nginx consumes these error codes without #ifdefs. Continue to define
them for compatibility, even though we never emit them.
BUG=95
Change-Id: I1e991987ce25fc4952cc85b98ffa050a8beab92e
Reviewed-on: https://boringssl-review.googlesource.com/10446
Reviewed-by: David Benjamin <davidben@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
Change-Id: Ie60744761f5aa434a71a998f5ca98a8f8b1c25d5
Reviewed-on: https://boringssl-review.googlesource.com/10447
Reviewed-by: David Benjamin <davidben@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
Having two copies of this is confusing. This field is inherently tied to
the certificate chain, which lives on SSL_SESSION, so this should live
there too. This also wasn't getting reset correctly on SSL_clear, but
this is now resolved.
Change-Id: I22b1734a93320bb0bf0dc31faa74d77a8e1de906
Reviewed-on: https://boringssl-review.googlesource.com/10283
Reviewed-by: David Benjamin <davidben@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
As documented by OpenSSL, it does not interact with session resumption
correctly:
https://www.openssl.org/docs/manmaster/ssl/SSL_set_verify_result.html
Sadly, netty-tcnative calls it, but we should be able to get them to
take it out because it doesn't do anything. Two of the three calls are
immediately after SSL_new. In OpenSSL and BoringSSL as of the previous
commit, this does nothing.
The final call is in verify_callback (see SSL_set_verify). This callback
is called in X509_verify_cert by way of X509_STORE_CTX_set_verify_cb.
As soon as X509_verify_cert returns, ssl->verify_result is clobbered
anyway, so it doesn't do anything.
Within OpenSSL, it's used in testdane.c. As far as I can tell, it does
not actually do a handshake and just uses this function to fake having
done one. (Regardless, we don't need to build against that.)
This is done in preparation for removing ssl->verify_result in favor of
session->verify_result.
Change-Id: I7e32d7f26c44f70136c72e58be05a3a43e62582b
Reviewed-on: https://boringssl-review.googlesource.com/10485
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: Adam Langley <agl@google.com>
Change-Id: I2e1ee319bb9852b9c686f2f297c470db54f72279
Reviewed-on: https://boringssl-review.googlesource.com/10370
Reviewed-by: David Benjamin <davidben@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
BUG=75
Change-Id: Ied864cfccbc0e68d71c55c5ab563da27b7253463
Reviewed-on: https://boringssl-review.googlesource.com/9043
Reviewed-by: David Benjamin <davidben@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
It's odd that a function like ssl_bytes_to_cipher_list secretly has side
effects all over the place. This removes the need for the TLS 1.3 code
to re-query the version range, and it removes the requirement that the
RI extension be first.
Change-Id: Ic9af549db3aaa8880f3c591b8a13ba9ae91d6a46
Reviewed-on: https://boringssl-review.googlesource.com/10220
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
Between TLS 1.2, TLS 1.3, and the early callback, we've got a lot of
ClientHello parsers. Unify everything on the early callback's parser. As
a side effect, this means we can parse a ClientHello fairly succinctly
from any function which will let us split up ClientHello states where
appropriate.
Change-Id: I2359b75f80926cc7d827570cf33f93029b39e525
Reviewed-on: https://boringssl-review.googlesource.com/10184
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: Adam Langley <agl@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
Only X509_up_ref left (it's still waiting on a few external callers).
BUG=89
Change-Id: Ia2aec2bb0a944356cb1ce29f3b58a26bdb8a9977
Reviewed-on: https://boringssl-review.googlesource.com/9141
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
Extend the DTLS mock clock to apply to sessions too and test that
resumption behaves as expected.
Change-Id: Ib8fdec91b36e11cfa032872b63cf589f93b3da13
Reviewed-on: https://boringssl-review.googlesource.com/9110
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: Adam Langley <agl@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
We broke this to varying degrees ages ago.
This is the logic to implement the variations of rules in TLS to discard
sessions after a failed connection, where a failed connection could be
one of:
- A connection that was not cleanly shut down.
- A connection that received a fatal alert.
The first one is nonsense since close_notify does not actually work in
the real world. The second is a vaguely more plausible but...
- A stateless ticket-based server can't drop sessions anyway.
- In TLS 1.3, a client may receive many tickets over the lifetime of a
single connection. With an external session cache like ours which may,
in theory, but multithreaded, this will be a huge hassle to track.
