023:                                                len = lmin(MCLBYTES - max_hdr,
1.54      lukem    1024:                                                    resid);
1.45      tv       1025:                                                m->m_data += max_hdr;
                   1026:                                        } else
1.58      jdolecek 1027:                                                len = lmin(MCLBYTES, resid);
1.45      tv       1028:                                        space -= len;
                   1029:                                } else {
1.64      thorpej  1030:  nopages:
                   1031:                                        SOSEND_COUNTER_INCR(&sosend_copy_small);
1.58      jdolecek 1032:                                        len = lmin(lmin(mlen, resid), space);
1.45      tv       1033:                                        space -= len;
                   1034:                                        /*
                   1035:                                         * For datagram protocols, leave room
                   1036:                                         * for protocol headers in first mbuf.
                   1037:                                         */
                   1038:                                        if (atomic && top == 0 && len < mlen)
                   1039:                                                MH_ALIGN(m, len);
                   1040:                                }
1.144     dyoung   1041:                                error = uiomove(mtod(m, void *), (int)len, uio);
1.64      thorpej  1042:  have_data:
1.45      tv       1043:                                resid = uio->uio_resid;
                   1044:                                m->m_len = len;
                   1045:                                *mp = m;
                   1046:                                top->m_pkthdr.len += len;
1.160     ad       1047:                                s = splsoftnet();
                   1048:                                solock(so);
1.144     dyoung   1049:                                if (error != 0)
1.45      tv       1050:                                        goto release;
                   1051:                                mp = &m->m_next;
                   1052:                                if (resid <= 0) {
                   1053:                                        if (flags & MSG_EOR)
                   1054:                                                top->m_flags |= M_EOR;
                   1055:                                        break;
                   1056:                                }
                   1057:                        } while (space > 0 && atomic);
1.108     perry    1058:
1.160     ad       1059:                        if (so->so_state & SS_CANTSENDMORE) {
                   1060:                                error = EPIPE;
                   1061:                                goto release;
                   1062:                        }
1.45      tv       1063:                        if (dontroute)
                   1064:                                so->so_options |= SO_DONTROUTE;
                   1065:                        if (resid > 0)
                   1066:                                so->so_state |= SS_MORETOCOME;
1.240     rtr      1067:                        if (flags & MSG_OOB) {
1.226     rtr      1068:                                error = (*so->so_proto->pr_usrreqs->pr_sendoob)(so,
                   1069:                                    top, control);
1.240     rtr      1070:                        } else {
1.232     rtr      1071:                                error = (*so->so_proto->pr_usrreqs->pr_send)(so,
1.245     rtr      1072:                                    top, addr, control, l);
1.240     rtr      1073:                        }
1.45      tv       1074:                        if (dontroute)
                   1075:                                so->so_options &= ~SO_DONTROUTE;
                   1076:                        if (resid > 0)
                   1077:                                so->so_state &= ~SS_MORETOCOME;
                   1078:                        clen = 0;
1.144     dyoung   1079:                        control = NULL;
                   1080:                        top = NULL;
1.45      tv       1081:                        mp = &top;
1.144     dyoung   1082:                        if (error != 0)
1.1       cgd      1083:                                goto release;
                   1084:                } while (resid && space > 0);
                   1085:        } while (resid);
                   1086:
1.54      lukem    1087:  release:
1.1       cgd      1088:        sbunlock(&so->so_snd);
1.54      lukem    1089:  out:
1.160     ad       1090:        sounlock(so);
                   1091:        splx(s);
1.1       cgd      1092:        if (top)
                   1093:                m_freem(top);
                   1094:        if (control)
                   1095:                m_freem(control);
                   1096:        return (error);
                   1097: }
                   1098:
                   1099: /*
1.159     ad       1100:  * Following replacement or removal of the first mbuf on the first
                   1101:  * mbuf chain of a socket buffer, push necessary state changes back
                   1102:  * into the socket buffer so that other consumers see the values
                   1103:  * consistently.  'nextrecord' is the callers locally stored value of
                   1104:  * the original value of sb->sb_mb->m_nextpkt which must be restored
                   1105:  * when the lead mbuf changes.  NOTE: 'nextrecord' may be NULL.
                   1106:  */
                   1107: static void
                   1108: sbsync(struct sockbuf *sb, struct mbuf *nextrecord)
                   1109: {
                   1110:
1.160     ad       1111:        KASSERT(solocked(sb->sb_so));
                   1112:
1.159     ad       1113:        /*
                   1114:         * First, update for the new value of nextrecord.  If necessary,
                   1115:         * make it the first record.
                   1116:         */
                   1117:        if (sb->sb_mb != NULL)
                   1118:                sb->sb_mb->m_nextpkt = nextrecord;
                   1119:        else
                   1120:                sb->sb_mb = nextrecord;
                   1121:
                   1122:         /*
                   1123:          * Now update any dependent socket buffer fields to reflect
                   1124:          * the new state.  This is an inline of SB_EMPTY_FIXUP, with
                   1125:          * the addition of a second clause that takes care of the
                   1126:          * case where sb_mb has been updated, but remains the last
                   1127:          * record.
                   1128:          */
                   1129:         if (sb->sb_mb == NULL) {
                   1130:                 sb->sb_mbtail = NULL;
                   1131:                 sb->sb_lastrecord = NULL;
                   1132:         } else if (sb->sb_mb->m_nextpkt == NULL)
                   1133:                 sb->sb_lastrecord = sb->sb_mb;
                   1134: }
                   1135:
                   1136: /*
1.1       cgd      1137:  * Implement receive operations on a socket.
                   1138:  * We depend on the way that records are added to the sockbuf
                   1139:  * by sbappend*.  In particular, each record (mbufs linked through m_next)
                   1140:  * must begin with an address if the protocol so specifies,
                   1141:  * followed by an optional mbuf or mbufs containing ancillary data,
                   1142:  * and then zero or more mbufs of data.
                   1143:  * In order to avoid blocking network interrupts for the entire time here,
                   1144:  * we splx() while doing the actual copy to user space.
                   1145:  * Although the sockbuf is locked, new data may still be appended,
                   1146:  * and thus we must maintain consistency of the sockbuf during that time.
                   1147:  *
                   1148:  * The caller may receive the data as a single mbuf chain by supplying
                   1149:  * an mbuf **mp0 for use in returning the chain.  The uio is then used
                   1150:  * only for the count in uio_resid.
                   1151:  */
1.3       andrew   1152: int
1.54      lukem    1153: soreceive(struct socket *so, struct mbuf **paddr, struct uio *uio,
                   1154:        struct mbuf **mp0, struct mbuf **controlp, int *flagsp)
1.1       cgd      1155: {
1.116     yamt     1156:        struct lwp *l = curlwp;
1.160     ad       1157:        struct mbuf     *m, **mp, *mt;
1.211     chs      1158:        size_t len, offset, moff, orig_resid;
                   1159:        int atomic, flags, error, s, type;
1.99      matt     1160:        const struct protosw    *pr;
1.54      lukem    1161:        struct mbuf     *nextrecord;
1.67      he       1162:        int             mbuf_removed = 0;
1.146     dyoung   1163:        const struct domain *dom;
1.196     dsl      1164:        short           wakeup_state = 0;
1.64      thorpej  1165:
1.54      lukem    1166:        pr = so->so_proto;
1.146     dyoung   1167:        atomic = pr->pr_flags & PR_ATOMIC;
                   1168:        dom = pr->pr_domain;
1.1       cgd      1169:        mp = mp0;
1.54      lukem    1170:        type = 0;
                   1171:        orig_resid = uio->uio_resid;
1.102     jonathan 1172:
1.144     dyoung   1173:        if (paddr != NULL)
                   1174:                *paddr = NULL;
                   1175:        if (controlp != NULL)
                   1176:                *controlp = NULL;
                   1177:        if (flagsp != NULL)
1.1       cgd      1178:                flags = *flagsp &~ MSG_EOR;
                   1179:        else
                   1180:                flags = 0;
1.66      enami    1181:
1.1       cgd      1182:        if (flags & MSG_OOB) {
                   1183:                m = m_get(M_WAIT, MT_DATA);
1.160     ad       1184:                solock(so);
1.226     rtr      1185:                error = (*pr->pr_usrreqs->pr_recvoob)(so, m, flags & MSG_PEEK);
1.160     ad       1186:                sounlock(so);
1.1       cgd      1187:                if (error)
                   1188:                        goto bad;
                   1189:                do {
1.134     christos 1190:                        error = uiomove(mtod(m, void *),
1.211     chs      1191:                            MIN(uio->uio_resid, m->m_len), uio);
1.1       cgd      1192:                        m = m_free(m);
1.144     dyoung   1193:                } while (uio->uio_resid > 0 && error == 0 && m);
1.54      lukem    1194:  bad:
1.144     dyoung   1195:                if (m != NULL)
1.1       cgd      1196:                        m_freem(m);
1.144     dyoung   1197:                return error;
1.1       cgd      1198:        }
1.144     dyoung   1199:        if (mp != NULL)
1.140     dyoung   1200:                *mp = NULL;
1.160     ad       1201:
                   1202:        /*
                   1203:         * solock() provides atomicity of access.  splsoftnet() prevents
                   1204:         * protocol processing soft interrupts from interrupting us and
                   1205:         * blocking (expensive).
