[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 1.144

1.144   ! dyoung      1: /*     $NetBSD: uipc_socket.c,v 1.143 2007/09/25 14:04:07 ad Exp $     */
1.64      thorpej     2:
                      3: /*-
1.137     ad          4:  * Copyright (c) 2002, 2007 The NetBSD Foundation, Inc.
1.64      thorpej     5:  * All rights reserved.
                      6:  *
                      7:  * This code is derived from software contributed to The NetBSD Foundation
                      8:  * by Jason R. Thorpe of Wasabi Systems, Inc.
                      9:  *
                     10:  * Redistribution and use in source and binary forms, with or without
                     11:  * modification, are permitted provided that the following conditions
                     12:  * are met:
                     13:  * 1. Redistributions of source code must retain the above copyright
                     14:  *    notice, this list of conditions and the following disclaimer.
                     15:  * 2. Redistributions in binary form must reproduce the above copyright
                     16:  *    notice, this list of conditions and the following disclaimer in the
                     17:  *    documentation and/or other materials provided with the distribution.
                     18:  * 3. All advertising materials mentioning features or use of this software
                     19:  *    must display the following acknowledgement:
                     20:  *     This product includes software developed by the NetBSD
                     21:  *     Foundation, Inc. and its contributors.
                     22:  * 4. Neither the name of The NetBSD Foundation nor the names of its
                     23:  *    contributors may be used to endorse or promote products derived
                     24:  *    from this software without specific prior written permission.
                     25:  *
                     26:  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
                     27:  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
                     28:  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
                     29:  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
                     30:  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
                     31:  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
                     32:  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
                     33:  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
                     34:  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
                     35:  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
                     36:  * POSSIBILITY OF SUCH DAMAGE.
                     37:  */
1.16      cgd        38:
1.1       cgd        39: /*
1.15      mycroft    40:  * Copyright (c) 1982, 1986, 1988, 1990, 1993
                     41:  *     The Regents of the University of California.  All rights reserved.
1.1       cgd        42:  *
                     43:  * Redistribution and use in source and binary forms, with or without
                     44:  * modification, are permitted provided that the following conditions
                     45:  * are met:
                     46:  * 1. Redistributions of source code must retain the above copyright
                     47:  *    notice, this list of conditions and the following disclaimer.
                     48:  * 2. Redistributions in binary form must reproduce the above copyright
                     49:  *    notice, this list of conditions and the following disclaimer in the
                     50:  *    documentation and/or other materials provided with the distribution.
1.85      agc        51:  * 3. Neither the name of the University nor the names of its contributors
1.1       cgd        52:  *    may be used to endorse or promote products derived from this software
                     53:  *    without specific prior written permission.
                     54:  *
                     55:  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
                     56:  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
                     57:  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
                     58:  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
                     59:  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
                     60:  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
                     61:  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
                     62:  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
                     63:  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
                     64:  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
                     65:  * SUCH DAMAGE.
                     66:  *
1.32      fvdl       67:  *     @(#)uipc_socket.c       8.6 (Berkeley) 5/2/95
1.1       cgd        68:  */
1.59      lukem      69:
                     70: #include <sys/cdefs.h>
1.144   ! dyoung     71: __KERNEL_RCSID(0, "$NetBSD: uipc_socket.c,v 1.143 2007/09/25 14:04:07 ad Exp $");
1.64      thorpej    72:
                     73: #include "opt_sock_counters.h"
                     74: #include "opt_sosend_loan.h"
1.81      martin     75: #include "opt_mbuftrace.h"
1.84      ragge      76: #include "opt_somaxkva.h"
1.1       cgd        77:
1.9       mycroft    78: #include <sys/param.h>
                     79: #include <sys/systm.h>
                     80: #include <sys/proc.h>
                     81: #include <sys/file.h>
1.142     dyoung     82: #include <sys/filedesc.h>
1.9       mycroft    83: #include <sys/malloc.h>
                     84: #include <sys/mbuf.h>
                     85: #include <sys/domain.h>
                     86: #include <sys/kernel.h>
                     87: #include <sys/protosw.h>
                     88: #include <sys/socket.h>
                     89: #include <sys/socketvar.h>
1.21      christos   90: #include <sys/signalvar.h>
1.9       mycroft    91: #include <sys/resourcevar.h>
1.37      thorpej    92: #include <sys/pool.h>
1.72      jdolecek   93: #include <sys/event.h>
1.89      christos   94: #include <sys/poll.h>
1.118     elad       95: #include <sys/kauth.h>
1.136     ad         96: #include <sys/mutex.h>
                     97: #include <sys/condvar.h>
1.37      thorpej    98:
1.64      thorpej    99: #include <uvm/uvm.h>
                    100:
1.135     ad        101: POOL_INIT(socket_pool, sizeof(struct socket), 0, 0, 0, "sockpl", NULL,
                    102:     IPL_SOFTNET);
1.77      thorpej   103:
                    104: MALLOC_DEFINE(M_SOOPTS, "soopts", "socket options");
                    105: MALLOC_DEFINE(M_SONAME, "soname", "socket name");
1.37      thorpej   106:
1.142     dyoung    107: extern const struct fileops socketops;
                    108:
1.54      lukem     109: extern int     somaxconn;                      /* patchable (XXX sysctl) */
                    110: int            somaxconn = SOMAXCONN;
1.49      jonathan  111:
1.64      thorpej   112: #ifdef SOSEND_COUNTERS
                    113: #include <sys/device.h>
                    114:
1.113     thorpej   115: static struct evcnt sosend_loan_big = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64      thorpej   116:     NULL, "sosend", "loan big");
1.113     thorpej   117: static struct evcnt sosend_copy_big = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64      thorpej   118:     NULL, "sosend", "copy big");
1.113     thorpej   119: static struct evcnt sosend_copy_small = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64      thorpej   120:     NULL, "sosend", "copy small");
1.113     thorpej   121: static struct evcnt sosend_kvalimit = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64      thorpej   122:     NULL, "sosend", "kva limit");
                    123:
                    124: #define        SOSEND_COUNTER_INCR(ev)         (ev)->ev_count++
                    125:
1.101     matt      126: EVCNT_ATTACH_STATIC(sosend_loan_big);
                    127: EVCNT_ATTACH_STATIC(sosend_copy_big);
                    128: EVCNT_ATTACH_STATIC(sosend_copy_small);
                    129: EVCNT_ATTACH_STATIC(sosend_kvalimit);
1.64      thorpej   130: #else
                    131:
                    132: #define        SOSEND_COUNTER_INCR(ev)         /* nothing */
                    133:
                    134: #endif /* SOSEND_COUNTERS */
                    135:
1.119     yamt      136: static struct callback_entry sokva_reclaimerentry;
1.1       cgd       137:
1.71      thorpej   138: #ifdef SOSEND_NO_LOAN
1.121     yamt      139: int sock_loan_thresh = -1;
1.71      thorpej   140: #else
1.121     yamt      141: int sock_loan_thresh = 4096;
1.65      thorpej   142: #endif
1.64      thorpej   143:
1.136     ad        144: static kmutex_t so_pendfree_lock;
1.113     thorpej   145: static struct mbuf *so_pendfree;
1.64      thorpej   146:
1.84      ragge     147: #ifndef SOMAXKVA
                    148: #define        SOMAXKVA (16 * 1024 * 1024)
                    149: #endif
                    150: int somaxkva = SOMAXKVA;
1.113     thorpej   151: static int socurkva;
1.136     ad        152: static kcondvar_t socurkva_cv;
1.64      thorpej   153:
                    154: #define        SOCK_LOAN_CHUNK         65536
                    155:
1.117     yamt      156: static size_t sodopendfree(void);
                    157: static size_t sodopendfreel(void);
1.93      yamt      158:
1.113     thorpej   159: static vsize_t
1.129     yamt      160: sokvareserve(struct socket *so, vsize_t len)
1.80      yamt      161: {
1.98      christos  162:        int error;
1.80      yamt      163:
1.136     ad        164:        mutex_enter(&so_pendfree_lock);
1.80      yamt      165:        while (socurkva + len > somaxkva) {
1.93      yamt      166:                size_t freed;
                    167:
                    168:                /*
                    169:                 * try to do pendfree.
                    170:                 */
                    171:
1.117     yamt      172:                freed = sodopendfreel();
1.93      yamt      173:
                    174:                /*
                    175:                 * if some kva was freed, try again.
                    176:                 */
                    177:
                    178:                if (freed)
1.80      yamt      179:                        continue;
1.93      yamt      180:
1.80      yamt      181:                SOSEND_COUNTER_INCR(&sosend_kvalimit);
1.136     ad        182:                error = cv_wait_sig(&socurkva_cv, &so_pendfree_lock);
1.98      christos  183:                if (error) {
                    184:                        len = 0;
                    185:                        break;
                    186:                }
1.80      yamt      187:        }
1.93      yamt      188:        socurkva += len;
1.136     ad        189:        mutex_exit(&so_pendfree_lock);
1.98      christos  190:        return len;
1.95      yamt      191: }
                    192:
1.113     thorpej   193: static void
1.95      yamt      194: sokvaunreserve(vsize_t len)
                    195: {
                    196:
1.136     ad        197:        mutex_enter(&so_pendfree_lock);
1.95      yamt      198:        socurkva -= len;
1.136     ad        199:        cv_broadcast(&socurkva_cv);
                    200:        mutex_exit(&so_pendfree_lock);
1.95      yamt      201: }
                    202:
                    203: /*
                    204:  * sokvaalloc: allocate kva for loan.
                    205:  */
                    206:
                    207: vaddr_t
                    208: sokvaalloc(vsize_t len, struct socket *so)
                    209: {
                    210:        vaddr_t lva;
                    211:
                    212:        /*
                    213:         * reserve kva.
                    214:         */
                    215:
1.98      christos  216:        if (sokvareserve(so, len) == 0)
                    217:                return 0;
1.93      yamt      218:
                    219:        /*
                    220:         * allocate kva.
                    221:         */
1.80      yamt      222:
1.109     yamt      223:        lva = uvm_km_alloc(kernel_map, len, 0, UVM_KMF_VAONLY | UVM_KMF_WAITVA);
1.95      yamt      224:        if (lva == 0) {
                    225:                sokvaunreserve(len);
1.80      yamt      226:                return (0);
1.95      yamt      227:        }
1.80      yamt      228:
                    229:        return lva;
                    230: }
                    231:
1.93      yamt      232: /*
                    233:  * sokvafree: free kva for loan.
                    234:  */
                    235:
1.80      yamt      236: void
                    237: sokvafree(vaddr_t sva, vsize_t len)
                    238: {
1.93      yamt      239:
                    240:        /*
                    241:         * free kva.
                    242:         */
1.80      yamt      243:
1.109     yamt      244:        uvm_km_free(kernel_map, sva, len, UVM_KMF_VAONLY);
1.93      yamt      245:
                    246:        /*
                    247:         * unreserve kva.
