Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files. =================================================================== RCS file: /ftp/cvs/cvsroot/src/sys/kern/uipc_socket.c,v rcsdiff: /ftp/cvs/cvsroot/src/sys/kern/uipc_socket.c,v: warning: Unknown phrases like `commitid ...;' are present. retrieving revision 1.135 retrieving revision 1.144.4.1 diff -u -p -r1.135 -r1.144.4.1 --- src/sys/kern/uipc_socket.c 2007/03/12 18:18:34 1.135 +++ src/sys/kern/uipc_socket.c 2007/11/19 00:48:58 1.144.4.1 @@ -1,7 +1,7 @@ -/* $NetBSD: uipc_socket.c,v 1.135 2007/03/12 18:18:34 ad Exp $ */ +/* $NetBSD: uipc_socket.c,v 1.144.4.1 2007/11/19 00:48:58 mjf Exp $ */ /*- - * Copyright (c) 2002 The NetBSD Foundation, Inc. + * Copyright (c) 2002, 2007 The NetBSD Foundation, Inc. * All rights reserved. * * This code is derived from software contributed to The NetBSD Foundation @@ -68,7 +68,7 @@ */ #include -__KERNEL_RCSID(0, "$NetBSD: uipc_socket.c,v 1.135 2007/03/12 18:18:34 ad Exp $"); +__KERNEL_RCSID(0, "$NetBSD: uipc_socket.c,v 1.144.4.1 2007/11/19 00:48:58 mjf Exp $"); #include "opt_sock_counters.h" #include "opt_sosend_loan.h" @@ -79,6 +79,7 @@ __KERNEL_RCSID(0, "$NetBSD: uipc_socket. #include #include #include +#include #include #include #include @@ -92,6 +93,8 @@ __KERNEL_RCSID(0, "$NetBSD: uipc_socket. #include #include #include +#include +#include #include @@ -101,6 +104,8 @@ POOL_INIT(socket_pool, sizeof(struct soc MALLOC_DEFINE(M_SOOPTS, "soopts", "socket options"); MALLOC_DEFINE(M_SONAME, "soname", "socket name"); +extern const struct fileops socketops; + extern int somaxconn; /* patchable (XXX sysctl) */ int somaxconn = SOMAXCONN; @@ -136,7 +141,7 @@ int sock_loan_thresh = -1; int sock_loan_thresh = 4096; #endif -static struct simplelock so_pendfree_slock = SIMPLELOCK_INITIALIZER; +static kmutex_t so_pendfree_lock; static struct mbuf *so_pendfree; #ifndef SOMAXKVA @@ -144,7 +149,7 @@ static struct mbuf *so_pendfree; #endif int somaxkva = SOMAXKVA; static int socurkva; -static int sokvawaiters; +static kcondvar_t socurkva_cv; #define SOCK_LOAN_CHUNK 65536 @@ -154,11 +159,9 @@ static size_t sodopendfreel(void); static vsize_t sokvareserve(struct socket *so, vsize_t len) { - int s; int error; - s = splvm(); - simple_lock(&so_pendfree_slock); + mutex_enter(&so_pendfree_lock); while (socurkva + len > somaxkva) { size_t freed; @@ -176,33 +179,25 @@ sokvareserve(struct socket *so, vsize_t continue; SOSEND_COUNTER_INCR(&sosend_kvalimit); - sokvawaiters++; - error = ltsleep(&socurkva, PVM | PCATCH, "sokva", 0, - &so_pendfree_slock); - sokvawaiters--; + error = cv_wait_sig(&socurkva_cv, &so_pendfree_lock); if (error) { len = 0; break; } } socurkva += len; - simple_unlock(&so_pendfree_slock); - splx(s); + mutex_exit(&so_pendfree_lock); return len; } static void sokvaunreserve(vsize_t len) { - int s; - s = splvm(); - simple_lock(&so_pendfree_slock); + mutex_enter(&so_pendfree_lock); socurkva -= len; - if (sokvawaiters) - wakeup(&socurkva); - simple_unlock(&so_pendfree_slock); - splx(s); + cv_broadcast(&socurkva_cv); + mutex_exit(&so_pendfree_lock); } /* @@ -287,43 +282,34 @@ sodoloanfree(struct vm_page **pgs, void static size_t sodopendfree() { - int s; size_t rv; - s = splvm(); - simple_lock(&so_pendfree_slock); + mutex_enter(&so_pendfree_lock); rv = sodopendfreel(); - simple_unlock(&so_pendfree_slock); - splx(s); + mutex_exit(&so_pendfree_lock); return rv; } /* * sodopendfreel: free mbufs on "pendfree" list. - * unlock and relock so_pendfree_slock when freeing mbufs. + * unlock and relock so_pendfree_lock when freeing mbufs. * - * => called with so_pendfree_slock held. - * => called at splvm. + * => called with so_pendfree_lock held. */ static size_t sodopendfreel() { + struct mbuf *m, *next; size_t rv = 0; - LOCK_ASSERT(simple_lock_held(&so_pendfree_slock)); - - for (;;) { - struct mbuf *m; - struct mbuf *next; + KASSERT(mutex_owned(&so_pendfree_lock)); + while (so_pendfree != NULL) { m = so_pendfree; - if (m == NULL) - break; so_pendfree = NULL; - simple_unlock(&so_pendfree_slock); - /* XXX splx */ + mutex_exit(&so_pendfree_lock); for (; m != NULL; m = next) { next = m->m_next; @@ -332,11 +318,10 @@ sodopendfreel() sodoloanfree((m->m_flags & M_EXT_PAGES) ? m->m_ext.ext_pgs : NULL, m->m_ext.ext_buf, m->m_ext.ext_size); - pool_cache_put(&mbpool_cache, m); + pool_cache_put(mb_cache, m); } - /* XXX splvm */ - simple_lock(&so_pendfree_slock); + mutex_enter(&so_pendfree_lock); } return (rv); @@ -345,7 +330,6 @@ sodopendfreel() void soloanfree(struct mbuf *m, void *buf, size_t size, void *arg) { - int s; if (m == NULL) { @@ -364,14 +348,11 @@ soloanfree(struct mbuf *m, void *buf, si * because we need to put kva back to kernel_map. */ - s = splvm(); - simple_lock(&so_pendfree_slock); + mutex_enter(&so_pendfree_lock); m->m_next = so_pendfree; so_pendfree = m; - if (sokvawaiters) - wakeup(&socurkva); - simple_unlock(&so_pendfree_slock); - splx(s); + cv_broadcast(&socurkva_cv); + mutex_exit(&so_pendfree_lock); } static long @@ -446,10 +427,34 @@ sokva_reclaim_callback(struct callback_e return CALLBACK_CHAIN_CONTINUE; } +struct mbuf * +getsombuf(struct socket *so) +{ + struct mbuf *m; + + m = m_get(M_WAIT, MT_SONAME); + MCLAIM(m, so->so_mowner); + return m; +} + +struct mbuf * +m_intopt(struct socket *so, int val) +{ + struct mbuf *m; + + m = getsombuf(so); + m->m_len = sizeof(int); + *mtod(m, int *) = val; + return m; +} + void soinit(void) { + mutex_init(&so_pendfree_lock, MUTEX_DRIVER, IPL_VM); + cv_init(&socurkva_cv, "sokva"); + /* Set the initial adjusted socket buffer size. */ if (sb_max_set(sb_max)) panic("bad initial sb_max value: %lu", sb_max); @@ -477,29 +482,29 @@ socreate(int dom, struct socket **aso, i error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET, KAUTH_REQ_NETWORK_SOCKET_OPEN, KAUTH_ARG(dom), KAUTH_ARG(type), KAUTH_ARG(proto)); - if (error) - return (error); + if (error != 0) + return error; if (proto) prp = pffindproto(dom, proto, type); else prp = pffindtype(dom, type); - if (prp == 0) { + if (prp == NULL) { /* no support for domain */ if (pffinddomain(dom) == 0) - return (EAFNOSUPPORT); + return EAFNOSUPPORT; /* no support for socket type */ if (proto == 0 && type != 0) - return (EPROTOTYPE); - return (EPROTONOSUPPORT); + return EPROTOTYPE; + return EPROTONOSUPPORT; } - if (prp->pr_usrreq == 0) - return (EPROTONOSUPPORT); + if (prp->pr_usrreq == NULL) + return EPROTONOSUPPORT; if (prp->pr_type != type) - return (EPROTOTYPE); + return EPROTOTYPE; s = splsoftnet(); so = pool_get(&socket_pool, PR_WAITOK); - memset((void *)so, 0, sizeof(*so)); + memset(so, 0, sizeof(*so)); TAILQ_INIT(&so->so_q0); TAILQ_INIT(&so->so_q); so->so_type = type; @@ -511,23 +516,56 @@ socreate(int dom, struct socket **aso, i so->so_snd.sb_mowner = &prp->pr_domain->dom_mowner; so->so_mowner = &prp->pr_domain->dom_mowner; #endif - if (l != NULL) { - uid = kauth_cred_geteuid(l->l_cred); - } else { - uid = 0; - } + selinit(&so->so_rcv.sb_sel); + selinit(&so->so_snd.sb_sel); + uid = kauth_cred_geteuid(l->l_cred); so->so_uidinfo = uid_find(uid); - error = (*prp->pr_usrreq)(so, PRU_ATTACH, (struct mbuf *)0, - (struct mbuf *)(long)proto, (struct mbuf *)0, l); - if (error) { + error = (*prp->pr_usrreq)(so, PRU_ATTACH, NULL, + (struct mbuf *)(long)proto, NULL, l); + if (error != 0) { so->so_state |= SS_NOFDREF; sofree(so); splx(s); - return (error); + return error; } splx(s); *aso = so; - return (0); + return 0; +} + +/* On success, write file descriptor to fdout and return zero. On + * failure, return non-zero; *fdout will be undefined. + */ +int +fsocreate(int domain, struct socket **sop, int type, int protocol, + struct lwp *l, int *fdout) +{ + struct filedesc *fdp; + struct socket *so; + struct file *fp; + int fd, error; + + fdp = l->l_proc->p_fd; + /* falloc() will use the desciptor for us */ + if ((error = falloc(l, &fp, &fd)) != 0) + return (error); + fp->f_flag = FREAD|FWRITE; + fp->f_type = DTYPE_SOCKET; + fp->f_ops = &socketops; + error = socreate(domain, &so, type, protocol, l); + if (error != 0) { + FILE_UNUSE(fp, l); + fdremove(fdp, fd); + ffree(fp); + } else { + if (sop != NULL) + *sop = so; + fp->f_data = so; + FILE_SET_MATURE(fp); + FILE_UNUSE(fp, l); + *fdout = fd; + } + return error; } int @@ -536,10 +574,9 @@ sobind(struct socket *so, struct mbuf *n int s, error; s = splsoftnet(); - error = (*so->so_proto->pr_usrreq)(so, PRU_BIND, (struct mbuf *)0, - nam, (struct mbuf *)0, l); + error = (*so->so_proto->pr_usrreq)(so, PRU_BIND, NULL, nam, NULL, l); splx(s); - return (error); + return error; } int @@ -548,11 +585,11 @@ solisten(struct socket *so, int backlog) int s, error; s = splsoftnet(); - error = (*so->so_proto->pr_usrreq)(so, PRU_LISTEN, (struct mbuf *)0, - (struct mbuf *)0, (struct mbuf *)0, (struct lwp *)0); - if (error) { + error = (*so->so_proto->pr_usrreq)(so, PRU_LISTEN, NULL, + NULL, NULL, NULL); + if (error != 0) { splx(s); - return (error); + return error; } if (TAILQ_EMPTY(&so->so_q)) so->so_options |= SO_ACCEPTCONN; @@ -560,7 +597,7 @@ solisten(struct socket *so, int backlog) backlog = 0; so->so_qlimit = min(backlog, somaxconn); splx(s); - return (0); + return 0; } void @@ -586,6 +623,8 @@ sofree(struct socket *so) RLIM_INFINITY); sbrelease(&so->so_snd, so); sorflush(so); + seldestroy(&so->so_rcv.sb_sel); + seldestroy(&so->so_snd.sb_sel); pool_put(&socket_pool, so); } @@ -636,8 +675,7 @@ soclose(struct socket *so) drop: if (so->so_pcb) { int error2 = (*so->so_proto->pr_usrreq)(so, PRU_DETACH, - (struct mbuf *)0, (struct mbuf *)0, (struct mbuf *)0, - (struct lwp *)0); + NULL, NULL, NULL, NULL); if (error == 0) error = error2; } @@ -656,9 +694,15 @@ soclose(struct socket *so) int soabort(struct socket *so) { + int error; - return (*so->so_proto->pr_usrreq)(so, PRU_ABORT, (struct mbuf *)0, - (struct mbuf *)0, (struct mbuf *)0, (struct lwp *)0); + KASSERT(so->so_head == NULL); + error = (*so->so_proto->pr_usrreq)(so, PRU_ABORT, NULL, + NULL, NULL, NULL); + if (error) { + sofree(so); + } + return error; } int @@ -674,7 +718,7 @@ soaccept(struct socket *so, struct mbuf if ((so->so_state & SS_ISDISCONNECTED) == 0 || (so->so_proto->pr_flags & PR_ABRTACPTDIS) == 0) error = (*so->so_proto->pr_usrreq)(so, PRU_ACCEPT, - (struct mbuf *)0, nam, (struct mbuf *)0, (struct lwp *)0); + NULL, nam, NULL, NULL); else error = ECONNABORTED; @@ -702,7 +746,7 @@ soconnect(struct socket *so, struct mbuf error = EISCONN; else error = (*so->so_proto->pr_usrreq)(so, PRU_CONNECT, - (struct mbuf *)0, nam, (struct mbuf *)0, l); + NULL, nam, NULL, l); splx(s); return (error); } @@ -714,8 +758,7 @@ soconnect2(struct socket *so1, struct so s = splsoftnet(); error = (*so1->so_proto->pr_usrreq)(so1, PRU_CONNECT2, - (struct mbuf *)0, (struct mbuf *)so2, (struct mbuf *)0, - (struct lwp *)0); + NULL, (struct mbuf *)so2, NULL, NULL); splx(s); return (error); } @@ -735,8 +778,7 @@ sodisconnect(struct socket *so) goto bad; } error = (*so->so_proto->pr_usrreq)(so, PRU_DISCONNECT, - (struct mbuf *)0, (struct mbuf *)0, (struct mbuf *)0, - (struct lwp *)0); + NULL, NULL, NULL, NULL); bad: splx(s); sodopendfree(); @@ -849,11 +891,11 @@ sosend(struct socket *so, struct mbuf *a if (flags & MSG_EOR) top->m_flags |= M_EOR; } else do { - if (top == 0) { + if (top == NULL) { m = m_gethdr(M_WAIT, MT_DATA); mlen = MHLEN; m->m_pkthdr.len = 0; - m->m_pkthdr.rcvif = (struct ifnet *)0; + m->m_pkthdr.rcvif = NULL; } else { m = m_get(M_WAIT, MT_DATA); mlen = MLEN; @@ -893,14 +935,13 @@ sosend(struct socket *so, struct mbuf *a if (atomic && top == 0 && len < mlen) MH_ALIGN(m, len); } - error = uiomove(mtod(m, void *), (int)len, - uio); + error = uiomove(mtod(m, void *), (int)len, uio); have_data: resid = uio->uio_resid; m->m_len = len; *mp = m; top->m_pkthdr.len += len; - if (error) + if (error != 0) goto release; mp = &m->m_next; if (resid <= 0) { @@ -929,10 +970,10 @@ sosend(struct socket *so, struct mbuf *a splx(s); clen = 0; - control = 0; - top = 0; + control = NULL; + top = NULL; mp = ⊤ - if (error) + if (error != 0) goto release; } while (resid && space > 0); } while (resid); @@ -979,11 +1020,11 @@ soreceive(struct socket *so, struct mbuf type = 0; orig_resid = uio->uio_resid; - if (paddr) - *paddr = 0; - if (controlp) - *controlp = 0; - if (flagsp) + if (paddr != NULL) + *paddr = NULL; + if (controlp != NULL) + *controlp = NULL; + if (flagsp != NULL) flags = *flagsp &~ MSG_EOR; else flags = 0; @@ -994,29 +1035,27 @@ soreceive(struct socket *so, struct mbuf if (flags & MSG_OOB) { m = m_get(M_WAIT, MT_DATA); error = (*pr->pr_usrreq)(so, PRU_RCVOOB, m, - (struct mbuf *)(long)(flags & MSG_PEEK), - (struct mbuf *)0, l); + (struct mbuf *)(long)(flags & MSG_PEEK), NULL, l); if (error) goto bad; do { error = uiomove(mtod(m, void *), (int) min(uio->uio_resid, m->m_len), uio); m = m_free(m); - } while (uio->uio_resid && error == 0 && m); + } while (uio->uio_resid > 0 && error == 0 && m); bad: - if (m) + if (m != NULL) m_freem(m); - return (error); + return error; } - if (mp) - *mp = (struct mbuf *)0; + if (mp != NULL) + *mp = NULL; if (so->so_state & SS_ISCONFIRMING && uio->uio_resid) - (*pr->pr_usrreq)(so, PRU_RCVD, (struct mbuf *)0, - (struct mbuf *)0, (struct mbuf *)0, l); + (*pr->pr_usrreq)(so, PRU_RCVD, NULL, NULL, NULL, l); restart: if ((error = sblock(&so->so_rcv, SBLOCKWAIT(flags))) != 