                   1206:         */
                   1207:        s = splsoftnet();
                   1208:        solock(so);
1.54      lukem    1209:  restart:
1.160     ad       1210:        if ((error = sblock(&so->so_rcv, SBLOCKWAIT(flags))) != 0) {
                   1211:                sounlock(so);
                   1212:                splx(s);
1.144     dyoung   1213:                return error;
1.160     ad       1214:        }
1.1       cgd      1215:
                   1216:        m = so->so_rcv.sb_mb;
                   1217:        /*
                   1218:         * If we have less data than requested, block awaiting more
                   1219:         * (subject to any timeout) if:
1.15      mycroft  1220:         *   1. the current count is less than the low water mark,
1.1       cgd      1221:         *   2. MSG_WAITALL is set, and it is possible to do the entire
1.15      mycroft  1222:         *      receive operation at once if we block (resid <= hiwat), or
                   1223:         *   3. MSG_DONTWAIT is not set.
1.1       cgd      1224:         * If MSG_WAITALL is set but resid is larger than the receive buffer,
                   1225:         * we have to do the receive in sections, and thus risk returning
                   1226:         * a short count if a timeout or signal occurs after we start.
                   1227:         */
1.144     dyoung   1228:        if (m == NULL ||
                   1229:            ((flags & MSG_DONTWAIT) == 0 &&
                   1230:             so->so_rcv.sb_cc < uio->uio_resid &&
                   1231:             (so->so_rcv.sb_cc < so->so_rcv.sb_lowat ||
                   1232:              ((flags & MSG_WAITALL) &&
                   1233:               uio->uio_resid <= so->so_rcv.sb_hiwat)) &&
1.146     dyoung   1234:             m->m_nextpkt == NULL && !atomic)) {
1.1       cgd      1235: #ifdef DIAGNOSTIC
1.144     dyoung   1236:                if (m == NULL && so->so_rcv.sb_cc)
1.1       cgd      1237:                        panic("receive 1");
                   1238: #endif
                   1239:                if (so->so_error) {
1.144     dyoung   1240:                        if (m != NULL)
1.15      mycroft  1241:                                goto dontblock;
1.1       cgd      1242:                        error = so->so_error;
                   1243:                        if ((flags & MSG_PEEK) == 0)
                   1244:                                so->so_error = 0;
                   1245:                        goto release;
                   1246:                }
                   1247:                if (so->so_state & SS_CANTRCVMORE) {
1.144     dyoung   1248:                        if (m != NULL)
1.15      mycroft  1249:                                goto dontblock;
1.1       cgd      1250:                        else
                   1251:                                goto release;
                   1252:                }
1.144     dyoung   1253:                for (; m != NULL; m = m->m_next)
1.1       cgd      1254:                        if (m->m_type == MT_OOBDATA  || (m->m_flags & M_EOR)) {
                   1255:                                m = so->so_rcv.sb_mb;
                   1256:                                goto dontblock;
                   1257:                        }
                   1258:                if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) == 0 &&
                   1259:                    (so->so_proto->pr_flags & PR_CONNREQUIRED)) {
                   1260:                        error = ENOTCONN;
                   1261:                        goto release;
                   1262:                }
                   1263:                if (uio->uio_resid == 0)
                   1264:                        goto release;
1.206     christos 1265:                if ((so->so_state & SS_NBIO) ||
                   1266:                    (flags & (MSG_DONTWAIT|MSG_NBIO))) {
1.1       cgd      1267:                        error = EWOULDBLOCK;
                   1268:                        goto release;
                   1269:                }
1.69      thorpej  1270:                SBLASTRECORDCHK(&so->so_rcv, "soreceive sbwait 1");
                   1271:                SBLASTMBUFCHK(&so->so_rcv, "soreceive sbwait 1");
1.1       cgd      1272:                sbunlock(&so->so_rcv);
1.196     dsl      1273:                if (wakeup_state & SS_RESTARTSYS)
                   1274:                        error = ERESTART;
                   1275:                else
                   1276:                        error = sbwait(&so->so_rcv);
1.160     ad       1277:                if (error != 0) {
                   1278:                        sounlock(so);
                   1279:                        splx(s);
1.144     dyoung   1280:                        return error;
1.160     ad       1281:                }
1.196     dsl      1282:                wakeup_state = so->so_state;
1.1       cgd      1283:                goto restart;
                   1284:        }
1.54      lukem    1285:  dontblock:
1.69      thorpej  1286:        /*
                   1287:         * On entry here, m points to the first record of the socket buffer.
1.159     ad       1288:         * From this point onward, we maintain 'nextrecord' as a cache of the
                   1289:         * pointer to the next record in the socket buffer.  We must keep the
                   1290:         * various socket buffer pointers and local stack versions of the
                   1291:         * pointers in sync, pushing out modifications before dropping the
1.160     ad       1292:         * socket lock, and re-reading them when picking it up.
1.159     ad       1293:         *
                   1294:         * Otherwise, we will race with the network stack appending new data
                   1295:         * or records onto the socket buffer by using inconsistent/stale
                   1296:         * versions of the field, possibly resulting in socket buffer
                   1297:         * corruption.
                   1298:         *
                   1299:         * By holding the high-level sblock(), we prevent simultaneous
                   1300:         * readers from pulling off the front of the socket buffer.
1.69      thorpej  1301:         */
1.144     dyoung   1302:        if (l != NULL)
1.157     ad       1303:                l->l_ru.ru_msgrcv++;
1.69      thorpej  1304:        KASSERT(m == so->so_rcv.sb_mb);
                   1305:        SBLASTRECORDCHK(&so->so_rcv, "soreceive 1");
                   1306:        SBLASTMBUFCHK(&so->so_rcv, "soreceive 1");
1.1       cgd      1307:        nextrecord = m->m_nextpkt;
                   1308:        if (pr->pr_flags & PR_ADDR) {
                   1309: #ifdef DIAGNOSTIC
                   1310:                if (m->m_type != MT_SONAME)
                   1311:                        panic("receive 1a");
                   1312: #endif
1.3       andrew   1313:                orig_resid = 0;
1.1       cgd      1314:                if (flags & MSG_PEEK) {
                   1315:                        if (paddr)
                   1316:                                *paddr = m_copy(m, 0, m->m_len);
                   1317:                        m = m->m_next;
                   1318:                } else {
                   1319:                        sbfree(&so->so_rcv, m);
1.67      he       1320:                        mbuf_removed = 1;
1.144     dyoung   1321:                        if (paddr != NULL) {
1.1       cgd      1322:                                *paddr = m;
                   1323:                                so->so_rcv.sb_mb = m->m_next;
1.144     dyoung   1324:                                m->m_next = NULL;
1.1       cgd      1325:                                m = so->so_rcv.sb_mb;
                   1326:                        } else {
1.249   ! christos 1327:                                m = so->so_rcv.sb_mb = m_free(m);
1.1       cgd      1328:                        }
1.159     ad       1329:                        sbsync(&so->so_rcv, nextrecord);
1.1       cgd      1330:                }
                   1331:        }
1.247     rjs      1332:        if (pr->pr_flags & PR_ADDR_OPT) {
                   1333:                /*
                   1334:                 * For SCTP we may be getting a
                   1335:                 * whole message OR a partial delivery.
                   1336:                 */
                   1337:                if (m->m_type == MT_SONAME) {
                   1338:                        orig_resid = 0;
                   1339:                        if (flags & MSG_PEEK) {
                   1340:                                if (paddr)
                   1341:                                        *paddr = m_copy(m, 0, m->m_len);
                   1342:                                m = m->m_next;
                   1343:                        } else {
                   1344:                                sbfree(&so->so_rcv, m);
                   1345:                                if (paddr) {
                   1346:                                        *paddr = m;
                   1347:                                        so->so_rcv.sb_mb = m->m_next;
                   1348:                                        m->m_next = 0;
                   1349:                                        m = so->so_rcv.sb_mb;
                   1350:                                } else {
1.249   ! christos 1351:                                        m = so->so_rcv.sb_mb = m_free(m);
1.247     rjs      1352:                                }
                   1353:                        }
                   1354:                }
                   1355:        }
1.159     ad       1356:
                   1357:        /*
                   1358:         * Process one or more MT_CONTROL mbufs present before any data mbufs
                   1359:         * in the first mbuf chain on the socket buffer.  If MSG_PEEK, we
                   1360:         * just copy the data; if !MSG_PEEK, we call into the protocol to
                   1361:         * perform externalization (or freeing if controlp == NULL).
