mirror of
https://github.com/monero-project/monero.git
synced 2024-12-14 12:26:31 +02:00
a85b5759f3
These files were pulled from the 1.6.3 release tarball. This new version builds against OpenSSL version 1.1 which will be the default in the new Debian Stable which is due to be released RealSoonNow (tm).
998 lines
28 KiB
C
998 lines
28 KiB
C
/*
|
|
* services/cache/infra.c - infrastructure cache, server rtt and capabilities
|
|
*
|
|
* Copyright (c) 2007, NLnet Labs. All rights reserved.
|
|
*
|
|
* This software is open source.
|
|
*
|
|
* Redistribution and use in source and binary forms, with or without
|
|
* modification, are permitted provided that the following conditions
|
|
* are met:
|
|
*
|
|
* Redistributions of source code must retain the above copyright notice,
|
|
* this list of conditions and the following disclaimer.
|
|
*
|
|
* 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.
|
|
*
|
|
* Neither the name of the NLNET LABS 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 COPYRIGHT HOLDERS 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 COPYRIGHT
|
|
* HOLDER 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.
|
|
*/
|
|
|
|
/**
|
|
* \file
|
|
*
|
|
* This file contains the infrastructure cache.
|
|
*/
|
|
#include "config.h"
|
|
#include "sldns/rrdef.h"
|
|
#include "sldns/str2wire.h"
|
|
#include "services/cache/infra.h"
|
|
#include "util/storage/slabhash.h"
|
|
#include "util/storage/lookup3.h"
|
|
#include "util/data/dname.h"
|
|
#include "util/log.h"
|
|
#include "util/net_help.h"
|
|
#include "util/config_file.h"
|
|
#include "iterator/iterator.h"
|
|
|
|
/** Timeout when only a single probe query per IP is allowed. */
|
|
#define PROBE_MAXRTO 12000 /* in msec */
|
|
|
|
/** number of timeouts for a type when the domain can be blocked ;
|
|
* even if another type has completely rtt maxed it, the different type
|
|
* can do this number of packets (until those all timeout too) */
|
|
#define TIMEOUT_COUNT_MAX 3
|
|
|
|
/** ratelimit value for delegation point */
|
|
int infra_dp_ratelimit = 0;
|
|
|
|
/** ratelimit value for client ip addresses,
|
|
* in queries per second. */
|
|
int infra_ip_ratelimit = 0;
|
|
|
|
size_t
|
|
infra_sizefunc(void* k, void* ATTR_UNUSED(d))
|
|
{
|
|
struct infra_key* key = (struct infra_key*)k;
|
|
return sizeof(*key) + sizeof(struct infra_data) + key->namelen
|
|
+ lock_get_mem(&key->entry.lock);
|
|
}
|
|
|
|
int
|
|
infra_compfunc(void* key1, void* key2)
|
|
{
|
|
struct infra_key* k1 = (struct infra_key*)key1;
|
|
struct infra_key* k2 = (struct infra_key*)key2;
|
|
int r = sockaddr_cmp(&k1->addr, k1->addrlen, &k2->addr, k2->addrlen);
|
|
if(r != 0)
|
|
return r;
|
|
if(k1->namelen != k2->namelen) {
|
|
if(k1->namelen < k2->namelen)
|
|
return -1;
|
|
return 1;
|
|
}
|
|
return query_dname_compare(k1->zonename, k2->zonename);
|
|
}
|
|
|
|
void
|
|
infra_delkeyfunc(void* k, void* ATTR_UNUSED(arg))
|
|
{
|
|
struct infra_key* key = (struct infra_key*)k;
|
|
if(!key)
|
|
return;
|
|
lock_rw_destroy(&key->entry.lock);
|
|
free(key->zonename);
|
|
free(key);
|
|
}
|
|
|
|
void
|
|
infra_deldatafunc(void* d, void* ATTR_UNUSED(arg))
|
|
{
|
|
struct infra_data* data = (struct infra_data*)d;
|
|
free(data);
|
|
}
|
|
|
|
size_t
|
|
rate_sizefunc(void* k, void* ATTR_UNUSED(d))
|
|
{
|
|
struct rate_key* key = (struct rate_key*)k;
|
|
return sizeof(*key) + sizeof(struct rate_data) + key->namelen
|
|
+ lock_get_mem(&key->entry.lock);
|
|
}
|
|
|
|
int
|
|
rate_compfunc(void* key1, void* key2)
|
|
{
|
|
struct rate_key* k1 = (struct rate_key*)key1;
|
|
struct rate_key* k2 = (struct rate_key*)key2;
|
|
if(k1->namelen != k2->namelen) {
|
|
if(k1->namelen < k2->namelen)
|
|