                    248:         */
                    249:
1.95      yamt      250:        sokvaunreserve(len);
1.80      yamt      251: }
                    252:
1.64      thorpej   253: static void
1.134     christos  254: sodoloanfree(struct vm_page **pgs, void *buf, size_t size)
1.64      thorpej   255: {
                    256:        vaddr_t va, sva, eva;
                    257:        vsize_t len;
                    258:        paddr_t pa;
                    259:        int i, npgs;
                    260:
                    261:        eva = round_page((vaddr_t) buf + size);
                    262:        sva = trunc_page((vaddr_t) buf);
                    263:        len = eva - sva;
                    264:        npgs = len >> PAGE_SHIFT;
                    265:
1.79      thorpej   266:        if (__predict_false(pgs == NULL)) {
                    267:                pgs = alloca(npgs * sizeof(*pgs));
1.64      thorpej   268:
1.79      thorpej   269:                for (i = 0, va = sva; va < eva; i++, va += PAGE_SIZE) {
1.133     thorpej   270:                        if (pmap_extract(pmap_kernel(), va, &pa) == false)
1.79      thorpej   271:                                panic("sodoloanfree: va 0x%lx not mapped", va);
                    272:                        pgs[i] = PHYS_TO_VM_PAGE(pa);
                    273:                }
1.64      thorpej   274:        }
                    275:
                    276:        pmap_kremove(sva, len);
                    277:        pmap_update(pmap_kernel());
                    278:        uvm_unloan(pgs, npgs, UVM_LOAN_TOPAGE);
1.80      yamt      279:        sokvafree(sva, len);
1.64      thorpej   280: }
                    281:
                    282: static size_t
1.117     yamt      283: sodopendfree()
1.64      thorpej   284: {
1.93      yamt      285:        size_t rv;
1.64      thorpej   286:
1.136     ad        287:        mutex_enter(&so_pendfree_lock);
1.117     yamt      288:        rv = sodopendfreel();
1.136     ad        289:        mutex_exit(&so_pendfree_lock);
1.93      yamt      290:
                    291:        return rv;
                    292: }
                    293:
                    294: /*
                    295:  * sodopendfreel: free mbufs on "pendfree" list.
1.136     ad        296:  * unlock and relock so_pendfree_lock when freeing mbufs.
1.93      yamt      297:  *
1.136     ad        298:  * => called with so_pendfree_lock held.
1.93      yamt      299:  */
                    300:
                    301: static size_t
1.117     yamt      302: sodopendfreel()
1.93      yamt      303: {
1.137     ad        304:        struct mbuf *m, *next;
1.93      yamt      305:        size_t rv = 0;
1.137     ad        306:        int s;
1.93      yamt      307:
1.136     ad        308:        KASSERT(mutex_owned(&so_pendfree_lock));
1.64      thorpej   309:
1.137     ad        310:        while (so_pendfree != NULL) {
1.64      thorpej   311:                m = so_pendfree;
1.93      yamt      312:                so_pendfree = NULL;
1.136     ad        313:                mutex_exit(&so_pendfree_lock);
1.93      yamt      314:
                    315:                for (; m != NULL; m = next) {
                    316:                        next = m->m_next;
                    317:
                    318:                        rv += m->m_ext.ext_size;
                    319:                        sodoloanfree((m->m_flags & M_EXT_PAGES) ?
                    320:                            m->m_ext.ext_pgs : NULL, m->m_ext.ext_buf,
                    321:                            m->m_ext.ext_size);
1.137     ad        322:                        s = splvm();
1.93      yamt      323:                        pool_cache_put(&mbpool_cache, m);
1.137     ad        324:                        splx(s);
1.93      yamt      325:                }
1.64      thorpej   326:
1.136     ad        327:                mutex_enter(&so_pendfree_lock);
1.64      thorpej   328:        }
                    329:
                    330:        return (rv);
                    331: }
                    332:
1.80      yamt      333: void
1.134     christos  334: soloanfree(struct mbuf *m, void *buf, size_t size, void *arg)
1.64      thorpej   335: {
                    336:
                    337:        if (m == NULL) {
1.93      yamt      338:
                    339:                /*
                    340:                 * called from MEXTREMOVE.
                    341:                 */
                    342:
1.79      thorpej   343:                sodoloanfree(NULL, buf, size);
1.64      thorpej   344:                return;
                    345:        }
                    346:
1.93      yamt      347:        /*
                    348:         * postpone freeing mbuf.
                    349:         *
                    350:         * we can't do it in interrupt context
                    351:         * because we need to put kva back to kernel_map.
                    352:         */
                    353:
1.136     ad        354:        mutex_enter(&so_pendfree_lock);
1.92      yamt      355:        m->m_next = so_pendfree;
                    356:        so_pendfree = m;
1.136     ad        357:        cv_broadcast(&socurkva_cv);
                    358:        mutex_exit(&so_pendfree_lock);
1.64      thorpej   359: }
                    360:
                    361: static long
                    362: sosend_loan(struct socket *so, struct uio *uio, struct mbuf *m, long space)
                    363: {
                    364:        struct iovec *iov = uio->uio_iov;
                    365:        vaddr_t sva, eva;
                    366:        vsize_t len;
                    367:        vaddr_t lva, va;
1.80      yamt      368:        int npgs, i, error;
1.64      thorpej   369:
1.116     yamt      370:        if (VMSPACE_IS_KERNEL_P(uio->uio_vmspace))
1.64      thorpej   371:                return (0);
                    372:
                    373:        if (iov->iov_len < (size_t) space)
                    374:                space = iov->iov_len;
                    375:        if (space > SOCK_LOAN_CHUNK)
                    376:                space = SOCK_LOAN_CHUNK;
                    377:
                    378:        eva = round_page((vaddr_t) iov->iov_base + space);
                    379:        sva = trunc_page((vaddr_t) iov->iov_base);
                    380:        len = eva - sva;
                    381:        npgs = len >> PAGE_SHIFT;
                    382:
1.79      thorpej   383:        /* XXX KDASSERT */
                    384:        KASSERT(npgs <= M_EXT_MAXPAGES);
                    385:
1.80      yamt      386:        lva = sokvaalloc(len, so);
1.64      thorpej   387:        if (lva == 0)
1.80      yamt      388:                return 0;
1.64      thorpej   389:
1.116     yamt      390:        error = uvm_loan(&uio->uio_vmspace->vm_map, sva, len,
1.79      thorpej   391:            m->m_ext.ext_pgs, UVM_LOAN_TOPAGE);
1.64      thorpej   392:        if (error) {
1.80      yamt      393:                sokvafree(lva, len);
1.64      thorpej   394:                return (0);
                    395:        }
                    396:
                    397:        for (i = 0, va = lva; i < npgs; i++, va += PAGE_SIZE)
1.79      thorpej   398:                pmap_kenter_pa(va, VM_PAGE_TO_PHYS(m->m_ext.ext_pgs[i]),
                    399:                    VM_PROT_READ);
1.64      thorpej   400:        pmap_update(pmap_kernel());
                    401:
                    402:        lva += (vaddr_t) iov->iov_base & PAGE_MASK;
                    403:
1.134     christos  404:        MEXTADD(m, (void *) lva, space, M_MBUF, soloanfree, so);
1.79      thorpej   405:        m->m_flags |= M_EXT_PAGES | M_EXT_ROMAP;
1.64      thorpej   406:
                    407:        uio->uio_resid -= space;
                    408:        /* uio_offset not updated, not set/used for write(2) */
1.134     christos  409:        uio->uio_iov->iov_base = (char *)uio->uio_iov->iov_base + space;
1.64      thorpej   410:        uio->uio_iov->iov_len -= space;
                    411:        if (uio->uio_iov->iov_len == 0) {
                    412:                uio->uio_iov++;
                    413:                uio->uio_iovcnt--;
                    414:        }
                    415:
                    416:        return (space);
                    417: }
                    418:
1.119     yamt      419: static int
1.129     yamt      420: sokva_reclaim_callback(struct callback_entry *ce, void *obj, void *arg)
1.119     yamt      421: {
                    422:
                    423:        KASSERT(ce == &sokva_reclaimerentry);
                    424:        KASSERT(obj == NULL);
                    425:
                    426:        sodopendfree();
                    427:        if (!vm_map_starved_p(kernel_map)) {
                    428:                return CALLBACK_CHAIN_ABORT;
                    429:        }
                    430:        return CALLBACK_CHAIN_CONTINUE;
                    431: }
                    432:
1.142     dyoung    433: struct mbuf *
                    434: getsombuf(struct socket *so)
                    435: {
                    436:        struct mbuf *m;
                    437:
                    438:        m = m_get(M_WAIT, MT_SONAME);
                    439:        MCLAIM(m, so->so_mowner);
                    440:        return m;
                    441: }
                    442:
                    443: struct mbuf *
                    444: m_intopt(struct socket *so, int val)
                    445: {
                    446:        struct mbuf *m;
                    447:
                    448:        m = getsombuf(so);
                    449:        m->m_len = sizeof(int);
                    450:        *mtod(m, int *) = val;
                    451:        return m;
                    452: }
                    453:
1.119     yamt      454: void
                    455: soinit(void)
                    456: {
                    457:
1.136     ad        458:        mutex_init(&so_pendfree_lock, MUTEX_DRIVER, IPL_VM);
                    459:        cv_init(&socurkva_cv, "sokva");
                    460:
1.119     yamt      461:        /* Set the initial adjusted socket buffer size. */
                    462:        if (sb_max_set(sb_max))
                    463:                panic("bad initial sb_max value: %lu", sb_max);
                    464:
                    465:        callback_register(&vm_map_to_kernel(kernel_map)->vmk_reclaim_callback,
                    466:            &sokva_reclaimerentry, NULL, sokva_reclaim_callback);
                    467: }
                    468:
1.1       cgd       469: /*
                    470:  * Socket operation routines.
                    471:  * These routines are called by the routines in
                    472:  * sys_socket.c or from a system process, and
                    473:  * implement the semantics of socket operations by
                    474:  * switching out to the protocol specific routines.
                    475:  */
                    476: /*ARGSUSED*/
1.3       andrew    477: int
1.114     christos  478: socreate(int dom, struct socket **aso, int type, int proto, struct lwp *l)
1.1       cgd       479: {
1.99      matt      480:        const struct protosw    *prp;
1.54      lukem     481:        struct socket   *so;
1.115     yamt      482:        uid_t           uid;
1.54      lukem     483:        int             error, s;
1.1       cgd       484:
1.132     elad      485:        error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET,
                    486:            KAUTH_REQ_NETWORK_SOCKET_OPEN, KAUTH_ARG(dom), KAUTH_ARG(type),
                    487:            KAUTH_ARG(proto));
1.140     dyoung    488:        if (error != 0)
                    489:                return error;
1.127     elad      490:
1.1       cgd       491:        if (proto)
                    492:                prp = pffindproto(dom, proto, type);
                    493:        else
                    494:                prp = pffindtype(dom, type);
1.140     dyoung    495:        if (prp == NULL) {
1.120     ginsbach  496:                /* no support for domain */
                    497:                if (pffinddomain(dom) == 0)
1.140     dyoung    498:                        return EAFNOSUPPORT;
1.120     ginsbach  499:                /* no support for socket type */
                    500:                if (proto == 0 && type != 0)
1.140     dyoung    501:                        return EPROTOTYPE;
                    502:                return EPROTONOSUPPORT;
1.120     ginsbach  503:        }
1.140     dyoung    504:        if (prp->pr_usrreq == NULL)
                    505:                return EPROTONOSUPPORT;
1.1       cgd       506:        if (prp->pr_type != type)
1.140     dyoung    507:                return EPROTOTYPE;
1.39      matt      508:        s = splsoftnet();
1.37      thorpej   509:        so = pool_get(&socket_pool, PR_WAITOK);
1.140     dyoung    510:        memset(so, 0, sizeof(*so));
1.31      thorpej   511:        TAILQ_INIT(&so->so_q0);
                    512:        TAILQ_INIT(&so->so_q);
1.1       cgd       513:        so->so_type = type;
                    514:        so->so_proto = prp;
1.33      matt      515:        so->so_send = sosend;
                    516:        so->so_receive = soreceive;
1.78      matt      517: #ifdef MBUFTRACE
                    518:        so->so_rcv.sb_mowner = &prp->pr_domain->dom_mowner;
                    519:        so->so_snd.sb_mowner = &prp->pr_domain->dom_mowner;
                    520:        so->so_mowner = &prp->pr_domain->dom_mowner;
                    521: #endif
1.143     ad        522:        selinit(&so->so_rcv.sb_sel);
                    523:        selinit(&so->so_snd.sb_sel);
1.138     rmind     524:        uid = kauth_cred_geteuid(l->l_cred);
1.115     yamt      525:        so->so_uidinfo = uid_find(uid);
1.140     dyoung    526:        error = (*prp->pr_usrreq)(so, PRU_ATTACH, NULL,
                    527:            (struct mbuf *)(long)proto, NULL, l);
                    528:        if (error != 0) {
1.1       cgd       529:                so->so_state |= SS_NOFDREF;
                    530:                sofree(so);
1.39      matt      531:                splx(s);
1.140     dyoung    532:                return error;
1.1       cgd       533:        }
1.39      matt      534:        splx(s);
1.1       cgd       535:        *aso = so;
1.140     dyoung    536:        return 0;
1.1       cgd       537: }
                    538:
1.142     dyoung    539: /* On success, write file descriptor to fdout and return zero.  On
                    540:  * failure, return non-zero; *fdout will be undefined.
