freebsd-nq/usr.sbin/named/ns_resp.c

2191 lines
55 KiB
C

#if !defined(lint) && !defined(SABER)
static char sccsid[] = "@(#)ns_resp.c 4.65 (Berkeley) 3/3/91";
static char rcsid[] = "$Id: ns_resp.c,v 4.9.1.24 1994/07/23 23:23:56 vixie Exp $";
#endif /* not lint */
/*
* ++Copyright++ 1986, 1988, 1990
* -
* Copyright (c) 1986, 1988, 1990
* 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. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by the University of
* California, Berkeley and its contributors.
* 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.
* -
* Portions Copyright (c) 1993 by Digital Equipment Corporation.
*
* Permission to use, copy, modify, and distribute this software for any
* purpose with or without fee is hereby granted, provided that the above
* copyright notice and this permission notice appear in all copies, and that
* the name of Digital Equipment Corporation not be used in advertising or
* publicity pertaining to distribution of the document or software without
* specific, written prior permission.
*
* THE SOFTWARE IS PROVIDED "AS IS" AND DIGITAL EQUIPMENT CORP. DISCLAIMS ALL
* WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES
* OF MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL DIGITAL EQUIPMENT
* CORPORATION BE LIABLE FOR ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL
* DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR
* PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS
* ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS
* SOFTWARE.
* -
* --Copyright--
*/
#include <sys/param.h>
#include <sys/socket.h>
#include <sys/file.h>
#include <netinet/in.h>
#include <arpa/nameser.h>
#include <arpa/inet.h>
#include <syslog.h>
#include <errno.h>
#include <stdio.h>
#include <resolv.h>
#include "named.h"
static void check_root __P((void)),
check_ns __P((void));
static u_int8_t norootlogged[MAXCLASS]; /* XXX- should be a bitmap */
static char skipnameFailedAnswer[] = "skipname failed in answer",
skipnameFailedQuery[] = "skipname failed in query",
outofDataQuery[] = "ran out of data in query",
outofDataAnswer[] = "ran out of data in answer",
#ifdef LAME_DELEGATION
expandFailedQuery[] = "dn_expand failed in query",
expandFailedAuth[] = "dn_expand failed in authority",
outofDataAuth[] = "ran out of data in authority",
#endif /* LAME_DELEGATION */
dlenOverrunAnswer[] = "dlen overrun in answer",
dlenUnderrunAnswer[] = "dlen underrun in answer",
outofDataFinal[] = "out of data in final pass",
outofDataAFinal[] = "out of data after final pass";
void
ns_resp(msg, msglen)
u_char *msg;
int msglen;
{
register struct qinfo *qp;
register HEADER *hp;
register struct qserv *qs;
register struct databuf *ns, *ns2;
register u_char *cp;
#ifdef VALIDATE
register u_char *tempcp;
struct sockaddr_in *server = &from_addr;
int *validatelist;
int lesscount;
#endif
struct sockaddr_in *nsa;
struct databuf *nsp[NSMAX], **nspp;
int i, c, n, ancount, aucount, nscount, arcount;
int old_ancount;
int type, class, dbflags;
int cname = 0; /* flag for processing cname response */
int count, founddata, foundname;
int buflen;
int newmsglen;
char name[MAXDNAME], *dname;
char *fname;
char *formerrmsg = "brain damage";
u_char newmsg[BUFSIZ];
u_char **dpp, *tp;
time_t rtrip;
struct hashbuf *htp;
struct namebuf *np;
struct netinfo *lp;
struct fwdinfo *fwd;
nameserIncr(from_addr.sin_addr, nssRcvdR);
#ifdef DATUMREFCNT
nsp[0] = NULL;
#endif
hp = (HEADER *) msg;
if ((qp = qfindid(hp->id)) == NULL ) {
dprintf(1, (ddt, "DUP? dropped (id %d)\n", ntohs(hp->id)));
nameserIncr(from_addr.sin_addr, nssRcvdDupR);
return;
}
dprintf(2, (ddt, "Response (%s %s %s) nsid=%d id=%d\n",
(qp->q_flags & Q_SYSTEM) ?"SYSTEM" :"USER",
(qp->q_flags & Q_PRIMING) ?"PRIMING" :"NORMAL",
(qp->q_flags & Q_ZSERIAL) ?"ZSERIAL" :"-",
ntohs(qp->q_nsid), ntohs(qp->q_id)));
/*
* Here we handle bad responses from servers.
* Several possibilities come to mind:
* The server is sick and returns SERVFAIL
* The server returns some garbage opcode (its sick)
* The server can't understand our query and return FORMERR
* In all these cases, we simply drop the packet and force
* a retry. This will make him look bad due to unresponsiveness.
*/
if ((hp->rcode != NOERROR && hp->rcode != NXDOMAIN)
#ifndef NCACHE
|| (hp->rcode == NXDOMAIN && !hp->aa) /* must accept this one if
* we allow negative caching
*/
#endif /*NCACHE*/
|| hp->opcode != QUERY) {
dprintf(2, (ddt, "resp: error (ret %d, op %d), dropped\n",
hp->rcode, hp->opcode));
switch (hp->rcode) {
case SERVFAIL:
nameserIncr(from_addr.sin_addr, nssRcvdFail);
break;
case FORMERR:
nameserIncr(from_addr.sin_addr, nssRcvdFErr);
break;
default:
nameserIncr(from_addr.sin_addr, nssRcvdErr);
break;
}
return;
}
#ifdef LAME_DELEGATION
/*
* Non-authoritative, no answer, no error
*/
if (hp->rcode == NOERROR && !hp->aa && ntohs(hp->ancount) == 0 &&
ntohs(hp->nscount) > 0) {
#ifdef LAME_LOGGING
char qname[MAXDNAME];
#endif /* LAME_LOGGING */
#ifdef DEBUG
if (debug > 0)
fp_query(msg, ddt);
#endif
cp = msg + HFIXEDSZ;
dpp = dnptrs;
*dpp++ = msg;
if ((*cp & INDIR_MASK) == 0)
*dpp++ = cp;
*dpp = NULL;
if (hp->qdcount) {
#ifdef LAME_LOGGING
n = dn_expand(msg, msg + msglen, cp, qname,
sizeof(qname));
if (n <= 0) {
formerrmsg = expandFailedQuery;
goto formerr;
}
#else /* LAME_LOGGING */
n = dn_skipname(cp, msg + msglen);
if (n <= 0) {
formerrmsg = skipnameFailedQuery;
goto formerr;
}
#endif /* LAME_LOGGING */
cp += n;
GETSHORT(type, cp);
GETSHORT(class, cp);
if (cp - msg > msglen) {
formerrmsg = outofDataQuery;
goto formerr;
}
#ifdef LAME_LOGGING
} else {
strcpy(qname, "[No query name!]");
#endif /* LAME_LOGGING */
}
n = dn_expand(msg, msg + msglen, cp, name, sizeof name);
if (n < 0) {
formerrmsg = expandFailedAuth;
goto formerr;
}
cp += n;
GETSHORT(type, cp);
if (cp - msg > msglen) {
formerrmsg = outofDataAuth;
goto formerr;
}
/*
* If the answer delegates us either to the same level in
* the hierarchy or closer to the root, we consider this
* server lame.
*/
if (type == T_NS && samedomain(qp->q_domain, name)) {
nameserIncr(from_addr.sin_addr, nssRcvdLDel);
#ifdef LAME_LOGGING
if (!haveComplained((char*)dhash((u_char*)name,
strlen(name)),
(char*)dhash((u_char*)qp->q_domain,
strlen(qp->q_domain)
)
)
) {
syslog(LAME_LOGGING,
"Lame delegation to '%s' from [%s] (server for '%s'?) on query on name '%s'\n",
name, inet_ntoa(from_addr.sin_addr),
qp->q_domain, qname);
}
#endif /* LAME_LOGGING */
return;
}
}
#endif /* LAME_DELEGATION */
#ifdef ALLOW_UPDATES
if ( (hp->rcode == NOERROR) &&
(hp->opcode == UPDATEA || hp->opcode == UPDATED ||
hp->opcode == UPDATEDA || hp->opcode == UPDATEM ||
hp->opcode == UPDATEMA) ) {
/*
* Update the secondary's copy, now that the primary
* successfully completed the update. Zone doesn't matter
* for dyn. update -- doupdate calls findzone to find it
*/
/* XXX - DB_C_AUTH may be wrong */
(void) doupdate(qp->q_msg, qp->q_msglen, qp->q_msg + HFIXEDSZ,
0, (struct databuf *)0, 0, DB_C_AUTH);
dprintf(3, (ddt, "resp: leaving, UPDATE*\n"));
/* return code filled in by doupdate */
goto return_msg;
}
#endif /* ALLOW_UPDATES */
/*
* Determine if the response came from a forwarder. Packets from
* anyplace not listed as a forwarder or as a server to whom we
* might have forwarded the query will be dropped.
