2002-02-11 10:05:58 +00:00
/*
protocol_key . c - - handle the meta - protocol , key exchange
2006-04-26 13:52:58 +00:00
Copyright ( C ) 1999 - 2005 Ivo Timmermans ,
2014-12-24 21:15:40 +00:00
2000 - 2014 Guus Sliepen < guus @ tinc - vpn . org >
2002-02-11 10:05:58 +00:00
This program is free software ; you can redistribute it and / or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation ; either version 2 of the License , or
( at your option ) any later version .
This program is distributed in the hope that it will be useful ,
but WITHOUT ANY WARRANTY ; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE . See the
GNU General Public License for more details .
2009-09-24 22:01:00 +00:00
You should have received a copy of the GNU General Public License along
with this program ; if not , write to the Free Software Foundation , Inc . ,
51 Franklin Street , Fifth Floor , Boston , MA 02110 - 1301 USA .
2002-02-11 10:05:58 +00:00
*/
2003-07-17 15:06:27 +00:00
# include "system.h"
2002-02-11 10:05:58 +00:00
2008-12-11 14:44:44 +00:00
# include "cipher.h"
2003-07-17 15:06:27 +00:00
# include "connection.h"
2009-06-05 21:03:28 +00:00
# include "crypto.h"
2003-07-17 15:06:27 +00:00
# include "logger.h"
2002-02-11 10:05:58 +00:00
# include "net.h"
# include "netutl.h"
# include "node.h"
2011-07-03 13:59:49 +00:00
# include "prf.h"
2003-07-17 15:06:27 +00:00
# include "protocol.h"
2012-07-30 16:36:59 +00:00
# include "sptps.h"
2003-07-17 15:06:27 +00:00
# include "utils.h"
# include "xalloc.h"
2002-02-11 10:05:58 +00:00
2008-12-11 14:44:44 +00:00
static bool mykeyused = false ;
2002-02-11 10:05:58 +00:00
2011-05-28 01:57:20 +00:00
void send_key_changed ( void ) {
2015-05-17 18:23:12 +00:00
# ifndef DISABLE_LEGACY
2012-02-20 16:12:48 +00:00
send_request ( everyone , " %d %x %s " , KEY_CHANGED , rand ( ) , myself - > name ) ;
2002-09-09 21:25:28 +00:00
2010-02-03 10:18:46 +00:00
/* Immediately send new keys to directly connected nodes to keep UDP mappings alive */
2002-09-09 21:25:28 +00:00
2012-10-07 22:35:38 +00:00
for list_each ( connection_t , c , connection_list )
2014-07-12 10:57:03 +00:00
if ( c - > edge & & c - > node & & c - > node - > status . reachable & & ! c - > node - > status . sptps )
2012-10-07 22:35:38 +00:00
send_ans_key ( c - > node ) ;
2015-05-17 18:23:12 +00:00
# endif
2012-07-30 16:36:59 +00:00
/* Force key exchange for connections using SPTPS */
if ( experimental ) {
2012-10-07 22:35:38 +00:00
for splay_each ( node_t , n , node_tree )
2012-07-31 19:43:49 +00:00
if ( n - > status . reachable & & n - > status . validkey & & n - > status . sptps )
2012-07-30 16:36:59 +00:00
sptps_force_kex ( & n - > sptps ) ;
2010-02-03 10:18:46 +00:00
}
2002-02-11 10:05:58 +00:00
}
2012-05-08 14:44:15 +00:00
bool key_changed_h ( connection_t * c , const char * request ) {
2002-09-09 21:25:28 +00:00
char name [ MAX_STRING_SIZE ] ;
node_t * n ;
2007-05-19 22:23:02 +00:00
if ( sscanf ( request , " %*d %*x " MAX_STRING , name ) ! = 1 ) {
2012-02-26 17:37:36 +00:00
logger ( DEBUG_ALWAYS , LOG_ERR , " Got bad %s from %s (%s) " , " KEY_CHANGED " ,
2002-09-09 21:25:28 +00:00
c - > name , c - > hostname ) ;
2003-07-22 20:55:21 +00:00
return false ;
2002-09-09 21:25:28 +00:00
}
2007-05-19 22:23:02 +00:00
if ( seen_request ( request ) )
2003-07-22 20:55:21 +00:00
return true ;
2002-09-09 21:25:28 +00:00
n = lookup_node ( name ) ;
if ( ! n ) {
2012-02-26 17:37:36 +00:00
logger ( DEBUG_ALWAYS , LOG_ERR , " Got %s from %s (%s) origin %s which does not exist " ,
2002-09-09 21:25:28 +00:00
" KEY_CHANGED " , c - > name , c - > hostname , name ) ;
2010-01-23 17:48:01 +00:00
return true ;
2002-09-09 21:25:28 +00:00
}
2012-07-31 19:43:49 +00:00
if ( ! n - > status . sptps ) {
2012-07-30 16:36:59 +00:00
n - > status . validkey = false ;
n - > last_req_key = 0 ;
}
2002-09-09 21:25:28 +00:00
/* Tell the others */
2003-11-17 15:30:18 +00:00
if ( ! tunnelserver )
2007-05-19 22:23:02 +00:00
forward_request ( c , request ) ;
2002-09-09 21:25:28 +00:00
2003-07-22 20:55:21 +00:00
return true ;
2002-02-11 10:05:58 +00:00
}
2015-05-09 16:54:34 +00:00
