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 1.215.4.2 retrieving revision 1.250 diff -u -p -r1.215.4.2 -r1.250 --- src/sys/kern/uipc_socket.c 2013/08/28 23:59:35 1.215.4.2 +++ src/sys/kern/uipc_socket.c 2016/10/10 01:22:08 1.250 @@ -1,4 +1,4 @@ -/* $NetBSD: uipc_socket.c,v 1.215.4.2 2013/08/28 23:59:35 rmind Exp $ */ +/* $NetBSD: uipc_socket.c,v 1.250 2016/10/10 01:22:08 dholland Exp $ */ /*- * Copyright (c) 2002, 2007, 2008, 2009 The NetBSD Foundation, Inc. @@ -71,14 +71,17 @@ */ #include -__KERNEL_RCSID(0, "$NetBSD: uipc_socket.c,v 1.215.4.2 2013/08/28 23:59:35 rmind Exp $"); +__KERNEL_RCSID(0, "$NetBSD: uipc_socket.c,v 1.250 2016/10/10 01:22:08 dholland Exp $"); +#ifdef _KERNEL_OPT #include "opt_compat_netbsd.h" #include "opt_sock_counters.h" #include "opt_sosend_loan.h" #include "opt_mbuftrace.h" #include "opt_somaxkva.h" #include "opt_multiprocessor.h" /* XXX */ +#include "opt_sctp.h" +#endif #include #include @@ -343,7 +346,7 @@ sosend_loan(struct socket *so, struct ui int i; if (VMSPACE_IS_KERNEL_P(uio->uio_vmspace)) - return (0); + return 0L; if (iov->iov_len < (size_t) space) space = iov->iov_len; @@ -359,13 +362,13 @@ sosend_loan(struct socket *so, struct ui lva = sokvaalloc(sva, len, so); if (lva == 0) - return 0; + return 0L; error = uvm_loan(&uio->uio_vmspace->vm_map, sva, len, m->m_ext.ext_pgs, UVM_LOAN_TOPAGE); if (error) { sokvafree(lva, len); - return (0); + return 0L; } for (i = 0, va = lva; i < npgs; i++, va += PAGE_SIZE) @@ -423,7 +426,7 @@ socket_listener_cb(kauth_cred_t cred, ka /* Normal users can only drop their own connections. */ struct socket *so = (struct socket *)arg1; - if (proc_uidmatch(cred, so->so_cred) == 0) + if (so->so_cred && proc_uidmatch(cred, so->so_cred) == 0) result = KAUTH_RESULT_ALLOW; break; @@ -540,24 +543,28 @@ socreate(int dom, struct socket **aso, i so->so_cpid = l->l_proc->p_pid; /* - * Lock assigned and taken during pr_attach, unless we share + * Lock assigned and taken during PCB attach, unless we share * the lock with another socket, e.g. socketpair(2) case. */ if (lockso) { lock = lockso->so_lock; so->so_lock = lock; mutex_obj_hold(lock); + mutex_enter(lock); } + /* Attach the PCB (returns with the socket lock held). */ error = (*prp->pr_usrreqs->pr_attach)(so, proto); + KASSERT(solocked(so)); + if (error) { - solock(so); KASSERT(so->so_pcb == NULL); so->so_state |= SS_NOFDREF; sofree(so); return error; } so->so_cred = kauth_cred_dup(l->l_cred); + sounlock(so); *aso = so; return 0; @@ -570,7 +577,7 @@ socreate(int dom, struct socket **aso, i * => On failure, return non-zero; *fdout will be undefined. */ int -fsocreate(int domain, struct socket **sop, int type, int protocol, int *fdout) +fsocreate(int domain, struct socket **sop, int type, int proto, int *fdout) { lwp_t *l = curlwp; int error, fd, flags; @@ -580,7 +587,6 @@ fsocreate(int domain, struct socket **so if ((error = fd_allocfile(&fp, &fd)) != 0) { return error; } - flags = type & SOCK_FLAGS_MASK; fd_set_exclose(l, fd, (flags & SOCK_CLOEXEC) != 0); fp->f_flag = FREAD|FWRITE|((flags & SOCK_NONBLOCK) ? FNONBLOCK : 0)| @@ -589,17 +595,17 @@ fsocreate(int domain, struct socket **so fp->f_ops = &socketops; type &= ~SOCK_FLAGS_MASK; - error = socreate(domain, &so, type, protocol, l, NULL); - if (error != 0) { + error = socreate(domain, &so, type, proto, l, NULL); + if (error) { fd_abort(curproc, fp, fd); return error; } if (flags & SOCK_NONBLOCK) { so->so_state |= SS_NBIO; } - - fp->f_data = so; + fp->f_socket = so; fd_affix(curproc, fp, fd); + if (sop != NULL) { *sop = so; } @@ -621,13 +627,16 @@ sofamily(const struct socket *so) } int -sobind(struct socket *so, struct mbuf *nam, struct lwp *l) +sobind(struct socket *so, struct sockaddr *nam, struct lwp *l) { - int error; + int error; solock(so); - error = (*so->so_proto->pr_usrreqs->pr_generic)(so, - PRU_BIND, NULL, nam, NULL, l); + if (nam->sa_family != so->so_proto->pr_domain->dom_family) { + sounlock(so); + return EAFNOSUPPORT; + } + error = (*so->so_proto->pr_usrreqs->pr_bind)(so, nam, l); sounlock(so); return error; } @@ -635,25 +644,30 @@ sobind(struct socket *so, struct mbuf *n int solisten(struct socket *so, int backlog, struct lwp *l) { - int error; + int error; + short oldopt, oldqlimit; solock(so); if ((so->so_state & (SS_ISCONNECTED | SS_ISCONNECTING | SS_ISDISCONNECTING)) != 0) { sounlock(so); - return (EINVAL); - } - error = (*so->so_proto->pr_usrreqs->pr_generic)(so, PRU_LISTEN, NULL, - NULL, NULL, l); - if (error != 0) { - sounlock(so); - return error; + return EINVAL; } + oldopt = so->so_options; + oldqlimit = so->so_qlimit; if (TAILQ_EMPTY(&so->so_q)) so->so_options |= SO_ACCEPTCONN; if (backlog < 0) backlog = 0; so->so_qlimit = min(backlog, somaxconn); + + error = (*so->so_proto->pr_usrreqs->pr_listen)(so, l); + if (error != 0) { + so->so_options = oldopt; + so->so_qlimit = oldqlimit; + sounlock(so); + return error; + } sounlock(so); return 0; } @@ -708,9 +722,8 @@ int soclose(struct socket *so) { struct socket *so2; - int error; + int error = 0; - error = 0; solock(so); if (so->so_options & SO_ACCEPTCONN) { for (;;) { @@ -754,6 +767,7 @@ soclose(struct socket *so) } drop: if (so->so_pcb) { + KASSERT(solocked(so)); (*so->so_proto->pr_usrreqs->pr_detach)(so); } discard: @@ -777,8 +791,7 @@ soabort(struct socket *so) KASSERT(so->so_head == NULL); so->so_aborting++; /* XXX */ - error = (*so->so_proto->pr_usrreqs->pr_generic)(so, - PRU_ABORT, NULL, NULL, NULL, NULL); + error = (*so->so_proto->pr_usrreqs->pr_abort)(so); refs = --so->so_aborting; /* XXX */ if (error || (refs == 0)) { sofree(so); @@ -789,35 +802,32 @@ soabort(struct socket *so) } int -soaccept(struct socket *so, struct mbuf *nam) +soaccept(struct socket *so, struct sockaddr *nam) { - int error; + int error; KASSERT(solocked(so)); - - error = 0; KASSERT((so->so_state & SS_NOFDREF) != 0); - so->so_state &= ~SS_NOFDREF; + so->so_state &= ~SS_NOFDREF; if ((so->so_state & SS_ISDISCONNECTED) == 