*/
for (fwd = fwdtab; fwd != (struct fwdinfo *)NULL; fwd = fwd->next) {
if (fwd->fwdaddr.sin_addr.s_addr ==
from_addr.sin_addr.s_addr) {
/* XXX - should put this in STATS somewhere */
break;
}
}
/* XXX: note bad ambiguity here. if one of our forwarders is also
* a delegated server for some domain, then we will not update
* the RTT information on any replies we get from those servers.
*/
/*
* If we were using nameservers, find the qinfo pointer and update
* the rtt and fact that we have called on this server before.
*/
if (fwd == (struct fwdinfo *)NULL) {
struct timeval *stp;
for (n = 0, qs = qp->q_addr; n < qp->q_naddr; n++, qs++)
if (qs->ns_addr.sin_addr.s_addr ==
from_addr.sin_addr.s_addr)
break;
if (n >= qp->q_naddr) {
if (!haveComplained((char*)from_addr.sin_addr.s_addr,
"unexpected source")) {
syslog(LOG_NOTICE,
"Response from unexpected source [%s].%d",
inet_ntoa(from_addr.sin_addr),
ntohs(from_addr.sin_port));
}
/*
* We don't know who this response came from so it
* gets dropped on the floor.
*/
return;
}
stp = &qs->stime;
/* Handle response from different (untried) interface */
if ((qs->ns != NULL) && (stp->tv_sec == 0)) {
ns = qs->ns;
while (qs > qp->q_addr &&
(qs->stime.tv_sec == 0 || qs->ns != ns))
qs--;
*stp = qs->stime;
/* XXX - sometimes stp still ends up pointing to
* a zero timeval, in spite of the above attempt.
* Why? What should we do about it?
*/
dprintf(1, (ddt,
"Response from unused address %s, assuming %s\n",
inet_ntoa(from_addr.sin_addr),
inet_ntoa(qs->ns_addr.sin_addr)));
/* XXX - catch aliases here */
}
/* compute query round trip time */
/* XXX - avoid integer overflow, which is quite likely if stp
* points to a zero timeval (see above).
* rtrip is of type time_t, which we assume is at least
* as big as an int.
*/
if ((tt.tv_sec - stp->tv_sec) > (INT_MAX-999)/1000) {
rtrip = INT_MAX;
} else {
rtrip = ((tt.tv_sec - stp->tv_sec) * 1000 +
(tt.tv_usec - stp->tv_usec) / 1000);
}
dprintf(3, (ddt, "stime %d/%d now %d/%d rtt %d\n",
stp->tv_sec, stp->tv_usec,
tt.tv_sec, tt.tv_usec, rtrip));
/* prevent floating point overflow, limit to 1000 sec */
if (rtrip > 1000000) {
rtrip = 1000000;
}
ns = qs->nsdata;
/*
* Don't update nstime if this doesn't look
* like an address databuf now. XXX
*/
if (ns && (ns->d_type==T_A) && (ns->d_class==qs->ns->d_class)){
if (ns->d_nstime == 0)
ns->d_nstime = (u_int32_t)rtrip;
else
ns->d_nstime = (u_int32_t)
(ns->d_nstime * ALPHA
+
(1-ALPHA) * (u_int32_t)rtrip);
/* prevent floating point overflow,
* limit to 1000 sec
*/
if (ns->d_nstime > 1000000)
ns->d_nstime = 1000000;
}
/*
* Record the source so that we do not use this NS again.
*/
if (ns && qs->ns && (qp->q_nusedns < NSMAX)) {
qp->q_usedns[qp->q_nusedns++] = qs->ns;
dprintf(2, (ddt, "NS #%d addr [%s] used, rtt %d\n",
n, inet_ntoa(qs->ns_addr.sin_addr),
ns->d_nstime));
}
/*
* Penalize those who had earlier chances but failed
* by multiplying round-trip times by BETA (>1).
* Improve nstime for unused addresses by applying GAMMA.
* The GAMMA factor makes unused entries slowly
* improve, so they eventually get tried again.
* GAMMA should be slightly less than 1.
* Watch out for records that may have timed out
* and are no longer the correct type. XXX
*/
for (n = 0, qs = qp->q_addr; n < qp->q_naddr; n++, qs++) {
ns2 = qs->nsdata;
if ((!ns2) || (ns2 == ns))
continue;
if (ns2->d_type != T_A ||
ns2->d_class != qs->ns->d_class) /* XXX */
continue;
if (qs->stime.tv_sec) {
if (ns2->d_nstime == 0)
ns2->d_nstime = (u_int32_t)(rtrip * BETA);
else
ns2->d_nstime = (u_int32_t)(
ns2->d_nstime * BETA + (1-ALPHA) * rtrip
);
if (ns2->d_nstime > 1000000)
ns2->d_nstime = 1000000;
} else
ns2->d_nstime = (u_int32_t)(ns2->d_nstime * GAMMA);
dprintf(2, (ddt, "NS #%d [%s] rtt now %d\n", n,
inet_ntoa(qs->ns_addr.sin_addr),
ns2->d_nstime));
}
}
/*************************************************************/
/*
* Skip query section
*/
free_addinfo(); /* sets addcount to zero */
cp = msg + HFIXEDSZ;
dpp = dnptrs;
*dpp++ = msg;
if ((*cp & INDIR_MASK) == 0)
*dpp++ = cp;
*dpp = NULL;
type = class = 0;
if (hp->qdcount) {
n = dn_skipname(cp, msg + msglen);
if (n <= 0) {
formerrmsg = skipnameFailedQuery;
goto formerr;
}
cp += n;
GETSHORT(type, cp);
GETSHORT(class, cp);
if (cp - msg > msglen) {
formerrmsg = outofDataQuery;
goto formerr;
}
}
/*
* Save answers, authority, and additional records for future use.
*/
ancount = ntohs(hp->ancount);
aucount = ntohs(hp->nscount);
arcount = ntohs(hp->arcount);
nscount = 0;
tp = cp;
dprintf(3, (ddt, "resp: ancount %d, aucount %d, arcount %d\n",
ancount, aucount, arcount));
/*
* If there's an answer, check if it's a CNAME response;
* if no answer but aucount > 0, see if there is an NS
* or just an SOA. (NOTE: ancount might be 1 with a CNAME,
* and NS records may still be in the authority section;
* we don't bother counting them, as we only use nscount
* if ancount == 0.)
*/
if (ancount == 1 || (ancount == 0 && aucount > 0)) {
c = aucount;
do {
if (tp - msg >= msglen) {
formerrmsg = outofDataAnswer;
goto formerr;
}
n = dn_skipname(tp, msg + msglen);
if (n <= 0) {
formerrmsg = skipnameFailedAnswer;
goto formerr;
}
tp += n; /* name */
GETSHORT(i, tp); /* type */
tp += INT16SZ; /* class */
tp += INT32SZ; /* ttl */
GETSHORT(count, tp); /* dlen */
if (tp - msg > msglen - count) {
formerrmsg = dlenOverrunAnswer;
goto formerr;
}
tp += count;
if (ancount && i == T_CNAME) {
cname++;
dprintf(1,
(ddt,
"CNAME - needs more processing\n"
)
);
if (!qp->q_cmsglen) {
qp->q_cmsg = qp->q_msg;
qp->q_cmsglen = qp->q_msglen;
qp->q_msg = NULL;
qp->q_msglen = 0;
}
}
/*
* See if authority record is a nameserver.
