This reverts commit e60e2a483b. Turns out we do
still use clock_gettime in speed.cc.
Change-Id: Idab42ef2863345c3d1409b5d33c3e36b41739e1f
Reviewed-on: https://boringssl-review.googlesource.com/4894
Reviewed-by: Adam Langley <agl@google.com>
Yes, OpenSSL lets you randomly change its internal state. This is used
as part of server-side renegotiation. Server-side renegotiation is gone.
BUG=429450
Change-Id: Ic1b013705734357acf64e8bf89a051b2b7521c64
Reviewed-on: https://boringssl-review.googlesource.com/4828
Reviewed-by: Adam Langley <agl@google.com>
They were added to avoid accidentally enabling renego for a consumer which set
them to zero to break the handshake on renego. Now that renego is off by
default, we can get rid of them again.
Change-Id: I2cc3bf567c55c6562352446a36f2b5af37f519ba
Reviewed-on: https://boringssl-review.googlesource.com/4827
Reviewed-by: Adam Langley <agl@google.com>
It's never called and the state is meaningless now.
Change-Id: I5429ec3eb7dc2b789c0584ea88323f0ff18920ae
Reviewed-on: https://boringssl-review.googlesource.com/4826
Reviewed-by: Adam Langley <agl@google.com>
When the peer or caller requests a renegotiation, OpenSSL doesn't
renegotiate immediately. It sets a flag to begin a renegotiation as soon
as record-layer read and write buffers are clear. One reason is that
OpenSSL's record layer cannot write a handshake record while an
application data record is being written. The buffer consistency checks
around partial writes will break.
None of these cases are relevant for the client auth hack. We already
require that renego come in at a quiescent part of the application
protocol by forbidding handshake/app_data interleave.
The new behavior is now: when a HelloRequest comes in, if the record
layer is not idle, the renegotiation is rejected as if
SSL_set_reject_peer_renegotiations were set. Otherwise we immediately
begin the new handshake. The server may not send any application data
between HelloRequest and completing the handshake. The HelloRequest may
not be consumed if an SSL_write is pending.
Note this does require that Chromium's HTTP stack not attempt to read
the HTTP response until the request has been written, but the
renegotiation logic already assumes it. Were Chromium to drive the
SSL_read state machine early and the server, say, sent a HelloRequest
after reading the request headers but before we've sent the whole POST
body, the SSL state machine may racily enter renegotiate early, block
writing the POST body on the new handshake, which would break Chromium's
ERR_SSL_CLIENT_AUTH_CERT_NEEDED plumbing.
BUG=429450
Change-Id: I6278240c3bceb5d2e1a2195bdb62dd9e0f4df718
Reviewed-on: https://boringssl-review.googlesource.com/4825
Reviewed-by: Adam Langley <agl@google.com>
The only case where renego is supported is if we are a client and the
server sends a HelloRequest. That is still needed to support the renego
+ client auth hack in Chrome. Beyond that, no other forms of renego will
work.
The messy logic where the handshake loop is repurposed to send
HelloRequest and the extremely confusing tri-state s->renegotiate (which
makes SSL_renegotiate_pending a lie during the initial handshake as a
server) are now gone. The next change will further simplify things by
removing ssl->s3->renegotiate and the renego deferral logic. There's
also some server-only renegotiation checks that can go now.
Also clean up ssl3_read_bytes' HelloRequest handling. The old logic relied on
the handshake state machine to reject bad HelloRequests which... actually that
code probably lets you initiate renego by sending the first four bytes of a
ServerHello and expecting the peer to read it later.
BUG=429450
Change-Id: Ie0f87d0c2b94e13811fe8e22e810ab2ffc8efa6c
Reviewed-on: https://boringssl-review.googlesource.com/4824
Reviewed-by: Adam Langley <agl@google.com>
With NO_ASM defined, the recent AEAD changes broke the tests. The
problem is that the generic CBC mode code tests whether in != out and
omits to save the IV, assuming that it'll be able to read the old
ciphertext block.
However, consider the case where out = in - 16:
1 2 3 4
|-------|-------|------|-------|
^ ^
| |
out in
First time around, 1 = decrypt(2) ^ iv and everything is fine, because
the IV was preconfigured. However, the next iteration of the loop sets
2 = decrypt(3) and tries to XOR it with the contents of the previous
ciphertext block… from 2.
