If while calling SSL_peek() we read an empty record then we go into an
infinite loop, continually trying to read data from the empty record and
never making any progress. This could be exploited by a malicious peer in
a Denial Of Service attack.
CVE-2016-6305
GitHub Issue #1563
Reviewed-by: Rich Salz <rsalz@openssl.org>
Certain warning alerts are ignored if they are received. This can mean that
no progress will be made if one peer continually sends those warning alerts.
Implement a count so that we abort the connection if we receive too many.
Issue reported by Shi Lei.
Reviewed-by: Rich Salz <rsalz@openssl.org>
This reverts commit 77a6be4dfc.
There were some unexpected side effects to this commit, e.g. in SSLv3 a
warning alert gets sent "no_certificate" if a client does not send a
Certificate during Client Auth. With the above commit this causes the
connection to abort, which is incorrect. There may be some other edge cases
like this so we need to have a rethink on this.
Reviewed-by: Tim Hudson <tjh@openssl.org>
A peer continually sending unrecognised warning alerts could mean that we
make no progress on a connection. We should abort rather than continuing if
we receive an unrecognised warning alert.
Thanks to Shi Lei for reporting this issue.
Reviewed-by: Rich Salz <rsalz@openssl.org>
It is never valid to call ssl3_read_bytes with
type == SSL3_RT_CHANGE_CIPHER_SPEC, and in fact we check for valid values
for type near the beginning of the function. Therefore this check will never
be true and can be removed.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Follow on from CVE-2016-2179
The investigation and analysis of CVE-2016-2179 highlighted a related flaw.
This commit fixes a security "near miss" in the buffered message handling
code. Ultimately this is not currently believed to be exploitable due to
the reasons outlined below, and therefore there is no CVE for this on its
own.
The issue this commit fixes is a MITM attack where the attacker can inject
a Finished message into the handshake. In the description below it is
assumed that the attacker injects the Finished message for the server to
receive it. The attack could work equally well the other way around (i.e
where the client receives the injected Finished message).
The MITM requires the following capabilities:
- The ability to manipulate the MTU that the client selects such that it
is small enough for the client to fragment Finished messages.
- The ability to selectively drop and modify records sent from the client
- The ability to inject its own records and send them to the server
The MITM forces the client to select a small MTU such that the client
will fragment the Finished message. Ideally for the attacker the first
fragment will contain all but the last byte of the Finished message,
with the second fragment containing the final byte.
During the handshake and prior to the client sending the CCS the MITM
injects a plaintext Finished message fragment to the server containing
all but the final byte of the Finished message. The message sequence
number should be the one expected to be used for the real Finished message.
OpenSSL will recognise that the received fragment is for the future and
will buffer it for later use.
After the client sends the CCS it then sends its own Finished message in
two fragments. The MITM causes the first of these fragments to be
dropped. The OpenSSL server will then receive the second of the fragments
and reassemble the complete Finished message consisting of the MITM
fragment and the final byte from the real client.
The advantage to the attacker in injecting a Finished message is that
this provides the capability to modify other handshake messages (e.g.
the ClientHello) undetected. A difficulty for the attacker is knowing in
advance what impact any of those changes might have on the final byte of
the handshake hash that is going to be sent in the "real" Finished
message. In the worst case for the attacker this means that only 1 in
256 of such injection attempts will succeed.
It may be possible in some situations for the attacker to improve this such
that all attempts succeed. For example if the handshake includes client
authentication then the final message flight sent by the client will
include a Certificate. Certificates are ASN.1 objects where the signed
portion is DER encoded. The non-signed portion could be BER encoded and so
the attacker could re-encode the certificate such that the hash for the
whole handshake comes to a different value. The certificate re-encoding
would not be detectable because only the non-signed portion is changed. As
this is the final flight of messages sent from the client the attacker
knows what the complete hanshake hash value will be that the client will
send - and therefore knows what the final byte will be. Through a process
of trial and error the attacker can re-encode the certificate until the
modified handhshake also has a hash with the same final byte. This means
that when the Finished message is verified by the server it will be
correct in all cases.
In practice the MITM would need to be able to perform the same attack
against both the client and the server. If the attack is only performed
against the server (say) then the server will not detect the modified
handshake, but the client will and will abort the connection.
