This makes the following changes:
- SSL_cutthrough_complete no longer rederives whether cutthrough happened and
just maintains a handshake bit.
- SSL_in_init no longer returns true if we are False Starting but haven't
completed the handshake. That logic was awkward as it depended on querying
in_read_app_data to force SSL_read to flush the entire handshake. Defaulting
SSL_in_init to continue querying the full handshake and special-casing
SSL_write is better. E.g. the check in bidirectional SSL_shutdown wants to know
if we're in a handshake. No internal consumer of
SSL_MODE_HANDSHAKE_CUTTHROUGH ever queries SSL_in_init directly.
- in_read_app_data is gone now that the final use is dead.
Change-Id: I05211a116d684054dfef53075cd277b1b30623b5
Reviewed-on: https://boringssl-review.googlesource.com/3336
Reviewed-by: Adam Langley <agl@google.com>
EVP_Digest can fail on malloc failure. May as well tidy that. Also make that
humongous comment less verbose.
Change-Id: I0ba74b901a5ac68711b9ed268b4202dc19242909
Reviewed-on: https://boringssl-review.googlesource.com/3331
Reviewed-by: Adam Langley <agl@google.com>
As of our 82b7da271f, an SSL_SESSION created
externally always has a cipher set. Unknown ciphers are rejected early. Prior
to that, an SSL_SESSION would only have a valid cipher or valid cipher_id
depending on whether it came from an internal or external session cache.
See upstream's 6a8afe2201cd888e472e44225d3c9ca5fae1ca62 and
c566205319beeaa196e247400c7eb0c16388372b for more context.
Since we don't get ourselves into this strange situation and s->cipher is now
always valid for established SSL_SESSION objects (the existence of
unestablished SSL_SESSION objects during a handshake is awkward, but something
to deal with later), do away with s->cipher_id altogether. An application
should be able to handle failing to parse an SSL_SESSION instead of parsing it
successfuly but rejecting all resumptions.
Change-Id: I2f064a815e0db657b109c7c9269ac6c726d1ffed
Reviewed-on: https://boringssl-review.googlesource.com/2703
Reviewed-by: Adam Langley <agl@google.com>
Add a dedicated error code to the queue for a handshake_failure alert in
response to ClientHello. This matches NSS's client behavior and gives a better
error on a (probable) failure to negotiate initial parameters.
BUG=https://crbug.com/446505
Change-Id: I34368712085a6cbf0031902daf2c00393783d96d
Reviewed-on: https://boringssl-review.googlesource.com/2751
Reviewed-by: Adam Langley <agl@google.com>
RAND_pseudo_bytes just calls RAND_bytes now and only returns 0 or 1. Switch all
callers within the library call the new one and use the simpler failure check.
This fixes a few error checks that no longer work (< 0) and some missing ones.
Change-Id: Id51c79deec80075949f73fa1fbd7b76aac5570c6
Reviewed-on: https://boringssl-review.googlesource.com/2621
Reviewed-by: Adam Langley <agl@google.com>
No current use of ssl_cert_type passes a NULL EVP_PKEY, so it can be simplified
a little.
Change-Id: I2052cc3b6069cd30e4685ba8a6d0014016a4d712
Reviewed-on: https://boringssl-review.googlesource.com/2620
Reviewed-by: Adam Langley <agl@google.com>
Those version checks are if renego tried to change the version, but at that
point we're out of the initial null cipher and should leave the version fixed.
(On the server end, the code in question was dead after the version negotiation
rewrite anyway.)
Change-Id: I3242ba11bc9981ccf7fdb867176d59846cc49dd9
Reviewed-on: https://boringssl-review.googlesource.com/2605
Reviewed-by: Adam Langley <agl@google.com>
This avoids needing a should_add_to_finished_hash boolean on do_write. The
logic in do_write was a little awkward because do_write would be called
multiple times if the write took several iterations. This also gets complex if
DTLS retransmits are involved. (At a glance, it's not obvious the
BIO_CTRL_DGRAM_MTU_EXCEEDED case actually works.)
Doing it as the handshake message is being prepared avoids this concern. It
also gives a natural point for the extended master secret logic which needs to
do work after the finished hash has been sampled.
As a bonus, we can remove s->d1->retransmitting which was only used to deal
with this issue.