- A client may well attempt to establish a connection and reuse the
session before we receive the fatal alert, so any application state we
hope to manage won't really work.
- An attacker can always close the connection before the fatal alert, so
whatever security policy clearing the session gave is easily
bypassable.
Implementation-wise, this has basically never worked. The
ssl_clear_bad_session logic called into SSL_CTX_remove_session which
relied on the internal session cache. (Sessions not in the internal
session cache don't get removed.) The internal session cache was only
useful for a server, where tickets prevent this mechanism from doing
anything. For a client, we since removed the internal session cache, so
nothing got removed. The API for a client also did not work as it gave
the SSL_SESSION, not the SSL, so a consumer would not know the key to
invalidate anyway.
The recent session state splitting change further broke this.
Moreover, calling into SSL_CTX_remove_session logic like that is
extremely dubious because it mutates the not_resumable flag on the
SSL_SESSION which isn't thread-safe.
Spec-wise, TLS 1.3 has downgraded the MUST to a SHOULD.
Given all that mess, just remove this code. It is no longer necessary to
call SSL_shutdown just to make session caching work.
Change-Id: Ib601937bfc5f6b40436941e1c86566906bb3165d
Reviewed-on: https://boringssl-review.googlesource.com/9091
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: Adam Langley <agl@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
OpenSSL 1.1.0 added a function to tell if an SSL* is DTLS or not. This
is probably a good idea, especially since SSL_version returns
non-normalized versions.
BUG=91
Change-Id: I25c6cf08b2ebabf0c610c74691de103399f729bc
Reviewed-on: https://boringssl-review.googlesource.com/9077
Reviewed-by: Steven Valdez <svaldez@google.com>
Reviewed-by: David Benjamin <davidben@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
SSL_set_bio is a nightmare.
In f715c423224a292d79ba0e3df373c828fbae29f7, we noticed that, among
other problems, SSL_set_bio's actual behavior did not match how
SSL_set_rfd was calling it due to an asymmetry in the rbio/wbio
handling. This resulted in SSL_set_fd/SSL_set_rfd calls to crash. We
decided that SSL_set_rfd's believed semantics were definitive and
changed SSL_set_bio.
Upstream, in 65e2d672548e7c4bcb28f1c5c835362830b1745b, decided that
SSL_set_bio's behavior, asymmetry and all, was definitive and that the
SSL_set_rfd crash was a bug in SSL_set_rfd. Accordingly, they switched
the fd callers to use the side-specific setters, new in 1.1.0.
Align with upstream's behavior and add tests for all of SSL_set_bio's
insanity. Also export the new side-specific setters in anticipation of
wanting to be mostly compatible with OpenSSL 1.1.0.
Change-Id: Iceac9508711f79750a3cc2ded081b2bb2cbf54d8
Reviewed-on: https://boringssl-review.googlesource.com/9064
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: Adam Langley <agl@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
It seems much safer for the default value of |verify_result| to be an
error value.
Change-Id: I372ec19c41d77516ed12d0169969994f7d23ed70
Reviewed-on: https://boringssl-review.googlesource.com/9063
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: Adam Langley <agl@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
Change-Id: I5cc194fc0a3ba8283049078e5671c924ee23036c
Reviewed-on: https://boringssl-review.googlesource.com/8980
Reviewed-by: David Benjamin <davidben@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
To prevent configuration/established session confusion, the handshake
session state is separated into the configured session (ssl->session)
and the newly created session (ssl->s3->new_session). Upon conclusion of
the handshake, the finalized session is stored
in (ssl->s3->established_session). During the handshake, any requests
for the session (SSL_get_session) return a non-resumable session, to
prevent resumption of a partially filled session. Sessions should only
be cached upon the completion of the full handshake, using the resulting
established_session. The semantics of accessors on the session are
maintained mid-renego.
Change-Id: I4358aecb71fce4fe14a6746c5af1416a69935078
Reviewed-on: https://boringssl-review.googlesource.com/8612
Reviewed-by: David Benjamin <davidben@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
This finishes getting rid of ssl_read_bytes! Now we have separate
entry-points for the various cases. For now, I've kept TLS handshake
consuming records partially. When we do the BIO-less API, I expect that
will need to change, since we won't have the record buffer available.