*/
if (ancount == 0 && i == T_NS)
nscount++;
} while (--c > 0);
tp = cp;
}
if (qp->q_flags & Q_ZSERIAL) {
if ((hp->aa)
&& (ancount != 0)
&& (hp->rcode == NOERROR)
&& (type == T_SOA)
&& ((class == C_IN) || (class == C_HS))
) { /* XXX - should check name, too */
int n;
u_int16_t dlen;
u_int32_t serial;
u_char *tp = cp;
if (0 >= (n = dn_skipname(tp, msg + msglen))) {
formerrmsg = skipnameFailedAnswer;
goto formerr;
}
tp += n /* name */
+ INT16SZ /* type */
+ INT16SZ /* class */
+ INT32SZ; /* ttl */
GETSHORT(dlen, tp); /* dlen */
if (dlen < (5 * INT32SZ)) {
formerrmsg = dlenUnderrunAnswer;
goto formerr;
}
if (0 >= (n = dn_skipname(tp, msg + msglen))) {
formerrmsg = skipnameFailedAnswer;
goto formerr;
}
tp += n; /* mname */
if (0 >= (n = dn_skipname(tp, msg + msglen))) {
formerrmsg = skipnameFailedAnswer;
goto formerr;
}
tp += n; /* rname */
GETLONG(serial, tp);
qserial_answer(qp, serial);
}
qremove(qp);
return;
}
/*
* Add the info received in the response to the data base.
*/
c = ancount + aucount + arcount;
#ifdef NCACHE
/* -ve $ing non-existence of record, must handle non-authoritative
* NOERRORs with c == 0.
*/
if (!hp->aa && hp->rcode == NOERROR && c == 0) {
goto return_msg;
} /*should ideally validate message before returning it*/
#endif /*NCACHE*/
#ifdef notdef
/*
* If the request was for a CNAME that doesn't exist,
* but the name is valid, fetch any other data for the name.
* DON'T do this now, as it will requery if data are already
* in the cache (maybe later with negative caching).
*/
if (hp->qdcount && type == T_CNAME && c == 0 && hp->rcode == NOERROR
&& !(qp->q_flags & Q_SYSTEM)) {
dprintf(4, (ddt, "resp: leaving, no CNAME\n"));
/* Cause us to put it in the cache later */
prime(class, T_ANY, qp);
/* Nothing to store, just give user the answer */
goto return_msg;
}
#endif /* notdef */
nspp = nsp;
if (qp->q_flags & Q_SYSTEM)
dbflags = DB_NOTAUTH | DB_NODATA;
else
dbflags = DB_NOTAUTH | DB_NODATA | DB_NOHINTS;
count = c;
if (hp->tc) {
count -= arcount; /* truncation had to affect this */
if (!arcount) {
count -= aucount; /* guess it got this too */
}
if (!(arcount || aucount)) {
count -= ancount; /* things are pretty grim */
}
/* XXX - should retry this query with TCP */
}
#ifdef VALIDATE
tempcp = cp;
validatelist = (int *)malloc(count * sizeof(int));
lesscount = 0; /*initialize*/
old_ancount = ancount;
for (i = 0; i < count; i++) {
int VCode;
if (tempcp >= msg + msglen) {
free((char *)validatelist);
formerrmsg = outofDataFinal;
goto formerr;
}
if ((n = dovalidate(msg, msglen, tempcp, 0,
dbflags, server, &VCode)) < 0) {
dprintf(1, (ddt,
"resp: leaving, dovalidate failed\n"));
free((char *)validatelist);
/* return code filled in by dovalidate */
goto return_msg;
}
validatelist[i] = VCode;
if (VCode == INVALID) lesscount++;
tempcp += n;
}
/* need to delete INVALID records from the message
* and change fields appropriately
*/
n = update_msg(msg, &msglen, validatelist, count);
free((char *)validatelist);
if (n < 0)
goto formerr;
count -= lesscount;
if (old_ancount && !hp->ancount) {
/* We lost all the answers */
dprintf(1, (ddt, "validate count -> 0"));
return;
}
ancount = ntohs(hp->ancount);
#endif
for (i = 0; i < count; i++) {
struct databuf *ns3;
u_char cred;
if (cp >= msg + msglen) {
formerrmsg = outofDataFinal;
goto formerr;
}
if (i < ancount) {
cred = hp->aa ? DB_C_AUTH : DB_C_ANSWER;
} else {
cred = DB_C_ADDITIONAL;
}
ns3 = 0;
n = doupdate(msg, msglen, cp, 0, &ns3, dbflags, cred);
if (n < 0) {
dprintf(1, (ddt, "resp: leaving, doupdate failed\n"));
/* return code filled in by doupdate */
goto return_msg;
}
/*
* Remember nameservers from the authority section
* for referrals.
* (This is usually overwritten by findns below(?). XXX
*/
if (ns3 && i >= ancount && i < ancount + aucount &&
nspp < &nsp[NSMAX-1]) {
*nspp++ = ns3;
#ifdef DATUMREFCNT
ns3->d_rcnt++;
*nspp = NULL;
#endif
}
cp += n;
}
if ((qp->q_flags & Q_SYSTEM) && ancount) {
if (qp->q_flags & Q_PRIMING)
check_root();
dprintf(3, (ddt, "resp: leaving, SYSQUERY ancount %d\n",
ancount));
qremove(qp);
#ifdef DATUMREFCNT
free_nsp(nsp);
#endif
return;
}
if (cp > msg + msglen) {
formerrmsg = outofDataAFinal;
goto formerr;
}
/*
* If there are addresses and this is a local query,
* sort them appropriately for the local context.
*/
if (ancount > 1 && (lp = local(&qp->q_from)) != NULL)
sort_response(tp, ancount, lp, msg + msglen);
/*
* An answer to a T_ANY query or a successful answer to a
* regular query with no indirection, then just return answer.
*/
if ((hp->qdcount && type == T_ANY && ancount) ||
(!cname && !qp->q_cmsglen && ancount)) {
dprintf(3, (ddt, "resp: got as much answer as there is\n"));
goto return_msg;
}
/*
* We might want to cache this negative answer.
*/
if (!ancount &&
(!nscount || hp->rcode == NXDOMAIN) &&
(hp->aa || fwd || class == C_ANY)) {
/* we have an authoritative NO */
dprintf(3, (ddt, "resp: leaving auth NO\n"));
if (qp->q_cmsglen) {
msg = qp->q_cmsg;
msglen = qp->q_cmsglen;
hp = (HEADER *)msg;
}
#ifdef NCACHE
/* answer was NO */
if (hp->aa &&
((hp->rcode == NXDOMAIN) || (hp->rcode == NOERROR))) {
cache_n_resp(msg, msglen);
}
#endif /*NCACHE*/
goto return_msg;
}
/*
* All messages in here need further processing. i.e. they
* are either CNAMEs or we got referred again.
*/
count = 0;
founddata = 0;
foundname = 0;
dname = name;
if (!cname && qp->q_cmsglen && ancount) {
dprintf(1, (ddt, "Cname second pass\n"));
newmsglen = qp->q_cmsglen;
bcopy(qp->q_cmsg, newmsg, newmsglen);
} else {
newmsglen = msglen;
bcopy(msg, newmsg, newmsglen);
}
hp = (HEADER *) newmsg;
hp->ancount = 0;
hp->nscount = 0;
hp->arcount = 0;
dnptrs[0] = newmsg;
dnptrs[1] = NULL;
cp = newmsg + HFIXEDSZ;
if (cname)
cp += dn_skipname(cp, newmsg + newmsglen) + QFIXEDSZ;
n = dn_expand(newmsg, newmsg + newmsglen, cp, dname, sizeof name);
if (n < 0) {
dprintf(1, (ddt, "dn_expand failed\n"));
goto servfail;
}
if (!cname)
cp += n + QFIXEDSZ;
buflen = sizeof(newmsg) - (cp - newmsg);
try_again:
dprintf(1, (ddt, "resp: nlookup(%s) type=%d\n", dname, type));
fname = "";
htp = hashtab; /* lookup relative to root */
np = nlookup(dname, &htp, &fname, 0);
dprintf(1, (ddt, "resp: %s '%s' as '%s' (cname=%d)\n",
np == NULL ? "missed" : "found", dname, fname, cname));
if (np == NULL || fname != dname)
goto fetch_ns;
foundname++;
count = cp - newmsg;
n = finddata(np, class, type, hp, &dname, &buflen, &count);
if (n == 0)
goto fetch_ns; /* NO data available */
cp += n;
buflen -= n;
hp->ancount += count;
if (fname != dname && type != T_CNAME && type != T_ANY) {
cname++;
goto try_again;
}
founddata = 1;
dprintf(3, (ddt,
"resp: foundname=%d, count=%d, founddata=%d, cname=%d\n",
foundname, count, founddata, cname));
fetch_ns:
hp->ancount = htons(hp->ancount);
/*
* Look for name servers to refer to and fill in the authority
* section or record the address for forwarding the query
* (recursion desired).
