The alignment calculation in ssl3_setup_write incorrectly results in an
alignment allowance of
(-SSL3_RT_HEADER_LENGTH) & (SSL3_ALIGN_PAYLOAD - 1) bytes. This equals 3
in almost all cases. The maximum alignment actually used in do_ssl3_write
is (SSL3_ALIGN_PAYLOAD - 1). This equals 7 bytes in almost all cases. So
there is a potential to overrun the buffer by up to 4 bytes.
Fortunately, the encryption overhead allowed for is 80 bytes which
consists of 16 bytes for the cipher block size and 64 bytes for the MAC
output. However the biggest MAC that we ever produce is HMAC-384 which is
48 bytes - so we have a headroom of 16 bytes (i.e. more than the 4 bytes
of potential overrun).
Thanks to Nagesh Hegde for reporting this.
Fixes#11766
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/11768)
Since the BIO_SSL structure was renewed by `ssl_free(b)/ssl_new(b)`,
the `bs` pointer needs to be updated before assigning to `bs->ssl`.
Thanks to @suishixingkong for reporting the issue and providing a fix.
Closes#10539
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/11746)
We were not correctly detecting whether TLSv1.3 ciphersuites could
actually be supported by the available provider implementations. For
example a FIPS client would still offer CHACHA20-POLY1305 based
ciphersuites even though it couldn't actually use them. Similarly on
the server would try to use CHACHA20-POLY1305 and then fail the
handshake.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/11700)
This API requests that the TLS stack generate a (TLS 1.3)
NewSessionTicket message the next time it is safe to do so (i.e., we do
not have other data pending write, which could be mid-record). For
efficiency, defer actually generating/writing the ticket until there
is other data to write, to avoid producing server-to-client traffic when
not needed.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11416)
An 'if' clause was nestled against a previous closing brace as it if was
an 'else if', but should properly stand on its own line.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11416)
... and only *define* them in the source files that need them.
Use DEFINE_OR_DECLARE which is set appropriately for internal builds
and not non-deprecated builds.
Deprecate stack-of-block
Better documentation
Move some ASN1 struct typedefs to types.h
Update ParseC to handle this. Most of all, ParseC needed to be more
consistent. The handlers are "recursive", in so far that they are called
again and again until they terminate, which depends entirely on what the
"massager" returns. There's a comment at the beginning of ParseC that
explains how that works. {Richard Levtte}
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/10669)
A couple of fetches of the MD5 and SHA1 digests were not using the
libctx in libssl and causing test_ssl_new to fail in travis. This
only occurs on builds with SSLv3 enabled (its disabled by default).
[extended tests]
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/11586)
In the tls1_check_sig_alg() helper function, we loop through the list of
"signature_algorithms_cert" values received from the client and attempt
to look up each one in turn in our internal table that maps wire
codepoint to string-form name, digest and/or signature NID, etc., in
order to compare the signature scheme from the peer's list against what
is used to sign the certificates in the certificate chain we're
checking. Unfortunately, when the peer sends a value that we don't
support, the lookup returns NULL, but we unconditionally dereference the
lookup result for the comparison, leading to an application crash
triggerable by an unauthenticated client.
Since we will not be able to say anything about algorithms we don't
recognize, treat NULL return from lookup as "does not match".
We currently only apply the "signature_algorithm_cert" checks on TLS 1.3
connections, so previous TLS versions are unaffected. SSL_check_chain()
is not called directly from libssl, but may be used by the application
inside a callback (e.g., client_hello or cert callback) to verify that a
candidate certificate chain will be acceptable to the client.
CVE-2020-1967
Reviewed-by: Matt Caswell <matt@openssl.org>
HMACs used via the legacy EVP_DigestSign interface are strange in
that they use legacy codepath's which eventually (under the covers)
transform the operation into a new style EVP_MAC. This can mean the
digest in use can be a legacy one, so we need to be careful with any
digest we extract from the ctx.
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/11511)
There were a few places where we were not passing through the libctx
when constructing and EVP_PKEY_CTX.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/11508)
We had a spot where a fatal error was occurring but we hadn't sent an
alert. This results in a later assertion failure.
Reviewed-by: Paul Yang <kaishen.yy@antfin.com>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/11537)
Since loading a private key might require algorithm fetches we should
make sure the correct libctx is used.
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/11494)
Ensure that when we create a CTLOG_STORE we use the new library context
aware function.
Also ensure that when we create a CT_POLICY_EVAL_CTX we associate it with
the library context.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/11483)
Make sure we cache the extensions for a cert using the right libctx.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/11457)
Libssl is OPENSSL_CTX aware so we should use it when creating an
X509_STORE_CTX.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/11457)
The transfer of TLS encodedpoint to backends isn't yet fully supported
in provider implementations. This is a temporary measure so as not to
get stuck in other development.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11358)
libssl code uses EVP_PKEY_get0_EC_KEY() to extract certain basic data
from the EC_KEY. We replace that with internal EVP_PKEY functions.