                   1362:         */
                   1363:        if (__predict_false(m != NULL && m->m_type == MT_CONTROL)) {
                   1364:                struct mbuf *cm = NULL, *cmn;
                   1365:                struct mbuf **cme = &cm;
                   1366:
                   1367:                do {
                   1368:                        if (flags & MSG_PEEK) {
                   1369:                                if (controlp != NULL) {
                   1370:                                        *controlp = m_copy(m, 0, m->m_len);
                   1371:                                        controlp = &(*controlp)->m_next;
                   1372:                                }
                   1373:                                m = m->m_next;
                   1374:                        } else {
                   1375:                                sbfree(&so->so_rcv, m);
1.1       cgd      1376:                                so->so_rcv.sb_mb = m->m_next;
1.144     dyoung   1377:                                m->m_next = NULL;
1.159     ad       1378:                                *cme = m;
                   1379:                                cme = &(*cme)->m_next;
1.1       cgd      1380:                                m = so->so_rcv.sb_mb;
1.159     ad       1381:                        }
                   1382:                } while (m != NULL && m->m_type == MT_CONTROL);
                   1383:                if ((flags & MSG_PEEK) == 0)
                   1384:                        sbsync(&so->so_rcv, nextrecord);
                   1385:                for (; cm != NULL; cm = cmn) {
                   1386:                        cmn = cm->m_next;
                   1387:                        cm->m_next = NULL;
                   1388:                        type = mtod(cm, struct cmsghdr *)->cmsg_type;
                   1389:                        if (controlp != NULL) {
                   1390:                                if (dom->dom_externalize != NULL &&
                   1391:                                    type == SCM_RIGHTS) {
1.160     ad       1392:                                        sounlock(so);
1.159     ad       1393:                                        splx(s);
1.204     christos 1394:                                        error = (*dom->dom_externalize)(cm, l,
                   1395:                                            (flags & MSG_CMSG_CLOEXEC) ?
                   1396:                                            O_CLOEXEC : 0);
1.159     ad       1397:                                        s = splsoftnet();
1.160     ad       1398:                                        solock(so);
1.159     ad       1399:                                }
                   1400:                                *controlp = cm;
                   1401:                                while (*controlp != NULL)
                   1402:                                        controlp = &(*controlp)->m_next;
1.1       cgd      1403:                        } else {
1.106     itojun   1404:                                /*
                   1405:                                 * Dispose of any SCM_RIGHTS message that went
                   1406:                                 * through the read path rather than recv.
                   1407:                                 */
1.159     ad       1408:                                if (dom->dom_dispose != NULL &&
                   1409:                                    type == SCM_RIGHTS) {
1.160     ad       1410:                                        sounlock(so);
1.159     ad       1411:                                        (*dom->dom_dispose)(cm);
1.160     ad       1412:                                        solock(so);
1.159     ad       1413:                                }
                   1414:                                m_freem(cm);
1.1       cgd      1415:                        }
                   1416:                }
1.159     ad       1417:                if (m != NULL)
                   1418:                        nextrecord = so->so_rcv.sb_mb->m_nextpkt;
                   1419:                else
                   1420:                        nextrecord = so->so_rcv.sb_mb;
                   1421:                orig_resid = 0;
1.1       cgd      1422:        }
1.69      thorpej  1423:
1.159     ad       1424:        /* If m is non-NULL, we have some data to read. */
                   1425:        if (__predict_true(m != NULL)) {
1.1       cgd      1426:                type = m->m_type;
                   1427:                if (type == MT_OOBDATA)
                   1428:                        flags |= MSG_OOB;
                   1429:        }
1.69      thorpej  1430:        SBLASTRECORDCHK(&so->so_rcv, "soreceive 2");
                   1431:        SBLASTMBUFCHK(&so->so_rcv, "soreceive 2");
                   1432:
1.1       cgd      1433:        moff = 0;
                   1434:        offset = 0;
1.144     dyoung   1435:        while (m != NULL && uio->uio_resid > 0 && error == 0) {
1.1       cgd      1436:                if (m->m_type == MT_OOBDATA) {
                   1437:                        if (type != MT_OOBDATA)
                   1438:                                break;
                   1439:                } else if (type == MT_OOBDATA)
                   1440:                        break;
                   1441: #ifdef DIAGNOSTIC
                   1442:                else if (m->m_type != MT_DATA && m->m_type != MT_HEADER)
                   1443:                        panic("receive 3");
                   1444: #endif
                   1445:                so->so_state &= ~SS_RCVATMARK;
1.196     dsl      1446:                wakeup_state = 0;
1.1       cgd      1447:                len = uio->uio_resid;
                   1448:                if (so->so_oobmark && len > so->so_oobmark - offset)
                   1449:                        len = so->so_oobmark - offset;
                   1450:                if (len > m->m_len - moff)
                   1451:                        len = m->m_len - moff;
                   1452:                /*
                   1453:                 * If mp is set, just pass back the mbufs.
                   1454:                 * Otherwise copy them out via the uio, then free.
                   1455:                 * Sockbuf must be consistent here (points to current mbuf,
                   1456:                 * it points to next record) when we drop priority;
                   1457:                 * we must note any additions to the sockbuf when we
                   1458:                 * block interrupts again.
                   1459:                 */
1.144     dyoung   1460:                if (mp == NULL) {
1.69      thorpej  1461:                        SBLASTRECORDCHK(&so->so_rcv, "soreceive uiomove");
                   1462:                        SBLASTMBUFCHK(&so->so_rcv, "soreceive uiomove");
1.160     ad       1463:                        sounlock(so);
1.1       cgd      1464:                        splx(s);
1.211     chs      1465:                        error = uiomove(mtod(m, char *) + moff, len, uio);
1.20      mycroft  1466:                        s = splsoftnet();
1.160     ad       1467:                        solock(so);
1.144     dyoung   1468:                        if (error != 0) {
1.67      he       1469:                                /*
                   1470:                                 * If any part of the record has been removed
                   1471:                                 * (such as the MT_SONAME mbuf, which will
                   1472:                                 * happen when PR_ADDR, and thus also
                   1473:                                 * PR_ATOMIC, is set), then drop the entire
                   1474:                                 * record to maintain the atomicity of the
                   1475:                                 * receive operation.
                   1476:                                 *
                   1477:                                 * This avoids a later panic("receive 1a")
                   1478:                                 * when compiled with DIAGNOSTIC.
                   1479:                                 */
1.146     dyoung   1480:                                if (m && mbuf_removed && atomic)
1.67      he       1481:                                        (void) sbdroprecord(&so->so_rcv);
                   1482:
1.57      jdolecek 1483:                                goto release;
1.67      he       1484:                        }
1.1       cgd      1485:                } else
                   1486:                        uio->uio_resid -= len;
                   1487:                if (len == m->m_len - moff) {
                   1488:                        if (m->m_flags & M_EOR)
                   1489:                                flags |= MSG_EOR;
1.247     rjs      1490: #ifdef SCTP
                   1491:                        if (m->m_flags & M_NOTIFICATION)
                   1492:                                flags |= MSG_NOTIFICATION;
                   1493: #endif /* SCTP */
1.1       cgd      1494:                        if (flags & MSG_PEEK) {
                   1495:                                m = m->m_next;
                   1496:                                moff = 0;
                   1497:                        } else {
                   1498:                                nextrecord = m->m_nextpkt;
                   1499:                                sbfree(&so->so_rcv, m);
                   1500:                                if (mp) {
                   1501:                                        *mp = m;
                   1502:                                        mp = &m->m_next;
                   1503:                                        so->so_rcv.sb_mb = m = m->m_next;
1.140     dyoung   1504:                                        *mp = NULL;
1.1       cgd      1505:                                } else {
1.249   ! christos 1506:                                        m = so->so_rcv.sb_mb = m_free(m);
1.1       cgd      1507:                                }
1.69      thorpej  1508:                                /*
                   1509:                                 * If m != NULL, we also know that
                   1510:                                 * so->so_rcv.sb_mb != NULL.
                   1511:                                 */
                   1512:                                KASSERT(so->so_rcv.sb_mb == m);
                   1513:                                if (m) {
1.1       cgd      1514:                                        m->m_nextpkt = nextrecord;
1.69      thorpej  1515:                                        if (nextrecord == NULL)
                   1516:                                                so->so_rcv.sb_lastrecord = m;
                   1517:                                } else {
                   1518:                                        so->so_rcv.sb_mb = nextrecord;
1.70      thorpej  1519:                                        SB_EMPTY_FIXUP(&so->so_rcv);
1.69      thorpej  1520:                                }
                   1521:                                SBLASTRECORDCHK(&so->so_rcv, "soreceive 3");
                   1522:                                SBLASTMBUFCHK(&so->so_rcv, "soreceive 3");
1.1       cgd      1523:                        }
1.144     dyoung   1524:                } else if (flags & MSG_PEEK)
                   1525:                        moff += len;
                   1526:                else {
1.160     ad       1527:                        if (mp != NULL) {
                   1528:                                mt = m_copym(m, 0, len, M_NOWAIT);
                   1529:                                if (__predict_false(mt == NULL)) {
                   1530:                                        sounlock(so);
                   1531:                                        mt = m_copym(m, 0, len, M_WAIT);
                   1532:                                        solock(so);
                   1533:                                }
                   1534:                                *mp = mt;
                   1535:                        }
1.144     dyoung   1536:                        m->m_data += len;
                   1537:                        m->m_len -= len;
                   1538:                        so->so_rcv.sb_cc -= len;
1.1       cgd      1539:                }
                   1540:                if (so->so_oobmark) {
                   1541:                        if ((flags & MSG_PEEK) == 0) {
                   1542:                                so->so_oobmark -= len;
                   1543:                                if (so->so_oobmark == 0) {
                   1544:                                        so->so_state |= SS_RCVATMARK;
                   1545:                                        break;
                   1546:                                }
1.7       cgd      1547:                        } else {
1.1       cgd      1548:                                offset += len;
1.7       cgd      1549:                                if (offset == so->so_oobmark)
                   1550:                                        break;
                   1551:                        }
1.1       cgd      1552:                }
                   1553:                if (flags & MSG_EOR)
                   1554:                        break;
                   1555:                /*
                   1556:                 * If the MSG_WAITALL flag is set (for non-atomic socket),
                   1557:                 * we must not quit until "uio->uio_resid == 0" or an error
                   1558:                 * termination.  If a signal/timeout occurs, return
                   1559:                 * with a short count but without error.
