minix/lib/libc/net/rcmd.c

1039 lines
24 KiB
C
Raw Normal View History

/* $NetBSD: rcmd.c,v 1.68 2012/07/14 15:06:26 darrenr Exp $ */
/*
* Copyright (c) 1983, 1993, 1994
* The Regents of the University of California. 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. 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.
*/
#include <sys/cdefs.h>
#if defined(LIBC_SCCS) && !defined(lint)
#if 0
static char sccsid[] = "@(#)rcmd.c 8.3 (Berkeley) 3/26/94";
#else
__RCSID("$NetBSD: rcmd.c,v 1.68 2012/07/14 15:06:26 darrenr Exp $");
#endif
#endif /* LIBC_SCCS and not lint */
#ifdef _LIBC
#include "namespace.h"
#endif
#include <sys/param.h>
#include <sys/socket.h>
#include <sys/stat.h>
#include <sys/poll.h>
#include <sys/wait.h>
#include <netinet/in.h>
#ifndef __minix
#include <rpc/rpc.h>
#endif
#include <arpa/inet.h>
#include <netgroup.h>
#include <assert.h>
#include <ctype.h>
#include <err.h>
#include <errno.h>
#include <fcntl.h>
#include <grp.h>
#include <netdb.h>
#include <paths.h>
#include <pwd.h>
#include <signal.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <syslog.h>
#include <unistd.h>
#include "pathnames.h"
int orcmd(char **, u_int, const char *, const char *, const char *, int *);
int orcmd_af(char **, u_int, const char *, const char *, const char *,
int *, int);
int __ivaliduser(FILE *, u_int32_t, const char *, const char *);
int __ivaliduser_sa(FILE *, const struct sockaddr *, socklen_t,
const char *, const char *);
static int rshrcmd(int, char **, u_int32_t, const char *,
const char *, const char *, int *, const char *);
static int resrcmd(struct addrinfo *, char **, u_int32_t, const char *,
const char *, const char *, int *);
static int __icheckhost(const struct sockaddr *, socklen_t,
const char *);
static char *__gethostloop(const struct sockaddr *, socklen_t);
int
rcmd(char **ahost, int rport, const char *locuser, const char *remuser,
const char *cmd, int *fd2p)
{
return rcmd_af(ahost, rport, locuser, remuser, cmd, fd2p, AF_INET);
}
int
rcmd_af(char **ahost, int rport, const char *locuser, const char *remuser,
const char *cmd, int *fd2p, int af)
{
static char hbuf[MAXHOSTNAMELEN];
char pbuf[NI_MAXSERV];
struct addrinfo hints, *res;
int error;
struct servent *sp;
_DIAGASSERT(ahost != NULL);
_DIAGASSERT(locuser != NULL);
_DIAGASSERT(remuser != NULL);
_DIAGASSERT(cmd != NULL);
/* fd2p may be NULL */
snprintf(pbuf, sizeof(pbuf), "%u", ntohs(rport));
memset(&hints, 0, sizeof(hints));
hints.ai_family = af;
hints.ai_socktype = SOCK_STREAM;
hints.ai_flags = AI_CANONNAME;
error = getaddrinfo(*ahost, pbuf, &hints, &res);
if (error) {
warnx("%s: %s", *ahost, gai_strerror(error)); /*XXX*/
return -1;
}
if (res->ai_canonname) {
/*
* Canonicalise hostname.
* XXX: Should we really do this?
*/
strlcpy(hbuf, res->ai_canonname, sizeof(hbuf));
*ahost = hbuf;
}
/*
* Check if rport is the same as the shell port, and that the fd2p. If
* it is not, the program isn't expecting 'rsh' and so we can't use the
* RCMD_CMD environment.
