[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 1.111.2.8

1.111.2.8! yamt        1: /*     $NetBSD: uipc_socket.c,v 1.111.2.7 2007/09/03 14:41:18 yamt Exp $       */
1.64      thorpej     2:
                      3: /*-
1.111.2.7  yamt        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>
1.111.2.8! yamt       71: __KERNEL_RCSID(0, "$NetBSD: uipc_socket.c,v 1.111.2.7 2007/09/03 14:41:18 yamt 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>
1.111.2.8! yamt       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>
1.111.2.3  yamt       95: #include <sys/kauth.h>
1.111.2.7  yamt       96: #include <sys/mutex.h>
                     97: #include <sys/condvar.h>
1.37      thorpej    98:
1.64      thorpej    99: #include <uvm/uvm.h>
                    100:
1.111.2.7  yamt      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:
1.111.2.8! yamt      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:
1.111.2.3  yamt      115: static struct evcnt sosend_loan_big = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64      thorpej   116:     NULL, "sosend", "loan big");
1.111.2.3  yamt      117: static struct evcnt sosend_copy_big = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64      thorpej   118:     NULL, "sosend", "copy big");
1.111.2.3  yamt      119: static struct evcnt sosend_copy_small = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
1.64      thorpej   120:     NULL, "sosend", "copy small");
1.111.2.3  yamt      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:
1.111.2.3  yamt      136: static struct callback_entry sokva_reclaimerentry;
1.1       cgd       137:
1.71      thorpej   138: #ifdef SOSEND_NO_LOAN
1.111.2.3  yamt      139: int sock_loan_thresh = -1;
1.71      thorpej   140: #else
1.111.2.3  yamt      141: int sock_loan_thresh = 4096;
1.65      thorpej   142: #endif
1.64      thorpej   143:
1.111.2.7  yamt      144: static kmutex_t so_pendfree_lock;
1.111.2.3  yamt      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;
1.111.2.3  yamt      151: static int socurkva;
1.111.2.7  yamt      152: static kcondvar_t socurkva_cv;
1.64      thorpej   153:
                    154: #define        SOCK_LOAN_CHUNK         65536
                    155:
1.111.2.3  yamt      156: static size_t sodopendfree(void);
                    157: static size_t sodopendfreel(void);
1.93      yamt      158:
1.111.2.3  yamt      159: static vsize_t
1.95      yamt      160: sokvareserve(struct socket *so, vsize_t len)
1.80      yamt      161: {
1.98      christos  162:        int error;
1.80      yamt      163:
1.111.2.7  yamt      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:
1.111.2.3  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);
1.111.2.7  yamt      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;
1.111.2.7  yamt      189:        mutex_exit(&so_pendfree_lock);
1.98      christos  190:        return len;
1.95      yamt      191: }
                    192:
1.111.2.3  yamt      193: static void
1.95      yamt      194: sokvaunreserve(vsize_t len)
                    195: {
                    196:
1.111.2.7  yamt      197:        mutex_enter(&so_pendfree_lock);
1.95      yamt      198:        socurkva -= len;
1.111.2.7  yamt      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
1.111.2.7  yamt      254: sodoloanfree(struct vm_page **pgs, void *buf, size_t size, bool mapped)
1.64      thorpej   255: {
1.111.2.1  yamt      256:        vaddr_t sva, eva;
1.64      thorpej   257:        vsize_t len;
1.111.2.1  yamt      258:        int npgs;
                    259:
                    260:        KASSERT(pgs != NULL);
1.64      thorpej   261:
                    262:        eva = round_page((vaddr_t) buf + size);
                    263:        sva = trunc_page((vaddr_t) buf);
                    264:        len = eva - sva;
                    265:        npgs = len >> PAGE_SHIFT;
                    266:
1.111.2.1  yamt      267:        if (mapped) {
                    268:                pmap_kremove(sva, len);
                    269:                pmap_update(pmap_kernel());
1.64      thorpej   270:        }
                    271:        uvm_unloan(pgs, npgs, UVM_LOAN_TOPAGE);
1.80      yamt      272:        sokvafree(sva, len);
1.64      thorpej   273: }
                    274:
                    275: static size_t
1.111.2.3  yamt      276: sodopendfree()
1.64      thorpej   277: {
1.93      yamt      278:        size_t rv;
1.64      thorpej   279:
1.111.2.7  yamt      280:        mutex_enter(&so_pendfree_lock);
1.111.2.3  yamt      281:        rv = sodopendfreel();
1.111.2.7  yamt      282:        mutex_exit(&so_pendfree_lock);
1.93      yamt      283:
                    284:        return rv;
                    285: }
                    286:
                    287: /*
                    288:  * sodopendfreel: free mbufs on "pendfree" list.
1.111.2.7  yamt      289:  * unlock and relock so_pendfree_lock when freeing mbufs.
1.93      yamt      290:  *
1.111.2.7  yamt      291:  * => called with so_pendfree_lock held.
1.93      yamt      292:  */
                    293:
                    294: static size_t
1.111.2.3  yamt      295: sodopendfreel()
1.93      yamt      296: {
1.111.2.7  yamt      297:        struct mbuf *m, *next;
1.93      yamt      298:        size_t rv = 0;
1.111.2.7  yamt      299:        int s;
1.93      yamt      300:
1.111.2.7  yamt      301:        KASSERT(mutex_owned(&so_pendfree_lock));
1.93      yamt      302:
1.111.2.7  yamt      303:        while (so_pendfree != NULL) {
1.64      thorpej   304:                m = so_pendfree;
1.93      yamt      305:                so_pendfree = NULL;
1.111.2.7  yamt      306:                mutex_exit(&so_pendfree_lock);
1.93      yamt      307:
                    308:                for (; m != NULL; m = next) {
                    309:                        next = m->m_next;
1.111.2.1  yamt      310:                        KASSERT((~m->m_flags & (M_EXT|M_EXT_PAGES)) == 0);
1.111.2.4  yamt      311:                        KASSERT(m->m_ext.ext_refcnt == 0);
1.93      yamt      312:
                    313:                        rv += m->m_ext.ext_size;
1.111.2.1  yamt      314:                        sodoloanfree(m->m_ext.ext_pgs, m->m_ext.ext_buf,
                    315:                            m->m_ext.ext_size,
                    316:                            (m->m_ext.ext_flags & M_EXT_LAZY) == 0);
1.111.2.7  yamt      317:                        s = splvm();
1.93      yamt      318:                        pool_cache_put(&mbpool_cache, m);
1.111.2.7  yamt      319:                        splx(s);
1.93      yamt      320:                }
1.64      thorpej   321:
1.111.2.7  yamt      322:                mutex_enter(&so_pendfree_lock);
1.64      thorpej   323:        }
                    324:
                    325:        return (rv);
                    326: }
                    327:
1.80      yamt      328: void
1.111.2.7  yamt      329: soloanfree(struct mbuf *m, void *buf, size_t size, void *arg)
1.64      thorpej   330: {
                    331:
1.111.2.1  yamt      332:        KASSERT(m != NULL);
1.64      thorpej   333:
1.93      yamt      334:        /*
                    335:         * postpone freeing mbuf.
                    336:         *
                    337:         * we can't do it in interrupt context
                    338:         * because we need to put kva back to kernel_map.
                    339:         */
                    340:
1.111.2.7  yamt      341:        mutex_enter(&so_pendfree_lock);
1.92      yamt      342:        m->m_next = so_pendfree;
                    343:        so_pendfree = m;
1.111.2.7  yamt      344:        cv_broadcast(&socurkva_cv);
                    345:        mutex_exit(&so_pendfree_lock);
1.64      thorpej   346: }
                    347:
                    348: static long
                    349: sosend_loan(struct socket *so, struct uio *uio, struct mbuf *m, long space)
                    350: {
                    351:        struct iovec *iov = uio->uio_iov;
                    352:        vaddr_t sva, eva;
                    353:        vsize_t len;
1.111.2.1  yamt      354:        vaddr_t lva;
                    355:        int npgs, error;
1.111.2.2  yamt      356: #if !defined(__HAVE_LAZY_MBUF)
1.111.2.1  yamt      357:        vaddr_t va;
                    358:        int i;
1.111.2.2  yamt      359: #endif /* !defined(__HAVE_LAZY_MBUF) */
1.64      thorpej   360:
1.111.2.3  yamt      361:        if (VMSPACE_IS_KERNEL_P(uio->uio_vmspace))
1.64      thorpej   362:                return (0);
                    363:
                    364:        if (iov->iov_len < (size_t) space)
                    365:                space = iov->iov_len;
                    366:        if (space > SOCK_LOAN_CHUNK)
                    367:                space = SOCK_LOAN_CHUNK;
                    368:
                    369:        eva = round_page((vaddr_t) iov->iov_base + space);
                    370:        sva = trunc_page((vaddr_t) iov->iov_base);
                    371:        len = eva - sva;
                    372:        npgs = len >> PAGE_SHIFT;
                    373:
1.79      thorpej   374:        /* XXX KDASSERT */
                    375:        KASSERT(npgs <= M_EXT_MAXPAGES);
                    376:
1.80      yamt      377:        lva = sokvaalloc(len, so);
1.64      thorpej   378:        if (lva == 0)
1.80      yamt      379:                return 0;
1.64      thorpej   380:
1.111.2.3  yamt      381:        error = uvm_loan(&uio->uio_vmspace->vm_map, sva, len,
1.79      thorpej   382:            m->m_ext.ext_pgs, UVM_LOAN_TOPAGE);
1.64      thorpej   383:        if (error) {
1.80      yamt      384:                sokvafree(lva, len);
1.64      thorpej   385:                return (0);
                    386:        }
                    387:
1.111.2.2  yamt      388: #if !defined(__HAVE_LAZY_MBUF)
1.64      thorpej   389:        for (i = 0, va = lva; i < npgs; i++, va += PAGE_SIZE)
1.79      thorpej   390:                pmap_kenter_pa(va, VM_PAGE_TO_PHYS(m->m_ext.ext_pgs[i]),
                    391:                    VM_PROT_READ);
1.64      thorpej   392:        pmap_update(pmap_kernel());
1.111.2.2  yamt      393: #endif /* !defined(__HAVE_LAZY_MBUF) */
1.64      thorpej   394:
                    395:        lva += (vaddr_t) iov->iov_base & PAGE_MASK;
                    396:
1.111.2.7  yamt      397:        MEXTADD(m, (void *) lva, space, M_MBUF, soloanfree, so);
1.79      thorpej   398:        m->m_flags |= M_EXT_PAGES | M_EXT_ROMAP;
1.64      thorpej   399:
1.111.2.2  yamt      400: #if defined(__HAVE_LAZY_MBUF)
1.111.2.1  yamt      401:        m->m_flags |= M_EXT_LAZY;
                    402:        m->m_ext.ext_flags |= M_EXT_LAZY;
1.111.2.2  yamt      403: #endif /* defined(__HAVE_LAZY_MBUF) */
1.111.2.1  yamt      404:
1.64      thorpej   405:        uio->uio_resid -= space;
                    406:        /* uio_offset not updated, not set/used for write(2) */
1.111.2.7  yamt      407:        uio->uio_iov->iov_base = (char *)uio->uio_iov->iov_base + space;
1.64      thorpej   408:        uio->uio_iov->iov_len -= space;
                    409:        if (uio->uio_iov->iov_len == 0) {
                    410:                uio->uio_iov++;
                    411:                uio->uio_iovcnt--;
                    412:        }
                    413:
                    414:        return (space);
                    415: }
                    416:
1.111.2.3  yamt      417: static int
                    418: sokva_reclaim_callback(struct callback_entry *ce, void *obj, void *arg)
                    419: {
                    420:
                    421:        KASSERT(ce == &sokva_reclaimerentry);
                    422:        KASSERT(obj == NULL);
                    423:
                    424:        sodopendfree();
                    425:        if (!vm_map_starved_p(kernel_map)) {
                    426:                return CALLBACK_CHAIN_ABORT;
                    427:        }
                    428:        return CALLBACK_CHAIN_CONTINUE;
                    429: }
                    430:
1.111.2.8! yamt      431: struct mbuf *
        !           432: getsombuf(struct socket *so)
        !           433: {
        !           434:        struct mbuf *m;
        !           435:
        !           436:        m = m_get(M_WAIT, MT_SONAME);
        !           437:        MCLAIM(m, so->so_mowner);
        !           438:        return m;
        !           439: }
        !           440:
        !           441: struct mbuf *
        !           442: m_intopt(struct socket *so, int val)
        !           443: {
        !           444:        struct mbuf *m;
        !           445:
        !           446:        m = getsombuf(so);
        !           447:        m->m_len = sizeof(int);
        !           448:        *mtod(m, int *) = val;
        !           449:        return m;
        !           450: }
        !           451:
1.111.2.3  yamt      452: void
                    453: soinit(void)
                    454: {
                    455:
1.111.2.7  yamt      456:        mutex_init(&so_pendfree_lock, MUTEX_DRIVER, IPL_VM);
                    457:        cv_init(&socurkva_cv, "sokva");
                    458:
1.111.2.3  yamt      459:        /* Set the initial adjusted socket buffer size. */
                    460:        if (sb_max_set(sb_max))
                    461:                panic("bad initial sb_max value: %lu", sb_max);
                    462:
                    463:        callback_register(&vm_map_to_kernel(kernel_map)->vmk_reclaim_callback,
                    464:            &sokva_reclaimerentry, NULL, sokva_reclaim_callback);
                    465: }
                    466:
1.1       cgd       467: /*
                    468:  * Socket operation routines.