(Instead, the ssl3_read_handshake_bytes and extend_handshake_buffer pair
will look more like the DTLS side or Go and pull the entire record into
init_buf.)
This change opts to make read_app_data drive the message to completion
in anticipation of DTLS 1.3. That hasn't been specified, but
NewSessionTicket certainly will exist. Knowing that DTLS necessarily has
interleave seems something better suited for the SSL_PROTOCOL_METHOD
internals to drive.
It needs refining, but SSL_PROTOCOL_METHOD is now actually a half-decent
abstraction boundary between the higher-level protocol logic and
DTLS/TLS-specific record-layer and message dispatchy bits.
BUG=83
Change-Id: I9b4626bb8a29d9cb30174d9e6912bb420ed45aff
Reviewed-on: https://boringssl-review.googlesource.com/9001
Reviewed-by: David Benjamin <davidben@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
This API needs to be improved but, for the time being, keep the
invariant reasonable.
Change-Id: If94d41e7e7936e44de5ecb36da45f89f80df7784
Reviewed-on: https://boringssl-review.googlesource.com/8984
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: Adam Langley <agl@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
WebRTC want to be able to send a random alert. Add an API for this.
Change-Id: Id3113d68f25748729fd9e9a91dbbfa93eead12c3
Reviewed-on: https://boringssl-review.googlesource.com/8950
Reviewed-by: Taylor Brandstetter <deadbeef@webrtc.org>
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: Adam Langley <agl@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
It used to give a sensible answer ("no") before version negotiation.
Change-Id: I85b778a48cca7a4b66a81384eb18c447982875d1
Reviewed-on: https://boringssl-review.googlesource.com/8900
Reviewed-by: Matt Braithwaite <mab@google.com>
Reviewed-by: David Benjamin <davidben@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
Alas, we will need a version fallback for TLS 1.3 again.
This deprecates SSL_MODE_SEND_FALLBACK_SCSV. Rather than supplying a
boolean, have BoringSSL be aware of the real maximum version so we can
change the TLS 1.3 anti-downgrade logic to kick in, even when
max_version is set to 1.2.
The fallback version replaces the maximum version when it is set for
almost all purposes, except for downgrade protection purposes.
BUG=chromium:630165
Change-Id: I4c841dcbc6e55a282b223dfe169ac89c83c8a01f
Reviewed-on: https://boringssl-review.googlesource.com/8882
Reviewed-by: David Benjamin <davidben@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
This adds three more formats to the SSLKEYLOGFILE format to support TLS
1.3:
EARLY_TRAFFIC_SECRET <client_random> <early_traffic_secret>
HANDSHAKE_TRAFFIC_SECRET <client_random> <handshake_traffic_secret>
TRAFFIC_SECRET_0 <client_random> <traffic_secret_0>
(We don't implement 0-RTT yet, so only the second two are implemented.)
Motivations:
1. If emitted the non-traffic secrets (early, handshake, and master) or
the IKMs, Wireshark needs to maintain a handshake hash. I don't
believe they need to do this today.
2. We don't store more than one non-traffic secret at a time and don't
keep traffic secrets for longer than needed. That suggests three
separate lines logged at different times rather than one line.
3. If 0-RTT isn't used, we probably won't even compute the early traffic
secret, so that further suggests three different lines.
4. If the handshake didn't get far enough to complete, we won't have an
TRAFFIC_SECRET_0 to log at all. That seems like exactly when
Wireshark would be handy, which means we want to log secrets as they
are computed.
MT from NSS has ACK'd over email that this format would be acceptable
for them, so let's go with it.
Change-Id: I4d685a1355dff4d4bd200310029d502bb6c511f9
Reviewed-on: https://boringssl-review.googlesource.com/8841
Reviewed-by: David Benjamin <davidben@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
This adds the machinery for doing TLS 1.3 1RTT.
Change-Id: I736921ffe9dc6f6e64a08a836df6bb166d20f504
Reviewed-on: https://boringssl-review.googlesource.com/8720
Reviewed-by: David Benjamin <davidben@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
This makes custom private keys and EVP_PKEYs symmetric again. There is
no longer a requirement that the caller pre-filter the configured
signing prefs.
Also switch EVP_PKEY_RSA to NID_rsaEncryption. These are identical, but
if some key types are to be NIDs, we should make them all NIDs.