return -1;
|
|
return 1;
|
|
}
|
|
return query_dname_compare(k1->name, k2->name);
|
|
}
|
|
|
|
void
|
|
rate_delkeyfunc(void* k, void* ATTR_UNUSED(arg))
|
|
{
|
|
struct rate_key* key = (struct rate_key*)k;
|
|
if(!key)
|
|
return;
|
|
lock_rw_destroy(&key->entry.lock);
|
|
free(key->name);
|
|
free(key);
|
|
}
|
|
|
|
void
|
|
rate_deldatafunc(void* d, void* ATTR_UNUSED(arg))
|
|
{
|
|
struct rate_data* data = (struct rate_data*)d;
|
|
free(data);
|
|
}
|
|
|
|
/** find or create element in domainlimit tree */
|
|
static struct domain_limit_data* domain_limit_findcreate(
|
|
struct infra_cache* infra, char* name)
|
|
{
|
|
uint8_t* nm;
|
|
int labs;
|
|
size_t nmlen;
|
|
struct domain_limit_data* d;
|
|
|
|
/* parse name */
|
|
nm = sldns_str2wire_dname(name, &nmlen);
|
|
if(!nm) {
|
|
log_err("could not parse %s", name);
|
|
return NULL;
|
|
}
|
|
labs = dname_count_labels(nm);
|
|
|
|
/* can we find it? */
|
|
d = (struct domain_limit_data*)name_tree_find(&infra->domain_limits,
|
|
nm, nmlen, labs, LDNS_RR_CLASS_IN);
|
|
if(d) {
|
|
free(nm);
|
|
return d;
|
|
}
|
|
|
|
/* create it */
|
|
d = (struct domain_limit_data*)calloc(1, sizeof(*d));
|
|
if(!d) {
|
|
free(nm);
|
|
return NULL;
|
|
}
|
|
d->node.node.key = &d->node;
|
|
d->node.name = nm;
|
|
d->node.len = nmlen;
|
|
d->node.labs = labs;
|
|
d->node.dclass = LDNS_RR_CLASS_IN;
|
|
d->lim = -1;
|
|
d->below = -1;
|
|
if(!name_tree_insert(&infra->domain_limits, &d->node, nm, nmlen,
|
|
labs, LDNS_RR_CLASS_IN)) {
|
|
log_err("duplicate element in domainlimit tree");
|
|
free(nm);
|
|
free(d);
|
|
return NULL;
|
|
}
|
|
return d;
|
|
}
|
|
|
|
/** insert rate limit configuration into lookup tree */
|
|
static int infra_ratelimit_cfg_insert(struct infra_cache* infra,
|
|
struct config_file* cfg)
|
|
{
|
|
struct config_str2list* p;
|
|
struct domain_limit_data* d;
|
|
for(p = cfg->ratelimit_for_domain; p; p = p->next) {
|
|
d = domain_limit_findcreate(infra, p->str);
|
|
if(!d)
|
|
return 0;
|
|
d->lim = atoi(p->str2);
|
|
}
|
|
for(p = cfg->ratelimit_below_domain; p; p = p->next) {
|
|
d = domain_limit_findcreate(infra, p->str);
|
|
if(!d)
|
|
return 0;
|
|
d->below = atoi(p->str2);
|
|
}
|
|
return 1;
|
|
}
|
|
|
|
struct infra_cache*
|
|
infra_create(struct config_file* cfg)
|
|
{
|
|
struct infra_cache* infra = (struct infra_cache*)calloc(1,
|
|
sizeof(struct infra_cache));
|
|
size_t maxmem = cfg->infra_cache_numhosts * (sizeof(struct infra_key)+
|
|
sizeof(struct infra_data)+INFRA_BYTES_NAME);
|
|
infra->hosts = slabhash_create(cfg->infra_cache_slabs,
|
|
INFRA_HOST_STARTSIZE, maxmem, &infra_sizefunc, &infra_compfunc,
|
|
&infra_delkeyfunc, &infra_deldatafunc, NULL);
|
|
if(!infra->hosts) {
|
|
free(infra);
|
|
return NULL;
|
|
}
|
|
infra->host_ttl = cfg->host_ttl;
|
|
name_tree_init(&infra->domain_limits);
|
|
infra_dp_ratelimit = cfg->ratelimit;
|
|
if(cfg->ratelimit != 0) {
|
|
infra->domain_rates = slabhash_create(cfg->ratelimit_slabs,
|
|
INFRA_HOST_STARTSIZE, cfg->ratelimit_size,
|
|
&rate_sizefunc, &rate_compfunc, &rate_delkeyfunc,
|
|
&rate_deldatafunc, NULL);
|
|
if(!infra->domain_rates) {
|
|
infra_delete(infra);
|
|
return NULL;
|
|
}
|
|
/* insert config data into ratelimits */
|
|
if(!infra_ratelimit_cfg_insert(infra, cfg)) {
|
|
infra_delete(infra);
|
|
return NULL;
|
|
}
|
|
name_tree_init_parents(&infra->domain_limits);
|
|
}
|
|
infra_ip_ratelimit = cfg->ip_ratelimit;
|
|
infra->client_ip_rates = slabhash_create(cfg->ratelimit_slabs,
|
|
INFRA_HOST_STARTSIZE, cfg->ip_ratelimit_size, &ip_rate_sizefunc,
|
|
&ip_rate_compfunc, &ip_rate_delkeyfunc, &ip_rate_deldatafunc, NULL);
|
|
if(!infra->client_ip_rates) {
|
|
infra_delete(infra);
|
|
return NULL;
|
|
}
|
|
