Calling SSL_free() will call BIO_free_all() on the rbio and wbio. We
keep references to the rbio and wbio inside the record layer object.
References to that object are held directly, as well as in fragment
retransmission queues. We need to ensure all record layer objects are
cleaned up before we call BIO_free_all() on rbio/wbio - otherwise the
"top" BIO may not have its reference count drop to 0 when BIO_free_all()
is called. This means that the rest of the BIOs in the chain don't get
freed and a memory leak can occur.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Hugo Landau <hlandau@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/19424)
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Hugo Landau <hlandau@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/19424)
We no longer use the old buffer management code now that it has all been
moved to the new record layer.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Hugo Landau <hlandau@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/19424)
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Hugo Landau <hlandau@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/19424)
Mostly revamped from #16712
- fall thru -> fall through
- time stamp -> timestamp
- host name -> hostname
- ipv6 -> IPv6
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/19059)
We also convert to passing COMP_METHOD rather than SSL_COMP to the record
layer. The former is a public type while the latter is internal only - and
the only thing we need from SSL_COMP is the method.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Hugo Landau <hlandau@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/19217)
Since OPENSSL_malloc() and friends report ERR_R_MALLOC_FAILURE, and
at least handle the file name and line number they are called from,
there's no need to report ERR_R_MALLOC_FAILURE where they are called
directly, or when SSLfatal() and RLAYERfatal() is used, the reason
`ERR_R_MALLOC_FAILURE` is changed to `ERR_R_CRYPTO_LIB`.
There were a number of places where `ERR_R_MALLOC_FAILURE` was reported
even though it was a function from a different sub-system that was
called. Those places are changed to report ERR_R_{lib}_LIB, where
{lib} is the name of that sub-system.
Some of them are tricky to get right, as we have a lot of functions
that belong in the ASN1 sub-system, and all the `sk_` calls or from
the CRYPTO sub-system.
Some extra adaptation was necessary where there were custom OPENSSL_malloc()
wrappers, and some bugs are fixed alongside these changes.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Hugo Landau <hlandau@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/19301)
Make sure we free the record layer before we free the connection BIOs
Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/19198)
Previously we were referencing the block_padding value through the
SSL_CONNECTION. Now it is held within OSSL_RECORD_LAYER.
Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/19198)
We use the record layer reference to the BIO rather than the SSL object
reference. This removes an unneeded SSL object usage.
Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/19198)
Make sure we set the write record layer method and create the object
where appropriate. Move the newly restructured writing code into the
record layer object.
For now we are cheating and still accessing the underlying SSL_CONNECTION
object. This will be removed in subsequent commits.
Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/19198)
Doing so, had to fix sloppiness in using the stack API in crypto/conf/conf_def.c,
ssl/ssl_ciph.c, ssl/statem/statem_srvr.c, and mostly in test/helpers/ssltestlib.c.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: David von Oheimb <david.von.oheimb@siemens.com>
(Merged from https://github.com/openssl/openssl/pull/18918)
This is instead of time_t and struct timeval. Some public APIs mandate a
presence of these two types, but they are converted to OSSL_TIME internally.
Reviewed-by: Todd Short <todd.short@me.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/19082)
We would like the capability for the options/mode/read_ahead settings
to be updateable after the record layer object has been instantiated.
Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/18132)
Fields such as rrlmethod and rrl are entirely related to the record layer,
and so should be in that object.
Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/18132)
Some TODO(RECLAYER) comments are no longer necessary and can be removed.
Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/18132)
We had some temporary function pointers in OSSL_RECORD_METHOD which were
only necessary during the process of refactoring the read record layer.
These are no longer required so can be removed.
Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/18132)
Sometimes data read by a record layer in one epoch is actually intended for
the next epoch. For example in a TLS with read_ahead, the read_ahead data
could contain a KeyUpdate message followed by application data encrypted
with new keys. Therefore we implement a mechanism for passing this data
across the epochs.
Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/18132)
Only done for the read side so far. Still need to do TLS1.3 and SSL3.0.
Also need to separate out KTLS.
Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/18132)
This transfers the low level function ssl3_read_n to the new record layer.
We temporarily make the read_n function a top level record layer function.
Eventually, in later commits in this refactor, we will remove it as a top
level function and it will just be called from read_record.
Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/18132)
If app data is received before a Finished message in DTLS then we buffer
it to return later. The function SSL_pending() is supposed to tell you
how much processed app data we have already buffered, and SSL_has_pending()
is supposed to tell you if we have any data buffered (whether processed or
not, and whether app data or not).
Neither SSL_pending() or SSL_has_pending() were taking account of this
DTLS specific app data buffer.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Hugo Landau <hlandau@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/18868)
Make the SSL object polymorphic based on whether this is
a traditional SSL connection, QUIC connection, or later
to be implemented a QUIC stream.
It requires adding if after every SSL_CONNECTION_FROM_SSL() call
which itself has to be added to almost every public SSL_ API call.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/18612)
Provide a different mechanism to indicate that the application wants
to retry the verification. The negative result of the callback function
now indicates an error again.
Instead the SSL_set_retry_verify() can be called from the callback
to indicate that the handshake should be suspended.