*/
sp = getservbyname("shell", "tcp");
if (sp != NULL && sp->s_port == rport)
error = rshrcmd(af, ahost, (u_int32_t)rport,
locuser, remuser, cmd, fd2p, getenv("RCMD_CMD"));
else
error = resrcmd(res, ahost, (u_int32_t)rport,
locuser, remuser, cmd, fd2p);
freeaddrinfo(res);
return error;
}
/* this is simply a wrapper around hprcmd() that handles ahost first */
int
orcmd(char **ahost, u_int rport, const char *locuser, const char *remuser,
const char *cmd, int *fd2p)
{
return orcmd_af(ahost, rport, locuser, remuser, cmd, fd2p, AF_INET);
}
int
orcmd_af(char **ahost, u_int rport, const char *locuser, const char *remuser,
const char *cmd, int *fd2p, int af)
{
static char hbuf[MAXHOSTNAMELEN];
char pbuf[NI_MAXSERV];
struct addrinfo hints, *res;
int error;
_DIAGASSERT(ahost != NULL);
_DIAGASSERT(locuser != NULL);
_DIAGASSERT(remuser != NULL);
_DIAGASSERT(cmd != NULL);
/* fd2p may be NULL */
snprintf(pbuf, sizeof(pbuf), "%u", ntohs(rport));
memset(&hints, 0, sizeof(hints));
hints.ai_family = af;
hints.ai_socktype = SOCK_STREAM;
hints.ai_flags = AI_CANONNAME;
error = getaddrinfo(*ahost, pbuf, &hints, &res);
if (error) {
warnx("%s: %s", *ahost, gai_strerror(error)); /*XXX*/
return -1;
}
if (res->ai_canonname) {
strlcpy(hbuf, res->ai_canonname, sizeof(hbuf));
*ahost = hbuf;
}
error = resrcmd(res, ahost, rport, locuser, remuser, cmd, fd2p);
freeaddrinfo(res);
return error;
}
/*ARGSUSED*/
static int
resrcmd(struct addrinfo *res, char **ahost, u_int32_t rport,
const char *locuser, const char *remuser, const char *cmd, int *fd2p)
{
struct addrinfo *r;
struct sockaddr_storage from;
struct pollfd reads[2];
#ifdef __minix
/* No support for OOB data in Minix. */
#else
sigset_t nmask, omask;
#endif /* !__minix */
pid_t pid;
int s, lport, timo;
int pollr;
char c;
int refused;
_DIAGASSERT(res != NULL);
_DIAGASSERT(ahost != NULL);
_DIAGASSERT(locuser != NULL);
_DIAGASSERT(remuser != NULL);
_DIAGASSERT(cmd != NULL);
/* fd2p may be NULL */
r = res;
refused = 0;
pid = getpid();
#ifndef __minix
/* Minix has no support for OOB data,
no need to block SIGURG. */
sigemptyset(&nmask);
sigaddset(&nmask, SIGURG);
if (sigprocmask(SIG_BLOCK, &nmask, &omask) == -1)
return -1;
#endif /* !__minix */
for (timo = 1, lport = IPPORT_RESERVED - 1;;) {
s = rresvport_af(&lport, r->ai_family);
if (s < 0) {
if (errno == EAGAIN)
warnx("rcmd: socket: All ports in use");
else
warn("rcmd: socket");
if (r->ai_next) {
r = r->ai_next;
continue;
} else {
#ifndef __minix
(void)sigprocmask(SIG_SETMASK, &omask, NULL);
#endif /* !__minix */
return -1;
}
}
#ifndef __minix
/* No OOB support in Minix. */
fcntl(s, F_SETOWN, pid);
#endif /* !__minix */
if (connect(s, r->ai_addr, r->ai_addrlen) >= 0)
break;
(void)close(s);
if (errno == EADDRINUSE) {
lport--;
continue;
} else if (errno == ECONNREFUSED)
refused++;
if (r->ai_next) {
int oerrno = errno;
char hbuf[NI_MAXHOST];
const int niflags = NI_NUMERICHOST;
hbuf[0] = '\0';
if (getnameinfo(r->ai_addr, r->ai_addrlen,
hbuf, (socklen_t)sizeof(hbuf), NULL, 0, niflags) !=
0)
strlcpy(hbuf, "(invalid)", sizeof(hbuf));
errno = oerrno;
warn("rcmd: connect to address %s", hbuf);
r = r->ai_next;
hbuf[0] = '\0';