static bool send_sptps_data_myself ( void * handle , uint8_t type , const void * data , size_t len ) {
return send_sptps_data ( handle , myself , type , data , len ) ;
}
2014-12-24 21:15:40 +00:00
static bool send_initial_sptps_data ( void * handle , uint8_t type , const void * data , size_t len ) {
2012-07-30 16:36:59 +00:00
node_t * to = handle ;
2015-05-09 16:54:34 +00:00
to - > sptps . send_data = send_sptps_data_myself ;
2012-07-30 16:36:59 +00:00
char buf [ len * 4 / 3 + 5 ] ;
b64encode ( data , buf , len ) ;
return send_request ( to - > nexthop - > connection , " %d %s %s %d %s " , REQ_KEY , myself - > name , to - > name , REQ_KEY , buf ) ;
}
2009-06-05 21:03:28 +00:00
bool send_req_key ( node_t * to ) {
2012-07-31 19:43:49 +00:00
if ( to - > status . sptps ) {
2012-07-30 16:36:59 +00:00
if ( ! node_read_ecdsa_public_key ( to ) ) {
2014-05-18 18:47:04 +00:00
logger ( DEBUG_PROTOCOL , LOG_DEBUG , " No Ed25519 key known for %s (%s) " , to - > name , to - > hostname ) ;
2012-07-19 23:02:51 +00:00
send_request ( to - > nexthop - > connection , " %d %s %s %d " , REQ_KEY , myself - > name , to - > name , REQ_PUBKEY ) ;
2012-07-30 16:36:59 +00:00
return true ;
}
2012-10-14 12:33:54 +00:00
2012-10-10 15:17:49 +00:00
char label [ 25 + strlen ( myself - > name ) + strlen ( to - > name ) ] ;
2012-07-30 16:36:59 +00:00
snprintf ( label , sizeof label , " tinc UDP key expansion %s %s " , myself - > name , to - > name ) ;
2012-08-02 15:23:51 +00:00
sptps_stop ( & to - > sptps ) ;
to - > status . validkey = false ;
2012-10-07 11:31:19 +00:00
to - > status . waitingforkey = true ;
2013-03-08 13:11:15 +00:00
to - > last_req_key = now . tv_sec ;
2012-08-02 15:44:59 +00:00
to - > incompression = myself - > incompression ;
2012-10-10 15:17:49 +00:00
return sptps_start ( & to - > sptps , to , true , true , myself - > connection - > ecdsa , to - > ecdsa , label , sizeof label , send_initial_sptps_data , receive_sptps_record ) ;
2012-07-19 23:02:51 +00:00
}
2012-07-30 16:36:59 +00:00
2012-07-19 23:02:51 +00:00
return send_request ( to - > nexthop - > connection , " %d %s %s " , REQ_KEY , myself - > name , to - > name ) ;
}
/* REQ_KEY is overloaded to allow arbitrary requests to be routed between two nodes. */
2015-05-09 17:09:23 +00:00
static bool req_key_ext_h ( connection_t * c , const char * request , node_t * from , node_t * to , int reqno ) {
/* If this is a SPTPS packet, see if sending UDP info helps.
Note that we only do this if we ' re the destination or the static relay ;
otherwise every hop would initiate its own UDP info message , resulting in elevated chatter . */
2015-05-10 17:05:19 +00:00
if ( ( reqno = = REQ_KEY | | reqno = = SPTPS_PACKET ) & & to - > via = = myself )
2015-05-09 17:09:23 +00:00
send_udp_info ( myself , from ) ;
2015-05-10 17:05:19 +00:00
if ( reqno = = SPTPS_PACKET ) {
2015-05-09 17:09:23 +00:00
/* This is a SPTPS data packet. */
char buf [ MAX_STRING_SIZE ] ;
int len ;
if ( sscanf ( request , " %*d %*s %*s %*d " MAX_STRING , buf ) ! = 1 | | ! ( len = b64decode ( buf , buf , strlen ( buf ) ) ) ) {
2015-05-10 17:05:19 +00:00
logger ( DEBUG_ALWAYS , LOG_ERR , " Got bad %s from %s (%s) to %s (%s): %s " , " SPTPS_PACKET " , from - > name , from - > hostname , to - > name , to - > hostname , " invalid SPTPS data " ) ;
2015-05-09 17:09:23 +00:00
return true ;
}
if ( to ! = myself ) {
/* We don't just forward the request, because we want to use UDP if it's available. */
send_sptps_data ( to , from , 0 , buf , len ) ;
try_tx ( to , true ) ;
} else {
/* The packet is for us */
Proactively restart the SPTPS tunnel if we get receive errors.
There are a number of ways a SPTPS tunnel can get into a corrupt state.
For example, during key regeneration, the KEX and SIG messages from
other nodes might arrive out of order, which confuses the hell out of
the SPTPS code. Another possible scenario is not noticing another node
crashed and restarted because there was no point in time where the node
was seen completely disconnected from *all* nodes; this could result in
using the wrong (old) key. There are probably other scenarios which have
not even been considered yet. Distributed systems are hard.