Change-Id: Ifedf23ee4a6c5e08f960d296a6eb1f337a16dc7a
Reviewed-on: https://boringssl-review.googlesource.com/2604
Reviewed-by: Adam Langley <agl@google.com>
Some of the messages did the computation manually which would bite us if we
tried to transplant them between DTLS and TLS. More importantly, it precludes
moving the handshake hash computation from ssl_do_write to
ssl_set_handshake_header.
Change-Id: I9d400deb0720e62cb1ab905242eb0679ad650a46
Reviewed-on: https://boringssl-review.googlesource.com/2600
Reviewed-by: Adam Langley <agl@google.com>
Ensure that both the client and the server emit a protocol_version alert
(except in SSLv3 where it doesn't exist) with a record-layer version which the
peer will recognize.
Change-Id: I31650a64fe9b027ff3d51e303711910a00b43d6f
This makes SSLv23_method go through DTLS_ANY_VERSION's version negotiation
logic. This allows us to get rid of duplicate ClientHello logic. For
compatibility, SSL_METHOD is now split into SSL_PROTOCOL_METHOD and a version.
The legacy version-locked methods set min_version and max_version based this
version field to emulate the original semantics.
As a bonus, we can now handle fragmented ClientHello versions now.
Because SSLv23_method is a silly name, deprecate that too and introduce
TLS_method.
Change-Id: I8b3df2b427ae34c44ecf972f466ad64dc3dbb171
These tests use both APIs. This also modifies the inline version negotiation's
error codes (currently only used for DTLS) to align with SSLv23's error codes.
Note: the peer should send a protocol_version alert which is currently untested
because it's broken.
Upstream would send such an alert if TLS 1.0 was supported but not otherwise,
which is somewhat bizarre. We've actually regressed and never send the alert in
SSLv23. When version negotiation is unified, we'll get the alerts back.
Change-Id: I4c77bcef3a3cd54a039a642f189785cd34387410
Reviewed-on: https://boringssl-review.googlesource.com/2584
Reviewed-by: Adam Langley <agl@google.com>
DTLS_method() can now negotiate versions without switching methods.
Change-Id: I0655b3221b6e7e4b3ed4acc45f1f41c594447021
Reviewed-on: https://boringssl-review.googlesource.com/2582
Reviewed-by: Adam Langley <agl@google.com>
SSL3_ENC_METHOD will remain version-specific while SSL_METHOD will become
protocol-specific. This finally removes all the version-specific portions of
SSL_METHOD but the version tag itself.
(SSL3_ENC_METHOD's version-specific bits themselves can probably be handled by
tracking a canonicalized protocol version. It would simplify version
comparisons anyway. The one catch is SSLv3 has a very different table. But
that's a cleanup for future. Then again, perhaps a version-specific method
table swap somewhere will be useful later for TLS 1.3.)
Much of this commit was generated with sed invocation:
s/method->ssl3_enc/enc_method/g
Change-Id: I2b192507876aadd4f9310240687e562e56e6c0b1
Reviewed-on: https://boringssl-review.googlesource.com/2581
Reviewed-by: Adam Langley <agl@google.com>
Now SSLv23 and DTLS_ANY_VERSION share version-related helper functions.
ssl3_get_method is temporary until the method switch is no longer necessary.
Put them all together so there's one place to refactor them when we add a new
version or implement min_version/max_version controls.
Change-Id: Ic28a145cad22db08a87fdb854480b22886c451c6
Reviewed-on: https://boringssl-review.googlesource.com/2580
Reviewed-by: Adam Langley <agl@google.com>
These may as well be replaced with assertions. Get them out of the way of the
initialization.
Change-Id: Ie4ab8bdc018e4a1def7d3f6b3b172a77896bfc0a
Reviewed-on: https://boringssl-review.googlesource.com/2563
Reviewed-by: Adam Langley <agl@google.com>
The client_version needs to be preserved, both for the RSA key exchange and
(when this codepath is used for TLS) for the SChannel renego workaround. Fix
the tests to enforce this so the cipher suite version tests catch this.
Change-Id: I0c42dc3ec4830f3724026b400e5066e7a7f1ee97
Reviewed-on: https://boringssl-review.googlesource.com/2551
Reviewed-by: Adam Langley <agl@google.com>
Comparing data is a much easier idiom than CBS_skip + a CBS_len check.
Change-Id: I3efe925734c76f3494cad682445291ae83750a7e
Reviewed-on: https://boringssl-review.googlesource.com/2500
Reviewed-by: Adam Langley <agl@google.com>
It just inserts extra flushes everywhere and isn't used.
