Index: head/sys/compat/svr4/svr4_stream.c =================================================================== --- head/sys/compat/svr4/svr4_stream.c (revision 160764) +++ head/sys/compat/svr4/svr4_stream.c (revision 160765) @@ -1,1993 +1,2007 @@ /*- * Copyright (c) 1998 Mark Newton. All rights reserved. * Copyright (c) 1994, 1996 Christos Zoulas. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by Christos Zoulas. * 4. The name of the author may not be used to endorse or promote products * derived from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ /* * Pretend that we have streams... * Yes, this is gross. * * ToDo: The state machine for getmsg needs re-thinking */ #include __FBSDID("$FreeBSD$"); #include "opt_compat.h" #include "opt_ktrace.h" #include "opt_mac.h" #include #include #include #include #include #include #include #include /* Must come after sys/malloc.h */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* Must come after sys/uio.h */ #include #include #include #include #include #include #include #include #include #include #include #include /* Utils */ static int clean_pipe(struct thread *, char *); static void getparm(struct file *, struct svr4_si_sockparms *); static int svr4_do_putmsg(struct thread *, struct svr4_sys_putmsg_args *, struct file *); static int svr4_do_getmsg(struct thread *, struct svr4_sys_getmsg_args *, struct file *); /* Address Conversions */ static void sockaddr_to_netaddr_in(struct svr4_strmcmd *, const struct sockaddr_in *); static void sockaddr_to_netaddr_un(struct svr4_strmcmd *, const struct sockaddr_un *); static void netaddr_to_sockaddr_in(struct sockaddr_in *, const struct svr4_strmcmd *); static void netaddr_to_sockaddr_un(struct sockaddr_un *, const struct svr4_strmcmd *); /* stream ioctls */ static int i_nread(struct file *, struct thread *, register_t *, int, u_long, caddr_t); static int i_fdinsert(struct file *, struct thread *, register_t *, int, u_long, caddr_t); static int i_str(struct file *, struct thread *, register_t *, int, u_long, caddr_t); static int i_setsig(struct file *, struct thread *, register_t *, int, u_long, caddr_t); static int i_getsig(struct file *, struct thread *, register_t *, int, u_long, caddr_t); static int _i_bind_rsvd(struct file *, struct thread *, register_t *, int, u_long, caddr_t); static int _i_rele_rsvd(struct file *, struct thread *, register_t *, int, u_long, caddr_t); /* i_str sockmod calls */ static int sockmod(struct file *, int, struct svr4_strioctl *, struct thread *); static int si_listen(struct file *, int, struct svr4_strioctl *, struct thread *); static int si_ogetudata(struct file *, int, struct svr4_strioctl *, struct thread *); static int si_sockparams(struct file *, int, struct svr4_strioctl *, struct thread *); static int si_shutdown (struct file *, int, struct svr4_strioctl *, struct thread *); static int si_getudata(struct file *, int, struct svr4_strioctl *, struct thread *); /* i_str timod calls */ static int timod(struct file *, int, struct svr4_strioctl *, struct thread *); static int ti_getinfo(struct file *, int, struct svr4_strioctl *, struct thread *); static int ti_bind(struct file *, int, struct svr4_strioctl *, struct thread *); #ifdef DEBUG_SVR4 static void bufprint(u_char *, size_t); static int show_ioc(const char *, struct svr4_strioctl *); static int show_strbuf(struct svr4_strbuf *); static void show_msg(const char *, int, struct svr4_strbuf *, struct svr4_strbuf *, int); static void bufprint(buf, len) u_char *buf; size_t len; { size_t i; uprintf("\n\t"); for (i = 0; i < len; i++) { uprintf("%x ", buf[i]); if (i && (i % 16) == 0) uprintf("\n\t"); } } static int show_ioc(str, ioc) const char *str; struct svr4_strioctl *ioc; { u_char *ptr = NULL; int len; int error; len = ioc->len; if (len > 1024) len = 1024; if (len > 0) { ptr = (u_char *) malloc(len, M_TEMP, M_WAITOK); if ((error = copyin(ioc->buf, ptr, len)) != 0) { free((char *) ptr, M_TEMP); return error; } } uprintf("%s cmd = %ld, timeout = %d, len = %d, buf = %p { ", str, ioc->cmd, ioc->timeout, ioc->len, ioc->buf); if (ptr != NULL) bufprint(ptr, len); uprintf("}\n"); if (ptr != NULL) free((char *) ptr, M_TEMP); return 0; } static int show_strbuf(str) struct svr4_strbuf *str; { int error; u_char *ptr = NULL; int maxlen = str->maxlen; int len = str->len; if (maxlen > 8192) maxlen = 8192; if (maxlen < 0) maxlen = 0; if (len >= maxlen) len = maxlen; if (len > 0) { ptr = (u_char *) malloc(len, M_TEMP, M_WAITOK); if ((error = copyin(str->buf, ptr, len)) != 0) { free((char *) ptr, M_TEMP); return error; } } uprintf(", { %d, %d, %p=[ ", str->maxlen, str->len, str->buf); if (ptr) bufprint(ptr, len); uprintf("]}"); if (ptr) free((char *) ptr, M_TEMP); return 0; } static void show_msg(str, fd, ctl, dat, flags) const char *str; int fd; struct svr4_strbuf *ctl; struct svr4_strbuf *dat; int flags; { struct svr4_strbuf buf; int error; uprintf("%s(%d", str, fd); if (ctl != NULL) { if ((error = copyin(ctl, &buf, sizeof(buf))) != 0) return; show_strbuf(&buf); } else uprintf(", NULL"); if (dat != NULL) { if ((error = copyin(dat, &buf, sizeof(buf))) != 0) return; show_strbuf(&buf); } else uprintf(", NULL"); uprintf(", %x);\n", flags); } #endif /* DEBUG_SVR4 */ /* * We are faced with an interesting situation. On svr4 unix sockets * are really pipes. But we really have sockets, and we might as * well use them. At the point where svr4 calls TI_BIND, it has * already created a named pipe for the socket using mknod(2). * We need to create a socket with the same name when we bind, * so we need to remove the pipe before, otherwise we'll get address * already in use. So we *carefully* remove the pipe, to avoid * using this as a random file removal tool. We use system calls * to avoid code duplication. */ static int clean_pipe(td, path) struct thread *td; char *path; { struct stat st; int error; error = kern_lstat(td, path, UIO_SYSSPACE, &st); /* * Make sure we are dealing with a mode 0 named pipe. */ if ((st.st_mode & S_IFMT) != S_IFIFO) return (0); if ((st.st_mode & ALLPERMS) != 0) return (0); error = kern_unlink(td, path, UIO_SYSSPACE); if (error) DPRINTF(("clean_pipe: unlink failed %d\n", error)); return (error); } static void sockaddr_to_netaddr_in(sc, sain) struct svr4_strmcmd *sc; const struct sockaddr_in *sain; { struct svr4_netaddr_in *na; na = SVR4_ADDROF(sc); na->family = sain->sin_family; na->port = sain->sin_port; na->addr = sain->sin_addr.s_addr; DPRINTF(("sockaddr_in -> netaddr %d %d %lx\n", na->family, na->port, na->addr)); } static void sockaddr_to_netaddr_un(sc, saun) struct svr4_strmcmd *sc; const struct sockaddr_un *saun; { struct svr4_netaddr_un *na; char *dst, *edst = ((char *) sc) + sc->offs + sizeof(na->family) + 1 - sizeof(*sc); const char *src; na = SVR4_ADDROF(sc); na->family = saun->sun_family; for (src = saun->sun_path, dst = na->path; (*dst++ = *src++) != '\0'; ) if (dst == edst) break; DPRINTF(("sockaddr_un -> netaddr %d %s\n", na->family, na->path)); } static void netaddr_to_sockaddr_in(sain, sc) struct sockaddr_in *sain; const struct svr4_strmcmd *sc; { const struct svr4_netaddr_in *na; na = SVR4_C_ADDROF(sc); memset(sain, 0, sizeof(*sain)); sain->sin_len = sizeof(*sain); sain->sin_family = na->family; sain->sin_port = na->port; sain->sin_addr.s_addr = na->addr; DPRINTF(("netaddr -> sockaddr_in %d %d %x\n", sain->sin_family, sain->sin_port, sain->sin_addr.s_addr)); } static void netaddr_to_sockaddr_un(saun, sc) struct sockaddr_un *saun; const struct svr4_strmcmd *sc; { const struct svr4_netaddr_un *na; char *dst, *edst = &saun->sun_path[sizeof(saun->sun_path) - 1]; const char *src; na = SVR4_C_ADDROF(sc); memset(saun, 0, sizeof(*saun)); saun->sun_family = na->family; for (src = na->path, dst = saun->sun_path; (*dst++ = *src++) != '\0'; ) if (dst == edst) break; saun->sun_len = dst - saun->sun_path; DPRINTF(("netaddr -> sockaddr_un %d %s\n", saun->sun_family, saun->sun_path)); } static void getparm(fp, pa) struct file *fp; struct svr4_si_sockparms *pa; { struct svr4_strm *st; struct socket *so; st = svr4_stream_get(fp); if (st == NULL) return; so = fp->f_data; pa->family = st->s_family; switch (so->so_type) { case SOCK_DGRAM: pa->type = SVR4_T_CLTS; pa->protocol = IPPROTO_UDP; DPRINTF(("getparm(dgram)\n")); return; case SOCK_STREAM: pa->type = SVR4_T_COTS; /* What about T_COTS_ORD? XXX */ pa->protocol = IPPROTO_IP; DPRINTF(("getparm(stream)\n")); return; case SOCK_RAW: pa->type = SVR4_T_CLTS; pa->protocol = IPPROTO_RAW; DPRINTF(("getparm(raw)\n")); return; default: pa->type = 0; pa->protocol = 0; DPRINTF(("getparm(type %d?)\n", so->so_type)); return; } } static int si_ogetudata(fp, fd, ioc, td) struct file *fp; int fd; struct svr4_strioctl *ioc; struct thread *td; { int error; struct svr4_si_oudata ud; struct svr4_si_sockparms pa; if (ioc->len != sizeof(ud) && ioc->len != sizeof(ud) - sizeof(int)) { DPRINTF(("SI_OGETUDATA: Wrong size %d != %d\n", sizeof(ud), ioc->len)); return EINVAL; } if ((error = copyin(ioc->buf, &ud, sizeof(ud))) != 0) return error; getparm(fp, &pa); switch (pa.family) { case AF_INET: ud.tidusize = 16384; ud.addrsize = sizeof(struct svr4_sockaddr_in); if (pa.type == SVR4_SOCK_STREAM) ud.etsdusize = 1; else ud.etsdusize = 0; break; case AF_LOCAL: ud.tidusize = 65536; ud.addrsize = 128; ud.etsdusize = 128; break; default: DPRINTF(("SI_OGETUDATA: Unsupported address family %d\n", pa.family)); return ENOSYS; } /* I have no idea what these should be! */ ud.optsize = 128; ud.tsdusize = 128; ud.servtype = pa.type; /* XXX: Fixme */ ud.so_state = 0; ud.so_options = 0; return copyout(&ud, ioc->buf, ioc->len); } static int si_sockparams(fp, fd, ioc, td) struct file *fp; int fd; struct svr4_strioctl *ioc; struct thread *td; { struct svr4_si_sockparms pa; getparm(fp, &pa); return copyout(&pa, ioc->buf, sizeof(pa)); } static int si_listen(fp, fd, ioc, td) struct file *fp; int fd; struct svr4_strioctl *ioc; struct thread *td; { int error; struct svr4_strm *st = svr4_stream_get(fp); struct svr4_strmcmd lst; struct listen_args la; if (st == NULL) return EINVAL; if (ioc->len < 0 || ioc->len > sizeof(lst)) return EINVAL; if ((error = copyin(ioc->buf, &lst, ioc->len)) != 0) return error; if (lst.cmd != SVR4_TI_OLD_BIND_REQUEST) { DPRINTF(("si_listen: bad request %ld\n", lst.cmd)); return EINVAL; } /* * We are making assumptions again... */ la.s = fd; DPRINTF(("SI_LISTEN: fileno %d backlog = %d\n", fd, 5)); la.backlog = 5; if ((error = listen(td, &la)) != 0) { DPRINTF(("SI_LISTEN: listen failed %d\n", error)); return error; } st->s_cmd = SVR4_TI__ACCEPT_WAIT; lst.cmd = SVR4_TI_BIND_REPLY; switch (st->s_family) { case AF_INET: /* XXX: Fill the length here */ break; case AF_LOCAL: lst.len = 140; lst.pad[28] = 0x00000000; /* magic again */ lst.pad[29] = 0x00000800; /* magic again */ lst.pad[30] = 0x80001400; /* magic again */ break; default: DPRINTF(("SI_LISTEN: Unsupported address family %d\n", st->s_family)); return