                    541:  */
                    542: int
                    543: fsocreate(int domain, struct socket **sop, int type, int protocol,
                    544:     struct lwp *l, int *fdout)
                    545: {
                    546:        struct filedesc *fdp;
                    547:        struct socket   *so;
                    548:        struct file     *fp;
                    549:        int             fd, error;
                    550:
                    551:        fdp = l->l_proc->p_fd;
                    552:        /* falloc() will use the desciptor for us */
                    553:        if ((error = falloc(l, &fp, &fd)) != 0)
                    554:                return (error);
                    555:        fp->f_flag = FREAD|FWRITE;
                    556:        fp->f_type = DTYPE_SOCKET;
                    557:        fp->f_ops = &socketops;
                    558:        error = socreate(domain, &so, type, protocol, l);
                    559:        if (error != 0) {
                    560:                FILE_UNUSE(fp, l);
                    561:                fdremove(fdp, fd);
                    562:                ffree(fp);
                    563:        } else {
                    564:                if (sop != NULL)
                    565:                        *sop = so;
                    566:                fp->f_data = so;
                    567:                FILE_SET_MATURE(fp);
                    568:                FILE_UNUSE(fp, l);
                    569:                *fdout = fd;
                    570:        }
                    571:        return error;
                    572: }
                    573:
1.3       andrew    574: int
1.114     christos  575: sobind(struct socket *so, struct mbuf *nam, struct lwp *l)
1.1       cgd       576: {
1.54      lukem     577:        int     s, error;
1.1       cgd       578:
1.54      lukem     579:        s = splsoftnet();
1.140     dyoung    580:        error = (*so->so_proto->pr_usrreq)(so, PRU_BIND, NULL, nam, NULL, l);
1.1       cgd       581:        splx(s);
1.140     dyoung    582:        return error;
1.1       cgd       583: }
                    584:
1.3       andrew    585: int
1.54      lukem     586: solisten(struct socket *so, int backlog)
1.1       cgd       587: {
1.54      lukem     588:        int     s, error;
1.1       cgd       589:
1.54      lukem     590:        s = splsoftnet();
1.140     dyoung    591:        error = (*so->so_proto->pr_usrreq)(so, PRU_LISTEN, NULL,
                    592:            NULL, NULL, NULL);
                    593:        if (error != 0) {
1.1       cgd       594:                splx(s);
1.140     dyoung    595:                return error;
1.1       cgd       596:        }
1.63      matt      597:        if (TAILQ_EMPTY(&so->so_q))
1.1       cgd       598:                so->so_options |= SO_ACCEPTCONN;
                    599:        if (backlog < 0)
                    600:                backlog = 0;
1.49      jonathan  601:        so->so_qlimit = min(backlog, somaxconn);
1.1       cgd       602:        splx(s);
1.140     dyoung    603:        return 0;
1.1       cgd       604: }
                    605:
1.21      christos  606: void
1.54      lukem     607: sofree(struct socket *so)
1.1       cgd       608: {
                    609:
1.43      mycroft   610:        if (so->so_pcb || (so->so_state & SS_NOFDREF) == 0)
1.1       cgd       611:                return;
1.43      mycroft   612:        if (so->so_head) {
                    613:                /*
                    614:                 * We must not decommission a socket that's on the accept(2)
                    615:                 * queue.  If we do, then accept(2) may hang after select(2)
                    616:                 * indicated that the listening socket was ready.
                    617:                 */
                    618:                if (!soqremque(so, 0))
                    619:                        return;
                    620:        }
1.98      christos  621:        if (so->so_rcv.sb_hiwat)
1.110     christos  622:                (void)chgsbsize(so->so_uidinfo, &so->so_rcv.sb_hiwat, 0,
1.98      christos  623:                    RLIM_INFINITY);
                    624:        if (so->so_snd.sb_hiwat)
1.110     christos  625:                (void)chgsbsize(so->so_uidinfo, &so->so_snd.sb_hiwat, 0,
1.98      christos  626:                    RLIM_INFINITY);
                    627:        sbrelease(&so->so_snd, so);
1.1       cgd       628:        sorflush(so);
1.143     ad        629:        seldestroy(&so->so_rcv.sb_sel);
                    630:        seldestroy(&so->so_snd.sb_sel);
1.37      thorpej   631:        pool_put(&socket_pool, so);
1.1       cgd       632: }
                    633:
                    634: /*
                    635:  * Close a socket on last file table reference removal.
                    636:  * Initiate disconnect if connected.
                    637:  * Free socket when disconnect complete.
                    638:  */
1.3       andrew    639: int
1.54      lukem     640: soclose(struct socket *so)
1.1       cgd       641: {
1.54      lukem     642:        struct socket   *so2;
                    643:        int             s, error;
1.1       cgd       644:
1.54      lukem     645:        error = 0;
                    646:        s = splsoftnet();               /* conservative */
1.1       cgd       647:        if (so->so_options & SO_ACCEPTCONN) {
1.63      matt      648:                while ((so2 = TAILQ_FIRST(&so->so_q0)) != 0) {
1.42      mycroft   649:                        (void) soqremque(so2, 0);
1.41      mycroft   650:                        (void) soabort(so2);
                    651:                }
1.63      matt      652:                while ((so2 = TAILQ_FIRST(&so->so_q)) != 0) {
1.42      mycroft   653:                        (void) soqremque(so2, 1);
1.41      mycroft   654:                        (void) soabort(so2);
                    655:                }
1.1       cgd       656:        }
                    657:        if (so->so_pcb == 0)
                    658:                goto discard;
                    659:        if (so->so_state & SS_ISCONNECTED) {
                    660:                if ((so->so_state & SS_ISDISCONNECTING) == 0) {
                    661:                        error = sodisconnect(so);
                    662:                        if (error)
                    663:                                goto drop;
                    664:                }
                    665:                if (so->so_options & SO_LINGER) {
                    666:                        if ((so->so_state & SS_ISDISCONNECTING) &&
                    667:                            (so->so_state & SS_NBIO))
                    668:                                goto drop;
1.21      christos  669:                        while (so->so_state & SS_ISCONNECTED) {
1.134     christos  670:                                error = tsleep((void *)&so->so_timeo,
1.21      christos  671:                                               PSOCK | PCATCH, netcls,
1.30      thorpej   672:                                               so->so_linger * hz);
1.21      christos  673:                                if (error)
1.1       cgd       674:                                        break;
1.21      christos  675:                        }
1.1       cgd       676:                }
                    677:        }
1.54      lukem     678:  drop:
1.1       cgd       679:        if (so->so_pcb) {
1.22      mycroft   680:                int error2 = (*so->so_proto->pr_usrreq)(so, PRU_DETACH,
1.140     dyoung    681:                    NULL, NULL, NULL, NULL);
1.1       cgd       682:                if (error == 0)
                    683:                        error = error2;
                    684:        }
1.54      lukem     685:  discard:
1.1       cgd       686:        if (so->so_state & SS_NOFDREF)
                    687:                panic("soclose: NOFDREF");
                    688:        so->so_state |= SS_NOFDREF;
                    689:        sofree(so);
                    690:        splx(s);
                    691:        return (error);
                    692: }
                    693:
                    694: /*
1.20      mycroft   695:  * Must be called at splsoftnet...
1.1       cgd       696:  */
1.3       andrew    697: int
1.54      lukem     698: soabort(struct socket *so)
1.1       cgd       699: {
1.139     yamt      700:        int error;
1.1       cgd       701:
1.139     yamt      702:        KASSERT(so->so_head == NULL);
1.140     dyoung    703:        error = (*so->so_proto->pr_usrreq)(so, PRU_ABORT, NULL,
                    704:            NULL, NULL, NULL);
1.139     yamt      705:        if (error) {
                    706:                sofree(so);
                    707:        }
                    708:        return error;
1.1       cgd       709: }
                    710:
1.3       andrew    711: int
1.54      lukem     712: soaccept(struct socket *so, struct mbuf *nam)
1.1       cgd       713: {
1.54      lukem     714:        int     s, error;
1.1       cgd       715:
1.54      lukem     716:        error = 0;
                    717:        s = splsoftnet();
1.1       cgd       718:        if ((so->so_state & SS_NOFDREF) == 0)
                    719:                panic("soaccept: !NOFDREF");
                    720:        so->so_state &= ~SS_NOFDREF;
1.55      thorpej   721:        if ((so->so_state & SS_ISDISCONNECTED) == 0 ||
                    722:            (so->so_proto->pr_flags & PR_ABRTACPTDIS) == 0)
1.41      mycroft   723:                error = (*so->so_proto->pr_usrreq)(so, PRU_ACCEPT,
1.140     dyoung    724:                    NULL, nam, NULL, NULL);
1.41      mycroft   725:        else
1.53      itojun    726:                error = ECONNABORTED;
1.52      itojun    727:
1.1       cgd       728:        splx(s);
                    729:        return (error);
                    730: }
                    731:
1.3       andrew    732: int
1.114     christos  733: soconnect(struct socket *so, struct mbuf *nam, struct lwp *l)
1.1       cgd       734: {
1.54      lukem     735:        int             s, error;
1.1       cgd       736:
                    737:        if (so->so_options & SO_ACCEPTCONN)
                    738:                return (EOPNOTSUPP);
1.20      mycroft   739:        s = splsoftnet();
1.1       cgd       740:        /*
                    741:         * If protocol is connection-based, can only connect once.
                    742:         * Otherwise, if connected, try to disconnect first.
                    743:         * This allows user to disconnect by connecting to, e.g.,
                    744:         * a null address.