Change-Id: I082e4bada405611f4986ba852dd5575265854036
Reviewed-on: https://boringssl-review.googlesource.com/2456
Reviewed-by: Adam Langley <agl@google.com>
Use it in ssl3_cert_verify_hash so signing a pre-TLS-1.2 handshake hash can go
through RSA_sign and be intercepted via RSA_METHOD appropriately. This avoids
Windows needing to intercept sign_raw. (CAPI keys cannot provide sign_raw,
unless the input size happens to be that of NID_md5_sha1.)
Also use it in processing ServerKeyExchange to avoid special-casing RSA.
BUG=crbug.com/437023
Change-Id: Ia07433f468b75fdf7bfc8fa90c9751639b2478e6
Reviewed-on: https://boringssl-review.googlesource.com/2420
Reviewed-by: David Benjamin <davidben@google.com>
Replace the comment with a clearer one and reimplement it much more tidily. The
mask thing was more complicated than was needed.
This slightly changes behavior on the DTLS_ANY_VERSION side in that, if only
one method is enabled, we no longer short-circuit to the version-locked method
early. This "optimization" seems unnecessary.
Change-Id: I571c8b60ed16bd4357c67d65df0dd1ef9cc5eb57
Reviewed-on: https://boringssl-review.googlesource.com/2451
Reviewed-by: Adam Langley <agl@google.com>
It should be set correctly prior to entering the handshake. Don't mask bugs by
assigning it.
Change-Id: Ib9bca8fad68916b3b242aad8819e3760e59e777a
Reviewed-on: https://boringssl-review.googlesource.com/2443
Reviewed-by: Adam Langley <agl@google.com>
first_packet is a temporary connection-global flag set for the duration of some
call and then queried from other code. This kind of logic is too difficult to
reason through. It also incorrectly treats renegotiate ClientHellos as
pre-version-negotiation records. This eliminates the need to query
enc_write_ctx (which wasn't EVP_AEAD-aware anyway).
Instead, take a leaf from Go TLS's book and add a have_version bit. This is
placed on s->s3 as it is connection state; s->s3 automatically gets reset on
SSL_clear while s doesn't.
This new flag will also be used to determine whether to do the V2ClientHello
sniff when the version-locked methods merge into SSLv23_method. It will also
replace needing to condition s->method against a dummy DTLS_ANY_VERSION value
to determine whether DTLS version negotiation has happened yet.
Change-Id: I5c8bc6258b182ba4ab175a48a84eab6d3a001333
Reviewed-on: https://boringssl-review.googlesource.com/2442
Reviewed-by: Adam Langley <agl@google.com>
Supporting both schemes seems pointless. Now that s->server and s->state are
set appropriately late and get_ssl_method is gone, the only difference is that
the client/server ones have non-functional ssl_accept or ssl_connect hooks. We
can't lose the generic ones, so let's unify on that.
Note: this means a static linker will no longer drop the client or server
handshake code if unused by a consumer linking statically. However, Chromium
needs the server half anyway for DTLS and WebRTC, so that's probably a lost
cause. Android also exposes server APIs.
Change-Id: I290f5fb4ed558f59fadb5d1f84e9d9c405004c23
Reviewed-on: https://boringssl-review.googlesource.com/2440
Reviewed-by: Adam Langley <agl@google.com>
If the state is SSL_ST_BEFORE, the SSL* was just initialized. Otherwise, we
don't want to call SSL_clear. The one case I found where we do is if a
handshake message is received and someone sets
SSL3_FLAGS_NO_RENEGOTIATE_CIPHERS. This is apparently intended for external
consumers to set, but I see no code in Google that does.
Which is fortunate because it'll trigger SSL_clear. This retains the BIOs but
drops all connection state, including the record. If the client just initiated
renego, that's the ClientHello that's lost. The connection then hangs: the now
reset SSL* wants a ClientHello (under the null cipher because that too's been
dropped) while the peer wants an encrypted ServerHello.
Change-Id: Iddb3e0bb86d39d98155b060f9273a0856f2d1409
Reviewed-on: https://boringssl-review.googlesource.com/2436
Reviewed-by: Adam Langley <agl@google.com>
SSL_ST_BEFORE is never standalone. As of upstream's
413c4f45ed0508d2242638696b7665f499d68265, SSL_ST_BEFORE is only ever set paired
with SSL_ST_ACCEPT or SSL_ST_CONNECT.
