Otherwise it still thinks this is an RFC 5114 prime and kicks in the (now
incorrect) validity check.
Change-Id: Ie78514211927f1f2d2549958621cb7896f68b5ce
Reviewed-on: https://boringssl-review.googlesource.com/7050
Reviewed-by: Adam Langley <agl@google.com>
This slightly simplifies the SSL_ECDH code and will be useful later on
in reimplementing the key parsing logic.
Change-Id: Ie41ea5fd3a9a734b3879b715fbf57bd991e23799
Reviewed-on: https://boringssl-review.googlesource.com/6858
Reviewed-by: Adam Langley <agl@google.com>
Change acb24518 renamed some functions, but there were some dangling
references in bn_test.c. Thanks to Brian Smith for noticing.
This change has no semantic effect.
Change-Id: Id149505090566583834be3abce2cee28b8c248e2
Reviewed-on: https://boringssl-review.googlesource.com/7040
Reviewed-by: David Benjamin <davidben@google.com>
This is CVE-2016-0701 for OpenSSL, reported by Antonio Sanso. It is a no-op for
us as we'd long removed SSL_OP_DH_SINGLE_USE and static DH cipher suites. (We
also do not parse or generate X9.42 DH parameters.)
However, we do still have the APIs which return RFC 5114 groups, so we should
perform the necessary checks in case later consumers reuse keys.
Unlike groups we generate, RFC 5114 groups do not use "safe primes" and have
many small subgroups. In those cases, the subprime q is available. Before using
a public key, ensure its order is q by checking y^q = 1 (mod p). (q is assumed
to be prime and the existing range checks ensure y is not 1.)
(Imported from upstream's 878e2c5b13010329c203f309ed0c8f2113f85648 and
75374adf8a6ff69d6718952121875a491ed2cd29, but with some bugs fixed. See
RT4278.)
Change-Id: Ib18c3e84819002fa36a127ac12ca00ee33ea018a
Reviewed-on: https://boringssl-review.googlesource.com/7001
Reviewed-by: Adam Langley <agl@google.com>
Add guards for the architecture and OPENSSL_NO_ASM to
the assembly-language files in crypto/curve25519/asm.
The Dart compilation of BoringSSL includes all files,
because the architecture is not known when gyp is run.
Change-Id: I66f5ae525266b63b0fe3a929012b771d545779b5
Reviewed-on: https://boringssl-review.googlesource.com/7030
Reviewed-by: Adam Langley <agl@google.com>
Take the mappings for MD5 and SHA-224 values out of the code altogether. This
aligns with the current TLS 1.3 draft.
For MD5, this is a no-op. It is not currently possible to configure accepted
signature algorithms, MD5 wasn't in the hardcoded list, and we already had a
test ensuring we enforced our preferences correctly. MD5 also wasn't in the
default list of hashes our keys could sign and no one overrides it with a
different hash.
For SHA-224, this is not quite a no-op. The hardcoded accepted signature
algorithms list included SHA-224, so this will break servers relying on that.
However, Chrome's metrics have zero data points of servers picking SHA-224 and
no other major browser includes it. Thus that should be safe.
SHA-224 was also in the default list of hashes we are willing to sign. For
client certificates, Chromium's abstractions already did not allow signing
SHA-224, so this is a no-op there. For servers, this will break any clients
which only accept SHA-224. But no major browsers do this and I am not aware of
any client implementation which does such ridiculous thing.
(SHA-1's still in there. Getting rid of that one is going to take more effort.)
Change-Id: I6a765fdeea9e19348e409d58a0eac770b318e599
Reviewed-on: https://boringssl-review.googlesource.com/7020
Reviewed-by: Adam Langley <agl@google.com>
The |inline| must appear before the type.
Change-Id: Iecebbcc50024a846d7804228a858acfc33d68efd
Reviewed-on: https://boringssl-review.googlesource.com/7010
Reviewed-by: David Benjamin <davidben@google.com>
Finally, we can stick ScopedFOO in containers.
Change-Id: I3ed166575822af9f182e8be8f4db723e1f08ea31
Reviewed-on: https://boringssl-review.googlesource.com/6553
Reviewed-by: Adam Langley <alangley@gmail.com>
Although RFC 3279 allows both, per RFC 5912, keys must use a named curve
rather than spelling out the curve parameters. Although we do not allow
arbitrary curves, we do have to (pretty hackishly) recognize built-in
curves in ECPrivateKeys.