When SPTPS got confused by a packet, it used to crash the entire
process; fortunately that was fixed by commit
2e7f68ad2b51648b89c4b5c61aeb4cec67c2fbbb. However, the error handling
(or lack thereof) leaves a lot to be desired. Currently, when SPTPS
encounters an error when receiving a packet, it just shrugs it off and
continues as if nothing happened. The problem is, sometimes getting
receive errors mean the tunnel is completely stuck and will not recover
on its own. In that case, the node will become unreachable - possibly
indefinitely.
The goal of this commit is to improve SPTPS error handling by taking
proactive action when an incoming packet triggers a failure, which is
often an indicator that the tunnel is stuck in some way. When that
happens, we simply restart SPTPS entirely, which should make the tunnel
recover quickly.
To prevent "storms" where two buggy nodes flood each other with invalid
packets and therefore spend all their time negotiating new tunnels, we
limit the frequency at which tunnel restarts happen to ten seconds.
It is likely this commit will solve the "Invalid KEX record length
during key regeneration" issue that has been seen in the wild. It is
difficult to be sure though because we do not have a full understanding
of all the possible conditions that can trigger this problem.
2015-05-17 17:50:11 +00:00
if ( ! sptps_receive_data ( & from - > sptps , buf , len ) ) {
/* Uh-oh. It might be that the tunnel is stuck in some corrupted state,
so let ' s restart SPTPS in case that helps . But don ' t do that too often
to prevent storms . */
if ( from - > last_req_key < now . tv_sec - 10 ) {
logger ( DEBUG_PROTOCOL , LOG_ERR , " Failed to decode TCP packet from %s (%s), restarting SPTPS " , from - > name , from - > hostname ) ;
send_req_key ( from ) ;
}
2015-05-09 17:09:23 +00:00
return true ;
}
send_mtu_info ( myself , from , MTU ) ;
}
return true ;
}
/* Requests that are not SPTPS data packets are forwarded as-is. */
if ( to ! = myself )
return send_request ( to - > nexthop - > connection , " %s " , request ) ;
/* The request is for us */
2012-07-19 23:02:51 +00:00
switch ( reqno ) {
case REQ_PUBKEY : {
2014-09-21 09:38:41 +00:00
if ( ! node_read_ecdsa_public_key ( from ) ) {
/* Request their key *before* we send our key back. Otherwise the first SPTPS packet from them will get dropped. */
logger ( DEBUG_PROTOCOL , LOG_DEBUG , " Preemptively requesting Ed25519 key for %s (%s) " , from - > name , from - > hostname ) ;
send_request ( from - > nexthop - > connection , " %d %s %s %d " , REQ_KEY , myself - > name , from - > name , REQ_PUBKEY ) ;
}
2013-05-01 15:17:22 +00:00
char * pubkey = ecdsa_get_base64_public_key ( myself - > connection - > ecdsa ) ;
2012-07-19 23:02:51 +00:00
send_request ( from - > nexthop - > connection , " %d %s %s %d %s " , REQ_KEY , myself - > name , from - > name , ANS_PUBKEY , pubkey ) ;
free ( pubkey ) ;
return true ;
}
case ANS_PUBKEY : {
if ( node_read_ecdsa_public_key ( from ) ) {
2012-09-28 15:51:48 +00:00
logger ( DEBUG_PROTOCOL , LOG_WARNING , " Got ANS_PUBKEY from %s (%s) even though we already have his pubkey " , from - > name , from - > hostname ) ;
2012-07-19 23:02:51 +00:00
return true ;
}
2012-07-30 16:36:59 +00:00
char pubkey [ MAX_STRING_SIZE ] ;
2013-05-01 15:17:22 +00:00
if ( sscanf ( request , " %*d %*s %*s %*d " MAX_STRING , pubkey ) ! = 1 | | ! ( from - > ecdsa = ecdsa_set_base64_public_key ( pubkey ) ) ) {
2012-07-19 23:02:51 +00:00
logger ( DEBUG_ALWAYS , LOG_ERR , " Got bad %s from %s (%s): %s " , " ANS_PUBKEY " , from - > name , from - > hostname , " invalid pubkey " ) ;
return true ;
}
2014-05-18 18:47:04 +00:00
logger ( DEBUG_PROTOCOL , LOG_INFO , " Learned Ed25519 public key from %s (%s) " , from - > name , from - > hostname ) ;
append_config_file ( from - > name , " Ed25519PublicKey " , pubkey ) ;
2012-07-19 23:02:51 +00:00
return true ;
}
2012-07-30 16:36:59 +00:00
case REQ_KEY : {
if ( ! node_read_ecdsa_public_key ( from ) ) {
2014-05-18 18:47:04 +00:00
logger ( DEBUG_PROTOCOL , LOG_DEBUG , " No Ed25519 key known for %s (%s) " , from - > name , from - > hostname ) ;