Conversely, SSL_ST_OK is never paired with SSL_ST_ACCEPT or SSL_ST_CONNECT. As
far as I can tell, this combination has never been possible.
Change-Id: Ifbc8f147be821026cf59f3d5038f0dbad3b0a1d2
Reviewed-on: https://boringssl-review.googlesource.com/2433
Reviewed-by: Adam Langley <agl@google.com>
It should already be assigned, as of upstream's
b31b04d951e9b65bde29657e1ae057b76f0f0a73. I believe these assignments are part
of the reason it used to appear to work. Replace them with assertions. So the
assertions are actually valid, check in SSL_connect / SSL_accept that they are
never called if the socket had been placed in the opposite state. (Or we'd be
in another place where it would have appeared to work with the handshake
functions fixing things afterwards.)
Now the only places handshake_func is set are in SSL_set_{connect,accept}_state
and the method switches.
Change-Id: Ib249212bf4aa889b94c35965a62ca06bdbcf52e1
Reviewed-on: https://boringssl-review.googlesource.com/2432
Reviewed-by: Adam Langley <agl@google.com>
When not offering to resume a session, the client populates s->session with a
fresh SSL_SESSION before the ServerHello is processed and, in DTLS_ANY_VERSION,
before the version is even determined. Don't create a fresh SSL_SESSION until
we know we are doing a full handshake.
This brings ssl3_send_client_hello closer to ssl23_client_hello in behavior. It
also fixes ssl_version in the client in DTLS_ANY_VERSION.
SSLv23_client_method is largely unchanged. If no session is offered, s->session
continues to be NULL until the ServerHello is received. The one difference is
that s->session isn't populated until the entire ServerHello is received,
rather than just the first half, in the case of a fragmented ServerHello. Apart
from info_callback, no external hooks get called between those points, so this
shouldn't expose new missing NULL checks.
The other client methods change significantly to match SSLv23_client_method's
behavior. For TLS, any exposed missing NULL checks are also in
SSLv23_client_method (and version-specific methods are already weird), so that
should be safe. For DTLS, I've verified that accesses in d1_*.c either handle
NULL or are after the ServerHello.
Change-Id: Idcae6bd242480e28a57dbba76ce67f1ac1ae1d1d
Reviewed-on: https://boringssl-review.googlesource.com/2404
Reviewed-by: Adam Langley <agl@google.com>
This fixes bugs that kept the tests from working:
- Resolve DTLS version and cookie before the session.
- In DTLS_ANY_VERSION, ServerHello should be read with first_packet = 1. This
is a regression from f2fedefdca. We'll want to
do the same for TLS, but first let's change this to a boolean has_version in a
follow-up.
Things not yet fixed:
- DTLS code is not EVP_AEAD-aware. Those ciphers are disabled for now.
- On the client, DTLS_ANY_VERSION creates SSL_SESSIONs with the wrong
ssl_version. The tests pass because we no longer enforce the match as of
e37216f56009fbf48c3a1e733b7a546ca6dfc2af. (In fact, we've gone from the server
ignoring ssl_version and client enforcing to the client mostly ignoring
ssl_version and the server enforcing.)
- ssl3_send_client_hello's ssl_version check checks for equality against
s->version rather than >.
Change-Id: I5a0dde221b2009413df9b9443882b9bf3b29519c
Reviewed-on: https://boringssl-review.googlesource.com/2403
Reviewed-by: Adam Langley <agl@google.com>
This is a bit of cleanup that probably should have been done at the same time
as 30ddb434bf.
For now, version negotiation is implemented with a method swap. It also
performs this swap on SSL_set_session, but this was neutered in
30ddb434bf. Rather than hackishly neuter it,
remove it outright. In addition, remove SSL_set_ssl_method. Now all method
swaps are internal: SSLv23_method switch to a version-specific method and
SSL_clear undoing it.
Note that this does change behavior: if an SSL* is created with one
version-specific method and we SSL_set_session to a session from a /different/
version, we would switch to the /other/ version-specific method. This is
extremely confusing, so it's unlikely anyone was actually expecting it.
Version-specific methods in general don't work well.
Change-Id: I72a5c1f321ca9aeb1b52ebe0317072950ba25092
Reviewed-on: https://boringssl-review.googlesource.com/2390
Reviewed-by: Adam Langley <agl@google.com>
This is only used for EAP-FAST which we apparently don't need to support.
