We fill in placeholder values of all zeros fairly often in TLS now,
as workarounds for messages being constructed in the wrong order.
draft-12 of ECH adds even more of these. Add a helper so we don't need
to interrupt an || chain with a memset.
Change-Id: Id4f9d988ee67598645a01637cc9515b475c1aec2
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/48909
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
The session ID field cannot exceed 32 bytes, and we size various buffers
based on this. Test that our parsers correctly handle this.
Also fix the -wait-for-debugger flag. I broke it recently by removing
the statusShimStarted message.
Change-Id: I29bb177f29a79bb4904fb5ba3cedfb0b6b856061
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/48907
Reviewed-by: Adam Langley <agl@google.com>
The cipher suite, like the version, is determined by the first server
message, independent of whether it's ServerHello or HelloRetryRequest.
We can simplify this by just processing it before we branch on which it
was.
Change-Id: I747f515e9e5b05a42cbed6e7844808d0fc79a30b
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/48906
Reviewed-by: Adam Langley <agl@google.com>
absl::Span, base::span, and std::span have first() and last() methods
which give prefixes and suffixes. first() just saves 5 characters, but
last() is nicer to write than subspan() for suffixes.
Unlike subspan(), they also do not have clipping behavior, so we're
guaranteed the length is correct. The clipping behavior comes from
absl::Span::subspan() and is not present in std::span or base::span.
I've left it in, in case we switch to absl::Span in the future, but I
imagine absl::Span will need to migrate this at some point.
Change-Id: I042dd6c566b6d753ec6de9d84e8c09ac7c270267
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/48905
Reviewed-by: Adam Langley <agl@google.com>
It's a bit of a mess, but BIO-like APIs typically return -1 on error and
0 for EOF.
Change-Id: Ibdcb70e1009ffebf6cc6df40804dc4a178c7199e
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/48845
Reviewed-by: Adam Langley <agl@google.com>
The stack consumption of the HRSS functions is causing issues in
stack-constrained environments. Therefore allocate many variables on the
heap. This means that several HRSS_ functions now allocate, and thus can
fail, where they couldn't before. Callers that ignore the return value
and don't have crash-on-failure mallocs will still be safe, although
things will fail to decrypt later on.
Somehow, this actually makes key generation _faster_ on my machine. (I
don't know. Better alignment? Fewer L1 collisions?) The other operations
are slightly slower, as expected.
Before:
Did 17390 HRSS generate operations in 3054088us (5694.0 ops/sec)
Did 225000 HRSS encap operations in 3000512us (74987.2 ops/sec)
Did 87000 HRSS decap operations in 3014525us (28860.3 ops/sec)
After:
Did 21300 HRSS generate operations in 3026637us (7037.5 ops/sec)
Did 221000 HRSS encap operations in 3008911us (73448.5 ops/sec)
Did 84000 HRSS decap operations in 3007622us (27929.0 ops/sec)
Change-Id: I2312df8909af7d8d250c7c483c65038123f21ad9
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/48345
Commit-Queue: Adam Langley <agl@google.com>
Reviewed-by: David Benjamin <davidben@google.com>
Found by OSS-Fuzz. This comes up if you enable client certificates and
the draft ECH implementation on the server.
Bug: 275, oss-fuzz:35815
Change-Id: I0b4fcc994f7238f8a3cf1f1934672bac0cee0cfb
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/48425
Reviewed-by: Adam Langley <agl@google.com>
If a client offers ECH, but the server rejects it, the client completes
the handshake with ClientHelloOuter in order to authenticate retry keys.
Implement this flow. This is largely allowing the existing handshake to
proceed, but with some changes:
- Certificate verification uses the other name. This CL routes this up to
the built-in verifier and adds SSL_get0_ech_name_override for the
callback.
- We need to disable False Start to pick up server Finished in TLS 1.2.
- Client certificates, notably in TLS 1.3 where they're encrypted,
should only be revealed to the true server. Fortunately, not sending
client certs is always an option, so do that.
Channel ID has a similar issue. I've just omitted the extension in
ClientHelloOuter because it's deprecated and is unlikely to be used
with ECH at this point. ALPS may be worth some pondering but, the way
it's currently used, is not sensitive.
(Possibly we should change the draft to terminate the handshake before
even sending that flight...)
- The session is never offered in ClientHelloOuter, but our internal
book-keeping doesn't quite notice.