ENOSYS; } if ((error = copyout(&lst, ioc->buf, ioc->len)) != 0) return error; return 0; } static int si_getudata(fp, fd, ioc, td) struct file *fp; int fd; struct svr4_strioctl *ioc; struct thread *td; { int error; struct svr4_si_udata ud; if (sizeof(ud) != ioc->len) { DPRINTF(("SI_GETUDATA: Wrong size %d != %d\n", sizeof(ud), ioc->len)); return EINVAL; } if ((error = copyin(ioc->buf, &ud, sizeof(ud))) != 0) return error; getparm(fp, &ud.sockparms); switch (ud.sockparms.family) { case AF_INET: DPRINTF(("getudata_inet\n")); ud.tidusize = 16384; ud.tsdusize = 16384; ud.addrsize = sizeof(struct svr4_sockaddr_in); if (ud.sockparms.type == SVR4_SOCK_STREAM) ud.etsdusize = 1; else ud.etsdusize = 0; ud.optsize = 0; break; case AF_LOCAL: DPRINTF(("getudata_local\n")); ud.tidusize = 65536; ud.tsdusize = 128; ud.addrsize = 128; ud.etsdusize = 128; ud.optsize = 128; break; default: DPRINTF(("SI_GETUDATA: Unsupported address family %d\n", ud.sockparms.family)); return ENOSYS; } ud.servtype = ud.sockparms.type; DPRINTF(("ud.servtype = %d\n", ud.servtype)); /* XXX: Fixme */ ud.so_state = 0; ud.so_options = 0; return copyout(&ud, ioc->buf, sizeof(ud)); } static int si_shutdown(fp, fd, ioc, td) struct file *fp; int fd; struct svr4_strioctl *ioc; struct thread *td; { int error; struct shutdown_args ap; if (ioc->len != sizeof(ap.how)) { DPRINTF(("SI_SHUTDOWN: Wrong size %d != %d\n", sizeof(ap.how), ioc->len)); return EINVAL; } if ((error = copyin(ioc->buf, &ap.how, ioc->len)) != 0) return error; ap.s = fd; return shutdown(td, &ap); } static int sockmod(fp, fd, ioc, td) struct file *fp; int fd; struct svr4_strioctl *ioc; struct thread *td; { switch (ioc->cmd) { case SVR4_SI_OGETUDATA: DPRINTF(("SI_OGETUDATA\n")); return si_ogetudata(fp, fd, ioc, td); case SVR4_SI_SHUTDOWN: DPRINTF(("SI_SHUTDOWN\n")); return si_shutdown(fp, fd, ioc, td); case SVR4_SI_LISTEN: DPRINTF(("SI_LISTEN\n")); return si_listen(fp, fd, ioc, td); case SVR4_SI_SETMYNAME: DPRINTF(("SI_SETMYNAME\n")); return 0; case SVR4_SI_SETPEERNAME: DPRINTF(("SI_SETPEERNAME\n")); return 0; case SVR4_SI_GETINTRANSIT: DPRINTF(("SI_GETINTRANSIT\n")); return 0; case SVR4_SI_TCL_LINK: DPRINTF(("SI_TCL_LINK\n")); return 0; case SVR4_SI_TCL_UNLINK: DPRINTF(("SI_TCL_UNLINK\n")); return 0; case SVR4_SI_SOCKPARAMS: DPRINTF(("SI_SOCKPARAMS\n")); return si_sockparams(fp, fd, ioc, td); case SVR4_SI_GETUDATA: DPRINTF(("SI_GETUDATA\n")); return si_getudata(fp, fd, ioc, td); default: DPRINTF(("Unknown sockmod ioctl %lx\n", ioc->cmd)); return 0; } } static int ti_getinfo(fp, fd, ioc, td) struct file *fp; int fd; struct svr4_strioctl *ioc; struct thread *td; { int error; struct svr4_infocmd info; memset(&info, 0, sizeof(info)); if (ioc->len < 0 || ioc->len > sizeof(info)) return EINVAL; if ((error = copyin(ioc->buf, &info, ioc->len)) != 0) return error; if (info.cmd != SVR4_TI_INFO_REQUEST) return EINVAL; info.cmd = SVR4_TI_INFO_REPLY; info.tsdu = 0; info.etsdu = 1; info.cdata = -2; info.ddata = -2; info.addr = 16; info.opt = -1; info.tidu = 16384; info.serv = 2; info.current = 0; info.provider = 2; ioc->len = sizeof(info); if ((error = copyout(&info, ioc->buf, ioc->len)) != 0) return error; return 0; } static int ti_bind(fp, fd, ioc, td) struct file *fp; int fd; struct svr4_strioctl *ioc; struct thread *td; { int error; struct svr4_strm *st = svr4_stream_get(fp); struct sockaddr_in sain; struct sockaddr_un saun; struct sockaddr *skp; int sasize; struct svr4_strmcmd bnd; if (st == NULL) { DPRINTF(("ti_bind: bad file descriptor\n")); return EINVAL; } if (ioc->len < 0 || ioc->len > sizeof(bnd)) return EINVAL; if ((error = copyin(ioc->buf, &bnd, ioc->len)) != 0) return error; if (bnd.cmd != SVR4_TI_OLD_BIND_REQUEST) { DPRINTF(("ti_bind: bad request %ld\n", bnd.cmd)); return EINVAL; } switch (st->s_family) { case AF_INET: skp = (struct sockaddr *)&sain; sasize = sizeof(sain); if (bnd.offs == 0) goto error; netaddr_to_sockaddr_in(&sain, &bnd); DPRINTF(("TI_BIND: fam %d, port %d, addr %x\n", sain.sin_family, sain.sin_port, sain.sin_addr.s_addr)); break; case AF_LOCAL: skp = (struct sockaddr *)&saun; sasize = sizeof(saun); if (bnd.offs == 0) goto error; netaddr_to_sockaddr_un(&saun, &bnd); if (saun.sun_path[0] == '\0') goto error; DPRINTF(("TI_BIND: fam %d, path %s\n", saun.sun_family, saun.sun_path)); if ((error = clean_pipe(td, saun.sun_path)) != 0) return error; bnd.pad[28] = 0x00001000; /* magic again */ break; default: DPRINTF(("TI_BIND: Unsupported address family %d\n", st->s_family)); return ENOSYS; } DPRINTF(("TI_BIND: fileno %d\n", fd)); if ((error = kern_bind(td, fd, skp)) != 0) { DPRINTF(("TI_BIND: bind failed %d\n", error)); return error; } goto reply; error: memset(&bnd, 0, sizeof(bnd)); bnd.len = sasize + 4; bnd.offs = 0x10; /* XXX */ reply: bnd.cmd = SVR4_TI_BIND_REPLY; if ((error = copyout(&bnd, ioc->buf, ioc->len)) != 0) return error; return 0; } static int timod(fp, fd, ioc, td) struct file *fp; int fd; struct svr4_strioctl *ioc; struct thread *td; { switch (ioc->cmd) { case SVR4_TI_GETINFO: DPRINTF(("TI_GETINFO\n")); return ti_getinfo(fp, fd, ioc, td); case SVR4_TI_OPTMGMT: DPRINTF(("TI_OPTMGMT\n")); return 0; case SVR4_TI_BIND: DPRINTF(("TI_BIND\n")); return ti_bind(fp, fd, ioc, td); case SVR4_TI_UNBIND: DPRINTF(("TI_UNBIND\n")); return 0; default: DPRINTF(("Unknown timod ioctl %lx\n", ioc->cmd)); return 0; } } int svr4_stream_ti_ioctl(fp, td, retval, fd, cmd, dat) struct file *fp; struct thread *td; register_t *retval; int fd; u_long cmd; caddr_t dat; { struct svr4_strbuf skb, *sub = (struct svr4_strbuf *) dat; struct svr4_strm *st = svr4_stream_get(fp); int error; struct sockaddr *sa; socklen_t sasize, oldsasize; struct svr4_strmcmd sc; DPRINTF(("svr4_stream_ti_ioctl\n")); if (st == NULL) return EINVAL; sc.offs = 0x10; if ((error = copyin(sub, &skb, sizeof(skb))) != 0) { DPRINTF(("ti_ioctl: error copying in strbuf\n")); return error; } switch (st->s_family) { case AF_INET: sasize = sizeof(struct sockaddr_in); break; case AF_LOCAL: sasize = sizeof(struct sockaddr_un); break; default: DPRINTF(("ti_ioctl: Unsupported address family %d\n", st->s_family)); return ENOSYS; } oldsasize = sasize; switch (cmd) { case SVR4_TI_GETMYNAME: DPRINTF(("TI_GETMYNAME\n")); { error = kern_getsockname(td, fd, &sa, &sasize); if (error) { DPRINTF(("ti_ioctl: getsockname error\n")); return error; } } break; case SVR4_TI_GETPEERNAME: DPRINTF(("TI_GETPEERNAME\n")); { error = kern_getpeername(td, fd, &sa, &sasize); if (error) { DPRINTF(("ti_ioctl: getpeername error\n")); return error; } } break; case SVR4_TI_SETMYNAME: DPRINTF(("TI_SETMYNAME\n")); return 0; case SVR4_TI_SETPEERNAME: DPRINTF(("TI_SETPEERNAME\n")); return 0; default: DPRINTF(("ti_ioctl: Unknown ioctl %lx\n", cmd)); return ENOSYS; } if (sasize < 0 || sasize > oldsasize) { free(sa, M_SONAME); return EINVAL; } switch (st->s_family) { case AF_INET: sockaddr_to_netaddr_in(&sc, (struct sockaddr_in *)sa); skb.len = sasize; break; case AF_LOCAL: sockaddr_to_netaddr_un(&sc, (struct sockaddr_un *)sa); skb.len = sasize + 4; break; default: free(sa, M_SONAME); return ENOSYS; } free(sa, M_SONAME); if ((error = copyout(SVR4_ADDROF(&sc), skb.buf, sasize)) != 0) { DPRINTF(("ti_ioctl: error copying out socket data\n")); return error; } if ((error = copyout(&skb, sub, sizeof(skb))) != 0) { DPRINTF(("ti_ioctl: error copying out strbuf\n")); return error; } return error; } static int i_nread(fp, td, retval, fd, cmd, dat) struct file *fp; struct thread *td; register_t *retval; int fd; u_long cmd; caddr_t dat; { int error; int nread = 0; /* * We are supposed to return the message length in nread, and the * number of messages in retval. We don't have the notion of number * of stream messages, so we just find out if we have any bytes waiting * for us, and if we do, then we assume that we have at least one * message waiting for us. */ if ((error = fo_ioctl(fp, FIONREAD, (caddr_t) &nread, td->td_ucred, td)) != 0) return error; if (nread != 0) *retval = 1; else *retval = 0; return copyout(&nread, dat, sizeof(nread)); } static int i_fdinsert(fp, td, retval, fd, cmd, dat) struct file *fp; struct thread *td; register_t *retval; int fd; u_long cmd; caddr_t dat; { /* * Major hack again here. We assume that we are using this to * implement accept(2). If that is the case, we have already * called accept, and we have stored the file descriptor in * afd. We find the file descriptor that the code wants to use * in fd insert, and then we dup2() our accepted file descriptor * to it. */ int error; struct svr4_strm *st = svr4_stream_get(fp); struct svr4_strfdinsert fdi; struct dup2_args d2p; if (st == NULL) { DPRINTF(("fdinsert: bad file type\n")); return EINVAL; } if (st->s_afd == -1) { DPRINTF(("fdinsert: accept fd not found\n")); return ENOENT; } if ((error = copyin(dat, &fdi, sizeof(fdi))) != 0) { DPRINTF(("fdinsert: copyin failed %d\n", error)); return error; } d2p.from = st->s_afd; d2p.to = fdi.fd; if ((error = dup2(td, &d2p)) != 0) { DPRINTF(("fdinsert: dup2(%d, %d) failed %d\n", st->s_afd, fdi.fd, error)); return error; } if ((error = kern_close(td, st->s_afd)) != 0) { DPRINTF(("fdinsert: close(%d) failed %d\n", st->s_afd, error)); return error; } st->s_afd = -1; *retval = 0; return 0; } static int _i_bind_rsvd(fp, td, retval, fd, cmd, dat) struct file *fp; struct thread *td; register_t *retval; int fd; u_long cmd; caddr_t dat; { struct mkfifo_args ap; /* * This is a supposed to be a kernel and library only ioctl. * It gets called before ti_bind, when we have a unix * socket, to physically create the socket transport and * ``reserve'' it. I don't know how this get reserved inside * the kernel, but we are going to create it nevertheless. */ ap.path = dat; ap.mode = S_IFIFO; return mkfifo(td, &ap); } static int _i_rele_rsvd(fp, td, retval, fd, cmd, dat) struct file *fp; struct thread *td; register_t *retval; int fd; u_long cmd; caddr_t dat; { struct unlink_args ap; /* * This is a supposed to be a kernel and library only ioctl. * I guess it is supposed to release the socket. */ ap.path = dat; return unlink(td, &ap); } static int i_str(fp, td, retval, fd, cmd, dat) struct file *fp; struct thread *td; register_t *retval; int fd; u_long cmd; caddr_t dat; { int error; struct svr4_strioctl ioc; if ((error = copyin(dat, &ioc, sizeof(ioc))) != 0) return error; #ifdef DEBUG_SVR4 if ((error = show_ioc(">", &ioc)) != 0) return error; #endif /* DEBUG_SVR4 */ switch (ioc.cmd & 0xff00) { case SVR4_SIMOD: if ((error = sockmod(fp, fd, &ioc, td)) != 0) return error; break; case SVR4_TIMOD: if ((error = timod(fp, fd, &ioc, td)) != 0) return error; break; default: DPRINTF(("Unimplemented module %c %ld\n", (char) (cmd >> 8), cmd & 0xff)); return 0; } #ifdef DEBUG_SVR4 if ((error = show_ioc("<", &ioc)) != 0) return error; #endif /* DEBUG_SVR4 */ return copyout(&ioc, dat, sizeof(ioc)); } static int i_setsig(fp, td, retval, fd, cmd, dat) struct file *fp; struct thread *td; register_t *retval; int fd; u_long cmd; caddr_t dat; { /* * This is the best we can do for now; we cannot generate * signals only for specific events so the signal mask gets * ignored; we save it just to pass it to a possible I_GETSIG... * * We alse have to fix the O_ASYNC fcntl bit, so the * process will get SIGPOLLs. */ int error; register_t oflags, flags; struct svr4_strm *st = svr4_stream_get(fp); if (st == NULL) { DPRINTF(("i_setsig: bad file descriptor\n")); return EINVAL; } /* get old status flags */ error = kern_fcntl(td, fd, F_GETFL, 0); if (error) return (error); oflags = td->td_retval[0]; /* update the flags */ if (dat != NULL) { int mask; flags = oflags | O_ASYNC; if ((error = copyin(dat, &mask, sizeof(mask))) != 0) { DPRINTF(("i_setsig: bad eventmask pointer\n")); return error; } if (mask & SVR4_S_ALLMASK) { DPRINTF(("i_setsig: bad eventmask data %x\n", mask)); return EINVAL; } st->s_eventmask = mask; } else { flags = oflags & ~O_ASYNC; st->s_eventmask = 0; } /* set the new flags, if changed */ if (flags != oflags) { error = kern_fcntl(td, fd, F_SETFL, flags); if (error) return (error); flags = td->td_retval[0]; } /* set up SIGIO receiver if needed */ if (dat != NULL) return (kern_fcntl(td, fd, F_SETOWN, td->td_proc->p_pid)); return 0; } static int i_getsig(fp, td, retval, fd, cmd, dat) struct file *fp; struct thread *td; register_t *retval; int fd; u_long cmd; caddr_t dat; { int error; if (dat != NULL) { struct svr4_strm *st = svr4_stream_get(fp); if (st == NULL) { DPRINTF(("i_getsig: bad file descriptor\n")); return EINVAL; } if ((error = copyout(&st->s_eventmask, dat, sizeof(st->s_eventmask))) != 0) { DPRINTF(("i_getsig: bad eventmask pointer\n")); return error; } } return 0; } int svr4_stream_ioctl(fp, td, retval, fd, cmd, dat) struct file *fp; struct thread *td; register_t *retval; int fd; u_long cmd; caddr_t dat; { *retval = 0; /* * All the following stuff assumes "sockmod" is pushed... */ switch (cmd) { case SVR4_I_NREAD: DPRINTF(("I_NREAD\n")); return i_nread(fp, td, retval, fd, cmd, dat); case SVR4_I_PUSH: DPRINTF(("I_PUSH %p\n", dat)); #if defined(DEBUG_SVR4) show_strbuf((struct svr4_strbuf *)dat); #endif return 0; case SVR4_I_POP: DPRINTF(("I_POP\n")); return 0; case SVR4_I_LOOK: DPRINTF(("I_LOOK\n")); return 0; case SVR4_I_FLUSH: DPRINTF(("I_FLUSH\n")); return 0; case SVR4_I_SRDOPT: DPRINTF(("I_SRDOPT\n")); return 0; case SVR4_I_GRDOPT: DPRINTF(("I_GRDOPT\n")); return 0; case SVR4_I_STR: DPRINTF(("I_STR\n")); return i_str(fp, td, retval, fd, cmd, dat); case SVR4_I_SETSIG: DPRINTF(("I_SETSIG\n")); return i_setsig(fp, td, retval, fd, cmd, dat); case SVR4_I_GETSIG: DPRINTF(("I_GETSIG\n")); return i_getsig(fp, td, retval, fd, cmd, dat); case SVR4_I_FIND: DPRINTF(("I_FIND\n")); /* * Here we are not pushing modules really, we just * pretend all are present */ *retval = 0; return 0; case SVR4_I_LINK: DPRINTF(("I_LINK\n")); return 0; case SVR4_I_UNLINK: DPRINTF(("I_UNLINK\n")); return 0; case SVR4_I_ERECVFD: DPRINTF(("I_ERECVFD\n")); return 0; case SVR4_I_PEEK: DPRINTF(("I_PEEK\n")); return 0; case SVR4_I_FDINSERT: DPRINTF(("I_FDINSERT\n")); return i_fdinsert(fp, td, retval, fd, cmd, dat); case SVR4_I_SENDFD: DPRINTF(("I_SENDFD\n")); return 0; case SVR4_I_RECVFD: DPRINTF(("I_RECVFD\n")); return 0; case SVR4_I_SWROPT: DPRINTF(("I_SWROPT\n")); return 0; case SVR4_I_GWROPT: DPRINTF(("I_GWROPT\n")); return 0; case SVR4_I_LIST: DPRINTF(("I_LIST\n")); return 0; case SVR4_I_PLINK: DPRINTF(("I_PLINK\n")); return 0; case SVR4_I_PUNLINK: DPRINTF(("I_PUNLINK\n")); return 0; case SVR4_I_SETEV: DPRINTF(("I_SETEV\n")); return 0; case SVR4_I_GETEV: DPRINTF(("I_GETEV\n")); return 0; case SVR4_I_STREV: DPRINTF(("I_STREV\n")); return 0; case SVR4_I_UNSTREV: DPRINTF(("I_UNSTREV\n")); return 0; case SVR4_I_FLUSHBAND: DPRINTF(("I_FLUSHBAND\n")); return 0; case SVR4_I_CKBAND: DPRINTF(("I_CKBAND\n")); return 0; case SVR4_I_GETBAND: DPRINTF(("I_GETBANK\n")); return 0; case SVR4_I_ATMARK: DPRINTF(("I_ATMARK\n")); return 0; case SVR4_I_SETCLTIME: DPRINTF(("I_SETCLTIME\n")); return 0; case SVR4_I_GETCLTIME: DPRINTF(("I_GETCLTIME\n")); return 0; case SVR4_I_CANPUT: DPRINTF(("I_CANPUT\n")); return 0; case SVR4__I_BIND_RSVD: DPRINTF(("_I_BIND_RSVD\n")); return _i_bind_rsvd(fp, td, retval, fd, cmd, dat); case SVR4__I_RELE_RSVD: DPRINTF(("_I_RELE_RSVD\n")); return _i_rele_rsvd(fp, td, retval, fd, cmd, dat); default: DPRINTF(("unimpl cmd = %lx\n", cmd)); break; } return 0; } int svr4_sys_putmsg(td, uap) register struct thread *td; struct svr4_sys_putmsg_args *uap; { struct file *fp; int error; if ((error = fget(td, uap->fd, &fp)) != 0) { #ifdef DEBUG_SVR4 uprintf("putmsg: bad fp\n"); #endif return EBADF; } error = svr4_do_putmsg(td, uap, fp); fdrop(fp, td); return (error); } static int svr4_do_putmsg(td, uap, fp) struct thread *td; struct svr4_sys_putmsg_args *uap; struct file *fp; { struct svr4_strbuf dat, ctl; struct svr4_strmcmd sc; struct sockaddr_in sain; struct sockaddr_un saun; struct sockaddr *sa; int sasize, *retval; struct svr4_strm *st; int error; retval = td->td_retval; #ifdef DEBUG_SVR4 show_msg(">putmsg", uap->fd, uap->ctl, uap->dat, uap->flags); #endif /* DEBUG_SVR4 */ FILE_LOCK_ASSERT(fp, MA_NOTOWNED); if (uap->ctl != NULL) { if ((error = copyin(uap->ctl, &ctl, sizeof(ctl))) != 0) { #ifdef DEBUG_SVR4 uprintf("putmsg: copyin(): %d\n", error); #endif return error; } } else ctl.len = -1; if (uap->dat != NULL) { if ((error = copyin(uap->dat, &dat, sizeof(dat))) != 0) { #ifdef DEBUG_SVR4 uprintf("putmsg: copyin(): %d (2)\n", error); #endif return error; } } else dat.len = -1; /* * Only for sockets for now. */ if ((st = svr4_stream_get(fp)) == NULL) { DPRINTF(("putmsg: bad file type\n")); return EINVAL; } if (ctl.len < 0 || ctl.len > sizeof(sc)) { DPRINTF(("putmsg: Bad control size %d != %d\n", ctl.len, sizeof(struct svr4_strmcmd))); return EINVAL; } if ((error = copyin(ctl.buf, &sc, ctl.len)) != 0) return error; switch (st->s_family) { case AF_INET: if (sc.len != sizeof(sain)) { if (sc.cmd == SVR4_TI_DATA_REQUEST) { struct write_args wa; /* Solaris seems to use sc.cmd = 3 to * send "expedited" data. telnet uses * this for options processing, sending EOF, * etc. I'm sure other things use it too. * I don't have any documentation * on it, so I'm making a guess that this * is how it works. newton@atdot.dotat.org XXX */ DPRINTF(("sending expedited data ??\n")); wa.fd = uap->fd; wa.buf = dat.buf; wa.nbyte = dat.len; return write(td, &wa); } DPRINTF(("putmsg: Invalid inet length %ld\n", sc.len)); return EINVAL; } netaddr_to_sockaddr_in(&sain, &sc); sa = (struct sockaddr *)&sain; sasize = sizeof(sain); if (sain.sin_family != st->s_family) error = EINVAL; break; case AF_LOCAL: if (ctl.len == 8) { /* We are doing an accept; succeed */ DPRINTF(("putmsg: Do nothing\n")); *retval = 0; return 0; } else { /* Maybe we've been given a device/inode pair */ dev_t *dev = SVR4_ADDROF(&sc); ino_t *ino = (ino_t *) &dev[1]; if (svr4_find_socket(td, fp, *dev, *ino, &saun) != 0) { /* I guess we have it by name */ netaddr_to_sockaddr_un(&saun, &sc); } sa = (struct sockaddr *)&saun; sasize = sizeof(saun); } break; default: DPRINTF(("putmsg: Unsupported address family %d\n", st->s_family)); return ENOSYS; } switch (st->s_cmd = sc.cmd) { case SVR4_TI_CONNECT_REQUEST: /* connect */ { return (kern_connect(td, uap->fd, sa)); } case SVR4_TI_SENDTO_REQUEST: /* sendto */ { struct msghdr msg; struct iovec aiov; msg.msg_name = sa; msg.msg_namelen = sasize; msg.msg_iov = &aiov; msg.msg_iovlen = 1; msg.msg_control = 0; msg.msg_flags = 0; aiov.iov_base = dat.buf; aiov.iov_len = dat.len; error = kern_sendit(td, uap->fd, &msg, uap->flags, NULL, UIO_USERSPACE); DPRINTF(("sendto_request error: %d\n", error)); *retval = 0; return error; } default: DPRINTF(("putmsg: Unimplemented command %lx\n", sc.cmd)); return ENOSYS; } } int svr4_sys_getmsg(td, uap) struct thread *td; struct svr4_sys_getmsg_args *uap; { struct file *fp; int error; if ((error = fget(td, uap->fd, &fp)) != 0) { #ifdef DEBUG_SVR4 uprintf("getmsg: bad fp\n"); #endif return EBADF; } error = svr4_do_getmsg(td, uap, fp); fdrop(fp, td); return (error); } int svr4_do_getmsg(td, uap, fp) register struct thread *td; struct svr4_sys_getmsg_args *uap; struct file *fp; { struct svr4_strbuf dat, ctl; struct svr4_strmcmd sc; int error, *retval; struct msghdr msg; struct iovec aiov; struct sockaddr_in sain; struct sockaddr_un saun; struct sockaddr *sa; socklen_t sasize; struct svr4_strm *st; + struct file *afp; int fl; retval = td->td_retval; error = 0; + afp = NULL; FILE_LOCK_ASSERT(fp, MA_NOTOWNED); memset(&sc, 0, sizeof(sc)); #ifdef DEBUG_SVR4 show_msg(">getmsg", uap->fd, uap->ctl, uap->dat, 0); #endif /* DEBUG_SVR4 */ if (uap->ctl != NULL) { if ((error = copyin(uap->ctl, &ctl, sizeof(ctl))) != 0) return error; if (ctl.len < 0) return EINVAL; } else { ctl.len = -1; ctl.maxlen = 0; } if (uap->dat != NULL) { if ((error = copyin(uap->dat, &dat, sizeof(dat))) != 0) return error; } else { dat.len = -1; dat.maxlen = 0; } /* * Only for sockets for now. */ if ((st = svr4_stream_get(fp)) == NULL) { DPRINTF(("getmsg: bad file type\n")); return EINVAL; } if (ctl.maxlen == -1 || dat.maxlen == -1) { DPRINTF(("getmsg: Cannot handle -1 maxlen (yet)\n")); return ENOSYS; } switch (st->s_family) { case AF_INET: sasize = sizeof(sain); break; case AF_LOCAL: sasize = sizeof(saun); break; default: DPRINTF(("getmsg: Unsupported address family %d\n", st->s_family)); return ENOSYS; } switch (st->s_cmd) { case SVR4_TI_CONNECT_REQUEST: DPRINTF(("getmsg: TI_CONNECT_REQUEST\n")); /* * We do the connect in one step, so the putmsg should * have gotten the error. */ sc.cmd = SVR4_TI_OK_REPLY; sc.len = 0; ctl.len = 8; dat.len = -1; fl = 1; st->s_cmd = sc.cmd; break; case SVR4_TI_OK_REPLY: DPRINTF(("getmsg: TI_OK_REPLY\n")); /* * We are immediately after a connect reply, so we send * a connect verification. */ error = kern_getpeername(td, uap->fd, &sa, &sasize); if (error) { DPRINTF(("getmsg: getpeername failed %d\n", error)); return error; } sc.cmd = SVR4_TI_CONNECT_REPLY; sc.pad[0] = 0x4; sc.offs = 0x18; sc.pad[1] = 0x14; sc.pad[2] = 0x04000402; switch (st->s_family) { case AF_INET: sc.len = sasize; sockaddr_to_netaddr_in(&sc, (struct sockaddr_in *)sa); break; case AF_LOCAL: sc.len = sasize + 4; sockaddr_to_netaddr_un(&sc, (struct sockaddr_un *)sa); break; default: free(sa, M_SONAME); return ENOSYS; } free(sa, M_SONAME); ctl.len = 40; dat.len = -1; fl = 0; st->s_cmd = sc.cmd; break; case SVR4_TI__ACCEPT_OK: DPRINTF(("getmsg: TI__ACCEPT_OK\n")); /* * We do the connect in one step, so the putmsg should * have gotten the error. */ sc.cmd = SVR4_TI_OK_REPLY; sc.len = 1; ctl.len = 8; dat.len = -1; fl = 1; st->s_cmd = SVR4_TI__ACCEPT_WAIT; break; case SVR4_TI__ACCEPT_WAIT: DPRINTF(("getmsg: TI__ACCEPT_WAIT\n")); /* * We are after a listen, so we try to accept... */ - error = kern_accept(td, uap->fd, &sa, &sasize); + error = kern_accept(td, uap->fd, &sa, &sasize, &afp); if (error) { DPRINTF(("getmsg: accept failed %d\n", error)); return error; } st->s_afd = *retval; DPRINTF(("getmsg: Accept fd = %d\n", st->s_afd)); sc.cmd = SVR4_TI_ACCEPT_REPLY; sc.offs = 0x18; sc.pad[0] = 0x0; switch (st->s_family) { case AF_INET: sc.pad[1] = 0x28; sockaddr_to_netaddr_in(&sc, (struct sockaddr_in *)&sa); ctl.len = 40; sc.len = sasize; break; case AF_LOCAL: sc.pad[1] = 0x00010000; sc.pad[2] = 0xf6bcdaa0; /* I don't know what that is */ sc.pad[3] = 0x00010000; ctl.len = 134; sc.len = sasize + 4; break; default: + fdclose(td->td_proc->p_fd, afp, st->s_afd, td); + fdrop(afp, td); + st->s_afd = -1; free(sa, M_SONAME); return ENOSYS; } free(sa, M_SONAME); dat.len = -1; fl = 0; st->s_cmd = SVR4_TI__ACCEPT_OK; break; case SVR4_TI_SENDTO_REQUEST: DPRINTF(("getmsg: TI_SENDTO_REQUEST\n")); if (ctl.maxlen > 36 && ctl.len < 36) ctl.len = 36; if (ctl.len > sizeof(sc)) ctl.len = sizeof(sc); if ((error = copyin(ctl.buf, &sc, ctl.len)) != 0) return error; switch (st->s_family) { case AF_INET: sa = (struct sockaddr *)&sain; sockaddr_to_netaddr_in(&sc, &sain); break; case AF_LOCAL: sa = (struct sockaddr *)&saun; sockaddr_to_netaddr_un(&sc, &saun); break; default: return ENOSYS; } msg.msg_name = sa; msg.msg_namelen = sasize; msg.msg_iov = &aiov; msg.msg_iovlen = 1; msg.msg_control = 0; aiov.iov_base = dat.buf; aiov.iov_len = dat.maxlen; msg.msg_flags = 0; error = kern_recvit(td, uap->fd, &msg, UIO_SYSSPACE, NULL); if (error) { DPRINTF(("getmsg: recvit failed %d\n", error)); return error; } sc.cmd = SVR4_TI_RECVFROM_IND; switch (st->s_family) { case AF_INET: sc.len = sasize; sockaddr_to_netaddr_in(&sc, &sain); break; case AF_LOCAL: sc.len = sasize + 4; sockaddr_to_netaddr_un(&sc, &saun); break; default: return ENOSYS; } dat.len = *retval; fl = 0; st->s_cmd = sc.cmd; break; default: st->s_cmd = sc.cmd; if (st->s_cmd == SVR4_TI_CONNECT_REQUEST) { struct read_args ra; /* More weirdness: Again, I can't find documentation * to back this up, but when a process does a generic * "getmsg()" call it seems that the command field is * zero and the length of the data area is zero. I * think processes expect getmsg() to fill in dat.len * after reading at most dat.maxlen octets from the * stream. Since we're using sockets I can let * read() look after it and frob return values * appropriately (or inappropriately :-) * -- newton@atdot.dotat.org XXX */ ra.fd = uap->fd; ra.buf = dat.buf; ra.nbyte = dat.maxlen; if ((error = read(td, &ra)) != 0) { return error; } dat.len = *retval; *retval = 0; st->s_cmd = SVR4_TI_SENDTO_REQUEST; break; } DPRINTF(("getmsg: Unknown state %x\n", st->s_cmd)); return EINVAL; } - /* XXX: We leak the accept fd if we get an error here. */ if (uap->ctl) { if (ctl.len > sizeof(sc)) ctl.len = sizeof(sc); if (ctl.len != -1) - if ((error = copyout(&sc, ctl.buf, ctl.len)) != 0) - return error; + error = copyout(&sc, ctl.buf, ctl.len); - if ((error = copyout(&ctl, uap->ctl, sizeof(ctl))) != 0) - return error; + if (error == 0) + error = copyout(&ctl, uap->ctl, sizeof(ctl)); } if (uap->dat) { - if ((error = copyout(&dat, uap->dat, sizeof(dat))) != 0) - return error; + if (error == 0) + error = copyout(&dat, uap->dat, sizeof(dat)); } if (uap->flags) { /* XXX: Need translation */ - if ((error = copyout(&fl, uap->flags, sizeof(fl))) != 0) - return error; + if (error == 0) + error = copyout(&fl, uap->flags, sizeof(fl)); } + + if (error) { + if (afp) { + fdclose(td->td_proc->p_fd, afp, st->s_afd, td); + fdrop(afp, td); + st->s_afd = -1; + } + return (error); + } + if (afp) + fdrop(afp, td); *retval = 0; #ifdef DEBUG_SVR4 show_msg("fd, uap->ctl, uap->dat, fl); #endif /* DEBUG_SVR4 */ return error; } int svr4_sys_send(td, uap) struct thread *td; struct svr4_sys_send_args *uap; { struct osend_args osa; osa.s = uap->s; osa.buf = uap->buf; osa.len = uap->len; osa.flags = uap->flags; return osend(td, &osa); } int svr4_sys_recv(td, uap) struct thread *td; struct svr4_sys_recv_args *uap; { struct orecv_args ora; ora.s = uap->s; ora.buf = uap->buf; ora.len = uap->len; ora.flags = uap->flags; return orecv(td, &ora); } /* * XXX This isn't necessary, but it's handy for inserting debug code into * sendto(). Let's leave it here for now... */ int svr4_sys_sendto(td, uap) struct thread *td; struct svr4_sys_sendto_args *uap; { struct sendto_args sa; sa.s = uap->s; sa.buf = uap->buf; sa.len = uap->len; sa.flags = uap->flags; sa.to = (caddr_t)uap->to; sa.tolen = uap->tolen; DPRINTF(("calling sendto()\n")); return sendto(td, &sa); } Index: head/sys/kern/uipc_syscalls.c =================================================================== --- head/sys/kern/uipc_syscalls.c (revision 160764) +++ head/sys/kern/uipc_syscalls.c (revision 160765) @@ -1,2276 +1,2286 @@ /*- * Copyright (c) 1982, 1986, 1989, 1990, 1993 * The Regents of the University of California. All rights reserved. * * sendfile(2) and related extensions: * Copyright (c) 1998, David Greenman. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)uipc_syscalls.c 8.4 (Berkeley) 2/21/94 */ #include __FBSDID("$FreeBSD$"); #include "opt_compat.h" #include "opt_ktrace.h" #include "opt_mac.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef KTRACE #include #endif #include #include #include #include #include #include static int sendit(struct thread *td, int s, struct msghdr *mp, int flags); static int recvit(struct thread *td, int s, struct msghdr *mp, void *namelenp); static int accept1(struct thread *td, struct accept_args *uap, int compat); static int do_sendfile(struct thread *td, struct sendfile_args *uap, int compat); static int getsockname1(struct thread *td, struct getsockname_args *uap, int compat); static int getpeername1(struct thread *td, struct getpeername_args *uap, int compat); /* * NSFBUFS-related variables and associated sysctls */ int nsfbufs; int nsfbufspeak; int nsfbufsused; SYSCTL_INT(_kern_ipc, OID_AUTO, nsfbufs, CTLFLAG_RDTUN, &nsfbufs, 0, "Maximum number of sendfile(2) sf_bufs available"); SYSCTL_INT(_kern_ipc, OID_AUTO, nsfbufspeak, CTLFLAG_RD, &nsfbufspeak, 0, "Number of sendfile(2) sf_bufs at peak usage"); SYSCTL_INT(_kern_ipc, OID_AUTO, nsfbufsused, CTLFLAG_RD, &nsfbufsused, 0, "Number of sendfile(2) sf_bufs in use"); /* * Convert a user file descriptor to a kernel file entry. A reference on the * file entry is held upon returning. This is lighter weight than * fgetsock(), which bumps the socket reference drops the file reference * count instead, as this approach avoids several additional mutex operations * associated with the additional reference count. If requested, return the * open file flags. */ static int getsock(struct filedesc *fdp, int fd, struct file **fpp, u_int *fflagp) { struct file *fp; int error; fp = NULL; if (fdp == NULL) error = EBADF; else { FILEDESC_LOCK_FAST(fdp); fp = fget_locked(fdp, fd); if (fp == NULL) error = EBADF; else if (fp->f_type != DTYPE_SOCKET) { fp = NULL; error = ENOTSOCK; } else { fhold(fp); if (fflagp != NULL) *fflagp = fp->f_flag; error = 0; } FILEDESC_UNLOCK_FAST(fdp); } *fpp = fp; return (error); } /* * System call interface to the socket abstraction. */ #if defined(COMPAT_43) #define COMPAT_OLDSOCK #endif /* * MPSAFE */ int socket(td, uap) struct thread *td; register struct socket_args /* { int domain; int type; int protocol; } */ *uap; { struct filedesc *fdp; struct socket *so; struct file *fp; int fd, error; #ifdef MAC error = mac_check_socket_create(td->td_ucred, uap->domain, uap->type, uap->protocol); if (error) return (error); #endif fdp = td->td_proc->p_fd; error = falloc(td, &fp, &fd); if (error) return (error); /* An extra reference on `fp' has been held for us by falloc(). */ NET_LOCK_GIANT(); error = socreate(uap->domain, &so, uap->type, uap->protocol, td->td_ucred, td); NET_UNLOCK_GIANT(); if (error) { fdclose(fdp, fp, fd, td); } else { FILEDESC_LOCK_FAST(fdp); fp->f_data = so; /* already has ref count */ fp->f_flag = FREAD|FWRITE; fp->f_ops = &socketops; fp->f_type = DTYPE_SOCKET; FILEDESC_UNLOCK_FAST(fdp); td->td_retval[0] = fd; } fdrop(fp, td); return (error); } /* * MPSAFE */ /* ARGSUSED */ int bind(td, uap) struct thread *td; register struct bind_args /* { int s; caddr_t name; int namelen; } */ *uap; { struct sockaddr *sa; int error; if ((error = getsockaddr(&sa, uap->name, uap->namelen)) != 0) return (error); error = kern_bind(td, uap->s, sa); free(sa, M_SONAME); return (error); } int kern_bind(td, fd, sa) struct thread *td; int fd; struct sockaddr *sa; { struct socket *so; struct file *fp; int error; NET_LOCK_GIANT(); error = getsock(td->td_proc->p_fd, fd, &fp, NULL); if (error) goto done2; so = fp->f_data; #ifdef MAC SOCK_LOCK(so); error = mac_check_socket_bind(td->td_ucred, so, sa); SOCK_UNLOCK(so); if (error) goto done1; #endif error = sobind(so, sa, td); #ifdef MAC done1: #endif fdrop(fp, td); done2: NET_UNLOCK_GIANT(); return (error); } /* * MPSAFE */ /* ARGSUSED */ int listen(td, uap) struct thread *td; register struct listen_args /* { int s; int backlog; } */ *uap; { struct socket *so; struct file *fp; int error; NET_LOCK_GIANT(); error = getsock(td->td_proc->p_fd, uap->s, &fp, NULL); if (error == 0) { so = fp->f_data; #ifdef MAC SOCK_LOCK(so); error = mac_check_socket_listen(td->td_ucred, so); SOCK_UNLOCK(so); if (error) goto done; #endif error = solisten(so, uap->backlog, td); #ifdef MAC done: #endif fdrop(fp, td); } NET_UNLOCK_GIANT(); return(error); } /* * accept1() * MPSAFE */ static int accept1(td, uap, compat) struct thread *td; register struct accept_args /* { int s; struct sockaddr * __restrict name; socklen_t * __restrict anamelen; } */ *uap; int compat; { struct sockaddr *name; socklen_t namelen; + struct file *fp; int error; if (uap->name == NULL) - return (kern_accept(td, uap->s, NULL, NULL)); + return (kern_accept(td, uap->s, NULL, NULL, NULL)); error = copyin(uap->anamelen, &namelen, sizeof (namelen)); if (error) return (error); - error = kern_accept(td, uap->s, &name, &namelen); + error = kern_accept(td, uap->s, &name, &namelen, &fp); /* * return a namelen of zero for older code which might * ignore the return value from accept. */ if (error) { (void) copyout(&namelen, uap->anamelen, sizeof(*uap->anamelen)); return (error); } if (error == 0 && name != NULL) { #ifdef COMPAT_OLDSOCK if (compat) ((struct osockaddr *)name)->sa_family = name->sa_family; #endif error = copyout(name, uap->name, namelen); } if (error == 0) error = copyout(&namelen, uap->anamelen, sizeof(namelen)); if (error) - kern_close(td, td->td_retval[0]); + fdclose(td->td_proc->p_fd, fp, td->td_retval[0], td); + fdrop(fp, td); free(name, M_SONAME); return (error); } int kern_accept(struct thread *td, int s, struct sockaddr **name, - socklen_t *namelen) + socklen_t *namelen, struct file **fp) { struct filedesc *fdp; struct file *headfp, *nfp = NULL; struct sockaddr *sa = NULL; int error; struct socket *head, *so; int fd; u_int fflag; pid_t pgid; int tmp; if (name) { *name = NULL; if (*namelen < 0) return (EINVAL); } fdp = td->td_proc->p_fd; NET_LOCK_GIANT(); error = getsock(fdp, s, &headfp, &fflag); if (error) goto done2; head = headfp->f_data; if ((head->so_options & SO_ACCEPTCONN) == 0) { error = EINVAL; goto done; } #ifdef MAC SOCK_LOCK(head); error = mac_check_socket_accept(td->td_ucred, head); SOCK_UNLOCK(head); if (error != 0) goto done; #endif error = falloc(td, &nfp, &fd); if (error) goto done; ACCEPT_LOCK(); if ((head->so_state & SS_NBIO) && TAILQ_EMPTY(&head->so_comp)) { ACCEPT_UNLOCK(); error = EWOULDBLOCK; goto noconnection; } while (TAILQ_EMPTY(&head->so_comp) && head->so_error == 0) { if (head->so_rcv.sb_state & SBS_CANTRCVMORE) { head->so_error = ECONNABORTED; break; } error = msleep(&head->so_timeo, &accept_mtx, PSOCK | PCATCH, "accept", 0); if (error) { ACCEPT_UNLOCK(); goto noconnection; } } if (head->so_error) { error = head->so_error; head->so_error = 0; ACCEPT_UNLOCK(); goto noconnection; } so = TAILQ_FIRST(&head->so_comp); KASSERT(!