                    745:         */
                    746:        if (so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING) &&
                    747:            ((so->so_proto->pr_flags & PR_CONNREQUIRED) ||
                    748:            (error = sodisconnect(so))))
                    749:                error = EISCONN;
                    750:        else
                    751:                error = (*so->so_proto->pr_usrreq)(so, PRU_CONNECT,
1.140     dyoung    752:                    NULL, nam, NULL, l);
1.1       cgd       753:        splx(s);
                    754:        return (error);
                    755: }
                    756:
1.3       andrew    757: int
1.54      lukem     758: soconnect2(struct socket *so1, struct socket *so2)
1.1       cgd       759: {
1.54      lukem     760:        int     s, error;
1.1       cgd       761:
1.54      lukem     762:        s = splsoftnet();
1.22      mycroft   763:        error = (*so1->so_proto->pr_usrreq)(so1, PRU_CONNECT2,
1.140     dyoung    764:            NULL, (struct mbuf *)so2, NULL, NULL);
1.1       cgd       765:        splx(s);
                    766:        return (error);
                    767: }
                    768:
1.3       andrew    769: int
1.54      lukem     770: sodisconnect(struct socket *so)
1.1       cgd       771: {
1.54      lukem     772:        int     s, error;
1.1       cgd       773:
1.54      lukem     774:        s = splsoftnet();
1.1       cgd       775:        if ((so->so_state & SS_ISCONNECTED) == 0) {
                    776:                error = ENOTCONN;
                    777:                goto bad;
                    778:        }
                    779:        if (so->so_state & SS_ISDISCONNECTING) {
                    780:                error = EALREADY;
                    781:                goto bad;
                    782:        }
1.22      mycroft   783:        error = (*so->so_proto->pr_usrreq)(so, PRU_DISCONNECT,
1.140     dyoung    784:            NULL, NULL, NULL, NULL);
1.54      lukem     785:  bad:
1.1       cgd       786:        splx(s);
1.117     yamt      787:        sodopendfree();
1.1       cgd       788:        return (error);
                    789: }
                    790:
1.15      mycroft   791: #define        SBLOCKWAIT(f)   (((f) & MSG_DONTWAIT) ? M_NOWAIT : M_WAITOK)
1.1       cgd       792: /*
                    793:  * Send on a socket.
                    794:  * If send must go all at once and message is larger than
                    795:  * send buffering, then hard error.
                    796:  * Lock against other senders.
                    797:  * If must go all at once and not enough room now, then
                    798:  * inform user that this would block and do nothing.
                    799:  * Otherwise, if nonblocking, send as much as possible.
                    800:  * The data to be sent is described by "uio" if nonzero,
                    801:  * otherwise by the mbuf chain "top" (which must be null
                    802:  * if uio is not).  Data provided in mbuf chain must be small
                    803:  * enough to send all at once.
                    804:  *
                    805:  * Returns nonzero on error, timeout or signal; callers
                    806:  * must check for short counts if EINTR/ERESTART are returned.
                    807:  * Data and control buffers are freed on return.
                    808:  */
1.3       andrew    809: int
1.54      lukem     810: sosend(struct socket *so, struct mbuf *addr, struct uio *uio, struct mbuf *top,
1.114     christos  811:        struct mbuf *control, int flags, struct lwp *l)
1.1       cgd       812: {
1.54      lukem     813:        struct mbuf     **mp, *m;
1.114     christos  814:        struct proc     *p;
1.58      jdolecek  815:        long            space, len, resid, clen, mlen;
                    816:        int             error, s, dontroute, atomic;
1.54      lukem     817:
1.114     christos  818:        p = l->l_proc;
1.117     yamt      819:        sodopendfree();
1.64      thorpej   820:
1.54      lukem     821:        clen = 0;
                    822:        atomic = sosendallatonce(so) || top;
1.1       cgd       823:        if (uio)
                    824:                resid = uio->uio_resid;
                    825:        else
                    826:                resid = top->m_pkthdr.len;
1.7       cgd       827:        /*
                    828:         * In theory resid should be unsigned.
                    829:         * However, space must be signed, as it might be less than 0
                    830:         * if we over-committed, and we must use a signed comparison
                    831:         * of space and resid.  On the other hand, a negative resid
                    832:         * causes us to loop sending 0-length segments to the protocol.
                    833:         */
1.29      mycroft   834:        if (resid < 0) {
                    835:                error = EINVAL;
                    836:                goto out;
                    837:        }
1.1       cgd       838:        dontroute =
                    839:            (flags & MSG_DONTROUTE) && (so->so_options & SO_DONTROUTE) == 0 &&
                    840:            (so->so_proto->pr_flags & PR_ATOMIC);
1.102     jonathan  841:        if (p)
                    842:                p->p_stats->p_ru.ru_msgsnd++;
1.1       cgd       843:        if (control)
                    844:                clen = control->m_len;
                    845: #define        snderr(errno)   { error = errno; splx(s); goto release; }
                    846:
1.54      lukem     847:  restart:
1.21      christos  848:        if ((error = sblock(&so->so_snd, SBLOCKWAIT(flags))) != 0)
1.1       cgd       849:                goto out;
                    850:        do {
1.20      mycroft   851:                s = splsoftnet();
1.1       cgd       852:                if (so->so_state & SS_CANTSENDMORE)
                    853:                        snderr(EPIPE);
1.48      thorpej   854:                if (so->so_error) {
                    855:                        error = so->so_error;
                    856:                        so->so_error = 0;
                    857:                        splx(s);
                    858:                        goto release;
                    859:                }
1.1       cgd       860:                if ((so->so_state & SS_ISCONNECTED) == 0) {
                    861:                        if (so->so_proto->pr_flags & PR_CONNREQUIRED) {
                    862:                                if ((so->so_state & SS_ISCONFIRMING) == 0 &&
                    863:                                    !(resid == 0 && clen != 0))
                    864:                                        snderr(ENOTCONN);
                    865:                        } else if (addr == 0)
                    866:                                snderr(EDESTADDRREQ);
                    867:                }
                    868:                space = sbspace(&so->so_snd);
                    869:                if (flags & MSG_OOB)
                    870:                        space += 1024;
1.21      christos  871:                if ((atomic && resid > so->so_snd.sb_hiwat) ||
1.11      mycroft   872:                    clen > so->so_snd.sb_hiwat)
                    873:                        snderr(EMSGSIZE);
1.96      mycroft   874:                if (space < resid + clen &&
1.1       cgd       875:                    (atomic || space < so->so_snd.sb_lowat || space < clen)) {
                    876:                        if (so->so_state & SS_NBIO)
                    877:                                snderr(EWOULDBLOCK);
                    878:                        sbunlock(&so->so_snd);
                    879:                        error = sbwait(&so->so_snd);
                    880:                        splx(s);
                    881:                        if (error)
                    882:                                goto out;
                    883:                        goto restart;
                    884:                }
                    885:                splx(s);
                    886:                mp = &top;
                    887:                space -= clen;
                    888:                do {
1.45      tv        889:                        if (uio == NULL) {
                    890:                                /*
                    891:                                 * Data is prepackaged in "top".
                    892:                                 */
                    893:                                resid = 0;
                    894:                                if (flags & MSG_EOR)
                    895:                                        top->m_flags |= M_EOR;
                    896:                        } else do {
1.144   ! dyoung    897:                                if (top == NULL) {
1.78      matt      898:                                        m = m_gethdr(M_WAIT, MT_DATA);
1.45      tv        899:                                        mlen = MHLEN;
                    900:                                        m->m_pkthdr.len = 0;
1.140     dyoung    901:                                        m->m_pkthdr.rcvif = NULL;
1.45      tv        902:                                } else {
1.78      matt      903:                                        m = m_get(M_WAIT, MT_DATA);
1.45      tv        904:                                        mlen = MLEN;
                    905:                                }
1.78      matt      906:                                MCLAIM(m, so->so_snd.sb_mowner);
1.121     yamt      907:                                if (sock_loan_thresh >= 0 &&
                    908:                                    uio->uio_iov->iov_len >= sock_loan_thresh &&
                    909:                                    space >= sock_loan_thresh &&
1.64      thorpej   910:                                    (len = sosend_loan(so, uio, m,
                    911:                                                       space)) != 0) {
                    912:                                        SOSEND_COUNTER_INCR(&sosend_loan_big);
                    913:                                        space -= len;
                    914:                                        goto have_data;
                    915:                                }
1.45      tv        916:                                if (resid >= MINCLSIZE && space >= MCLBYTES) {
1.64      thorpej   917:                                        SOSEND_COUNTER_INCR(&sosend_copy_big);
1.78      matt      918:                                        m_clget(m, M_WAIT);
1.45      tv        919:                                        if ((m->m_flags & M_EXT) == 0)
                    920:                                                goto nopages;
                    921:                                        mlen = MCLBYTES;
                    922:                                        if (atomic && top == 0) {
1.58      jdolecek  923:                                                len = lmin(MCLBYTES - max_hdr,
1.54      lukem     924:                                                    resid);
1.45      tv        925:                                                m->m_data += max_hdr;
                    926:                                        } else
1.58      jdolecek  927:                                                len = lmin(MCLBYTES, resid);
1.45      tv        928:                                        space -= len;
                    929:                                } else {
1.64      thorpej   930:  nopages:
                    931:                                        SOSEND_COUNTER_INCR(&sosend_copy_small);
1.58      jdolecek  932:                                        len = lmin(lmin(mlen, resid), space);
1.45      tv        933:                                        space -= len;
                    934:                                        /*
                    935:                                         * For datagram protocols, leave room
                    936:                                         * for protocol headers in first mbuf.
                    937:                                         */
                    938:                                        if (atomic && top == 0 && len < mlen)
                    939:                                                MH_ALIGN(m, len);
                    940:                                }
1.144   ! dyoung    941:                                error = uiomove(mtod(m, void *), (int)len, uio);
1.64      thorpej   942:  have_data:
1.45      tv        943:                                resid = uio->uio_resid;
                    944:                                m->m_len = len;
                    945:                                *mp = m;
                    946:                                top->m_pkthdr.len += len;
1.144   ! dyoung    947:                                if (error != 0)
1.45      tv        948:                                        goto release;
                    949:                                mp = &m->m_next;
                    950:                                if (resid <= 0) {
                    951:                                        if (flags & MSG_EOR)
                    952:                                                top->m_flags |= M_EOR;
                    953:                                        break;
                    954:                                }
                    955:                        } while (space > 0 && atomic);
1.108     perry     956:
1.46      sommerfe  957:                        s = splsoftnet();
                    958:
                    959:                        if (so->so_state & SS_CANTSENDMORE)
                    960:                                snderr(EPIPE);
1.45      tv        961:
                    962:                        if (dontroute)
                    963:                                so->so_options |= SO_DONTROUTE;
                    964:                        if (resid > 0)
                    965:                                so->so_state |= SS_MORETOCOME;
1.46      sommerfe  966:                        error = (*so->so_proto->pr_usrreq)(so,
                    967:                            (flags & MSG_OOB) ? PRU_SENDOOB : PRU_SEND,
1.114     christos  968:                            top, addr, control, curlwp);        /* XXX */
1.45      tv        969:                        if (dontroute)
                    970:                                so->so_options &= ~SO_DONTROUTE;
                    971:                        if (resid > 0)
                    972:                                so->so_state &= ~SS_MORETOCOME;
1.46      sommerfe  973:                        splx(s);
                    974:
1.45      tv        975:                        clen = 0;
1.144   ! dyoung    976:                        control = NULL;
        !           977:                        top = NULL;
1.45      tv        978:                        mp = &top;
1.144   ! dyoung    979:                        if (error != 0)
1.1       cgd       980:                                goto release;
                    981:                } while (resid && space > 0);
                    982:        } while (resid);
                    983:
1.54      lukem     984:  release:
1.1       cgd       985:        sbunlock(&so->so_snd);
1.54      lukem     986:  out:
1.1       cgd       987:        if (top)
                    988:                m_freem(top);
                    989:        if (control)
                    990:                m_freem(control);
                    991:        return (error);
                    992: }
                    993:
                    994: /*
                    995:  * Implement receive operations on a socket.