It seems the cause of this was that OpenSSL, unless you set asn1_flag on
the EC_GROUP, likes to encode keys by spelling out the parameters. This
is in violation of RFC 5915, though probably not in violation of one of
the other redundant ECC specifications. For more fun, it appears
asn1_flag defaults to *off* in the API and *on* in the command-line
tools.
I think the original cause was these defaults meant the pre-BoringSSL
Android/OpenSSL Chromium port wrote out Channel ID keys in this format.
By now this should no longer by an issue, but it'll warrant a bit more
investigation to be sure we can drop it.
For now, keep this logic out of SPKIs by not calling d2i_ECParameters.
d2i_ECParameters is a fairly pointless function when only named curves
are allowed. In testing other implementations, none of Firefox, Safari,
or IE11/Win will parse such certificates (i.e. the error is fatal and
unbypassable). Likewise, because Mac and Windows' underlying libraries
reject this, Chrome on Mac and Windows already rejects such things. Thus
this change should be compatible.
The following is the certificate and key I constructed to test with:
-----BEGIN CERTIFICATE-----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-----END CERTIFICATE-----
-----BEGIN EC PARAMETERS-----
MIH3AgEBMCwGByqGSM49AQECIQD/////AAAAAQAAAAAAAAAAAAAAAP//////////
/////zBbBCD/////AAAAAQAAAAAAAAAAAAAAAP///////////////AQgWsY12Ko6
k+ez671VdpiGvGUdBrDMU7D2O848PifSYEsDFQDEnTYIhucEk2pmeOETnSa3gZ9+
kARBBGsX0fLhLEJH+Lzm5WOkQPJ3A32BLeszoPShOUXYmMKWT+NC4v4af5uO5+tK
fA+eFivOM1drMV7Oy7ZAaDe/UfUCIQD/////AAAAAP//////////vOb6racXnoTz
ucrC/GMlUQIBAQ==
-----END EC PARAMETERS-----
-----BEGIN EC PRIVATE KEY-----
MHcCAQEEIAcPCHJ61KBKnN1ZyU2JaHcItW/JXTB3DujRyc4Ki7RqoAoGCCqGSM49
AwEHoUQDQgAE5itp4r9ln5e+Lx4NlIpM1Zdrt6keDUb73ampHp3culoB59aXqAoY
+cPEox5W4nyDSNsWGhz1HX7xlC1Lz3IiwQ==
-----END EC PRIVATE KEY-----
BUG=522228
Change-Id: I3723411a633dc07c4640027de07500293f8f7913
Reviewed-on: https://boringssl-review.googlesource.com/6853
Reviewed-by: Adam Langley <alangley@gmail.com>
OpenSSL accepts both OID 2.5.8.1.1 and OID 1.2.840.113549.1.1.1 for RSA
public keys. The latter comes from RFC 3279 and is widely implemented.
The former comes from the ITU-T version of X.509. Interestingly,
2.5.8.1.1 actually has a parameter, which OpenSSL ignores:
rsa ALGORITHM ::= {
KeySize
IDENTIFIED BY id-ea-rsa
}
KeySize ::= INTEGER
Remove support for 2.5.8.1.1 completely. In tests with a self-signed
certificate and code inspection:
- IE11 on Win8 does not accept the certificate in a TLS handshake at
all. Such a certificate is fatal and unbypassable. However Microsoft's
libraries do seem to parse it, so Chrome on Windows allows one to
click through the error. I'm guessing either the X.509 stack accepts
it while the TLS stack doesn't recognize it as RSA or the X.509 stack
is able to lightly parse it but not actually understand the key. (The
system certificate UI didn't display it as an RSA key, so probably the
latter?)
- Apple's certificate library on 10.11.2 does not parse the certificate
at all. Both Safari and Chrome on Mac treat it as a fatal and
unbypassable error.
- mozilla::pkix, from code inspection, does not accept such
certificates. However, Firefox does allow clicking through the error.