0) - return (error); + return error; s = splsoftnet(); m = so->so_rcv.sb_mb; @@ -1031,17 +1070,20 @@ soreceive(struct socket *so, struct mbuf * we have to do the receive in sections, and thus risk returning * a short count if a timeout or signal occurs after we start. */ - if (m == 0 || (((flags & MSG_DONTWAIT) == 0 && - so->so_rcv.sb_cc < uio->uio_resid) && - (so->so_rcv.sb_cc < so->so_rcv.sb_lowat || - ((flags & MSG_WAITALL) && uio->uio_resid <= so->so_rcv.sb_hiwat)) && - m->m_nextpkt == 0 && (pr->pr_flags & PR_ATOMIC) == 0)) { + if (m == NULL || + ((flags & MSG_DONTWAIT) == 0 && + so->so_rcv.sb_cc < uio->uio_resid && + (so->so_rcv.sb_cc < so->so_rcv.sb_lowat || + ((flags & MSG_WAITALL) && + uio->uio_resid <= so->so_rcv.sb_hiwat)) && + m->m_nextpkt == NULL && + (pr->pr_flags & PR_ATOMIC) == 0)) { #ifdef DIAGNOSTIC - if (m == 0 && so->so_rcv.sb_cc) + if (m == NULL && so->so_rcv.sb_cc) panic("receive 1"); #endif if (so->so_error) { - if (m) + if (m != NULL) goto dontblock; error = so->so_error; if ((flags & MSG_PEEK) == 0) @@ -1049,12 +1091,12 @@ soreceive(struct socket *so, struct mbuf goto release; } if (so->so_state & SS_CANTRCVMORE) { - if (m) + if (m != NULL) goto dontblock; else goto release; } - for (; m; m = m->m_next) + for (; m != NULL; m = m->m_next) if (m->m_type == MT_OOBDATA || (m->m_flags & M_EOR)) { m = so->so_rcv.sb_mb; goto dontblock; @@ -1075,8 +1117,8 @@ soreceive(struct socket *so, struct mbuf sbunlock(&so->so_rcv); error = sbwait(&so->so_rcv); splx(s); - if (error) - return (error); + if (error != 0) + return error; goto restart; } dontblock: @@ -1085,7 +1127,7 @@ soreceive(struct socket *so, struct mbuf * While we process the initial mbufs containing address and control * info, we save a copy of m->m_nextpkt into nextrecord. */ - if (l) + if (l != NULL) l->l_proc->p_stats->p_ru.ru_msgrcv++; KASSERT(m == so->so_rcv.sb_mb); SBLASTRECORDCHK(&so->so_rcv, "soreceive 1"); @@ -1104,10 +1146,10 @@ soreceive(struct socket *so, struct mbuf } else { sbfree(&so->so_rcv, m); mbuf_removed = 1; - if (paddr) { + if (paddr != NULL) { *paddr = m; so->so_rcv.sb_mb = m->m_next; - m->m_next = 0; + m->m_next = NULL; m = so->so_rcv.sb_mb; } else { MFREE(m, so->so_rcv.sb_mb); @@ -1115,15 +1157,15 @@ soreceive(struct socket *so, struct mbuf } } } - while (m && m->m_type == MT_CONTROL && error == 0) { + while (m != NULL && m->m_type == MT_CONTROL && error == 0) { if (flags & MSG_PEEK) { - if (controlp) + if (controlp != NULL) *controlp = m_copy(m, 0, m->m_len); m = m->m_next; } else { sbfree(&so->so_rcv, m); mbuf_removed = 1; - if (controlp) { + if (controlp != NULL) { struct domain *dom = pr->pr_domain; if (dom->dom_externalize && l && mtod(m, struct cmsghdr *)->cmsg_type == @@ -1131,7 +1173,7 @@ soreceive(struct socket *so, struct mbuf error = (*dom->dom_externalize)(m, l); *controlp = m; so->so_rcv.sb_mb = m->m_next; - m->m_next = 0; + m->m_next = NULL; m = so->so_rcv.sb_mb; } else { /* @@ -1145,7 +1187,7 @@ soreceive(struct socket *so, struct mbuf m = so->so_rcv.sb_mb; } } - if (controlp) { + if (controlp != NULL) { orig_resid = 0; controlp = &(*controlp)->m_next; } @@ -1157,7 +1199,7 @@ soreceive(struct socket *so, struct mbuf * the last packet on the chain (nextrecord == NULL) and we * change m->m_nextpkt. */ - if (m) { + if (m != NULL) { if ((flags & MSG_PEEK) == 0) { m->m_nextpkt = nextrecord; /* @@ -1185,7 +1227,7 @@ soreceive(struct socket *so, struct mbuf moff = 0; offset = 0; - while (m && uio->uio_resid > 0 && error == 0) { + while (m != NULL && uio->uio_resid > 0 && error == 0) { if (m->m_type == MT_OOBDATA) { if (type != MT_OOBDATA) break; @@ -1209,13 +1251,13 @@ soreceive(struct socket *so, struct mbuf * we must note any additions to the sockbuf when we * block interrupts again. */ - if (mp == 0) { + if (mp == NULL) { SBLASTRECORDCHK(&so->so_rcv, "soreceive uiomove"); SBLASTMBUFCHK(&so->so_rcv, "soreceive uiomove"); splx(s); error = uiomove(mtod(m, char *) + moff, (int)len, uio); s = splsoftnet(); - if (error) { + if (error != 0) { /* * If any part of the record has been removed * (such as the MT_SONAME mbuf, which will @@ -1248,7 +1290,7 @@ soreceive(struct socket *so, struct mbuf *mp = m; mp = &m->m_next; so->so_rcv.sb_mb = m = m->m_next; - *mp = (struct mbuf *)0; + *mp = NULL; } else { MFREE(m, so->so_rcv.sb_mb); m = so->so_rcv.sb_mb; @@ -1269,16 +1311,14 @@ soreceive(struct socket *so, struct mbuf SBLASTRECORDCHK(&so->so_rcv, "soreceive 3"); SBLASTMBUFCHK(&so->so_rcv, "soreceive 3"); } - } else { - if (flags & MSG_PEEK) - moff += len; - else { - if (mp) - *mp = m_copym(m, 0, len, M_WAIT); - m->m_data += len; - m->m_len -= len; - so->so_rcv.sb_cc -= len; - } + } else if (flags & MSG_PEEK) + moff += len; + else { + if (mp != NULL) + *mp = m_copym(m, 0, len, M_WAIT); + m->m_data += len; + m->m_len -= len; + so->so_rcv.sb_cc -= len; } if (so->so_oobmark) { if ((flags & MSG_PEEK) == 0) { @@ -1302,7 +1342,7 @@ soreceive(struct socket *so, struct mbuf * with a short count but without error. * Keep sockbuf locked against other readers. */ - while (flags & MSG_WAITALL && m == 0 && uio->uio_resid > 0 && + while (flags & MSG_WAITALL && m == NULL && uio->uio_resid > 0 && !sosendallatonce(so) && !nextrecord) { if (so->so_error || so->so_state & SS_CANTRCVMORE) break; @@ -1319,16 +1359,14 @@ soreceive(struct socket *so, struct mbuf */ if ((pr->pr_flags & PR_WANTRCVD) && so->so_pcb) (*pr->pr_usrreq)(so, PRU_RCVD, - (struct mbuf *)0, - (struct mbuf *)(long)flags, - (struct mbuf *)0, l); + NULL, (struct mbuf *)(long)flags, NULL, l); SBLASTRECORDCHK(&so->so_rcv, "soreceive sbwait 2"); SBLASTMBUFCHK(&so->so_rcv, "soreceive sbwait 2"); error = sbwait(&so->so_rcv); - if (error) { + if (error != 0) { sbunlock(&so->so_rcv); splx(s); - return (0); + return 0; } if ((m = so->so_rcv.sb_mb) != NULL) nextrecord = m->m_nextpkt; @@ -1341,7 +1379,7 @@ soreceive(struct socket *so, struct mbuf (void) sbdroprecord(&so->so_rcv); } if ((flags & MSG_PEEK) == 0) { - if (m == 0) { + if (m == NULL) { /* * First part is an inline SB_EMPTY_FIXUP(). Second * part makes sure sb_lastrecord is up-to-date if @@ -1357,8 +1395,8 @@ soreceive(struct socket *so, struct mbuf SBLASTRECORDCHK(&so->so_rcv, "soreceive 4"); SBLASTMBUFCHK(&so->so_rcv, "soreceive 4"); if (pr->pr_flags & PR_WANTRCVD && so->so_pcb) - (*pr->pr_usrreq)(so, PRU_RCVD, (struct mbuf *)0, - (struct mbuf *)(long)flags, (struct mbuf *)0, l); + (*pr->pr_usrreq)(so, PRU_RCVD, NULL, + (struct mbuf *)(long)flags, NULL, l); } if (orig_resid == uio->uio_resid && orig_resid && (flags & MSG_EOR) == 0 && (so->so_state & SS_CANTRCVMORE) == 0) { @@ -1367,12 +1405,12 @@ soreceive(struct socket *so, struct mbuf goto restart; } - if (flagsp) + if (flagsp != NULL) *flagsp |= flags; release: sbunlock(&so->so_rcv); splx(s); - return (error); + return error; } int @@ -1387,9 +1425,9 @@ soshutdown(struct socket *so, int how) if (how == SHUT_RD || how == SHUT_RDWR) sorflush(so); if (how == SHUT_WR || how == SHUT_RDWR) - return (*pr->pr_usrreq)(so, PRU_SHUTDOWN, (struct mbuf *)0, - (struct mbuf *)0, (struct mbuf *)0, (struct lwp *)0); - return (0); + return (*pr->pr_usrreq)(so, PRU_SHUTDOWN, NULL, + NULL, NULL, NULL); + return 0; } void @@ -1419,157 +1457,140 @@ sorflush(struct socket *so) sbrelease(&asb, so); } -int -sosetopt(struct socket *so, int level, int optname, struct mbuf *m0) +static int +sosetopt1(struct socket *so, int level, int optname, struct mbuf *m) { - int error; - struct mbuf *m; + int optval, val; struct linger *l; + struct sockbuf *sb; + struct timeval *tv; - error = 0; - m = m0; - if (level != SOL_SOCKET) { - if (so->so_proto && so->so_proto->pr_ctloutput) - return ((*so->so_proto->pr_ctloutput) - (PRCO_SETOPT, so, level, optname, &m0)); - error = ENOPROTOOPT; - } else { - switch (optname) { + switch (optname) { - case SO_LINGER: - if (m == NULL || m->m_len != sizeof(struct linger)) { - error = EINVAL; - goto bad; - } - l = mtod(m, struct linger *); - if (l->l_linger < 0 || l->l_linger > USHRT_MAX || - l->l_linger > (INT_MAX / hz)) { - error = EDOM; - goto bad; - } - so->so_linger = l->l_linger; - if (l->l_onoff) - so->so_options |= SO_LINGER; - else - so->so_options &= ~SO_LINGER; - break; + case SO_LINGER: + if (m == NULL || m->m_len != sizeof(struct linger)) + return EINVAL; + l = mtod(m, struct linger *); + if (l->l_linger < 0 || l->l_linger > USHRT_MAX || + l->l_linger > (INT_MAX / hz)) + return EDOM; + so->so_linger = l->l_linger; + if (l->l_onoff) + so->so_options |= SO_LINGER; + else + so->so_options &= ~SO_LINGER; + break; - case SO_DEBUG: - case SO_KEEPALIVE: - case SO_DONTROUTE: - case SO_USELOOPBACK: - case SO_BROADCAST: - case SO_REUSEADDR: - case SO_REUSEPORT: - case SO_OOBINLINE: - case SO_TIMESTAMP: - if (m == NULL || m->m_len < sizeof(int)) { - error = EINVAL; - goto bad; - } - if (*mtod(m, int *)) - so->so_options |= optname; - else - so->so_options &= ~optname; - break; + case SO_DEBUG: + case SO_KEEPALIVE: + case SO_DONTROUTE: + case SO_USELOOPBACK: + case SO_BROADCAST: + case SO_REUSEADDR: + case SO_REUSEPORT: + case SO_OOBINLINE: + case SO_TIMESTAMP: + if (m == NULL || m->m_len < sizeof(int)) + return EINVAL; + if (*mtod(m, int *)) + so->so_options |= optname; + else + so->so_options &= ~optname; + break; + + case SO_SNDBUF: + case SO_RCVBUF: + case SO_SNDLOWAT: + case SO_RCVLOWAT: + if (m == NULL || m->m_len < sizeof(int)) + return EINVAL; + + /* + * Values < 1 make no sense for any of these + * options, so disallow them. + */ + optval = *mtod(m, int *); + if (optval < 1) + return EINVAL; + + switch (optname) { case SO_SNDBUF: case SO_RCVBUF: + sb = (optname == SO_SNDBUF) ? + &so->so_snd : &so->so_rcv; + if (sbreserve(sb, (u_long)optval, so) == 0) + return ENOBUFS; + sb->sb_flags &= ~SB_AUTOSIZE; + break; + + /* + * Make sure the low-water is never greater than + * the high-water. + */ case SO_SNDLOWAT: + so->so_snd.sb_lowat = + (optval > so->so_snd.sb_hiwat) ? + so->so_snd.sb_hiwat : optval; + break; case SO_RCVLOWAT: - { - int optval; - - if (m == NULL || m->m_len < sizeof(int)) { - error = EINVAL; - goto bad; - } - - /* - * Values < 1 make no sense for any of these - * options, so disallow them. - */ - optval = *mtod(m, int *); - if (optval < 1) { - error = EINVAL; - goto bad; - } - - switch (optname) { + so->so_rcv.sb_lowat = + (optval > so->so_rcv.sb_hiwat) ? + so->so_rcv.sb_hiwat : optval; + break; + } + break; - case SO_SNDBUF: - case SO_RCVBUF: - if (sbreserve(optname == SO_SNDBUF ? - &so->so_snd : &so->so_rcv, - (u_long) optval, so) == 0) { - error = ENOBUFS; - goto bad; - } - break; + case SO_SNDTIMEO: + case SO_RCVTIMEO: + if (m == NULL || m->m_len < sizeof(*tv)) + return EINVAL; + tv = mtod(m, struct timeval *); + if (tv->tv_sec > (INT_MAX - tv->tv_usec / tick) / hz) + return EDOM; + val = tv->tv_sec * hz + tv->tv_usec / tick; + if (val == 0 && tv->tv_usec != 0) + val = 1; - /* - * Make sure the low-water is never greater than - * the high-water. - */ - case SO_SNDLOWAT: - so->so_snd.sb_lowat = - (optval > so->so_snd.sb_hiwat) ? - so->so_snd.sb_hiwat : optval; - break; - case SO_RCVLOWAT: - so->so_rcv.sb_lowat = - (optval > so->so_rcv.sb_hiwat) ? - so->so_rcv.sb_hiwat : optval; - break; - } - break; - } + switch (optname) { case SO_SNDTIMEO: + so->so_snd.sb_timeo = val; + break; case SO_RCVTIMEO: - { - struct timeval *tv; - int val; + so->so_rcv.sb_timeo = val; + break; + } + break; - if (m == NULL || m->m_len < sizeof(*tv)) { - error = EINVAL; - goto bad; - } - tv = mtod(m, struct timeval *); - if (tv->tv_sec > (INT_MAX - tv->tv_usec / tick) / hz) { - error = EDOM; - goto bad; - } - val = tv->tv_sec * hz + tv->tv_usec / tick; - if (val == 0 && tv->tv_usec != 0) - val = 1; + default: + return ENOPROTOOPT; + } + return 0; +} - switch (optname) { +int +sosetopt(struct socket *so, int level, int optname, struct mbuf *m) +{ + int error, prerr; - case SO_SNDTIMEO: - so->so_snd.sb_timeo = val; - break; - case SO_RCVTIMEO: - so->so_rcv.sb_timeo = val; - break; - } - break; - } + if (level == SOL_SOCKET) + error = sosetopt1(so, level, optname, m); + else + error = ENOPROTOOPT; - default: - error = ENOPROTOOPT; - break; - } - if (error == 0 && so->so_proto && so->so_proto->pr_ctloutput) { - (void) ((*so->so_proto->pr_ctloutput) - (PRCO_SETOPT, so, level, optname, &m0)); - m = NULL; /* freed by protocol */ - } - } - bad: - if (m) - (void) m_free(m); - return (error); + if ((error == 0 || error == ENOPROTOOPT) && + so->so_proto != NULL && so->so_proto->pr_ctloutput != NULL) { + /* give the protocol stack a shot */ + prerr = (*so->so_proto->pr_ctloutput)(PRCO_SETOPT, so, level, + optname, &m); + if (prerr == 0) + error = 0; + else if (prerr != ENOPROTOOPT) + error = prerr; + } else if (m != NULL) + (void)m_free(m); + return error; } int @@ -1795,7 +1816,6 @@ sysctl_kern_somaxkva(SYSCTLFN_ARGS) { int error, new_somaxkva; struct sysctlnode node; - int s; new_somaxkva = somaxkva; node = *rnode; @@ -1807,12 +1827,10 @@ sysctl_kern_somaxkva(SYSCTLFN_ARGS) if (new_somaxkva < (16 * 1024 * 1024)) /* sanity */ return (EINVAL); - s = splvm(); - simple_lock(&so_pendfree_slock); + mutex_enter(&so_pendfree_lock); somaxkva = new_somaxkva; - wakeup(&socurkva); - simple_unlock(&so_pendfree_slock); - splx(s); + cv_broadcast(&socurkva_cv); + mutex_exit(&so_pendfree_lock); return (error); }