*/
#ifdef DATUMREFCNT
free_nsp(nsp);
#endif
switch (findns(&np, class, nsp, &count, 0)) {
case NXDOMAIN: /* shouldn't happen */
dprintf(3, (ddt, "req: leaving (%s, rcode %d)\n",
dname, hp->rcode));
if (!foundname)
hp->rcode = NXDOMAIN;
if (class != C_ANY) {
hp->aa = 1;
/* XXX: should return SOA if founddata == 0,
* but old named's are confused by an SOA
* in the auth. section if there's no error.
*/
if (foundname == 0 && np) {
n = doaddauth(hp, cp, buflen, np, nsp[0]);
cp += n;
buflen -= n;
}
}
goto return_newmsg;
case SERVFAIL:
goto servfail;
}
if (founddata) {
hp = (HEADER *)newmsg;
n = add_data(np, nsp, cp, buflen);
if (n < 0) {
hp->tc = 1;
n = (-n);
}
cp += n;
buflen -= n;
hp->nscount = htons((u_int16_t)count);
goto return_newmsg;
}
/*
* If we get here, we don't have the answer yet and are about
* to iterate to try and get it. First, infinite loop avoidance.
*/
if (qp->q_nqueries++ > MAXQUERIES) {
dprintf(1,
(ddt,
"resp: MAXQUERIES exceeded (%s, class %d, type %d)\n",
dname, class, type
)
);
syslog(LOG_NOTICE,
"MAXQUERIES exceeded, possible data loop in resolving (%s)",
dname);
goto servfail;
}
/* Reset the query control structure */
#ifdef DATUMREFCNT
for (i = 0 ; i < qp->q_naddr ; i++) {
if ((--(qp->q_addr[i].ns->d_rcnt))) {
dprintf(1 ,(ddt, "ns_resp: ns %s rcnt %d\n",
qp->q_addr[i].ns->d_data,
qp->q_addr[i].ns->d_rcnt));
} else {
dprintf(1 ,(ddt, "ns_resp: ns %s rcnt %d delayed\n",
qp->q_addr[i].ns->d_data,
qp->q_addr[i].ns->d_rcnt));
free((char*)qp->q_addr[i].ns);
}
if ((--(qp->q_addr[i].nsdata->d_rcnt))) {
dprintf(1 ,(ddt, "ns_resp: nsdata %08.8X rcnt %d\n",
*(int32_t *)(qp->q_addr[i].nsdata->d_data),
qp->q_addr[i].nsdata->d_rcnt));
} else {
dprintf(1 ,(ddt, "ns_resp: nsdata %08.8X rcnt %d delayed\n",
*(int32_t *)(qp->q_addr[i].nsdata->d_data),
qp->q_addr[i].nsdata->d_rcnt));
free((char*)qp->q_addr[i].nsdata);
}
}
#endif
qp->q_naddr = 0;
qp->q_curaddr = 0;
qp->q_fwd = fwdtab;
#ifdef LAME_DELEGATION
getname(np, qp->q_domain, sizeof(qp->q_domain));
#endif /* LAME_DELEGATION */
if ((n = nslookup(nsp, qp, dname, "ns_resp")) <= 0) {
if (n < 0) {
dprintf(3, (ddt, "resp: nslookup reports danger\n"));
} else {
dprintf(3, (ddt, "resp: no addrs found for NS's\n"));
}
if (cname) /* a remote CNAME that does not have data */
goto return_newmsg;
goto servfail;
}
for (n = 0; n < qp->q_naddr; n++)
qp->q_addr[n].stime.tv_sec = 0;
if (!qp->q_fwd)
qp->q_addr[0].stime = tt;
if (cname) {
if (qp->q_cname++ == MAXCNAMES) {
dprintf(3, (ddt,
"resp: leaving, MAXCNAMES exceeded\n"));
goto servfail;
}
dprintf(1, (ddt, "q_cname = %d\n",qp->q_cname));
dprintf(3, (ddt,
"resp: building recursive query; nslookup\n"));
if (qp->q_msg)
(void) free(qp->q_msg);
if ((qp->q_msg = (u_char *)malloc(BUFSIZ)) == NULL) {
dprintf(1, (ddt, "resp: malloc error\n"));
goto servfail;
}
qp->q_msglen = res_mkquery(QUERY, dname, class,
type, NULL, 0, NULL,
qp->q_msg, BUFSIZ);
hp = (HEADER *) qp->q_msg;
hp->rd = 0;
} else
hp = (HEADER *)qp->q_msg;
hp->id = qp->q_nsid = htons((u_int16_t)++nsid);
if (qp->q_fwd)
hp->rd = 1;
unsched(qp);
schedretry(qp, retrytime(qp));
nsa = Q_NEXTADDR(qp, 0);
dprintf(1, (ddt, "resp: forw -> [%s].%d ds=%d nsid=%d id=%d %dms\n",
inet_ntoa(nsa->sin_addr),
ntohs(nsa->sin_port), ds,
ntohs(qp->q_nsid), ntohs(qp->q_id),
(qp->q_addr[0].nsdata != NULL)
? qp->q_addr[0].nsdata->d_nstime
: (-1)));
#ifdef DEBUG
if (debug >= 10)
fp_query(msg, ddt);
#endif
if (sendto(ds, qp->q_msg, qp->q_msglen, 0,
(struct sockaddr *)nsa,
sizeof(struct sockaddr_in)) < 0) {
if (!haveComplained((char*)nsa->sin_addr.s_addr, sendtoStr))
syslog(LOG_NOTICE, "ns_resp: sendto([%s].%d): %m",
inet_ntoa(nsa->sin_addr), ntohs(nsa->sin_port));
nameserIncr(nsa->sin_addr, nssSendtoErr);
}
hp->rd = 0; /* leave set to 0 for dup detection */
nameserIncr(nsa->sin_addr, nssSentFwdR);
nameserIncr(qp->q_from.sin_addr, nssRcvdFwdR);
dprintf(3, (ddt, "resp: Query sent.\n"));
#ifdef DATUMREFCNT
free_nsp(nsp);
#endif
return;
formerr:
dprintf(3, (ddt,
"FORMERR resp() from [%s].%d size err %d, msglen %d\n",
inet_ntoa(from_addr.sin_addr),
ntohs(from_addr.sin_port),
cp - msg, msglen));
if (!haveComplained((char*)from_addr.sin_addr.s_addr,
(char*)dhash((u_char *)formerrmsg,
strlen(formerrmsg)
)
)
) {
syslog(LOG_INFO, "Malformed response from [%s].%d (%s)\n",
inet_ntoa(from_addr.sin_addr),
ntohs(from_addr.sin_port),
formerrmsg);
}
nameserIncr(from_addr.sin_addr, nssSentFErr);
#ifdef DATUMREFCNT
free_nsp(nsp);
#endif
return;
return_msg:
nameserIncr(from_addr.sin_addr, nssRcvdFwdR);
nameserIncr(qp->q_from.sin_addr, nssSentFwdR);
/* The "standard" return code */
hp->qr = 1;
hp->id = qp->q_id;
hp->rd = 1;
hp->ra = 1;
(void) send_msg(msg, msglen, qp);
qremove(qp);
#ifdef DATUMREFCNT
free_nsp(nsp);
#endif
return;
return_newmsg:
nameserIncr(qp->q_from.sin_addr, nssSentAns);
if (addcount) {
n = doaddinfo(hp, cp, buflen);
cp += n;
buflen -= n;
}
hp->qr = 1;
hp->id = qp->q_id;
hp->rd = 1;
hp->ra = 1;
(void) send_msg(newmsg, cp - newmsg, qp);
qremove(qp);
#ifdef DATUMREFCNT
free_nsp(nsp);
#endif
return;
servfail:
nameserIncr(qp->q_from.sin_addr, nssSentFail);
hp = (HEADER *)(cname ? qp->q_cmsg : qp->q_msg);
hp->rcode = SERVFAIL;
hp->id = qp->q_id;
hp->rd = 1;
hp->ra = 1;
hp->qr = 1;
(void) send_msg((u_char *)hp, (cname ? qp->q_cmsglen : qp->q_msglen),
qp);
qremove(qp);
#ifdef DATUMREFCNT
free_nsp(nsp);
#endif
return;
}
/*
* Decode the resource record 'rrp' and update the database.