if (getnameinfo(r->ai_addr, r->ai_addrlen,
hbuf, (socklen_t)sizeof(hbuf), NULL, 0, niflags) !=
0)
strlcpy(hbuf, "(invalid)", sizeof(hbuf));
(void)fprintf(stderr, "Trying %s...\n", hbuf);
continue;
}
if (refused && timo <= 16) {
(void)sleep((unsigned int)timo);
timo *= 2;
r = res;
refused = 0;
continue;
}
(void)fprintf(stderr, "%s: %s\n", res->ai_canonname,
strerror(errno));
#ifndef __minix
/* No OOB support in Minix. */
(void)sigprocmask(SIG_SETMASK, &omask, NULL);
#endif /* !__minix */
return -1;
}
lport--;
if (fd2p == 0) {
write(s, "", 1);
lport = 0;
} else {
char num[8];
int s2 = rresvport_af(&lport, r->ai_family), s3;
socklen_t len = sizeof(from);
if (s2 < 0)
goto bad;
listen(s2, 1);
(void)snprintf(num, sizeof(num), "%d", lport);
if (write(s, num, strlen(num) + 1) !=
(ssize_t) (strlen(num) + 1)) {
warn("rcmd: write (setting up stderr)");
(void)close(s2);
goto bad;
}
reads[0].fd = s;
reads[0].events = POLLIN;
reads[1].fd = s2;
reads[1].events = POLLIN;
errno = 0;
pollr = poll(reads, 2, INFTIM);
if (pollr < 1 || (reads[1].revents & POLLIN) == 0) {
if (errno != 0)
warn("poll: setting up stderr");
else
warnx(
"poll: protocol failure in circuit setup");
(void)close(s2);
goto bad;
}
s3 = accept(s2, (struct sockaddr *)(void *)&from, &len);
(void)close(s2);
if (s3 < 0) {
warn("rcmd: accept");
lport = 0;
goto bad;
}
*fd2p = s3;
switch (((struct sockaddr *)(void *)&from)->sa_family) {
case AF_INET:
#ifdef INET6
case AF_INET6:
#endif
if (getnameinfo((struct sockaddr *)(void *)&from, len,
NULL, 0, num, (socklen_t)sizeof(num),
NI_NUMERICSERV) != 0 ||
(atoi(num) >= IPPORT_RESERVED ||
atoi(num) < IPPORT_RESERVED / 2)) {
warnx(
"rcmd: protocol failure in circuit setup.");
goto bad2;
}
break;
default:
break;
}
}
(void)write(s, locuser, strlen(locuser)+1);
(void)write(s, remuser, strlen(remuser)+1);
(void)write(s, cmd, strlen(cmd)+1);
if (read(s, &c, 1) != 1) {
warn("%s", *ahost);
goto bad2;
}
if (c != 0) {
while (read(s, &c, 1) == 1) {
(void)write(STDERR_FILENO, &c, 1);
if (c == '\n')
break;
}
goto bad2;
}
#ifndef __minix
(void)sigprocmask(SIG_SETMASK, &omask, NULL);
#endif /* __minix */
return s;
bad2:
if (lport)
(void)close(*fd2p);
bad:
(void)close(s);
#ifndef __minix
(void)sigprocmask(SIG_SETMASK, &omask, NULL);
#endif /* __minix */
return -1;
}
/*
* based on code written by Chris Siebenmann <cks@utcc.utoronto.ca>
*/
/* ARGSUSED */
static int
rshrcmd(int af, char **ahost, u_int32_t rport, const char *locuser,
const char *remuser, const char *cmd, int *fd2p, const char *rshcmd)
{
pid_t pid;
int sp[2], ep[2];
char *p;
struct passwd *pw, pwres;
char pwbuf[1024];
_DIAGASSERT(ahost != NULL);
_DIAGASSERT(locuser != NULL);
_DIAGASSERT(remuser != NULL);
_DIAGASSERT(cmd != NULL);
/* fd2p may be NULL */
/* What rsh/shell to use. */
if (rshcmd == NULL)
rshcmd = _PATH_BIN_RCMD;
/* locuser must exist on this host. */
if (getpwnam_r(locuser, &pwres, pwbuf, sizeof(pwbuf), &pw) != 0 ||
pw == NULL) {
warnx("%s: unknown user: %s", __func__, locuser);
return -1;
}
/* get a socketpair we'll use for stdin and stdout. */
if (socketpair(AF_LOCAL, SOCK_STREAM, 0, sp) < 0) {
warn("%s: socketpair", __func__);
return -1;
}