                    469:  * These routines are called by the routines in
                    470:  * sys_socket.c or from a system process, and
                    471:  * implement the semantics of socket operations by
                    472:  * switching out to the protocol specific routines.
                    473:  */
                    474: /*ARGSUSED*/
1.3       andrew    475: int
1.111.2.3  yamt      476: socreate(int dom, struct socket **aso, int type, int proto, struct lwp *l)
1.1       cgd       477: {
1.99      matt      478:        const struct protosw    *prp;
1.54      lukem     479:        struct socket   *so;
1.111.2.3  yamt      480:        uid_t           uid;
1.54      lukem     481:        int             error, s;
1.1       cgd       482:
1.111.2.6  yamt      483:        error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET,
                    484:            KAUTH_REQ_NETWORK_SOCKET_OPEN, KAUTH_ARG(dom), KAUTH_ARG(type),
                    485:            KAUTH_ARG(proto));
1.111.2.7  yamt      486:        if (error != 0)
                    487:                return error;
1.111.2.5  yamt      488:
1.1       cgd       489:        if (proto)
                    490:                prp = pffindproto(dom, proto, type);
                    491:        else
                    492:                prp = pffindtype(dom, type);
1.111.2.7  yamt      493:        if (prp == NULL) {
1.111.2.3  yamt      494:                /* no support for domain */
                    495:                if (pffinddomain(dom) == 0)
1.111.2.7  yamt      496:                        return EAFNOSUPPORT;
1.111.2.3  yamt      497:                /* no support for socket type */
                    498:                if (proto == 0 && type != 0)
1.111.2.7  yamt      499:                        return EPROTOTYPE;
                    500:                return EPROTONOSUPPORT;
1.111.2.3  yamt      501:        }
1.111.2.7  yamt      502:        if (prp->pr_usrreq == NULL)
                    503:                return EPROTONOSUPPORT;
1.1       cgd       504:        if (prp->pr_type != type)
1.111.2.7  yamt      505:                return EPROTOTYPE;
1.39      matt      506:        s = splsoftnet();
1.37      thorpej   507:        so = pool_get(&socket_pool, PR_WAITOK);
1.111.2.7  yamt      508:        memset(so, 0, sizeof(*so));
1.31      thorpej   509:        TAILQ_INIT(&so->so_q0);
                    510:        TAILQ_INIT(&so->so_q);
1.1       cgd       511:        so->so_type = type;
                    512:        so->so_proto = prp;
1.33      matt      513:        so->so_send = sosend;
                    514:        so->so_receive = soreceive;
1.78      matt      515: #ifdef MBUFTRACE
                    516:        so->so_rcv.sb_mowner = &prp->pr_domain->dom_mowner;
                    517:        so->so_snd.sb_mowner = &prp->pr_domain->dom_mowner;
                    518:        so->so_mowner = &prp->pr_domain->dom_mowner;
                    519: #endif
1.111.2.8! yamt      520:        selinit(&so->so_rcv.sb_sel);
        !           521:        selinit(&so->so_snd.sb_sel);
1.111.2.7  yamt      522:        uid = kauth_cred_geteuid(l->l_cred);
1.111.2.3  yamt      523:        so->so_uidinfo = uid_find(uid);
1.111.2.7  yamt      524:        error = (*prp->pr_usrreq)(so, PRU_ATTACH, NULL,
                    525:            (struct mbuf *)(long)proto, NULL, l);
                    526:        if (error != 0) {
1.1       cgd       527:                so->so_state |= SS_NOFDREF;
                    528:                sofree(so);
1.39      matt      529:                splx(s);
1.111.2.7  yamt      530:                return error;
1.1       cgd       531:        }
1.39      matt      532:        splx(s);
1.1       cgd       533:        *aso = so;
1.111.2.7  yamt      534:        return 0;
1.1       cgd       535: }
                    536:
1.111.2.8! yamt      537: /* On success, write file descriptor to fdout and return zero.  On
        !           538:  * failure, return non-zero; *fdout will be undefined.
        !           539:  */
        !           540: int
        !           541: fsocreate(int domain, struct socket **sop, int type, int protocol,
        !           542:     struct lwp *l, int *fdout)
        !           543: {
        !           544:        struct filedesc *fdp;
        !           545:        struct socket   *so;
        !           546:        struct file     *fp;
        !           547:        int             fd, error;
        !           548:
        !           549:        fdp = l->l_proc->p_fd;
        !           550:        /* falloc() will use the desciptor for us */
        !           551:        if ((error = falloc(l, &fp, &fd)) != 0)
        !           552:                return (error);
        !           553:        fp->f_flag = FREAD|FWRITE;
        !           554:        fp->f_type = DTYPE_SOCKET;
        !           555:        fp->f_ops = &socketops;
        !           556:        error = socreate(domain, &so, type, protocol, l);
        !           557:        if (error != 0) {
        !           558:                FILE_UNUSE(fp, l);
        !           559:                fdremove(fdp, fd);
        !           560:                ffree(fp);
        !           561:        } else {
        !           562:                if (sop != NULL)
        !           563:                        *sop = so;
        !           564:                fp->f_data = so;
        !           565:                FILE_SET_MATURE(fp);
        !           566:                FILE_UNUSE(fp, l);
        !           567:                *fdout = fd;
        !           568:        }
        !           569:        return error;
        !           570: }
        !           571:
1.3       andrew    572: int
1.111.2.3  yamt      573: sobind(struct socket *so, struct mbuf *nam, struct lwp *l)
1.1       cgd       574: {
1.54      lukem     575:        int     s, error;
1.1       cgd       576:
1.54      lukem     577:        s = splsoftnet();
1.111.2.7  yamt      578:        error = (*so->so_proto->pr_usrreq)(so, PRU_BIND, NULL, nam, NULL, l);
1.1       cgd       579:        splx(s);
1.111.2.7  yamt      580:        return error;
1.1       cgd       581: }
                    582:
1.3       andrew    583: int
1.54      lukem     584: solisten(struct socket *so, int backlog)
1.1       cgd       585: {
1.54      lukem     586:        int     s, error;
1.1       cgd       587:
1.54      lukem     588:        s = splsoftnet();
1.111.2.7  yamt      589:        error = (*so->so_proto->pr_usrreq)(so, PRU_LISTEN, NULL,
                    590:            NULL, NULL, NULL);
                    591:        if (error != 0) {
1.1       cgd       592:                splx(s);
1.111.2.7  yamt      593:                return error;
1.1       cgd       594:        }
1.63      matt      595:        if (TAILQ_EMPTY(&so->so_q))
1.1       cgd       596:                so->so_options |= SO_ACCEPTCONN;
                    597:        if (backlog < 0)
                    598:                backlog = 0;
1.49      jonathan  599:        so->so_qlimit = min(backlog, somaxconn);
1.1       cgd       600:        splx(s);
1.111.2.7  yamt      601:        return 0;
1.1       cgd       602: }
                    603:
1.21      christos  604: void
1.54      lukem     605: sofree(struct socket *so)
1.1       cgd       606: {
                    607:
1.43      mycroft   608:        if (so->so_pcb || (so->so_state & SS_NOFDREF) == 0)
1.1       cgd       609:                return;
1.43      mycroft   610:        if (so->so_head) {
                    611:                /*
                    612:                 * We must not decommission a socket that's on the accept(2)
                    613:                 * queue.  If we do, then accept(2) may hang after select(2)
                    614:                 * indicated that the listening socket was ready.