* If savens is non-nil, record pointer for forwarding queries a second time.
*/
int
doupdate(msg, msglen, rrp, zone, savens, flags, cred)
u_char *msg, *rrp;
struct databuf **savens;
int msglen, zone, flags;
u_int cred;
{
register u_char *cp;
register int n;
int class, type, dlen, n1;
u_int32_t ttl;
struct databuf *dp;
char dname[MAXDNAME];
u_char *cp1;
u_char data[BUFSIZ];
register HEADER *hp = (HEADER *)msg;
#ifdef ALLOW_UPDATES
int zonenum;
#endif
dprintf(3, (ddt, "doupdate(zone %d, savens %x, flags %x)\n",
zone, savens, flags));
cp = rrp;
if ((n = dn_expand(msg, msg + msglen, cp, dname, sizeof dname)) < 0) {
hp->rcode = FORMERR;
return (-1);
}
cp += n;
GETSHORT(type, cp);
GETSHORT(class, cp);
GETLONG(ttl, cp);
GETSHORT(dlen, cp);
dprintf(3, (ddt, "doupdate: dname %s type %d class %d ttl %d\n",
dname, type, class, ttl));
/*
* Convert the resource record data into the internal
* database format.
*/
switch (type) {
case T_A:
if (dlen != INT32SZ) {
hp->rcode = FORMERR;
return (-1);
}
/*FALLTHROUGH*/
case T_WKS:
case T_HINFO:
case T_UINFO:
case T_UID:
case T_GID:
case T_TXT:
case T_X25:
case T_ISDN:
case T_NSAP:
#ifdef ALLOW_T_UNSPEC
case T_UNSPEC:
#endif
cp1 = cp;
n = dlen;
cp += n;
break;
case T_CNAME:
case T_MB:
case T_MG:
case T_MR:
case T_NS:
case T_PTR:
n = dn_expand(msg, msg + msglen, cp,
(char *)data, sizeof data);
if (n < 0) {
hp->rcode = FORMERR;
return (-1);
}
cp += n;
cp1 = data;
n = strlen((char *)data) + 1;
break;
case T_MINFO:
case T_SOA:
case T_RP:
n = dn_expand(msg, msg + msglen, cp,
(char *)data, sizeof data);
if (n < 0) {
hp->rcode = FORMERR;
return (-1);
}
cp += n;
cp1 = data + (n = strlen((char *)data) + 1);
n1 = sizeof(data) - n;
if (type == T_SOA)
n1 -= 5 * INT32SZ;
n = dn_expand(msg, msg + msglen, cp, (char *)cp1, n1);
if (n < 0) {
hp->rcode = FORMERR;
return (-1);
}
cp += n;
cp1 += strlen((char *)cp1) + 1;
if (type == T_SOA) {
bcopy(cp, cp1, n = 5 * INT32SZ);
cp += n;
cp1 += n;
}
n = cp1 - data;
cp1 = data;
break;
case T_MX:
case T_AFSDB:
case T_RT:
/* grab preference */
bcopy(cp, data, INT16SZ);
cp1 = data + INT16SZ;
cp += INT16SZ;
/* get name */
n = dn_expand(msg, msg + msglen, cp, (char *)cp1,
sizeof data - INT16SZ);
if (n < 0) {
hp->rcode = FORMERR;
return (-1);
}
cp += n;
/* compute end of data */
cp1 += strlen((char *)cp1) + 1;
/* compute size of data */
n = cp1 - data;
cp1 = data;
break;
default:
dprintf(3, (ddt, "unknown type %d\n", type));
return ((cp - rrp) + dlen);
}
if (n > MAXDATA) {
dprintf(1, (ddt,
"update type %d: %d bytes is too much data\n",
type, n));
hp->rcode = NOCHANGE; /* XXX - FORMERR ??? */
return (-1);
}
#ifdef ALLOW_UPDATES
/*
* If this is a dynamic update request, process it specially; else,
* execute normal update code.
*/
switch(hp->opcode) {
/* For UPDATEM and UPDATEMA, do UPDATED/UPDATEDA followed by UPDATEA */
case UPDATEM:
case UPDATEMA:
/*
* The named code for UPDATED and UPDATEDA is the same except that for
* UPDATEDA we we ignore any data that was passed: we just delete all
* RRs whose name, type, and class matches
*/
case UPDATED:
case UPDATEDA:
if (type == T_SOA) { /* Not allowed */
dprintf(1, (ddt, "UDPATE: REFUSED - SOA delete\n"));
hp->rcode = REFUSED;
return (-1);
}
/*
* Don't check message length if doing UPDATEM/UPDATEMA,
* since the whole message wont have been demarshalled until
* we reach the code for UPDATEA
*/
if ( (hp->opcode == UPDATED) || (hp->opcode == UPDATEDA) ) {
if (cp != (u_char *)(msg + msglen)) {
dprintf(1,
(ddt,
"FORMERR UPDATE message length off\n"
)
);
hp->rcode = FORMERR;
return (-1);
}
}
if ((zonenum = findzone(dname, class)) == 0) {
hp->rcode = NXDOMAIN;
return (-1);
}
if (zones[zonenum].z_flags & Z_DYNADDONLY) {
hp->rcode = NXDOMAIN;
return (-1);
}
if ( (hp->opcode == UPDATED) || (hp->opcode == UPDATEM) ) {
/* Make a dp for use in db_update, as old dp */
dp = savedata(class, type, 0, cp1, n);
dp->d_zone = zonenum;
dp->d_cred = cred;
dp->d_clev = db_getclev(zones[zonenum].z_origin);
n = db_update(dname, dp, NULL, DB_MEXIST | DB_DELETE,
hashtab);
if (n != OK) {
dprintf(1, (ddt,
"UPDATE: db_update failed\n"));
free((char*) dp);
hp->rcode = NOCHANGE;
return (-1);
}
} else { /* UPDATEDA or UPDATEMA */
int DeletedOne = 0;
/* Make a dp for use in db_update, as old dp */
dp = savedata(class, type, 0, NULL, 0);
dp->d_zone = zonenum;
dp->d_cred = cred;
dp->d_clev = db_getclev(zones[zonenum].z_origin);
do { /* Loop and delete all matching RR(s) */
n = db_update(dname, dp, NULL, DB_DELETE,
hashtab);
if (n != OK)
break;
DeletedOne++;
} while (1);
free((char*) dp);
/* Ok for UPDATEMA not to have deleted any RRs */
if (!DeletedOne && hp->opcode == UPDATEDA) {
dprintf(1, (ddt,
"UPDATE: db_update failed\n"));
hp->rcode = NOCHANGE;
return (-1);
}
}
if ( (hp->opcode == UPDATED) || (hp->opcode == UPDATEDA) )
return (cp - rrp);;
/*
* Else unmarshal the RR to be added and continue on to
* UPDATEA code for UPDATEM/UPDATEMA
*/
if ((n =
dn_expand(msg, msg+msglen, cp, dname, sizeof(dname))) < 0) {
dprintf(1, (ddt,
"FORMERR UPDATE expand name failed\n"));
hp->rcode = FORMERR;
return (-1);
}
cp += n;
GETSHORT(type, cp);
GETSHORT(class, cp);
GETLONG(ttl, cp);
GETSHORT(n, cp);
cp1 = cp;
/**** XXX - need bounds checking here ****/
cp += n;
case UPDATEA:
if (n > MAXDATA) {
dprintf(1, (ddt, "UPDATE: too much data\n"));
hp->rcode = NOCHANGE;
return (-1);
}
if (cp != (u_char *)(msg + msglen)) {
dprintf(1, (ddt,
"FORMERR UPDATE message length off\n"));
hp->rcode = FORMERR;
return (-1);
}
if ((zonenum = findzone(dname, class)) == 0) {
hp->rcode = NXDOMAIN;
return (-1);
}
if (zones[zonenum].z_flags & Z_DYNADDONLY) {
struct hashbuf *htp = hashtab;
char *fname;
if (nlookup(dname, &htp, &fname, 0) &&
!strcasecmp(dname, fname)) {
dprintf(1, (ddt,
"refusing add of existing name\n"
));
hp->rcode = REFUSED;
return (-1);
}
}
dp = savedata(class, type, ttl, cp1, n);
dp->d_zone = zonenum;
dp->d_cred = cred;
dp->d_clev = db_getclev(zones[zonenum].z_origin);
if ((n = db_update(dname, NULL, dp, DB_NODATA,
hashtab)) != OK) {
dprintf(1, (ddt, "UPDATE: db_update failed\n"));
hp->rcode = NOCHANGE;
free((char*) dp);
return (-1);
}
else
return (cp - rrp);
}
#endif /* ALLOW_UPDATES */
if (zone == 0)
ttl += tt.tv_sec;
#if defined(TRACEROOT) || defined(BOGUSNS)
if ((type == T_NS) && (savens != NULL)) {
char qname[MAXDNAME], *temp;
register int bogus = 0;
int bogusns = 0;
#ifdef BOGUSNS
if (addr_on_netlist(from_addr.sin_addr, boglist)) {
bogusns++;
bogus++;
}
#endif
if (!bogus &&
((temp = strrchr((char *)data, '.')) != NULL) &&
!strcasecmp(temp, ".arpa")
)
bogus++;