/* we will use this for the fd2 pointer */
if (fd2p) {
if (socketpair(AF_LOCAL, SOCK_STREAM, 0, ep) < 0) {
warn("%s: socketpair", __func__);
return -1;
}
*fd2p = ep[0];
}
pid = fork();
if (pid < 0) {
warn("%s: fork", __func__);
return -1;
}
if (pid == 0) {
/*
* child
* - we use sp[1] to be stdin/stdout, and close sp[0]
* - with fd2p, we use ep[1] for stderr, and close ep[0]
*/
(void)close(sp[0]);
if (dup2(sp[1], 0) < 0 || dup2(0, 1) < 0) {
warn("%s: dup2", __func__);
_exit(1);
}
(void)close(sp[1]);
if (fd2p) {
if (dup2(ep[1], 2) < 0) {
warn("%s: dup2", __func__);
_exit(1);
}
(void)close(ep[0]);
(void)close(ep[1]);
} else if (dup2(0, 2) < 0) {
warn("%s: dup2", __func__);
_exit(1);
}
/* fork again to lose parent. */
pid = fork();
if (pid < 0) {
warn("%s: second fork", __func__);
_exit(1);
}
if (pid > 0)
_exit(0);
/* Orphan. Become local user for rshprog. */
if (setuid(pw->pw_uid)) {
warn("%s: setuid(%lu)", __func__, (u_long)pw->pw_uid);
_exit(1);
}
/*
* If we are rcmd'ing to "localhost" as the same user as we
* are, then avoid running remote shell for efficiency.
*/
if (strcmp(*ahost, "localhost") == 0 &&
strcmp(locuser, remuser) == 0) {
if (pw->pw_shell[0] == '\0')
rshcmd = _PATH_BSHELL;
else
rshcmd = pw->pw_shell;
p = strrchr(rshcmd, '/');
execlp(rshcmd, p ? p + 1 : rshcmd, "-c", cmd, NULL);
} else {
const char *program;
program = strrchr(rshcmd, '/');
program = program ? program + 1 : rshcmd;
switch (af) {
case AF_INET:
execlp(rshcmd, program, "-4", "-l", remuser,
*ahost, cmd, NULL);
break;
case AF_INET6:
execlp(rshcmd, program, "-6", "-l", remuser,
*ahost, cmd, NULL);
break;
default:
/* typically AF_UNSPEC, plus whatever */
execlp(rshcmd, program, "-l", remuser,
*ahost, cmd, NULL);
break;
}
}
warn("%s: exec %s", __func__, rshcmd);
_exit(1);
}
/* Parent */
(void)close(sp[1]);
if (fd2p)
(void)close(ep[1]);
(void)waitpid(pid, NULL, 0);
return sp[0];
}
int
rresvport(int *alport)
{
_DIAGASSERT(alport != NULL);
return rresvport_af(alport, AF_INET);
}
int
rresvport_af(int *alport, int family)
{
return rresvport_af_addr(alport, family, NULL);
}
int
rresvport_af_addr(int *alport, int family, void *addr)
{
struct sockaddr_storage ss;
struct sockaddr *sa;
socklen_t salen;
int s;
u_int16_t *portp;
_DIAGASSERT(alport != NULL);
memset(&ss, 0, sizeof(ss));
sa = (struct sockaddr *)(void *)&ss;
switch (family) {
case AF_INET:
#ifdef BSD4_4
sa->sa_len =
#endif
salen = sizeof(struct sockaddr_in);
if (addr)
((struct sockaddr_in *)(void *)sa)->sin_addr =
((struct sockaddr_in *)addr)->sin_addr;
portp = &((struct sockaddr_in *)(void *)sa)->sin_port;
break;
#ifdef INET6
case AF_INET6:
#ifdef BSD4_4
sa->sa_len =
#endif
salen = sizeof(struct sockaddr_in6);
if (addr)
((struct sockaddr_in6 *)(void *)sa)->sin6_addr =
((struct sockaddr_in6 *)addr)->sin6_addr;
portp = &((struct sockaddr_in6 *)(void *)sa)->sin6_port;
break;
#endif
default:
errno = EAFNOSUPPORT;
return -1;
}
sa->sa_family = family;
s = socket(family, SOCK_STREAM, 0);
if (s < 0)
return -1;
#ifdef BSD4_4
switch (family) {
case AF_INET:
case AF_INET6:
*portp = 0;
if (bindresvport(s, (struct sockaddr_in *)(void *)sa) < 0) {
int sverr = errno;
(void)close(s);
errno = sverr;