2012-07-30 16:36:59 +00:00
send_request ( from - > nexthop - > connection , " %d %s %s %d " , REQ_KEY , myself - > name , from - > name , REQ_PUBKEY ) ;
return true ;
}
2012-10-14 12:33:54 +00:00
if ( from - > sptps . label )
logger ( DEBUG_ALWAYS , LOG_DEBUG , " Got REQ_KEY from %s while we already started a SPTPS session! " , from - > name ) ;
2012-07-30 16:36:59 +00:00
char buf [ MAX_STRING_SIZE ] ;
2013-05-28 11:39:15 +00:00
int len ;
if ( sscanf ( request , " %*d %*s %*s %*d " MAX_STRING , buf ) ! = 1 | | ! ( len = b64decode ( buf , buf , strlen ( buf ) ) ) ) {
2012-10-14 12:33:54 +00:00
logger ( DEBUG_ALWAYS , LOG_ERR , " Got bad %s from %s (%s): %s " , " REQ_SPTPS_START " , from - > name , from - > hostname , " invalid SPTPS data " ) ;
2012-07-30 16:36:59 +00:00
return true ;
}
char label [ 25 + strlen ( from - > name ) + strlen ( myself - > name ) ] ;
snprintf ( label , sizeof label , " tinc UDP key expansion %s %s " , from - > name , myself - > name ) ;
2012-08-02 15:23:51 +00:00
sptps_stop ( & from - > sptps ) ;
from - > status . validkey = false ;
2012-10-07 11:31:19 +00:00
from - > status . waitingforkey = true ;
2013-03-08 13:11:15 +00:00
from - > last_req_key = now . tv_sec ;
2015-05-09 16:54:34 +00:00
sptps_start ( & from - > sptps , from , false , true , myself - > connection - > ecdsa , from - > ecdsa , label , sizeof label , send_sptps_data_myself , receive_sptps_record ) ;
2012-07-30 16:36:59 +00:00
sptps_receive_data ( & from - > sptps , buf , len ) ;
Add MTU_INFO protocol message.
In this commit, nodes use MTU_INFO messages to provide MTU information.
The issue this code is meant to address is the non-trivial problem of
finding the proper MTU when UDP SPTPS relays are involved. Currently,
tinc has no idea what the MTU looks like beyond the first relay, and
will arbitrarily use the first relay's MTU as the limit. This will fail
miserably if the MTU decreases after the first relay, forcing relays to
fall back to TCP. More generally, one should keep in mind that relay
paths can be arbitrarily complex, resulting in packets taking "epic
journeys" through the graph, switching back and forth between UDP (with
variable MTUs) and TCP multiple times along the path.
A solution that was considered consists in sending standard MTU probes
through the relays. This is inefficient (if there are 3 nodes on one
side of relay and 3 nodes on the other side, we end up with 3*3=9 MTU
discoveries taking place at the same time, while technically only
3+3=6 are needed) and would involve eyebrow-raising behaviors such as
probes being sent over TCP.
This commit implements an alternative solution, which consists in
the packet receiver sending MTU_INFO messages to the packet sender.
The message contains an MTU value which is set to maximum when the
message is originally sent. The message gets altered as it travels
through the metagraph, such that when the message arrives to the
destination, the MTU value contained in the message can be used to
send packets while making sure no relays will be forced to fall back to
TCP to deliver them.
The operating principles behind such a protocol message are similar to
how the UDP_INFO message works, but there is a key difference that
prevents us from simply reusing the same message: the UDP_INFO message
only cares about relay-to-relay links (i.e. it is sent between static
relays and the information it contains only makes sense between two
adjacent static relays), while the MTU_INFO cares about the end-to-end
MTU, including the entire relay path. Therefore, UDP_INFO messages stop
when they encounter static relays, while MTU_INFO messages don't stop
until they get to the original packet sender.
Note that, technically, the MTU that is obtained through this mechanism
can be slightly pessimistic, because it can be lowered by an
intermediate node that is not being used as a relay. Since nodes have no
way of knowing whether they'll be used as dynamic relays or not (and
have no say in the matter), this is not a trivial problem. That said,
this is highly unlikely to result in noticeable issues in realistic
scenarios.