I had to replace ech_accept with a tri-state ech_status to correctly
handle an edge case in SSL_get0_ech_name_override: when ECH + 0-RTT +
reverify_on_resume are all enabled, the first certificate verification
is for the 0-RTT session and should be against the true name, yet we
have selected_ech_config && !ech_accept. A tri-state tracks when ECH is
actually rejected. I've maintained this on the server as well, though
the server never actually cares.
Bug: 275
Change-Id: Ie55966ca3dc4ffcc8c381479f0fe9bcacd34d0f8
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/48135
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
base64.StdEncoding.EncodeToString is very long to write out.
Change-Id: Ie987d483513e4192a31c8562b9cf25e99f8a838b
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/48134
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
ssl_update_cache takes the cache lock to add to the session cache,
releases it, and then immediately takes and releases the lock to
increment handshakes_since_cache_flush. Then, in 1/255 connections, does
the same thing again to flush stale sessions.
Merge the first two into one lock. In doing so, move ssl_update_cache to
ssl_session.cc, so it can access a newly-extracted add_session_lock.
Also remove the mode parameter (the SSL knows if it's a client or
server), and move the established_session != session check to the
caller, which more directly knows whether there was a new session.
Also add some TSan coverage for this path in the tests. In an earlier
iteration of this patch, I managed to introduce a double-locking bug
because we weren't testing it at all. Confirmed this test catches both
double-locking and insufficient locking. (It doesn't seem able to catch
using a read lock instead of a write lock in SSL_CTX_flush_sessions,
however. I suspect the hash table is distributing the cells each thread
touches.)
Update-Note: This reshuffles some locks around the session cache.
(Hopefully for the better.)
Change-Id: I78dca53fda74e036b90110cca7fbcc306a5c8ebe
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/48133
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
In renegotiation handshakes and, later, ECH ClientHelloOuter handshakes,
we don't want to add sessions to the session cache. We also don't want
to release a session as resumable until the handshake completes.
Ideally we'd only construct SSL_SESSION at the end of the handshake, but
existing APIs like SSL_get_session must work mid-handshake, so
SSL_SESSION is both a handle to immutable resumption state, and a
container for in-progress connection properties. We manage this with a
not_resumable flag that's only cleared after the handshake is done and
the SSL_SESSION finalized.
However, TLS 1.2 ticket renewal currently clears the flag too early and
breaks the invariant. This won't actually affect renegotiation or
ClientHelloOuter because those handshakes never resume. Still, we can
maintain the invariant storing the copy in hs->new_session. Note this
does sacrifice a different invariant: previously, ssl->session and
hs->new_session were never set at the same time.
This change also means ssl_update_cache does not need to special-case
ticket renewal.
Change-Id: I03230cd9c63e5bee6bd60cd05c0439e16533c6d4
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/48132
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
Some tests run three connections, resuming a renewed ticket.
Particularly the way TLS 1.2 ticket renewal works, the client logic
could accidentally report the old session up to the application. Our
runner tests would not currently notice (though one of the tests in
ssl_tests does).
Make runner tests also check this by cycling ticket keys between
connection attempts. This also makes newSessionsOnResume apply even if
the test did not specify a resumeConfig.
Change-Id: I95375c01adf6ad62de65ecf8aed3c286a0571875
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/48131
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
This addresses some feedback in
https://boringssl-review.googlesource.com/c/boringssl/+/48131/1/ssl/test/runner/runner.go#1555,
pulled into a separate CL for clarity:
First, take the listener, waitChan, exec.Cmd trio and wrap them into a
shimProcess type. shimProcess is now responsible for the -port flag, so
it can manage the TCPListener internally.
Next, take the core test loop and moves it into a doExchanges()
function, so that it can use a more usual early return pattern for
errors, rather than thread err == nil through all the control flow. With
shimProcess pulled out, doExchanges() can just take a *shimProcess.
Finally, unacted-on err variable has gotten very far from where it's
actually used. Rename it to localErr, to align with our
expectedLocalError machinery.
Change-Id: I63697a5d79040ad77fa06c125253ec5031aeaf5c
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/48186
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
The one place where LHASH_OF(T) appears in public APIs is
X509V3_EXT_conf_nid. This is only ever called with conf = NULL, but
cryptography.io needs to utter the type name as part of bindings. Thus
this CL keeps DECLARE_LHASH_OF and LHASH_OF macros public and the others
private.