                    996:  * We depend on the way that records are added to the sockbuf
                    997:  * by sbappend*.  In particular, each record (mbufs linked through m_next)
                    998:  * must begin with an address if the protocol so specifies,
                    999:  * followed by an optional mbuf or mbufs containing ancillary data,
                   1000:  * and then zero or more mbufs of data.
                   1001:  * In order to avoid blocking network interrupts for the entire time here,
                   1002:  * we splx() while doing the actual copy to user space.
                   1003:  * Although the sockbuf is locked, new data may still be appended,
                   1004:  * and thus we must maintain consistency of the sockbuf during that time.
                   1005:  *
                   1006:  * The caller may receive the data as a single mbuf chain by supplying
                   1007:  * an mbuf **mp0 for use in returning the chain.  The uio is then used
                   1008:  * only for the count in uio_resid.
                   1009:  */
1.3       andrew   1010: int
1.54      lukem    1011: soreceive(struct socket *so, struct mbuf **paddr, struct uio *uio,
                   1012:        struct mbuf **mp0, struct mbuf **controlp, int *flagsp)
1.1       cgd      1013: {
1.116     yamt     1014:        struct lwp *l = curlwp;
1.54      lukem    1015:        struct mbuf     *m, **mp;
                   1016:        int             flags, len, error, s, offset, moff, type, orig_resid;
1.99      matt     1017:        const struct protosw    *pr;
1.54      lukem    1018:        struct mbuf     *nextrecord;
1.67      he       1019:        int             mbuf_removed = 0;
1.64      thorpej  1020:
1.54      lukem    1021:        pr = so->so_proto;
1.1       cgd      1022:        mp = mp0;
1.54      lukem    1023:        type = 0;
                   1024:        orig_resid = uio->uio_resid;
1.102     jonathan 1025:
1.144   ! dyoung   1026:        if (paddr != NULL)
        !          1027:                *paddr = NULL;
        !          1028:        if (controlp != NULL)
        !          1029:                *controlp = NULL;
        !          1030:        if (flagsp != NULL)
1.1       cgd      1031:                flags = *flagsp &~ MSG_EOR;
                   1032:        else
                   1033:                flags = 0;
1.66      enami    1034:
                   1035:        if ((flags & MSG_DONTWAIT) == 0)
1.117     yamt     1036:                sodopendfree();
1.66      enami    1037:
1.1       cgd      1038:        if (flags & MSG_OOB) {
                   1039:                m = m_get(M_WAIT, MT_DATA);
1.17      cgd      1040:                error = (*pr->pr_usrreq)(so, PRU_RCVOOB, m,
1.140     dyoung   1041:                    (struct mbuf *)(long)(flags & MSG_PEEK), NULL, l);
1.1       cgd      1042:                if (error)
                   1043:                        goto bad;
                   1044:                do {
1.134     christos 1045:                        error = uiomove(mtod(m, void *),
1.1       cgd      1046:                            (int) min(uio->uio_resid, m->m_len), uio);
                   1047:                        m = m_free(m);
1.144   ! dyoung   1048:                } while (uio->uio_resid > 0 && error == 0 && m);
1.54      lukem    1049:  bad:
1.144   ! dyoung   1050:                if (m != NULL)
1.1       cgd      1051:                        m_freem(m);
1.144   ! dyoung   1052:                return error;
1.1       cgd      1053:        }
1.144   ! dyoung   1054:        if (mp != NULL)
1.140     dyoung   1055:                *mp = NULL;
1.1       cgd      1056:        if (so->so_state & SS_ISCONFIRMING && uio->uio_resid)
1.140     dyoung   1057:                (*pr->pr_usrreq)(so, PRU_RCVD, NULL, NULL, NULL, l);
1.1       cgd      1058:
1.54      lukem    1059:  restart:
1.21      christos 1060:        if ((error = sblock(&so->so_rcv, SBLOCKWAIT(flags))) != 0)
1.144   ! dyoung   1061:                return error;
1.20      mycroft  1062:        s = splsoftnet();
1.1       cgd      1063:
                   1064:        m = so->so_rcv.sb_mb;
                   1065:        /*
                   1066:         * If we have less data than requested, block awaiting more
                   1067:         * (subject to any timeout) if:
1.15      mycroft  1068:         *   1. the current count is less than the low water mark,
1.1       cgd      1069:         *   2. MSG_WAITALL is set, and it is possible to do the entire
1.15      mycroft  1070:         *      receive operation at once if we block (resid <= hiwat), or
                   1071:         *   3. MSG_DONTWAIT is not set.
1.1       cgd      1072:         * If MSG_WAITALL is set but resid is larger than the receive buffer,
                   1073:         * we have to do the receive in sections, and thus risk returning
                   1074:         * a short count if a timeout or signal occurs after we start.
                   1075:         */
1.144   ! dyoung   1076:        if (m == NULL ||
        !          1077:            ((flags & MSG_DONTWAIT) == 0 &&
        !          1078:             so->so_rcv.sb_cc < uio->uio_resid &&
        !          1079:             (so->so_rcv.sb_cc < so->so_rcv.sb_lowat ||
        !          1080:              ((flags & MSG_WAITALL) &&
        !          1081:               uio->uio_resid <= so->so_rcv.sb_hiwat)) &&
        !          1082:             m->m_nextpkt == NULL &&
        !          1083:             (pr->pr_flags & PR_ATOMIC) == 0)) {
1.1       cgd      1084: #ifdef DIAGNOSTIC
1.144   ! dyoung   1085:                if (m == NULL && so->so_rcv.sb_cc)
1.1       cgd      1086:                        panic("receive 1");
                   1087: #endif
                   1088:                if (so->so_error) {
1.144   ! dyoung   1089:                        if (m != NULL)
1.15      mycroft  1090:                                goto dontblock;
1.1       cgd      1091:                        error = so->so_error;
                   1092:                        if ((flags & MSG_PEEK) == 0)
                   1093:                                so->so_error = 0;
                   1094:                        goto release;
                   1095:                }
                   1096:                if (so->so_state & SS_CANTRCVMORE) {
1.144   ! dyoung   1097:                        if (m != NULL)
1.15      mycroft  1098:                                goto dontblock;
1.1       cgd      1099:                        else
                   1100:                                goto release;
                   1101:                }
1.144   ! dyoung   1102:                for (; m != NULL; m = m->m_next)
1.1       cgd      1103:                        if (m->m_type == MT_OOBDATA  || (m->m_flags & M_EOR)) {
                   1104:                                m = so->so_rcv.sb_mb;
                   1105:                                goto dontblock;
                   1106:                        }
                   1107:                if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) == 0 &&
                   1108:                    (so->so_proto->pr_flags & PR_CONNREQUIRED)) {
                   1109:                        error = ENOTCONN;
                   1110:                        goto release;
                   1111:                }
                   1112:                if (uio->uio_resid == 0)
                   1113:                        goto release;
1.15      mycroft  1114:                if ((so->so_state & SS_NBIO) || (flags & MSG_DONTWAIT)) {
1.1       cgd      1115:                        error = EWOULDBLOCK;
                   1116:                        goto release;
                   1117:                }
1.69      thorpej  1118:                SBLASTRECORDCHK(&so->so_rcv, "soreceive sbwait 1");
                   1119:                SBLASTMBUFCHK(&so->so_rcv, "soreceive sbwait 1");
1.1       cgd      1120:                sbunlock(&so->so_rcv);
                   1121:                error = sbwait(&so->so_rcv);
                   1122:                splx(s);
1.144   ! dyoung   1123:                if (error != 0)
        !          1124:                        return error;
1.1       cgd      1125:                goto restart;
                   1126:        }
1.54      lukem    1127:  dontblock:
1.69      thorpej  1128:        /*
                   1129:         * On entry here, m points to the first record of the socket buffer.
                   1130:         * While we process the initial mbufs containing address and control
                   1131:         * info, we save a copy of m->m_nextpkt into nextrecord.
                   1132:         */
1.144   ! dyoung   1133:        if (l != NULL)
1.114     christos 1134:                l->l_proc->p_stats->p_ru.ru_msgrcv++;
1.69      thorpej  1135:        KASSERT(m == so->so_rcv.sb_mb);
                   1136:        SBLASTRECORDCHK(&so->so_rcv, "soreceive 1");
                   1137:        SBLASTMBUFCHK(&so->so_rcv, "soreceive 1");
1.1       cgd      1138:        nextrecord = m->m_nextpkt;
                   1139:        if (pr->pr_flags & PR_ADDR) {
                   1140: #ifdef DIAGNOSTIC
                   1141:                if (m->m_type != MT_SONAME)
                   1142:                        panic("receive 1a");
                   1143: #endif
1.3       andrew   1144:                orig_resid = 0;
1.1       cgd      1145:                if (flags & MSG_PEEK) {
                   1146:                        if (paddr)
                   1147:                                *paddr = m_copy(m, 0, m->m_len);
                   1148:                        m = m->m_next;
                   1149:                } else {
                   1150:                        sbfree(&so->so_rcv, m);
1.67      he       1151:                        mbuf_removed = 1;
1.144   ! dyoung   1152:                        if (paddr != NULL) {
1.1       cgd      1153:                                *paddr = m;
                   1154:                                so->so_rcv.sb_mb = m->m_next;
1.144   ! dyoung   1155:                                m->m_next = NULL;
1.1       cgd      1156:                                m = so->so_rcv.sb_mb;
                   1157:                        } else {
                   1158:                                MFREE(m, so->so_rcv.sb_mb);
                   1159:                                m = so->so_rcv.sb_mb;
                   1160:                        }
                   1161:                }
                   1162:        }
1.144   ! dyoung   1163:        while (m != NULL && m->m_type == MT_CONTROL && error == 0) {
1.1       cgd      1164:                if (flags & MSG_PEEK) {
1.144   ! dyoung   1165:                        if (controlp != NULL)
1.1       cgd      1166:                                *controlp = m_copy(m, 0, m->m_len);
                   1167:                        m = m->m_next;
                   1168:                } else {
                   1169:                        sbfree(&so->so_rcv, m);
1.67      he       1170:                        mbuf_removed = 1;
1.144   ! dyoung   1171:                        if (controlp != NULL) {
1.102     jonathan 1172:                                struct domain *dom = pr->pr_domain;
1.114     christos 1173:                                if (dom->dom_externalize && l &&
1.1       cgd      1174:                                    mtod(m, struct cmsghdr *)->cmsg_type ==
                   1175:                                    SCM_RIGHTS)
1.114     christos 1176:                                        error = (*dom->dom_externalize)(m, l);
1.1       cgd      1177:                                *controlp = m;
                   1178:                                so->so_rcv.sb_mb = m->m_next;
1.144   ! dyoung   1179:                                m->m_next = NULL;
1.1       cgd      1180:                                m = so->so_rcv.sb_mb;
                   1181:                        } else {
1.106     itojun   1182:                                /*
                   1183:                                 * Dispose of any SCM_RIGHTS message that went
                   1184:                                 * through the read path rather than recv.