                    615:                 */
                    616:                if (!soqremque(so, 0))
                    617:                        return;
                    618:        }
1.98      christos  619:        if (so->so_rcv.sb_hiwat)
1.110     christos  620:                (void)chgsbsize(so->so_uidinfo, &so->so_rcv.sb_hiwat, 0,
1.98      christos  621:                    RLIM_INFINITY);
                    622:        if (so->so_snd.sb_hiwat)
1.110     christos  623:                (void)chgsbsize(so->so_uidinfo, &so->so_snd.sb_hiwat, 0,
1.98      christos  624:                    RLIM_INFINITY);
                    625:        sbrelease(&so->so_snd, so);
1.1       cgd       626:        sorflush(so);
1.111.2.8! yamt      627:        seldestroy(&so->so_rcv.sb_sel);
        !           628:        seldestroy(&so->so_snd.sb_sel);
1.37      thorpej   629:        pool_put(&socket_pool, so);
1.1       cgd       630: }
                    631:
                    632: /*
                    633:  * Close a socket on last file table reference removal.
                    634:  * Initiate disconnect if connected.
                    635:  * Free socket when disconnect complete.
                    636:  */
1.3       andrew    637: int
1.54      lukem     638: soclose(struct socket *so)
1.1       cgd       639: {
1.54      lukem     640:        struct socket   *so2;
                    641:        int             s, error;
1.1       cgd       642:
1.54      lukem     643:        error = 0;
                    644:        s = splsoftnet();               /* conservative */
1.1       cgd       645:        if (so->so_options & SO_ACCEPTCONN) {
1.63      matt      646:                while ((so2 = TAILQ_FIRST(&so->so_q0)) != 0) {
1.42      mycroft   647:                        (void) soqremque(so2, 0);
1.41      mycroft   648:                        (void) soabort(so2);
                    649:                }
1.63      matt      650:                while ((so2 = TAILQ_FIRST(&so->so_q)) != 0) {
1.42      mycroft   651:                        (void) soqremque(so2, 1);
1.41      mycroft   652:                        (void) soabort(so2);
                    653:                }
1.1       cgd       654:        }
                    655:        if (so->so_pcb == 0)
                    656:                goto discard;
                    657:        if (so->so_state & SS_ISCONNECTED) {
                    658:                if ((so->so_state & SS_ISDISCONNECTING) == 0) {
                    659:                        error = sodisconnect(so);
                    660:                        if (error)
                    661:                                goto drop;
                    662:                }
                    663:                if (so->so_options & SO_LINGER) {
                    664:                        if ((so->so_state & SS_ISDISCONNECTING) &&
                    665:                            (so->so_state & SS_NBIO))
                    666:                                goto drop;
1.21      christos  667:                        while (so->so_state & SS_ISCONNECTED) {
1.111.2.7  yamt      668:                                error = tsleep((void *)&so->so_timeo,
1.21      christos  669:                                               PSOCK | PCATCH, netcls,
1.30      thorpej   670:                                               so->so_linger * hz);
1.21      christos  671:                                if (error)
1.1       cgd       672:                                        break;
1.21      christos  673:                        }
1.1       cgd       674:                }
                    675:        }
1.54      lukem     676:  drop:
1.1       cgd       677:        if (so->so_pcb) {
1.22      mycroft   678:                int error2 = (*so->so_proto->pr_usrreq)(so, PRU_DETACH,
1.111.2.7  yamt      679:                    NULL, NULL, NULL, NULL);
1.1       cgd       680:                if (error == 0)
                    681:                        error = error2;
                    682:        }
1.54      lukem     683:  discard:
1.1       cgd       684:        if (so->so_state & SS_NOFDREF)
                    685:                panic("soclose: NOFDREF");
                    686:        so->so_state |= SS_NOFDREF;
                    687:        sofree(so);
                    688:        splx(s);
                    689:        return (error);
                    690: }
                    691:
                    692: /*
1.20      mycroft   693:  * Must be called at splsoftnet...
1.1       cgd       694:  */
1.3       andrew    695: int
1.54      lukem     696: soabort(struct socket *so)
1.1       cgd       697: {
1.111.2.7  yamt      698:        int error;
1.1       cgd       699:
1.111.2.7  yamt      700:        KASSERT(so->so_head == NULL);
                    701:        error = (*so->so_proto->pr_usrreq)(so, PRU_ABORT, NULL,
                    702:            NULL, NULL, NULL);
                    703:        if (error) {
                    704:                sofree(so);
                    705:        }
                    706:        return error;
1.1       cgd       707: }
                    708:
1.3       andrew    709: int
1.54      lukem     710: soaccept(struct socket *so, struct mbuf *nam)
1.1       cgd       711: {
1.54      lukem     712:        int     s, error;
1.1       cgd       713:
1.54      lukem     714:        error = 0;
                    715:        s = splsoftnet();
1.1       cgd       716:        if ((so->so_state & SS_NOFDREF) == 0)
                    717:                panic("soaccept: !NOFDREF");
                    718:        so->so_state &= ~SS_NOFDREF;
1.55      thorpej   719:        if ((so->so_state & SS_ISDISCONNECTED) == 0 ||
                    720:            (so->so_proto->pr_flags & PR_ABRTACPTDIS) == 0)
1.41      mycroft   721:                error = (*so->so_proto->pr_usrreq)(so, PRU_ACCEPT,
1.111.2.7  yamt      722:                    NULL, nam, NULL, NULL);
1.41      mycroft   723:        else
1.53      itojun    724:                error = ECONNABORTED;
1.52      itojun    725:
1.1       cgd       726:        splx(s);
                    727:        return (error);
                    728: }
                    729:
1.3       andrew    730: int
1.111.2.3  yamt      731: soconnect(struct socket *so, struct mbuf *nam, struct lwp *l)
1.1       cgd       732: {
1.54      lukem     733:        int             s, error;
1.1       cgd       734:
                    735:        if (so->so_options & SO_ACCEPTCONN)
                    736:                return (EOPNOTSUPP);
1.20      mycroft   737:        s = splsoftnet();
1.1       cgd       738:        /*
                    739:         * If protocol is connection-based, can only connect once.
                    740:         * Otherwise, if connected, try to disconnect first.
                    741:         * This allows user to disconnect by connecting to, e.g.,
                    742:         * a null address.
                    743:         */
                    744:        if (so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING) &&
                    745:            ((so->so_proto->pr_flags & PR_CONNREQUIRED) ||
                    746:            (error = sodisconnect(so))))
                    747:                error = EISCONN;
                    748:        else
                    749:                error = (*so->so_proto->pr_usrreq)(so, PRU_CONNECT,
1.111.2.7  yamt      750:                    NULL, nam, NULL, l);
1.1       cgd       751:        splx(s);
                    752:        return (error);
                    753: }
                    754:
1.3       andrew    755: int
1.54      lukem     756: soconnect2(struct socket *so1, struct socket *so2)
1.1       cgd       757: {
1.54      lukem     758:        int     s, error;
1.1       cgd       759:
1.54      lukem     760:        s = splsoftnet();
1.22      mycroft   761:        error = (*so1->so_proto->pr_usrreq)(so1, PRU_CONNECT2,
1.111.2.7  yamt      762:            NULL, (struct mbuf *)so2, NULL, NULL);
1.1       cgd       763:        splx(s);
                    764:        return (error);
                    765: }
                    766:
1.3       andrew    767: int
1.54      lukem     768: sodisconnect(struct socket *so)
1.1       cgd       769: {
1.54      lukem     770:        int     s, error;
1.1       cgd       771:
1.54      lukem     772:        s = splsoftnet();
1.1       cgd       773:        if ((so->so_state & SS_ISCONNECTED) == 0) {
                    774:                error = ENOTCONN;
                    775:                goto bad;
                    776:        }
                    777:        if (so->so_state & SS_ISDISCONNECTING) {
                    778:                error = EALREADY;
                    779:                goto bad;
                    780:        }
1.22      mycroft   781:        error = (*so->so_proto->pr_usrreq)(so, PRU_DISCONNECT,
1.111.2.7  yamt      782:            NULL, NULL, NULL, NULL);
1.54      lukem     783:  bad:
1.1       cgd       784:        splx(s);
1.111.2.3  yamt      785:        sodopendfree();
1.1       cgd       786:        return (error);
                    787: }
                    788:
1.15      mycroft   789: #define        SBLOCKWAIT(f)   (((f) & MSG_DONTWAIT) ? M_NOWAIT : M_WAITOK)
1.1       cgd       790: /*
                    791:  * Send on a socket.
                    792:  * If send must go all at once and message is larger than
                    793:  * send buffering, then hard error.
                    794:  * Lock against other senders.
                    795:  * If must go all at once and not enough room now, then
                    796:  * inform user that this would block and do nothing.
                    797:  * Otherwise, if nonblocking, send as much as possible.
                    798:  * The data to be sent is described by "uio" if nonzero,
                    799:  * otherwise by the mbuf chain "top" (which must be null
                    800:  * if uio is not).  Data provided in mbuf chain must be small
                    801:  * enough to send all at once.
                    802:  *
                    803:  * Returns nonzero on error, timeout or signal; callers
                    804:  * must check for short counts if EINTR/ERESTART are returned.
                    805:  * Data and control buffers are freed on return.
                    806:  */
1.3       andrew    807: int
1.54      lukem     808: sosend(struct socket *so, struct mbuf *addr, struct uio *uio, struct mbuf *top,
1.111.2.3  yamt      809:        struct mbuf *control, int flags, struct lwp *l)
1.1       cgd       810: {
1.54      lukem     811:        struct mbuf     **mp, *m;
1.111.2.3  yamt      812:        struct proc     *p;
1.58      jdolecek  813:        long            space, len, resid, clen, mlen;
                    814:        int             error, s, dontroute, atomic;
1.54      lukem     815:
1.111.2.3  yamt      816:        p = l->l_proc;
                    817:        sodopendfree();
1.64      thorpej   818:
1.54      lukem     819:        clen = 0;
                    820:        atomic = sosendallatonce(so) || top;
1.1       cgd       821:        if (uio)
                    822:                resid = uio->uio_resid;
                    823:        else
                    824:                resid = top->m_pkthdr.len;
1.7       cgd       825:        /*
                    826:         * In theory resid should be unsigned.
                    827:         * However, space must be signed, as it might be less than 0
                    828:         * if we over-committed, and we must use a signed comparison
                    829:         * of space and resid.  On the other hand, a negative resid
                    830:         * causes us to loop sending 0-length segments to the protocol.
