The handshake state machine is still rather messy (we should switch to CBB,
split the key exchanges apart, and also pull reading and writing out), but this
version makes it more obvious to the compiler that |p| and |sig_len| are
initialized. The old logic created a synchronous-only state which, if enterred
directly, resulted in some variables being uninitialized.
Change-Id: Ia3ac9397d523fe299c50a95dc82a9b26304cea96
Reviewed-on: https://boringssl-review.googlesource.com/5765
Reviewed-by: Adam Langley <agl@google.com>
Our tests shouldn't panic if the program misbehaves.
Change-Id: I113e050222bcf48e5f25883f860dbc1c5c77e77e
Reviewed-on: https://boringssl-review.googlesource.com/5764
Reviewed-by: Adam Langley <agl@google.com>
Some compilers complain and it's worth checking. Maybe the file changed in size
between ftell and fread.
Change-Id: I7898b8517556ec6899bd6e8866ba3d1cd7efd5f4
Reviewed-on: https://boringssl-review.googlesource.com/5763
Reviewed-by: Adam Langley <agl@google.com>
Move cert_chain to the SSL_SESSION. Now everything on an SSL_SESSION is
properly serialized. The cert_chain field is, unfortunately, messed up
since it means different things between client and server.
There exists code which calls SSL_get_peer_cert_chain as both client and
server and assumes the existing semantics for each. Since that function
doesn't return a newly-allocated STACK_OF(X509), normalizing between the
two formats is a nuisance (we'd either need to store both cert_chain and
cert_chain_full on the SSL_SESSION or create one of the two variants
on-demand and stash it into the SSL).
This CL does not resolve this and retains the client/server difference
in SSL_SESSION. The SSL_SESSION serialization is a little inefficient
(two copies of the leaf certificate) for a client, but clients don't
typically serialize sessions. Should we wish to resolve it in the
future, we can use a different tag number. Because this was historically
unserialized, existing code must already allow for cert_chain not being
preserved across i2d/d2i.
In keeping with the semantics of retain_only_sha256_of_client_certs,
cert_chain is not retained when that flag is set.
Change-Id: Ieb72fc62c3076dd59750219e550902f1ad039651
Reviewed-on: https://boringssl-review.googlesource.com/5759
Reviewed-by: Adam Langley <agl@google.com>
The difference of two pointers is signed, even though it's always
non-negative here, so MSVC is complaining about signedness mismatch.
Change-Id: I5a042d06ed348540706b93310af3f60f3ab5f303
Reviewed-on: https://boringssl-review.googlesource.com/5766
Reviewed-by: Adam Langley <agl@google.com>
Rather than parse the fields in two passes, group the code relating to
one field together. Somewhat less annoying to add new fields. To keep
this from getting too unwieldy, add a few more helper functions for the
common field types.
Change-Id: Ia86c6bbca9dd212d5c35029363ea4d6b6426164a
Reviewed-on: https://boringssl-review.googlesource.com/5758
Reviewed-by: Adam Langley <agl@google.com>
It's completely redundant with the copy in the SSL_SESSION except it
isn't serialized.
Change-Id: I1d95a14cae064c599e4bab576df1dd156da4b81c
Reviewed-on: https://boringssl-review.googlesource.com/5757
Reviewed-by: Adam Langley <agl@google.com>
Gets another field out of the SSL_SESSION.
Change-Id: I9a27255533f8e43e152808427466ec1306cfcc60
Reviewed-on: https://boringssl-review.googlesource.com/5756
Reviewed-by: Adam Langley <agl@google.com>
This is analogous to openssl s_client's -sess_in and -sess_out. Use PEM to
align with OpenSSL. This is useful for debugging session resumption and also
generating things to test serialization against.
Change-Id: Idc58e8fa3dd4c2385f6a2d647e66ef11427be60d
Reviewed-on: https://boringssl-review.googlesource.com/5761
Reviewed-by: Adam Langley <agl@google.com>
It's supposed to be void*. The only reason this was working was that it was
only called in C which happily casts from void* to T*. (But if called in C++ in
a macro, it breaks.)