This is likely a consequence of mozilla::pkix and NSS having different
ASN.1 stacks. I did not test this, but I expect this means Chrome on
Linux also accepts it.
Given IE and Safari's results, it should be safe to simply remove this.
Firefox's data point is weak (perhaps someone is relying on being able
to click-through a self-signed 2.5.8.1.1 certificate), but it does
further ensure no valid certificate could be doing this.
The following is the 2.5.8.1.1 certificate I constructed to test with.
The private key is key.pem from ssl/test/runner:
-----BEGIN CERTIFICATE-----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-----END CERTIFICATE-----
BUG=522228
Change-Id: I031d03c0f53a16cbc749c4a5d8be6efca50dc863
Reviewed-on: https://boringssl-review.googlesource.com/6852
Reviewed-by: Adam Langley <alangley@gmail.com>
Normally this would be pretty scary:
if (...) {
} if (...) {
}
But it's an early return anyway.
Change-Id: I0a8965b5e294d3aaa803be47f4006ea0311c431d
Reviewed-on: https://boringssl-review.googlesource.com/6851
Reviewed-by: Adam Langley <alangley@gmail.com>
It takes ownership of the buffer, so it's not actually const. The
const-ness gets dropped once it transits through EVP_PKEY_CTX_ctrl.
Also compare against INT_MAX explicitly for the overflow check. I'm not sure
whether the casting version is undefined, but comparing against INT_MAX matches
the rest of the codebase when transiting in and out of signed ints.
Change-Id: I131165a4b5f0ebe02c6db3e7e3e0d1af5b771710
Reviewed-on: https://boringssl-review.googlesource.com/6850
Reviewed-by: Adam Langley <alangley@gmail.com>
It's never used. It's not clear why one would want such a thing.
EVP_PKEY_CTX has no way for callers to register callbacks, which means
there shouldn't be a way for the library to present you an EVP_PKEY_CTX
out-of-context. (Whereas app_data/ex_data makes sense on SSL because of
its numerous callbacks or RSA because of RSA_METHOD.)
Change-Id: I55af537ab101682677af34f6ac1f2c27b5899a89
Reviewed-on: https://boringssl-review.googlesource.com/6849
Reviewed-by: Adam Langley <alangley@gmail.com>
This removes the last caller of EVP_PKEY_copy_parameters within the
library.
Change-Id: I6af138d364973b18f52baf55c36c50a24a56bd44
Reviewed-on: https://boringssl-review.googlesource.com/6848
Reviewed-by: Adam Langley <alangley@gmail.com>
foo_init hooks are never implemented. Even upstream never uses them. The
flags member is also never used. We also don't expose paramgen, so
remove it.
Change-Id: I51d9439316c5163520ab7168693c457f33e59417
Reviewed-on: https://boringssl-review.googlesource.com/6846
Reviewed-by: Adam Langley <alangley@gmail.com>
These are never called. Group parameters are not secret anyway. This is
a remnant of upstream's EC_GROUP_clear_free.
Change-Id: I23a4076eae8e4561abddbe74d0ba72641532f229
Reviewed-on: https://boringssl-review.googlesource.com/6823
Reviewed-by: Adam Langley <alangley@gmail.com>
There's many ways to serialize a BIGNUM, so not including asn1 in the name is
confusing (and collides with BN_bn2cbb_padded). Since BN_asn12bn looks
ridiculous, match the parse/marshal naming scheme of other modules instead.
Change-Id: I53d22ae0537a98e223ed943e943c48cb0743cf51
Reviewed-on: https://boringssl-review.googlesource.com/6822
Reviewed-by: Adam Langley <alangley@gmail.com>
The code was using `#define INLINE` instead, but we have `inline` so
use it.
Change-Id: Id05eaec4720061c5d9a7278e20127c2bebcb2495
Reviewed-on: https://boringssl-review.googlesource.com/6976
Reviewed-by: David Benjamin <davidben@google.com>
Commit 75a64c08fc missed one case where
the GCC syntax should have been replaced with |alignas|.
Change-Id: Iebdaa9c9a2c0aff171f0b5d4daac607e351a4b7e
Reviewed-on: https://boringssl-review.googlesource.com/6974
Reviewed-by: David Benjamin <davidben@google.com>
The uses of |memcpy| to cast pointer-to-function to pointer-to-data and
back again did not have well-defined semantics. Use a union instead to
avoid the need for such a conversion get well-defined semantics.