                   1185:                                 */
                   1186:                                if (pr->pr_domain->dom_dispose &&
                   1187:                                    mtod(m, struct cmsghdr *)->cmsg_type == SCM_RIGHTS)
                   1188:                                        (*pr->pr_domain->dom_dispose)(m);
1.1       cgd      1189:                                MFREE(m, so->so_rcv.sb_mb);
                   1190:                                m = so->so_rcv.sb_mb;
                   1191:                        }
                   1192:                }
1.144   ! dyoung   1193:                if (controlp != NULL) {
1.3       andrew   1194:                        orig_resid = 0;
1.1       cgd      1195:                        controlp = &(*controlp)->m_next;
1.3       andrew   1196:                }
1.1       cgd      1197:        }
1.69      thorpej  1198:
                   1199:        /*
                   1200:         * If m is non-NULL, we have some data to read.  From now on,
                   1201:         * make sure to keep sb_lastrecord consistent when working on
                   1202:         * the last packet on the chain (nextrecord == NULL) and we
                   1203:         * change m->m_nextpkt.
                   1204:         */
1.144   ! dyoung   1205:        if (m != NULL) {
1.69      thorpej  1206:                if ((flags & MSG_PEEK) == 0) {
1.1       cgd      1207:                        m->m_nextpkt = nextrecord;
1.69      thorpej  1208:                        /*
                   1209:                         * If nextrecord == NULL (this is a single chain),
                   1210:                         * then sb_lastrecord may not be valid here if m
                   1211:                         * was changed earlier.
                   1212:                         */
                   1213:                        if (nextrecord == NULL) {
                   1214:                                KASSERT(so->so_rcv.sb_mb == m);
                   1215:                                so->so_rcv.sb_lastrecord = m;
                   1216:                        }
                   1217:                }
1.1       cgd      1218:                type = m->m_type;
                   1219:                if (type == MT_OOBDATA)
                   1220:                        flags |= MSG_OOB;
1.69      thorpej  1221:        } else {
                   1222:                if ((flags & MSG_PEEK) == 0) {
                   1223:                        KASSERT(so->so_rcv.sb_mb == m);
                   1224:                        so->so_rcv.sb_mb = nextrecord;
1.70      thorpej  1225:                        SB_EMPTY_FIXUP(&so->so_rcv);
1.69      thorpej  1226:                }
1.1       cgd      1227:        }
1.69      thorpej  1228:        SBLASTRECORDCHK(&so->so_rcv, "soreceive 2");
                   1229:        SBLASTMBUFCHK(&so->so_rcv, "soreceive 2");
                   1230:
1.1       cgd      1231:        moff = 0;
                   1232:        offset = 0;
1.144   ! dyoung   1233:        while (m != NULL && uio->uio_resid > 0 && error == 0) {
1.1       cgd      1234:                if (m->m_type == MT_OOBDATA) {
                   1235:                        if (type != MT_OOBDATA)
                   1236:                                break;
                   1237:                } else if (type == MT_OOBDATA)
                   1238:                        break;
                   1239: #ifdef DIAGNOSTIC
                   1240:                else if (m->m_type != MT_DATA && m->m_type != MT_HEADER)
                   1241:                        panic("receive 3");
                   1242: #endif
                   1243:                so->so_state &= ~SS_RCVATMARK;
                   1244:                len = uio->uio_resid;
                   1245:                if (so->so_oobmark && len > so->so_oobmark - offset)
                   1246:                        len = so->so_oobmark - offset;
                   1247:                if (len > m->m_len - moff)
                   1248:                        len = m->m_len - moff;
                   1249:                /*
                   1250:                 * If mp is set, just pass back the mbufs.
                   1251:                 * Otherwise copy them out via the uio, then free.
                   1252:                 * Sockbuf must be consistent here (points to current mbuf,
                   1253:                 * it points to next record) when we drop priority;
                   1254:                 * we must note any additions to the sockbuf when we
                   1255:                 * block interrupts again.
                   1256:                 */
1.144   ! dyoung   1257:                if (mp == NULL) {
1.69      thorpej  1258:                        SBLASTRECORDCHK(&so->so_rcv, "soreceive uiomove");
                   1259:                        SBLASTMBUFCHK(&so->so_rcv, "soreceive uiomove");
1.1       cgd      1260:                        splx(s);
1.134     christos 1261:                        error = uiomove(mtod(m, char *) + moff, (int)len, uio);
1.20      mycroft  1262:                        s = splsoftnet();
1.144   ! dyoung   1263:                        if (error != 0) {
1.67      he       1264:                                /*
                   1265:                                 * If any part of the record has been removed
                   1266:                                 * (such as the MT_SONAME mbuf, which will
                   1267:                                 * happen when PR_ADDR, and thus also
                   1268:                                 * PR_ATOMIC, is set), then drop the entire
                   1269:                                 * record to maintain the atomicity of the
                   1270:                                 * receive operation.
                   1271:                                 *
                   1272:                                 * This avoids a later panic("receive 1a")
                   1273:                                 * when compiled with DIAGNOSTIC.
                   1274:                                 */
                   1275:                                if (m && mbuf_removed
                   1276:                                    && (pr->pr_flags & PR_ATOMIC))
                   1277:                                        (void) sbdroprecord(&so->so_rcv);
                   1278:
1.57      jdolecek 1279:                                goto release;
1.67      he       1280:                        }
1.1       cgd      1281:                } else
                   1282:                        uio->uio_resid -= len;
                   1283:                if (len == m->m_len - moff) {
                   1284:                        if (m->m_flags & M_EOR)
                   1285:                                flags |= MSG_EOR;
                   1286:                        if (flags & MSG_PEEK) {
                   1287:                                m = m->m_next;
                   1288:                                moff = 0;
                   1289:                        } else {
                   1290:                                nextrecord = m->m_nextpkt;
                   1291:                                sbfree(&so->so_rcv, m);
                   1292:                                if (mp) {
                   1293:                                        *mp = m;
                   1294:                                        mp = &m->m_next;
                   1295:                                        so->so_rcv.sb_mb = m = m->m_next;
1.140     dyoung   1296:                                        *mp = NULL;
1.1       cgd      1297:                                } else {
                   1298:                                        MFREE(m, so->so_rcv.sb_mb);
                   1299:                                        m = so->so_rcv.sb_mb;
                   1300:                                }
1.69      thorpej  1301:                                /*
                   1302:                                 * If m != NULL, we also know that
                   1303:                                 * so->so_rcv.sb_mb != NULL.
                   1304:                                 */
                   1305:                                KASSERT(so->so_rcv.sb_mb == m);
                   1306:                                if (m) {
1.1       cgd      1307:                                        m->m_nextpkt = nextrecord;
1.69      thorpej  1308:                                        if (nextrecord == NULL)
                   1309:                                                so->so_rcv.sb_lastrecord = m;
                   1310:                                } else {
                   1311:                                        so->so_rcv.sb_mb = nextrecord;
1.70      thorpej  1312:                                        SB_EMPTY_FIXUP(&so->so_rcv);
1.69      thorpej  1313:                                }
                   1314:                                SBLASTRECORDCHK(&so->so_rcv, "soreceive 3");
                   1315:                                SBLASTMBUFCHK(&so->so_rcv, "soreceive 3");
1.1       cgd      1316:                        }
1.144   ! dyoung   1317:                } else if (flags & MSG_PEEK)
        !          1318:                        moff += len;
        !          1319:                else {
        !          1320:                        if (mp != NULL)
        !          1321:                                *mp = m_copym(m, 0, len, M_WAIT);
        !          1322:                        m->m_data += len;
        !          1323:                        m->m_len -= len;
        !          1324:                        so->so_rcv.sb_cc -= len;
1.1       cgd      1325:                }
                   1326:                if (so->so_oobmark) {
                   1327:                        if ((flags & MSG_PEEK) == 0) {
                   1328:                                so->so_oobmark -= len;
                   1329:                                if (so->so_oobmark == 0) {
                   1330:                                        so->so_state |= SS_RCVATMARK;
                   1331:                                        break;
                   1332:                                }
1.7       cgd      1333:                        } else {
1.1       cgd      1334:                                offset += len;
1.7       cgd      1335:                                if (offset == so->so_oobmark)
                   1336:                                        break;
                   1337:                        }
1.1       cgd      1338:                }
                   1339:                if (flags & MSG_EOR)
                   1340:                        break;
                   1341:                /*
                   1342:                 * If the MSG_WAITALL flag is set (for non-atomic socket),
                   1343:                 * we must not quit until "uio->uio_resid == 0" or an error
                   1344:                 * termination.  If a signal/timeout occurs, return
                   1345:                 * with a short count but without error.
                   1346:                 * Keep sockbuf locked against other readers.
                   1347:                 */
1.144   ! dyoung   1348:                while (flags & MSG_WAITALL && m == NULL && uio->uio_resid > 0 &&
1.3       andrew   1349:                    !sosendallatonce(so) && !nextrecord) {
1.1       cgd      1350:                        if (so->so_error || so->so_state & SS_CANTRCVMORE)
                   1351:                                break;
1.68      matt     1352:                        /*
                   1353:                         * If we are peeking and the socket receive buffer is
                   1354:                         * full, stop since we can't get more data to peek at.
                   1355:                         */
                   1356:                        if ((flags & MSG_PEEK) && sbspace(&so->so_rcv) <= 0)
                   1357:                                break;
                   1358:                        /*
                   1359:                         * If we've drained the socket buffer, tell the
                   1360:                         * protocol in case it needs to do something to
                   1361:                         * get it filled again.
                   1362:                         */
                   1363:                        if ((pr->pr_flags & PR_WANTRCVD) && so->so_pcb)
                   1364:                                (*pr->pr_usrreq)(so, PRU_RCVD,
1.140     dyoung   1365:                                    NULL, (struct mbuf *)(long)flags, NULL, l);
1.69      thorpej  1366:                        SBLASTRECORDCHK(&so->so_rcv, "soreceive sbwait 2");
                   1367:                        SBLASTMBUFCHK(&so->so_rcv, "soreceive sbwait 2");
1.1       cgd      1368:                        error = sbwait(&so->so_rcv);
1.144   ! dyoung   1369:                        if (error != 0) {
1.1       cgd      1370:                                sbunlock(&so->so_rcv);
                   1371:                                splx(s);
1.144   ! dyoung   1372:                                return 0;
1.1       cgd      1373:                        }
1.21      christos 1374:                        if ((m = so->so_rcv.sb_mb) != NULL)
1.1       cgd      1375:                                nextrecord = m->m_nextpkt;
                   1376:                }
                   1377:        }
1.3       andrew   1378:
                   1379:        if (m && pr->pr_flags & PR_ATOMIC) {
                   1380:                flags |= MSG_TRUNC;
                   1381:                if ((flags & MSG_PEEK) == 0)
                   1382:                        (void) sbdroprecord(&so->so_rcv);
                   1383:        }
1.1       cgd      1384:        if ((flags & MSG_PEEK) == 0) {
1.144   ! dyoung   1385:                if (m == NULL) {
1.69      thorpej  1386:                        /*
1.70      thorpej  1387:                         * First part is an inline SB_EMPTY_FIXUP().  Second
1.69      thorpej  1388:                         * part makes sure sb_lastrecord is up-to-date if
                   1389:                         * there is still data in the socket buffer.