return infra;
|
|
}
|
|
|
|
/** delete domain_limit entries */
|
|
static void domain_limit_free(rbnode_type* n, void* ATTR_UNUSED(arg))
|
|
{
|
|
if(n) {
|
|
free(((struct domain_limit_data*)n)->node.name);
|
|
free(n);
|
|
}
|
|
}
|
|
|
|
void
|
|
infra_delete(struct infra_cache* infra)
|
|
{
|
|
if(!infra)
|
|
return;
|
|
slabhash_delete(infra->hosts);
|
|
slabhash_delete(infra->domain_rates);
|
|
traverse_postorder(&infra->domain_limits, domain_limit_free, NULL);
|
|
slabhash_delete(infra->client_ip_rates);
|
|
free(infra);
|
|
}
|
|
|
|
struct infra_cache*
|
|
infra_adjust(struct infra_cache* infra, struct config_file* cfg)
|
|
{
|
|
size_t maxmem;
|
|
if(!infra)
|
|
return infra_create(cfg);
|
|
infra->host_ttl = cfg->host_ttl;
|
|
maxmem = cfg->infra_cache_numhosts * (sizeof(struct infra_key)+
|
|
sizeof(struct infra_data)+INFRA_BYTES_NAME);
|
|
if(maxmem != slabhash_get_size(infra->hosts) ||
|
|
cfg->infra_cache_slabs != infra->hosts->size) {
|
|
infra_delete(infra);
|
|
infra = infra_create(cfg);
|
|
}
|
|
return infra;
|
|
}
|
|
|
|
/** calculate the hash value for a host key
|
|
* set use_port to a non-0 number to use the port in
|
|
* the hash calculation; 0 to ignore the port.*/
|
|
static hashvalue_type
|
|
hash_addr(struct sockaddr_storage* addr, socklen_t addrlen,
|
|
int use_port)
|
|
{
|
|
hashvalue_type h = 0xab;
|
|
/* select the pieces to hash, some OS have changing data inside */
|
|
if(addr_is_ip6(addr, addrlen)) {
|
|
struct sockaddr_in6* in6 = (struct sockaddr_in6*)addr;
|
|
h = hashlittle(&in6->sin6_family, sizeof(in6->sin6_family), h);
|
|
if(use_port){
|
|
h = hashlittle(&in6->sin6_port, sizeof(in6->sin6_port), h);
|
|
}
|
|
h = hashlittle(&in6->sin6_addr, INET6_SIZE, h);
|
|
} else {
|
|
struct sockaddr_in* in = (struct sockaddr_in*)addr;
|
|
h = hashlittle(&in->sin_family, sizeof(in->sin_family), h);
|
|
if(use_port){
|
|
h = hashlittle(&in->sin_port, sizeof(in->sin_port), h);
|
|
}
|
|
h = hashlittle(&in->sin_addr, INET_SIZE, h);
|
|
}
|
|
return h;
|
|
}
|
|
|
|
/** calculate infra hash for a key */
|
|
static hashvalue_type
|
|
hash_infra(struct sockaddr_storage* addr, socklen_t addrlen, uint8_t* name)
|
|
{
|
|
return dname_query_hash(name, hash_addr(addr, addrlen, 1));
|
|
}
|
|
|
|
/** lookup version that does not check host ttl (you check it) */
|
|
struct lruhash_entry*
|
|
infra_lookup_nottl(struct infra_cache* infra, struct sockaddr_storage* addr,
|
|
socklen_t addrlen, uint8_t* name, size_t namelen, int wr)
|
|
{
|
|
struct infra_key k;
|
|
k.addrlen = addrlen;
|
|
memcpy(&k.addr, addr, addrlen);
|
|
k.namelen = namelen;
|
|
k.zonename = name;
|
|
k.entry.hash = hash_infra(addr, addrlen, name);
|
|
k.entry.key = (void*)&k;
|
|
k.entry.data = NULL;
|
|
return slabhash_lookup(infra->hosts, k.entry.hash, &k, wr);
|
|
}
|
|
|
|
/** init the data elements */
|
|
static void
|
|
data_entry_init(struct infra_cache* infra, struct lruhash_entry* e,
|
|
time_t timenow)
|
|
{
|
|
struct infra_data* data = (struct infra_data*)e->data;
|
|
data->ttl = timenow + infra->host_ttl;
|
|
rtt_init(&data->rtt);
|
|
data->edns_version = 0;
|
|
data->edns_lame_known = 0;
|
|
data->probedelay = 0;
|
|
data->isdnsseclame = 0;
|
|
data->rec_lame = 0;
|
|
data->lame_type_A = 0;
|
|
data->lame_other = 0;
|
|
data->timeout_A = 0;
|
|
data->timeout_AAAA = 0;
|
|
data->timeout_other = 0;
|
|
}
|
|
|
|
/**
|
|
* Create and init a new entry for a host
|
|
* @param infra: infra structure with config parameters.
|
|
* @param addr: host address.
|
|
* @param addrlen: length of addr.
|
|
* @param name: name of zone
|
|
* @param namelen: length of name.
|
|
* @param tm: time now.
|
|
* @return: the new entry or NULL on malloc failure.