                   1560:                 * Keep sockbuf locked against other readers.
                   1561:                 */
1.144     dyoung   1562:                while (flags & MSG_WAITALL && m == NULL && uio->uio_resid > 0 &&
1.3       andrew   1563:                    !sosendallatonce(so) && !nextrecord) {
1.1       cgd      1564:                        if (so->so_error || so->so_state & SS_CANTRCVMORE)
                   1565:                                break;
1.68      matt     1566:                        /*
                   1567:                         * If we are peeking and the socket receive buffer is
                   1568:                         * full, stop since we can't get more data to peek at.
                   1569:                         */
                   1570:                        if ((flags & MSG_PEEK) && sbspace(&so->so_rcv) <= 0)
                   1571:                                break;
                   1572:                        /*
                   1573:                         * If we've drained the socket buffer, tell the
                   1574:                         * protocol in case it needs to do something to
                   1575:                         * get it filled again.
                   1576:                         */
                   1577:                        if ((pr->pr_flags & PR_WANTRCVD) && so->so_pcb)
1.233     rtr      1578:                                (*pr->pr_usrreqs->pr_rcvd)(so, flags, l);
1.69      thorpej  1579:                        SBLASTRECORDCHK(&so->so_rcv, "soreceive sbwait 2");
                   1580:                        SBLASTMBUFCHK(&so->so_rcv, "soreceive sbwait 2");
1.196     dsl      1581:                        if (wakeup_state & SS_RESTARTSYS)
                   1582:                                error = ERESTART;
                   1583:                        else
                   1584:                                error = sbwait(&so->so_rcv);
1.144     dyoung   1585:                        if (error != 0) {
1.1       cgd      1586:                                sbunlock(&so->so_rcv);
1.160     ad       1587:                                sounlock(so);
1.1       cgd      1588:                                splx(s);
1.144     dyoung   1589:                                return 0;
1.1       cgd      1590:                        }
1.21      christos 1591:                        if ((m = so->so_rcv.sb_mb) != NULL)
1.1       cgd      1592:                                nextrecord = m->m_nextpkt;
1.196     dsl      1593:                        wakeup_state = so->so_state;
1.1       cgd      1594:                }
                   1595:        }
1.3       andrew   1596:
1.146     dyoung   1597:        if (m && atomic) {
1.3       andrew   1598:                flags |= MSG_TRUNC;
                   1599:                if ((flags & MSG_PEEK) == 0)
                   1600:                        (void) sbdroprecord(&so->so_rcv);
                   1601:        }
1.1       cgd      1602:        if ((flags & MSG_PEEK) == 0) {
1.144     dyoung   1603:                if (m == NULL) {
1.69      thorpej  1604:                        /*
1.70      thorpej  1605:                         * First part is an inline SB_EMPTY_FIXUP().  Second
1.69      thorpej  1606:                         * part makes sure sb_lastrecord is up-to-date if
                   1607:                         * there is still data in the socket buffer.
                   1608:                         */
1.1       cgd      1609:                        so->so_rcv.sb_mb = nextrecord;
1.69      thorpej  1610:                        if (so->so_rcv.sb_mb == NULL) {
                   1611:                                so->so_rcv.sb_mbtail = NULL;
                   1612:                                so->so_rcv.sb_lastrecord = NULL;
                   1613:                        } else if (nextrecord->m_nextpkt == NULL)
                   1614:                                so->so_rcv.sb_lastrecord = nextrecord;
                   1615:                }
                   1616:                SBLASTRECORDCHK(&so->so_rcv, "soreceive 4");
                   1617:                SBLASTMBUFCHK(&so->so_rcv, "soreceive 4");
1.1       cgd      1618:                if (pr->pr_flags & PR_WANTRCVD && so->so_pcb)
1.233     rtr      1619:                        (*pr->pr_usrreqs->pr_rcvd)(so, flags, l);
1.1       cgd      1620:        }
1.3       andrew   1621:        if (orig_resid == uio->uio_resid && orig_resid &&
                   1622:            (flags & MSG_EOR) == 0 && (so->so_state & SS_CANTRCVMORE) == 0) {
                   1623:                sbunlock(&so->so_rcv);
                   1624:                goto restart;
                   1625:        }
1.108     perry    1626:
1.144     dyoung   1627:        if (flagsp != NULL)
1.1       cgd      1628:                *flagsp |= flags;
1.54      lukem    1629:  release:
1.1       cgd      1630:        sbunlock(&so->so_rcv);
1.160     ad       1631:        sounlock(so);
1.1       cgd      1632:        splx(s);
1.144     dyoung   1633:        return error;
1.1       cgd      1634: }
                   1635:
1.14      mycroft  1636: int
1.54      lukem    1637: soshutdown(struct socket *so, int how)
1.1       cgd      1638: {
1.99      matt     1639:        const struct protosw    *pr;
1.160     ad       1640:        int     error;
                   1641:
                   1642:        KASSERT(solocked(so));
1.34      kleink   1643:
1.54      lukem    1644:        pr = so->so_proto;
1.34      kleink   1645:        if (!(how == SHUT_RD || how == SHUT_WR || how == SHUT_RDWR))
                   1646:                return (EINVAL);
1.1       cgd      1647:
1.160     ad       1648:        if (how == SHUT_RD || how == SHUT_RDWR) {
1.1       cgd      1649:                sorflush(so);
1.160     ad       1650:                error = 0;
                   1651:        }
1.34      kleink   1652:        if (how == SHUT_WR || how == SHUT_RDWR)
1.229     rtr      1653:                error = (*pr->pr_usrreqs->pr_shutdown)(so);
1.160     ad       1654:
                   1655:        return error;
1.1       cgd      1656: }
                   1657:
1.195     dsl      1658: void
1.196     dsl      1659: sorestart(struct socket *so)
1.188     ad       1660: {
1.196     dsl      1661:        /*
                   1662:         * An application has called close() on an fd on which another
                   1663:         * of its threads has called a socket system call.
                   1664:         * Mark this and wake everyone up, and code that would block again
                   1665:         * instead returns ERESTART.
                   1666:         * On system call re-entry the fd is validated and EBADF returned.
                   1667:         * Any other fd will block again on the 2nd syscall.
                   1668:         */
1.188     ad       1669:        solock(so);
1.196     dsl      1670:        so->so_state |= SS_RESTARTSYS;
1.188     ad       1671:        cv_broadcast(&so->so_cv);
1.196     dsl      1672:        cv_broadcast(&so->so_snd.sb_cv);
                   1673:        cv_broadcast(&so->so_rcv.sb_cv);
1.188     ad       1674:        sounlock(so);
                   1675: }
                   1676:
1.14      mycroft  1677: void
1.54      lukem    1678: sorflush(struct socket *so)
1.1       cgd      1679: {
1.54      lukem    1680:        struct sockbuf  *sb, asb;
1.99      matt     1681:        const struct protosw    *pr;
1.160     ad       1682:
                   1683:        KASSERT(solocked(so));
1.1       cgd      1684:
1.54      lukem    1685:        sb = &so->so_rcv;
                   1686:        pr = so->so_proto;
1.160     ad       1687:        socantrcvmore(so);
1.1       cgd      1688:        sb->sb_flags |= SB_NOINTR;
1.160     ad       1689:        (void )sblock(sb, M_WAITOK);
1.1       cgd      1690:        sbunlock(sb);
                   1691:        asb = *sb;
1.86      wrstuden 1692:        /*
                   1693:         * Clear most of the sockbuf structure, but leave some of the
                   1694:         * fields valid.