2015-03-08 18:54:50 +00:00
send_mtu_info ( myself , from , MTU ) ;
2012-07-30 16:36:59 +00:00
return true ;
}
2012-07-19 23:02:51 +00:00
default :
logger ( DEBUG_ALWAYS , LOG_ERR , " Unknown extended REQ_KEY request from %s (%s): %s " , from - > name , from - > hostname , request ) ;
return true ;
}
2002-02-11 10:05:58 +00:00
}
2012-05-08 14:44:15 +00:00
bool req_key_h ( connection_t * c , const char * request ) {
2002-09-09 21:25:28 +00:00
char from_name [ MAX_STRING_SIZE ] ;
char to_name [ MAX_STRING_SIZE ] ;
node_t * from , * to ;
2012-07-19 23:02:51 +00:00
int reqno = 0 ;
2002-09-09 21:25:28 +00:00
2012-07-19 23:02:51 +00:00
if ( sscanf ( request , " %*d " MAX_STRING " " MAX_STRING " %d " , from_name , to_name , & reqno ) < 2 ) {
2012-02-26 17:37:36 +00:00
logger ( DEBUG_ALWAYS , LOG_ERR , " Got bad %s from %s (%s) " , " REQ_KEY " , c - > name ,
2002-09-09 21:25:28 +00:00
c - > hostname ) ;
2003-07-22 20:55:21 +00:00
return false ;
2002-09-09 21:25:28 +00:00
}
2010-01-23 17:48:01 +00:00
if ( ! check_id ( from_name ) | | ! check_id ( to_name ) ) {
2012-02-26 17:37:36 +00:00
logger ( DEBUG_ALWAYS , LOG_ERR , " Got bad %s from %s (%s): %s " , " REQ_KEY " , c - > name , c - > hostname , " invalid name " ) ;
2010-01-23 17:48:01 +00:00
return false ;
}
2002-09-09 21:25:28 +00:00
from = lookup_node ( from_name ) ;
if ( ! from ) {
2012-02-26 17:37:36 +00:00
logger ( DEBUG_ALWAYS , LOG_ERR , " Got %s from %s (%s) origin %s which does not exist in our connection list " ,
2002-09-09 21:25:28 +00:00
" REQ_KEY " , c - > name , c - > hostname , from_name ) ;
2010-01-23 17:48:01 +00:00
return true ;
2002-09-09 21:25:28 +00:00
}
to = lookup_node ( to_name ) ;
if ( ! to ) {
2012-02-26 17:37:36 +00:00
logger ( DEBUG_ALWAYS , LOG_ERR , " Got %s from %s (%s) destination %s which does not exist in our connection list " ,
2002-09-09 21:25:28 +00:00
" REQ_KEY " , c - > name , c - > hostname , to_name ) ;
2010-01-23 17:48:01 +00:00
return true ;
2002-09-09 21:25:28 +00:00
}
/* Check if this key request is for us */
2012-10-10 15:17:49 +00:00
if ( to = = myself ) { /* Yes */
2012-07-31 19:43:49 +00:00
/* Is this an extended REQ_KEY message? */
2012-07-19 23:02:51 +00:00
if ( experimental & & reqno )
2015-05-09 17:09:23 +00:00
return req_key_ext_h ( c , request , from , to , reqno ) ;
2012-07-19 23:02:51 +00:00
2012-07-31 19:43:49 +00:00
/* No, just send our key back */
2009-04-02 23:05:23 +00:00
send_ans_key ( from ) ;
2002-09-09 21:25:28 +00:00
} else {
2003-11-17 15:30:18 +00:00
if ( tunnelserver )
2010-01-23 17:48:01 +00:00
return true ;
2003-11-17 15:30:18 +00:00
2008-12-11 15:21:40 +00:00
if ( ! to - > status . reachable ) {
2012-09-28 15:51:48 +00:00
logger ( DEBUG_PROTOCOL , LOG_WARNING , " Got %s from %s (%s) destination %s which is not reachable " ,
2008-12-11 15:21:40 +00:00
" REQ_KEY " , c - > name , c - > hostname , to_name ) ;
return true ;
}
2015-05-09 17:09:23 +00:00
/* Is this an extended REQ_KEY message? */
if ( experimental & & reqno )
return req_key_ext_h ( c , request , from , to , reqno ) ;
2009-06-05 21:03:28 +00:00
send_request ( to - > nexthop - > connection , " %s " , request ) ;
2002-09-09 21:25:28 +00:00
}
2002-02-11 10:05:58 +00:00
2003-07-22 20:55:21 +00:00
return true ;
2002-02-11 10:05:58 +00:00
}
2009-06-05 21:03:28 +00:00
bool send_ans_key ( node_t * to ) {
2012-07-31 19:43:49 +00:00
if ( to - > status . sptps )
2012-07-30 16:36:59 +00:00
abort ( ) ;
2011-07-03 11:17:28 +00:00
2014-12-29 21:57:18 +00:00
# ifdef DISABLE_LEGACY
return false ;
# else
2014-05-18 19:51:42 +00:00
size_t keylen = myself - > incipher ? cipher_keylength ( myself - > incipher ) : 1 ;
2008-12-11 14:44:44 +00:00
char key [ keylen * 2 + 1 ] ;
2002-09-09 21:25:28 +00:00
2014-05-18 19:51:42 +00:00
randomize ( key , keylen ) ;
2013-05-01 15:17:22 +00:00
cipher_close ( to - > incipher ) ;
digest_close ( to - > indigest ) ;
2012-10-09 14:27:28 +00:00
2014-05-18 19:51:42 +00:00
if ( myself - > incipher ) {
to - > incipher = cipher_open_by_nid ( cipher_get_nid ( myself - > incipher ) ) ;
if ( ! to - > incipher )
abort ( ) ;
if ( ! cipher_set_key ( to - > incipher , key , false ) )
abort ( ) ;
}
2002-09-09 21:25:28 +00:00
2014-05-18 19:51:42 +00:00
if ( myself - > indigest ) {
to - > indigest = digest_open_by_nid ( digest_get_nid ( myself - > indigest ) , digest_length ( myself - > indigest ) ) ;
if ( ! to - > indigest )
abort ( ) ;
if ( ! digest_set_key ( to - > indigest , key , keylen ) )
abort ( ) ;
}
2013-05-01 15:17:22 +00:00