0 || (so->so_proto->pr_flags & PR_ABRTACPTDIS) == 0) - error = (*so->so_proto->pr_usrreqs->pr_generic)(so, - PRU_ACCEPT, NULL, nam, NULL, NULL); + error = (*so->so_proto->pr_usrreqs->pr_accept)(so, nam); else error = ECONNABORTED; - return (error); + return error; } int -soconnect(struct socket *so, struct mbuf *nam, struct lwp *l) +soconnect(struct socket *so, struct sockaddr *nam, struct lwp *l) { int error; KASSERT(solocked(so)); if (so->so_options & SO_ACCEPTCONN) - return (EOPNOTSUPP); + return EOPNOTSUPP; /* * If protocol is connection-based, can only connect once. * Otherwise, if connected, try to disconnect first. @@ -826,11 +836,14 @@ soconnect(struct socket *so, struct mbuf */ if (so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING) && ((so->so_proto->pr_flags & PR_CONNREQUIRED) || - (error = sodisconnect(so)))) + (error = sodisconnect(so)))) { error = EISCONN; - else - error = (*so->so_proto->pr_usrreqs->pr_generic)(so, - PRU_CONNECT, NULL, nam, NULL, l); + } else { + if (nam->sa_family != so->so_proto->pr_domain->dom_family) { + return EAFNOSUPPORT; + } + error = (*so->so_proto->pr_usrreqs->pr_connect)(so, nam, l); + } return error; } @@ -838,13 +851,9 @@ soconnect(struct socket *so, struct mbuf int soconnect2(struct socket *so1, struct socket *so2) { - int error; - KASSERT(solocked2(so1, so2)); - error = (*so1->so_proto->pr_usrreqs->pr_generic)(so1, PRU_CONNECT2, - NULL, (struct mbuf *)so2, NULL, NULL); - return (error); + return (*so1->so_proto->pr_usrreqs->pr_connect2)(so1, so2); } int @@ -859,8 +868,7 @@ sodisconnect(struct socket *so) } else if (so->so_state & SS_ISDISCONNECTING) { error = EALREADY; } else { - error = (*so->so_proto->pr_usrreqs->pr_generic)(so, - PRU_DISCONNECT, NULL, NULL, NULL, NULL); + error = (*so->so_proto->pr_usrreqs->pr_disconnect)(so); } return (error); } @@ -884,8 +892,8 @@ sodisconnect(struct socket *so) * Data and control buffers are freed on return. */ int -sosend(struct socket *so, struct mbuf *addr, struct uio *uio, struct mbuf *top, - struct mbuf *control, int flags, struct lwp *l) +sosend(struct socket *so, struct sockaddr *addr, struct uio *uio, + struct mbuf *top, struct mbuf *control, int flags, struct lwp *l) { struct mbuf **mp, *m; long space, len, resid, clen, mlen; @@ -938,12 +946,11 @@ sosend(struct socket *so, struct mbuf *a } if ((so->so_state & SS_ISCONNECTED) == 0) { if (so->so_proto->pr_flags & PR_CONNREQUIRED) { - if ((so->so_state & SS_ISCONFIRMING) == 0 && - !(resid == 0 && clen != 0)) { + if (resid || clen == 0) { error = ENOTCONN; goto release; } - } else if (addr == 0) { + } else if (addr == NULL) { error = EDESTADDRREQ; goto release; } @@ -991,7 +998,7 @@ sosend(struct socket *so, struct mbuf *a m = m_gethdr(M_WAIT, MT_DATA); mlen = MHLEN; m->m_pkthdr.len = 0; - m->m_pkthdr.rcvif = NULL; + m_reset_rcvif(m); } else { m = m_get(M_WAIT, MT_DATA); mlen = MLEN; @@ -1001,7 +1008,7 @@ sosend(struct socket *so, struct mbuf *a uio->uio_iov->iov_len >= sock_loan_thresh && space >= sock_loan_thresh && (len = sosend_loan(so, uio, m, - space)) != 0) { + space)) != 0L) { SOSEND_COUNTER_INCR(&sosend_loan_big); space -= len; goto have_data; @@ -1057,9 +1064,13 @@ sosend(struct socket *so, struct mbuf *a so->so_options |= SO_DONTROUTE; if (resid > 0) so->so_state |= SS_MORETOCOME; - error = (*so->so_proto->pr_usrreqs->pr_generic)(so, - (flags & MSG_OOB) ? PRU_SENDOOB : PRU_SEND, - top, addr, control, curlwp); + if (flags & MSG_OOB) { + error = (*so->so_proto->pr_usrreqs->pr_sendoob)(so, + top, control); + } else { + error = (*so->so_proto->pr_usrreqs->pr_send)(so, + top, addr, control, l); + } if (dontroute) so->so_options &= ~SO_DONTROUTE; if (resid > 0) @@ -1171,8 +1182,7 @@ soreceive(struct socket *so, struct mbuf if (flags & MSG_OOB) { m = m_get(M_WAIT, MT_DATA); solock(so); - error = (*pr->pr_usrreqs->pr_generic)(so, PRU_RCVOOB, m, - (struct mbuf *)(long)(flags & MSG_PEEK), NULL, l); + error = (*pr->pr_usrreqs->pr_recvoob)(so, m, flags & MSG_PEEK); sounlock(so); if (error) goto bad; @@ -1196,9 +1206,6 @@ soreceive(struct socket *so, struct mbuf */ s = splsoftnet(); solock(so); - if (so->so_state & SS_ISCONFIRMING && uio->uio_resid) - (*pr->pr_usrreqs->pr_generic)(so, PRU_RCVD, NULL, NULL, NULL, l); - restart: if ((error = sblock(&so->so_rcv, SBLOCKWAIT(flags))) != 0) { sounlock(so); @@ -1317,12 +1324,35 @@ soreceive(struct socket *so, struct mbuf m->m_next = NULL; m = so->so_rcv.sb_mb; } else { - MFREE(m, so->so_rcv.sb_mb); - m = so->so_rcv.sb_mb; + m = so->so_rcv.sb_mb = m_free(m); } sbsync(&so->so_rcv, nextrecord); } } + if (pr->pr_flags & PR_ADDR_OPT) { + /* + * For SCTP we may be getting a + * whole message OR a partial delivery. + */ + if (m->m_type == MT_SONAME) { + orig_resid = 0; + if (flags & MSG_PEEK) { + if (paddr) + *paddr = m_copy(m, 0, m->m_len); + m = m->m_next; + } else { + sbfree(&so->so_rcv, m); + if (paddr) { + *paddr = m; + so->so_rcv.sb_mb = m->m_next; + m->m_next = 0; + m = so->so_rcv.sb_mb; + } else { + m = so->so_rcv.sb_mb = m_free(m); + } + } + } + } /* * Process one or more MT_CONTROL mbufs present before any data mbufs @@ -1457,6 +1487,10 @@ soreceive(struct socket *so, struct mbuf if (len == m->m_len - moff) { if (m->m_flags & M_EOR) flags |= MSG_EOR; +#ifdef SCTP + if (m->m_flags & M_NOTIFICATION) + flags |= MSG_NOTIFICATION; +#endif /* SCTP */ if (flags & MSG_PEEK) { m = m->m_next; moff = 0; @@ -1469,8 +1503,7 @@ soreceive(struct socket *so, struct mbuf so->so_rcv.sb_mb = m = m->m_next; *mp = NULL; } else { - MFREE(m, so->so_rcv.sb_mb); - m = so->so_rcv.sb_mb; + m = so->so_rcv.sb_mb = m_free(m); } /* * If m != NULL, we also know that @@ -1542,8 +1575,7 @@ soreceive(struct socket *so, struct mbuf * get it filled again. */ if ((pr->pr_flags & PR_WANTRCVD) && so->so_pcb) - (*pr->pr_usrreqs->pr_generic)(so, PRU_RCVD, - NULL, (struct mbuf *)(long)flags, NULL, l); + (*pr->pr_usrreqs->pr_rcvd)(so, flags, l); SBLASTRECORDCHK(&so->so_rcv, "soreceive