Fixes#17568
Reviewed-by: David von Oheimb <david.von.oheimb@siemens.com>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17825)
Including e_os.h with a path from a header file doesn't work well on
certain exotic platform. It simply fails to build.
Since we don't seem to be able to stop ourselves, the better move is
to move e_os.h to an include directory that's part of the inclusion
path given to the compiler.
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17641)
The function/macro allow user get groups/extensions without memory allcations.
So we could calculate the ssl fignerprint(ja3) in low cost.
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16910)
For platforms that do not have native TSAN support, locking needs to be used
instead. This adds the locking.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17489)
We should not be freeing the caller's key in the event of error.
Fixes#17196
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17209)
If an async job pauses while processing a TLS connection then the
rwstate gets set to SSL_ASYNC_PAUSED. When resuming the job we should
reset the rwstate back to SSL_NOTHING. In fact we can do this
unconditionally since if we're about to call ASYNC_start_job() then either
we are about to start the async job for the first time (in which case the
rwstate should already by SSL_NOTHING), or we are restarting it after a
pause (in which case reseting it to SSL_NOTHING is the correct action).
Fixes#16809
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17013)
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16918)
Various comments referred to s->packet and s->packet_length instead of
s->rlayer.packet and s->rlayer.packet_length. Also fixed is a spot where
RECORD_LAYER_write_pending() should have been used. Based on the review
comments in #16077.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/16086)
If an application is halfway through writing application data it should
not be allowed to attempt an SSL_key_update() operation. Instead the
SSL_write() operation should be completed.
Fixes#12485
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16077)
When EVP_MD_CTX_new fails call SSLfatal before the goto err.
This resolves a state machine issue on the out of memory condition.
Fixes#15491.
CLA: trivial
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15520)
For functions that exist in 1.1.1 provide a simple aliases via #define.
Fixes#15236
Functions with OSSL_DECODER_, OSSL_ENCODER_, OSSL_STORE_LOADER_,
EVP_KEYEXCH_, EVP_KEM_, EVP_ASYM_CIPHER_, EVP_SIGNATURE_,
EVP_KEYMGMT_, EVP_RAND_, EVP_MAC_, EVP_KDF_, EVP_PKEY_,
EVP_MD_, and EVP_CIPHER_ prefixes are renamed.
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15405)
The initial implementation always deferred the generation of the
requested ticket(s) until the next application write, but this
is not a great fit for what it actually does, architecturally wise.
A request to send a session ticket means entering back into the
handshake state machine (or "in init", as it's known in the
implementation). The state machine transition is not something that
only occurs at an application-data write, and in general could occur at
any time. The only constraint is that we can't enter "init" while in
the middle of writing application data. In such cases we will need to
wait until the next TLS record boundary to enter the state machine,
as is currently done.
However, there is no reason why we cannot enter the handshake state
machine immediately in SSL_new_session_ticket() if there are no
application writes pending. Doing so provides a cleaner API surface to
the application, as then calling SSL_do_handshake() suffices to drive
the actual ticket generation. In the previous state of affairs a dummy
zero-length SSL_write() would be needed to trigger the ticket
generation, which is a logical mismatch in the type of operation being
performed.
This commit should only change whether SSL_do_handshake() vs zero-length
SSL_write() is needed to immediately generate a ticket after the
SSL_new_session_ticket() call -- the default behavior is still to defer
the actual write until there is other application data to write, unless
the application requests otherwise.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14817)
Add -client_renegotiation flag support. The -client_renegotiation flag is
equivalent to SSL_OP_ALLOW_CLIENT_RENEGOTIATION. Add support to the app,
the config code, and the documentation.
Add SSL_OP_ALLOW_CLIENT_RENEGOTIATION to the SSL tests. We don't need to
always enable it, but there are so many tests so this is the easiest thing
to do.
Add a test where client tries to renegotiate and it fails as expected. Add
a test where server tries to renegotiate and it succeeds. The second test
is supported by a new flag, -immediate_renegotiation, which is ignored on
the client.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15184)
Less tersely: converted SSL_get_options, SSL_set_options,
SSL_CTX_get_options and SSL_CTX_get_options to take and return uint64_t
since we were running out of 32 bits.
Fixes: 15145
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15230)
Previously we would set SSL_OP_LEGACY_SERVER_CONNECT by default in
SSL_CTX_new(), to allow connections to legacy servers that did not
implement RFC 5746.
It has been more than a decade since RFC 5746 was published, so
there has been plenty of time for implmentation support to roll out.
Change the default behavior to be to require peers to support
secure renegotiation. Existing applications that already cleared
SSL_OP_LEGACY_SERVER_CONNECT will see no behavior change, as
re-clearing the flag is just a little bit of redundant work.
The old behavior is still available by explicitly setting the flag
in the application.
Also remove SSL_OP_LEGACY_SERVER_CONNECT from SSL_OP_ALL, for
similar reasons.
Document the behavior change in CHANGES.md, and update the
SSL_CTX_set_options() and SSL_CONF_cmd manuals to reflect the change
in default behavior.
Fixes: 14848
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15127)
Matt Caswell
Todd Short
FdaSilvaYY
Richard Levitte
Dr. David von Oheimb
Pauli
Tomas Mraz
Phus Lu
Pauli
x2018
Trev Larock
Benjamin Kaduk
Rich Salz