Change-Id: Ibabff430704fad246de132b4d6d514f6a0362734
This dates back to upstream's a2a0158959e597188c10fbfeaf61888b2df2e587.
It seems to be a remnant of those SSL_clear calls in the handshake state
machine which... were also bizarre and since gone.
Since SSL_clear is to drop the current connection but retain the
configuration, it doesn't really make sense to forbid it while you're
mid-handshake.
This removes another consumer of s->renegotiate.
BUG=429450
Change-Id: Ifac6bf11644447fd5571262bed7421684739bc39
Reviewed-on: https://boringssl-review.googlesource.com/4823
Reviewed-by: Adam Langley <agl@google.com>
ssl_cipher_list_to_bytes is client-only, so s->renegotiate worked, but
the only reason the other two worked is because s->renegotiate isn't a
lie on the server before ServerHello.
BUG=429450
Change-Id: If68a986c6ec4a0f16e57a6187238e05b50ecedfc
Reviewed-on: https://boringssl-review.googlesource.com/4822
Reviewed-by: Adam Langley <agl@google.com>
Now that WebRTC honors packet boundaries (https://crbug.com/447431), we
can start enforcing them correctly. Configuring read-ahead now does
nothing. Instead DTLS will always set "read-ahead" and also correctly
enforce packet boundaries when reading records. Add tests to ensure that
badly fragmented packets are ignored. Because such packets don't fail
the handshake, the tests work by injecting an alert in the front of the
handshake stream and ensuring the DTLS implementation ignores them.
ssl3_read_n can be be considerably unraveled now, but leave that for
future cleanup. For now, make it correct.
BUG=468889
Change-Id: I800cfabe06615af31c2ccece436ca52aed9fe899
Reviewed-on: https://boringssl-review.googlesource.com/4820
Reviewed-by: Adam Langley <agl@google.com>
This isn't exhaustive. There are still failures in some tests which probably
ought to get C++'d first.
Change-Id: Iac58df9d98cdfd94603d54374a531b2559df64c3
Reviewed-on: https://boringssl-review.googlesource.com/4795
Reviewed-by: Adam Langley <agl@google.com>
Currently far from passing and I haven't even tried with a leak checker yet.
Also bn_test is slow.
Change-Id: I4fe2783aa5f7897839ca846062ae7e4a367d2469
Reviewed-on: https://boringssl-review.googlesource.com/4794
Reviewed-by: Adam Langley <agl@google.com>
This is considerably less scary than swapping out connection state. It also
fixes a minor bug where, if dtls1_do_write had an alert to dispatch and we
happened to retry during a rexmit, it would use the wrong epoch.
BUG=468889
Change-Id: I754b0d46bfd02f797f4c3f7cfde28d3e5f30c52b
Reviewed-on: https://boringssl-review.googlesource.com/4793
Reviewed-by: Adam Langley <agl@google.com>
tls1_enc is now SSL_AEAD_CTX_{open,seal}. This starts tidying up a bit
of the record-layer logic. This removes rr->input, as encrypting and
decrypting records no longer refers to various globals. It also removes
wrec altogether. SSL3_RECORD is now only used to maintain state about
the current incoming record. Outgoing records go straight to the write
buffer.
This also removes the outgoing alignment memcpy and simply calls
SSL_AEAD_CTX_seal with the parameters as appropriate. From bssl speed
tests, this seems to be faster on non-ARM and a bit of a wash on ARM.
Later it may be worth recasting these open/seal functions to write into
a CBB (tweaked so it can be malloc-averse), but for now they take an
out/out_len/max_out trio like their EVP_AEAD counterparts.
BUG=468889
Change-Id: Ie9266a818cc053f695d35ef611fd74c5d4def6c3
Reviewed-on: https://boringssl-review.googlesource.com/4792
Reviewed-by: Adam Langley <agl@google.com>
(This makes it possible to include opensslv.h when not linking SSL.)
Change-Id: Id88c5ff44a7099d33d8d4672f7ba88986ffd1526
Reviewed-on: https://boringssl-review.googlesource.com/4831
Reviewed-by: Adam Langley <agl@google.com>
SSL3_VERSION_MAJOR is the only MAJOR/MINOR number used internally or
externally.