                   1390:                         */
1.1       cgd      1391:                        so->so_rcv.sb_mb = nextrecord;
1.69      thorpej  1392:                        if (so->so_rcv.sb_mb == NULL) {
                   1393:                                so->so_rcv.sb_mbtail = NULL;
                   1394:                                so->so_rcv.sb_lastrecord = NULL;
                   1395:                        } else if (nextrecord->m_nextpkt == NULL)
                   1396:                                so->so_rcv.sb_lastrecord = nextrecord;
                   1397:                }
                   1398:                SBLASTRECORDCHK(&so->so_rcv, "soreceive 4");
                   1399:                SBLASTMBUFCHK(&so->so_rcv, "soreceive 4");
1.1       cgd      1400:                if (pr->pr_flags & PR_WANTRCVD && so->so_pcb)
1.140     dyoung   1401:                        (*pr->pr_usrreq)(so, PRU_RCVD, NULL,
                   1402:                            (struct mbuf *)(long)flags, NULL, l);
1.1       cgd      1403:        }
1.3       andrew   1404:        if (orig_resid == uio->uio_resid && orig_resid &&
                   1405:            (flags & MSG_EOR) == 0 && (so->so_state & SS_CANTRCVMORE) == 0) {
                   1406:                sbunlock(&so->so_rcv);
                   1407:                splx(s);
                   1408:                goto restart;
                   1409:        }
1.108     perry    1410:
1.144   ! dyoung   1411:        if (flagsp != NULL)
1.1       cgd      1412:                *flagsp |= flags;
1.54      lukem    1413:  release:
1.1       cgd      1414:        sbunlock(&so->so_rcv);
                   1415:        splx(s);
1.144   ! dyoung   1416:        return error;
1.1       cgd      1417: }
                   1418:
1.14      mycroft  1419: int
1.54      lukem    1420: soshutdown(struct socket *so, int how)
1.1       cgd      1421: {
1.99      matt     1422:        const struct protosw    *pr;
1.34      kleink   1423:
1.54      lukem    1424:        pr = so->so_proto;
1.34      kleink   1425:        if (!(how == SHUT_RD || how == SHUT_WR || how == SHUT_RDWR))
                   1426:                return (EINVAL);
1.1       cgd      1427:
1.34      kleink   1428:        if (how == SHUT_RD || how == SHUT_RDWR)
1.1       cgd      1429:                sorflush(so);
1.34      kleink   1430:        if (how == SHUT_WR || how == SHUT_RDWR)
1.140     dyoung   1431:                return (*pr->pr_usrreq)(so, PRU_SHUTDOWN, NULL,
                   1432:                    NULL, NULL, NULL);
1.144   ! dyoung   1433:        return 0;
1.1       cgd      1434: }
                   1435:
1.14      mycroft  1436: void
1.54      lukem    1437: sorflush(struct socket *so)
1.1       cgd      1438: {
1.54      lukem    1439:        struct sockbuf  *sb, asb;
1.99      matt     1440:        const struct protosw    *pr;
1.54      lukem    1441:        int             s;
1.1       cgd      1442:
1.54      lukem    1443:        sb = &so->so_rcv;
                   1444:        pr = so->so_proto;
1.1       cgd      1445:        sb->sb_flags |= SB_NOINTR;
1.15      mycroft  1446:        (void) sblock(sb, M_WAITOK);
1.56      thorpej  1447:        s = splnet();
1.1       cgd      1448:        socantrcvmore(so);
                   1449:        sbunlock(sb);
                   1450:        asb = *sb;
1.86      wrstuden 1451:        /*
                   1452:         * Clear most of the sockbuf structure, but leave some of the
                   1453:         * fields valid.
                   1454:         */
                   1455:        memset(&sb->sb_startzero, 0,
                   1456:            sizeof(*sb) - offsetof(struct sockbuf, sb_startzero));
1.1       cgd      1457:        splx(s);
                   1458:        if (pr->pr_flags & PR_RIGHTS && pr->pr_domain->dom_dispose)
                   1459:                (*pr->pr_domain->dom_dispose)(asb.sb_mb);
1.98      christos 1460:        sbrelease(&asb, so);
1.1       cgd      1461: }
                   1462:
1.142     dyoung   1463: static int
                   1464: sosetopt1(struct socket *so, int level, int optname, struct mbuf *m)
1.1       cgd      1465: {
1.142     dyoung   1466:        int optval, val;
1.130     christos 1467:        struct linger   *l;
1.141     yamt     1468:        struct sockbuf  *sb;
1.142     dyoung   1469:        struct timeval *tv;
                   1470:
                   1471:        switch (optname) {
                   1472:
                   1473:        case SO_LINGER:
                   1474:                if (m == NULL || m->m_len != sizeof(struct linger))
                   1475:                        return EINVAL;
                   1476:                l = mtod(m, struct linger *);
                   1477:                if (l->l_linger < 0 || l->l_linger > USHRT_MAX ||
                   1478:                    l->l_linger > (INT_MAX / hz))
                   1479:                        return EDOM;
                   1480:                so->so_linger = l->l_linger;
                   1481:                if (l->l_onoff)
                   1482:                        so->so_options |= SO_LINGER;
                   1483:                else
                   1484:                        so->so_options &= ~SO_LINGER;
                   1485:                break;
1.1       cgd      1486:
1.142     dyoung   1487:        case SO_DEBUG:
                   1488:        case SO_KEEPALIVE:
                   1489:        case SO_DONTROUTE:
                   1490:        case SO_USELOOPBACK:
                   1491:        case SO_BROADCAST:
                   1492:        case SO_REUSEADDR:
                   1493:        case SO_REUSEPORT:
                   1494:        case SO_OOBINLINE:
                   1495:        case SO_TIMESTAMP:
                   1496:                if (m == NULL || m->m_len < sizeof(int))
                   1497:                        return EINVAL;
                   1498:                if (*mtod(m, int *))
                   1499:                        so->so_options |= optname;
                   1500:                else
                   1501:                        so->so_options &= ~optname;
                   1502:                break;
                   1503:
                   1504:        case SO_SNDBUF:
                   1505:        case SO_RCVBUF:
                   1506:        case SO_SNDLOWAT:
                   1507:        case SO_RCVLOWAT:
                   1508:                if (m == NULL || m->m_len < sizeof(int))
                   1509:                        return EINVAL;
1.1       cgd      1510:
1.142     dyoung   1511:                /*
                   1512:                 * Values < 1 make no sense for any of these
                   1513:                 * options, so disallow them.
                   1514:                 */
                   1515:                optval = *mtod(m, int *);
                   1516:                if (optval < 1)
                   1517:                        return EINVAL;
1.1       cgd      1518:
1.142     dyoung   1519:                switch (optname) {
1.1       cgd      1520:
                   1521:                case SO_SNDBUF:
                   1522:                case SO_RCVBUF:
1.142     dyoung   1523:                        sb = (optname == SO_SNDBUF) ?
                   1524:                            &so->so_snd : &so->so_rcv;
                   1525:                        if (sbreserve(sb, (u_long)optval, so) == 0)
                   1526:                                return ENOBUFS;
                   1527:                        sb->sb_flags &= ~SB_AUTOSIZE;
                   1528:                        break;
                   1529:
                   1530:                /*
                   1531:                 * Make sure the low-water is never greater than
                   1532:                 * the high-water.
                   1533:                 */
1.1       cgd      1534:                case SO_SNDLOWAT:
1.142     dyoung   1535:                        so->so_snd.sb_lowat =
                   1536:                            (optval > so->so_snd.sb_hiwat) ?
                   1537:                            so->so_snd.sb_hiwat : optval;
                   1538:                        break;
1.1       cgd      1539:                case SO_RCVLOWAT:
1.142     dyoung   1540:                        so->so_rcv.sb_lowat =
                   1541:                            (optval > so->so_rcv.sb_hiwat) ?
                   1542:                            so->so_rcv.sb_hiwat : optval;
                   1543:                        break;
                   1544:                }
                   1545:                break;
1.28      thorpej  1546:
1.142     dyoung   1547:        case SO_SNDTIMEO:
                   1548:        case SO_RCVTIMEO:
                   1549:                if (m == NULL || m->m_len < sizeof(*tv))
                   1550:                        return EINVAL;
                   1551:                tv = mtod(m, struct timeval *);
                   1552:                if (tv->tv_sec > (INT_MAX - tv->tv_usec / tick) / hz)
                   1553:                        return EDOM;
                   1554:                val = tv->tv_sec * hz + tv->tv_usec / tick;
                   1555:                if (val == 0 && tv->tv_usec != 0)
                   1556:                        val = 1;
1.28      thorpej  1557:
1.142     dyoung   1558:                switch (optname) {
1.28      thorpej  1559:
1.142     dyoung   1560:                case SO_SNDTIMEO:
                   1561:                        so->so_snd.sb_timeo = val;
1.1       cgd      1562:                        break;
                   1563:                case SO_RCVTIMEO:
1.142     dyoung   1564:                        so->so_rcv.sb_timeo = val;
                   1565:                        break;
                   1566:                }
                   1567:                break;
1.1       cgd      1568:
1.142     dyoung   1569:        default:
                   1570:                return ENOPROTOOPT;
                   1571:        }
                   1572:        return 0;
                   1573: }
1.1       cgd      1574:
1.142     dyoung   1575: int
                   1576: sosetopt(struct socket *so, int level, int optname, struct mbuf *m)
                   1577: {
                   1578:        int error, prerr;
1.1       cgd      1579:
1.142     dyoung   1580:        if (level == SOL_SOCKET)
                   1581:                error = sosetopt1(so, level, optname, m);
                   1582:        else
                   1583:                error = ENOPROTOOPT;
1.1       cgd      1584:
1.142     dyoung   1585:        if ((error == 0 || error == ENOPROTOOPT) &&
                   1586:            so->so_proto != NULL && so->so_proto->pr_ctloutput != NULL) {
                   1587:                /* give the protocol stack a shot */
                   1588:                prerr = (*so->so_proto->pr_ctloutput)(PRCO_SETOPT, so, level,
                   1589:                    optname, &m);
                   1590:                if (prerr == 0)
                   1591:                        error = 0;
                   1592:                else if (prerr != ENOPROTOOPT)
                   1593:                        error = prerr;
                   1594:        } else if (m != NULL)
                   1595:                (void)m_free(m);
                   1596:        return error;
1.1       cgd      1597: }
                   1598:
1.14      mycroft  1599: int
1.54      lukem    1600: sogetopt(struct socket *so, int level, int optname, struct mbuf **mp)
1.1       cgd      1601: {
1.54      lukem    1602:        struct mbuf     *m;
1.1       cgd      1603:
                   1604:        if (level != SOL_SOCKET) {
                   1605:                if (so->so_proto && so->so_proto->pr_ctloutput) {
                   1606:                        return ((*so->so_proto->pr_ctloutput)
                   1607:                                  (PRCO_GETOPT, so, level, optname, mp));
                   1608:                } else
                   1609:                        return (ENOPROTOOPT);
                   1610:        } else {
                   1611:                m = m_get(M_WAIT, MT_SOOPTS);
1.36      perry    1612:                m->m_len = sizeof(int);
1.1       cgd      1613:
                   1614:                switch (optname) {
                   1615:
                   1616:                case SO_LINGER:
1.36      perry    1617:                        m->m_len = sizeof(struct linger);
1.1       cgd      1618:                        mtod(m, struct linger *)->l_onoff =
1.131     christos 1619:                            (so->so_options & SO_LINGER) ? 1 : 0;
1.1       cgd      1620:                        mtod(m, struct linger *)->l_linger = so->so_linger;
                   1621:                        break;
                   1622:
                   1623:                case SO_USELOOPBACK:
                   1624:                case SO_DONTROUTE:
                   1625:                case SO_DEBUG:
                   1626:                case SO_KEEPALIVE:
                   1627:                case SO_REUSEADDR:
1.15      mycroft  1628:                case SO_REUSEPORT:
1.1       cgd      1629:                case SO_BROADCAST:
                   1630:                case SO_OOBINLINE:
1.26      thorpej  1631:                case SO_TIMESTAMP:
1.131     christos 1632:                        *mtod(m, int *) = (so->so_options & optname) ? 1 : 0;
1.1       cgd      1633:                        break;
                   1634:
                   1635:                case SO_TYPE:
                   1636:                        *mtod(m, int *) = so->so_type;
                   1637:                        break;
                   1638:
                   1639:                case SO_ERROR:
                   1640:                        *mtod(m, int *) = so->so_error;
                   1641:                        so->so_error = 0;
                   1642:                        break;
                   1643:
                   1644:                case SO_SNDBUF:
                   1645:                        *mtod(m, int *) = so->so_snd.sb_hiwat;
                   1646:                        break;
                   1647:
                   1648:                case SO_RCVBUF:
                   1649:                        *mtod(m, int *) = so->so_rcv.sb_hiwat;
                   1650:                        break;
                   1651:
                   1652:                case SO_SNDLOWAT:
                   1653:                        *mtod(m, int *) = so->so_snd.sb_lowat;
                   1654:                        break;
                   1655:
                   1656:                case SO_RCVLOWAT:
                   1657:                        *mtod(m, int *) = so->so_rcv.sb_lowat;
                   1658:                        break;
                   1659:
                   1660:                case SO_SNDTIMEO:
                   1661:                case SO_RCVTIMEO:
                   1662:                    {
                   1663:                        int val = (optname == SO_SNDTIMEO ?