Change-Id: I82ea41c27a3c57f4c4401ffe1ccad406783e4c64
Reviewed-on: https://boringssl-review.googlesource.com/8785
Reviewed-by: David Benjamin <davidben@google.com>
Upstream have added |EVP_PKEY_up_ref|, but their version returns an int.
Having this function with a different signature like that is dangerous
so this change aligns BoringSSL with upstream. Users of this function in
Chromium and internally should already have been updated.
Change-Id: I0a7aeaf1a1ca3b0f0c635e2ee3826aa100b18157
Reviewed-on: https://boringssl-review.googlesource.com/8736
Reviewed-by: David Benjamin <davidben@google.com>
Since they include an ECDHE exchange in them, they are equally-well
suited to False Start.
Change-Id: I75d31493a614a78ccbf337574c359271831d654d
Reviewed-on: https://boringssl-review.googlesource.com/8732
Reviewed-by: David Benjamin <davidben@google.com>
Chromium no longer uses it.
Change-Id: I50cc55bad4124305686d299032a2e8ed2cb9d0d7
Reviewed-on: https://boringssl-review.googlesource.com/8691
Reviewed-by: David Benjamin <davidben@google.com>
Upstream added this in a18a31e49d266. The various *_up_ref functions
return a variety of types, but this one returns int because upstream
appears to be trying to unify around that. (See upstream's c5ebfcab713.)
Change-Id: I7e1cfe78c3a32f5a85b1b3c14428bd91548aba6d
Reviewed-on: https://boringssl-review.googlesource.com/8581
Reviewed-by: Adam Langley <alangley@gmail.com>
This replaces the old key_exchange_info APIs and does not require the
caller be aware of the mess around SSL_SESSION management. They
currently have the same bugs around renegotiation as before, but later
work to fix up SSL_SESSION tracking will fix their internals.
For consistency with the existing functions, I've kept the public API at
'curve' rather than 'group' for now. I think it's probably better to
have only one name with a single explanation in the section header
rather than half and half. (I also wouldn't be surprised if the IETF
ends up renaming 'group' again to 'key exchange' at some point. We'll
see what happens.)
Change-Id: I8e90a503bc4045d12f30835c86de64ef9f2d07c8
Reviewed-on: https://boringssl-review.googlesource.com/8565
Reviewed-by: Adam Langley <agl@google.com>
OpenSSL's SSL_OP_NO_* flags allow discontinuous version ranges. This is a
nuisance for two reasons. First it makes it unnecessarily difficult to answer
"are any versions below TLS 1.3 enabled?". Second the protocol does not allow
discontinuous version ranges on the client anyway. OpenSSL instead picks the
first continous range of enabled versions on the client, but not the server.
This is bizarrely inconsistent. It also doesn't quite do this as the
ClientHello sending logic does this, but not the ServerHello processing logic.
So we actually break some invariants slightly. The logic is also cumbersome in
DTLS which kindly inverts the comparison logic.
First, switch min_version/max_version's storage to normalized versions. Next
replace all the ad-hoc version-related functions with a single
ssl_get_version_range function. Client and server now consistently pick a
contiguous range of versions. Note this is a slight behavior change for
servers. Version-range-sensitive logic is rewritten to use this new function.
BUG=66
Change-Id: Iad0d64f2b7a917603fc7da54c9fc6656c5fbdb24
Reviewed-on: https://boringssl-review.googlesource.com/8513
Reviewed-by: David Benjamin <davidben@google.com>
As part of the SignatureAlgorithm change in the TLS 1.3 specification,
the existing signature/hash combinations are replaced with a combined
signature algorithm identifier. This change maintains the existing APIs
while fixing the internal representations. The signing code currently
still treats the SignatureAlgorithm as a decomposed value, which will be
fixed as part of a separate CL.
Change-Id: I0cd1660d74ad9bcf55ce5da4449bf2922660be36
Reviewed-on: https://boringssl-review.googlesource.com/8480
Reviewed-by: Steven Valdez <svaldez@google.com>
Reviewed-by: David Benjamin <davidben@google.com>
It doesn't really convey anything useful. Leave ssl_get_message alone for now
since it's called everywhere in the handshake and I'm about to tweak it
further.