|
|
*/
|
|
static struct lruhash_entry*
|
|
new_entry(struct infra_cache* infra, struct sockaddr_storage* addr,
|
|
socklen_t addrlen, uint8_t* name, size_t namelen, time_t tm)
|
|
{
|
|
struct infra_data* data;
|
|
struct infra_key* key = (struct infra_key*)malloc(sizeof(*key));
|
|
if(!key)
|
|
return NULL;
|
|
data = (struct infra_data*)malloc(sizeof(struct infra_data));
|
|
if(!data) {
|
|
free(key);
|
|
return NULL;
|
|
}
|
|
key->zonename = memdup(name, namelen);
|
|
if(!key->zonename) {
|
|
free(key);
|
|
free(data);
|
|
return NULL;
|
|
}
|
|
key->namelen = namelen;
|
|
lock_rw_init(&key->entry.lock);
|
|
key->entry.hash = hash_infra(addr, addrlen, name);
|
|
key->entry.key = (void*)key;
|
|
key->entry.data = (void*)data;
|
|
key->addrlen = addrlen;
|
|
memcpy(&key->addr, addr, addrlen);
|
|
data_entry_init(infra, &key->entry, tm);
|
|
return &key->entry;
|
|
}
|
|
|
|
int
|
|
infra_host(struct infra_cache* infra, struct sockaddr_storage* addr,
|
|
socklen_t addrlen, uint8_t* nm, size_t nmlen, time_t timenow,
|
|
int* edns_vs, uint8_t* edns_lame_known, int* to)
|
|
{
|
|
struct lruhash_entry* e = infra_lookup_nottl(infra, addr, addrlen,
|
|
nm, nmlen, 0);
|
|
struct infra_data* data;
|
|
int wr = 0;
|
|
if(e && ((struct infra_data*)e->data)->ttl < timenow) {
|
|
/* it expired, try to reuse existing entry */
|
|
int old = ((struct infra_data*)e->data)->rtt.rto;
|
|
uint8_t tA = ((struct infra_data*)e->data)->timeout_A;
|
|
uint8_t tAAAA = ((struct infra_data*)e->data)->timeout_AAAA;
|
|
uint8_t tother = ((struct infra_data*)e->data)->timeout_other;
|
|
lock_rw_unlock(&e->lock);
|
|
e = infra_lookup_nottl(infra, addr, addrlen, nm, nmlen, 1);
|
|
if(e) {
|
|
/* if its still there we have a writelock, init */
|
|
/* re-initialise */
|
|
/* do not touch lameness, it may be valid still */
|
|
data_entry_init(infra, e, timenow);
|
|
wr = 1;
|
|
/* TOP_TIMEOUT remains on reuse */
|
|
if(old >= USEFUL_SERVER_TOP_TIMEOUT) {
|
|
((struct infra_data*)e->data)->rtt.rto
|
|
= USEFUL_SERVER_TOP_TIMEOUT;
|
|
((struct infra_data*)e->data)->timeout_A = tA;
|
|
((struct infra_data*)e->data)->timeout_AAAA = tAAAA;
|
|
((struct infra_data*)e->data)->timeout_other = tother;
|
|
}
|
|
}
|
|
}
|
|
if(!e) {
|
|
/* insert new entry */
|
|
if(!(e = new_entry(infra, addr, addrlen, nm, nmlen, timenow)))
|
|
return 0;
|
|
data = (struct infra_data*)e->data;
|
|
*edns_vs = data->edns_version;
|
|
*edns_lame_known = data->edns_lame_known;
|
|
*to = rtt_timeout(&data->rtt);
|
|
slabhash_insert(infra->hosts, e->hash, e, data, NULL);
|
|
return 1;
|
|
}
|
|
/* use existing entry */
|
|
data = (struct infra_data*)e->data;
|
|
*edns_vs = data->edns_version;
|
|
*edns_lame_known = data->edns_lame_known;
|
|
*to = rtt_timeout(&data->rtt);
|
|
if(*to >= PROBE_MAXRTO && rtt_notimeout(&data->rtt)*4 <= *to) {
|
|
/* delay other queries, this is the probe query */
|
|
if(!wr) {
|
|
lock_rw_unlock(&e->lock);
|
|
e = infra_lookup_nottl(infra, addr,addrlen,nm,nmlen, 1);
|
|
if(!e) { /* flushed from cache real fast, no use to
|
|
allocate just for the probedelay */
|
|
return 1;
|
|
}
|
|
data = (struct infra_data*)e->data;
|
|
}
|
|
/* add 999 to round up the timeout value from msec to sec,
|
|
* then add a whole second so it is certain that this probe
|
|
* has timed out before the next is allowed */
|
|
data->probedelay = timenow + ((*to)+1999)/1000;
|
|
}
|
|
lock_rw_unlock(&e->lock);
|
|
return 1;
|
|
}
|
|
|
|
int
|
|
infra_set_lame(struct infra_cache* infra, struct sockaddr_storage* addr,
|
|
socklen_t addrlen, uint8_t* nm, size_t nmlen, time_t timenow,
|
|
int dnsseclame, int reclame, uint16_t qtype)
|
|
{
|
|
struct infra_data* data;
|
|
struct lruhash_entry* e;
|
|
int needtoinsert = 0;
|
|
e = infra_lookup_nottl(infra, addr, addrlen, nm, nmlen, 1);
|
|
if(!e) {
|
|
/* insert it */
|
|
if(!(e = new_entry(infra, addr, addrlen, nm, nmlen, timenow))) {