return -1;
}
*alport = (int)ntohs(*portp);
return s;
default:
/* is it necessary to try keep code for other AFs? */
break;
}
#endif
for (;;) {
*portp = htons((u_short)*alport);
if (bind(s, sa, salen) >= 0)
return s;
if (errno != EADDRINUSE) {
(void)close(s);
return -1;
}
(*alport)--;
if (*alport == IPPORT_RESERVED/2) {
(void)close(s);
errno = EAGAIN; /* close */
return -1;
}
}
}
int __check_rhosts_file = 1;
const char *__rcmd_errstr;
int
ruserok(const char *rhost, int superuser, const char *ruser, const char *luser)
{
struct addrinfo hints, *res, *r;
int error;
_DIAGASSERT(rhost != NULL);
_DIAGASSERT(ruser != NULL);
_DIAGASSERT(luser != NULL);
memset(&hints, 0, sizeof(hints));
hints.ai_family = PF_UNSPEC;
hints.ai_socktype = SOCK_DGRAM; /*dummy*/
error = getaddrinfo(rhost, "0", &hints, &res);
if (error)
return -1;
for (r = res; r; r = r->ai_next) {
if (iruserok_sa(r->ai_addr, (int)r->ai_addrlen, superuser,
ruser, luser) == 0) {
freeaddrinfo(res);
return 0;
}
}
freeaddrinfo(res);
return -1;
}
/*
* New .rhosts strategy: We are passed an ip address. We spin through
* hosts.equiv and .rhosts looking for a match. When the .rhosts only
* has ip addresses, we don't have to trust a nameserver. When it
* contains hostnames, we spin through the list of addresses the nameserver
* gives us and look for a match.
*
* Returns 0 if ok, -1 if not ok.
*/
int
iruserok(u_int32_t raddr, int superuser, const char *ruser, const char *luser)
{
struct sockaddr_in irsin;
memset(&irsin, 0, sizeof(irsin));
irsin.sin_family = AF_INET;
#ifdef BSD4_4
irsin.sin_len = sizeof(irsin);
#endif
memcpy(&irsin.sin_addr, &raddr, sizeof(irsin.sin_addr));
return iruserok_sa(&irsin, (socklen_t)sizeof(irsin), superuser, ruser,
luser);
}
/*
* 2nd and 3rd arguments are typed like this, to avoid dependency between
* unistd.h and sys/socket.h. There's no better way.
*/
int
iruserok_sa(const void *raddr, int rlen, int superuser, const char *ruser,
const char *luser)
{
const struct sockaddr *sa;
struct stat sbuf;
struct passwd *pwd, pwres;
FILE *hostf;
uid_t uid;
gid_t gid;
int isvaliduser;
char pbuf[MAXPATHLEN];
char pwbuf[1024];
_DIAGASSERT(raddr != NULL);
_DIAGASSERT(ruser != NULL);
_DIAGASSERT(luser != NULL);
sa = raddr;
__rcmd_errstr = NULL;
hostf = superuser ? NULL : fopen(_PATH_HEQUIV, "r");
if (hostf) {
if (__ivaliduser_sa(hostf, sa, (socklen_t)rlen, luser,
ruser) == 0) {
(void)fclose(hostf);
return 0;
}
(void)fclose(hostf);
}
isvaliduser = -1;
if (__check_rhosts_file || superuser) {
if (getpwnam_r(luser, &pwres, pwbuf, sizeof(pwbuf), &pwd) != 0
|| pwd == NULL)
return -1;
(void)strlcpy(pbuf, pwd->pw_dir, sizeof(pbuf));
(void)strlcat(pbuf, "/.rhosts", sizeof(pbuf));
/*
* Change effective uid while opening and reading .rhosts.
* If root and reading an NFS mounted file system, can't
* read files that are protected read/write owner only.
*/
uid = geteuid();
gid = getegid();
(void)setegid(pwd->pw_gid);
initgroups(pwd->pw_name, pwd->pw_gid);
(void)seteuid(pwd->pw_uid);
hostf = fopen(pbuf, "r");
if (hostf != NULL) {
/*
* If not a regular file, or is owned by someone other
* than user or root or if writable by anyone but the
* owner, quit.