                    831:         */
1.29      mycroft   832:        if (resid < 0) {
                    833:                error = EINVAL;
                    834:                goto out;
                    835:        }
1.1       cgd       836:        dontroute =
                    837:            (flags & MSG_DONTROUTE) && (so->so_options & SO_DONTROUTE) == 0 &&
                    838:            (so->so_proto->pr_flags & PR_ATOMIC);
1.102     jonathan  839:        if (p)
                    840:                p->p_stats->p_ru.ru_msgsnd++;
1.1       cgd       841:        if (control)
                    842:                clen = control->m_len;
                    843: #define        snderr(errno)   { error = errno; splx(s); goto release; }
                    844:
1.54      lukem     845:  restart:
1.21      christos  846:        if ((error = sblock(&so->so_snd, SBLOCKWAIT(flags))) != 0)
1.1       cgd       847:                goto out;
                    848:        do {
1.20      mycroft   849:                s = splsoftnet();
1.1       cgd       850:                if (so->so_state & SS_CANTSENDMORE)
                    851:                        snderr(EPIPE);
1.48      thorpej   852:                if (so->so_error) {
                    853:                        error = so->so_error;
                    854:                        so->so_error = 0;
                    855:                        splx(s);
                    856:                        goto release;
                    857:                }
1.1       cgd       858:                if ((so->so_state & SS_ISCONNECTED) == 0) {
                    859:                        if (so->so_proto->pr_flags & PR_CONNREQUIRED) {
                    860:                                if ((so->so_state & SS_ISCONFIRMING) == 0 &&
                    861:                                    !(resid == 0 && clen != 0))
                    862:                                        snderr(ENOTCONN);
                    863:                        } else if (addr == 0)
                    864:                                snderr(EDESTADDRREQ);
                    865:                }
                    866:                space = sbspace(&so->so_snd);
                    867:                if (flags & MSG_OOB)
                    868:                        space += 1024;
1.21      christos  869:                if ((atomic && resid > so->so_snd.sb_hiwat) ||
1.11      mycroft   870:                    clen > so->so_snd.sb_hiwat)
                    871:                        snderr(EMSGSIZE);
1.96      mycroft   872:                if (space < resid + clen &&
1.1       cgd       873:                    (atomic || space < so->so_snd.sb_lowat || space < clen)) {
                    874:                        if (so->so_state & SS_NBIO)
                    875:                                snderr(EWOULDBLOCK);
                    876:                        sbunlock(&so->so_snd);
                    877:                        error = sbwait(&so->so_snd);
                    878:                        splx(s);
                    879:                        if (error)
                    880:                                goto out;
                    881:                        goto restart;
                    882:                }
                    883:                splx(s);
                    884:                mp = &top;
                    885:                space -= clen;
                    886:                do {
1.45      tv        887:                        if (uio == NULL) {
                    888:                                /*
                    889:                                 * Data is prepackaged in "top".
                    890:                                 */
                    891:                                resid = 0;
                    892:                                if (flags & MSG_EOR)
                    893:                                        top->m_flags |= M_EOR;
                    894:                        } else do {
1.111.2.8! yamt      895:                                if (top == NULL) {
1.78      matt      896:                                        m = m_gethdr(M_WAIT, MT_DATA);
1.45      tv        897:                                        mlen = MHLEN;
                    898:                                        m->m_pkthdr.len = 0;
1.111.2.7  yamt      899:                                        m->m_pkthdr.rcvif = NULL;
1.45      tv        900:                                } else {
1.78      matt      901:                                        m = m_get(M_WAIT, MT_DATA);
1.45      tv        902:                                        mlen = MLEN;
                    903:                                }
1.78      matt      904:                                MCLAIM(m, so->so_snd.sb_mowner);
1.111.2.3  yamt      905:                                if (sock_loan_thresh >= 0 &&
                    906:                                    uio->uio_iov->iov_len >= sock_loan_thresh &&
                    907:                                    space >= sock_loan_thresh &&
1.64      thorpej   908:                                    (len = sosend_loan(so, uio, m,
                    909:                                                       space)) != 0) {
                    910:                                        SOSEND_COUNTER_INCR(&sosend_loan_big);
                    911:                                        space -= len;
                    912:                                        goto have_data;
                    913:                                }
1.45      tv        914:                                if (resid >= MINCLSIZE && space >= MCLBYTES) {
1.64      thorpej   915:                                        SOSEND_COUNTER_INCR(&sosend_copy_big);
1.78      matt      916:                                        m_clget(m, M_WAIT);
1.45      tv        917:                                        if ((m->m_flags & M_EXT) == 0)
                    918:                                                goto nopages;
                    919:                                        mlen = MCLBYTES;
                    920:                                        if (atomic && top == 0) {
1.58      jdolecek  921:                                                len = lmin(MCLBYTES - max_hdr,
1.54      lukem     922:                                                    resid);
1.45      tv        923:                                                m->m_data += max_hdr;
                    924:                                        } else
1.58      jdolecek  925:                                                len = lmin(MCLBYTES, resid);
1.45      tv        926:                                        space -= len;
                    927:                                } else {
1.64      thorpej   928:  nopages:
                    929:                                        SOSEND_COUNTER_INCR(&sosend_copy_small);
1.58      jdolecek  930:                                        len = lmin(lmin(mlen, resid), space);
1.45      tv        931:                                        space -= len;
                    932:                                        /*
                    933:                                         * For datagram protocols, leave room
                    934:                                         * for protocol headers in first mbuf.
                    935:                                         */
                    936:                                        if (atomic && top == 0 && len < mlen)
                    937:                                                MH_ALIGN(m, len);
                    938:                                }
1.111.2.8! yamt      939:                                error = uiomove(mtod(m, void *), (int)len, uio);
1.64      thorpej   940:  have_data:
1.45      tv        941:                                resid = uio->uio_resid;
                    942:                                m->m_len = len;
                    943:                                *mp = m;
                    944:                                top->m_pkthdr.len += len;
1.111.2.8! yamt      945:                                if (error != 0)
1.45      tv        946:                                        goto release;
                    947:                                mp = &m->m_next;
                    948:                                if (resid <= 0) {
                    949:                                        if (flags & MSG_EOR)
                    950:                                                top->m_flags |= M_EOR;
                    951:                                        break;
                    952:                                }
                    953:                        } while (space > 0 && atomic);
1.108     perry     954:
1.46      sommerfe  955:                        s = splsoftnet();
                    956:
                    957:                        if (so->so_state & SS_CANTSENDMORE)
                    958:                                snderr(EPIPE);
1.45      tv        959:
                    960:                        if (dontroute)
                    961:                                so->so_options |= SO_DONTROUTE;
                    962:                        if (resid > 0)
                    963:                                so->so_state |= SS_MORETOCOME;
1.46      sommerfe  964:                        error = (*so->so_proto->pr_usrreq)(so,
                    965:                            (flags & MSG_OOB) ? PRU_SENDOOB : PRU_SEND,
1.111.2.3  yamt      966:                            top, addr, control, curlwp);        /* XXX */
1.45      tv        967:                        if (dontroute)
                    968:                                so->so_options &= ~SO_DONTROUTE;
                    969:                        if (resid > 0)
                    970:                                so->so_state &= ~SS_MORETOCOME;
1.46      sommerfe  971:                        splx(s);
                    972:
1.45      tv        973:                        clen = 0;
1.111.2.8! yamt      974:                        control = NULL;
        !           975:                        top = NULL;
1.45      tv        976:                        mp = &top;
1.111.2.8! yamt      977:                        if (error != 0)
1.1       cgd       978:                                goto release;
                    979:                } while (resid && space > 0);
                    980:        } while (resid);
                    981:
1.54      lukem     982:  release:
1.1       cgd       983:        sbunlock(&so->so_snd);
1.54      lukem     984:  out:
1.1       cgd       985:        if (top)
                    986:                m_freem(top);
                    987:        if (control)
                    988:                m_freem(control);
                    989:        return (error);
                    990: }
                    991:
                    992: /*
                    993:  * Implement receive operations on a socket.
                    994:  * We depend on the way that records are added to the sockbuf
                    995:  * by sbappend*.  In particular, each record (mbufs linked through m_next)
                    996:  * must begin with an address if the protocol so specifies,
                    997:  * followed by an optional mbuf or mbufs containing ancillary data,
                    998:  * and then zero or more mbufs of data.
                    999:  * In order to avoid blocking network interrupts for the entire time here,
                   1000:  * we splx() while doing the actual copy to user space.
                   1001:  * Although the sockbuf is locked, new data may still be appended,
                   1002:  * and thus we must maintain consistency of the sockbuf during that time.
                   1003:  *
                   1004:  * The caller may receive the data as a single mbuf chain by supplying
                   1005:  * an mbuf **mp0 for use in returning the chain.  The uio is then used
                   1006:  * only for the count in uio_resid.
                   1007:  */
1.3       andrew   1008: int
1.54      lukem    1009: soreceive(struct socket *so, struct mbuf **paddr, struct uio *uio,
                   1010:        struct mbuf **mp0, struct mbuf **controlp, int *flagsp)
1.1       cgd      1011: {
1.111.2.3  yamt     1012:        struct lwp *l = curlwp;
1.54      lukem    1013:        struct mbuf     *m, **mp;
                   1014:        int             flags, len, error, s, offset, moff, type, orig_resid;
1.99      matt     1015:        const struct protosw    *pr;
1.54      lukem    1016:        struct mbuf     *nextrecord;
1.67      he       1017:        int             mbuf_removed = 0;
1.64      thorpej  1018:
1.54      lukem    1019:        pr = so->so_proto;
1.1       cgd      1020:        mp = mp0;
1.54      lukem    1021:        type = 0;
                   1022:        orig_resid = uio->uio_resid;
1.102     jonathan 1023:
1.111.2.8! yamt     1024:        if (paddr != NULL)
        !          1025:                *paddr = NULL;
        !          1026:        if (controlp != NULL)
        !          1027:                *controlp = NULL;
        !          1028:        if (flagsp != NULL)
1.1       cgd      1029:                flags = *flagsp &~ MSG_EOR;
                   1030:        else
                   1031:                flags = 0;
1.66      enami    1032:
                   1033:        if ((flags & MSG_DONTWAIT) == 0)
1.111.2.3  yamt     1034:                sodopendfree();
1.66      enami    1035:
1.1       cgd      1036:        if (flags & MSG_OOB) {
                   1037:                m = m_get(M_WAIT, MT_DATA);
1.17      cgd      1038:                error = (*pr->pr_usrreq)(so, PRU_RCVOOB, m,
1.111.2.7  yamt     1039:                    (struct mbuf *)(long)(flags & MSG_PEEK), NULL, l);
1.1       cgd      1040:                if (error)
                   1041:                        goto bad;
                   1042:                do {
1.111.2.7  yamt     1043:                        error = uiomove(mtod(m, void *),
1.1       cgd      1044:                            (int) min(uio->uio_resid, m->m_len), uio);
                   1045:                        m = m_free(m);
1.111.2.8! yamt     1046:                } while (uio->uio_resid > 0 && error == 0 && m);
1.54      lukem    1047:  bad:
1.111.2.8! yamt     1048:                if (m != NULL)
1.1       cgd      1049:                        m_freem(m);
1.111.2.8! yamt     1050:                return error;
1.1       cgd      1051:        }
1.111.2.8! yamt     1052:        if (mp != NULL)
1.111.2.7  yamt     1053:                *mp = NULL;
1.1       cgd      1054:        if (so->so_state & SS_ISCONFIRMING && uio->uio_resid)
1.111.2.7  yamt     1055:                (*pr->pr_usrreq)(so, PRU_RCVD, NULL, NULL, NULL, l);
1.1       cgd      1056:
1.54      lukem    1057:  restart:
1.21      christos 1058:        if ((error = sblock(&so->so_rcv, SBLOCKWAIT(flags))) != 0)
1.111.2.8! yamt     1059:                return error;
1.20      mycroft  1060:        s = splsoftnet();
1.1       cgd      1061:
                   1062:        m = so->so_rcv.sb_mb;
                   1063:        /*
                   1064:         * If we have less data than requested, block awaiting more
                   1065:         * (subject to any timeout) if:
1.15      mycroft  1066:         *   1. the current count is less than the low water mark,
1.1       cgd      1067:         *   2. MSG_WAITALL is set, and it is possible to do the entire
1.15      mycroft  1068:         *      receive operation at once if we block (resid <= hiwat), or
                   1069:         *   3. MSG_DONTWAIT is not set.