Change-Id: I3f17175e73fd89887665accf1bfa680581f42dfe
Reviewed-on: https://boringssl-review.googlesource.com/4790
Reviewed-by: Adam Langley <agl@google.com>
Chromium's session cache has since been rewritten and no longer needs to
muck with those functions in tests.
Change-Id: I2defad81513210dca5e105757e04cbb677583251
Reviewed-on: https://boringssl-review.googlesource.com/4788
Reviewed-by: Adam Langley <agl@google.com>
One of these days we may need to get a more aggressive C parser...
Change-Id: I7c6a848fb3b7f41083ac70542aa17e971baf10a4
Reviewed-on: https://boringssl-review.googlesource.com/4786
Reviewed-by: Adam Langley <agl@google.com>
Fleshes out the table of contents more.
Change-Id: I8f8f0e43bdf7419f978b4fc66de80922ed1ae425
Reviewed-on: https://boringssl-review.googlesource.com/4785
Reviewed-by: Adam Langley <agl@google.com>
Current thought is to organize this by:
- Core SSL_CTX APIs (creating, destroying)
- Core SSL APIs (creating destroying, maybe handshake, read, write as
well)
- APIs to configure SSL_CTX/SSL, roughly grouped by feature. Probably
options and modes are the first two sections. SSL_TXT_* constants can
be part of documenting cipher suite configuration.
- APIs to query state from SSL_CTX/SSL, roughly grouped by feature. (Or
perhaps these should be folded into the configuration sections?)
The functions themselves aren't reordered or reorganized to match the
eventual header order yet. Though I did do the s -> ssl rename on the
ones I've touched.
Also formally deprecate SSL_clear. It would be a core SSL API
except it's horrible.
Change-Id: Ia7e4fdcb7bad4e9ccdee8cf8c3136dc63aaaa772
Reviewed-on: https://boringssl-review.googlesource.com/4784
Reviewed-by: Adam Langley <agl@google.com>
The type names are perfectly serviceable. Most of them are
forward-declared in base.h.
Change-Id: Id03f5039a2d1bab82c68ade074a0e26cd3ab5ad9
Reviewed-on: https://boringssl-review.googlesource.com/4783
Reviewed-by: Adam Langley <agl@google.com>
Rather than duplicate all the various modifiers, which is quite
error-prone, write all the tests to a temporary array and then apply
modifiers afterwards.
Change-Id: I19bfeb83b722ed34e973f17906c5e071471a926a
Reviewed-on: https://boringssl-review.googlesource.com/4782
Reviewed-by: Adam Langley <agl@google.com>
We should be testing asynchronous renego.
BUG=429450
Change-Id: Ib7a5d42f2ac728f9ea0d80158eef63ad77cd77a4
Reviewed-on: https://boringssl-review.googlesource.com/4781
Reviewed-by: Adam Langley <agl@google.com>
Since we hope to eventually lose server-side renegotiation support
altogether, get the client-side version of those tests. We should have
had those anyway to test that the default is to allow it.
BUG=429450
Change-Id: I4a18f339b55f3f07d77e22e823141e10a12bc9ff
Reviewed-on: https://boringssl-review.googlesource.com/4780
Reviewed-by: Adam Langley <agl@google.com>
Looks like it was the use in type_check.h that was still causing
problems, not that MSVC doesn't short-circuit #if statements.
Change-Id: I574e8dd463c46b0133a989b221a7bb8861b3eed9
6e1f6456 tried to do this, but MSVC doesn't short-circuit #if
statements. So this change tries having the test be in a different #if.
Change-Id: Id0074770c166a2b7cd9ba2c8cd06245a68b77af8
While the compiler on OS X sets the macros as if it supports C11
atomics, stdatomic.h is actually missing.
Change-Id: Ifecaf1c8df6390e6b994663adedc284d9b8130b7
At this point, none of these functions or macros are used so they can
just be deleted.
Change-Id: I8ed1aae7a252e886864bf43e3096eff2228183cd
Reviewed-on: https://boringssl-review.googlesource.com/4777
Reviewed-by: Adam Langley <agl@google.com>
These ASN.1 macros are the last references to the old-style OpenSSL
locks that remain. The ASN.1 reference count handling was changed in a
previous commit to use |CRYPTO_refcount_*| so these lock references were
unused anyway.