                   1695:         */
                   1696:        memset(&sb->sb_startzero, 0,
                   1697:            sizeof(*sb) - offsetof(struct sockbuf, sb_startzero));
1.160     ad       1698:        if (pr->pr_flags & PR_RIGHTS && pr->pr_domain->dom_dispose) {
                   1699:                sounlock(so);
1.1       cgd      1700:                (*pr->pr_domain->dom_dispose)(asb.sb_mb);
1.160     ad       1701:                solock(so);
                   1702:        }
1.98      christos 1703:        sbrelease(&asb, so);
1.1       cgd      1704: }
                   1705:
1.171     plunky   1706: /*
                   1707:  * internal set SOL_SOCKET options
                   1708:  */
1.142     dyoung   1709: static int
1.171     plunky   1710: sosetopt1(struct socket *so, const struct sockopt *sopt)
1.1       cgd      1711: {
1.219     christos 1712:        int error = EINVAL, opt;
                   1713:        int optval = 0; /* XXX: gcc */
1.171     plunky   1714:        struct linger l;
                   1715:        struct timeval tv;
1.142     dyoung   1716:
1.179     christos 1717:        switch ((opt = sopt->sopt_name)) {
1.142     dyoung   1718:
1.170     tls      1719:        case SO_ACCEPTFILTER:
1.177     ad       1720:                error = accept_filt_setopt(so, sopt);
                   1721:                KASSERT(solocked(so));
1.170     tls      1722:                break;
                   1723:
1.171     plunky   1724:        case SO_LINGER:
                   1725:                error = sockopt_get(sopt, &l, sizeof(l));
1.177     ad       1726:                solock(so);
1.171     plunky   1727:                if (error)
1.177     ad       1728:                        break;
1.171     plunky   1729:                if (l.l_linger < 0 || l.l_linger > USHRT_MAX ||
1.177     ad       1730:                    l.l_linger > (INT_MAX / hz)) {
                   1731:                        error = EDOM;
                   1732:                        break;
                   1733:                }
1.171     plunky   1734:                so->so_linger = l.l_linger;
                   1735:                if (l.l_onoff)
                   1736:                        so->so_options |= SO_LINGER;
                   1737:                else
                   1738:                        so->so_options &= ~SO_LINGER;
1.177     ad       1739:                break;
1.1       cgd      1740:
1.142     dyoung   1741:        case SO_DEBUG:
                   1742:        case SO_KEEPALIVE:
                   1743:        case SO_DONTROUTE:
                   1744:        case SO_USELOOPBACK:
                   1745:        case SO_BROADCAST:
                   1746:        case SO_REUSEADDR:
                   1747:        case SO_REUSEPORT:
                   1748:        case SO_OOBINLINE:
                   1749:        case SO_TIMESTAMP:
1.207     christos 1750:        case SO_NOSIGPIPE:
1.184     christos 1751: #ifdef SO_OTIMESTAMP
                   1752:        case SO_OTIMESTAMP:
                   1753: #endif
1.171     plunky   1754:                error = sockopt_getint(sopt, &optval);
1.177     ad       1755:                solock(so);
1.171     plunky   1756:                if (error)
1.177     ad       1757:                        break;
1.171     plunky   1758:                if (optval)
1.179     christos 1759:                        so->so_options |= opt;
1.142     dyoung   1760:                else
1.179     christos 1761:                        so->so_options &= ~opt;
1.142     dyoung   1762:                break;
                   1763:
                   1764:        case SO_SNDBUF:
                   1765:        case SO_RCVBUF:
                   1766:        case SO_SNDLOWAT:
                   1767:        case SO_RCVLOWAT:
1.171     plunky   1768:                error = sockopt_getint(sopt, &optval);
1.177     ad       1769:                solock(so);
1.171     plunky   1770:                if (error)
1.177     ad       1771:                        break;
1.1       cgd      1772:
1.142     dyoung   1773:                /*
                   1774:                 * Values < 1 make no sense for any of these
                   1775:                 * options, so disallow them.
                   1776:                 */
1.177     ad       1777:                if (optval < 1) {
                   1778:                        error = EINVAL;
                   1779:                        break;
                   1780:                }
1.1       cgd      1781:
1.179     christos 1782:                switch (opt) {
1.171     plunky   1783:                case SO_SNDBUF:
1.177     ad       1784:                        if (sbreserve(&so->so_snd, (u_long)optval, so) == 0) {
                   1785:                                error = ENOBUFS;
                   1786:                                break;
                   1787:                        }
1.171     plunky   1788:                        so->so_snd.sb_flags &= ~SB_AUTOSIZE;
                   1789:                        break;
1.1       cgd      1790:
                   1791:                case SO_RCVBUF:
1.177     ad       1792:                        if (sbreserve(&so->so_rcv, (u_long)optval, so) == 0) {
                   1793:                                error = ENOBUFS;
                   1794:                                break;
                   1795:                        }
1.171     plunky   1796:                        so->so_rcv.sb_flags &= ~SB_AUTOSIZE;
1.142     dyoung   1797:                        break;
                   1798:
                   1799:                /*
                   1800:                 * Make sure the low-water is never greater than
                   1801:                 * the high-water.
                   1802:                 */
1.1       cgd      1803:                case SO_SNDLOWAT:
1.171     plunky   1804:                        if (optval > so->so_snd.sb_hiwat)
                   1805:                                optval = so->so_snd.sb_hiwat;
                   1806:
                   1807:                        so->so_snd.sb_lowat = optval;
1.142     dyoung   1808:                        break;
1.171     plunky   1809:
1.1       cgd      1810:                case SO_RCVLOWAT:
1.171     plunky   1811:                        if (optval > so->so_rcv.sb_hiwat)
                   1812:                                optval = so->so_rcv.sb_hiwat;
                   1813:
                   1814:                        so->so_rcv.sb_lowat = optval;
1.142     dyoung   1815:                        break;
                   1816:                }
                   1817:                break;
1.28      thorpej  1818:
1.179     christos 1819: #ifdef COMPAT_50
                   1820:        case SO_OSNDTIMEO:
                   1821:        case SO_ORCVTIMEO: {
                   1822:                struct timeval50 otv;
                   1823:                error = sockopt_get(sopt, &otv, sizeof(otv));
1.186     pooka    1824:                if (error) {
                   1825:                        solock(so);
1.183     christos 1826:                        break;
1.186     pooka    1827:                }
1.179     christos 1828:                timeval50_to_timeval(&otv, &tv);
                   1829:                opt = opt == SO_OSNDTIMEO ? SO_SNDTIMEO : SO_RCVTIMEO;
1.182     christos 1830:                error = 0;
1.179     christos 1831:                /*FALLTHROUGH*/
                   1832:        }
                   1833: #endif /* COMPAT_50 */
                   1834:
1.142     dyoung   1835:        case SO_SNDTIMEO:
                   1836:        case SO_RCVTIMEO:
1.182     christos 1837:                if (error)
1.179     christos 1838:                        error = sockopt_get(sopt, &tv, sizeof(tv));
1.177     ad       1839:                solock(so);
1.171     plunky   1840:                if (error)
1.177     ad       1841:                        break;
1.171     plunky   1842:
1.177     ad       1843:                if (tv.tv_sec > (INT_MAX - tv.tv_usec / tick) / hz) {
                   1844:                        error = EDOM;
                   1845:                        break;
                   1846:                }
1.28      thorpej  1847:
1.171     plunky   1848:                optval = tv.tv_sec * hz + tv.tv_usec / tick;
                   1849:                if (optval == 0 && tv.tv_usec != 0)
                   1850:                        optval = 1;
1.28      thorpej  1851:
1.179     christos 1852:                switch (opt) {
1.142     dyoung   1853:                case SO_SNDTIMEO:
1.171     plunky   1854:                        so->so_snd.sb_timeo = optval;
1.1       cgd      1855:                        break;
                   1856:                case SO_RCVTIMEO:
1.171     plunky   1857:                        so->so_rcv.sb_timeo = optval;
1.142     dyoung   1858:                        break;
                   1859:                }
                   1860:                break;
1.1       cgd      1861:
1.142     dyoung   1862:        default:
1.177     ad       1863:                solock(so);
                   1864:                error = ENOPROTOOPT;
                   1865:                break;
1.142     dyoung   1866:        }
1.177     ad       1867:        KASSERT(solocked(so));
                   1868:        return error;
1.142     dyoung   1869: }
1.1       cgd      1870:
1.142     dyoung   1871: int
1.171     plunky   1872: sosetopt(struct socket *so, struct sockopt *sopt)
1.142     dyoung   1873: {
                   1874:        int error, prerr;
1.1       cgd      1875:
1.177     ad       1876:        if (sopt->sopt_level == SOL_SOCKET) {
1.171     plunky   1877:                error = sosetopt1(so, sopt);
1.177     ad       1878:                KASSERT(solocked(so));
                   1879:        } else {
1.142     dyoung   1880:                error = ENOPROTOOPT;
1.177     ad       1881:                solock(so);
                   1882:        }
1.1       cgd      1883:
1.142     dyoung   1884:        if ((error == 0 || error == ENOPROTOOPT) &&
                   1885:            so->so_proto != NULL && so->so_proto->pr_ctloutput != NULL) {