qname[0] = qname[1] = '\0';
if (dn_expand(msg, msg + msglen, msg + HFIXEDSZ,
qname, sizeof(qname)) < 0)
qname[0] = '?';
else if (qname[0] == '\0')
qname[0] = '.';
if (bogus && ((dname[0] == '\0') && (zone == 0))) {
if (!haveComplained((char*)from_addr.sin_addr.s_addr,
"bogus root NS"))
syslog(LOG_NOTICE,
"bogus root NS %s rcvd from [%s] on query for \"%s\"",
data, inet_ntoa(from_addr.sin_addr),
qname);
return (cp - rrp);
}
#ifdef BOGUSNS
if (bogusns) {
if (!haveComplained((char*)from_addr.sin_addr.s_addr,
"bogus nonroot NS"))
syslog(LOG_NOTICE,
"bogus nonroot NS %s rcvd from [%s] on query for \"%s\"",
data, inet_ntoa(from_addr.sin_addr),
qname);
return (cp - rrp);
}
#endif
}
#endif /*TRACEROOT || BOGUSNS*/
dp = savedata(class, type, ttl, cp1, n);
dp->d_zone = zone;
dp->d_cred = cred;
dp->d_clev = 0; /* We trust what is on disk more, except root srvrs */
if ((n = db_update(dname, dp, dp, flags, hashtab)) != OK) {
#ifdef DEBUG
if (debug && (n != DATAEXISTS))
fprintf(ddt,"update failed (%d)\n", n);
else if (debug >= 3)
fprintf(ddt,"update failed (DATAEXISTS)\n");
#endif
free((char *)dp);
} else if (type == T_NS && savens != NULL)
*savens = dp;
return (cp - rrp);
}
int
send_msg(msg, msglen, qp)
u_char *msg;
int msglen;
struct qinfo *qp;
{
if (qp->q_flags & Q_SYSTEM)
return (1);
#ifdef DEBUG
if (debug) {
fprintf(ddt,"send_msg -> [%s] (%s %d %d) id=%d\n",
inet_ntoa(qp->q_from.sin_addr),
qp->q_stream == QSTREAM_NULL ? "UDP" : "TCP",
qp->q_stream == QSTREAM_NULL ? qp->q_dfd
: qp->q_stream->s_rfd,
ntohs(qp->q_from.sin_port),
ntohs(qp->q_id));
}
if (debug>4) {
struct qinfo *tqp;
for (tqp = qhead; tqp!=QINFO_NULL; tqp = tqp->q_link) {
fprintf(ddt, "qp %x q_id: %d q_nsid: %d q_msglen: %d ",
tqp, tqp->q_id,tqp->q_nsid,tqp->q_msglen);
fprintf(ddt,"q_naddr: %d q_curaddr: %d\n", tqp->q_naddr,
tqp->q_curaddr);
fprintf(ddt,"q_next: %x q_link: %x\n", qp->q_next,
qp->q_link);
}
}
if (debug >= 10)
fp_query(msg, ddt);
#endif /* DEBUG */
if (qp->q_stream == QSTREAM_NULL) {
if (sendto(qp->q_dfd, msg, msglen, 0,
(struct sockaddr *)&qp->q_from,
sizeof(qp->q_from)) < 0) {
if (!haveComplained((char*)qp->q_from.sin_addr.s_addr,
sendtoStr))
syslog(LOG_NOTICE,
"send_msg: sendto([%s].%d): %m",
inet_ntoa(qp->q_from.sin_addr),
ntohs(qp->q_from.sin_port));
nameserIncr(qp->q_from.sin_addr, nssSendtoErr);
return (1);
}
} else {
(void) writemsg(qp->q_stream->s_rfd, (u_char*)msg, msglen);
sq_done(qp->q_stream);
}
return (0);
}
#ifdef notdef
/* i don't quite understand this but the only ref to it is notdef'd --vix */
prime(class, type, oqp)
int class, type;
register struct qinfo *oqp;
{
char dname[BUFSIZ];
if (oqp->q_msg == NULL)
return;
if (dn_expand((u_char *)oqp->q_msg,
(u_char *)oqp->q_msg + oqp->q_msglen,
(u_char *)oqp->q_msg + HFIXEDSZ, (u_char *)dname,
sizeof(dname)) < 0)
return;
dprintf(2, (ddt, "prime: %s\n", dname));
(void) sysquery(dname, class, type, NULL, 0);
}
#endif
void
prime_cache()
{
register struct qinfo *qp;
dprintf(1, (ddt, "prime_cache: priming = %d\n", priming));
if (!priming && fcachetab->h_tab[0] != NULL && !forward_only) {
priming++;
if ((qp = sysquery("", C_IN, T_NS, NULL, 0)) == NULL)
priming = 0;
else
qp->q_flags |= (Q_SYSTEM | Q_PRIMING);
}
needs_prime_cache = 0;
return;
}
struct qinfo *
sysquery(dname, class, type, nss, nsc)
char *dname;
int class, type;
struct in_addr *nss;
int nsc;
{
register struct qinfo *qp, *oqp;
register HEADER *hp;
struct namebuf *np;
struct databuf *nsp[NSMAX];
struct hashbuf *htp;
struct sockaddr_in *nsa;
char *fname;
int count;
#ifdef DATUMREFCNT
nsp[0] = NULL;
#endif
dprintf(3, (ddt, "sysquery(%s, %d, %d, 0x%x, %d)\n",
dname, class, type, nss, nsc));
qp = qnew();
if (nss && nsc) {
np = NULL;
} else {
htp = hashtab;
if (priming && dname[0] == '\0') {
np = NULL;
} else if ((np = nlookup(dname, &htp, &fname, 1)) == NULL) {
dprintf(1, (ddt,
"sysquery: nlookup error on %s?\n",
dname));
qfree(qp);
return (0);
}
switch (findns(&np, class, nsp, &count, 0)) {
case NXDOMAIN:
case SERVFAIL:
dprintf(1, (ddt,
"sysquery: findns error on %s?\n", dname));
qfree(qp);
#ifdef DATUMREFCNT
free_nsp(nsp);
#endif
return (0);
}
}
/* build new qinfo struct */
qp->q_cmsg = qp->q_msg = NULL;
qp->q_dfd = ds;
if (nss && nsc)
qp->q_fwd = NULL;
else
qp->q_fwd = fwdtab;
qp->q_expire = tt.tv_sec + RETRY_TIMEOUT*2;
qp->q_flags |= Q_SYSTEM;
#ifdef LAME_DELEGATION
getname(np, qp->q_domain, sizeof(qp->q_domain));
#endif /* LAME_DELEGATION */
if ((qp->q_msg = (u_char *)malloc(BUFSIZ)) == NULL) {
qfree(qp);
#ifdef DATUMREFCNT
free_nsp(nsp);
#endif
return (0);
}
qp->q_msglen = res_mkquery(QUERY, dname, class,
type, NULL, 0, NULL,
qp->q_msg, BUFSIZ);
hp = (HEADER *) qp->q_msg;
hp->id = qp->q_nsid = htons((u_int16_t)++nsid);
hp->rd = (qp->q_fwd ? 1 : 0);
/* First check for an already pending query for this data */
for (oqp = qhead; oqp != QINFO_NULL; oqp = oqp->q_link) {
if ((oqp != qp)
&& (oqp->q_msglen == qp->q_msglen)
&& bcmp((char *)oqp->q_msg+2,
qp->q_msg+2,
qp->q_msglen-2) == 0
) {
dprintf(3, (ddt, "sysquery: duplicate\n"));
qfree(qp);
#ifdef DATUMREFCNT
free_nsp(nsp);
#endif
return (0);
}
}
if (nss && nsc) {
int i;
struct qserv *qs;
for (i = 0, qs = qp->q_addr;
i < nsc;
i++, qs++) {
qs->ns_addr.sin_family = AF_INET;
qs->ns_addr.sin_addr = nss[i];
qs->ns_addr.sin_port = ns_port;
qs->ns = NULL;
qs->nsdata = NULL;
qs->stime = tt;
qs->nretry = 0;
}
qp->q_naddr = nsc;
} else {
if ((count = nslookup(nsp, qp, dname, "sysquery")) <= 0) {
if (count < 0) {
dprintf(1, (ddt,
"sysquery: nslookup reports danger\n"));
} else {
dprintf(1, (ddt,
"sysquery: no addrs found for NS's\n"));
}
qfree(qp);
#ifdef DATUMREFCNT
free_nsp(nsp);
#endif
return (0);
}
}
schedretry(qp, retrytime(qp));
if (qp->q_fwd == NULL)
qp->q_addr[0].stime = tt; /* XXX - why not every? */
nsa = Q_NEXTADDR(qp, 0);
dprintf(1, (ddt,
"sysquery: send -> [%s].%d dfd=%d nsid=%d id=%d retry=%ld\n",
inet_ntoa(nsa->sin_addr),
ntohs(nsa->sin_port), qp->q_dfd,
ntohs(qp->q_nsid), ntohs(qp->q_id),
qp->q_time));
#ifdef DEBUG
if (debug >= 10)
fp_query(qp->q_msg, ddt);
#endif
if (sendto(qp->q_dfd, qp->q_msg, qp->q_msglen, 0,
(struct sockaddr *)nsa,
sizeof(struct sockaddr_in)) < 0) {
if (!haveComplained((char*)nsa->sin_addr.s_addr, sendtoStr))
syslog(LOG_NOTICE, "sysquery: sendto([%s].%d): %m",
inet_ntoa(nsa->sin_addr), ntohs(nsa->sin_port));
nameserIncr(nsa->sin_addr, nssSendtoErr);
}
nameserIncr(nsa->sin_addr, nssSentSysQ);
#ifdef DATUMREFCNT
free_nsp(nsp);
#endif
return (qp);
}
/*
* Check the list of root servers after receiving a response
* to a query for the root servers.