|
|
log_err("set_lame: malloc failure");
|
|
return 0;
|
|
}
|
|
needtoinsert = 1;
|
|
} else if( ((struct infra_data*)e->data)->ttl < timenow) {
|
|
/* expired, reuse existing entry */
|
|
data_entry_init(infra, e, timenow);
|
|
}
|
|
/* got an entry, now set the zone lame */
|
|
data = (struct infra_data*)e->data;
|
|
/* merge data (if any) */
|
|
if(dnsseclame)
|
|
data->isdnsseclame = 1;
|
|
if(reclame)
|
|
data->rec_lame = 1;
|
|
if(!dnsseclame && !reclame && qtype == LDNS_RR_TYPE_A)
|
|
data->lame_type_A = 1;
|
|
if(!dnsseclame && !reclame && qtype != LDNS_RR_TYPE_A)
|
|
data->lame_other = 1;
|
|
/* done */
|
|
if(needtoinsert)
|
|
slabhash_insert(infra->hosts, e->hash, e, e->data, NULL);
|
|
else { lock_rw_unlock(&e->lock); }
|
|
return 1;
|
|
}
|
|
|
|
void
|
|
infra_update_tcp_works(struct infra_cache* infra,
|
|
struct sockaddr_storage* addr, socklen_t addrlen, uint8_t* nm,
|
|
size_t nmlen)
|
|
{
|
|
struct lruhash_entry* e = infra_lookup_nottl(infra, addr, addrlen,
|
|
nm, nmlen, 1);
|
|
struct infra_data* data;
|
|
if(!e)
|
|
return; /* doesn't exist */
|
|
data = (struct infra_data*)e->data;
|
|
if(data->rtt.rto >= RTT_MAX_TIMEOUT)
|
|
/* do not disqualify this server altogether, it is better
|
|
* than nothing */
|
|
data->rtt.rto = RTT_MAX_TIMEOUT-1000;
|
|
lock_rw_unlock(&e->lock);
|
|
}
|
|
|
|
int
|
|
infra_rtt_update(struct infra_cache* infra, struct sockaddr_storage* addr,
|
|
socklen_t addrlen, uint8_t* nm, size_t nmlen, int qtype,
|
|
int roundtrip, int orig_rtt, time_t timenow)
|
|
{
|
|
struct lruhash_entry* e = infra_lookup_nottl(infra, addr, addrlen,
|
|
nm, nmlen, 1);
|
|
struct infra_data* data;
|
|
int needtoinsert = 0;
|
|
int rto = 1;
|
|
if(!e) {
|
|
if(!(e = new_entry(infra, addr, addrlen, nm, nmlen, timenow)))
|
|
return 0;
|
|
needtoinsert = 1;
|
|
} else if(((struct infra_data*)e->data)->ttl < timenow) {
|
|
data_entry_init(infra, e, timenow);
|
|
}
|
|
/* have an entry, update the rtt */
|
|
data = (struct infra_data*)e->data;
|
|
if(roundtrip == -1) {
|
|
rtt_lost(&data->rtt, orig_rtt);
|
|
if(qtype == LDNS_RR_TYPE_A) {
|
|
if(data->timeout_A < TIMEOUT_COUNT_MAX)
|
|
data->timeout_A++;
|
|
} else if(qtype == LDNS_RR_TYPE_AAAA) {
|
|
if(data->timeout_AAAA < TIMEOUT_COUNT_MAX)
|
|
data->timeout_AAAA++;
|
|
} else {
|
|
if(data->timeout_other < TIMEOUT_COUNT_MAX)
|
|
data->timeout_other++;
|
|
}
|
|
} else {
|
|
/* if we got a reply, but the old timeout was above server
|
|
* selection height, delete the timeout so the server is
|
|
* fully available again */
|
|
if(rtt_unclamped(&data->rtt) >= USEFUL_SERVER_TOP_TIMEOUT)
|
|
rtt_init(&data->rtt);
|
|
rtt_update(&data->rtt, roundtrip);
|
|
data->probedelay = 0;
|
|
if(qtype == LDNS_RR_TYPE_A)
|
|
data->timeout_A = 0;
|
|
else if(qtype == LDNS_RR_TYPE_AAAA)
|
|
data->timeout_AAAA = 0;
|
|
else data->timeout_other = 0;
|
|
}
|
|
if(data->rtt.rto > 0)
|
|
rto = data->rtt.rto;
|
|
|
|
if(needtoinsert)
|
|
slabhash_insert(infra->hosts, e->hash, e, e->data, NULL);
|
|
else { lock_rw_unlock(&e->lock); }
|
|
return rto;
|
|
}
|
|
|
|
long long infra_get_host_rto(struct infra_cache* infra,
|
|
struct sockaddr_storage* addr, socklen_t addrlen, uint8_t* nm,
|
|
size_t nmlen, struct rtt_info* rtt, int* delay, time_t timenow,
|
|
int* tA, int* tAAAA, int* tother)
|
|
{
|
|
struct lruhash_entry* e = infra_lookup_nottl(infra, addr, addrlen,
|
|
nm, nmlen, 0);
|
|
struct infra_data* data;
|
|
long long ttl = -2;
|
|
if(!e) return -1;
|
|
data = (struct infra_data*)e->data;
|
|
if(data->ttl >= timenow) {
|
|
ttl = (long long)(data->ttl - timenow);
|
|
memmove(rtt, &data->rtt, sizeof(*rtt));
|
|
if(timenow < data->probedelay)
|
|
*delay = (int)(data->probedelay - timenow);
|
|
else *delay = 0;
|
|
}
|
|
*tA = (int)data->timeout_A;
|
|
*tAAAA = (int)data->timeout_AAAA;
|
|
*tother = (int)data->timeout_other;
|
|
