DTLS can handle out of order record delivery. Additionally since
handshake messages can be bigger than will fit into a single packet, the
messages can be fragmented across multiple records (as with normal TLS).
That means that the messages can arrive mixed up, and we have to
reassemble them. We keep a queue of buffered messages that are "from the
future", i.e. messages we're not ready to deal with yet but have arrived
early. The messages held there may not be full yet - they could be one
or more fragments that are still in the process of being reassembled.
The code assumes that we will eventually complete the reassembly and
when that occurs the complete message is removed from the queue at the
point that we need to use it.
However, DTLS is also tolerant of packet loss. To get around that DTLS
messages can be retransmitted. If we receive a full (non-fragmented)
message from the peer after previously having received a fragment of
that message, then we ignore the message in the queue and just use the
non-fragmented version. At that point the queued message will never get
removed.
Additionally the peer could send "future" messages that we never get to
in order to complete the handshake. Each message has a sequence number
(starting from 0). We will accept a message fragment for the current
message sequence number, or for any sequence up to 10 into the future.
However if the Finished message has a sequence number of 2, anything
greater than that in the queue is just left there.
So, in those two ways we can end up with "orphaned" data in the queue
that will never get removed - except when the connection is closed. At
that point all the queues are flushed.
An attacker could seek to exploit this by filling up the queues with
lots of large messages that are never going to be used in order to
attempt a DoS by memory exhaustion.
I will assume that we are only concerned with servers here. It does not
seem reasonable to be concerned about a memory exhaustion attack on a
client. They are unlikely to process enough connections for this to be
an issue.
A "long" handshake with many messages might be 5 messages long (in the
incoming direction), e.g. ClientHello, Certificate, ClientKeyExchange,
CertificateVerify, Finished. So this would be message sequence numbers 0
to 4. Additionally we can buffer up to 10 messages in the future.
Therefore the maximum number of messages that an attacker could send
that could get orphaned would typically be 15.
The maximum size that a DTLS message is allowed to be is defined by
max_cert_list, which by default is 100k. Therefore the maximum amount of
"orphaned" memory per connection is 1500k.
Message sequence numbers get reset after the Finished message, so
renegotiation will not extend the maximum number of messages that can be
orphaned per connection.
As noted above, the queues do get cleared when the connection is closed.
Therefore in order to mount an effective attack, an attacker would have
to open many simultaneous connections.
Issue reported by Quan Luo.
CVE-2016-2179
Reviewed-by: Richard Levitte <levitte@openssl.org>
In DTLS if an IO retry occurs during writing of a fragmented ClientHello
then we can end up reseting the finish mac variables on the retry, which
causes a handshake failure. We should only reset on the first attempt not
on retries.
Thanks to BoringSSL for reporting this issue.
RT#4119
Reviewed-by: Tim Hudson <tjh@openssl.org>
(cherry picked from commit 15a7164eb7)
A DTLS client will abort a handshake if the server attempts to renew the
session ticket. This is caused by a state machine discrepancy between DTLS
and TLS discovered during the state machine rewrite work.
The bug can be demonstrated as follows:
Start a DTLS s_server instance:
openssl s_server -dtls
Start a client and obtain a session but no ticket:
openssl s_client -dtls -sess_out session.pem -no_ticket
Now start a client reusing the session, but allow a ticket:
openssl s_client -dtls -sess_in session.pem
The client will abort the handshake.
Reviewed-by: Tim Hudson <tjh@openssl.org>
(cherry picked from commit ee4ffd6fcc)
Conflicts:
ssl/d1_clnt.c
There are some missing return value checks in the SCTP code. In master this
was causing a compilation failure when config'd with
"--strict-warnings sctp".
Reviewed-by: Tim Hudson <tjh@openssl.org>
(cherry picked from commit d8e8590ed9)
Ensure all fatal errors transition into the new error state for DTLS.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(cherry picked from commit cefc93910c)
Conflicts:
ssl/d1_srvr.c
Conflicts:
ssl/d1_srvr.c
This should be a one off operation (subsequent invokation of the
script should not move them)
This commit is for the 1.0.1 changes
Reviewed-by: Tim Hudson <tjh@openssl.org>
once the ChangeCipherSpec message is received. Previously, the server would
set the flag once at SSL3_ST_SR_CERT_VRFY and again at SSL3_ST_SR_FINISHED.
This would allow a second CCS to arrive and would corrupt the server state.
(Because the first CCS would latch the correct keys and subsequent CCS
messages would have to be encrypted, a MitM attacker cannot exploit this,
though.)
Thanks to Joeri de Ruiter for reporting this issue.
Reviewed-by: Matt Caswell <matt@openssl.org>
(cherry picked from commit e94a6c0ede)
Conflicts:
CHANGES
ssl/s3_srvr.c
The client sends a session ID with the session ticket, and uses
the returned ID to detect resumption, so we do not need to peek
at handshake messages: s->hit tells us explicitly if we're resuming.
An equivalent change was independently made in BoringSSL, see commit
407886f589cf2dbaed82db0a44173036c3bc3317.
Reviewed-by: Matt Caswell <matt@openssl.org>
(cherry picked from commit 980bc1ec61)
Conflicts:
ssl/d1_clnt.c
ssl/s3_clnt.c
PR: 2808
With DTLS/SCTP the SCTP extension SCTP-AUTH is used to protect DATA and
FORWARD-TSN chunks. The key for this extension is derived from the
master secret and changed with the next ChangeCipherSpec, whenever a new
key has been negotiated. The following Finished then already uses the
new key. Unfortunately, the ChangeCipherSpec and Finished are part of
the same flight as the ClientKeyExchange, which is necessary for the
computation of the new secret. Hence, these messages are sent
immediately following each other, leaving the server very little time to
compute the new secret and pass it to SCTP before the finished arrives.
So the Finished is likely to be discarded by SCTP and a retransmission
becomes necessary. To prevent this issue, the Finished of the client is
still sent with the old key.
(cherry picked from commit 9fb523adce)
(cherry picked from commit b9ef52b078)
Instead, send random bytes.
While the gmt_unix_time record was added in an ostensible attempt to
mitigate the dangers of a bad RNG, its presence leaks the host's view
of the current time in the clear. This minor leak can help
fingerprint TLS instances across networks and protocols... and what's
worse, it's doubtful thet the gmt_unix_time record does any good at
all for its intended purpose, since:
* It's quite possible to open two TLS connections in one second.
* If the PRNG output is prone to repeat itself, ephemeral
* handshakes (and who knows what else besides) are broken.
Submitted by: steve@openssl.org
More robust fix and workaround for PR#1949. Don't try to work out if there
is any write pending data as this can be unreliable: always flush.
This change resolves a number of problems and obviates multiple kludges.
A new feature is that you can now say "AES256" or "AES128" (not just
"AES", which enables both).
In some cases the ciphersuite list generated from a given string is
affected by this change. I hope this is just in those cases where the
previous behaviour did not make sense.