                   1886:                /* give the protocol stack a shot */
1.171     plunky   1887:                prerr = (*so->so_proto->pr_ctloutput)(PRCO_SETOPT, so, sopt);
1.142     dyoung   1888:                if (prerr == 0)
                   1889:                        error = 0;
                   1890:                else if (prerr != ENOPROTOOPT)
                   1891:                        error = prerr;
1.171     plunky   1892:        }
1.160     ad       1893:        sounlock(so);
1.142     dyoung   1894:        return error;
1.1       cgd      1895: }
                   1896:
1.171     plunky   1897: /*
                   1898:  * so_setsockopt() is a wrapper providing a sockopt structure for sosetopt()
                   1899:  */
                   1900: int
                   1901: so_setsockopt(struct lwp *l, struct socket *so, int level, int name,
                   1902:     const void *val, size_t valsize)
                   1903: {
                   1904:        struct sockopt sopt;
                   1905:        int error;
                   1906:
                   1907:        KASSERT(valsize == 0 || val != NULL);
                   1908:
                   1909:        sockopt_init(&sopt, level, name, valsize);
                   1910:        sockopt_set(&sopt, val, valsize);
                   1911:
                   1912:        error = sosetopt(so, &sopt);
                   1913:
                   1914:        sockopt_destroy(&sopt);
                   1915:
                   1916:        return error;
                   1917: }
                   1918:
                   1919: /*
                   1920:  * internal get SOL_SOCKET options
                   1921:  */
                   1922: static int
                   1923: sogetopt1(struct socket *so, struct sockopt *sopt)
                   1924: {
1.179     christos 1925:        int error, optval, opt;
1.171     plunky   1926:        struct linger l;
                   1927:        struct timeval tv;
                   1928:
1.179     christos 1929:        switch ((opt = sopt->sopt_name)) {
1.171     plunky   1930:
                   1931:        case SO_ACCEPTFILTER:
1.177     ad       1932:                error = accept_filt_getopt(so, sopt);
1.171     plunky   1933:                break;
                   1934:
                   1935:        case SO_LINGER:
                   1936:                l.l_onoff = (so->so_options & SO_LINGER) ? 1 : 0;
                   1937:                l.l_linger = so->so_linger;
                   1938:
                   1939:                error = sockopt_set(sopt, &l, sizeof(l));
                   1940:                break;
                   1941:
                   1942:        case SO_USELOOPBACK:
                   1943:        case SO_DONTROUTE:
                   1944:        case SO_DEBUG:
                   1945:        case SO_KEEPALIVE:
                   1946:        case SO_REUSEADDR:
                   1947:        case SO_REUSEPORT:
                   1948:        case SO_BROADCAST:
                   1949:        case SO_OOBINLINE:
                   1950:        case SO_TIMESTAMP:
1.207     christos 1951:        case SO_NOSIGPIPE:
1.184     christos 1952: #ifdef SO_OTIMESTAMP
                   1953:        case SO_OTIMESTAMP:
                   1954: #endif
1.218     seanb    1955:        case SO_ACCEPTCONN:
1.179     christos 1956:                error = sockopt_setint(sopt, (so->so_options & opt) ? 1 : 0);
1.171     plunky   1957:                break;
                   1958:
                   1959:        case SO_TYPE:
                   1960:                error = sockopt_setint(sopt, so->so_type);
                   1961:                break;
                   1962:
                   1963:        case SO_ERROR:
                   1964:                error = sockopt_setint(sopt, so->so_error);
                   1965:                so->so_error = 0;
                   1966:                break;
                   1967:
                   1968:        case SO_SNDBUF:
                   1969:                error = sockopt_setint(sopt, so->so_snd.sb_hiwat);
                   1970:                break;
                   1971:
                   1972:        case SO_RCVBUF:
                   1973:                error = sockopt_setint(sopt, so->so_rcv.sb_hiwat);
                   1974:                break;
                   1975:
                   1976:        case SO_SNDLOWAT:
                   1977:                error = sockopt_setint(sopt, so->so_snd.sb_lowat);
                   1978:                break;
                   1979:
                   1980:        case SO_RCVLOWAT:
                   1981:                error = sockopt_setint(sopt, so->so_rcv.sb_lowat);
                   1982:                break;
                   1983:
1.179     christos 1984: #ifdef COMPAT_50
                   1985:        case SO_OSNDTIMEO:
                   1986:        case SO_ORCVTIMEO: {
                   1987:                struct timeval50 otv;
                   1988:
                   1989:                optval = (opt == SO_OSNDTIMEO ?
                   1990:                     so->so_snd.sb_timeo : so->so_rcv.sb_timeo);
                   1991:
                   1992:                otv.tv_sec = optval / hz;
                   1993:                otv.tv_usec = (optval % hz) * tick;
                   1994:
                   1995:                error = sockopt_set(sopt, &otv, sizeof(otv));
                   1996:                break;
                   1997:        }
                   1998: #endif /* COMPAT_50 */
                   1999:
1.171     plunky   2000:        case SO_SNDTIMEO:
                   2001:        case SO_RCVTIMEO:
1.179     christos 2002:                optval = (opt == SO_SNDTIMEO ?
1.171     plunky   2003:                     so->so_snd.sb_timeo : so->so_rcv.sb_timeo);
                   2004:
                   2005:                tv.tv_sec = optval / hz;
                   2006:                tv.tv_usec = (optval % hz) * tick;
                   2007:
                   2008:                error = sockopt_set(sopt, &tv, sizeof(tv));
                   2009:                break;
                   2010:
                   2011:        case SO_OVERFLOWED:
                   2012:                error = sockopt_setint(sopt, so->so_rcv.sb_overflowed);
                   2013:                break;
                   2014:
                   2015:        default:
                   2016:                error = ENOPROTOOPT;
                   2017:                break;
                   2018:        }
                   2019:
                   2020:        return (error);
                   2021: }
                   2022:
1.14      mycroft  2023: int
1.171     plunky   2024: sogetopt(struct socket *so, struct sockopt *sopt)
1.1       cgd      2025: {
1.160     ad       2026:        int             error;
1.1       cgd      2027:
1.160     ad       2028:        solock(so);
1.171     plunky   2029:        if (sopt->sopt_level != SOL_SOCKET) {
1.1       cgd      2030:                if (so->so_proto && so->so_proto->pr_ctloutput) {
1.160     ad       2031:                        error = ((*so->so_proto->pr_ctloutput)
1.171     plunky   2032:                            (PRCO_GETOPT, so, sopt));
1.1       cgd      2033:                } else
1.160     ad       2034:                        error = (ENOPROTOOPT);
1.1       cgd      2035:        } else {
1.171     plunky   2036:                error = sogetopt1(so, sopt);
                   2037:        }
                   2038:        sounlock(so);
                   2039:        return (error);
                   2040: }
                   2041:
                   2042: /*
                   2043:  * alloc sockopt data buffer buffer
                   2044:  *     - will be released at destroy
                   2045:  */
1.176     plunky   2046: static int
                   2047: sockopt_alloc(struct sockopt *sopt, size_t len, km_flag_t kmflag)
1.171     plunky   2048: {
                   2049:
                   2050:        KASSERT(sopt->sopt_size == 0);
                   2051:
1.176     plunky   2052:        if (len > sizeof(sopt->sopt_buf)) {
                   2053:                sopt->sopt_data = kmem_zalloc(len, kmflag);
                   2054:                if (sopt->sopt_data == NULL)
                   2055:                        return ENOMEM;
                   2056:        } else
1.171     plunky   2057:                sopt->sopt_data = sopt->sopt_buf;
                   2058:
                   2059:        sopt->sopt_size = len;
1.176     plunky   2060:        return 0;
1.171     plunky   2061: }
                   2062:
                   2063: /*
                   2064:  * initialise sockopt storage
1.176     plunky   2065:  *     - MAY sleep during allocation
1.171     plunky   2066:  */
                   2067: void
                   2068: sockopt_init(struct sockopt *sopt, int level, int name, size_t size)
                   2069: {
1.1       cgd      2070:
1.171     plunky   2071:        memset(sopt, 0, sizeof(*sopt));
1.1       cgd      2072:
1.171     plunky   2073:        sopt->sopt_level = level;
                   2074:        sopt->sopt_name = name;
1.176     plunky   2075:        (void)sockopt_alloc(sopt, size, KM_SLEEP);
1.171     plunky   2076: }
                   2077:
                   2078: /*
                   2079:  * destroy sockopt storage
                   2080:  *     - will release any held memory references
                   2081:  */
                   2082: void
                   2083: sockopt_destroy(struct sockopt *sopt)
                   2084: {
                   2085:
                   2086:        if (sopt->sopt_data != sopt->sopt_buf)
1.173     plunky   2087:                kmem_free(sopt->sopt_data, sopt->sopt_size);
1.171     plunky   2088:
                   2089:        memset(sopt, 0, sizeof(*sopt));
                   2090: }
                   2091:
                   2092: /*
                   2093:  * set sockopt value
                   2094:  *     - value is copied into sockopt