lock_rw_unlock(&e->lock);
|
|
return ttl;
|
|
}
|
|
|
|
int
|
|
infra_edns_update(struct infra_cache* infra, struct sockaddr_storage* addr,
|
|
socklen_t addrlen, uint8_t* nm, size_t nmlen, int edns_version,
|
|
time_t timenow)
|
|
{
|
|
struct lruhash_entry* e = infra_lookup_nottl(infra, addr, addrlen,
|
|
nm, nmlen, 1);
|
|
struct infra_data* data;
|
|
int needtoinsert = 0;
|
|
if(!e) {
|
|
if(!(e = new_entry(infra, addr, addrlen, nm, nmlen, timenow)))
|
|
return 0;
|
|
needtoinsert = 1;
|
|
} else if(((struct infra_data*)e->data)->ttl < timenow) {
|
|
data_entry_init(infra, e, timenow);
|
|
}
|
|
/* have an entry, update the rtt, and the ttl */
|
|
data = (struct infra_data*)e->data;
|
|
/* do not update if noEDNS and stored is yesEDNS */
|
|
if(!(edns_version == -1 && (data->edns_version != -1 &&
|
|
data->edns_lame_known))) {
|
|
data->edns_version = edns_version;
|
|
data->edns_lame_known = 1;
|
|
}
|
|
|
|
if(needtoinsert)
|
|
slabhash_insert(infra->hosts, e->hash, e, e->data, NULL);
|
|
else { lock_rw_unlock(&e->lock); }
|
|
return 1;
|
|
}
|
|
|
|
int
|
|
infra_get_lame_rtt(struct infra_cache* infra,
|
|
struct sockaddr_storage* addr, socklen_t addrlen,
|
|
uint8_t* name, size_t namelen, uint16_t qtype,
|
|
int* lame, int* dnsseclame, int* reclame, int* rtt, time_t timenow)
|
|
{
|
|
struct infra_data* host;
|
|
struct lruhash_entry* e = infra_lookup_nottl(infra, addr, addrlen,
|
|
name, namelen, 0);
|
|
if(!e)
|
|
return 0;
|
|
host = (struct infra_data*)e->data;
|
|
*rtt = rtt_unclamped(&host->rtt);
|
|
if(host->rtt.rto >= PROBE_MAXRTO && timenow < host->probedelay
|
|
&& rtt_notimeout(&host->rtt)*4 <= host->rtt.rto) {
|
|
/* single probe for this domain, and we are not probing */
|
|
/* unless the query type allows a probe to happen */
|
|
if(qtype == LDNS_RR_TYPE_A) {
|
|
if(host->timeout_A >= TIMEOUT_COUNT_MAX)
|
|
*rtt = USEFUL_SERVER_TOP_TIMEOUT;
|
|
else *rtt = USEFUL_SERVER_TOP_TIMEOUT-1000;
|
|
} else if(qtype == LDNS_RR_TYPE_AAAA) {
|
|
if(host->timeout_AAAA >= TIMEOUT_COUNT_MAX)
|
|
*rtt = USEFUL_SERVER_TOP_TIMEOUT;
|
|
else *rtt = USEFUL_SERVER_TOP_TIMEOUT-1000;
|
|
} else {
|
|
if(host->timeout_other >= TIMEOUT_COUNT_MAX)
|
|
*rtt = USEFUL_SERVER_TOP_TIMEOUT;
|
|
else *rtt = USEFUL_SERVER_TOP_TIMEOUT-1000;
|
|
}
|
|
}
|
|
if(timenow > host->ttl) {
|
|
/* expired entry */
|
|
/* see if this can be a re-probe of an unresponsive server */
|
|
/* minus 1000 because that is outside of the RTTBAND, so
|
|
* blacklisted servers stay blacklisted if this is chosen */
|
|
if(host->rtt.rto >= USEFUL_SERVER_TOP_TIMEOUT) {
|
|
lock_rw_unlock(&e->lock);
|
|
*rtt = USEFUL_SERVER_TOP_TIMEOUT-1000;
|
|
*lame = 0;
|
|
*dnsseclame = 0;
|
|
*reclame = 0;
|
|
return 1;
|
|
}
|
|
lock_rw_unlock(&e->lock);
|
|
return 0;
|
|
}
|
|
/* check lameness first */
|
|
if(host->lame_type_A && qtype == LDNS_RR_TYPE_A) {
|
|
lock_rw_unlock(&e->lock);
|
|
*lame = 1;
|
|
*dnsseclame = 0;
|
|
*reclame = 0;
|
|
return 1;
|
|
} else if(host->lame_other && qtype != LDNS_RR_TYPE_A) {
|
|
lock_rw_unlock(&e->lock);
|
|
*lame = 1;
|
|
*dnsseclame = 0;
|
|
*reclame = 0;
|
|
return 1;
|
|
} else if(host->isdnsseclame) {
|
|
lock_rw_unlock(&e->lock);
|
|
*lame = 0;
|
|
*dnsseclame = 1;
|
|
*reclame = 0;
|
|
return 1;
|
|
} else if(host->rec_lame) {
|
|
lock_rw_unlock(&e->lock);
|
|
*lame = 0;
|
|
*dnsseclame = 0;
|
|
*reclame = 1;
|
|
return 1;
|
|
}
|
|
/* no lameness for this type of query */
|
|
lock_rw_unlock(&e->lock);
|
|
*lame = 0;
|
|
*dnsseclame = 0;
|
|
*reclame = 0;
|
|
return 1;
|
|
}
|
|
|
|
int infra_find_ratelimit(struct infra_cache* infra, uint8_t* name,
|
|
size_t namelen)
|
|
{
|
|
int labs = dname_count_labels(name);
|
|
struct domain_limit_data* d = (struct domain_limit_data*)
|
|
name_tree_lookup(&infra->domain_limits, name, namelen, labs,
|
|
LDNS_RR_CLASS_IN);
|
|
if(!d) return infra_dp_ratelimit;
|
|
|
|