Change-Id: I7f765c3572b9b4815ae58da852be1e742de1bd96
Reviewed-on: https://boringssl-review.googlesource.com/5760
Reviewed-by: Adam Langley <agl@google.com>
The old empty record logic discarded the records at a very low-level.
Let the error bubble up to ssl3_read_bytes so the type mismatch logic
may kick in before the empty record is skipped.
Add tests for when the record in question is application data, before
before the handshake and post ChangeCipherSpec.
BUG=521840
Change-Id: I47dff389cda65d6672b9be39d7d89490331063fa
Reviewed-on: https://boringssl-review.googlesource.com/5754
Reviewed-by: Adam Langley <agl@google.com>
This begins decoupling the transport from the SSL state machine. The buffering
logic is hidden behind an opaque API. Fields like ssl->packet and
ssl->packet_length are gone.
ssl3_get_record and dtls1_get_record now call low-level tls_open_record and
dtls_open_record functions that unpack a single record independent of who owns
the buffer. Both may be called in-place. This removes ssl->rstate which was
redundant with the buffer length.
Future work will push the buffer up the stack until it is above the handshake.
Then we can expose SSL_open and SSL_seal APIs which act like *_open_record but
return a slightly larger enum due to other events being possible. Likewise the
handshake state machine will be detached from its buffer. The existing
SSL_read, SSL_write, etc., APIs will be implemented on top of SSL_open, etc.,
combined with ssl_read_buffer_* and ssl_write_buffer_*. (Which is why
ssl_read_buffer_extend still tries to abstract between TLS's and DTLS's fairly
different needs.)
The new buffering logic does not support read-ahead (removed previously) since
it lacks a memmove on ssl_read_buffer_discard for TLS, but this could be added
if desired. The old buffering logic wasn't quite right anyway; it tried to
avoid the memmove in some cases and could get stuck too far into the buffer and
not accept records. (The only time the memmove is optional is in DTLS or if
enough of the record header is available to know that the entire next record
would fit in the buffer.)
The new logic also now actually decrypts the ciphertext in-place again, rather
than almost in-place when there's an explicit nonce/IV. (That accidentally
switched in https://boringssl-review.googlesource.com/#/c/4792/; see
3d59e04bce96474099ba76786a2337e99ae14505.)
BUG=468889
Change-Id: I403c1626253c46897f47c7ae93aeab1064b767b2
Reviewed-on: https://boringssl-review.googlesource.com/5715
Reviewed-by: Adam Langley <agl@google.com>
This consists mostly of re-adding OpenSSL's implementation of PBKDF2
(very loosely based upon e0d26bb3). The meat of it, namely
|PKCS5_PBKDF2_HMAC|, was already present, but unused.
In addition, |PKCS8_encrypt| and |PKCS8_decrypt| must be changed to
not perform UCS-2 conversion in the PBES2 case.
Change-Id: Id170ecabc43c79491600051147d1d6d3c7273dbc
Reviewed-on: https://boringssl-review.googlesource.com/5745
Reviewed-by: Adam Langley <agl@google.com>
arm_arch.h is included from ARM asm files, but lives in crypto/, not
openssl/include/. Since the asm files are often built from a different
location than their position in the source tree, relative include paths
are unlikely to work so, rather than having crypto/ be a de-facto,
second global include path, this change moves arm_arch.h to
include/openssl/.
It also removes entries from many include paths because they should be
needed as relative includes are always based on the locations of the
source file.
Change-Id: I638ff43d641ca043a4fc06c0d901b11c6ff73542
Reviewed-on: https://boringssl-review.googlesource.com/5746
Reviewed-by: Adam Langley <agl@google.com>
Match the other stack-allocated types in that we expose a wrapper function to
get them into the zero state. Makes it more amenable to templates like
ScopedOpenSSLContext.
Change-Id: Ibc7b2b1bc0421ce5ccc84760c78c0b143441ab0f
Reviewed-on: https://boringssl-review.googlesource.com/5753
Reviewed-by: Adam Langley <agl@google.com>
Apparently we left those as zero. Oh well. This only affects
SSL_CIPHER_get_bits, but so long as we have the field, it ought to be correct.