1.176     plunky   2095:  *     - memory is allocated when necessary, will not sleep
1.171     plunky   2096:  */
                   2097: int
                   2098: sockopt_set(struct sockopt *sopt, const void *buf, size_t len)
                   2099: {
1.176     plunky   2100:        int error;
1.171     plunky   2101:
1.176     plunky   2102:        if (sopt->sopt_size == 0) {
                   2103:                error = sockopt_alloc(sopt, len, KM_NOSLEEP);
                   2104:                if (error)
                   2105:                        return error;
                   2106:        }
1.171     plunky   2107:
                   2108:        KASSERT(sopt->sopt_size == len);
                   2109:        memcpy(sopt->sopt_data, buf, len);
                   2110:        return 0;
                   2111: }
                   2112:
                   2113: /*
                   2114:  * common case of set sockopt integer value
                   2115:  */
                   2116: int
                   2117: sockopt_setint(struct sockopt *sopt, int val)
                   2118: {
                   2119:
                   2120:        return sockopt_set(sopt, &val, sizeof(int));
                   2121: }
                   2122:
                   2123: /*
                   2124:  * get sockopt value
                   2125:  *     - correct size must be given
                   2126:  */
                   2127: int
                   2128: sockopt_get(const struct sockopt *sopt, void *buf, size_t len)
                   2129: {
1.170     tls      2130:
1.171     plunky   2131:        if (sopt->sopt_size != len)
                   2132:                return EINVAL;
1.1       cgd      2133:
1.171     plunky   2134:        memcpy(buf, sopt->sopt_data, len);
                   2135:        return 0;
                   2136: }
1.1       cgd      2137:
1.171     plunky   2138: /*
                   2139:  * common case of get sockopt integer value
                   2140:  */
                   2141: int
                   2142: sockopt_getint(const struct sockopt *sopt, int *valp)
                   2143: {
1.1       cgd      2144:
1.171     plunky   2145:        return sockopt_get(sopt, valp, sizeof(int));
                   2146: }
1.1       cgd      2147:
1.171     plunky   2148: /*
                   2149:  * set sockopt value from mbuf
                   2150:  *     - ONLY for legacy code
                   2151:  *     - mbuf is released by sockopt
1.176     plunky   2152:  *     - will not sleep
1.171     plunky   2153:  */
                   2154: int
                   2155: sockopt_setmbuf(struct sockopt *sopt, struct mbuf *m)
                   2156: {
                   2157:        size_t len;
1.176     plunky   2158:        int error;
1.1       cgd      2159:
1.171     plunky   2160:        len = m_length(m);
1.1       cgd      2161:
1.176     plunky   2162:        if (sopt->sopt_size == 0) {
                   2163:                error = sockopt_alloc(sopt, len, KM_NOSLEEP);
                   2164:                if (error)
                   2165:                        return error;
                   2166:        }
1.1       cgd      2167:
1.171     plunky   2168:        KASSERT(sopt->sopt_size == len);
                   2169:        m_copydata(m, 0, len, sopt->sopt_data);
                   2170:        m_freem(m);
1.1       cgd      2171:
1.171     plunky   2172:        return 0;
                   2173: }
1.1       cgd      2174:
1.171     plunky   2175: /*
                   2176:  * get sockopt value into mbuf
                   2177:  *     - ONLY for legacy code
                   2178:  *     - mbuf to be released by the caller
1.176     plunky   2179:  *     - will not sleep
1.171     plunky   2180:  */
                   2181: struct mbuf *
                   2182: sockopt_getmbuf(const struct sockopt *sopt)
                   2183: {
                   2184:        struct mbuf *m;
1.107     darrenr  2185:
1.176     plunky   2186:        if (sopt->sopt_size > MCLBYTES)
                   2187:                return NULL;
                   2188:
                   2189:        m = m_get(M_DONTWAIT, MT_SOOPTS);
1.171     plunky   2190:        if (m == NULL)
                   2191:                return NULL;
                   2192:
1.176     plunky   2193:        if (sopt->sopt_size > MLEN) {
                   2194:                MCLGET(m, M_DONTWAIT);
                   2195:                if ((m->m_flags & M_EXT) == 0) {
                   2196:                        m_free(m);
                   2197:                        return NULL;
                   2198:                }
1.1       cgd      2199:        }
1.176     plunky   2200:
                   2201:        memcpy(mtod(m, void *), sopt->sopt_data, sopt->sopt_size);
                   2202:        m->m_len = sopt->sopt_size;
1.160     ad       2203:
1.171     plunky   2204:        return m;
1.1       cgd      2205: }
                   2206:
1.14      mycroft  2207: void
1.54      lukem    2208: sohasoutofband(struct socket *so)
1.1       cgd      2209: {
1.153     rmind    2210:
1.90      christos 2211:        fownsignal(so->so_pgid, SIGURG, POLL_PRI, POLLPRI|POLLRDBAND, so);
1.189     ad       2212:        selnotify(&so->so_rcv.sb_sel, POLLPRI | POLLRDBAND, NOTE_SUBMIT);
1.1       cgd      2213: }
1.72      jdolecek 2214:
                   2215: static void
                   2216: filt_sordetach(struct knote *kn)
                   2217: {
                   2218:        struct socket   *so;
                   2219:
1.235     matt     2220:        so = ((file_t *)kn->kn_obj)->f_socket;
1.160     ad       2221:        solock(so);
1.73      christos 2222:        SLIST_REMOVE(&so->so_rcv.sb_sel.sel_klist, kn, knote, kn_selnext);
                   2223:        if (SLIST_EMPTY(&so->so_rcv.sb_sel.sel_klist))
1.72      jdolecek 2224:                so->so_rcv.sb_flags &= ~SB_KNOTE;
1.160     ad       2225:        sounlock(so);
1.72      jdolecek 2226: }
                   2227:
                   2228: /*ARGSUSED*/
                   2229: static int
1.129     yamt     2230: filt_soread(struct knote *kn, long hint)
1.72      jdolecek 2231: {
                   2232:        struct socket   *so;
1.160     ad       2233:        int rv;
1.72      jdolecek 2234:
1.235     matt     2235:        so = ((file_t *)kn->kn_obj)->f_socket;
1.160     ad       2236:        if (hint != NOTE_SUBMIT)
                   2237:                solock(so);
1.72      jdolecek 2238:        kn->kn_data = so->so_rcv.sb_cc;
                   2239:        if (so->so_state & SS_CANTRCVMORE) {
1.108     perry    2240:                kn->kn_flags |= EV_EOF;
1.72      jdolecek 2241:                kn->kn_fflags = so->so_error;
1.160     ad       2242:                rv = 1;
                   2243:        } else if (so->so_error)        /* temporary udp error */
                   2244:                rv = 1;
                   2245:        else if (kn->kn_sfflags & NOTE_LOWAT)
                   2246:                rv = (kn->kn_data >= kn->kn_sdata);
                   2247:        else
                   2248:                rv = (kn->kn_data >= so->so_rcv.sb_lowat);
                   2249:        if (hint != NOTE_SUBMIT)
                   2250:                sounlock(so);
                   2251:        return rv;
1.72      jdolecek 2252: }
                   2253:
                   2254: static void
                   2255: filt_sowdetach(struct knote *kn)
                   2256: {
                   2257:        struct socket   *so;
                   2258:
1.235     matt     2259:        so = ((file_t *)kn->kn_obj)->f_socket;
1.160     ad       2260:        solock(so);
1.73      christos 2261:        SLIST_REMOVE(&so->so_snd.sb_sel.sel_klist, kn, knote, kn_selnext);
                   2262:        if (SLIST_EMPTY(&so->so_snd.sb_sel.sel_klist))
1.72      jdolecek 2263:                so->so_snd.sb_flags &= ~SB_KNOTE;
1.160     ad       2264:        sounlock(so);
1.72      jdolecek 2265: }
                   2266:
                   2267: /*ARGSUSED*/
                   2268: static int
1.129     yamt     2269: filt_sowrite(struct knote *kn, long hint)
1.72      jdolecek 2270: {
                   2271:        struct socket   *so;
1.160     ad       2272:        int rv;
1.72      jdolecek 2273:
1.235     matt     2274:        so = ((file_t *)kn->kn_obj)->f_socket;
1.160     ad       2275:        if (hint != NOTE_SUBMIT)
                   2276:                solock(so);
1.72      jdolecek 2277:        kn->kn_data = sbspace(&so->so_snd);
                   2278:        if (so->so_state & SS_CANTSENDMORE) {
1.108     perry    2279:                kn->kn_flags |= EV_EOF;
1.72      jdolecek 2280:                kn->kn_fflags = so->so_error;
1.160     ad       2281:                rv = 1;
                   2282:        } else if (so->so_error)        /* temporary udp error */
                   2283:                rv = 1;
                   2284:        else if (((so->so_state & SS_ISCONNECTED) == 0) &&
1.72      jdolecek 2285:            (so->so_proto->pr_flags & PR_CONNREQUIRED))
1.160     ad       2286:                rv = 0;
                   2287:        else if (kn->kn_sfflags & NOTE_LOWAT)
                   2288:                rv = (kn->kn_data >= kn->kn_sdata);
                   2289:        else
                   2290:                rv = (kn->kn_data >= so->so_snd.sb_lowat);
                   2291:        if (hint != NOTE_SUBMIT)
                   2292:                sounlock(so);
                   2293:        return rv;
1.72      jdolecek 2294: }
                   2295:
                   2296: /*ARGSUSED*/
                   2297: static int
1.129     yamt     2298: filt_solisten(struct knote *kn, long hint)
1.72      jdolecek 2299: {
                   2300:        struct socket   *so;
1.160     ad       2301:        int rv;
1.72      jdolecek 2302:
1.235     matt     2303:        so = ((file_t *)kn->kn_obj)->f_socket;
1.72      jdolecek 2304:
                   2305:        /*
                   2306:         * Set kn_data to number of incoming connections, not
                   2307:         * counting partial (incomplete) connections.