Change-Id: I1b27eef140723050a8e6878a1bea11da3409d0eb
Reviewed-on: https://boringssl-review.googlesource.com/4776
Reviewed-by: Adam Langley <agl@google.com>
|SSL_CTX| and |X509_STORE| have grown their own locks. Several static
locks have been added to hack around not being able to use a
|CRYPTO_once_t| in public headers. Lastly, support for calling
|SSL_CTX_set_generate_session_id| concurrently with active connections
has been removed. No other property of an |SSL_CTX| works like that.
Change-Id: Iff5fe3ee3fdd6ea9c9daee96f850b107ad8a6bca
Reviewed-on: https://boringssl-review.googlesource.com/4775
Reviewed-by: Adam Langley <agl@google.com>
It's no longer needed after the conversion to |CRYPTO_refcount_t|.
Change-Id: Ied129c4c247fcd426745fa016350528b7571aaaa
Reviewed-on: https://boringssl-review.googlesource.com/4774
Reviewed-by: Adam Langley <agl@google.com>
Convert reference counts in ssl/ to use |CRYPTO_refcount_t|.
Change-Id: I5d60f641b0c89b1ddfe38bfbd9d7285c60377f4c
Reviewed-on: https://boringssl-review.googlesource.com/4773
Reviewed-by: Adam Langley <agl@google.com>
This change converts the reference counts in crypto/ to use
|CRYPTO_refcount_t|. The reference counts in |X509_PKEY| and |X509_INFO|
were never actually used and so were dropped.
Change-Id: I75d572cdac1f8c1083c482e29c9519282d7fd16c
Reviewed-on: https://boringssl-review.googlesource.com/4772
Reviewed-by: Adam Langley <agl@google.com>
OpenSSL has traditionally done reference counting with |int|s and the
|CRYPTO_add| function. Unless a special callback is installed (rare),
this is implemented by doing the reference count operations under a
lock.
This change adds infrastructure for handling reference counts and uses
atomic operations when C11 support is available.
Change-Id: Ia023ce432319efd00f77a7340da27d16ee4b63c3
Reviewed-on: https://boringssl-review.googlesource.com/4771
Reviewed-by: Adam Langley <agl@google.com>
OPENSSL_COMPILE_ASSERT implements a static assertion, but the error
message is a little weird because it's a hack around the fact that C,
traditionally, doesn't have static assertions.
C11 now does have _Static_assert (a.k.a. static_assert when one includes
assert.h) so we can use that when provided to get cleaner error
messages.
Change-Id: Ia3625dfb2988de11fd95ddba957f118c0d3183ff
Reviewed-on: https://boringssl-review.googlesource.com/4770
Reviewed-by: Adam Langley <agl@google.com>
DH groups less than 1024 bits are clearly not very safe. Ideally servers
would switch to ECDHE because 1024 isn't great either, but this will
serve for the short term.
BUG=490240
Change-Id: Ic9aac714cdcdcbfae319b5eb1410675d3b903a69
Reviewed-on: https://boringssl-review.googlesource.com/4813
Reviewed-by: David Benjamin <davidben@chromium.org>
Reviewed-by: Adam Langley <agl@google.com>
Change-Id: I1bade6ad22596857b208b2c8ad54b239b648b86e
Reviewed-on: https://boringssl-review.googlesource.com/4812
Reviewed-by: David Benjamin <davidben@chromium.org>
Reviewed-by: Adam Langley <agl@google.com>
SSLeay is a compatibility function for OpenSSL, but I got it wrong. It
doesn't return a string, it returns a number. This doesn't end up making
any difference, but it fixes a warning when building OpenSSH.
Change-Id: I327ab4f70313c93c18f81d8804ba4acdc3bc1a4a
Reviewed-on: https://boringssl-review.googlesource.com/4811
Reviewed-by: David Benjamin <davidben@chromium.org>
Reviewed-by: Adam Langley <agl@google.com>
This change import's upstream's beeb0fa7 and fixes a UAF in X509.
Thankfully, this shouldn't impact Chromium, which doesn't use OpenSSL
for certificate verification.
BUG=489764
Change-Id: I0ce2ec05083f7c588ba5504bb12151437dec593e
Reviewed-on: https://boringssl-review.googlesource.com/4810
Reviewed-by: David Benjamin <davidben@chromium.org>
Reviewed-by: Adam Langley <agl@google.com>