*/
if (lstat(pbuf, &sbuf) < 0)
__rcmd_errstr = ".rhosts lstat failed";
else if (!S_ISREG(sbuf.st_mode))
__rcmd_errstr = ".rhosts not regular file";
else if (fstat(fileno(hostf), &sbuf) < 0)
__rcmd_errstr = ".rhosts fstat failed";
else if (sbuf.st_uid && sbuf.st_uid != pwd->pw_uid)
__rcmd_errstr = "bad .rhosts owner";
else if (sbuf.st_mode & (S_IWGRP|S_IWOTH))
__rcmd_errstr =
".rhosts writable by other than owner";
else
isvaliduser =
__ivaliduser_sa(hostf, sa, (socklen_t)rlen,
luser, ruser);
(void)fclose(hostf);
}
(void)seteuid(uid);
(void)setegid(gid);
}
return isvaliduser;
}
/*
* XXX
* Don't make static, used by lpd(8). We will be able to change the function
* into static function, when we bump libc major #.
*
* Returns 0 if ok, -1 if not ok.
*/
#ifdef notdef /*_LIBC*/
static
#endif
int
__ivaliduser(FILE *hostf, u_int32_t raddr, const char *luser,
const char *ruser)
{
struct sockaddr_in ivusin;
memset(&ivusin, 0, sizeof(ivusin));
ivusin.sin_family = AF_INET;
#ifdef BSD4_4
ivusin.sin_len = sizeof(ivusin);
#endif
memcpy(&ivusin.sin_addr, &raddr, sizeof(ivusin.sin_addr));
return __ivaliduser_sa(hostf, (struct sockaddr *)(void *)&ivusin,
(socklen_t)sizeof(ivusin), luser, ruser);
}
#ifdef notdef /*_LIBC*/
static
#endif
int
__ivaliduser_sa(FILE *hostf, const struct sockaddr *raddr, socklen_t salen,
const char *luser, const char *ruser)
{
char *user, *p;
int ch;
char buf[MAXHOSTNAMELEN + 128]; /* host + login */
const char *auser, *ahost;
int hostok, userok;
char *rhost = NULL;
int firsttime = 1;
char domain[MAXHOSTNAMELEN];
getdomainname(domain, sizeof(domain));
_DIAGASSERT(hostf != NULL);
_DIAGASSERT(luser != NULL);
_DIAGASSERT(ruser != NULL);
while (fgets(buf, (int)sizeof(buf), hostf)) {
p = buf;
/* Skip lines that are too long. */
if (strchr(p, '\n') == NULL) {
while ((ch = getc(hostf)) != '\n' && ch != EOF)
;
continue;
}
while (*p != '\n' && *p != ' ' && *p != '\t' && *p != '\0') {
*p = isupper((unsigned char)*p) ?
tolower((unsigned char)*p) : *p;
p++;
}
if (*p == ' ' || *p == '\t') {
*p++ = '\0';
while (*p == ' ' || *p == '\t')
p++;
user = p;
while (*p != '\n' && *p != ' ' &&
*p != '\t' && *p != '\0')
p++;
} else
user = p;
*p = '\0';
if (p == buf)
continue;
auser = *user ? user : luser;
ahost = buf;
if (ahost[0] == '+')
switch (ahost[1]) {
case '\0':
hostok = 1;
break;
case '@':
if (firsttime) {
rhost = __gethostloop(raddr, salen);
firsttime = 0;
}
if (rhost)
hostok = innetgr(&ahost[2], rhost,
NULL, domain);
else
hostok = 0;
break;
default:
hostok = __icheckhost(raddr, salen, &ahost[1]);
break;
}
else if (ahost[0] == '-')
switch (ahost[1]) {
case '\0':
hostok = -1;
break;
case '@':
if (firsttime) {
rhost = __gethostloop(raddr, salen);
firsttime = 0;
}
if (rhost)
hostok = -innetgr(&ahost[2], rhost,
NULL, domain);
else
hostok = 0;
break;
default:
hostok =
-__icheckhost(raddr, salen, &ahost[1]);
break;
}
else
hostok = __icheckhost(raddr, salen, ahost);
if (auser[0] == '+')
switch (auser[1]) {
case '\0':
userok = 1;
break;
case '@':
userok = innetgr(&auser[2], NULL, ruser,
domain);
break;
default:
userok = strcmp(ruser, &auser[1]) == 0;
break;
}
else if (auser[0] == '-')
switch (auser[1]) {
case '\0':
userok = -1;
break;
case '@':
userok = -innetgr(&auser[2], NULL, ruser,
domain);
break;
default:
userok =
-(strcmp(ruser, &auser[1]) == 0 ? 1 : 0);
break;
}
else
userok = strcmp(ruser, auser) == 0;
/* Check if one component did not match */
if (hostok == 0 || userok == 0)
continue;
/* Check if we got a forbidden pair */
if (userok == -1 || hostok == -1)
return -1;
/* Check if we got a valid pair */
if (hostok == 1 && userok == 1)
return 0;
}
return -1;
}
/*
* Returns "true" if match, 0 if no match.