1.108     perry    2308:         */
1.160     ad       2309:        if (hint != NOTE_SUBMIT)
                   2310:                solock(so);
1.72      jdolecek 2311:        kn->kn_data = so->so_qlen;
1.160     ad       2312:        rv = (kn->kn_data > 0);
                   2313:        if (hint != NOTE_SUBMIT)
                   2314:                sounlock(so);
                   2315:        return rv;
1.72      jdolecek 2316: }
                   2317:
                   2318: static const struct filterops solisten_filtops =
                   2319:        { 1, NULL, filt_sordetach, filt_solisten };
                   2320: static const struct filterops soread_filtops =
                   2321:        { 1, NULL, filt_sordetach, filt_soread };
                   2322: static const struct filterops sowrite_filtops =
                   2323:        { 1, NULL, filt_sowdetach, filt_sowrite };
                   2324:
                   2325: int
1.129     yamt     2326: soo_kqfilter(struct file *fp, struct knote *kn)
1.72      jdolecek 2327: {
                   2328:        struct socket   *so;
                   2329:        struct sockbuf  *sb;
                   2330:
1.235     matt     2331:        so = ((file_t *)kn->kn_obj)->f_socket;
1.160     ad       2332:        solock(so);
1.72      jdolecek 2333:        switch (kn->kn_filter) {
                   2334:        case EVFILT_READ:
                   2335:                if (so->so_options & SO_ACCEPTCONN)
                   2336:                        kn->kn_fop = &solisten_filtops;
                   2337:                else
                   2338:                        kn->kn_fop = &soread_filtops;
                   2339:                sb = &so->so_rcv;
                   2340:                break;
                   2341:        case EVFILT_WRITE:
                   2342:                kn->kn_fop = &sowrite_filtops;
                   2343:                sb = &so->so_snd;
                   2344:                break;
                   2345:        default:
1.160     ad       2346:                sounlock(so);
1.149     pooka    2347:                return (EINVAL);
1.72      jdolecek 2348:        }
1.73      christos 2349:        SLIST_INSERT_HEAD(&sb->sb_sel.sel_klist, kn, kn_selnext);
1.72      jdolecek 2350:        sb->sb_flags |= SB_KNOTE;
1.160     ad       2351:        sounlock(so);
1.72      jdolecek 2352:        return (0);
                   2353: }
                   2354:
1.154     ad       2355: static int
                   2356: sodopoll(struct socket *so, int events)
                   2357: {
                   2358:        int revents;
                   2359:
                   2360:        revents = 0;
                   2361:
                   2362:        if (events & (POLLIN | POLLRDNORM))
                   2363:                if (soreadable(so))
                   2364:                        revents |= events & (POLLIN | POLLRDNORM);
                   2365:
                   2366:        if (events & (POLLOUT | POLLWRNORM))
                   2367:                if (sowritable(so))
                   2368:                        revents |= events & (POLLOUT | POLLWRNORM);
                   2369:
                   2370:        if (events & (POLLPRI | POLLRDBAND))
                   2371:                if (so->so_oobmark || (so->so_state & SS_RCVATMARK))
                   2372:                        revents |= events & (POLLPRI | POLLRDBAND);
                   2373:
                   2374:        return revents;
                   2375: }
                   2376:
                   2377: int
                   2378: sopoll(struct socket *so, int events)
                   2379: {
                   2380:        int revents = 0;
                   2381:
1.160     ad       2382: #ifndef DIAGNOSTIC
                   2383:        /*
                   2384:         * Do a quick, unlocked check in expectation that the socket
                   2385:         * will be ready for I/O.  Don't do this check if DIAGNOSTIC,
                   2386:         * as the solocked() assertions will fail.
                   2387:         */
1.154     ad       2388:        if ((revents = sodopoll(so, events)) != 0)
                   2389:                return revents;
1.160     ad       2390: #endif
1.154     ad       2391:
1.160     ad       2392:        solock(so);
1.154     ad       2393:        if ((revents = sodopoll(so, events)) == 0) {
                   2394:                if (events & (POLLIN | POLLPRI | POLLRDNORM | POLLRDBAND)) {
                   2395:                        selrecord(curlwp, &so->so_rcv.sb_sel);
1.160     ad       2396:                        so->so_rcv.sb_flags |= SB_NOTIFY;
1.154     ad       2397:                }
                   2398:
                   2399:                if (events & (POLLOUT | POLLWRNORM)) {
                   2400:                        selrecord(curlwp, &so->so_snd.sb_sel);
1.160     ad       2401:                        so->so_snd.sb_flags |= SB_NOTIFY;
1.154     ad       2402:                }
                   2403:        }
1.160     ad       2404:        sounlock(so);
1.154     ad       2405:
                   2406:        return revents;
                   2407: }
                   2408:
                   2409:
1.94      yamt     2410: #include <sys/sysctl.h>
                   2411:
                   2412: static int sysctl_kern_somaxkva(SYSCTLFN_PROTO);
1.212     pooka    2413: static int sysctl_kern_sbmax(SYSCTLFN_PROTO);
1.94      yamt     2414:
                   2415: /*
                   2416:  * sysctl helper routine for kern.somaxkva.  ensures that the given
                   2417:  * value is not too small.
                   2418:  * (XXX should we maybe make sure it's not too large as well?)
                   2419:  */
                   2420: static int
                   2421: sysctl_kern_somaxkva(SYSCTLFN_ARGS)
                   2422: {
                   2423:        int error, new_somaxkva;
                   2424:        struct sysctlnode node;
                   2425:
                   2426:        new_somaxkva = somaxkva;
                   2427:        node = *rnode;
                   2428:        node.sysctl_data = &new_somaxkva;
                   2429:        error = sysctl_lookup(SYSCTLFN_CALL(&node));
                   2430:        if (error || newp == NULL)
                   2431:                return (error);
                   2432:
                   2433:        if (new_somaxkva < (16 * 1024 * 1024)) /* sanity */
                   2434:                return (EINVAL);
                   2435:
1.136     ad       2436:        mutex_enter(&so_pendfree_lock);
1.94      yamt     2437:        somaxkva = new_somaxkva;
1.136     ad       2438:        cv_broadcast(&socurkva_cv);
                   2439:        mutex_exit(&so_pendfree_lock);
1.94      yamt     2440:
                   2441:        return (error);
                   2442: }
                   2443:
1.212     pooka    2444: /*
                   2445:  * sysctl helper routine for kern.sbmax. Basically just ensures that
                   2446:  * any new value is not too small.
                   2447:  */
                   2448: static int
                   2449: sysctl_kern_sbmax(SYSCTLFN_ARGS)
                   2450: {
                   2451:        int error, new_sbmax;
                   2452:        struct sysctlnode node;
                   2453:
                   2454:        new_sbmax = sb_max;
                   2455:        node = *rnode;
                   2456:        node.sysctl_data = &new_sbmax;
                   2457:        error = sysctl_lookup(SYSCTLFN_CALL(&node));
                   2458:        if (error || newp == NULL)
                   2459:                return (error);
                   2460:
                   2461:        KERNEL_LOCK(1, NULL);
                   2462:        error = sb_max_set(new_sbmax);
                   2463:        KERNEL_UNLOCK_ONE(NULL);
                   2464:
                   2465:        return (error);
                   2466: }
                   2467:
1.178     pooka    2468: static void
1.212     pooka    2469: sysctl_kern_socket_setup(void)
1.94      yamt     2470: {
                   2471:
1.178     pooka    2472:        KASSERT(socket_sysctllog == NULL);
1.97      atatat   2473:
1.178     pooka    2474:        sysctl_createv(&socket_sysctllog, 0, NULL, NULL,
1.97      atatat   2475:                       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.103     atatat   2476:                       CTLTYPE_INT, "somaxkva",
                   2477:                       SYSCTL_DESCR("Maximum amount of kernel memory to be "
                   2478:                                    "used for socket buffers"),
1.94      yamt     2479:                       sysctl_kern_somaxkva, 0, NULL, 0,
                   2480:                       CTL_KERN, KERN_SOMAXKVA, CTL_EOL);
1.212     pooka    2481:
                   2482:        sysctl_createv(&socket_sysctllog, 0, NULL, NULL,
                   2483:                       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
                   2484:                       CTLTYPE_INT, "sbmax",
                   2485:                       SYSCTL_DESCR("Maximum socket buffer size"),
                   2486:                       sysctl_kern_sbmax, 0, NULL, 0,
                   2487:                       CTL_KERN, KERN_SBMAX, CTL_EOL);
1.94      yamt     2488: }

CVSweb <webmaster@jp.NetBSD.org>