The server should not be allowed select a protocol that wasn't
advertised. Callers tend to not really notice and act as if some default
were chosen which is unlikely to work very well.
Change-Id: Ib6388db72f05386f854d275bab762ca79e8174e6
Reviewed-on: https://boringssl-review.googlesource.com/10284
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>
Since we are eliminating DHE support in TLS, this is just a waste of
bytes.
Change-Id: I3a23ece564e43f7e8874d1ec797def132ba59504
Reviewed-on: https://boringssl-review.googlesource.com/10260
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>
These are probably a good idea to ship so long as we have the PSK
callbacks at all, but they're not *completely* standard yet and Android
tests otherwise need updating to know about them. We don't care enough
about PSK to be in a rush to ship them, and taking them out is an easier
default action until then.
Change-Id: Ic646053d29b69a114e2efea61d593d5e912bdcd0
Reviewed-on: https://boringssl-review.googlesource.com/10225
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>
People seem to like adding ifdefs for us for random initialization
functions that are cheap enough to add no-ops stubs for.
Change-Id: I7fb4e978e035329cd81d9bf33ab0d64fde6cc05f
Reviewed-on: https://boringssl-review.googlesource.com/10280
Reviewed-by: Adam Langley <agl@google.com>
If cert_cb runs asynchronously, we end up repeating a large part of very
stateful ClientHello processing. This seems to be mostly fine and there
are few users of server-side cert_cb (it's a new API in 1.0.2), but it's
a little scary.
This is also visible to external consumers because some callbacks get
called multiple times. We especially should try to avoid that as there
is no guarantee that these callbacks are idempotent and give the same
answer each time.
Change-Id: I212b2325eae2cfca0fb423dace101e466c5e5d4e
Reviewed-on: https://boringssl-review.googlesource.com/10224
Commit-Queue: David Benjamin <davidben@google.com>
Commit-Queue: Adam Langley <agl@google.com>
Reviewed-by: Adam Langley <agl@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
Now that ssl_bytes_to_cipher_list is uninteresting, it can be an
implementation detail of ssl3_choose_cipher. This removes a tiny amount
of duplicated TLS 1.2 / TLS 1.3 code.
Change-Id: I116a6bb08bbc43da573d4b7b5626c556e1a7452d
Reviewed-on: https://boringssl-review.googlesource.com/10221
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>
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>
This more accurately reflects the documented contract for
|BN_mod_inverse_odd|.
Change-Id: Iae98dabe3943231859eaa5e798d06ebe0231b9f1
Reviewed-on: https://boringssl-review.googlesource.com/9160
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
Reviewed-by: David Benjamin <davidben@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
In OpenSSL 1.1.0, this API has been renamed to gain a BN prefix. Now
that it's no longer squatting on a namespace, provide the function so
wpa_supplicant needn't carry a BoringSSL #ifdef here.
BUG=91
Change-Id: Iac8e90238c816caae6acf0e359893c14a7a970f1
Reviewed-on: https://boringssl-review.googlesource.com/10223
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>
The name of this has been annoying me every time I've seen it over the
past couple of days. Having a flag with a negation in the name isn't
always bad, but I think this case was.
Change-Id: I5922bf4cc94eab8c59256042a9d9acb575bd40aa
Reviewed-on: https://boringssl-review.googlesource.com/10242
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>
The previous code was not an impressive demonstration of clear thinking
and could reject cases where STARTTLS was actually supported.
Change-Id: I27ce8b401447a49be93f58c9e4eb5c5d8e7b73d4
Reviewed-on: https://boringssl-review.googlesource.com/10241
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>
This gets cURL building against both BoringSSL as it is and BoringSSL
with OPENSSL_VERSION_NUMBER set to 1.1.0.
BUG=91
Change-Id: I5be73b84df701fe76f3055b1239ae4704a931082
Reviewed-on: https://boringssl-review.googlesource.com/10180
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
Commit-Queue: David Benjamin <davidben@google.com>
Commit-Queue: Adam Langley <agl@google.com>
Reviewed-by: Adam Langley <agl@google.com>
The old one was rather confusing. Switch to returning 1/0 for whether
the padding is publicly invalid and then add an output argument which
returns a constant_time_eq-style boolean.
Change-Id: Ieba89d352faf80e9bcea993b716f4b2df5439d4b
Reviewed-on: https://boringssl-review.googlesource.com/10222
Commit-Queue: David Benjamin <davidben@google.com>
Commit-Queue: Adam Langley <agl@google.com>
Reviewed-by: Adam Langley <agl@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
Add the following cases:
- Maximal padding
- Maximal padding with each possible byte position wrong.
- When the input is not publicly too short to find a MAC, but the
unpadded value is too short. (This tests that
EVP_tls_cbc_remove_padding and EVP_tls_cbc_copy_mac coordinate
correctly. EVP_tls_cbc_remove_padding promises to also consider it
invalid padding if there is no room for a MAC.)