                   1664:                             so->so_snd.sb_timeo : so->so_rcv.sb_timeo);
                   1665:
                   1666:                        m->m_len = sizeof(struct timeval);
                   1667:                        mtod(m, struct timeval *)->tv_sec = val / hz;
                   1668:                        mtod(m, struct timeval *)->tv_usec =
1.27      kleink   1669:                            (val % hz) * tick;
1.1       cgd      1670:                        break;
                   1671:                    }
                   1672:
1.107     darrenr  1673:                case SO_OVERFLOWED:
                   1674:                        *mtod(m, int *) = so->so_rcv.sb_overflowed;
                   1675:                        break;
                   1676:
1.1       cgd      1677:                default:
                   1678:                        (void)m_free(m);
                   1679:                        return (ENOPROTOOPT);
                   1680:                }
                   1681:                *mp = m;
                   1682:                return (0);
                   1683:        }
                   1684: }
                   1685:
1.14      mycroft  1686: void
1.54      lukem    1687: sohasoutofband(struct socket *so)
1.1       cgd      1688: {
1.90      christos 1689:        fownsignal(so->so_pgid, SIGURG, POLL_PRI, POLLPRI|POLLRDBAND, so);
1.2       cgd      1690:        selwakeup(&so->so_rcv.sb_sel);
1.1       cgd      1691: }
1.72      jdolecek 1692:
                   1693: static void
                   1694: filt_sordetach(struct knote *kn)
                   1695: {
                   1696:        struct socket   *so;
                   1697:
                   1698:        so = (struct socket *)kn->kn_fp->f_data;
1.73      christos 1699:        SLIST_REMOVE(&so->so_rcv.sb_sel.sel_klist, kn, knote, kn_selnext);
                   1700:        if (SLIST_EMPTY(&so->so_rcv.sb_sel.sel_klist))
1.72      jdolecek 1701:                so->so_rcv.sb_flags &= ~SB_KNOTE;
                   1702: }
                   1703:
                   1704: /*ARGSUSED*/
                   1705: static int
1.129     yamt     1706: filt_soread(struct knote *kn, long hint)
1.72      jdolecek 1707: {
                   1708:        struct socket   *so;
                   1709:
                   1710:        so = (struct socket *)kn->kn_fp->f_data;
                   1711:        kn->kn_data = so->so_rcv.sb_cc;
                   1712:        if (so->so_state & SS_CANTRCVMORE) {
1.108     perry    1713:                kn->kn_flags |= EV_EOF;
1.72      jdolecek 1714:                kn->kn_fflags = so->so_error;
                   1715:                return (1);
                   1716:        }
                   1717:        if (so->so_error)       /* temporary udp error */
                   1718:                return (1);
                   1719:        if (kn->kn_sfflags & NOTE_LOWAT)
                   1720:                return (kn->kn_data >= kn->kn_sdata);
                   1721:        return (kn->kn_data >= so->so_rcv.sb_lowat);
                   1722: }
                   1723:
                   1724: static void
                   1725: filt_sowdetach(struct knote *kn)
                   1726: {
                   1727:        struct socket   *so;
                   1728:
                   1729:        so = (struct socket *)kn->kn_fp->f_data;
1.73      christos 1730:        SLIST_REMOVE(&so->so_snd.sb_sel.sel_klist, kn, knote, kn_selnext);
                   1731:        if (SLIST_EMPTY(&so->so_snd.sb_sel.sel_klist))
1.72      jdolecek 1732:                so->so_snd.sb_flags &= ~SB_KNOTE;
                   1733: }
                   1734:
                   1735: /*ARGSUSED*/
                   1736: static int
1.129     yamt     1737: filt_sowrite(struct knote *kn, long hint)
1.72      jdolecek 1738: {
                   1739:        struct socket   *so;
                   1740:
                   1741:        so = (struct socket *)kn->kn_fp->f_data;
                   1742:        kn->kn_data = sbspace(&so->so_snd);
                   1743:        if (so->so_state & SS_CANTSENDMORE) {
1.108     perry    1744:                kn->kn_flags |= EV_EOF;
1.72      jdolecek 1745:                kn->kn_fflags = so->so_error;
                   1746:                return (1);
                   1747:        }
                   1748:        if (so->so_error)       /* temporary udp error */
                   1749:                return (1);
                   1750:        if (((so->so_state & SS_ISCONNECTED) == 0) &&
                   1751:            (so->so_proto->pr_flags & PR_CONNREQUIRED))
                   1752:                return (0);
                   1753:        if (kn->kn_sfflags & NOTE_LOWAT)
                   1754:                return (kn->kn_data >= kn->kn_sdata);
                   1755:        return (kn->kn_data >= so->so_snd.sb_lowat);
                   1756: }
                   1757:
                   1758: /*ARGSUSED*/
                   1759: static int
1.129     yamt     1760: filt_solisten(struct knote *kn, long hint)
1.72      jdolecek 1761: {
                   1762:        struct socket   *so;
                   1763:
                   1764:        so = (struct socket *)kn->kn_fp->f_data;
                   1765:
                   1766:        /*
                   1767:         * Set kn_data to number of incoming connections, not
                   1768:         * counting partial (incomplete) connections.
1.108     perry    1769:         */
1.72      jdolecek 1770:        kn->kn_data = so->so_qlen;
                   1771:        return (kn->kn_data > 0);
                   1772: }
                   1773:
                   1774: static const struct filterops solisten_filtops =
                   1775:        { 1, NULL, filt_sordetach, filt_solisten };
                   1776: static const struct filterops soread_filtops =
                   1777:        { 1, NULL, filt_sordetach, filt_soread };
                   1778: static const struct filterops sowrite_filtops =
                   1779:        { 1, NULL, filt_sowdetach, filt_sowrite };
                   1780:
                   1781: int
1.129     yamt     1782: soo_kqfilter(struct file *fp, struct knote *kn)
1.72      jdolecek 1783: {
                   1784:        struct socket   *so;
                   1785:        struct sockbuf  *sb;
                   1786:
                   1787:        so = (struct socket *)kn->kn_fp->f_data;
                   1788:        switch (kn->kn_filter) {
                   1789:        case EVFILT_READ:
                   1790:                if (so->so_options & SO_ACCEPTCONN)
                   1791:                        kn->kn_fop = &solisten_filtops;
                   1792:                else
                   1793:                        kn->kn_fop = &soread_filtops;
                   1794:                sb = &so->so_rcv;
                   1795:                break;
                   1796:        case EVFILT_WRITE:
                   1797:                kn->kn_fop = &sowrite_filtops;
                   1798:                sb = &so->so_snd;
                   1799:                break;
                   1800:        default:
                   1801:                return (1);
                   1802:        }
1.73      christos 1803:        SLIST_INSERT_HEAD(&sb->sb_sel.sel_klist, kn, kn_selnext);
1.72      jdolecek 1804:        sb->sb_flags |= SB_KNOTE;
                   1805:        return (0);
                   1806: }
                   1807:
1.94      yamt     1808: #include <sys/sysctl.h>
                   1809:
                   1810: static int sysctl_kern_somaxkva(SYSCTLFN_PROTO);
                   1811:
                   1812: /*
                   1813:  * sysctl helper routine for kern.somaxkva.  ensures that the given
                   1814:  * value is not too small.
                   1815:  * (XXX should we maybe make sure it's not too large as well?)
                   1816:  */
                   1817: static int
                   1818: sysctl_kern_somaxkva(SYSCTLFN_ARGS)
                   1819: {
                   1820:        int error, new_somaxkva;
                   1821:        struct sysctlnode node;
                   1822:
                   1823:        new_somaxkva = somaxkva;
                   1824:        node = *rnode;
                   1825:        node.sysctl_data = &new_somaxkva;
                   1826:        error = sysctl_lookup(SYSCTLFN_CALL(&node));
                   1827:        if (error || newp == NULL)
                   1828:                return (error);
                   1829:
                   1830:        if (new_somaxkva < (16 * 1024 * 1024)) /* sanity */
                   1831:                return (EINVAL);
                   1832:
1.136     ad       1833:        mutex_enter(&so_pendfree_lock);
1.94      yamt     1834:        somaxkva = new_somaxkva;
1.136     ad       1835:        cv_broadcast(&socurkva_cv);
                   1836:        mutex_exit(&so_pendfree_lock);
1.94      yamt     1837:
                   1838:        return (error);
                   1839: }
                   1840:
                   1841: SYSCTL_SETUP(sysctl_kern_somaxkva_setup, "sysctl kern.somaxkva setup")
                   1842: {
                   1843:
1.97      atatat   1844:        sysctl_createv(clog, 0, NULL, NULL,
                   1845:                       CTLFLAG_PERMANENT,
                   1846:                       CTLTYPE_NODE, "kern", NULL,
                   1847:                       NULL, 0, NULL, 0,
                   1848:                       CTL_KERN, CTL_EOL);
                   1849:
                   1850:        sysctl_createv(clog, 0, NULL, NULL,
                   1851:                       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.103     atatat   1852:                       CTLTYPE_INT, "somaxkva",
                   1853:                       SYSCTL_DESCR("Maximum amount of kernel memory to be "
                   1854:                                    "used for socket buffers"),
1.94      yamt     1855:                       sysctl_kern_somaxkva, 0, NULL, 0,
                   1856:                       CTL_KERN, KERN_SOMAXKVA, CTL_EOL);
                   1857: }

CVSweb <webmaster@jp.NetBSD.org>