(so->so_qstate & SQ_INCOMP), ("accept1: so SQ_INCOMP")); KASSERT(so->so_qstate & SQ_COMP, ("accept1: so not SQ_COMP")); /* * Before changing the flags on the socket, we have to bump the * reference count. Otherwise, if the protocol calls sofree(), * the socket will be released due to a zero refcount. */ SOCK_LOCK(so); /* soref() and so_state update */ soref(so); /* file descriptor reference */ TAILQ_REMOVE(&head->so_comp, so, so_list); head->so_qlen--; so->so_state |= (head->so_state & SS_NBIO); so->so_qstate &= ~SQ_COMP; so->so_head = NULL; SOCK_UNLOCK(so); ACCEPT_UNLOCK(); /* An extra reference on `nfp' has been held for us by falloc(). */ td->td_retval[0] = fd; /* connection has been removed from the listen queue */ KNOTE_UNLOCKED(&head->so_rcv.sb_sel.si_note, 0); pgid = fgetown(&head->so_sigio); if (pgid != 0) fsetown(pgid, &so->so_sigio); FILE_LOCK(nfp); nfp->f_data = so; /* nfp has ref count from falloc */ nfp->f_flag = fflag; nfp->f_ops = &socketops; nfp->f_type = DTYPE_SOCKET; FILE_UNLOCK(nfp); /* Sync socket nonblocking/async state with file flags */ tmp = fflag & FNONBLOCK; (void) fo_ioctl(nfp, FIONBIO, &tmp, td->td_ucred, td); tmp = fflag & FASYNC; (void) fo_ioctl(nfp, FIOASYNC, &tmp, td->td_ucred, td); sa = 0; error = soaccept(so, &sa); if (error) { /* * return a namelen of zero for older code which might * ignore the return value from accept. */ if (name) *namelen = 0; goto noconnection; } if (sa == NULL) { if (name) *namelen = 0; goto done; } if (name) { /* check sa_len before it is destroyed */ if (*namelen > sa->sa_len) *namelen = sa->sa_len; *name = sa; sa = NULL; } noconnection: if (sa) FREE(sa, M_SONAME); /* * close the new descriptor, assuming someone hasn't ripped it * out from under us. */ if (error) fdclose(fdp, nfp, fd, td); /* - * Release explicitly held references before returning. + * Release explicitly held references before returning. We return + * a reference on nfp to the caller on success if they request it. */ done: + if (fp != NULL) { + if (error == 0) { + *fp = nfp; + nfp = NULL; + } else + *fp = NULL; + } if (nfp != NULL) fdrop(nfp, td); fdrop(headfp, td); done2: NET_UNLOCK_GIANT(); return (error); } /* * MPSAFE (accept1() is MPSAFE) */ int accept(td, uap) struct thread *td; struct accept_args *uap; { return (accept1(td, uap, 0)); } #ifdef COMPAT_OLDSOCK /* * MPSAFE (accept1() is MPSAFE) */ int oaccept(td, uap) struct thread *td; struct accept_args *uap; { return (accept1(td, uap, 1)); } #endif /* COMPAT_OLDSOCK */ /* * MPSAFE */ /* ARGSUSED */ int connect(td, uap) struct thread *td; register struct connect_args /* { int s; caddr_t name; int namelen; } */ *uap; { struct sockaddr *sa; int error; error = getsockaddr(&sa, uap->name, uap->namelen); if (error) return (error); error = kern_connect(td, uap->s, sa); free(sa, M_SONAME); return (error); } int kern_connect(td, fd, sa) struct thread *td; int fd; struct sockaddr *sa; { struct socket *so; struct file *fp; int error; int interrupted = 0; NET_LOCK_GIANT(); error = getsock(td->td_proc->p_fd, fd, &fp, NULL); if (error) goto done2; so = fp->f_data; if (so->so_state & SS_ISCONNECTING) { error = EALREADY; goto done1; } #ifdef MAC SOCK_LOCK(so); error = mac_check_socket_connect(td->td_ucred, so, sa); SOCK_UNLOCK(so); if (error) goto bad; #endif error = soconnect(so, sa, td); if (error) goto bad; if ((so->so_state & SS_NBIO) && (so->so_state & SS_ISCONNECTING)) { error = EINPROGRESS; goto done1; } SOCK_LOCK(so); while ((so->so_state & SS_ISCONNECTING) && so->so_error == 0) { error = msleep(&so->so_timeo, SOCK_MTX(so), PSOCK | PCATCH, "connec", 0); if (error) { if (error == EINTR || error == ERESTART) interrupted = 1; break; } } if (error == 0) { error = so->so_error; so->so_error = 0; } SOCK_UNLOCK(so); bad: if (!interrupted) so->so_state &= ~SS_ISCONNECTING; if (error == ERESTART) error = EINTR; done1: fdrop(fp, td); done2: NET_UNLOCK_GIANT(); return (error); } /* * MPSAFE */ int socketpair(td, uap) struct thread *td; register struct socketpair_args /* { int domain; int type; int protocol; int *rsv; } */ *uap; { register struct filedesc *fdp = td->td_proc->p_fd; struct file *fp1, *fp2; struct socket *so1, *so2; int fd, error, sv[2]; #ifdef MAC /* We might want to have a separate check for socket pairs. */ error = mac_check_socket_create(td->td_ucred, uap->domain, uap->type, uap->protocol); if (error) return (error); #endif NET_LOCK_GIANT(); error = socreate(uap->domain, &so1, uap->type, uap->protocol, td->td_ucred, td); if (error) goto done2; error = socreate(uap->domain, &so2, uap->type, uap->protocol, td->td_ucred, td); if (error) goto free1; /* On success extra reference to `fp1' and 'fp2' is set by falloc. */ error = falloc(td, &fp1, &fd); if (error) goto free2; sv[0] = fd; fp1->f_data = so1; /* so1 already has ref count */ error = falloc(td, &fp2, &fd); if (error) goto free3; fp2->f_data = so2; /* so2 already has ref count */ sv[1] = fd; error = soconnect2(so1, so2); if (error) goto free4; if (uap->type == SOCK_DGRAM) { /* * Datagram socket connection is asymmetric. */ error = soconnect2(so2, so1); if (error) goto free4; } FILE_LOCK(fp1); fp1->f_flag = FREAD|FWRITE; fp1->f_ops = &socketops; fp1->f_type = DTYPE_SOCKET; FILE_UNLOCK(fp1); FILE_LOCK(fp2); fp2->f_flag = FREAD|FWRITE; fp2->f_ops = &socketops; fp2->f_type = DTYPE_SOCKET; FILE_UNLOCK(fp2); error = copyout(sv, uap->rsv, 2 * sizeof (int)); fdrop(fp1, td); fdrop(fp2, td); goto done2; free4: fdclose(fdp, fp2, sv[1], td); fdrop(fp2, td); free3: fdclose(fdp, fp1, sv[0], td); fdrop(fp1, td); free2: (void)soclose(so2); free1: (void)soclose(so1); done2: NET_UNLOCK_GIANT(); return (error); } static int sendit(td, s, mp, flags) register struct thread *td; int s; register struct msghdr *mp; int flags; { struct mbuf *control; struct sockaddr *to; int error; if (mp->msg_name != NULL) { error = getsockaddr(&to, mp->msg_name, mp->msg_namelen); if (error) { to = NULL; goto bad; } mp->msg_name = to; } else { to = NULL; } if (mp->msg_control) { if (mp->msg_controllen < sizeof(struct cmsghdr) #ifdef COMPAT_OLDSOCK && mp->msg_flags != MSG_COMPAT #endif ) { error = EINVAL; goto bad; } error = sockargs(&control, mp->msg_control, mp->msg_controllen, MT_CONTROL); if (error) goto bad; #ifdef COMPAT_OLDSOCK if (mp->msg_flags == MSG_COMPAT) { register struct cmsghdr *cm; M_PREPEND(control, sizeof(*cm), M_TRYWAIT); if (control == 0) { error = ENOBUFS; goto bad; } else { cm = mtod(control, struct cmsghdr *); cm->cmsg_len = control->m_len; cm->cmsg_level = SOL_SOCKET; cm->cmsg_type = SCM_RIGHTS; } } #endif } else { control = NULL; } error = kern_sendit(td, s, mp, flags, control, UIO_USERSPACE); bad: if (to) FREE(to, M_SONAME); return (error); } int kern_sendit(td, s, mp, flags, control, segflg) struct thread *td; int s; struct msghdr *mp; int flags; struct mbuf *control; enum uio_seg segflg; { struct file *fp; struct uio auio; struct iovec *iov; struct socket *so; int i; int len, error; #ifdef KTRACE struct uio *ktruio = NULL; #endif NET_LOCK_GIANT(); error = getsock(td->td_proc->p_fd, s, &fp, NULL); if (error) goto bad2; so = (struct socket *)fp->f_data; #ifdef MAC SOCK_LOCK(so); error = mac_check_socket_send(td->td_ucred, so); SOCK_UNLOCK(so); if (error) goto bad; #endif auio.uio_iov = mp->msg_iov; auio.uio_iovcnt = mp->msg_iovlen; auio.uio_segflg = segflg; auio.uio_rw = UIO_WRITE; auio.uio_td = td; auio.uio_offset = 0; /* XXX */ auio.uio_resid = 0; iov = mp->msg_iov; for (i = 0; i < mp->msg_iovlen; i++, iov++) { if ((auio.uio_resid += iov->iov_len) < 0) { error = EINVAL; goto bad; } } #ifdef KTRACE if (KTRPOINT(td, KTR_GENIO)) ktruio = cloneuio(&auio); #endif len = auio.uio_resid; error = sosend(so, mp->msg_name, &auio, 0, control, flags, td); if (error) { if (auio.uio_resid != len && (error == ERESTART || error == EINTR || error == EWOULDBLOCK)) error = 0; /* Generation of SIGPIPE can be controlled per socket */ if (error == EPIPE && !(so->so_options & SO_NOSIGPIPE) && !(flags & MSG_NOSIGNAL)) { PROC_LOCK(td->td_proc); psignal(td->td_proc, SIGPIPE); PROC_UNLOCK(td->td_proc); } } if (error == 0) td->td_retval[0] = len - auio.uio_resid; #ifdef KTRACE if (ktruio != NULL) { ktruio->uio_resid = td->td_retval[0]; ktrgenio(s, UIO_WRITE, ktruio, error); } #endif bad: fdrop(fp, td); bad2: NET_UNLOCK_GIANT(); return (error); } /* * MPSAFE */ int sendto(td, uap) struct thread *td; register struct sendto_args /* { int s; caddr_t buf; size_t len; int flags; caddr_t to; int tolen; } */ *uap; { struct msghdr msg; struct iovec aiov; int error; msg.msg_name = uap->to; msg.msg_namelen = uap->tolen; msg.msg_iov = &aiov; msg.msg_iovlen = 1; msg.msg_control = 0; #ifdef COMPAT_OLDSOCK msg.msg_flags = 0; #endif aiov.iov_base = uap->buf; aiov.iov_len = uap->len; error = sendit(td, uap->s, &msg, uap->flags); return (error); } #ifdef COMPAT_OLDSOCK /* * MPSAFE */ int osend(td, uap) struct thread *td; register struct osend_args /* { int s; caddr_t buf; int len; int flags; } */ *uap; { struct msghdr msg; struct iovec aiov; int error; msg.msg_name = 0; msg.msg_namelen = 0; msg.msg_iov = &aiov; msg.msg_iovlen = 1; aiov.iov_base = uap->buf; aiov.iov_len = uap->len; msg.msg_control = 0; msg.msg_flags = 0; error = sendit(td, uap->s, &msg, uap->flags); return (error); } /* * MPSAFE */ int osendmsg(td, uap) struct thread *td; struct osendmsg_args /* { int s; caddr_t msg; int flags; } */ *uap; { struct msghdr msg; struct iovec *iov; int error; error = copyin(uap->msg, &msg, sizeof (struct omsghdr)); if (error) return (error); error = copyiniov(msg.msg_iov, msg.msg_iovlen, &iov, EMSGSIZE); if (error) return (error); msg.msg_iov = iov; msg.msg_flags = MSG_COMPAT; error = sendit(td, uap->s, &msg, uap->flags); free(iov, M_IOV); return (error); } #endif /* * MPSAFE */ int sendmsg(td, uap) struct thread *td; struct sendmsg_args /* { int s; caddr_t msg; int flags; } */ *uap; { struct msghdr msg; struct iovec *iov; int error; error = copyin(uap->msg, &msg, sizeof (msg)); if (error) return (error); error = copyiniov(msg.msg_iov, msg.msg_iovlen, &iov, EMSGSIZE); if (error) return (error); msg.msg_iov = iov; #ifdef COMPAT_OLDSOCK msg.msg_flags = 0; #endif error = sendit(td, uap->s, &msg, uap->flags); free(iov, M_IOV); return (error); } int kern_recvit(td, s, mp, fromseg, controlp) struct thread *td; int s; struct msghdr *mp; enum uio_seg fromseg; struct mbuf **controlp; { struct uio auio; struct iovec *iov; int i; socklen_t len; int error; struct mbuf *m, *control = 0; caddr_t ctlbuf; struct file *fp; struct socket *so; struct sockaddr *fromsa = 0; #ifdef KTRACE struct uio *ktruio = NULL; #endif if(controlp != NULL) *controlp = 0; NET_LOCK_GIANT(); error = getsock(td->td_proc->p_fd, s, &fp, NULL); if (error) { NET_UNLOCK_GIANT(); return (error); } so = fp->f_data; #ifdef MAC SOCK_LOCK(so); error = mac_check_socket_receive(td->td_ucred, so); SOCK_UNLOCK(so); if (error) { fdrop(fp, td); NET_UNLOCK_GIANT(); return (error); } #endif auio.uio_iov = mp->msg_iov; auio.uio_iovcnt = mp->msg_iovlen; auio.uio_segflg = UIO_USERSPACE; auio.uio_rw = UIO_READ; auio.uio_td = td; auio.uio_offset = 0; /* XXX */ auio.uio_resid = 0; iov = mp->msg_iov; for (i = 0; i < mp->msg_iovlen; i++, iov++) { if ((auio.uio_resid += iov->iov_len) < 0) { fdrop(fp, td); NET_UNLOCK_GIANT(); return (EINVAL); } } #ifdef KTRACE if (KTRPOINT(td, KTR_GENIO)) ktruio = cloneuio(&auio); #endif len = auio.uio_resid; error = soreceive(so, &fromsa, &auio, (struct mbuf **)0, (mp->msg_control || controlp) ? &control : (struct mbuf **)0, &mp->msg_flags); if (error) { if (auio.uio_resid != (int)len && (error == ERESTART || error == EINTR || error == EWOULDBLOCK)) error = 0; } #ifdef KTRACE if (ktruio != NULL) { ktruio->uio_resid = (int)len - auio.uio_resid; ktrgenio(s, UIO_READ, ktruio, error); } #endif if (error) goto out; td->td_retval[0] = (int)len - auio.uio_resid; if (mp->msg_name) { len = mp->msg_namelen; if (len <= 0 || fromsa == 0) len = 0; else { /* save sa_len before it is destroyed by MSG_COMPAT */ len = MIN(len, fromsa->sa_len); #ifdef COMPAT_OLDSOCK if (mp->msg_flags & MSG_COMPAT) ((struct osockaddr *)fromsa)->sa_family = fromsa->sa_family; #endif if (fromseg == UIO_USERSPACE) { error = copyout(fromsa, mp->msg_name, (unsigned)len); if (error) goto out; } else bcopy(fromsa, mp->msg_name, len); } mp->msg_namelen = len; } if (mp->msg_control && controlp == NULL) { #ifdef COMPAT_OLDSOCK /* * We assume that old recvmsg calls won't receive access * rights and other control info, esp. as control info * is always optional and those options didn't exist in 4.3. * If we receive rights, trim the cmsghdr; anything else * is tossed. */ if (control && mp->msg_flags & MSG_COMPAT) { if (mtod(control, struct cmsghdr *)->cmsg_level != SOL_SOCKET || mtod(control, struct cmsghdr *)->cmsg_type != SCM_RIGHTS) { mp->msg_controllen = 0; goto out; } control->m_len -= sizeof (struct cmsghdr); control->m_data += sizeof (struct cmsghdr); } #endif len = mp->msg_controllen; m = control; mp->msg_controllen = 0; ctlbuf = mp->msg_control; while (m && len > 0) { unsigned int tocopy; if (len >= m->m_len) tocopy = m->m_len; else { mp->msg_flags |= MSG_CTRUNC; tocopy = len; } if ((error = copyout(mtod(m, caddr_t), ctlbuf, tocopy)) != 0) goto out; ctlbuf += tocopy; len -= tocopy; m = m->m_next; } mp->msg_controllen = ctlbuf - (caddr_t)mp->msg_control; } out: fdrop(fp, td); NET_UNLOCK_GIANT(); if (fromsa) FREE(fromsa, M_SONAME); if (error == 0 && controlp != NULL) *controlp = control; else if (control) m_freem(control); return (error); } static int recvit(td, s, mp, namelenp) struct thread *td; int s; struct msghdr *mp; void *namelenp; { int error; error = kern_recvit(td, s, mp, UIO_USERSPACE, NULL); if (error) return (error); if (namelenp) { error = copyout(&mp->msg_namelen, namelenp, sizeof (socklen_t)); #ifdef COMPAT_OLDSOCK if (mp->msg_flags & MSG_COMPAT) error = 0; /* old recvfrom didn't check */ #endif } return (error); } /* * MPSAFE */ int recvfrom(td, uap) struct thread *td; register struct recvfrom_args /* { int s; caddr_t buf; size_t len; int flags; struct sockaddr * __restrict from; socklen_t * __restrict fromlenaddr; } */ *uap; { struct msghdr msg; struct iovec aiov; int error; if (uap->fromlenaddr) { error = copyin(uap->fromlenaddr, &msg.msg_namelen, sizeof (msg.msg_namelen)); if (error) goto done2; } else { msg.msg_namelen = 0; } msg.msg_name = uap->from; msg.msg_iov = &aiov; msg.msg_iovlen = 1; aiov.iov_base = uap->buf; aiov.iov_len = uap->len; msg.msg_control = 0; msg.msg_flags = uap->flags; error = recvit(td, uap->s, &msg, uap->fromlenaddr); done2: return(error); } #ifdef COMPAT_OLDSOCK /* * MPSAFE */ int orecvfrom(td, uap) struct thread *td; struct recvfrom_args *uap; { uap->flags |= MSG_COMPAT; return (recvfrom(td, uap)); } #endif #ifdef COMPAT_OLDSOCK /* * MPSAFE */ int orecv(td, uap) struct thread *td; register struct orecv_args /* { int s; caddr_t buf; int len; int flags; } */ *uap; { struct msghdr msg; struct iovec aiov; int error; msg.msg_name = 0; msg.msg_namelen = 0; msg.msg_iov = &aiov; msg.msg_iovlen = 1; aiov.iov_base = uap->buf; aiov.iov_len = uap->len; msg.msg_control = 0; msg.msg_flags = uap->flags; error = recvit(td, uap->s, &msg, NULL); return (error); } /* * Old recvmsg. This code takes advantage of the fact that the old msghdr * overlays the new one, missing only the flags, and with the (old) access * rights where the control fields are now. * * MPSAFE */ int orecvmsg(td, uap) struct thread *td; struct orecvmsg_args /* { int s; struct omsghdr *msg; int flags; } */ *uap; { struct msghdr msg; struct iovec *iov; int error; error = copyin(uap->msg, &msg, sizeof (struct omsghdr)); if (error) return (error); error = copyiniov(msg.msg_iov, msg.msg_iovlen, &iov, EMSGSIZE); if (error) return (error); msg.msg_flags = uap->flags | MSG_COMPAT; msg.msg_iov = iov; error = recvit(td, uap->s, &msg, &uap->msg->msg_namelen); if (msg.msg_controllen && error == 0) error = copyout(&msg.msg_controllen, &uap->msg->msg_accrightslen, sizeof (int)); free(iov, M_IOV); return (error); } #endif /* * MPSAFE */ int recvmsg(td, uap) struct thread *td; struct recvmsg_args /* { int s; struct msghdr *msg; int flags; } */ *uap; { struct msghdr msg; struct iovec *uiov, *iov; int error; error = copyin(uap->msg, &msg, sizeof (msg)); if (error) return (error); error = copyiniov(msg.msg_iov, msg.msg_iovlen, &iov, EMSGSIZE); if (error) return (error); msg.msg_flags = uap->flags; #ifdef COMPAT_OLDSOCK msg.msg_flags &= ~MSG_COMPAT; #endif uiov = msg.msg_iov; msg.msg_iov = iov; error = recvit(td, uap->s, &msg, NULL); if (error == 0) { msg.msg_iov = uiov; error = copyout(&msg, uap->msg, sizeof(msg)); } free(iov, M_IOV); return (error); } /* * MPSAFE */ /* ARGSUSED */ int shutdown(td, uap) struct thread *td; register struct shutdown_args /* { int s; int how; } */ *uap; { struct socket *so; struct file *fp; int error; NET_LOCK_GIANT(); error = getsock(td->td_proc->p_fd, uap->s, &fp, NULL); if (error == 0) { so = fp->f_data; error = soshutdown(so, uap->how); fdrop(fp, td); } NET_UNLOCK_GIANT(); return (error); } /* * MPSAFE */ /* ARGSUSED */ int setsockopt(td, uap) struct thread *td; register struct setsockopt_args /* { int s; int level; int name; caddr_t val; int valsize; } */ *uap; { return (kern_setsockopt(td, uap->s, uap->level, uap->name, uap->val, UIO_USERSPACE, uap->valsize)); } int kern_setsockopt(td, s, level, name, val, valseg, valsize) struct thread *td; int s; int level; int name; void *val; enum uio_seg valseg; socklen_t valsize; { int error; struct socket *so; struct file *fp; struct sockopt sopt; if (val == NULL && valsize != 0) return (EFAULT); if ((int)valsize < 0) return (EINVAL); sopt.sopt_dir = SOPT_SET; sopt.sopt_level = level; sopt.sopt_name = name; sopt.sopt_val = val; sopt.sopt_valsize = valsize; switch (valseg) { case UIO_USERSPACE: sopt.sopt_td = td; break; case UIO_SYSSPACE: sopt.sopt_td = NULL; break; default: panic("kern_setsockopt called with bad valseg"); } NET_LOCK_GIANT(); error = getsock(td->td_proc->p_fd, s, &fp, NULL); if (error == 0) { so = fp->f_data; error = sosetopt(so, &sopt); fdrop(fp, td); } NET_UNLOCK_GIANT(); return(error); } /* * MPSAFE */ /* ARGSUSED */ int getsockopt(td, uap) struct thread *td; register struct getsockopt_args /* { int s; int level; int name; void * __restrict val; socklen_t * __restrict avalsize; } */ *uap; { socklen_t valsize; int error; if (uap->val) { error = copyin(uap->avalsize, &valsize, sizeof (valsize)); if (error) return (error); } error = kern_getsockopt(td, uap->s, uap->level, uap->name, uap->val, UIO_USERSPACE, &valsize); if (error == 0) error = copyout(&valsize, uap->avalsize, sizeof (valsize)); return (error); } /* * Kernel version of getsockopt. * optval can be a userland or userspace. optlen is always a kernel pointer. */ int kern_getsockopt(td, s, level, name, val, valseg, valsize) struct thread *td; int s; int level; int name; void *val; enum uio_seg valseg; socklen_t *valsize; { int error; struct socket *so; struct file *fp; struct sockopt sopt; if (val == NULL) *valsize = 0; if ((int)*valsize < 0) return (EINVAL); sopt.sopt_dir = SOPT_GET; sopt.sopt_level = level; sopt.sopt_name = name; sopt.sopt_val = val; sopt.sopt_valsize = (size_t)*valsize; /* checked non-negative above */ switch (valseg) { case UIO_USERSPACE: sopt.sopt_td = td; break; case UIO_SYSSPACE: sopt.sopt_td = NULL; break; default: panic("kern_getsockopt called with bad valseg"); } NET_LOCK_GIANT(); error = getsock(td->td_proc->p_fd, s, &fp, NULL); if (error == 0) { so = fp->f_data; error = sogetopt(so, &sopt); *valsize = sopt.sopt_valsize; fdrop(fp, td); } NET_UNLOCK_GIANT(); return (error); } /* * getsockname1() - Get socket name. * * MPSAFE */ /* ARGSUSED */ static int getsockname1(td, uap, compat) struct thread *td; register struct getsockname_args /* { int fdes; struct sockaddr * __restrict asa; socklen_t * __restrict alen; } */ *uap; int compat; { struct sockaddr *sa; socklen_t len; int error; error = copyin(uap->alen, &len, sizeof(len)); if (error) return (error); error = kern_getsockname(td, uap->fdes, &sa, &len); if (error) return (error); if (len != 0) { #ifdef COMPAT_OLDSOCK if (compat) ((struct osockaddr *)sa)->sa_family = sa->sa_family; #endif error = copyout(sa, uap->asa, (u_int)len); } free(sa, M_SONAME); if (error == 0) error = copyout(&len, uap->alen, sizeof(len)); return (error); } int kern_getsockname(struct thread *td, int fd, struct sockaddr **sa, socklen_t *alen) { struct socket *so; struct file *fp; socklen_t len; int error; if (*alen < 0) return (EINVAL); NET_LOCK_GIANT(); error = getsock(td->td_proc->p_fd, fd, &fp, NULL); if (error) goto done; so = fp->f_data; *sa = NULL; error = (*so->so_proto->pr_usrreqs->pru_sockaddr)(so, sa); if (error) goto bad; if (*sa == NULL) len = 0; else len = MIN(*alen, (*sa)->sa_len); *alen = len; bad: fdrop(fp, td); if (error && *sa) { free(*sa, M_SONAME); *sa = NULL; } done: NET_UNLOCK_GIANT(); return (error); } /* * MPSAFE */ int getsockname(td, uap) struct thread *td; struct getsockname_args *uap; { return (getsockname1(td, uap, 0)); } #ifdef COMPAT_OLDSOCK /* * MPSAFE */ int ogetsockname(td, uap) struct thread *td; struct getsockname_args *uap; { return (getsockname1(td, uap, 1)); } #endif /* COMPAT_OLDSOCK */ /* * getpeername1() - Get name of peer for connected socket. * * MPSAFE */ /* ARGSUSED */ static int getpeername1(td, uap, compat) struct thread *td; register struct getpeername_args /* { int fdes; struct sockaddr * __restrict asa; socklen_t * __restrict alen; } */ *uap; int compat; { struct sockaddr *sa; socklen_t len; int error; error = copyin(uap->alen, &len, sizeof (len)); if (error) return (error); error = kern_getpeername(td, uap->fdes, &sa, &len); if (error) return (error); if (len != 0) { #ifdef COMPAT_OLDSOCK if (compat) ((struct osockaddr *)sa)->sa_family = sa->sa_family; #endif error = copyout(sa, uap->asa, (u_int)len); } free(sa, M_SONAME); if (error == 0) error = copyout(&len, uap->alen, sizeof(len)); return (error); } int kern_getpeername(struct thread *td, int fd, struct sockaddr **sa, socklen_t *alen) { struct socket *so; struct file *fp; socklen_t len; int error; if (*alen < 0) return (EINVAL); NET_LOCK_GIANT(); error = getsock(td->td_proc->p_fd, fd, &fp, NULL); if (error) goto done2; so = fp->f_data; if ((so->so_state & (SS_ISCONNECTED|SS_ISCONFIRMING)) == 0) { error = ENOTCONN; goto done1; } *sa = NULL; error = (*so->so_proto->pr_usrreqs->pru_peeraddr)(so, sa); if (error) goto bad; if (*sa == NULL) len = 0; else len = MIN(*alen, (*sa)->sa_len); *alen = len; bad: if (error && *sa) { free(*sa, M_SONAME); *sa = NULL; } done1: fdrop(fp, td); done2: NET_UNLOCK_GIANT(); return (error); } /* * MPSAFE */ int getpeername(td, uap) struct thread *td; struct getpeername_args *uap; { return (getpeername1(td, uap, 0)); } #ifdef COMPAT_OLDSOCK /* * MPSAFE */ int ogetpeername(td, uap) struct thread *td; struct ogetpeername_args *uap; { /* XXX uap should have type `getpeername_args *' to begin with. */ return (getpeername1(td, (struct getpeername_args *)uap, 1)); } #endif /* COMPAT_OLDSOCK */ int sockargs(mp, buf, buflen, type) struct mbuf **mp; caddr_t buf; int buflen, type; { register struct sockaddr *sa; register struct mbuf *m; int error; if ((u_int)buflen > MLEN) { #ifdef COMPAT_OLDSOCK if (type == MT_SONAME && (u_int)buflen <= 112) buflen = MLEN; /* unix domain compat. hack */ else #endif if ((u_int)buflen > MCLBYTES) return (EINVAL); } m = m_get(M_TRYWAIT, type); if (m == NULL) return (ENOBUFS); if ((u_int)buflen > MLEN) { MCLGET(m, M_TRYWAIT); if ((m->m_flags & M_EXT) == 0) { m_free(m); return (ENOBUFS); } } m->m_len = buflen; error = copyin(buf, mtod(m, caddr_t), (u_int)buflen); if (error) (void) m_free(m); else { *mp = m; if (type == MT_SONAME) { sa = mtod(m, struct sockaddr *); #if defined(COMPAT_OLDSOCK) && BYTE_ORDER != BIG_ENDIAN if (sa->sa_family == 0 && sa->sa_len < AF_MAX) sa->sa_family = sa->sa_len; #endif sa->sa_len = buflen; } } return (error); } int getsockaddr(namp, uaddr, len) struct sockaddr **namp; caddr_t uaddr; size_t len; { struct sockaddr *sa; int error; if (len > SOCK_MAXADDRLEN) return (ENAMETOOLONG); if (len < offsetof(struct sockaddr, sa_data[0])) return (EINVAL); MALLOC(sa, struct sockaddr *, len, M_SONAME, M_WAITOK); error = copyin(uaddr, sa, len); if (error) { FREE(sa, M_SONAME); } else { #if defined(COMPAT_OLDSOCK) && BYTE_ORDER != BIG_ENDIAN if (sa->sa_family == 0 && sa->sa_len < AF_MAX) sa->sa_family = sa->sa_len; #endif sa->sa_len = len; *namp = sa; } return (error); } /* * Detach mapped page and release resources back to the system. */ void sf_buf_mext(void *addr, void *args) { vm_page_t m; m = sf_buf_page(args); sf_buf_free(args); vm_page_lock_queues(); vm_page_unwire(m, 0); /* * Check for the object going away on us. This can * happen since we don't hold a reference to it. * If so, we're responsible for freeing the page. */ if (m->wire_count == 0 && m->object == NULL) vm_page_free(m); vm_page_unlock_queues(); } /* * sendfile(2) * * MPSAFE * * int sendfile(int fd, int s, off_t offset, size_t nbytes, * struct sf_hdtr *hdtr, off_t *sbytes, int flags) * * Send a file specified by 'fd' and starting at 'offset' to a socket * specified by 's'. Send only 'nbytes' of the file or until EOF if * nbytes == 0. Optionally add a header and/or trailer to the socket * output. If specified, write the total number of bytes sent into *sbytes. * */ int sendfile(struct thread *td, struct sendfile_args *uap) { return (do_sendfile(td, uap, 0)); } static int do_sendfile(struct thread *td, struct sendfile_args *uap, int compat) { struct sf_hdtr hdtr; struct uio *hdr_uio, *trl_uio; int error; hdr_uio = trl_uio = NULL; if (uap->hdtr != NULL) { error = copyin(uap->hdtr, &hdtr, sizeof(hdtr)); if (error) goto out; if (hdtr.headers != NULL) { error = copyinuio(hdtr.headers, hdtr.hdr_cnt, &hdr_uio); if (error) goto out; } if (hdtr.trailers != NULL) { error = copyinuio(hdtr.trailers, hdtr.trl_cnt, &trl_uio); if (error) goto out; } } error = kern_sendfile(td, uap, hdr_uio, trl_uio, compat); out: if (hdr_uio) free(hdr_uio, M_IOV); if (trl_uio) free(trl_uio, M_IOV); return (error); } #ifdef COMPAT_FREEBSD4 int freebsd4_sendfile(struct thread *td, struct freebsd4_sendfile_args *uap) { struct sendfile_args args; args.fd = uap->fd; args.s = uap->s; args.offset = uap->offset; args.nbytes = uap->nbytes; args.hdtr = uap->hdtr; args.sbytes = uap->sbytes; args.flags = uap->flags; return (do_sendfile(td, &args, 1)); } #endif /* COMPAT_FREEBSD4 */ int kern_sendfile(struct thread *td, struct sendfile_args *uap, struct uio *hdr_uio, struct uio *trl_uio, int compat) { struct file *sock_fp; struct vnode *vp; struct vm_object *obj = NULL; struct socket *so = NULL; struct mbuf *m, *m_header = NULL; struct sf_buf *sf; struct vm_page *pg; off_t off, xfsize, hdtr_size, sbytes = 0; int error, headersize = 0, headersent = 0; int vfslocked; NET_LOCK_GIANT(); hdtr_size = 0; /* * The descriptor must be a regular file and have a backing VM object. */ if ((error = fgetvp_read(td, uap->fd, &vp)) != 0) goto done; vfslocked = VFS_LOCK_GIANT(vp->v_mount); vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td); obj = vp->v_object; if (obj != NULL) { /* * Temporarily increase the backing VM object's reference * count so that a forced reclamation of its vnode does not * immediately destroy it. */ VM_OBJECT_LOCK(obj); if ((obj->flags & OBJ_DEAD) == 0) { vm_object_reference_locked(obj); VM_OBJECT_UNLOCK(obj); } else { VM_OBJECT_UNLOCK(obj); obj = NULL; } } VOP_UNLOCK(vp, 0, td); VFS_UNLOCK_GIANT(vfslocked); if (obj == NULL) { error = EINVAL; goto done; } if ((error = getsock(td->td_proc->p_fd, uap->s, &sock_fp, NULL)) != 0) goto done; so = sock_fp->f_data; if (so->so_type != SOCK_STREAM) { error = EINVAL; goto done; } if ((so->so_state & SS_ISCONNECTED) == 0) { error = ENOTCONN; goto done; } if (uap->offset < 0) { error = EINVAL; goto done; } #ifdef MAC SOCK_LOCK(so); error = mac_check_socket_send(td->td_ucred, so); SOCK_UNLOCK(so); if (error) goto done; #endif /* * If specified, get the pointer to the sf_hdtr struct for * any headers/trailers. */ if (hdr_uio != NULL) { hdr_uio->uio_td = td; hdr_uio->uio_rw = UIO_WRITE; if (hdr_uio->uio_resid > 0) { m_header = m_uiotombuf(hdr_uio, M_DONTWAIT, 0, 0); if (m_header == NULL) goto done; headersize = m_header->m_pkthdr.len; if (compat) sbytes += headersize; } } /* * Protect against multiple writers to the socket. */ SOCKBUF_LOCK(&so->so_snd); (void) sblock(&so->so_snd, M_WAITOK); SOCKBUF_UNLOCK(&so->so_snd); /* * Loop through the pages in the file, starting with the requested * offset. Get a file page (do I/O if necessary), map the file page * into an sf_buf, attach an mbuf header to the sf_buf, and queue * it on the socket. */ for (off = uap->offset; ; off += xfsize, sbytes += xfsize) { vm_pindex_t pindex; vm_offset_t pgoff; pindex = OFF_TO_IDX(off); VM_OBJECT_LOCK(obj); retry_lookup: /* * Calculate the amount to transfer. Not to exceed a page, * the EOF, or the passed in nbytes. */ xfsize = obj->un_pager.vnp.vnp_size - off; VM_OBJECT_UNLOCK(obj); if (xfsize > PAGE_SIZE) xfsize = PAGE_SIZE; pgoff = (vm_offset_t)(off & PAGE_MASK); if (PAGE_SIZE - pgoff < xfsize) xfsize = PAGE_SIZE - pgoff; if (uap->nbytes && xfsize > (uap->nbytes - sbytes)) xfsize = uap->nbytes - sbytes; if (xfsize <= 0) { if (m_header != NULL) { m = m_header; m_header = NULL; SOCKBUF_LOCK(&so->so_snd); goto retry_space; } else break; } /* * Optimize the non-blocking case by looking at the socket space * before going to the extra work of constituting the sf_buf. */ SOCKBUF_LOCK(&so->so_snd); if ((so->so_state & SS_NBIO) && sbspace(&so->so_snd) <= 0) { if (so->so_snd.sb_state & SBS_CANTSENDMORE) error = EPIPE; else error = EAGAIN; sbunlock(&so->so_snd); SOCKBUF_UNLOCK(&so->so_snd); goto done; } SOCKBUF_UNLOCK(&so->so_snd); VM_OBJECT_LOCK(obj); /* * Attempt to look up the page. * * Allocate if not found * * Wait and loop if busy. */ pg = vm_page_lookup(obj, pindex); if (pg == NULL) { pg = vm_page_alloc(obj, pindex, VM_ALLOC_NOBUSY | VM_ALLOC_NORMAL | VM_ALLOC_WIRED); if (pg == NULL) { VM_OBJECT_UNLOCK(obj); VM_WAIT; VM_OBJECT_LOCK(obj); goto retry_lookup; } vm_page_lock_queues(); } else { vm_page_lock_queues(); if (vm_page_sleep_if_busy(pg, TRUE, "sfpbsy")) goto retry_lookup; /* * Wire the page so it does not get ripped out from * under us. */ vm_page_wire(pg); } /* * If page is not valid for what we need, initiate I/O */ if (pg->valid && vm_page_is_valid(pg, pgoff, xfsize)) { VM_OBJECT_UNLOCK(obj); } else if (uap->flags & SF_NODISKIO) { error = EBUSY; } else { int bsize, resid; /* * Ensure that our page is still around when the I/O * completes. */ vm_page_io_start(pg); vm_page_unlock_queues(); VM_OBJECT_UNLOCK(obj); /* * Get the page from backing store. */ bsize = vp->v_mount->mnt_stat.f_iosize; vfslocked = VFS_LOCK_GIANT(vp->v_mount); vn_lock(vp, LK_SHARED | LK_RETRY, td); /* * XXXMAC: Because we don't have fp->f_cred here, * we pass in NOCRED. This is probably wrong, but * is consistent with our original implementation. */ error = vn_rdwr(UIO_READ, vp, NULL, MAXBSIZE, trunc_page(off), UIO_NOCOPY, IO_NODELOCKED | IO_VMIO | ((MAXBSIZE / bsize) << IO_SEQSHIFT), td->td_ucred, NOCRED, &resid, td); VOP_UNLOCK(vp, 0, td); VFS_UNLOCK_GIANT(vfslocked); VM_OBJECT_LOCK(obj); vm_page_lock_queues(); vm_page_io_finish(pg); if (!error) VM_OBJECT_UNLOCK(obj); mbstat.sf_iocnt++; } if (error) { vm_page_unwire(pg, 0); /* * See if anyone else might know about this page. * If not and it is not valid, then free it. */ if (pg->wire_count == 0 && pg->valid == 0 && pg->busy == 0 && !