Change-Id: I8fe18121afb915e579a8236d0e3ef354f1f835bc
Reviewed-on: https://boringssl-review.googlesource.com/10182
Commit-Queue: David Benjamin <davidben@google.com>
Commit-Queue: Adam Langley <agl@google.com>
Reviewed-by: Adam Langley <agl@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
Change-Id: I44bc5979cb8c15ad8c4f9bef17049312b6f23a41
Reviewed-on: https://boringssl-review.googlesource.com/10200
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 is more progress in letting other stacks use the test runner.
You can provide a per-shim configuration file that includes:
- A list of test patterns to be suppressed (presumably because
they don't work). This setting is ignored if -test is used.
- A translation table of expected errors to shim-specific errors.
BUG=92
Change-Id: I3c31d136e35c282e05d4919e18ba41d44ea9cf2a
Reviewed-on: https://boringssl-review.googlesource.com/9161
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>
Use a separate |size_t| variable for all logic that happens after the
special casing of the negative values of the signed parameter, to
minimize the amount of mixed signed/unsigned math used.
Change-Id: I4aeb1ffce47f889f340f9583684910b0fb2ca7c7
Reviewed-on: https://boringssl-review.googlesource.com/9173
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>
There is a comment "Note from a test above this value is guaranteed to
be non-negative". Reorganize the code to make it more clear that that
is actually the case, especially in the case where sLen == -1.
Change-Id: I09a3dd99458e34102c42d8d3a2f22c16c684c673
Reviewed-on: https://boringssl-review.googlesource.com/9172
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 was renamed to ticket_liftetime_hint in
1e6f11a7ff, which breaks Qt.
Change-Id: I9c6d3097fe96e669f06a4e0880bd4d7d82b03ba8
Reviewed-on: https://boringssl-review.googlesource.com/10181
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>
Initial stab at moving contents of scoped_types.h into
include/openssl/c++ and into the |bssl| namespace.
Started with one file. Will do the remaining ones once this looks good.
Change-Id: I51e2f7c1acbe52d508f1faee7740645f91f56386
Reviewed-on: https://boringssl-review.googlesource.com/9175
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>
EKR is unlikely to resolve this TODO anytime soon.
Change-Id: I2cf6b4ad4f643048d1a683d60b4b90e2b1230aae
Reviewed-on: https://boringssl-review.googlesource.com/9155
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>
This makes it easier to understand the |sLen|-related logic.
Change-Id: I98da4f4f7c82d5481544940407e6cc6a963f7e5b
Reviewed-on: https://boringssl-review.googlesource.com/9171
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
Reviewed-by: David Benjamin <davidben@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
We handle this correctly but never wrote a test for it. Noticed this in
chatting about the second ClientHello.version bug workaround with Eric
Rescorla.
Change-Id: I09bc6c995d07c0f2c9936031b52c3c639ed3695e
Reviewed-on: https://boringssl-review.googlesource.com/9154
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>
The old implementation had a lot of size_t/int confusion. It also
accepted non-minimally-encoded OIDs. Unlike the old implementation, the
new one does not fall back to BIGNUMs and does not attempt to
pretty-print OIDs with components which do not fit in a uint64_t. Add
tests for these cases.
With this new implementation, hopefully we'll have a much easier time
enabling MSVC's size_t truncation warning later.
Change-Id: I602102b97cf9b02d874644f8ef67fe9bac70e45e
Reviewed-on: https://boringssl-review.googlesource.com/9131
Reviewed-by: Adam Langley <agl@google.com>
This eliminates duplicate logic.
Change-Id: I283273ae152f3644df4384558ee4a021f8c2d454
Reviewed-on: https://boringssl-review.googlesource.com/9104
Reviewed-by: David Benjamin <davidben@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
Commit-Queue: David Benjamin <davidben@google.com>
BN_mod_inverse_odd was always being used on 64-bit platforms and was being used
for all curves with an order of 450 bits or smaller (basically, everything but
P-521). We generally don't care much about minor differences in the speed of
verifying signatures using curves other than P-256 and P-384. It is better to
always use the same algorithm.
This also allows |bn_mod_inverse_general|, |bn_mod_inverse_no_branch|, and
|BN_mod_inverse| to be dropped from programs that can somehow avoid linking in
the RSA key generation and RSA CRT recovery code.
Change-Id: I79b94bff23d2b07d5e0c704f7d44538797f8c7a0
Reviewed-on: https://boringssl-review.googlesource.com/9103
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>
The main RSA public modulus size of concern is 2048 bits.
bn_mod_inverse_odd is already used for public moduli of 2048 bits and
smaller on 64-bit platforms, so for 64-bit it is a no-op. For 32-bit
x86, this seems to slightly decrease the speed of RSA signing, but not
by a lot, and plus we don't care about RSA signing performance much on
32-bit platforms. It's better to have all platforms using the same
algorithms.
Change-Id: I869dbfc98994e36a04a535c1fe63b14a902a4f13
Reviewed-on: https://boringssl-review.googlesource.com/9102
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 is a step towards exposing |bn_mod_inverse_odd| for use outside
of crypto/bn/gcd.c.