1.1       cgd      1070:         * If MSG_WAITALL is set but resid is larger than the receive buffer,
                   1071:         * we have to do the receive in sections, and thus risk returning
                   1072:         * a short count if a timeout or signal occurs after we start.
                   1073:         */
1.111.2.8! yamt     1074:        if (m == NULL ||
        !          1075:            ((flags & MSG_DONTWAIT) == 0 &&
        !          1076:             so->so_rcv.sb_cc < uio->uio_resid &&
        !          1077:             (so->so_rcv.sb_cc < so->so_rcv.sb_lowat ||
        !          1078:              ((flags & MSG_WAITALL) &&
        !          1079:               uio->uio_resid <= so->so_rcv.sb_hiwat)) &&
        !          1080:             m->m_nextpkt == NULL &&
        !          1081:             (pr->pr_flags & PR_ATOMIC) == 0)) {
1.1       cgd      1082: #ifdef DIAGNOSTIC
1.111.2.8! yamt     1083:                if (m == NULL && so->so_rcv.sb_cc)
1.1       cgd      1084:                        panic("receive 1");
                   1085: #endif
                   1086:                if (so->so_error) {
1.111.2.8! yamt     1087:                        if (m != NULL)
1.15      mycroft  1088:                                goto dontblock;
1.1       cgd      1089:                        error = so->so_error;
                   1090:                        if ((flags & MSG_PEEK) == 0)
                   1091:                                so->so_error = 0;
                   1092:                        goto release;
                   1093:                }
                   1094:                if (so->so_state & SS_CANTRCVMORE) {
1.111.2.8! yamt     1095:                        if (m != NULL)
1.15      mycroft  1096:                                goto dontblock;
1.1       cgd      1097:                        else
                   1098:                                goto release;
                   1099:                }
1.111.2.8! yamt     1100:                for (; m != NULL; m = m->m_next)
1.1       cgd      1101:                        if (m->m_type == MT_OOBDATA  || (m->m_flags & M_EOR)) {
                   1102:                                m = so->so_rcv.sb_mb;
                   1103:                                goto dontblock;
                   1104:                        }
                   1105:                if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) == 0 &&
                   1106:                    (so->so_proto->pr_flags & PR_CONNREQUIRED)) {
                   1107:                        error = ENOTCONN;
                   1108:                        goto release;
                   1109:                }
                   1110:                if (uio->uio_resid == 0)
                   1111:                        goto release;
1.15      mycroft  1112:                if ((so->so_state & SS_NBIO) || (flags & MSG_DONTWAIT)) {
1.1       cgd      1113:                        error = EWOULDBLOCK;
                   1114:                        goto release;
                   1115:                }
1.69      thorpej  1116:                SBLASTRECORDCHK(&so->so_rcv, "soreceive sbwait 1");
                   1117:                SBLASTMBUFCHK(&so->so_rcv, "soreceive sbwait 1");
1.1       cgd      1118:                sbunlock(&so->so_rcv);
                   1119:                error = sbwait(&so->so_rcv);
                   1120:                splx(s);
1.111.2.8! yamt     1121:                if (error != 0)
        !          1122:                        return error;
1.1       cgd      1123:                goto restart;
                   1124:        }
1.54      lukem    1125:  dontblock:
1.69      thorpej  1126:        /*
                   1127:         * On entry here, m points to the first record of the socket buffer.
                   1128:         * While we process the initial mbufs containing address and control
                   1129:         * info, we save a copy of m->m_nextpkt into nextrecord.
                   1130:         */
1.111.2.8! yamt     1131:        if (l != NULL)
1.111.2.3  yamt     1132:                l->l_proc->p_stats->p_ru.ru_msgrcv++;
1.69      thorpej  1133:        KASSERT(m == so->so_rcv.sb_mb);
                   1134:        SBLASTRECORDCHK(&so->so_rcv, "soreceive 1");
                   1135:        SBLASTMBUFCHK(&so->so_rcv, "soreceive 1");
1.1       cgd      1136:        nextrecord = m->m_nextpkt;
                   1137:        if (pr->pr_flags & PR_ADDR) {
                   1138: #ifdef DIAGNOSTIC
                   1139:                if (m->m_type != MT_SONAME)
                   1140:                        panic("receive 1a");
                   1141: #endif
1.3       andrew   1142:                orig_resid = 0;
1.1       cgd      1143:                if (flags & MSG_PEEK) {
                   1144:                        if (paddr)
                   1145:                                *paddr = m_copy(m, 0, m->m_len);
                   1146:                        m = m->m_next;
                   1147:                } else {
                   1148:                        sbfree(&so->so_rcv, m);
1.67      he       1149:                        mbuf_removed = 1;
1.111.2.8! yamt     1150:                        if (paddr != NULL) {
1.1       cgd      1151:                                *paddr = m;
                   1152:                                so->so_rcv.sb_mb = m->m_next;
1.111.2.8! yamt     1153:                                m->m_next = NULL;
1.1       cgd      1154:                                m = so->so_rcv.sb_mb;
                   1155:                        } else {
                   1156:                                MFREE(m, so->so_rcv.sb_mb);
                   1157:                                m = so->so_rcv.sb_mb;
                   1158:                        }
                   1159:                }
                   1160:        }
1.111.2.8! yamt     1161:        while (m != NULL && m->m_type == MT_CONTROL && error == 0) {
1.1       cgd      1162:                if (flags & MSG_PEEK) {
1.111.2.8! yamt     1163:                        if (controlp != NULL)
1.1       cgd      1164:                                *controlp = m_copy(m, 0, m->m_len);
                   1165:                        m = m->m_next;
                   1166:                } else {
                   1167:                        sbfree(&so->so_rcv, m);
1.67      he       1168:                        mbuf_removed = 1;
1.111.2.8! yamt     1169:                        if (controlp != NULL) {
1.102     jonathan 1170:                                struct domain *dom = pr->pr_domain;
1.111.2.3  yamt     1171:                                if (dom->dom_externalize && l &&
1.1       cgd      1172:                                    mtod(m, struct cmsghdr *)->cmsg_type ==
                   1173:                                    SCM_RIGHTS)
1.111.2.3  yamt     1174:                                        error = (*dom->dom_externalize)(m, l);
1.1       cgd      1175:                                *controlp = m;
                   1176:                                so->so_rcv.sb_mb = m->m_next;
1.111.2.8! yamt     1177:                                m->m_next = NULL;
1.1       cgd      1178:                                m = so->so_rcv.sb_mb;
                   1179:                        } else {
1.106     itojun   1180:                                /*
                   1181:                                 * Dispose of any SCM_RIGHTS message that went
                   1182:                                 * through the read path rather than recv.
                   1183:                                 */
                   1184:                                if (pr->pr_domain->dom_dispose &&
                   1185:                                    mtod(m, struct cmsghdr *)->cmsg_type == SCM_RIGHTS)
                   1186:                                        (*pr->pr_domain->dom_dispose)(m);
1.1       cgd      1187:                                MFREE(m, so->so_rcv.sb_mb);
                   1188:                                m = so->so_rcv.sb_mb;
                   1189:                        }
                   1190:                }
1.111.2.8! yamt     1191:                if (controlp != NULL) {
1.3       andrew   1192:                        orig_resid = 0;
1.1       cgd      1193:                        controlp = &(*controlp)->m_next;
1.3       andrew   1194:                }
1.1       cgd      1195:        }
1.69      thorpej  1196:
                   1197:        /*
                   1198:         * If m is non-NULL, we have some data to read.  From now on,
                   1199:         * make sure to keep sb_lastrecord consistent when working on
                   1200:         * the last packet on the chain (nextrecord == NULL) and we
                   1201:         * change m->m_nextpkt.
                   1202:         */
1.111.2.8! yamt     1203:        if (m != NULL) {
1.69      thorpej  1204:                if ((flags & MSG_PEEK) == 0) {
1.1       cgd      1205:                        m->m_nextpkt = nextrecord;
1.69      thorpej  1206:                        /*
                   1207:                         * If nextrecord == NULL (this is a single chain),
                   1208:                         * then sb_lastrecord may not be valid here if m
                   1209:                         * was changed earlier.
                   1210:                         */
                   1211:                        if (nextrecord == NULL) {
                   1212:                                KASSERT(so->so_rcv.sb_mb == m);
                   1213:                                so->so_rcv.sb_lastrecord = m;
                   1214:                        }
                   1215:                }
1.1       cgd      1216:                type = m->m_type;
                   1217:                if (type == MT_OOBDATA)
                   1218:                        flags |= MSG_OOB;
1.69      thorpej  1219:        } else {
                   1220:                if ((flags & MSG_PEEK) == 0) {
                   1221:                        KASSERT(so->so_rcv.sb_mb == m);
                   1222:                        so->so_rcv.sb_mb = nextrecord;
1.70      thorpej  1223:                        SB_EMPTY_FIXUP(&so->so_rcv);
1.69      thorpej  1224:                }
1.1       cgd      1225:        }
1.69      thorpej  1226:        SBLASTRECORDCHK(&so->so_rcv, "soreceive 2");
                   1227:        SBLASTMBUFCHK(&so->so_rcv, "soreceive 2");
                   1228:
1.1       cgd      1229:        moff = 0;
                   1230:        offset = 0;
1.111.2.8! yamt     1231:        while (m != NULL && uio->uio_resid > 0 && error == 0) {
1.1       cgd      1232:                if (m->m_type == MT_OOBDATA) {
                   1233:                        if (type != MT_OOBDATA)
                   1234:                                break;
                   1235:                } else if (type == MT_OOBDATA)
                   1236:                        break;
                   1237: #ifdef DIAGNOSTIC
                   1238:                else if (m->m_type != MT_DATA && m->m_type != MT_HEADER)
                   1239:                        panic("receive 3");
                   1240: #endif
                   1241:                so->so_state &= ~SS_RCVATMARK;
                   1242:                len = uio->uio_resid;
                   1243:                if (so->so_oobmark && len > so->so_oobmark - offset)
                   1244:                        len = so->so_oobmark - offset;
                   1245:                if (len > m->m_len - moff)
                   1246:                        len = m->m_len - moff;
                   1247:                /*
                   1248:                 * If mp is set, just pass back the mbufs.