Change-Id: I8ee54a83ba75440f7bc78c194eb55e2cf09b05d8
Reviewed-on: https://boringssl-review.googlesource.com/6972
Reviewed-by: David Benjamin <davidben@google.com>
Casting a pointer-to-non-volatile to pointer-to-volatile can be a no-op
as the compiler only requires volatile semantics when the pointed-to
object is a volatile object and there are no pointers-to-non-volatile
involved. This probably doesn't matter unless building with the MSVC
-volatile:iso flag, and maybe not even then, but it is good practice
anyway.
Change-Id: I94900d3dc61de3b8ce2ddecab2811907a9a7adbf
Reviewed-on: https://boringssl-review.googlesource.com/6973
Reviewed-by: David Benjamin <davidben@google.com>
Division isn't constant-time on Intel chips so the code was adding a
large multiple of md_size to try and force the operation to always take
the maximum amount of time.
I'm less convinced, these days, that compilers aren't going to get smart
enough to optimise that away so use Barrett reduction instead.
Change-Id: Ib8c514192682a2fcb4b1fb7e7c6dd1301d9888d0
Reviewed-on: https://boringssl-review.googlesource.com/6906
Reviewed-by: David Benjamin <davidben@chromium.org>
Reviewed-by: Adam Langley <alangley@gmail.com>
OpenSSL 1.1.0 doesn't seem to have these two, so this isn't based on anything.
Have them return uint64_t in preparation for switching the internal
representation to uint64_t so ssl_record_sequence_update can go away.
Change-Id: I21d55e9a29861c992f409ed293e0930a7aaef7a3
Reviewed-on: https://boringssl-review.googlesource.com/6941
Reviewed-by: Adam Langley <alangley@gmail.com>
We have the hook on the SSL_CTX, but it should be possible to set it without
reaching into SSL_CTX.
Change-Id: I93db070c7c944be374543442a8de3ce655a28928
Reviewed-on: https://boringssl-review.googlesource.com/6880
Reviewed-by: Adam Langley <alangley@gmail.com>
They should use the same P-256 check.
Change-Id: I66dd63663e638cba35b8f70f9cf119c718af4aec
Reviewed-on: https://boringssl-review.googlesource.com/6845
Reviewed-by: Adam Langley <alangley@gmail.com>
Move it into ssl->s3 so it automatically behaves correctly on SSL_clear.
ssl->version is still a mess though.
Change-Id: I17a692a04a845886ec4f8de229fa6cf99fa7e24a
Reviewed-on: https://boringssl-review.googlesource.com/6844
Reviewed-by: Adam Langley <alangley@gmail.com>
For TLS, this machinery only exists to swallow no_certificate alerts
which only get sent in an SSL 3.0 codepath anyway. It's much less a
no-op for SSL 3.0 which, strictly speaking, has only a subset of TLS's
alerts.
This gets messy around version negotiation because of the complex
relationship between enc_method, have_version, and version which all get
set at different times. Given that SSL 3.0 is nearly dead and all these
alerts are fatal to the connection anyway, this doesn't seem worth
carrying around. (It doesn't work very well anyway. An SSLv3-only server
may still send a record_overflow alert before version negotiation.)
This removes the last place enc_method is accessed prior to version
negotiation.
Change-Id: I79a704259fca69e4df76bd5a6846c9373f46f5a9
Reviewed-on: https://boringssl-review.googlesource.com/6843
Reviewed-by: Adam Langley <alangley@gmail.com>
This removes the various non-PRF checks from SSL3_ENC_METHOD so that can
have a clearer purpose. It also makes TLS 1.0 through 1.2's
SSL3_ENC_METHOD tables identical and gives us an assert to ensure
nothing accesses the version bits before version negotiation.
Accordingly, ssl_needs_record_splitting was reordered slightly so we
don't rely on enc_method being initialized to TLS 1.2
pre-version-negotiation.
This leaves alert_value as the only part of SSL3_ENC_METHOD which may be
accessed before version negotiation.