2014-05-18 19:51:42 +00:00
to - > incompression = myself - > incompression ;
2009-05-24 17:31:31 +00:00
2008-12-11 14:44:44 +00:00
bin2hex ( key , key , keylen ) ;
2002-09-09 21:25:28 +00:00
2009-05-24 17:31:31 +00:00
// Reset sequence number and late packet window
mykeyused = true ;
to - > received_seqno = 0 ;
2013-01-15 12:33:16 +00:00
to - > received = 0 ;
2010-11-13 18:05:50 +00:00
if ( replaywin ) memset ( to - > late , 0 , replaywin ) ;
2009-04-02 23:05:23 +00:00
2015-01-10 21:26:33 +00:00
to - > status . validkey_in = true ;
2012-07-21 10:51:53 +00:00
return send_request ( to - > nexthop - > connection , " %d %s %s %s %d %d %d %d " , ANS_KEY ,
2009-06-05 21:03:28 +00:00
myself - > name , to - > name , key ,
2013-05-01 15:17:22 +00:00
cipher_get_nid ( to - > incipher ) ,
digest_get_nid ( to - > indigest ) ,
( int ) digest_length ( to - > indigest ) ,
2009-06-05 21:03:28 +00:00
to - > incompression ) ;
2014-12-29 21:57:18 +00:00
# endif
2002-02-11 10:05:58 +00:00
}
2012-05-08 14:44:15 +00:00
bool ans_key_h ( connection_t * c , const char * request ) {
2002-09-09 21:25:28 +00:00
char from_name [ MAX_STRING_SIZE ] ;
char to_name [ MAX_STRING_SIZE ] ;
char key [ MAX_STRING_SIZE ] ;
2012-10-10 15:17:49 +00:00
char address [ MAX_STRING_SIZE ] = " " ;
char port [ MAX_STRING_SIZE ] = " " ;
2009-12-18 00:15:25 +00:00
int cipher , digest , maclength , compression , keylen ;
2002-09-09 21:25:28 +00:00
node_t * from , * to ;
2011-07-13 20:52:52 +00:00
if ( sscanf ( request , " %*d " MAX_STRING " " MAX_STRING " " MAX_STRING " %d %d %d %d " MAX_STRING " " MAX_STRING ,
2002-09-09 21:25:28 +00:00
from_name , to_name , key , & cipher , & digest , & maclength ,
2010-02-01 23:51:44 +00:00
& compression , address , port ) < 7 ) {
2012-02-26 17:37:36 +00:00
logger ( DEBUG_ALWAYS , LOG_ERR , " Got bad %s from %s (%s) " , " ANS_KEY " , c - > name ,
2002-09-09 21:25:28 +00:00
c - > hostname ) ;
2003-07-22 20:55:21 +00:00
return false ;
2002-02-11 10:05:58 +00:00
}
2002-09-09 21:25:28 +00:00
2010-01-23 17:48:01 +00:00
if ( ! check_id ( from_name ) | | ! check_id ( to_name ) ) {
2012-02-26 17:37:36 +00:00
logger ( DEBUG_ALWAYS , LOG_ERR , " Got bad %s from %s (%s): %s " , " ANS_KEY " , c - > name , c - > hostname , " invalid name " ) ;
2010-01-23 17:48:01 +00:00
return false ;
}
2002-09-09 21:25:28 +00:00
from = lookup_node ( from_name ) ;
if ( ! from ) {
2012-02-26 17:37:36 +00:00
logger ( DEBUG_ALWAYS , LOG_ERR , " Got %s from %s (%s) origin %s which does not exist in our connection list " ,
2002-09-09 21:25:28 +00:00
" ANS_KEY " , c - > name , c - > hostname , from_name ) ;
2010-01-23 17:48:01 +00:00
return true ;
2002-09-09 21:25:28 +00:00
}
to = lookup_node ( to_name ) ;
if ( ! to ) {
2012-02-26 17:37:36 +00:00
logger ( DEBUG_ALWAYS , LOG_ERR , " Got %s from %s (%s) destination %s which does not exist in our connection list " ,
2002-09-09 21:25:28 +00:00
" ANS_KEY " , c - > name , c - > hostname , to_name ) ;
2010-01-23 17:48:01 +00:00
return true ;
2002-02-11 10:05:58 +00:00
}
2002-09-09 21:25:28 +00:00
/* Forward it if necessary */
if ( to ! = myself ) {
2003-11-17 15:30:18 +00:00
if ( tunnelserver )
2010-01-23 17:48:01 +00:00
return true ;
2003-11-17 15:30:18 +00:00
2008-12-11 15:21:40 +00:00
if ( ! to - > status . reachable ) {
2012-02-26 17:37:36 +00:00
logger ( DEBUG_ALWAYS , LOG_WARNING , " Got %s from %s (%s) destination %s which is not reachable " ,
2008-12-11 15:21:40 +00:00
" ANS_KEY " , c - > name , c - > hostname , to_name ) ;
return true ;
}
2010-06-04 14:03:19 +00:00
if ( ! * address & & from - > address . sa . sa_family ! = AF_UNSPEC ) {
2010-04-17 10:33:15 +00:00
char * address , * port ;
2012-02-26 17:37:36 +00:00
logger ( DEBUG_PROTOCOL , LOG_DEBUG , " Appending reflexive UDP address to ANS_KEY from %s to %s " , from - > name , to - > name ) ;
2010-04-17 10:33:15 +00:00
sockaddr2str ( & from - > address , & address , & port ) ;
send_request ( to - > nexthop - > connection , " %s %s %s " , request , address , port ) ;
free ( address ) ;
free ( port ) ;
return true ;
}
2007-05-19 22:23:02 +00:00
return send_request ( to - > nexthop - > connection , " %s " , request ) ;
2002-02-11 10:05:58 +00:00
}
2002-09-09 21:25:28 +00:00
2014-12-29 21:57:18 +00:00
# ifndef DISABLE_LEGACY
2012-09-30 11:45:47 +00:00
/* Don't use key material until every check has passed. */
2013-05-01 15:17:22 +00:00
cipher_close ( from - > outcipher ) ;
digest_close ( from - > outdigest ) ;
2014-12-29 21:57:18 +00:00
# endif
2012-09-30 11:45:47 +00:00