                   1249:                 * Otherwise copy them out via the uio, then free.
                   1250:                 * Sockbuf must be consistent here (points to current mbuf,
                   1251:                 * it points to next record) when we drop priority;
                   1252:                 * we must note any additions to the sockbuf when we
                   1253:                 * block interrupts again.
                   1254:                 */
1.111.2.8! yamt     1255:                if (mp == NULL) {
1.69      thorpej  1256:                        SBLASTRECORDCHK(&so->so_rcv, "soreceive uiomove");
                   1257:                        SBLASTMBUFCHK(&so->so_rcv, "soreceive uiomove");
1.1       cgd      1258:                        splx(s);
1.111.2.7  yamt     1259:                        error = uiomove(mtod(m, char *) + moff, (int)len, uio);
1.20      mycroft  1260:                        s = splsoftnet();
1.111.2.8! yamt     1261:                        if (error != 0) {
1.67      he       1262:                                /*
                   1263:                                 * If any part of the record has been removed
                   1264:                                 * (such as the MT_SONAME mbuf, which will
                   1265:                                 * happen when PR_ADDR, and thus also
                   1266:                                 * PR_ATOMIC, is set), then drop the entire
                   1267:                                 * record to maintain the atomicity of the
                   1268:                                 * receive operation.
                   1269:                                 *
                   1270:                                 * This avoids a later panic("receive 1a")
                   1271:                                 * when compiled with DIAGNOSTIC.
                   1272:                                 */
                   1273:                                if (m && mbuf_removed
                   1274:                                    && (pr->pr_flags & PR_ATOMIC))
                   1275:                                        (void) sbdroprecord(&so->so_rcv);
                   1276:
1.57      jdolecek 1277:                                goto release;
1.67      he       1278:                        }
1.1       cgd      1279:                } else
                   1280:                        uio->uio_resid -= len;
                   1281:                if (len == m->m_len - moff) {
                   1282:                        if (m->m_flags & M_EOR)
                   1283:                                flags |= MSG_EOR;
                   1284:                        if (flags & MSG_PEEK) {
                   1285:                                m = m->m_next;
                   1286:                                moff = 0;
                   1287:                        } else {
                   1288:                                nextrecord = m->m_nextpkt;
                   1289:                                sbfree(&so->so_rcv, m);
                   1290:                                if (mp) {
                   1291:                                        *mp = m;
                   1292:                                        mp = &m->m_next;
                   1293:                                        so->so_rcv.sb_mb = m = m->m_next;
1.111.2.7  yamt     1294:                                        *mp = NULL;
1.1       cgd      1295:                                } else {
                   1296:                                        MFREE(m, so->so_rcv.sb_mb);
                   1297:                                        m = so->so_rcv.sb_mb;
                   1298:                                }
1.69      thorpej  1299:                                /*
                   1300:                                 * If m != NULL, we also know that
                   1301:                                 * so->so_rcv.sb_mb != NULL.
                   1302:                                 */
                   1303:                                KASSERT(so->so_rcv.sb_mb == m);
                   1304:                                if (m) {
1.1       cgd      1305:                                        m->m_nextpkt = nextrecord;
1.69      thorpej  1306:                                        if (nextrecord == NULL)
                   1307:                                                so->so_rcv.sb_lastrecord = m;
                   1308:                                } else {
                   1309:                                        so->so_rcv.sb_mb = nextrecord;
1.70      thorpej  1310:                                        SB_EMPTY_FIXUP(&so->so_rcv);
1.69      thorpej  1311:                                }
                   1312:                                SBLASTRECORDCHK(&so->so_rcv, "soreceive 3");
                   1313:                                SBLASTMBUFCHK(&so->so_rcv, "soreceive 3");
1.1       cgd      1314:                        }
1.111.2.8! yamt     1315:                } else if (flags & MSG_PEEK)
        !          1316:                        moff += len;
        !          1317:                else {
        !          1318:                        if (mp != NULL)
        !          1319:                                *mp = m_copym(m, 0, len, M_WAIT);
        !          1320:                        m->m_data += len;
        !          1321:                        m->m_len -= len;
        !          1322:                        so->so_rcv.sb_cc -= len;
1.1       cgd      1323:                }
                   1324:                if (so->so_oobmark) {
                   1325:                        if ((flags & MSG_PEEK) == 0) {
                   1326:                                so->so_oobmark -= len;
                   1327:                                if (so->so_oobmark == 0) {
                   1328:                                        so->so_state |= SS_RCVATMARK;
                   1329:                                        break;
                   1330:                                }
1.7       cgd      1331:                        } else {
1.1       cgd      1332:                                offset += len;
1.7       cgd      1333:                                if (offset == so->so_oobmark)
                   1334:                                        break;
                   1335:                        }
1.1       cgd      1336:                }
                   1337:                if (flags & MSG_EOR)
                   1338:                        break;
                   1339:                /*
                   1340:                 * If the MSG_WAITALL flag is set (for non-atomic socket),
                   1341:                 * we must not quit until "uio->uio_resid == 0" or an error
                   1342:                 * termination.  If a signal/timeout occurs, return
                   1343:                 * with a short count but without error.
                   1344:                 * Keep sockbuf locked against other readers.
                   1345:                 */
1.111.2.8! yamt     1346:                while (flags & MSG_WAITALL && m == NULL && uio->uio_resid > 0 &&
1.3       andrew   1347:                    !sosendallatonce(so) && !nextrecord) {
1.1       cgd      1348:                        if (so->so_error || so->so_state & SS_CANTRCVMORE)
                   1349:                                break;
1.68      matt     1350:                        /*
                   1351:                         * If we are peeking and the socket receive buffer is
                   1352:                         * full, stop since we can't get more data to peek at.
                   1353:                         */
                   1354:                        if ((flags & MSG_PEEK) && sbspace(&so->so_rcv) <= 0)
                   1355:                                break;
                   1356:                        /*
                   1357:                         * If we've drained the socket buffer, tell the
                   1358:                         * protocol in case it needs to do something to
                   1359:                         * get it filled again.
                   1360:                         */
                   1361:                        if ((pr->pr_flags & PR_WANTRCVD) && so->so_pcb)
                   1362:                                (*pr->pr_usrreq)(so, PRU_RCVD,
1.111.2.7  yamt     1363:                                    NULL, (struct mbuf *)(long)flags, NULL, l);
1.69      thorpej  1364:                        SBLASTRECORDCHK(&so->so_rcv, "soreceive sbwait 2");
                   1365:                        SBLASTMBUFCHK(&so->so_rcv, "soreceive sbwait 2");
1.1       cgd      1366:                        error = sbwait(&so->so_rcv);
1.111.2.8! yamt     1367:                        if (error != 0) {
1.1       cgd      1368:                                sbunlock(&so->so_rcv);
                   1369:                                splx(s);
1.111.2.8! yamt     1370:                                return 0;
1.1       cgd      1371:                        }
1.21      christos 1372:                        if ((m = so->so_rcv.sb_mb) != NULL)
1.1       cgd      1373:                                nextrecord = m->m_nextpkt;
                   1374:                }
                   1375:        }
1.3       andrew   1376:
                   1377:        if (m && pr->pr_flags & PR_ATOMIC) {
                   1378:                flags |= MSG_TRUNC;
                   1379:                if ((flags & MSG_PEEK) == 0)
                   1380:                        (void) sbdroprecord(&so->so_rcv);
                   1381:        }
1.1       cgd      1382:        if ((flags & MSG_PEEK) == 0) {
1.111.2.8! yamt     1383:                if (m == NULL) {
1.69      thorpej  1384:                        /*
1.70      thorpej  1385:                         * First part is an inline SB_EMPTY_FIXUP().  Second
1.69      thorpej  1386:                         * part makes sure sb_lastrecord is up-to-date if
                   1387:                         * there is still data in the socket buffer.
                   1388:                         */
1.1       cgd      1389:                        so->so_rcv.sb_mb = nextrecord;
1.69      thorpej  1390:                        if (so->so_rcv.sb_mb == NULL) {
                   1391:                                so->so_rcv.sb_mbtail = NULL;
                   1392:                                so->so_rcv.sb_lastrecord = NULL;
                   1393:                        } else if (nextrecord->m_nextpkt == NULL)
                   1394:                                so->so_rcv.sb_lastrecord = nextrecord;
                   1395:                }
                   1396:                SBLASTRECORDCHK(&so->so_rcv, "soreceive 4");
                   1397:                SBLASTMBUFCHK(&so->so_rcv, "soreceive 4");
1.1       cgd      1398:                if (pr->pr_flags & PR_WANTRCVD && so->so_pcb)
1.111.2.7  yamt     1399:                        (*pr->pr_usrreq)(so, PRU_RCVD, NULL,
                   1400:                            (struct mbuf *)(long)flags, NULL, l);
1.1       cgd      1401:        }
1.3       andrew   1402:        if (orig_resid == uio->uio_resid && orig_resid &&
                   1403:            (flags & MSG_EOR) == 0 && (so->so_state & SS_CANTRCVMORE) == 0) {
                   1404:                sbunlock(&so->so_rcv);
                   1405:                splx(s);
                   1406:                goto restart;
                   1407:        }
1.108     perry    1408:
1.111.2.8! yamt     1409:        if (flagsp != NULL)
1.1       cgd      1410:                *flagsp |= flags;
1.54      lukem    1411:  release:
1.1       cgd      1412:        sbunlock(&so->so_rcv);
                   1413:        splx(s);
1.111.2.8! yamt     1414:        return error;
1.1       cgd      1415: }
                   1416:
1.14      mycroft  1417: int
1.54      lukem    1418: soshutdown(struct socket *so, int how)
1.1       cgd      1419: {
1.99      matt     1420:        const struct protosw    *pr;
1.34      kleink   1421:
1.54      lukem    1422:        pr = so->so_proto;
1.34      kleink   1423:        if (!(how == SHUT_RD || how == SHUT_WR || how == SHUT_RDWR))
                   1424:                return (EINVAL);
1.1       cgd      1425:
1.34      kleink   1426:        if (how == SHUT_RD || how == SHUT_RDWR)
1.1       cgd      1427:                sorflush(so);
1.34      kleink   1428:        if (how == SHUT_WR || how == SHUT_RDWR)
1.111.2.7  yamt     1429:                return (*pr->pr_usrreq)(so, PRU_SHUTDOWN, NULL,
                   1430:                    NULL, NULL, NULL);
1.111.2.8! yamt     1431:        return 0;
1.1       cgd      1432: }
                   1433:
1.14      mycroft  1434: void
1.54      lukem    1435: sorflush(struct socket *so)
1.1       cgd      1436: {
1.54      lukem    1437:        struct sockbuf  *sb, asb;
1.99      matt     1438:        const struct protosw    *pr;
1.54      lukem    1439:        int             s;
1.1       cgd      1440:
1.54      lukem    1441:        sb = &so->so_rcv;
                   1442:        pr = so->so_proto;
1.1       cgd      1443:        sb->sb_flags |= SB_NOINTR;
1.15      mycroft  1444:        (void) sblock(sb, M_WAITOK);
1.56      thorpej  1445:        s = splnet();
1.1       cgd      1446:        socantrcvmore(so);
                   1447:        sbunlock(sb);
                   1448:        asb = *sb;
1.86      wrstuden 1449:        /*
                   1450:         * Clear most of the sockbuf structure, but leave some of the
                   1451:         * fields valid.