Fortunately, although OpenSSL is vulnerable to Finished message
injection, it is not vulnerable if *both* client and server are OpenSSL.
The reason is that OpenSSL has a hard "floor" for a minimum MTU size
that it will never go below. This minimum means that a Finished message
will never be sent in a fragmented form and therefore the MITM does not
have one of its pre-requisites. Therefore this could only be exploited
if using OpenSSL and some other DTLS peer that had its own and separate
Finished message injection flaw.
The fix is to ensure buffered messages are cleared on epoch change.
Reviewed-by: Richard Levitte <levitte@openssl.org>
The DTLS implementation provides some protection against replay attacks
in accordance with RFC6347 section 4.1.2.6.
A sliding "window" of valid record sequence numbers is maintained with
the "right" hand edge of the window set to the highest sequence number we
have received so far. Records that arrive that are off the "left" hand
edge of the window are rejected. Records within the window are checked
against a list of records received so far. If we already received it then
we also reject the new record.
If we have not already received the record, or the sequence number is off
the right hand edge of the window then we verify the MAC of the record.
If MAC verification fails then we discard the record. Otherwise we mark
the record as received. If the sequence number was off the right hand edge
of the window, then we slide the window along so that the right hand edge
is in line with the newly received sequence number.
Records may arrive for future epochs, i.e. a record from after a CCS being
sent, can arrive before the CCS does if the packets get re-ordered. As we
have not yet received the CCS we are not yet in a position to decrypt or
validate the MAC of those records. OpenSSL places those records on an
unprocessed records queue. It additionally updates the window immediately,
even though we have not yet verified the MAC. This will only occur if
currently in a handshake/renegotiation.
This could be exploited by an attacker by sending a record for the next
epoch (which does not have to decrypt or have a valid MAC), with a very
large sequence number. This means the right hand edge of the window is
moved very far to the right, and all subsequent legitimate packets are
dropped causing a denial of service.
A similar effect can be achieved during the initial handshake. In this
case there is no MAC key negotiated yet. Therefore an attacker can send a
message for the current epoch with a very large sequence number. The code
will process the record as normal. If the hanshake message sequence number
(as opposed to the record sequence number that we have been talking about
so far) is in the future then the injected message is bufferred to be
handled later, but the window is still updated. Therefore all subsequent
legitimate handshake records are dropped. This aspect is not considered a
security issue because there are many ways for an attacker to disrupt the
initial handshake and prevent it from completing successfully (e.g.
injection of a handshake message will cause the Finished MAC to fail and
the handshake to be aborted). This issue comes about as a result of trying
to do replay protection, but having no integrity mechanism in place yet.
Does it even make sense to have replay protection in epoch 0? That
issue isn't addressed here though.
This addressed an OCAP Audit issue.
CVE-2016-2181
Reviewed-by: Richard Levitte <levitte@openssl.org>
During a DTLS handshake we may get records destined for the next epoch
arrive before we have processed the CCS. In that case we can't decrypt or
verify the record yet, so we buffer it for later use. When we do receive
the CCS we work through the queue of unprocessed records and process them.
Unfortunately the act of processing wipes out any existing packet data
that we were still working through. This includes any records from the new
epoch that were in the same packet as the CCS. We should only process the
buffered records if we've not got any data left.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Run util/openssl-format-source on ssl/
Some comments and hand-formatted tables were fixed up
manually by disabling auto-formatting.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Feedback on the previous SSLv2 ClientHello processing fix was that it
breaks layering by reading init_num in the record layer. It also does not
detect if there was a previous non-fatal warning.
This is an alternative approach that directly tracks in the record layer
whether this is the first record.
GitHub Issue #1298
Reviewed-by: Tim Hudson <tjh@openssl.org>
Thanks to Peter Gijsels for pointing out that if a CBC record has 255
bytes of padding, the first was not being checked.
(This is an import of change 80842bdb from BoringSSL.)
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1431)
Baroque, almost uncommented code triggers behaviour which is undefined
by the C standard. You might quite reasonably not care that the code was
broken on ones-complement machines, but if we support a ubsan build then
we need to at least pretend to care.