Update-Note: BoringSSL no longer provides a general-purpose hash table
to callers. Use the language's standard library, or another
implementation.
Change-Id: Ibfc65c4b4bf35abf5b1919658d0c52e4004e6629
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/48205
Reviewed-by: Adam Langley <agl@google.com>
This is a bit more self-explanatory, especially now that TLS 1.0 is the
minimum version we implement anyway.
Change-Id: Ic65e9f90bb5cd747328bd9e30b976d1e124c7764
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/48130
Reviewed-by: Adam Langley <agl@google.com>
This was added in draft-11, which I'll update to more broadly in a
follow-up CL. This is an easily separable component: we don't want to
allow the DNS to arbitrarily insert strings in the ClientHello, so
invalid public names are rejected.
Unfortunately, we have a bit of a mess because DNS syntax does not
exclude IPv4 literals, yet everyone sticks DNS and IP literals in the
same string. The RFC3986 rules are alright, but don't match reality.
Reality is (probably?) the WHATWG rules, which are a mess.
The load-bearing bit of the spec is that, at certificate verification,
you should reject whatever strings your application refuses to represent
as a DNS name. I'll have Chromium call into its URL parser.
https://www.ietf.org/archive/id/draft-ietf-tls-esni-11.html#section-6.1.4.3-3
But there's still a bit at the validation step where clients "SHOULD"
run the IPv4 parser. In case downstream logic forgets, I've gone ahead
and implemented the WHATWG IPv4 parser.
https://www.ietf.org/archive/id/draft-ietf-tls-esni-11.html#section-4-6.6.1
Bug: 275
Change-Id: I15aa1ac0391df9c3859c44b8a259296e1907b7d4
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/48085
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
This CL fixes a couple of things. First, we never tested that SSL_write
refuses to write application data after a fatal alert, so add some tests
here. With those tests, we can revise some of this logic:
Next, this removes the write_shutdown check in SSL_write and instead
relies on the lower-level versions of the check in the write_app_data,
etc., hooks. This improves error-reporting on handshake errors:
We generally try to make SSL_do_handshake errors sticky, analogous to
handshakeErr in the Go implementation. SSL_write and SSL_read both
implicitly call SSL_do_handshake. Callers driving the two in parallel
will naturally call SSL_do_handshake twice. Since the error effectively
applies to both operations, we save and replay handshake errors
(hs->error).
Handshake errors typically come with sending alerts, which also sets
write_shutdown so we don't try to send more data over the channel.
Checking this early in SSL_write means we don't get a chance to replay
the handshake error. So this CL defers it, and the test ensures we still
ultimately get it right.
Finally, https://crbug.com/1078515 is a particular incarnation of this.
If the server enables 0-RTT and then reverts to TLS 1.2, clients need
to catch the error and retry. There, deferring the SSL_write check
isn't sufficient, because the can_early_write bit removes the write
path's dependency on the handshake, so we don't call into
SSL_do_handshake at all.
For now, I've made this error path clear can_early_write. I suspect
we want it to apply to all handshake errors, though it's weird because
the handshake error is effectively a read error in 0-RTT. We don't
currently replay record decryption failures at SSL_write, even though
those also send a fatal alert and thus break all subsequent writes.
Bug: chromium:1078515
Change-Id: Icdfae6a8f2e7c1b1c921068dca244795a670807f
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/48065
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
Although not permitted by the TLS specification, systems sometimes
ossify TLS extension order, or byte offsets of various fields. To
keep the ecosystem healthy, add an API to reorder ClientHello
extensions.
Since ECH, HelloRetryRequest, and HelloVerifyRequest are sensitive to
extension order, I've implemented this by per-connection permutation of
the indices in the kExtensions structure. This ensures that all
ClientHellos within a connection are consistently ordered. As follow-up
work, permuting the other messages would also be nice, though any server
messages would need to be incorporated in handshake hints.
Change-Id: I18ce39b4df5ee376c654943f07ec26a50e0923a9
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/48045
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
This would have caught an issue with some tests I was working on. It
also catches an issue with some per-message tests, so fix those.