                   1452:         */
                   1453:        memset(&sb->sb_startzero, 0,
                   1454:            sizeof(*sb) - offsetof(struct sockbuf, sb_startzero));
1.1       cgd      1455:        splx(s);
                   1456:        if (pr->pr_flags & PR_RIGHTS && pr->pr_domain->dom_dispose)
                   1457:                (*pr->pr_domain->dom_dispose)(asb.sb_mb);
1.98      christos 1458:        sbrelease(&asb, so);
1.1       cgd      1459: }
                   1460:
1.111.2.8! yamt     1461: static int
        !          1462: sosetopt1(struct socket *so, int level, int optname, struct mbuf *m)
1.1       cgd      1463: {
1.111.2.8! yamt     1464:        int optval, val;
1.111.2.5  yamt     1465:        struct linger   *l;
1.111.2.7  yamt     1466:        struct sockbuf  *sb;
1.111.2.8! yamt     1467:        struct timeval *tv;
1.1       cgd      1468:
1.111.2.8! yamt     1469:        switch (optname) {
1.1       cgd      1470:
1.111.2.8! yamt     1471:        case SO_LINGER:
        !          1472:                if (m == NULL || m->m_len != sizeof(struct linger))
        !          1473:                        return EINVAL;
        !          1474:                l = mtod(m, struct linger *);
        !          1475:                if (l->l_linger < 0 || l->l_linger > USHRT_MAX ||
        !          1476:                    l->l_linger > (INT_MAX / hz))
        !          1477:                        return EDOM;
        !          1478:                so->so_linger = l->l_linger;
        !          1479:                if (l->l_onoff)
        !          1480:                        so->so_options |= SO_LINGER;
        !          1481:                else
        !          1482:                        so->so_options &= ~SO_LINGER;
        !          1483:                break;
1.1       cgd      1484:
1.111.2.8! yamt     1485:        case SO_DEBUG:
        !          1486:        case SO_KEEPALIVE:
        !          1487:        case SO_DONTROUTE:
        !          1488:        case SO_USELOOPBACK:
        !          1489:        case SO_BROADCAST:
        !          1490:        case SO_REUSEADDR:
        !          1491:        case SO_REUSEPORT:
        !          1492:        case SO_OOBINLINE:
        !          1493:        case SO_TIMESTAMP:
        !          1494:                if (m == NULL || m->m_len < sizeof(int))
        !          1495:                        return EINVAL;
        !          1496:                if (*mtod(m, int *))
        !          1497:                        so->so_options |= optname;
        !          1498:                else
        !          1499:                        so->so_options &= ~optname;
        !          1500:                break;
        !          1501:
        !          1502:        case SO_SNDBUF:
        !          1503:        case SO_RCVBUF:
        !          1504:        case SO_SNDLOWAT:
        !          1505:        case SO_RCVLOWAT:
        !          1506:                if (m == NULL || m->m_len < sizeof(int))
        !          1507:                        return EINVAL;
        !          1508:
        !          1509:                /*
        !          1510:                 * Values < 1 make no sense for any of these
        !          1511:                 * options, so disallow them.
        !          1512:                 */
        !          1513:                optval = *mtod(m, int *);
        !          1514:                if (optval < 1)
        !          1515:                        return EINVAL;
        !          1516:
        !          1517:                switch (optname) {
1.1       cgd      1518:
                   1519:                case SO_SNDBUF:
                   1520:                case SO_RCVBUF:
1.111.2.8! yamt     1521:                        sb = (optname == SO_SNDBUF) ?
        !          1522:                            &so->so_snd : &so->so_rcv;
        !          1523:                        if (sbreserve(sb, (u_long)optval, so) == 0)
        !          1524:                                return ENOBUFS;
        !          1525:                        sb->sb_flags &= ~SB_AUTOSIZE;
        !          1526:                        break;
        !          1527:
        !          1528:                /*
        !          1529:                 * Make sure the low-water is never greater than
        !          1530:                 * the high-water.
        !          1531:                 */
1.1       cgd      1532:                case SO_SNDLOWAT:
1.111.2.8! yamt     1533:                        so->so_snd.sb_lowat =
        !          1534:                            (optval > so->so_snd.sb_hiwat) ?
        !          1535:                            so->so_snd.sb_hiwat : optval;
        !          1536:                        break;
1.1       cgd      1537:                case SO_RCVLOWAT:
1.111.2.8! yamt     1538:                        so->so_rcv.sb_lowat =
        !          1539:                            (optval > so->so_rcv.sb_hiwat) ?
        !          1540:                            so->so_rcv.sb_hiwat : optval;
        !          1541:                        break;
        !          1542:                }
        !          1543:                break;
1.28      thorpej  1544:
1.111.2.8! yamt     1545:        case SO_SNDTIMEO:
        !          1546:        case SO_RCVTIMEO:
        !          1547:                if (m == NULL || m->m_len < sizeof(*tv))
        !          1548:                        return EINVAL;
        !          1549:                tv = mtod(m, struct timeval *);
        !          1550:                if (tv->tv_sec > (INT_MAX - tv->tv_usec / tick) / hz)
        !          1551:                        return EDOM;
        !          1552:                val = tv->tv_sec * hz + tv->tv_usec / tick;
        !          1553:                if (val == 0 && tv->tv_usec != 0)
        !          1554:                        val = 1;
1.1       cgd      1555:
1.111.2.8! yamt     1556:                switch (optname) {
1.1       cgd      1557:
                   1558:                case SO_SNDTIMEO:
1.111.2.8! yamt     1559:                        so->so_snd.sb_timeo = val;
        !          1560:                        break;
1.1       cgd      1561:                case SO_RCVTIMEO:
1.111.2.8! yamt     1562:                        so->so_rcv.sb_timeo = val;
        !          1563:                        break;
        !          1564:                }
        !          1565:                break;
1.1       cgd      1566:
1.111.2.8! yamt     1567:        default:
        !          1568:                return ENOPROTOOPT;
        !          1569:        }
        !          1570:        return 0;
        !          1571: }
1.1       cgd      1572:
1.111.2.8! yamt     1573: int
        !          1574: sosetopt(struct socket *so, int level, int optname, struct mbuf *m)
        !          1575: {
        !          1576:        int error, prerr;
1.1       cgd      1577:
1.111.2.8! yamt     1578:        if (level == SOL_SOCKET)
        !          1579:                error = sosetopt1(so, level, optname, m);
        !          1580:        else
        !          1581:                error = ENOPROTOOPT;
1.1       cgd      1582:
1.111.2.8! yamt     1583:        if ((error == 0 || error == ENOPROTOOPT) &&
        !          1584:            so->so_proto != NULL && so->so_proto->pr_ctloutput != NULL) {
        !          1585:                /* give the protocol stack a shot */
        !          1586:                prerr = (*so->so_proto->pr_ctloutput)(PRCO_SETOPT, so, level,
        !          1587:                    optname, &m);
        !          1588:                if (prerr == 0)
        !          1589:                        error = 0;
        !          1590:                else if (prerr != ENOPROTOOPT)
        !          1591:                        error = prerr;
        !          1592:        } else if (m != NULL)
        !          1593:                (void)m_free(m);
        !          1594:        return error;
1.1       cgd      1595: }
                   1596:
1.14      mycroft  1597: int
1.54      lukem    1598: sogetopt(struct socket *so, int level, int optname, struct mbuf **mp)
1.1       cgd      1599: {
1.54      lukem    1600:        struct mbuf     *m;
1.1       cgd      1601:
                   1602:        if (level != SOL_SOCKET) {
                   1603:                if (so->so_proto && so->so_proto->pr_ctloutput) {
                   1604:                        return ((*so->so_proto->pr_ctloutput)
                   1605:                                  (PRCO_GETOPT, so, level, optname, mp));
                   1606:                } else
                   1607:                        return (ENOPROTOOPT);
                   1608:        } else {
                   1609:                m = m_get(M_WAIT, MT_SOOPTS);
1.36      perry    1610:                m->m_len = sizeof(int);
1.1       cgd      1611:
                   1612:                switch (optname) {
                   1613:
                   1614:                case SO_LINGER:
1.36      perry    1615:                        m->m_len = sizeof(struct linger);
1.1       cgd      1616:                        mtod(m, struct linger *)->l_onoff =
1.111.2.5  yamt     1617:                            (so->so_options & SO_LINGER) ? 1 : 0;
1.1       cgd      1618:                        mtod(m, struct linger *)->l_linger = so->so_linger;
                   1619:                        break;
                   1620:
                   1621:                case SO_USELOOPBACK:
                   1622:                case SO_DONTROUTE:
                   1623:                case SO_DEBUG:
                   1624:                case SO_KEEPALIVE:
                   1625:                case SO_REUSEADDR:
1.15      mycroft  1626:                case SO_REUSEPORT:
1.1       cgd      1627:                case SO_BROADCAST:
                   1628:                case SO_OOBINLINE:
1.26      thorpej  1629:                case SO_TIMESTAMP:
1.111.2.5  yamt     1630:                        *mtod(m, int *) = (so->so_options & optname) ? 1 : 0;
1.1       cgd      1631:                        break;
                   1632:
                   1633:                case SO_TYPE:
                   1634:                        *mtod(m, int *) = so->so_type;
                   1635:                        break;
                   1636:
                   1637:                case SO_ERROR:
                   1638:                        *mtod(m, int *) = so->so_error;
                   1639:                        so->so_error = 0;
                   1640:                        break;
                   1641:
                   1642:                case SO_SNDBUF:
                   1643:                        *mtod(m, int *) = so->so_snd.sb_hiwat;
                   1644:                        break;
                   1645:
                   1646:                case SO_RCVBUF:
                   1647:                        *mtod(m, int *) = so->so_rcv.sb_hiwat;
                   1648:                        break;
                   1649:
                   1650:                case SO_SNDLOWAT:
                   1651:                        *mtod(m, int *) = so->so_snd.sb_lowat;
                   1652:                        break;
                   1653:
                   1654:                case SO_RCVLOWAT:
                   1655:                        *mtod(m, int *) = so->so_rcv.sb_lowat;
                   1656:                        break;
                   1657:
                   1658:                case SO_SNDTIMEO:
                   1659:                case SO_RCVTIMEO:
                   1660:                    {
                   1661:                        int val = (optname == SO_SNDTIMEO ?