if(d->node.labs == labs && d->lim != -1)
|
|
return d->lim; /* exact match */
|
|
|
|
/* find 'below match' */
|
|
if(d->node.labs == labs)
|
|
d = (struct domain_limit_data*)d->node.parent;
|
|
while(d) {
|
|
if(d->below != -1)
|
|
return d->below;
|
|
d = (struct domain_limit_data*)d->node.parent;
|
|
}
|
|
return infra_dp_ratelimit;
|
|
}
|
|
|
|
size_t ip_rate_sizefunc(void* k, void* ATTR_UNUSED(d))
|
|
{
|
|
struct ip_rate_key* key = (struct ip_rate_key*)k;
|
|
return sizeof(*key) + sizeof(struct ip_rate_data)
|
|
+ lock_get_mem(&key->entry.lock);
|
|
}
|
|
|
|
int ip_rate_compfunc(void* key1, void* key2)
|
|
{
|
|
struct ip_rate_key* k1 = (struct ip_rate_key*)key1;
|
|
struct ip_rate_key* k2 = (struct ip_rate_key*)key2;
|
|
return sockaddr_cmp_addr(&k1->addr, k1->addrlen,
|
|
&k2->addr, k2->addrlen);
|
|
}
|
|
|
|
void ip_rate_delkeyfunc(void* k, void* ATTR_UNUSED(arg))
|
|
{
|
|
struct ip_rate_key* key = (struct ip_rate_key*)k;
|
|
if(!key)
|
|
return;
|
|
lock_rw_destroy(&key->entry.lock);
|
|
free(key);
|
|
}
|
|
|
|
/** find data item in array, for write access, caller unlocks */
|
|
static struct lruhash_entry* infra_find_ratedata(struct infra_cache* infra,
|
|
uint8_t* name, size_t namelen, int wr)
|
|
{
|
|
struct rate_key key;
|
|
hashvalue_type h = dname_query_hash(name, 0xab);
|
|
memset(&key, 0, sizeof(key));
|
|
key.name = name;
|
|
key.namelen = namelen;
|
|
key.entry.hash = h;
|
|
return slabhash_lookup(infra->domain_rates, h, &key, wr);
|
|
}
|
|
|
|
/** find data item in array for ip addresses */
|
|
struct lruhash_entry* infra_find_ip_ratedata(struct infra_cache* infra,
|
|
struct comm_reply* repinfo, int wr)
|
|
{
|
|
struct ip_rate_key key;
|
|
hashvalue_type h = hash_addr(&(repinfo->addr),
|
|
repinfo->addrlen, 0);
|
|
memset(&key, 0, sizeof(key));
|
|
key.addr = repinfo->addr;
|
|
key.addrlen = repinfo->addrlen;
|
|
key.entry.hash = h;
|
|
return slabhash_lookup(infra->client_ip_rates, h, &key, wr);
|
|
}
|
|
|
|
/** create rate data item for name, number 1 in now */
|
|
static void infra_create_ratedata(struct infra_cache* infra,
|
|
uint8_t* name, size_t namelen, time_t timenow)
|
|
{
|
|
hashvalue_type h = dname_query_hash(name, 0xab);
|
|
struct rate_key* k = (struct rate_key*)calloc(1, sizeof(*k));
|
|
struct rate_data* d = (struct rate_data*)calloc(1, sizeof(*d));
|
|
if(!k || !d) {
|
|
free(k);
|
|
free(d);
|
|
return; /* alloc failure */
|
|
}
|
|
k->namelen = namelen;
|
|
k->name = memdup(name, namelen);
|
|
if(!k->name) {
|
|
free(k);
|
|
free(d);
|
|
return; /* alloc failure */
|
|
}
|
|
lock_rw_init(&k->entry.lock);
|
|
k->entry.hash = h;
|
|
k->entry.key = k;
|
|
k->entry.data = d;
|
|
d->qps[0] = 1;
|
|
d->timestamp[0] = timenow;
|
|
slabhash_insert(infra->domain_rates, h, &k->entry, d, NULL);
|
|
}
|
|
|
|
/** create rate data item for ip address */
|
|
static void infra_ip_create_ratedata(struct infra_cache* infra,
|
|
struct comm_reply* repinfo, time_t timenow)
|
|
{
|
|
hashvalue_type h = hash_addr(&(repinfo->addr),
|
|
repinfo->addrlen, 0);
|
|
struct ip_rate_key* k = (struct ip_rate_key*)calloc(1, sizeof(*k));
|
|
struct ip_rate_data* d = (struct ip_rate_data*)calloc(1, sizeof(*d));
|
|
if(!k || !d) {
|
|
free(k);
|
|
free(d);
|
|
return; /* alloc failure */
|
|
}
|
|
k->addr = repinfo->addr;
|
|
k->addrlen = repinfo->addrlen;
|
|
lock_rw_init(&k->entry.lock);
|
|
k->entry.hash = h;
|
|
k->entry.key = k;
|
|
k->entry.data = d;
|
|
d->qps[0] = 1;
|
|
d->timestamp[0] = timenow;
|
|
slabhash_insert(infra->client_ip_rates, h, &k->entry, d, NULL);
|
|
}
|
|
|
|
/** find the second and return its rate counter, if none, remove oldest */
|
|
static int* infra_rate_find_second(void* data, time_t t)
|
|
{
|
|
struct rate_data* d = (struct rate_data*)data;
|
|
int i, oldest;
|
|
for(i=0; i<RATE_WINDOW; i++) {
|
|
if(d->timestamp[i] == t)
|
|
return &(d->qps[i]);
|