Change-Id: I6b3ad8e0db0b1a6ccac4b346dcc652b16b73e006
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/48046
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
Based on an initial implementation by Dan McArdle at
https://boringssl-review.googlesource.com/c/boringssl/+/46784
This CL contains most of a client implementation for
draft-ietf-tls-esni-10. The pieces missing so far, which will be done in
follow-up CLs are:
1. While the ClientHelloInner is padded, the server Certificate message
is not. I'll add that once we resolve the spec discussions on how to
do that. (We were originally going to use TLS record-level padding,
but that doesn't work well with QUIC.)
2. The client should check the public name is a valid DNS name before
copying it into ClientHelloOuter.server_name.
3. The ClientHelloOuter handshake flow is not yet implemented. This CL
can detect when the server selects ClientHelloOuter, but for now the
handshake immediately fails. A follow-up CL will remove that logic
and instead add the APIs and extra checks needed.
Otherwise, this should be complete, including padding and compression.
The main interesting point design-wise is that we run through
ClientHello construction multiple times. We need to construct
ClientHelloInner and ClientHelloOuter. Then each of those has slight
variants: EncodedClientHelloInner is the compressed form, and
ClientHelloOuterAAD just has the ECH extension erased to avoid a
circular dependency.
I've computed ClientHelloInner and EncodedClientHelloInner concurrently
because the compression scheme requires shifting the extensions around
to be contiguous. However, I've computed ClientHelloOuterAAD and
ClientHelloOuter by running through the logic twice. This probably can
be done better, but the next draft revises the construction anyway, so
I'm thinking I'll rework it then. (In the next draft, we use a
placeholder payload of the same length, so we can construct the
ClientHello once and fill in the payload.)
Additionally, now that we have a client available in ssl_test, this adds
a threading test to confirm that SSL_CTX_set1_ech_keys is properly
synchronized. (Confirmed that, if I drop the lock in
SSL_CTX_set1_ech_keys, TSan notices.)
Change-Id: Icaff68b595035bdcc73c468ff638e67c84239ef4
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/48004
Reviewed-by: Adam Langley <agl@google.com>
We'll probably need to make this more complex later, but this should be
a start. I had hoped this would also simplify tests, MakeECHConfig() was
still needed to generate weird inputs for tests. I've instead tidied
that up a bit with a params structure. Now the only hard-coded ECHConfig
in tests is to check the output of the new API.
Bug: 275
Change-Id: I640a224fb4b7a7d20e8a2cd7a1e75d1e3fe69936
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/48003
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
Previously we would extract the KEM ID from the ECHConfig and then parse
the private key using the corresponding KEM type. This CL makes it take
a pre-pared EVP_HPKE_KEY and checks it matches. This does require the
caller pass the key type through externally, which is probably prudent?
(On the other hand we are still inferring config from the rest of the
ECHConfig... maybe we can add an API to extract the EVP_HPKE_KEM from a
serialized ECHConfig if it becomes a problem. I could see runner or tool
wanting that out of convenience.)
The immediate motivation is to add APIs to programmatically construct
ECHConfigs. I'm thinking we can pass a const EVP_HPKE_KEY * to specify
the key, at which point it's weird for SSL_ECH_KEYS_add to look
different.
Bug: 275
Change-Id: I2d424323885103d3fe0a99a9012c160baa8653bd
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/48002
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
We didn't correctly handle tests where the versions figure into
resumeConfig and got by because the test didn't actually check the
version. Run it more accurately, and check more fields. Also add a
skipVersionNameCheck option for when the heuristic doesn't work.
(Some of the tests specify a TLS maximum version by passing in all the
-no-tls1, etc., flags for the other versions. Moreover, some of them
will set no flags for a maximum of TLS 1.3. Suppress the check on those
tests.)
Change-Id: I72c069b1baed09e29bf502036957fe0e90edbe60
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/48000
Reviewed-by: Adam Langley <agl@google.com>
While the previous CL fixed a bug in computing this padding, we don't
actually need to pad the second (cleartext) ClientHello anyway. This
padding is to work around bugs in old F5 and WebSphere servers, which do
not speak TLS 1.3. Save a few bytes.
Change-Id: I9b5d9bb1c0d880f1b1c9182667a9d3d82588c04c
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/47999
Reviewed-by: Adam Langley <agl@google.com>
If we're dropping the PSK extension due to an HRR with mismatched hash
(looking back at that, we could easily have avoided that nuisance...
I've filed [0] on rfc8446bis), we don't predict the length correctly.