*/
static void
check_root()
{
register struct databuf *dp, *pdp;
register struct namebuf *np;
int count = 0;
priming = 0;
for (np = hashtab->h_tab[0]; np != NULL; np = np->n_next)
if (np->n_dname[0] == '\0')
break;
if (np == NULL) {
syslog(LOG_ERR, "check_root: Can't find root!\n");
return;
}
for (dp = np->n_data; dp != NULL; dp = dp->d_next)
if (dp->d_type == T_NS)
count++;
dprintf(1, (ddt, "%d root servers\n", count));
if (count < MINROOTS) {
syslog(LOG_WARNING,
"check_root: %d root servers after query to root server < min",
count);
return;
}
pdp = NULL;
dp = np->n_data;
while (dp != NULL) {
if (dp->d_type == T_NS && dp->d_zone == 0 &&
dp->d_ttl < tt.tv_sec) {
dprintf(1, (ddt, "deleting old root server '%s'\n",
dp->d_data));
dp = rm_datum(dp, np, pdp);
/* SHOULD DELETE FROM HINTS ALSO */
continue;
}
pdp = dp;
dp = dp->d_next;
}
check_ns();
}
/*
* Check the root to make sure that for each NS record we have a A RR
*/
static void
check_ns()
{
register struct databuf *dp, *tdp;
register struct namebuf *np, *tnp;
struct hashbuf *htp;
char *dname;
int found_arr;
char *fname;
time_t curtime;
dprintf(2, (ddt, "check_ns()\n"));
curtime = (u_int32_t) tt.tv_sec;
for (np = hashtab->h_tab[0]; np != NULL; np = np->n_next) {
if (np->n_dname[0] != 0)
continue;
for (dp = np->n_data; dp != NULL; dp = dp->d_next) {
if (dp->d_type != T_NS)
continue;
/* look for A records */
dname = (caddr_t) dp->d_data;
htp = hashtab;
tnp = nlookup(dname, &htp, &fname, 0);
if (tnp == NULL || fname != dname) {
dprintf(3, (ddt,
"check_ns: %s: not found %s %x\n",
dname, fname, tnp));
sysquery(dname, dp->d_class, T_A, NULL, 0);
continue;
}
/* look for name server addresses */
found_arr = 0;
for (tdp=tnp->n_data; tdp!=NULL; tdp=tdp->d_next) {
if (tdp->d_type != T_A ||
tdp->d_class != dp->d_class)
continue;
if ((tdp->d_zone == 0) &&
(tdp->d_ttl < curtime)) {
dprintf(3, (ddt,
"check_ns: stale entry '%s'\n",
tnp->n_dname));
/* Cache invalidate the address RR's */
delete_all(tnp, dp->d_class, T_A);
found_arr = 0;
break;
}
found_arr++;
}
if (!found_arr)
(void) sysquery(dname, dp->d_class, T_A, NULL, 0);
}
}
}
/* int findns(npp, class, nsp, countp, flag)
* Find NS' or an SOA
* npp, class:
* dname whose least-superior NS is wanted
* nsp, countp:
* result array and count; array will also be NULL terminated
* flag:
* boolean: we're being called from ADDAUTH, bypass authority checks
* return value:
* NXDOMAIN: we are authoritative for this {dname,class}
* SERVFAIL: we are auth but zone isn't loaded; or, no root servers found
* OK: success (this is the only case where *countp and nsp[] are valid)
*/
int
findns(npp, class, nsp, countp, flag)
register struct namebuf **npp;
int class;
struct databuf **nsp;
int *countp;
int flag;
{
register struct namebuf *np = *npp;
register struct databuf *dp;
register struct databuf **nspp;
struct hashbuf *htp;
#ifdef DATUMREFCNT
nsp[0] = NULL;
#endif
if (priming && (np == NULL || np->n_dname[0] == '\0'))
htp = fcachetab;
else
htp = hashtab;
try_again:
if (htp == fcachetab)
needs_prime_cache = 1;
while (np == NULL && htp != NULL) {
dprintf(3, (ddt, "findns: using %s\n",
htp == hashtab ? "cache" : "hints"));
for (np = htp->h_tab[0]; np != NULL; np = np->n_next)
if (np->n_dname[0] == '\0')
break;
htp = (htp == hashtab ? fcachetab : NULL); /* Fallback */
}
while (np != NULL) {
dprintf(5, (ddt, "findns: np 0x%x '%s'\n", np, np->n_dname));
/* Look first for SOA records. */
#ifdef ADDAUTH
if (!flag)
#endif
for (dp = np->n_data; dp != NULL; dp = dp->d_next) {
if (dp->d_zone != 0 && match(dp, class, T_SOA)) {
dprintf(3, (ddt, "findns: SOA found\n"));
if (zones[dp->d_zone].z_flags & Z_AUTH) {
*npp = np;
nsp[0] = dp;
#ifdef DATUMREFCNT
nsp[1] = NULL;
dp->d_rcnt++;
#endif
return (NXDOMAIN);
} else {
/* XXX: zone isn't loaded but we're
* primary or secondary for it.
* should we fwd this?
*/
return (SERVFAIL);
}
}
}
/* If no SOA records, look for NS records. */
nspp = &nsp[0];
*nspp = NULL;
for (dp = np->n_data; dp != NULL; dp = dp->d_next) {
if (!match(dp, class, T_NS))
continue;
/*
* Don't use records that may become invalid to
* reference later when we do the rtt computation.
* Never delete our safety-belt information!
*
* XXX: this is horribly bogus.