Change-Id: If9e299e2ef5511b5fa442b2af654eed054c3e675
Reviewed-on: https://boringssl-review.googlesource.com/6842
Reviewed-by: Adam Langley <alangley@gmail.com>
node.js is, effectively, another bindings library. However, it's better
written than most and, with these changes, only a couple of tiny fixes
are needed in node.js. Some of these changes are a little depressing
however so we'll need to push node.js to use APIs where possible.
Changes:
∙ Support verify_recover. This is very obscure and the motivation
appears to be https://github.com/nodejs/node/issues/477 – where it's
not clear that anyone understands what it means :(
∙ Add a few, no-op #defines
∙ Add some members to |SSL_CTX| and |SSL| – node.js needs to not
reach into these structs in the future.
∙ Add EC_get_builtin_curves.
∙ Add EVP_[CIPHER|MD]_do_all_sorted – these functions are limited to
decrepit.
Change-Id: I9a3566054260d6c4db9d430beb7c46cc970a9d46
Reviewed-on: https://boringssl-review.googlesource.com/6952
Reviewed-by: Adam Langley <agl@google.com>
In code, structs that happened to have a '(' somewhere in their body
would cause the parser to go wrong. This change fixes that and updates
the comments on a number of structs.
Change-Id: Ia76ead266615a3d5875b64a0857a0177fec2bd00
Reviewed-on: https://boringssl-review.googlesource.com/6970
Reviewed-by: Adam Langley <agl@google.com>
We haven't had problems with this, but make sure it stays that way.
Bogus signature algorithms are already covered.
Change-Id: I085350d89d79741dba3f30fc7c9f92de16bf242a
Reviewed-on: https://boringssl-review.googlesource.com/6910
Reviewed-by: David Benjamin <davidben@google.com>
Conscrypt needs to, in the certificate verification callback, know the key
exchange + auth method of the current cipher suite to pass into
X509TrustManager.checkServerTrusted. Currently it reaches into the struct to
get it. Add an API for this.
Change-Id: Ib4e0a1fbf1d9ea24e0114f760b7524e1f7bafe33
Reviewed-on: https://boringssl-review.googlesource.com/6881
Reviewed-by: David Benjamin <davidben@google.com>
Apparently OpenSSL's API is made entirely of initialization functions.
Some external libraries like to initialize with OPENSSL_config instead.
Change-Id: I28efe97fc5eb21309f560c84112b80e947f8bb17
Reviewed-on: https://boringssl-review.googlesource.com/6981
Reviewed-by: Adam Langley <agl@google.com>
With these stubs, cURL should not need any BoringSSL #ifdefs at all,
except for their OCSP #ifdefs (which can switch to the more generally
useful OPENSSL_NO_OCSP) and the workaround for wincrypt.h macro
collisions. That we intentionally leave to the consumer rather than add
a partial hack that makes the build sensitive to include order.
(I'll send them a patch upstream once this cycles in.)
Change-Id: I815fe67e51e80e9aafa9b91ae68867ca1ff1d623
Reviewed-on: https://boringssl-review.googlesource.com/6980
Reviewed-by: Adam Langley <agl@google.com>
Since the error string logic was rewritten, this hasn't done anything.
Change-Id: Icb73dca65e852bb3c7d04c260d591906ec72c15f
Reviewed-on: https://boringssl-review.googlesource.com/6961
Reviewed-by: Adam Langley <agl@google.com>
MSVC doesn't have stdalign.h and so doesn't support |alignas| in C
code. Define |alignas(x)| as a synonym for |__decltype(align(x))|
instead for it.
This also fixes -Wcast-qual warnings in rsaz_exp.c.
Change-Id: Ifce9031724cb93f5a4aa1f567e7af61b272df9d5
Reviewed-on: https://boringssl-review.googlesource.com/6924
Reviewed-by: Adam Langley <agl@google.com>
Reviewed-by: David Benjamin <davidben@google.com>
After its initial assignment, |e| is immediately reassigned another
value and so the initial assignment from |BN_CTX_get| is useless. If
that were not the case, then the |BN_free(e)| at the end of the
function would be very bad.
Change-Id: Id63a172073501c8ac157db9188a22f55ee36b205
Reviewed-on: https://boringssl-review.googlesource.com/6951
Reviewed-by: David Benjamin <davidben@google.com>