This may or may not be refactored later on.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11358)
Having created a DH object and assigned it to an EVP_PKEY - we should
not free both the EVP_PKEY and the original DH. This will lead to a
double free occurring.
This issue was discovered and reported by GitHub Security Lab team member
Agustin Gianni.
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/11441)
Some scenarios where we could not find a suitable sig alg just
gave "internal error" as the reason - which isn't very helpful. A
more suitable reason code already exists - so we use that.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/11425)
Caching the X509v3 extensions requires an explicit libctx. We do that
where required in libssl.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/11409)
EVP_CIPHERs in the ssl_cipher_methods table can be NULL if
they are not available. We shouldn't attempt to up-ref a
cipher if it is NULL.
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/11426)
Some fetch failurs are ok and should be ignored.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11405)
We should use an explicitly fetched cipher to ensure that we are using
the correct libctx and property query.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/11402)
We need to make sure we are using the correct libctx and property query.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/11402)
We use AES-256-CBC to encrypt stateless session tickets. We should
ensure that the implementation is fetched from the appropriate provider.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/11356)
We use the SHA256 digest of the ticket as a "fake" session id. We should
ensure that the SHA256 implementation is fetched from the appropriate
provider.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/11356)
When constructing an RSA ClientKeyExchange make sure we construct our
EVP_PKEY_CTX using the correct libctx and properties
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/11357)
in particular X509_NAME*, X509_STORE{,_CTX}*, and ASN1_INTEGER *,
also some result types of new functions, which does not break compatibility
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
Reviewed-by: David von Oheimb <david.von.oheimb@siemens.com>
(Merged from https://github.com/openssl/openssl/pull/10504)
With KTLS, writes to an SSL connection store the application buffer
pointer directly in the 'buf' member instead of allocating a separate
buffer to hold the encrypted data. As a result,
ssl3_release_write_buffer() has to avoid freeing these 'buf' pointers.
Previously, ssl3_release_write_buffer() checked for KTLS being enabled
on the write BIO to determine if a buffer should be freed. However, a
buffer can outlive a BIO. For example, 'openssl s_time' creates new
write BIOs when reusing sessions. Since the new BIO did not have KTLS
enabled at the start of a connection, ssl3_release_write_buffer()
would incorrectly try to free the 'buf' pointer from the previous KTLS
connection. To fix, track the state of 'buf' explicitly in
SSL3_BUFFER to determine if the 'buf' should be freed or simply
cleared.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/10489)
The server-side ChangeCipherState processing stores the new cipher
in the SSL_SESSION object, so that the new state can be used if
this session gets resumed. However, writing to the session is only
thread-safe for initial handshakes, as at other times the session
object may be in a shared cache and in use by another thread at the
same time. Reflect this invariant in the code by only writing to
s->session->cipher when it is currently NULL (we do not cache sessions
with no cipher). The code prior to this change would never actually
change the (non-NULL) cipher value in a session object, since our
server enforces that (pre-TLS-1.3) resumptions use the exact same
cipher as the initial connection, and non-abbreviated renegotiations
have produced a new session object before we get to this point.
Regardless, include logic to detect such a condition and abort the
handshake if it occurs, to avoid any risk of inadvertently using
the wrong cipher on a connection.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/10943)
TLS 1.3 maintains a separate keys chedule in the SSL object, but
was writing to the 'master_key_length' field in the SSL_SESSION
when generating the per-SSL master_secret. (The generate_master_secret
SSL3_ENC_METHOD function needs an output variable for the master secret
length, but the TLS 1.3 implementation just uses the output size of
the handshake hash function to get the lengths, so the only natural-looking
thing to use as the output length was the field in the session.
This would potentially involve writing to a SSL_SESSION object that was
in the cache (i.e., resumed) and shared with other threads, though.
The thread-safety impact should be minimal, since TLS 1.3 requires the
hash from the original handshake to be associated with the resumption
PSK and used for the subsequent connection. This means that (in the
resumption case) the value being written would be the same value that was
previously there, so the only risk would be on architectures that can
produce torn writes/reads for aligned size_t values.
Since the value is essentially ignored anyway, just provide the
address of a local dummy variable to generate_master_secret() instead.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/10943)
Use of the low level DH functions has been informally discouraged for a
long time. We now formally deprecate them.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11024)
Use of the low level RSA functions has been informally discouraged for a
long time. We now formally deprecate them.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11063)