The consequence is we don't pad the second ClientHello to the desired
range. Fix this and add an assert.
[0] https://github.com/tlswg/tls13-spec/issues/1227
Change-Id: I47d880b6bdafa95840f7513b6b7718851f22554d
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/47998
Reviewed-by: Adam Langley <agl@google.com>
Computing the binders on ClientHelloInner is a little interesting. While
I'm in the area, tidy this up a bit. The exploded parameters may as well
be an SSL_SESSION, and hash_transcript_and_truncated_client_hello can
just get folded in.
Change-Id: I9d3a7e0ae9f391d6b9a23b51b5d7198e15569b11
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/47997
Reviewed-by: Adam Langley <agl@google.com>
This makes calls to ssl_add_clienthello_tlsext a hair easier. Also we
only apply the [256, 511) compatibility hack to TLS, so we can just use
a constant.
Bug: 275
Change-Id: Ia2b5192aeef0cd8848ecfa1ea3b89a0a7382ff1a
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/47996
Reviewed-by: Adam Langley <agl@google.com>
ssl_add_clienthello_tlsext is about to get kinda messy with ECH. Move
the padding and GREASE extensions into a few helpers.
Bug: 275
Change-Id: I3bb702fb79dce4be68490c4a8fd889121ecdae58
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/47995
Reviewed-by: Adam Langley <agl@google.com>
For TLS 1.3, since the bulk of extensions move to EncryptedExtensions,
we made the extension callbacks apply to EncryptedExtensions and pulled
the few ServerHello extensions out of the callback system. This means
the ServerHello naming is a little confusing.
We probably should rename these callbacks, though parse_server is a bit
ambiguous as to whether this is "parse the extension from the server" or
"parse as a server". For now, add a comment.
Change-Id: If1aa0846426de2cc8dcb2253695a8dd3285f7b76
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/47994
Reviewed-by: Adam Langley <agl@google.com>
May not be strictly necessary, but similarly easier to reason about when
we need to interweave multiple ClientHellos.
Bug: 275
Change-Id: I9f85787860f3e8ce1653331ce52343d5bf5def23
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/47992
Reviewed-by: Adam Langley <agl@google.com>
This is less effective than it seems because both
((const SSL_HANDSHAKE*)hs)->ssl and ((const SSL*)ssl)->s3 are both
non-const pointers. I considered moving hs->ssl to hs->ssl_ and adding
const-overloaded accessors, but I'd need to repeat the same with
ssl->s3, at which point this seemed not worth it yet. Maybe later if we
rewrite everything to more idiomatic C++.
Bug: 275
Change-Id: I9912a3df205916fdf2191a3687013d717528038d
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/47991
Reviewed-by: Adam Langley <agl@google.com>
This is kinda annoying and, like the grease_seed, demonstrates a
shortcoming with the idea of making add_clienthello completely const.
Strictly speaking, we only need idempotence. But I think this is okay:
const is much easier to enforce than idempotence, and we'll likely need
to do this anyway:
- While not in the current draft, I expect the draft's eventual HRR
resolution to retain the ECH extension, GREASE or not, on ECH reject.
Things are somewhat violating RFC8446 HRR rules right now. That means
we'll need to stash the ECH payload regardless.
- ECH binds all the other extensions in the outer ClientHello, so
computing the payload will need to move outside the callback system
anyway.
In some sense, all this is shifting our ClientHello output from the
"immediate mode" end of the spectrum (as we usually use) to the
"retained mode" end, which I suppose makes sense as this message becomes
more intricate.
Bug: 275
Change-Id: I9eb8cd1cde2ce264345b6ed3ee526d4eab81e911
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/47990
Reviewed-by: Adam Langley <agl@google.com>
This lets ssl_get_grease_value be const. The lazy thing isn't a
deal-breaker (we only need idempotence, and a non-thread-safe const also
works fine), but just initializing it earlier seems simpler.
Bug: 275
Change-Id: Iad228ea4a9146ede9a3849f3339f7ec9e698e6eb
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/47988
Reviewed-by: Adam Langley <agl@google.com>
This is now never used. Instead, we rely on each renegotiation creating
a new handshake structure with fresh state. This simplifies things for
ECH.
(We probably could make an init hook work with ECH's two-ClientHello
scheme by either maintaining separate state per ClientHello or calling
init once for both ClientHellos. But the few uses of init were
removable, so this is easier.)