from - > status . validkey = false ;
2012-08-02 15:44:59 +00:00
if ( compression < 0 | | compression > 11 ) {
logger ( DEBUG_ALWAYS , LOG_ERR , " Node %s (%s) uses bogus compression level! " , from - > name , from - > hostname ) ;
return true ;
}
from - > outcompression = compression ;
2012-07-30 16:36:59 +00:00
/* SPTPS or old-style key exchange? */
2012-07-31 19:43:49 +00:00
if ( from - > status . sptps ) {
2012-07-30 16:36:59 +00:00
char buf [ strlen ( key ) ] ;
int len = b64decode ( key , buf , strlen ( key ) ) ;
Proactively restart the SPTPS tunnel if we get receive errors.
There are a number of ways a SPTPS tunnel can get into a corrupt state.
For example, during key regeneration, the KEX and SIG messages from
other nodes might arrive out of order, which confuses the hell out of
the SPTPS code. Another possible scenario is not noticing another node
crashed and restarted because there was no point in time where the node
was seen completely disconnected from *all* nodes; this could result in
using the wrong (old) key. There are probably other scenarios which have
not even been considered yet. Distributed systems are hard.
When SPTPS got confused by a packet, it used to crash the entire
process; fortunately that was fixed by commit
2e7f68ad2b51648b89c4b5c61aeb4cec67c2fbbb. However, the error handling
(or lack thereof) leaves a lot to be desired. Currently, when SPTPS
encounters an error when receiving a packet, it just shrugs it off and
continues as if nothing happened. The problem is, sometimes getting
receive errors mean the tunnel is completely stuck and will not recover
on its own. In that case, the node will become unreachable - possibly
indefinitely.
The goal of this commit is to improve SPTPS error handling by taking
proactive action when an incoming packet triggers a failure, which is
often an indicator that the tunnel is stuck in some way. When that
happens, we simply restart SPTPS entirely, which should make the tunnel
recover quickly.
To prevent "storms" where two buggy nodes flood each other with invalid
packets and therefore spend all their time negotiating new tunnels, we
limit the frequency at which tunnel restarts happen to ten seconds.
It is likely this commit will solve the "Invalid KEX record length
during key regeneration" issue that has been seen in the wild. It is
difficult to be sure though because we do not have a full understanding
of all the possible conditions that can trigger this problem.
2015-05-17 17:50:11 +00:00
if ( ! len | | ! sptps_receive_data ( & from - > sptps , buf , len ) ) {
/* Uh-oh. It might be that the tunnel is stuck in some corrupted state,
so let ' s restart SPTPS in case that helps . But don ' t do that too often
to prevent storms .
Note that simply relying on handshake timeout is not enough , because
that doesn ' t apply to key regeneration . */
if ( from - > last_req_key < now . tv_sec - 10 ) {
logger ( DEBUG_PROTOCOL , LOG_ERR , " Failed to decode handshake TCP packet from %s (%s), restarting SPTPS " , from - > name , from - > hostname ) ;
send_req_key ( from ) ;
}
return true ;
}
2012-07-30 16:36:59 +00:00
if ( from - > status . validkey ) {
if ( * address & & * port ) {
logger ( DEBUG_PROTOCOL , LOG_DEBUG , " Using reflexive UDP address from %s: %s port %s " , from - > name , address , port ) ;
sockaddr_t sa = str2sockaddr ( address , port ) ;
update_node_udp ( from , & sa ) ;
}
}
Add MTU_INFO protocol message.
In this commit, nodes use MTU_INFO messages to provide MTU information.
The issue this code is meant to address is the non-trivial problem of
finding the proper MTU when UDP SPTPS relays are involved. Currently,
tinc has no idea what the MTU looks like beyond the first relay, and
will arbitrarily use the first relay's MTU as the limit. This will fail
miserably if the MTU decreases after the first relay, forcing relays to
fall back to TCP. More generally, one should keep in mind that relay
paths can be arbitrarily complex, resulting in packets taking "epic
journeys" through the graph, switching back and forth between UDP (with
variable MTUs) and TCP multiple times along the path.
A solution that was considered consists in sending standard MTU probes
through the relays. This is inefficient (if there are 3 nodes on one
side of relay and 3 nodes on the other side, we end up with 3*3=9 MTU
discoveries taking place at the same time, while technically only
3+3=6 are needed) and would involve eyebrow-raising behaviors such as
probes being sent over TCP.