Change-Id: I6f3a74c170e818f624be8fbe5cf6b796353406df
Reviewed-on: https://boringssl-review.googlesource.com/8430
Reviewed-by: Adam Langley <agl@google.com>
Saves us some mess if they're never zero. This also fixes a bug in
ssl3_get_max_client_version where it didn't account for all versions being
disabled properly.
Change-Id: I4c95ff57cf8953cb4a528263b252379f252f3e01
Reviewed-on: https://boringssl-review.googlesource.com/8512
Reviewed-by: David Benjamin <davidben@google.com>
That both exist with nearly the same name is unfortunate. This also does away
with cert_req being unnecessarily tri-state.
Change-Id: Id83e13d0249b80700d9258b363d43b15d22898d8
Reviewed-on: https://boringssl-review.googlesource.com/8247
Reviewed-by: Adam Langley <agl@google.com>
SSL_set_bio has some rather complex ownership story because whether rbio/wbio
are both owning depends on whether they are equal. Moreover, whether
SSL_set_bio(ssl, rbio, wbio) frees ssl->rbio depends on whether rbio is the
existing rbio or not. The current logic doesn't even get it right; see tests.
Simplify this. First, rbio and wbio are always owning. All the weird ownership
cases which we're stuck with for compatibility will live in SSL_set_bio. It
will internally BIO_up_ref if necessary and appropriately no-op the left or
right side as needed. It will then call more well-behaved ssl_set_rbio or
ssl_set_wbio functions as necessary.
Change-Id: I6b4b34e23ed01561a8c0aead8bb905363ee413bb
Reviewed-on: https://boringssl-review.googlesource.com/8240
Reviewed-by: Adam Langley <agl@google.com>
This callback is used by BoringSSL tests in order to simulate the time,
so that the tests have repeatable results. This API will allow consumers
of BoringSSL to write the same sort of tests.
Change-Id: I79d72bce5510bbd83c307915cd2cc937579ce948
Reviewed-on: https://boringssl-review.googlesource.com/8200
Reviewed-by: David Benjamin <davidben@google.com>
The separation is purely historical (what happened to use an SSL_ctrl hook), so
put them all in one place. Make a vague attempt to match the order of the
header file, though we're still very far from matching.
Change-Id: Iba003ff4a06684a6be342e438d34bc92cab1cd14
Reviewed-on: https://boringssl-review.googlesource.com/8189
Reviewed-by: Steven Valdez <svaldez@google.com>
Reviewed-by: David Benjamin <davidben@google.com>
Reorder states and functions by where they appear in the handshake. Remove
unnecessary hooks on SSL_PROTOCOL_METHOD.
Change-Id: I78dae9cf70792170abed6f38510ce870707e82ff
Reviewed-on: https://boringssl-review.googlesource.com/8184
Reviewed-by: David Benjamin <davidben@google.com>
This is getting a little repetitive.
Change-Id: Ib0fa8ab10149557c2d728b88648381b9368221d9
Reviewed-on: https://boringssl-review.googlesource.com/8126
Reviewed-by: Steven Valdez <svaldez@google.com>
Reviewed-by: Adam Langley <agl@google.com>
Reviewed-by: David Benjamin <davidben@google.com>
We've got it in entry points. That should be sufficient. (Do we even need it
there?)
Change-Id: I39b245a08fcde7b57e61b0bfc595c6ff4ce2a07a
Reviewed-on: https://boringssl-review.googlesource.com/8127
Reviewed-by: David Benjamin <davidben@google.com>
Windows SRWLOCK requires you call different functions here. Split
them up in preparation for switching Windows from CRITICAL_SECTION.
BUG=37
Change-Id: I7b5c6a98eab9ae5bb0734b805cfa1ff334918f35
Reviewed-on: https://boringssl-review.googlesource.com/8080
Reviewed-by: Adam Langley <agl@google.com>
This is easier to deploy, and more obvious. This commit reverts a few
pieces of e25775bc, but keeps most of it.
Change-Id: If8d657a4221c665349c06041bb12fffca1527a2c
Reviewed-on: https://boringssl-review.googlesource.com/8061
Reviewed-by: Adam Langley <agl@google.com>
This reverts commit c7eae5a326. pyOpenSSL
expects to be able to call |SSL_read| after a shutdown and get EOF.