Change-Id: I2878ec22c2f5a6263f805e04d9fd8448994639b7
Reviewed-on: https://boringssl-review.googlesource.com/5752
Reviewed-by: Adam Langley <agl@google.com>
This made sense when the cipher might have been standardized as-is, so a
DHE_RSA variant could appease the IETF. Since the standardized variant is going
to have some nonce tweaks anyway, there's no sense in keeping this around. Get
rid of one non-standard cipher suite value early. (Even if they were to be
standardized as-is, it's not clear we should implement new DHE cipher suites at
this point.)
Chrome UMA, unsurprisingly, shows that it's unused.
Change-Id: Id83d73a4294b470ec2e94d5308fba135d6eeb228
Reviewed-on: https://boringssl-review.googlesource.com/5750
Reviewed-by: Adam Langley <agl@google.com>
The fact that |value_free| expects to free() value->section is
inconsistent with the behavior of |add_string|, which adds a reference
to an existing string.
Along the way, add a |CONF_VALUE_new| method to simplify things a bit.
Change-Id: I438abc80575394e4d8df62a4fe2ff1050e3ba039
Reviewed-on: https://boringssl-review.googlesource.com/5744
Reviewed-by: Adam Langley <agl@google.com>
As I read it:
1. |_LHASH| contains
2. buckets of |LHASH_ITEMS|, which contain
3. |CONF_VALUE|s, which contain
4. various bits of data.
The previous code was freeing #1 and #2 in |lh_free|, and #4 in
|value_free_contents|, but was failing to free the |CONF_VALUE|s
themselves. The fix is to call |value_free| rather than
|value_free_contents|.
Change-Id: I1d5b48692ca9ac04df688e45d7fc113dc5cd6ddf
Reviewed-on: https://boringssl-review.googlesource.com/5742
Reviewed-by: Adam Langley <agl@google.com>
This change makes |EVP_PKEY_asn1_find_str|, which is used by
|PEM_read_bio_PrivateKey|, recognize "DSA" as well as "EC" and "RSA".
Change-Id: I39cce12f600cec6a71df75312a41f8395429af62
Reviewed-on: https://boringssl-review.googlesource.com/5743
Reviewed-by: Adam Langley <agl@google.com>
MSAN appears to have a bug that causes this code to be miscompiled when
compiled with optimisations. In order to prevent that bug from holding
everything up, this change disables that code when MEMORY_SANITIZER is
defined. The generic elliptic-curve code can pick up the slack in that
case.
Change-Id: I7ce26969b3ee0bc0b0496506f06a8cf9b2523cfa
(I couldn't find an authoritative source of test data, including in
OpenSSL's source, so I used OpenSSL's implementation to produce the
test ciphertext.)
This benefits globalplatform.
Change-Id: Ifb79e77afb7efed1c329126a1a459bbf7ce6ca00
Reviewed-on: https://boringssl-review.googlesource.com/5725
Reviewed-by: Adam Langley <agl@google.com>
Note that while |DES_ede2_cbc_encrypt| exists, I didn't use it: I
think it's easier to see what's happening this way.
(I couldn't find an authoritative source of test data, including in
OpenSSL's source, so I used OpenSSL's implementation to produce the
test ciphertext.)
This benefits globalplatform.
Change-Id: I7e17ca0b69067d7b3f4bc213b4616eb269882ae0
Reviewed-on: https://boringssl-review.googlesource.com/5724
Reviewed-by: Adam Langley <agl@google.com>
|DES_ecb_encrypt| was already present.
This benefits globalplatform.
Change-Id: I2ab41eb1936b3026439b5981fb27e29a12672b66
Reviewed-on: https://boringssl-review.googlesource.com/5723
Reviewed-by: Adam Langley <agl@google.com>
This is a simpler implementation than OpenSSL's, lacking responder IDs
and request extensions support. This mirrors the client implementation
already present.
Change-Id: I54592b60e0a708bfb003d491c9250401403c9e69
Reviewed-on: https://boringssl-review.googlesource.com/5700
Reviewed-by: Adam Langley <agl@google.com>
This is harmless, but it wasn't annoted with |(void)| so Coverity
complained about it.