*/
static int
__icheckhost(const struct sockaddr *raddr, socklen_t salen, const char *lhost)
{
struct addrinfo hints, *res, *r;
char h1[NI_MAXHOST], h2[NI_MAXHOST];
int error;
const int niflags = NI_NUMERICHOST;
_DIAGASSERT(raddr != NULL);
_DIAGASSERT(lhost != NULL);
h1[0] = '\0';
if (getnameinfo(raddr, salen, h1, (socklen_t)sizeof(h1), NULL, 0,
niflags) != 0)
return 0;
/* Resolve laddr into sockaddr */
memset(&hints, 0, sizeof(hints));
hints.ai_family = raddr->sa_family;
hints.ai_socktype = SOCK_DGRAM; /*dummy*/
res = NULL;
error = getaddrinfo(lhost, "0", &hints, &res);
if (error)
return 0;
/*
* Try string comparisons between raddr and laddr.
*/
for (r = res; r; r = r->ai_next) {
h2[0] = '\0';
if (getnameinfo(r->ai_addr, r->ai_addrlen, h2,
(socklen_t)sizeof(h2), NULL, 0, niflags) != 0)
continue;
if (strcmp(h1, h2) == 0) {
freeaddrinfo(res);
return 1;
}
}
/* No match. */
freeaddrinfo(res);
return 0;
}
/*
* Return the hostname associated with the supplied address.
* Do a reverse lookup as well for security. If a loop cannot
* be found, pack the numeric IP address into the string.
*/
static char *
__gethostloop(const struct sockaddr *raddr, socklen_t salen)
{
static char remotehost[NI_MAXHOST];
char h1[NI_MAXHOST], h2[NI_MAXHOST];
struct addrinfo hints, *res, *r;
int error;
const int niflags = NI_NUMERICHOST;
_DIAGASSERT(raddr != NULL);
h1[0] = remotehost[0] = '\0';
if (getnameinfo(raddr, salen, remotehost, (socklen_t)sizeof(remotehost),
NULL, 0, NI_NAMEREQD) != 0)
return NULL;
if (getnameinfo(raddr, salen, h1, (socklen_t)sizeof(h1), NULL, 0,
niflags) != 0)
return NULL;
/*
* Look up the name and check that the supplied
* address is in the list
*/
memset(&hints, 0, sizeof(hints));
hints.ai_family = raddr->sa_family;
hints.ai_socktype = SOCK_DGRAM; /*dummy*/
hints.ai_flags = AI_CANONNAME;
res = NULL;
error = getaddrinfo(remotehost, "0", &hints, &res);
if (error)
return NULL;
for (r = res; r; r = r->ai_next) {
h2[0] = '\0';
if (getnameinfo(r->ai_addr, r->ai_addrlen, h2,
(socklen_t)sizeof(h2), NULL, 0, niflags) != 0)
continue;
if (strcmp(h1, h2) == 0) {
freeaddrinfo(res);
return remotehost;
}
}
/*
* either the DNS adminstrator has made a configuration
* mistake, or someone has attempted to spoof us
*/
syslog(LOG_NOTICE, "rcmd: address %s not listed for host %s",
h1, res->ai_canonname ? res->ai_canonname : remotehost);
freeaddrinfo(res);
return NULL;
}