This commit implements an alternative solution, which consists in
the packet receiver sending MTU_INFO messages to the packet sender.
The message contains an MTU value which is set to maximum when the
message is originally sent. The message gets altered as it travels
through the metagraph, such that when the message arrives to the
destination, the MTU value contained in the message can be used to
send packets while making sure no relays will be forced to fall back to
TCP to deliver them.
The operating principles behind such a protocol message are similar to
how the UDP_INFO message works, but there is a key difference that
prevents us from simply reusing the same message: the UDP_INFO message
only cares about relay-to-relay links (i.e. it is sent between static
relays and the information it contains only makes sense between two
adjacent static relays), while the MTU_INFO cares about the end-to-end
MTU, including the entire relay path. Therefore, UDP_INFO messages stop
when they encounter static relays, while MTU_INFO messages don't stop
until they get to the original packet sender.
Note that, technically, the MTU that is obtained through this mechanism
can be slightly pessimistic, because it can be lowered by an
intermediate node that is not being used as a relay. Since nodes have no
way of knowing whether they'll be used as dynamic relays or not (and
have no say in the matter), this is not a trivial problem. That said,
this is highly unlikely to result in noticeable issues in realistic
scenarios.
2015-03-08 18:54:50 +00:00
send_mtu_info ( myself , from , MTU ) ;
2012-07-30 16:36:59 +00:00
return true ;
}
2014-12-29 21:57:18 +00:00
# ifdef DISABLE_LEGACY
2014-12-30 09:56:30 +00:00
logger ( DEBUG_ALWAYS , LOG_ERR , " Node %s (%s) uses legacy protocol! " , from - > name , from - > hostname ) ;
2014-12-29 21:57:18 +00:00
return false ;
# else
2002-09-09 21:25:28 +00:00
/* Check and lookup cipher and digest algorithms */
2013-11-28 13:19:55 +00:00
if ( cipher ) {
if ( ! ( from - > outcipher = cipher_open_by_nid ( cipher ) ) ) {
logger ( DEBUG_ALWAYS , LOG_ERR , " Node %s (%s) uses unknown cipher! " , from - > name , from - > hostname ) ;
return false ;
}
} else {
from - > outcipher = NULL ;
2008-12-11 14:44:44 +00:00
}
2007-11-07 02:47:05 +00:00
2013-11-28 13:19:55 +00:00
if ( digest ) {
if ( ! ( from - > outdigest = digest_open_by_nid ( digest , maclength ) ) ) {
logger ( DEBUG_ALWAYS , LOG_ERR , " Node %s (%s) uses unknown digest! " , from - > name , from - > hostname ) ;
return false ;
}
} else {
from - > outdigest = NULL ;
2008-12-11 14:44:44 +00:00
}
2007-11-07 02:47:05 +00:00
2013-05-01 15:17:22 +00:00
if ( maclength ! = digest_length ( from - > outdigest ) ) {
2012-02-26 17:37:36 +00:00
logger ( DEBUG_ALWAYS , LOG_ERR , " Node %s (%s) uses bogus MAC length! " , from - > name , from - > hostname ) ;
2008-12-11 14:44:44 +00:00
return false ;
2002-09-09 21:25:28 +00:00
}
2012-07-30 16:36:59 +00:00
/* Process key */
2011-07-16 18:21:44 +00:00
2012-07-30 16:36:59 +00:00
keylen = hex2bin ( key , key , sizeof key ) ;
2011-07-03 11:17:28 +00:00
2014-05-18 19:51:42 +00:00
if ( keylen ! = ( from - > outcipher ? cipher_keylength ( from - > outcipher ) : 1 ) ) {
2012-07-30 16:36:59 +00:00
logger ( DEBUG_ALWAYS , LOG_ERR , " Node %s (%s) uses wrong keylength! " , from - > name , from - > hostname ) ;
return true ;
}
2008-12-11 14:44:44 +00:00
2012-07-30 16:36:59 +00:00
/* Update our copy of the origin's packet key */
2011-07-03 11:17:28 +00:00
2014-05-18 19:51:42 +00:00
if ( from - > outcipher & & ! cipher_set_key ( from - > outcipher , key , true ) )
2013-05-10 18:30:47 +00:00
return false ;
2014-05-18 19:51:42 +00:00
if ( from - > outdigest & & ! digest_set_key ( from - > outdigest , key , keylen ) )
2013-05-10 18:30:47 +00:00
return false ;
2008-12-11 14:44:44 +00:00
from - > status . validkey = true ;
from - > sent_seqno = 0 ;
2003-10-11 12:16:13 +00:00
2010-02-01 23:51:44 +00:00
if ( * address & & * port ) {
2012-02-26 17:37:36 +00:00
logger ( DEBUG_PROTOCOL , LOG_DEBUG , " Using reflexive UDP address from %s: %s port %s " , from - > name , address , port ) ;
2010-02-01 23:51:44 +00:00
sockaddr_t sa = str2sockaddr ( address , port ) ;
update_node_udp ( from , & sa ) ;
}
2003-07-22 20:55:21 +00:00
return true ;
2014-12-29 21:57:18 +00:00
# endif
2002-02-11 10:05:58 +00:00
}