sbwait 2"); SBLASTMBUFCHK(&so->so_rcv, "soreceive sbwait 2"); if (wakeup_state & SS_RESTARTSYS) @@ -1584,8 +1616,7 @@ 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_usrreqs->pr_generic)(so, PRU_RCVD, NULL, - (struct mbuf *)(long)flags, NULL, l); + (*pr->pr_usrreqs->pr_rcvd)(so, flags, l); } if (orig_resid == uio->uio_resid && orig_resid && (flags & MSG_EOR) == 0 && (so->so_state & SS_CANTRCVMORE) == 0) { @@ -1619,8 +1650,7 @@ soshutdown(struct socket *so, int how) error = 0; } if (how == SHUT_WR || how == SHUT_RDWR) - error = (*pr->pr_usrreqs->pr_generic)(so, PRU_SHUTDOWN, NULL, - NULL, NULL, NULL); + error = (*pr->pr_usrreqs->pr_shutdown)(so); return error; } @@ -1679,7 +1709,8 @@ sorflush(struct socket *so) static int sosetopt1(struct socket *so, const struct sockopt *sopt) { - int error = EINVAL, optval, opt; + int error = EINVAL, opt; + int optval = 0; /* XXX: gcc */ struct linger l; struct timeval tv; @@ -1921,6 +1952,7 @@ sogetopt1(struct socket *so, struct sock #ifdef SO_OTIMESTAMP case SO_OTIMESTAMP: #endif + case SO_ACCEPTCONN: error = sockopt_setint(sopt, (so->so_options & opt) ? 1 : 0); break; @@ -2185,7 +2217,7 @@ filt_sordetach(struct knote *kn) { struct socket *so; - so = ((file_t *)kn->kn_obj)->f_data; + so = ((file_t *)kn->kn_obj)->f_socket; solock(so); SLIST_REMOVE(&so->so_rcv.sb_sel.sel_klist, kn, knote, kn_selnext); if (SLIST_EMPTY(&so->so_rcv.sb_sel.sel_klist)) @@ -2200,7 +2232,7 @@ filt_soread(struct knote *kn, long hint) struct socket *so; int rv; - so = ((file_t *)kn->kn_obj)->f_data; + so = ((file_t *)kn->kn_obj)->f_socket; if (hint != NOTE_SUBMIT) solock(so); kn->kn_data = so->so_rcv.sb_cc; @@ -2224,7 +2256,7 @@ filt_sowdetach(struct knote *kn) { struct socket *so; - so = ((file_t *)kn->kn_obj)->f_data; + so = ((file_t *)kn->kn_obj)->f_socket; solock(so); SLIST_REMOVE(&so->so_snd.sb_sel.sel_klist, kn, knote, kn_selnext); if (SLIST_EMPTY(&so->so_snd.sb_sel.sel_klist)) @@ -2239,7 +2271,7 @@ filt_sowrite(struct knote *kn, long hint struct socket *so; int rv; - so = ((file_t *)kn->kn_obj)->f_data; + so = ((file_t *)kn->kn_obj)->f_socket; if (hint != NOTE_SUBMIT) solock(so); kn->kn_data = sbspace(&so->so_snd); @@ -2268,7 +2300,7 @@ filt_solisten(struct knote *kn, long hin struct socket *so; int rv; - so = ((file_t *)kn->kn_obj)->f_data; + so = ((file_t *)kn->kn_obj)->f_socket; /* * Set kn_data to number of incoming connections, not @@ -2296,7 +2328,7 @@ soo_kqfilter(struct file *fp, struct kno struct socket *so; struct sockbuf *sb; - so = ((file_t *)kn->kn_obj)->f_data; + so = ((file_t *)kn->kn_obj)->f_socket; solock(so); switch (kn->kn_filter) { case EVFILT_READ: @@ -2438,11 +2470,6 @@ sysctl_kern_socket_setup(void) { KASSERT(socket_sysctllog == NULL); - sysctl_createv(&socket_sysctllog, 0, NULL, NULL, - CTLFLAG_PERMANENT, - CTLTYPE_NODE, "kern", NULL, - NULL, 0, NULL, 0, - CTL_KERN, CTL_EOL); sysctl_createv(&socket_sysctllog, 0, NULL, NULL, CTLFLAG_PERMANENT|CTLFLAG_READWRITE,