Change-Id: Icc5faa09d644ec29aac99b181dac0db197f283e3
Reviewed-on: https://boringssl-review.googlesource.com/8060
Reviewed-by: Adam Langley <agl@google.com>
Constants representing TLS 1.3 are added to allow for future work to be
flagged on TLS1_3_VERSION. To prevent BoringSSL from negotiating the
non-existent TLS 1.3 version, it is explicitly disabled using
SSL_OP_NO_TLSv1_3.
Change-Id: Ie5258a916f4c19ef21646c4073d5b4a7974d6f3f
Reviewed-on: https://boringssl-review.googlesource.com/8041
Reviewed-by: Adam Langley <agl@google.com>
OpenSSL's bbio logic is kind of crazy. It would be good to eventually do the
buffering in a better way (notably, bbio is fragile, if not outright broken,
for DTLS). In the meantime, this fixes a number of bugs where the existence of
bbio was leaked in the public API and broke things.
- SSL_get_wbio returned the bbio during the handshake. It must always return
the BIO the consumer configured. In doing so, internal accesses of
SSL_get_wbio should be switched to ssl->wbio since those want to see bbio.
For consistency, do the same with rbio.
- The logic in SSL_set_rfd, etc. (which I doubt is quite right since
SSL_set_bio's lifetime is unclear) would get confused once wbio got wrapped.
Those want to compare to SSL_get_wbio.
- If SSL_set_bio was called mid-handshake, bbio would get disconnected and lose
state. It forgets to reattach the bbio afterwards. Unfortunately, Conscrypt
does this a lot. It just never ended up calling it at a point where the bbio
would cause problems.
- Make more explicit the invariant that any bbio's which exist are always
attached. Simplify a few things as part of that.
Change-Id: Ia02d6bdfb9aeb1e3021a8f82dcbd0629f5c7fb8d
Reviewed-on: https://boringssl-review.googlesource.com/8023
Reviewed-by: Kenny Root <kroot@google.com>
Reviewed-by: Adam Langley <agl@google.com>
The 'elliptic_curves' extension is being renamed to 'supported_groups'
in the TLS 1.3 draft, and most of the curve-specific methods are
generalized to groups/group IDs.
Change-Id: Icd1a1cf7365c8a4a64ae601993dc4273802610fb
Reviewed-on: https://boringssl-review.googlesource.com/7955
Reviewed-by: David Benjamin <davidben@google.com>
This explicitly forbids an API pattern which formerly kind of worked, but was
extremely buggy (see preceding commits). Depending on how one interprets
close_notify and our API, one might wish to call SSL_shutdown only once
(morally shutdown(SHUT_WR)) and then SSL_read until EOF.
However, this exposes additional confusing states where we might try to send an
alert post-SHUT_WR, etc. Early commits made us more robust here (whether one is
allowed to touch the SSL* after an operattion failed because it read an alert
is... unclear), so we could support it if we wanted to, but this doesn't seem
worth the additional statespace. See if we can get away with not allowing it.
Change-Id: Ie7a7e5520b464360b1e6316c34ec9854b571782f
Reviewed-on: https://boringssl-review.googlesource.com/7433
Reviewed-by: David Benjamin <davidben@google.com>
The existing logic gets confused in a number of cases around close_notify vs.
fatal alert. SSL_shutdown, while still pushing to the error queue, will fail to
notice alerts. We also get confused if we try to send a fatal alert when we've
already sent something else.
Change-Id: I9b1d217fbf1ee8a9c59efbebba60165b7de9689e
Reviewed-on: https://boringssl-review.googlesource.com/7952
Reviewed-by: David Benjamin <davidben@google.com>
SSL_RECEIVED_SHUTDOWN checks in the record layer happen in two different
places. Some operations (but not all) check it, and so does read_bytes. Move it
to get_record.
This check should be at a low-level since it is otherwise duplicated in every
operation. It is also a signal which originates from around the peer's record
layer, so it makes sense to check it near the same code. (This one's in
get_record which is technically lower-level than read_bytes, but we're trying
to get rid of read_bytes. They're very coupled functions.)
Also, if we've seen a fatal alert, replay an error, not an EOF.
Change-Id: Idec35c5068ddabe5b1a9145016d8f945da2421cf
Reviewed-on: https://boringssl-review.googlesource.com/7436
Reviewed-by: David Benjamin <davidben@google.com>
David Benjamin
Steven Valdez
Adam Langley
Adam Langley
Taylor Brandstetter
Matt Braithwaite