Change-Id: Ie3405b0c0545944d49973d4bf29f8aeb6b965211
Reviewed-on: https://boringssl-review.googlesource.com/5612
Reviewed-by: David Benjamin <davidben@chromium.org>
Reviewed-by: Adam Langley <agl@google.com>
Previously doc/doc.css was a symlink to util/doc.css, but symlinks
don't work well on Windows. Now util/doc.css is copied to the output
directory when the documentation is generated.
Change-Id: I2c9f4fee4f4307cc3dd70c4be380b4551d5e9ab5
Reviewed-on: https://boringssl-review.googlesource.com/5677
Reviewed-by: David Benjamin <davidben@chromium.org>
Reviewed-by: Adam Langley <agl@google.com>
They get initialized in SSL_new and SSL_CTX_new, respectively.
Change-Id: Ib484108987a99f654d1a77fc473103f5cb393bd7
Reviewed-on: https://boringssl-review.googlesource.com/5676
Reviewed-by: Adam Langley <agl@google.com>
They're not called (new in 1.0.2). We actually may well need to
configure these later to strike ECDSA from the list on Chrome/XP
depending on what TLS 1.3 does, but for now striking it from the cipher
suite list is both necessary and sufficient. I think we're better off
removing these for now and adding new APIs later if we need them.
(This API is weird. You pass in an array of NIDs that must be even
length and alternating between hash and signature NID. We'd also need a
way to query the configured set of sigalgs to filter away. Those used to
exist but were removed in
https://boringssl-review.googlesource.com/#/c/5347/. SSL_get_sigalgs is
an even uglier API and doesn't act on the SSL_CTX.)
And with that, SSL_ctrl and SSL_CTX_ctrl can *finally* be dropped. Don't
leave no-op wrappers; anything calling SSL_ctrl and SSL_CTX_ctrl should
instead switch to the wrapper macros.
BUG=404754
Change-Id: I5d465cd27eef30d108eeb6de075330c9ef5c05e8
Reviewed-on: https://boringssl-review.googlesource.com/5675
Reviewed-by: Adam Langley <agl@google.com>
I'm not sure why one would ever want to externally know the curve list
supported by the server. The API is new as of 1.0.2 and has no callers.
Configuring curves will be much more useful when Curve25519 exists and the API
isn't terribly crazy, so keep that API around and promote it to a real
function.
BUG=404754
Change-Id: Ibd5858791d3dfb30d53dd680cb75b0caddcbb7df
Reviewed-on: https://boringssl-review.googlesource.com/5674
Reviewed-by: Adam Langley <agl@google.com>
This change stores the size of the group/modulus (for RSA/DHE) or curve
ID (for ECDHE) in the |SSL_SESSION|. This makes it available for UIs
where desired.
Change-Id: I354141da432a08f71704c9683f298b361362483d
Reviewed-on: https://boringssl-review.googlesource.com/5280
Reviewed-by: Adam Langley <agl@google.com>
This guarantees that we never read beyond the first record, even if the
first record is empty. Between removing SSL_set_read_ahead and DTLS
enforcing record boundaries, this means the buffer need never memmove
data.
The memmove isn't really much of a burden and we can probably just put
SSL_set_read_ahead back after the cleanup if desired. But while the
non-existant read_ahead is off, we should avoid reading more than needed.
(Also the current memmove logic is completely wrong for TLS. Checking
align != 0 doesn't make sense. The real reason to memmove is that the
next record may still be full size. So now line 209 of s3_pkt.c should
*actually* be unreachable.)
SSL_R_HTTPS_PROXY_REQUEST detection is now slightly less accurate, but
OpenSSL was already not parsing HTTP completely. We could asynchronously
read the extra 3 bytes once the first 5 match, but that seems
unnecessary. (Shall we just get rid of all these HTTP detectors? The
only consumer of those error codes is some diagnostics logic.)
BUG=468889
Change-Id: Ie3bf148ae7274795e1d048d78282d1d8063278ea
Reviewed-on: https://boringssl-review.googlesource.com/5714
Reviewed-by: Adam Langley <agl@google.com>
I'm not sure why I made a separate one. (Not quite how the V2ClientHello
code will look in the buffer-free API yet. Probably the future
refactored SSL_HANDSHAKE gadget will need separate entry points to
consume a handshake message or V2ClientHello and the driver deals with
framing.)