(pg->flags & PG_BUSY) && pg->hold_count == 0) { vm_page_free(pg); } vm_page_unlock_queues(); VM_OBJECT_UNLOCK(obj); SOCKBUF_LOCK(&so->so_snd); sbunlock(&so->so_snd); SOCKBUF_UNLOCK(&so->so_snd); goto done; } vm_page_unlock_queues(); /* * Get a sendfile buf. We usually wait as long as necessary, * but this wait can be interrupted. */ if ((sf = sf_buf_alloc(pg, SFB_CATCH)) == NULL) { mbstat.sf_allocfail++; vm_page_lock_queues(); vm_page_unwire(pg, 0); if (pg->wire_count == 0 && pg->object == NULL) vm_page_free(pg); vm_page_unlock_queues(); SOCKBUF_LOCK(&so->so_snd); sbunlock(&so->so_snd); SOCKBUF_UNLOCK(&so->so_snd); error = EINTR; goto done; } /* * Get an mbuf header and set it up as having external storage. */ if (m_header) MGET(m, M_TRYWAIT, MT_DATA); else MGETHDR(m, M_TRYWAIT, MT_DATA); if (m == NULL) { error = ENOBUFS; sf_buf_mext((void *)sf_buf_kva(sf), sf); SOCKBUF_LOCK(&so->so_snd); sbunlock(&so->so_snd); SOCKBUF_UNLOCK(&so->so_snd); goto done; } /* * Setup external storage for mbuf. */ MEXTADD(m, sf_buf_kva(sf), PAGE_SIZE, sf_buf_mext, sf, M_RDONLY, EXT_SFBUF); m->m_data = (char *)sf_buf_kva(sf) + pgoff; m->m_pkthdr.len = m->m_len = xfsize; if (m_header) { m_cat(m_header, m); m = m_header; m_header = NULL; m_fixhdr(m); } /* * Add the buffer to the socket buffer chain. */ SOCKBUF_LOCK(&so->so_snd); retry_space: /* * Make sure that the socket is still able to take more data. * CANTSENDMORE being true usually means that the connection * was closed. so_error is true when an error was sensed after * a previous send. * The state is checked after the page mapping and buffer * allocation above since those operations may block and make * any socket checks stale. From this point forward, nothing * blocks before the pru_send (or more accurately, any blocking * results in a loop back to here to re-check). */ SOCKBUF_LOCK_ASSERT(&so->so_snd); if ((so->so_snd.sb_state & SBS_CANTSENDMORE) || so->so_error) { if (so->so_snd.sb_state & SBS_CANTSENDMORE) { error = EPIPE; } else { error = so->so_error; so->so_error = 0; } m_freem(m); sbunlock(&so->so_snd); SOCKBUF_UNLOCK(&so->so_snd); goto done; } /* * Wait for socket space to become available. We do this just * after checking the connection state above in order to avoid * a race condition with sbwait(). */ if (sbspace(&so->so_snd) < so->so_snd.sb_lowat) { if (so->so_state & SS_NBIO) { m_freem(m); sbunlock(&so->so_snd); SOCKBUF_UNLOCK(&so->so_snd); error = EAGAIN; goto done; } error = sbwait(&so->so_snd); /* * An error from sbwait usually indicates that we've * been interrupted by a signal. If we've sent anything * then return bytes sent, otherwise return the error. */ if (error) { m_freem(m); sbunlock(&so->so_snd); SOCKBUF_UNLOCK(&so->so_snd); goto done; } goto retry_space; } SOCKBUF_UNLOCK(&so->so_snd); error = (*so->so_proto->pr_usrreqs->pru_send)(so, 0, m, 0, 0, td); if (error) { SOCKBUF_LOCK(&so->so_snd); sbunlock(&so->so_snd); SOCKBUF_UNLOCK(&so->so_snd); goto done; } headersent = 1; } SOCKBUF_LOCK(&so->so_snd); sbunlock(&so->so_snd); SOCKBUF_UNLOCK(&so->so_snd); /* * Send trailers. Wimp out and use writev(2). */ if (trl_uio != NULL) { error = kern_writev(td, uap->s, trl_uio); if (error) goto done; if (compat) sbytes += td->td_retval[0]; else hdtr_size += td->td_retval[0]; } done: if (headersent) { if (!compat) hdtr_size += headersize; } else { if (compat) sbytes -= headersize; } /* * If there was no error we have to clear td->td_retval[0] * because it may have been set by writev. */ if (error == 0) { td->td_retval[0] = 0; } if (uap->sbytes != NULL) { if (!compat) sbytes += hdtr_size; copyout(&sbytes, uap->sbytes, sizeof(off_t)); } if (obj != NULL) vm_object_deallocate(obj); if (vp != NULL) { vfslocked = VFS_LOCK_GIANT(vp->v_mount); vrele(vp); VFS_UNLOCK_GIANT(vfslocked); } if (so) fdrop(sock_fp, td); if (m_header) m_freem(m_header); NET_UNLOCK_GIANT(); if (error == ERESTART) error = EINTR; return (error); } Index: head/sys/sys/syscallsubr.h =================================================================== --- head/sys/sys/syscallsubr.h (revision 160764) +++ head/sys/sys/syscallsubr.h (revision 160765) @@ -1,181 +1,182 @@ /*- * Copyright (c) 2002 Ian Dowse. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #ifndef _SYS_SYSCALLSUBR_H_ #define _SYS_SYSCALLSUBR_H_ #include #include #include #include #include +struct file; struct itimerval; struct image_args; struct mbuf; struct msghdr; struct msqid_ds; struct rlimit; struct rusage; union semun; struct sockaddr; struct stat; struct kevent; struct kevent_copyops; struct sendfile_args; int kern___getcwd(struct thread *td, u_char *buf, enum uio_seg bufseg, u_int buflen); int kern_accept(struct thread *td, int s, struct sockaddr **name, - socklen_t *namelen); + socklen_t *namelen, struct file **fp); int kern_access(struct thread *td, char *path, enum uio_seg pathseg, int flags); int kern_adjtime(struct thread *td, struct timeval *delta, struct timeval *olddelta); int kern_alternate_path(struct thread *td, const char *prefix, char *path, enum uio_seg pathseg, char **pathbuf, int create); int kern_bind(struct thread *td, int fd, struct sockaddr *sa); int kern_chdir(struct thread *td, char *path, enum uio_seg pathseg); int kern_chmod(struct thread *td, char *path, enum uio_seg pathseg, int mode); int kern_chown(struct thread *td, char *path, enum uio_seg pathseg, int uid, int gid); int kern_clock_getres(struct thread *td, clockid_t clock_id, struct timespec *ts); int kern_clock_gettime(struct thread *td, clockid_t clock_id, struct timespec *ats); int kern_clock_settime(struct thread *td, clockid_t clock_id, struct timespec *ats); int kern_close(struct thread *td, int fd); int kern_connect(struct thread *td, int fd, struct sockaddr *sa); int kern_eaccess(struct thread *td, char *path, enum uio_seg pathseg, int flags); int kern_execve(struct thread *td, struct image_args *args, struct mac *mac_p); int kern_fcntl(struct thread *td, int fd, int cmd, intptr_t arg); int kern_fhstatfs(struct thread *td, fhandle_t fh, struct statfs *buf); int kern_fstat(struct thread *td, int fd, struct stat *sbp); int kern_fstatfs(struct thread *td, int fd, struct statfs *buf); int kern_futimes(struct thread *td, int fd, struct timeval *tptr, enum uio_seg tptrseg); int kern_getfsstat(struct thread *td, struct statfs **buf, size_t bufsize, enum uio_seg bufseg, int flags); int kern_getgroups(struct thread *td, u_int *ngrp, gid_t *groups); int kern_getitimer(struct thread *, u_int, struct itimerval *); int kern_getpeername(struct thread *td, int fd, struct sockaddr **sa, socklen_t *alen); int kern_getrusage(struct thread *td, int who, struct rusage *rup); int kern_getsockname(struct thread *td, int fd, struct sockaddr **sa, socklen_t *alen); int kern_getsockopt(struct thread *td, int s, int level, int name, void *optval, enum uio_seg valseg, socklen_t *valsize); int kern_ioctl(struct thread *td, int fd, u_long com, caddr_t data); int kern_kevent(struct thread *td, int fd, int nchanges, int nevents, struct kevent_copyops *k_ops, const struct timespec *timeout); int kern_kldload(struct thread *td, const char *file, int *fileid); int kern_kldunload(struct thread *td, int fileid, int flags); int kern_lchown(struct thread *td, char *path, enum uio_seg pathseg, int uid, int gid); int kern_link(struct thread *td, char *path, char *link, enum uio_seg segflg); int kern_lstat(struct thread *td, char *path, enum uio_seg pathseg, struct stat *sbp); int kern_lutimes(struct thread *td, char *path, enum uio_seg pathseg, struct timeval *tptr, enum uio_seg tptrseg); int kern_mkdir(struct thread *td, char *path, enum uio_seg segflg, int mode); int kern_mkfifo(struct thread *td, char *path, enum uio_seg pathseg, int mode); int kern_mknod(struct thread *td, char *path, enum uio_seg pathseg, int mode, int dev); int kern_msgctl(struct thread *, int, int, struct msqid_ds *); int kern_nanosleep(struct thread *td, struct timespec *rqt, struct timespec *rmt); int kern_open(struct thread *td, char *path, enum uio_seg pathseg, int flags, int mode); int kern_pathconf(struct thread *td, char *path, enum uio_seg pathseg, int name); int kern_preadv(struct thread *td, int fd, struct uio *auio, off_t offset); int kern_ptrace(struct thread *td, int req, pid_t pid, void *addr, int data); int kern_pwritev(struct thread *td, int fd, struct uio *auio, off_t offset); int kern_readlink(struct thread *td, char *path, enum uio_seg pathseg, char *buf, enum uio_seg bufseg, int count); int kern_readv(struct thread *td, int fd, struct uio *auio); int kern_recvit(struct thread *td, int s, struct msghdr *mp, enum uio_seg fromseg, struct mbuf **controlp); int kern_rename(struct thread *td, char *from, char *to, enum uio_seg pathseg); int kern_rmdir(struct thread *td, char *path, enum uio_seg pathseg); int kern_sched_rr_get_interval(struct thread *td, pid_t pid, struct timespec *ts); int kern_semctl(struct thread *td, int semid, int semnum, int cmd, union semun *arg, register_t *rval); int kern_select(struct thread *td, int nd, fd_set *fd_in, fd_set *fd_ou, fd_set *fd_ex, struct timeval *tvp); int kern_sendfile(struct thread *td, struct sendfile_args *uap, struct uio *hdr_uio, struct uio *trl_uio, int compat); int kern_sendit(struct thread *td, int s, struct msghdr *mp, int flags, struct mbuf *control, enum uio_seg segflg); int kern_setgroups(struct thread *td, u_int ngrp, gid_t *groups); int kern_setitimer(struct thread *, u_int, struct itimerval *, struct itimerval *); int kern_setrlimit(struct thread *, u_int, struct rlimit *); int kern_setsockopt(struct thread *td, int s, int level, int name, void *optval, enum uio_seg valseg, socklen_t valsize); int kern_settimeofday(struct thread *td, struct timeval *tv, struct timezone *tzp); int kern_shmat(struct thread *td, int shmid, const void *shmaddr, int shmflg); int kern_shmctl(struct thread *td, int shmid, int cmd, void *buf, size_t *bufsz); int kern_sigaction(struct thread *td, int sig, struct sigaction *act, struct sigaction *oact, int flags); int kern_sigaltstack(struct thread *td, stack_t *ss, stack_t *oss); int kern_sigprocmask(struct thread *td, int how, sigset_t *set, sigset_t *oset, int old); int kern_sigsuspend(struct thread *td, sigset_t mask); int kern_stat(struct thread *td, char *path, enum uio_seg pathseg, struct stat *sbp); int kern_statfs(struct thread *td, char *path, enum uio_seg pathseg, struct statfs *buf); int kern_symlink(struct thread *td, char *path, char *link, enum uio_seg segflg); int kern_truncate(struct thread *td, char *path, enum uio_seg pathseg, off_t length); int kern_unlink(struct thread *td, char *path, enum uio_seg pathseg); int kern_utimes(struct thread *td, char *path, enum uio_seg pathseg, struct timeval *tptr, enum uio_seg tptrseg); int kern_wait(struct thread *td, pid_t pid, int *status, int options, struct rusage *rup); int kern_writev(struct thread *td, int fd, struct uio *auio); /* flags for kern_sigaction */ #define KSA_OSIGSET 0x0001 /* uses osigact_t */ #define KSA_FREEBSD4 0x0002 /* uses ucontext4 */ #endif /* !_SYS_SYSCALLSUBR_H_ */