Remove it outright. We broke it in 9eaeef81fa by
failing to account for session misses.
If this changes and we need it later, we can resurrect it. Preferably
implemented differently: the current implementation is bolted badly onto the
handshake. Ideally use the supplied callbacks to fabricate an appropriate
SSL_SESSION and resume that with as much of the normal session ticket flow as
possible.
The one difference is that EAP-FAST seems to require the probing mechanism for
session tickets rather than the sane session ID echoing version. We can
reimplement that by asking the record layer to probe ahead for one byte.
Change-Id: I38304953cc36b2020611556a91e8ac091691edac
Reviewed-on: https://boringssl-review.googlesource.com/2360
Reviewed-by: Adam Langley <agl@google.com>
This commit fixes a number of crashes caused by malloc failures. They
were found using the -malloc-test=0 option to runner.go which runs tests
many times, causing a different allocation call to fail in each case.
(This test only works on Linux and only looks for crashes caused by
allocation failures, not memory leaks or other errors.)
This is not the complete set of crashes! More can be found by collecting
core dumps from running with -malloc-test=0.
Change-Id: Ia61d19f51e373bccb7bc604642c51e043a74bd83
Reviewed-on: https://boringssl-review.googlesource.com/2320
Reviewed-by: Adam Langley <agl@google.com>
tls1_process_sigalgs now only determines the intersection between the peer
algorithms and those configured locally. That list is queried later to
determine the hash algorithm to use when signing CertificateVerify or
ServerKeyExchange.
This is needed to support client auth on Windows where smartcards or CAPI may
not support all hash functions.
As a bonus, this does away with more connection-global state. This avoids the
current situation where digests are chosen before keys are known (for
CertificateVerify) or for slots that don't exist.
Change-Id: Iec3619a103d691291d8ebe08ef77d574f2faf0e8
Reviewed-on: https://boringssl-review.googlesource.com/2280
Reviewed-by: Adam Langley <agl@google.com>
CERT_PKEY_SIGN isn't meaningful since, without strict mode, we always fall back
to SHA-1 anyway. So the digest is never NULL when CERT_PKEY_SIGN is computed.
The entire valid_flags is now back to it's pre-1.0.2 check of seeing if the
certificate and key are configured.
This finally removes the sensitivity between valid_flags and selecting the
digest, so we can defer choosing the digest all we like.
Change-Id: I9f9952498f512d7f0cc799497f7c5b52145a48af
Reviewed-on: https://boringssl-review.googlesource.com/2288
Reviewed-by: Adam Langley <agl@google.com>
It's new in OpenSSL 1.0.2 so it's never set by existing code. This removes gobs
and gobs of complexity from tls1_check_chain. It only checks the local
certificate, not the peer certificate. The uses appear to be:
- Sanity-check configuration. Not worth the complexity.
- Guide in selecting ciphers based on ClientHello parameters and which
certificates in the CERT_PKEY are compatible. This isn't very useful one its
own since the CERT_PKEY array only stores one slot per type (e.g. you cannot
configure RSA/SHA-1 and RSA/SHA-256).
- For the (currently removed) SSL_check_chain to return more information based
on ClientHello parameters and guide selecting a certificate. This is
potentially useful but, as noted in the commit which removed it, redundant
with ssl_early_callback_ctx.
This CL is largely mechanical removing of dead codepaths. The follow-up will
clean up the now unnecessary parts of this function.
Change-Id: I2ebfa17e4f73e59aa1ee9e4ae7f615af2c6cf590
Reviewed-on: https://boringssl-review.googlesource.com/2285
Reviewed-by: Adam Langley <agl@google.com>
Both of these are newly-exported in OpenSSL 1.0.2, so they cannot be used by
current consumers.
This was added in upstream's 18d7158809c9722f4c6d2a8af7513577274f9b56 to
support custom selection of certificates. The intent seems to be that you
listen to cert_cb and use SSL_check_chain to lean on OpenSSL to process
signature algorithms list for you.
Unfortunately, the implementation is slightly suspect: it uses the same
function as the codepath which mutates and refers to the CERT_PKEY of the
matching type. Some access was guarded by check_flags, but this is too
complex. Part of it is also because the matching digest is selected early and
we intend to connect this to EVP_PKEY_supports_digest so it is no longer a
property of just the key type.
Let's remove the hook for now, to unblock removing a lot of complexity. After
cleaning up this area, a function like this could be cleaner to support, but
we already have a version of this: select_certificate_cb and
ssl_early_callback_ctx.