It looks like the special-case code for 64-bit big-endian is going to
behave differently (and wrongly) on wrap-around, because it treats the
values as signed. That seems wrong, and allows replay and other attacks.
Surely you need to renegotiate and start a new epoch rather than
wrapping around to sequence number zero again?
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
DTLSv1_client_method() is deprecated, but it was the only way to obtain
DTLS1_BAD_VER support. The SSL_OP_CISCO_ANYCONNECT hack doesn't work with
DTLS_client_method(), and it's relatively non-trivial to make it work without
expanding the hack into lots of places.
So deprecate SSL_OP_CISCO_ANYCONNECT with DTLSv1_client_method(), and make
it work with SSL_CTX_set_{min,max}_proto_version(DTLS1_BAD_VER) instead.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
The MULTIBLOCK code uses a "jumbo" sized write buffer which it allocates
and then frees later. Pipelining however introduced multiple pipelines. It
keeps track of how many pipelines are initialised using numwpipes.
Unfortunately the MULTIBLOCK code was not updating this when in deallocated
its buffers, leading to a buffer being marked as initialised but set to
NULL.
RT#4618
Reviewed-by: Rich Salz <rsalz@openssl.org>
SSLv2 is no longer supported in 1.1.0, however we *do* still accept an SSLv2
style ClientHello, as long as we then subsequently negotiate a protocol
version >= SSLv3. The record format for SSLv2 style ClientHellos is quite
different to SSLv3+. We only accept this format in the first record of an
initial ClientHello. Previously we checked this by confirming
s->first_packet is set and s->server is true. However, this really only
tells us that we are dealing with an initial ClientHello, not that it is
the first record (s->first_packet is badly named...it really means this is
the first message). To check this is the first record of the initial
ClientHello we should also check that we've not received any data yet
(s->init_num == 0), and that we've not had any empty records.
GitHub Issue #1298
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Fix some indentation at the same time
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1292)
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1264)
In some situations (such as when we receive a fragment of an alert)
we try to get the next packet but did not mark the current one as read,
meaning that we got the same record back again - leading to an infinite
loop.
Found using the BoringSSL test suite.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Sessions are stored on the session_ctx, which doesn't change after
SSL_set_SSL_CTX().
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Previously if we received an empty record we just threw it away and
ignored it. Really though if we get an empty record of a different content
type to what we are expecting then that should be an error, i.e. we should
reject out of context empty records. This commit makes the necessary changes
to achieve that.
RT#4395
Reviewed-by: Andy Polyakov <appro@openssl.org>
Windows was complaining about a unary minus operator being applied to an
unsigned type. It did seem to go on and do the right thing anyway, but the
code does look a little suspect. This fixes it.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
In the SSLV2ClientHello processing code in ssl3_get_record, the value of
|num_recs| will always be 0. This isn't obvious from the code so a comment
is added to explain it.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
The function ssl3_get_record() can obtain multiple records in one go
as long as we are set up for pipelining and all the records are app
data records. The logic in the while loop which reads in each record is
supposed to only continue looping if the last record we read was app data
and we have an app data record waiting in the buffer to be processed. It
was actually checking that the first record had app data and we have an
app data record waiting. This actually amounts to the same thing so wasn't
wrong - but it looks a bit odd because it uses the |rr| array without an
offset.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Pipelining introduced the concept of multiple records being read in one
go. Therefore we work with an array of SSL3_RECORD objects. The pipelining
change erroneously made a change in ssl3_get_record() to apply the current
record offset to the SSL3_BUFFER we are using for reading. This is wrong -
there is only ever one read buffer. This reverts that change. In practice
this should make little difference because the code block in question is
only ever used when we are processing a single record.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
The numpipes argument to ssl3_enc/tls1_enc is actually the number of
records passed in the array. To make this clearer rename the argument to
|n_recs|.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Rename the have_whole_app_data_record_waiting() function to include the
ssl3_record prefix...and make it a bit shorter.
Reviewed-by: Tim Hudson <tjh@openssl.org>
We used to use the wrec field in the record layer for keeping track of the
current record that we are writing out. As part of the pipelining changes
this has been moved to stack allocated variables to do the same thing,
therefore the field is no longer needed.