                   1662:                             so->so_snd.sb_timeo : so->so_rcv.sb_timeo);
                   1663:
                   1664:                        m->m_len = sizeof(struct timeval);
                   1665:                        mtod(m, struct timeval *)->tv_sec = val / hz;
                   1666:                        mtod(m, struct timeval *)->tv_usec =
1.27      kleink   1667:                            (val % hz) * tick;
1.1       cgd      1668:                        break;
                   1669:                    }
                   1670:
1.107     darrenr  1671:                case SO_OVERFLOWED:
                   1672:                        *mtod(m, int *) = so->so_rcv.sb_overflowed;
                   1673:                        break;
                   1674:
1.1       cgd      1675:                default:
                   1676:                        (void)m_free(m);
                   1677:                        return (ENOPROTOOPT);
                   1678:                }
                   1679:                *mp = m;
                   1680:                return (0);
                   1681:        }
                   1682: }
                   1683:
1.14      mycroft  1684: void
1.54      lukem    1685: sohasoutofband(struct socket *so)
1.1       cgd      1686: {
1.90      christos 1687:        fownsignal(so->so_pgid, SIGURG, POLL_PRI, POLLPRI|POLLRDBAND, so);
1.2       cgd      1688:        selwakeup(&so->so_rcv.sb_sel);
1.1       cgd      1689: }
1.72      jdolecek 1690:
                   1691: static void
                   1692: filt_sordetach(struct knote *kn)
                   1693: {
                   1694:        struct socket   *so;
                   1695:
                   1696:        so = (struct socket *)kn->kn_fp->f_data;
1.73      christos 1697:        SLIST_REMOVE(&so->so_rcv.sb_sel.sel_klist, kn, knote, kn_selnext);
                   1698:        if (SLIST_EMPTY(&so->so_rcv.sb_sel.sel_klist))
1.72      jdolecek 1699:                so->so_rcv.sb_flags &= ~SB_KNOTE;
                   1700: }
                   1701:
                   1702: /*ARGSUSED*/
                   1703: static int
                   1704: filt_soread(struct knote *kn, long hint)
                   1705: {
                   1706:        struct socket   *so;
                   1707:
                   1708:        so = (struct socket *)kn->kn_fp->f_data;
                   1709:        kn->kn_data = so->so_rcv.sb_cc;
                   1710:        if (so->so_state & SS_CANTRCVMORE) {
1.108     perry    1711:                kn->kn_flags |= EV_EOF;
1.72      jdolecek 1712:                kn->kn_fflags = so->so_error;
                   1713:                return (1);
                   1714:        }
                   1715:        if (so->so_error)       /* temporary udp error */
                   1716:                return (1);
                   1717:        if (kn->kn_sfflags & NOTE_LOWAT)
                   1718:                return (kn->kn_data >= kn->kn_sdata);
                   1719:        return (kn->kn_data >= so->so_rcv.sb_lowat);
                   1720: }
                   1721:
                   1722: static void
                   1723: filt_sowdetach(struct knote *kn)
                   1724: {
                   1725:        struct socket   *so;
                   1726:
                   1727:        so = (struct socket *)kn->kn_fp->f_data;
1.73      christos 1728:        SLIST_REMOVE(&so->so_snd.sb_sel.sel_klist, kn, knote, kn_selnext);
                   1729:        if (SLIST_EMPTY(&so->so_snd.sb_sel.sel_klist))
1.72      jdolecek 1730:                so->so_snd.sb_flags &= ~SB_KNOTE;
                   1731: }
                   1732:
                   1733: /*ARGSUSED*/
                   1734: static int
                   1735: filt_sowrite(struct knote *kn, long hint)
                   1736: {
                   1737:        struct socket   *so;
                   1738:
                   1739:        so = (struct socket *)kn->kn_fp->f_data;
                   1740:        kn->kn_data = sbspace(&so->so_snd);
                   1741:        if (so->so_state & SS_CANTSENDMORE) {
1.108     perry    1742:                kn->kn_flags |= EV_EOF;
1.72      jdolecek 1743:                kn->kn_fflags = so->so_error;
                   1744:                return (1);
                   1745:        }
                   1746:        if (so->so_error)       /* temporary udp error */
                   1747:                return (1);
                   1748:        if (((so->so_state & SS_ISCONNECTED) == 0) &&
                   1749:            (so->so_proto->pr_flags & PR_CONNREQUIRED))
                   1750:                return (0);
                   1751:        if (kn->kn_sfflags & NOTE_LOWAT)
                   1752:                return (kn->kn_data >= kn->kn_sdata);
                   1753:        return (kn->kn_data >= so->so_snd.sb_lowat);
                   1754: }
                   1755:
                   1756: /*ARGSUSED*/
                   1757: static int
                   1758: filt_solisten(struct knote *kn, long hint)
                   1759: {
                   1760:        struct socket   *so;
                   1761:
                   1762:        so = (struct socket *)kn->kn_fp->f_data;
                   1763:
                   1764:        /*
                   1765:         * Set kn_data to number of incoming connections, not
                   1766:         * counting partial (incomplete) connections.
1.108     perry    1767:         */
1.72      jdolecek 1768:        kn->kn_data = so->so_qlen;
                   1769:        return (kn->kn_data > 0);
                   1770: }
                   1771:
                   1772: static const struct filterops solisten_filtops =
                   1773:        { 1, NULL, filt_sordetach, filt_solisten };
                   1774: static const struct filterops soread_filtops =
                   1775:        { 1, NULL, filt_sordetach, filt_soread };
                   1776: static const struct filterops sowrite_filtops =
                   1777:        { 1, NULL, filt_sowdetach, filt_sowrite };
                   1778:
                   1779: int
                   1780: soo_kqfilter(struct file *fp, struct knote *kn)
                   1781: {
                   1782:        struct socket   *so;
                   1783:        struct sockbuf  *sb;
                   1784:
                   1785:        so = (struct socket *)kn->kn_fp->f_data;
                   1786:        switch (kn->kn_filter) {
                   1787:        case EVFILT_READ:
                   1788:                if (so->so_options & SO_ACCEPTCONN)
                   1789:                        kn->kn_fop = &solisten_filtops;
                   1790:                else
                   1791:                        kn->kn_fop = &soread_filtops;
                   1792:                sb = &so->so_rcv;
                   1793:                break;
                   1794:        case EVFILT_WRITE:
                   1795:                kn->kn_fop = &sowrite_filtops;
                   1796:                sb = &so->so_snd;
                   1797:                break;
                   1798:        default:
                   1799:                return (1);
                   1800:        }
1.73      christos 1801:        SLIST_INSERT_HEAD(&sb->sb_sel.sel_klist, kn, kn_selnext);
1.72      jdolecek 1802:        sb->sb_flags |= SB_KNOTE;
                   1803:        return (0);
                   1804: }
                   1805:
1.94      yamt     1806: #include <sys/sysctl.h>
                   1807:
                   1808: static int sysctl_kern_somaxkva(SYSCTLFN_PROTO);
                   1809:
                   1810: /*
                   1811:  * sysctl helper routine for kern.somaxkva.  ensures that the given
                   1812:  * value is not too small.
                   1813:  * (XXX should we maybe make sure it's not too large as well?)
                   1814:  */
                   1815: static int
                   1816: sysctl_kern_somaxkva(SYSCTLFN_ARGS)
                   1817: {
                   1818:        int error, new_somaxkva;
                   1819:        struct sysctlnode node;
                   1820:
                   1821:        new_somaxkva = somaxkva;
                   1822:        node = *rnode;
                   1823:        node.sysctl_data = &new_somaxkva;
                   1824:        error = sysctl_lookup(SYSCTLFN_CALL(&node));
                   1825:        if (error || newp == NULL)
                   1826:                return (error);
                   1827:
                   1828:        if (new_somaxkva < (16 * 1024 * 1024)) /* sanity */
                   1829:                return (EINVAL);
                   1830:
1.111.2.7  yamt     1831:        mutex_enter(&so_pendfree_lock);
1.94      yamt     1832:        somaxkva = new_somaxkva;
1.111.2.7  yamt     1833:        cv_broadcast(&socurkva_cv);
                   1834:        mutex_exit(&so_pendfree_lock);
1.94      yamt     1835:
                   1836:        return (error);
                   1837: }
                   1838:
                   1839: SYSCTL_SETUP(sysctl_kern_somaxkva_setup, "sysctl kern.somaxkva setup")
                   1840: {
                   1841:
1.97      atatat   1842:        sysctl_createv(clog, 0, NULL, NULL,
                   1843:                       CTLFLAG_PERMANENT,
                   1844:                       CTLTYPE_NODE, "kern", NULL,
                   1845:                       NULL, 0, NULL, 0,
                   1846:                       CTL_KERN, CTL_EOL);
                   1847:
                   1848:        sysctl_createv(clog, 0, NULL, NULL,
                   1849:                       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.103     atatat   1850:                       CTLTYPE_INT, "somaxkva",
                   1851:                       SYSCTL_DESCR("Maximum amount of kernel memory to be "
                   1852:                                    "used for socket buffers"),
1.94      yamt     1853:                       sysctl_kern_somaxkva, 0, NULL, 0,
                   1854:                       CTL_KERN, KERN_SOMAXKVA, CTL_EOL);
                   1855: }

CVSweb <webmaster@jp.NetBSD.org>