Change-Id: I3add425b3996e5e32d4a88e14cc607b4fdaa5aec
Reviewed-on: https://boringssl-review.googlesource.com/2283
Reviewed-by: Adam Langley <agl@google.com>
All of NSS, upstream OpenSSL, SChannel, and Secure Transport require, on the
client, that the ServerHello version match the session's version on resumption.
OpenSSL's current behavior is incompatible with all of these. Fall back to a
full handshake on the server instead of mismatch.
Add a comment on the client for why we are, as of
30ddb434bf, not currently enforcing the same in
the client.
Change-Id: I60aec972d81368c4ec30e2fd515dabd69401d175
Reviewed-on: https://boringssl-review.googlesource.com/2244
Reviewed-by: Adam Langley <agl@google.com>
There's not much point in retaining the identity hint in the SSL_SESSION. This
avoids the complexity around setting psk_identity hint on either the SSL or the
SSL_SESSION. Introduce a peer_psk_identity_hint for the client to store the one
received from the server.
This changes the semantics of SSL_get_psk_identity_hint; it now only returns
the value configured for the server. The client learns the hint through the
callback. This is compatible with the one use of this API in conscrypt (it
pulls the hint back out to pass to a callback).
Change-Id: I6d9131636b47f13ac5800b4451436a057021054a
Reviewed-on: https://boringssl-review.googlesource.com/2213
Reviewed-by: Adam Langley <agl@google.com>
This is an experimental flag that dates back to SSLeay 0.8.1b or earlier. It's
never set internally and never set in consumers.
Change-Id: I922583635c9f3d8d93f08f1707531ad22a26ae6a
Reviewed-on: https://boringssl-review.googlesource.com/2214
Reviewed-by: Adam Langley <agl@google.com>
If psk_len were 0, it would already have been an error earlier. The PSK cipher
suites don't lose the other_secret || psk construction if the PSK happens to be
empty.
Change-Id: I1917236720d0862658562bc8f014cb827ee9aed5
Reviewed-on: https://boringssl-review.googlesource.com/2233
Reviewed-by: Adam Langley <agl@google.com>
Server-side support was removed in 77a942b7fe,
but client-side support was retained as it appeared NSS supported this.
However, this is not the case: ssl3_HandleServerKeyExchange only allows a
ServerKeyExchange message if hs.ws is in an appropriate state.
ssl3_AuthCertificate only sets it to allow ServerKeyExchange if it is a key
exchange that normally uses it or if is_limited is set. is_limited is only set
for the export cipher suites.
Thus we can safely remove this without waiting on gathering UMA data.
BUG=chromium:400587
Change-Id: I9aefb742dbb2d99c13340ab48017e1ceee04bc2f
Reviewed-on: https://boringssl-review.googlesource.com/2230
Reviewed-by: Adam Langley <agl@google.com>
Check the return value while we're here. This avoids some arithmetic and
appease scan-build's dead assignment flagger.
Change-Id: If3615076e091eb44b9e3e9d50cd64f80e645337e
Reviewed-on: https://boringssl-review.googlesource.com/2204
Reviewed-by: Adam Langley <agl@google.com>
Minor change, but they're the users of the old API left within
BoringSSL.
Change-Id: Ic24e0d006c97fa5265abc3373d3f98aa8d2f8b1e
Reviewed-on: https://boringssl-review.googlesource.com/2100
Reviewed-by: Adam Langley <agl@google.com>
Like ssl3_get_client_key_exchange, it is split into three parts:
- If PSK, query the PSK and write out the PSK identity.
- Compute the base pre-master secret.
- If PSK, compute the final pre-master secret.
This also fixes some double-frees on malloc failures in the ECDHE case. And it
avoids using the handshake output buffer to start the premaster secret.
Change-Id: I8631ee33c1e9c19604b3dcce2c676c83893c308d
Reviewed-on: https://boringssl-review.googlesource.com/2062
Reviewed-by: Adam Langley <agl@google.com>
The current implementation switches the order of other_secret and psk;
other_secret is first. Fix it and rewrite with CBB instead. The server half got
fixed on accident in a prior refactor.
Change-Id: Ib52a756aadd66e4bf22c66794447f71f4772da09
Reviewed-on: https://boringssl-review.googlesource.com/2052
Reviewed-by: Adam Langley <agl@google.com>