*/
if ((dp->d_zone == 0) &&
(dp->d_ttl < (tt.tv_sec+900)) &&
!(dp->d_flags & DB_F_HINT)) {
dprintf(1, (ddt, "findns: stale entry '%s'\n",
np->n_dname));
/* Cache invalidate the NS RR's. */
if (dp->d_ttl < tt.tv_sec)
delete_all(np, class, T_NS);
goto try_parent;
}
if (nspp < &nsp[NSMAX-1]) {
*nspp++ = dp;
#ifdef DATUMREFCNT
dp->d_rcnt++;
#endif
}
}
*countp = nspp - nsp;
if (*countp > 0) {
dprintf(3, (ddt, "findns: %d NS's added for '%s'\n",
*countp, np->n_dname));
*nspp = NULL;
*npp = np;
return (OK); /* Success, got some NS's */
}
try_parent:
np = np->n_parent;
}
if (htp)
goto try_again;
dprintf(1, (ddt, "findns: No root nameservers for class %s?\n",
p_class(class)));
if ((unsigned)class < MAXCLASS && norootlogged[class] == 0) {
norootlogged[class] = 1;
syslog(LOG_ERR, "No root nameservers for class %s\n",
p_class(class));
}
return (SERVFAIL);
}
/*
* Extract RR's from the given node that match class and type.
* Return number of bytes added to response.
* If no matching data is found, then 0 is returned.
*/
int
finddata(np, class, type, hp, dnamep, lenp, countp)
struct namebuf *np;
int class, type;
register HEADER *hp;
char **dnamep;
int *lenp, *countp;
{
register struct databuf *dp;
register char *cp;
int buflen, n, count = 0, foundstale = 0;
#ifdef ROUND_ROBIN
if (type != T_ANY && type != T_PTR) {
/* cycle order of RRs, for a load balancing effect... */
register struct databuf **dpp;
for (dpp = &np->n_data; dp = *dpp; dpp = &dp->d_next) {
if (dp->d_next && wanted(dp, class, type)) {
register struct databuf *lp;
*dpp = lp = dp->d_next;
dp->d_next = NULL;
for (dpp = &lp->d_next;
*dpp;
dpp = &lp->d_next)
lp = *dpp;
*dpp = dp;
break;
}
}
}
#endif /*ROUND_ROBIN*/
buflen = *lenp;
cp = ((char *)hp) + *countp;
for (dp = np->n_data; dp != NULL; dp = dp->d_next) {
if (!wanted(dp, class, type)) {
#ifndef NCACHE /*if no negative caching then cname => nothing else*/
if (type == T_CNAME && class == dp->d_class) {
/* any data means no CNAME exists */
*countp = 0;
return 0;
}
#endif /*NCACHE*/
continue;
}
if (stale(dp)) {
/*
* Don't use stale data.
* Would like to call delete_all here
* and continue, but the data chain would get
* munged; can't restart, as make_rr has side
* effects (leaving pointers in dnptr).
* Just skip this entry for now
* and call delete_all at the end.
*/
dprintf(3, (ddt,
"finddata: stale entry '%s'\n",
np->n_dname));
if (dp->d_zone == 0)
foundstale++;
continue;
}
if (dp->d_cred == DB_C_ADDITIONAL) {
/* we want to expire additional data very
* quickly. current strategy is to cut 5%
* off each time it is accessed. this makes
* stale(dp) true faster when this datum is
* used often.
*/
dp->d_ttl = tt.tv_sec
+
0.95 * (int) (dp->d_ttl - tt.tv_sec);
}
#ifdef NCACHE
/* -ve $ing stuff, anant@isi.edu
* if we have a -ve $ed record, change the rcode on the
* header to reflect that
*/
if (dp->d_rcode == NOERROR_NODATA) {
if (count != 0) {
/*
* This should not happen, yet it does...
*/
syslog(LOG_WARNING,
"NODATA & data for \"%s\" type %d class %d",
*dnamep, type, class);
continue;
}
if (type != T_ANY) {
hp->rcode = NOERROR_NODATA;
*countp = 0;
return 1; /* XXX - we have to report success */
}
/* don't satisfy T_ANY queries from -$ info */
continue;
}
#ifndef RETURNSOA
if (dp->d_rcode == NXDOMAIN) {
if (count != 0) {
/*
* This should not happen, yet it might...
*/
syslog(LOG_WARNING,
"NXDOMAIN & data for \"%s\" type %d class %d",
*dnamep, type, class);
continue;
}
if (type != T_ANY) {
hp->rcode = NXDOMAIN;
*countp = 0;
return 1; /* XXX - we have to report success */
}
/* don't satisfy T_ANY queries from -$ info */
continue;
}
#endif
#endif /*NCACHE*/
if ((n = make_rr(*dnamep, dp, (u_char *)cp, buflen, 1)) < 0) {
hp->tc = 1;
*countp = count;
return (*lenp - buflen);
}
cp += n;
buflen -= n;
count++;
#ifdef notdef
/* this isn't right for glue records, aa is set in ns_req */
if (dp->d_zone &&
(zones[dp->d_zone].z_flags & Z_AUTH) &&
class != C_ANY)
hp->aa = 1; /* XXX */
#endif
if (dp->d_type == T_CNAME) {
if (type != T_ANY) { /* or T_NS? */
*dnamep = (caddr_t) dp->d_data;
if (dp->d_zone &&
(zones[dp->d_zone].z_flags & Z_AUTH) &&
class != C_ANY) /* XXX */
hp->aa = 1; /* XXX */
}
break;
}
}
/*
* Cache invalidate the other RR's of same type
* if some have timed out
*/
if (foundstale) {
delete_all(np, class, type);
/* XXX this isn't right if 'type' is something special
* such as T_AXFR or T_MAILB, since the matching done
* by match() in delete_all() is different from that
* done by wanted() above.
*/
}
dprintf(3, (ddt, "finddata: added %d class %d type %d RRs\n",
count, class, type));
*countp = count;
return (*lenp - buflen);
}
/*
* Do we want this data record based on the class and type?
*/
int
wanted(dp, class, type)
struct databuf *dp;
int class, type;
{
dprintf(3, (ddt, "wanted(%x, %d, %d) %d, %d\n", dp, class, type,
dp->d_class, dp->d_type));
if (dp->d_class != class && class != C_ANY)
return (0);
if (type == dp->d_type)
return (1);
#ifdef NCACHE
/*-ve $ing stuff, for a T_ANY query, we do not want to return
* -ve $ed RRs.
*/
if (type == T_ANY && dp->d_rcode == NOERROR_NODATA)
return (0);
#endif
switch (dp->d_type) {
case T_ANY:
return (1);
case T_CNAME:
#ifdef NCACHE
if (dp->d_rcode != NOERROR_NODATA)
#endif
return (1);
#ifdef NCACHE
else
break;
#endif
}
switch (type) {
case T_ANY:
return (1);
case T_MAILB:
switch (dp->d_type) {
case T_MR:
case T_MB:
case T_MG:
case T_MINFO:
return (1);
}
break;
case T_AXFR:
/* T_AXFR needs an authoritative SOA */
if (dp->d_type == T_SOA && dp->d_zone != 0
&& (zones[dp->d_zone].z_flags & Z_AUTH))
return (1);
break;
}
return (0);
}
/*
* Add RR entries from dpp array to a query/response.
* Return the number of bytes added or negative the amount
* added if truncation was required. Typically you are
* adding NS records to a response.
*/
int
add_data(np, dpp, cp, buflen)
struct namebuf *np;
struct databuf **dpp;
register u_char *cp;
int buflen;
{
register struct databuf *dp;
char dname[MAXDNAME];
register int n, count = 0;
getname(np, dname, sizeof(dname));
for(dp = *dpp++; dp != NULL; dp = *dpp++) {
if (stale(dp))
continue; /* ignore old cache entry */
#ifdef NCACHE
if (dp->d_rcode)
continue;
#endif
if ((n = make_rr(dname, dp, cp, buflen, 1)) < 0)
return (-count); /* Truncation */
cp += n;
buflen -= n;
count += n;
}
return (count);
}
/*
* This is best thought of as a "cache invalidate" function.
* It is called whenever a piece of data is determined to have
* timed out. It is better to have no information, than to
* have partial information you pass off as complete.
*/
void
delete_all(np, class, type)
register struct namebuf *np;
int class, type;
{
register struct databuf *dp, *pdp;
dprintf(3, (ddt, "delete_all: '%s' 0x%x class %d type %d\n",
np->n_dname, np, class, type));
pdp = NULL;
dp = np->n_data;
while (dp != NULL) {
if ((dp->d_zone == 0) && !(dp->d_flags & DB_F_HINT)
&& match(dp, class, type)) {
dp = rm_datum(dp, np, pdp);
continue;
}
pdp = dp;
dp = dp->d_next;
}
}