|
}
|
|
/* remove oldest timestamp, and insert it at t with 0 qps */
|
|
oldest = 0;
|
|
for(i=0; i<RATE_WINDOW; i++) {
|
|
if(d->timestamp[i] < d->timestamp[oldest])
|
|
oldest = i;
|
|
}
|
|
d->timestamp[oldest] = t;
|
|
d->qps[oldest] = 0;
|
|
return &(d->qps[oldest]);
|
|
}
|
|
|
|
int infra_rate_max(void* data, time_t now)
|
|
{
|
|
struct rate_data* d = (struct rate_data*)data;
|
|
int i, max = 0;
|
|
for(i=0; i<RATE_WINDOW; i++) {
|
|
if(now-d->timestamp[i] <= RATE_WINDOW) {
|
|
if(d->qps[i] > max)
|
|
max = d->qps[i];
|
|
}
|
|
}
|
|
return max;
|
|
}
|
|
|
|
int infra_ratelimit_inc(struct infra_cache* infra, uint8_t* name,
|
|
size_t namelen, time_t timenow)
|
|
{
|
|
int lim, max;
|
|
struct lruhash_entry* entry;
|
|
|
|
if(!infra_dp_ratelimit)
|
|
return 1; /* not enabled */
|
|
|
|
/* find ratelimit */
|
|
lim = infra_find_ratelimit(infra, name, namelen);
|
|
|
|
/* find or insert ratedata */
|
|
entry = infra_find_ratedata(infra, name, namelen, 1);
|
|
if(entry) {
|
|
int premax = infra_rate_max(entry->data, timenow);
|
|
int* cur = infra_rate_find_second(entry->data, timenow);
|
|
(*cur)++;
|
|
max = infra_rate_max(entry->data, timenow);
|
|
lock_rw_unlock(&entry->lock);
|
|
|
|
if(premax < lim && max >= lim) {
|
|
char buf[257];
|
|
dname_str(name, buf);
|
|
verbose(VERB_OPS, "ratelimit exceeded %s %d", buf, lim);
|
|
}
|
|
return (max < lim);
|
|
}
|
|
|
|
/* create */
|
|
infra_create_ratedata(infra, name, namelen, timenow);
|
|
return (1 < lim);
|
|
}
|
|
|
|
void infra_ratelimit_dec(struct infra_cache* infra, uint8_t* name,
|
|
size_t namelen, time_t timenow)
|
|
{
|
|
struct lruhash_entry* entry;
|
|
int* cur;
|
|
if(!infra_dp_ratelimit)
|
|
return; /* not enabled */
|
|
entry = infra_find_ratedata(infra, name, namelen, 1);
|
|
if(!entry) return; /* not cached */
|
|
cur = infra_rate_find_second(entry->data, timenow);
|
|
if((*cur) > 0)
|
|
(*cur)--;
|
|
lock_rw_unlock(&entry->lock);
|
|
}
|
|
|
|
int infra_ratelimit_exceeded(struct infra_cache* infra, uint8_t* name,
|
|
size_t namelen, time_t timenow)
|
|
{
|
|
struct lruhash_entry* entry;
|
|
int lim, max;
|
|
if(!infra_dp_ratelimit)
|
|
return 0; /* not enabled */
|
|
|
|
/* find ratelimit */
|
|
lim = infra_find_ratelimit(infra, name, namelen);
|
|
|
|
/* find current rate */
|
|
entry = infra_find_ratedata(infra, name, namelen, 0);
|
|
if(!entry)
|
|
return 0; /* not cached */
|
|
max = infra_rate_max(entry->data, timenow);
|
|
lock_rw_unlock(&entry->lock);
|
|
|
|
return (max >= lim);
|
|
}
|
|
|
|
size_t
|
|
infra_get_mem(struct infra_cache* infra)
|
|
{
|
|
size_t s = sizeof(*infra) + slabhash_get_mem(infra->hosts);
|
|
if(infra->domain_rates) s += slabhash_get_mem(infra->domain_rates);
|
|
if(infra->client_ip_rates) s += slabhash_get_mem(infra->client_ip_rates);
|
|
/* ignore domain_limits because walk through tree is big */
|
|
return s;
|
|
}
|
|
|
|
int infra_ip_ratelimit_inc(struct infra_cache* infra,
|
|
struct comm_reply* repinfo, time_t timenow)
|
|
{
|
|
int max;
|
|
struct lruhash_entry* entry;
|
|
|
|
/* not enabled */
|
|
if(!infra_ip_ratelimit) {
|
|
return 1;
|
|
}
|
|
/* find or insert ratedata */
|
|
entry = infra_find_ip_ratedata(infra, repinfo, 1);
|
|
if(entry) {
|
|
int premax = infra_rate_max(entry->data, timenow);
|
|
int* cur = infra_rate_find_second(entry->data, timenow);
|
|
(*cur)++;
|
|
max = infra_rate_max(entry->data, timenow);
|
|
lock_rw_unlock(&entry->lock);
|
|
|
|
if(premax < infra_ip_ratelimit && max >= infra_ip_ratelimit) {
|
|
char client_ip[128];
|
|
addr_to_str((struct sockaddr_storage *)&repinfo->addr,
|
|
repinfo->addrlen, client_ip, sizeof(client_ip));
|
|
verbose(VERB_OPS, "ratelimit exceeded %s %d", client_ip,
|
|
infra_ip_ratelimit);
|
|
}
|
|
return (max <= infra_ip_ratelimit);
|
|
}
|
|
|
|
/* create */
|
|
infra_ip_create_ratedata(infra, repinfo, timenow);
|
|
return 1;
|
|
}
|