Change-Id: I2968f1e43306c03775b3573a022edd92f4e91df2
Reviewed-on: https://boringssl-review.googlesource.com/9101
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 is in preparation for factoring out the binary Euclidean
implementation (the one used for odd numbers that aren't too big) for
direct use from outside of crypto/bn/gcd.c. The goal is to make the
resultant |BN_mod_inverse_odd|'s signature similar to
|BN_mod_inverse_blinded|. Thus, the logic for reducing the final result
isn't factored out because that yet-to-be-created |BN_mod_inverse_odd|
will need to do it itself.
Change-Id: Iaecb79fb17d13c774c4fb6ade8742937780b0006
Reviewed-on: https://boringssl-review.googlesource.com/9100
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 is very far from all of it, but I did some easy ones before I got
bored. Snapshot the progress until someone else wants to continue this.
BUG=22
Change-Id: I2609e9766d883a273e53e01a75a4b1d4700e2436
Reviewed-on: https://boringssl-review.googlesource.com/9132
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>
Inch towards OpenSSL 1.1.0 compatibility.
BUG=91
Change-Id: Ia45b6bdb5114d0891fdffdef0b5868920324ecec
Reviewed-on: https://boringssl-review.googlesource.com/9140
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
Fix non-standard variable names, return value convention, unsigned vs
size_t, etc. This also fixes one size_t truncation warning.
BUG=22
Change-Id: Ibe083db90e8dac45d64da9ead8f519dd2fea96ea
Reviewed-on: https://boringssl-review.googlesource.com/9133
Commit-Queue: David Benjamin <davidben@google.com>
Commit-Queue: Adam Langley <agl@google.com>
Reviewed-by: Adam Langley <agl@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
OBJ_obj2txt's implementation is kind of scary. Also it casts between int
and size_t a lot. In preparation for rewriting it, add a test.
Change-Id: Iefb1d0cddff58d67e5b04ec332477aab8aa687b6
Reviewed-on: https://boringssl-review.googlesource.com/9130
Reviewed-by: Adam Langley <agl@google.com>
We're starting to get quite a lot of these ALL-CAPS.md documents.
There's been enough questions around how to properly use types like
EVP_MD_CTX that we probably should write down some of these common
rules.
Change-Id: I125f4e82efb168a071b54ff76c5af34c42ff4800
Reviewed-on: https://boringssl-review.googlesource.com/9115
Reviewed-by: Adam Langley <agl@google.com>
tls13_process_certificate can take a boolean for whether anonymous is
allowed. This does change the error on the client slightly, but I think
this is correct anyway. It is not a syntax error for the server to send
no certificates in so far as the Certificate message allows it. It's
just illegal.
Change-Id: I1af80dacf23f50aad0b1fbd884bc068a40714399
Reviewed-on: https://boringssl-review.googlesource.com/9072
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 have tests for this as a server, but none as a client. Extend the
certificate verification tests here. This is in preparation for ensuring
that TLS 1.3 session resumption works correctly.
Change-Id: I9ab9f42838ffd69f73fbd877b0cdfaf31caea707
Reviewed-on: https://boringssl-review.googlesource.com/9111
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>
No sense in having it in both the 1.2 and 1.3 code.
Change-Id: Ib3854714afed24253af7f4bcee26d25e95a10211
Reviewed-on: https://boringssl-review.googlesource.com/9071
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>
While the sanity check isn't insane (one should arrange for sessions to
be invalidated once client auth settings change, and a sid_ctx is one
way to do it), this check lives in a poor place to enforce configuration
mistakes. To be effective, it needs to happen at the start of the
handshake, independent of the ClientHello from the peer.
But the benefit this check gives is low compared to the trouble it will
be to continually maintain this difference from OpenSSL (our own
ssl_test and bssl_shim forget to set a dummy sid_ctx). Instead, remove
it so we don't have to duplicate it across TLS 1.2 and TLS 1.3. Also so
we don't have weird failures which only manifest once a resuming client
connects.
Change-Id: Ia7f88711701afde5e26b7782c2264ce78dccc89b
Reviewed-on: https://boringssl-review.googlesource.com/9112
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'd gotten rid of the macros, but not the underlying asn1_GetSequence
which is unused. Sadly this doesn't quite get rid of ASN1_(const_)?CTX.
There's still some code in the rest of crypto/asn1 that uses it.
Change-Id: I2ba8708ac5b20982295fbe9c898fef8f9b635704
Reviewed-on: https://boringssl-review.googlesource.com/9113
Commit-Queue: David Benjamin <davidben@google.com>
Commit-Queue: Adam Langley <agl@google.com>
Reviewed-by: Adam Langley <agl@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
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 almost forgot to handle this in TLS 1.3, so add a test for it.
Change-Id: I28600325d8fb6c09365e909db607cbace12ecac7
Reviewed-on: https://boringssl-review.googlesource.com/9093
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>
Save a little bit of typing at the call site.
Change-Id: I818535409b57a694e5e0ea0e9741d89f2be89375
Reviewed-on: https://boringssl-review.googlesource.com/9090
Reviewed-by: Adam Langley <agl@google.com>