This also means that ssl3_setup_read_buffer is never called external to
ssl3_read_n.
BUG=468889
Change-Id: I872f1188270968bf53ee9d0488a761c772a11e9e
Reviewed-on: https://boringssl-review.googlesource.com/5713
Reviewed-by: Adam Langley <agl@google.com>
Rather than sniff for ClientHello, just fall through to standard logic
once weird cases are resolved.
This means that garbage will now read as WRONG_VERSION rather than
UNKNOWN_PROTOCOL, but the rules here were slightly odd anyway. This also
means we'll now accept empty records before the ClientHello (up to the
empty record limit), and process records of the wrong type with the
usual codepath during the handshake.
This shouldn't be any more risk as it just makes the ClientHello more
consistent with the rest of the protocol. A TLS implementation that
doesn't parse V2ClientHello would do the same unless it still
special-cased the first record. All newly-exposed states are reachable
by fragmenting ClientHello by one byte and then sending the record in
question.
BUG=468889
Change-Id: Ib701ae5d8adb663e158c391639b232a9d9cd1c6e
Reviewed-on: https://boringssl-review.googlesource.com/5712
Reviewed-by: Adam Langley <agl@google.com>
If get_issuer fails, some of these calls return rather than jumping to common
cleanup code.
Change-Id: Iacd59747fb11e9bfaae86f2eeed88798ee08203e
Reviewed-on: https://boringssl-review.googlesource.com/5711
Reviewed-by: Adam Langley <agl@google.com>
(Imported from upstream's 25efcb44ac88ab34f60047e16a96c9462fad39c1 and
56353962e7da7e385c3d577581ccc3015ed6d1dc.)
Change-Id: I2ff22fc9da23868de02e6f31c50a3f1d0c6dec1a
Reviewed-on: https://boringssl-review.googlesource.com/5710
Reviewed-by: Adam Langley <agl@google.com>
The function BN_MONT_CTX_set was assuming that the modulus was non-zero
and therefore that |mod->top| > 0. In an error situation that may not be
the case and could cause a seg fault.
This is a follow on from CVE-2015-1794.
(Imported from upstream's 512368c9ed4d53fb230000e83071eb81bf628b22.)
The CVE itself doesn't affect us as the bit strength check in the DHE logic
excludes zero.
Also add tests to bn_test for a couple of division by zero cases. (This and
BN_div.)
Change-Id: Ibd8ef98d6be48eb95110021c23cd8e278656764d
Reviewed-on: https://boringssl-review.googlesource.com/5690
Reviewed-by: Adam Langley <agl@google.com>
BN_bin2bn takes a size_t as it should, but it passes that into bn_wexpand which
takes unsigned. Switch bn_wexpand and bn_expand to take size_t before they
check bounds against INT_MAX.
BIGNUM itself still uses int everywhere and we may want to audit all the
arithmetic at some point. Although I suspect having bn_expand require that the
number of bits fit in an int is sufficient to make everything happy, unless
we're doing interesting arithmetic on the number of bits somewhere.
Change-Id: Id191a4a095adb7c938cde6f5a28bee56644720c6
Reviewed-on: https://boringssl-review.googlesource.com/5680
Reviewed-by: Adam Langley <agl@google.com>
Move the bn_expand call inside decode_hex; it's an implementation detail of
hex-decoding. decode_dec instead works with BN_mul_word and BN_add_word so it
can just rely on BN internally expanding things and check the return value.
Also clean up the decode_hex loop so it's somewhat more readable and check for
INT_MAX in bn_x2bn. It uses int over size_t rather pervasively, but while I'm
here at least make that function check overflow.
BUG=517474
Change-Id: I4f043973ee43071a02ea5d4313a8fdaf12404e84
Reviewed-on: https://boringssl-review.googlesource.com/5679
Reviewed-by: Adam Langley <agl@google.com>
This isn't called and, with the fixed-DH client cert types removed, is
only useful if a server wishes to not accept ECDSA certificates or
something.
BUG=404754
Change-Id: I21d8e1a71aedf446ce974fbeadc62f311ae086db
Reviewed-on: https://boringssl-review.googlesource.com/5673
Reviewed-by: Adam Langley <agl@google.com>