Bug: 275
Change-Id: Ie5e132fe072e5ea8db21ca16aa53fcd0895d8e48
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/47987
Reviewed-by: Adam Langley <agl@google.com>
Like the early_data CL, this does shift a bit of logic that was
previously hidden away in the callbacks. For key_share, this is probably
a good move independent of ECH. The logic around HRR, etc., was a little
messy.
Bug: 275
Change-Id: Iafbcebdf66ce1f7957d798a98ee6b996fff24639
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/47986
Reviewed-by: Adam Langley <agl@google.com>
Also add ECH GREASE state into the mix. Clearing this isn't critical,
especially now that we have an SSL_HANDSHAKE structure, but it's easy
enough.
Bug: 275
Change-Id: If1aa8d5c0c8fdb5af710852778ce452c507a2524
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/47985
Reviewed-by: Adam Langley <agl@google.com>
ECH requires that we construct two ClientHellos. That means our
add_clienthello callbacks will need to be called multiple times and
should be const. (They already are called multiple times for
HelloRetryRequest, but we currently thread that through the callbacks a
bit. With ECH, I think we need to make them pure serialization.)
Bug: 275
Change-Id: I11f8195fd2ec4b8639f0a2af01a24d4974445580
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/47984
Reviewed-by: Adam Langley <agl@google.com>
While decompression is deterministic, compression is not. New revisions
of the compression algorithm may start using different (hopefully
smaller!) compressions. So this doesn't cause hint mismatches, add a
certificate compression hint. If the shim's Certificate message matches
the handshaker, we'll reuse the already compressed message.
This also adds what appears to be a missing test for when the server
cannot find compression algorithms in common.
Change-Id: Idbedaceb20208463d8f61581ee27971c17fcd126
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/48005
Reviewed-by: Adam Langley <agl@google.com>
This implements draft-ietf-tls-esni-10.
This will be used to test the client implementation. While I'm here,
I've switched the setup logic in the server tests to use the new
ServerECHConfig type. I'll probably need to patch in various features
later for testing, but this should be a usable starting point.
Based on an initial implementation by Dan McArdle in
https://boringssl-review.googlesource.com/c/boringssl/+/46786
Bug: 275
Change-Id: I69523cda70c3da2ae505bcab837fd358195fb9e9
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/47967
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
Noticed this while I was in the area. We currently use an extremely lax
parse that even tolerates syntax errors. Instead use a strict parse that
ensures our client only sends what we expect.
Change-Id: Ifb0e1e1698489ff217db0c7a0317caa885e20759
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/47966
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
Having the nil vs. non-nil []byte for the sake of a couple tests with
invalid payloads is tedious. Use separate fields instead.
Bug: 275
Change-Id: I557d914d60ce94d68796c05162ff3dd2ab7684db
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/47965
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
An ECHConfig is like a certificate in that knowing the fields isn't
sufficient. The exact byte representation is significant. (The ECHConfig
is bound into the encryption.) But the ECHConfig type only has fields,
so runner can only represent ECHConfigs that are the output of our
serialization function.
This matters less as a client testing a server because the server can
only parse ECHConfigs with fields we support. But as a server testing a
client, we need to see how the client reacts to extra extensions, etc.
Just using []byte to represent ECHConfigs is inconvenient, so instead
pattern this after x509.Certificate: you can parse one from a byte
string (not currently included since we don't need it yet), or you can
construct a new one from a template with the fields you want.
Bug: 275
Change-Id: I6602d0780b1cef12b6c4b442999bdff7b3d7dd70
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/47964
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
The test was not actually using a repeated config ID.
Bug: 275
Change-Id: I69519fde196247abb07dceba1da1bac22188f13f
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/47912
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
The first thing any deployment will want to monitor is whether ECH was
actually used. Also it's useful if the command-line tool can output
this. (The alert is how the client signals it discarded the connection
due to ECH reject.)
This also disables ECH with the handoff mechanism for now. (The
immediate cause being that ech_accept isn't serialized.) We'll probably
need to make some decisions around the ordering here, since ECH affects
where the true ClientHello is available.
Bug: 275
Change-Id: Ie4559733290e653a514fcd94431090bf86bc3172
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/47911
Reviewed-by: Adam Langley <agl@google.com>
David Benjamin
Adam Langley