Reviewed-by: Tim Hudson <tjh@openssl.org>
This is similar to SSL_pending() but just returns a 1 if there is data
pending in the internal OpenSSL buffers or 0 otherwise (as opposed to
SSL_pending() which returns the number of bytes available). Unlike
SSL_pending() this will work even if "read_ahead" is set (which is the
case if you are using read pipelining, or if you are doing DTLS). A 1
return value means that we have unprocessed data. It does *not* necessarily
indicate that there will be application data returned from a call to
SSL_read(). The unprocessed data may not be application data or there
could be errors when we attempt to parse the records.
Reviewed-by: Tim Hudson <tjh@openssl.org>
This capability is required for read pipelining. We will only read in as
many records as will fit in the read buffer (and the network can provide
in one go). The bigger the buffer the more records we can process in
parallel.
Reviewed-by: Tim Hudson <tjh@openssl.org>
With read pipelining we use multiple SSL3_RECORD structures for reading.
There are SSL_MAX_PIPELINES (32) of them defined (typically not all of these
would be used). Each one has a 16k compression buffer allocated! This
results in a significant amount of memory being consumed which, most of the
time, is not needed. This change swaps the allocation of the compression
buffer to be lazy so that it is only done immediately before it is actually
used.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Read pipelining is controlled in a slightly different way than with write
pipelining. While reading we are constrained by the number of records that
the peer (and the network) can provide to us in one go. The more records
we can get in one go the more opportunity we have to parallelise the
processing.
There are two parameters that affect this:
* The number of pipelines that we are willing to process in one go. This is
controlled by max_pipelines (as for write pipelining)
* The size of our read buffer. A subsequent commit will provide an API for
adjusting the size of the buffer.
Another requirement for this to work is that "read_ahead" must be set. The
read_ahead parameter will attempt to read as much data into our read buffer
as the network can provide. Without this set, data is read into the read
buffer on demand. Setting the max_pipelines parameter to a value greater
than 1 will automatically also turn read_ahead on.
Finally, the read pipelining as currently implemented will only parallelise
the processing of application data records. This would only make a
difference for renegotiation so is unlikely to have a significant impact.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Use the new pipeline cipher capability to encrypt multiple records being
written out all in one go. Two new SSL/SSL_CTX parameters can be used to
control how this works: max_pipelines and split_send_fragment.
max_pipelines defines the maximum number of pipelines that can ever be used
in one go for a single connection. It must always be less than or equal to
SSL_MAX_PIPELINES (currently defined to be 32). By default only one
pipeline will be used (i.e. normal non-parallel operation).
split_send_fragment defines how data is split up into pipelines. The number
of pipelines used will be determined by the amount of data provided to the
SSL_write call divided by split_send_fragment. For example if
split_send_fragment is set to 2000 and max_pipelines is 4 then:
SSL_write called with 0-2000 bytes == 1 pipeline used
SSL_write called with 2001-4000 bytes == 2 pipelines used
SSL_write called with 4001-6000 bytes == 3 pipelines used
SSL_write_called with 6001+ bytes == 4 pipelines used
split_send_fragment must always be less than or equal to max_send_fragment.
By default it is set to be equal to max_send_fragment. This will mean that
the same number of records will always be created as would have been
created in the non-parallel case, although the data will be apportioned
differently. In the parallel case data will be spread equally between the
pipelines.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Add -DBIO_DEBUG to --strict-warnings.
Remove comments about outdated debugging ifdef guards.
Remove md_rand ifdef guarding an assert; it doesn't seem used.
Remove the conf guards in conf_api since we use OPENSSL_assert, not assert.
For pkcs12 stuff put OPENSSL_ in front of the macro name.
Merge TLS_DEBUG into SSL_DEBUG.
Various things just turned on/off asserts, mainly for checking non-NULL
arguments, which is now removed: camellia, bn_ctx, crypto/modes.
Remove some old debug code, that basically just printed things to stderr:
DEBUG_PRINT_UNKNOWN_CIPHERSUITES, DEBUG_ZLIB, OPENSSL_RI_DEBUG,
RL_DEBUG, RSA_DEBUG, SCRYPT_DEBUG.
Remove OPENSSL_SSL_DEBUG_BROKEN_PROTOCOL.
Reviewed-by: Richard Levitte <levitte@openssl.org>
