It's a prime, so computing a constant-time mod inverse is straight-forward.
Change-Id: Ie09b84363c3d5da827989300a844c470437fd8f2
Reviewed-on: https://boringssl-review.googlesource.com/8308
Reviewed-by: Adam Langley <agl@google.com>
The case where ec_group_get_mont_data is NULL is only for arbitrary groups
which we now require to be prime order. BN_mod_exp_mont is fine with a NULL
BN_MONT_CTX. It will just compute it. Saves a bit of special-casing.
Also don't mark p-2 as BN_FLG_CONSTTIME as the exponent is public anyway.
Change-Id: Ie868576d52fc9ae5f5c9f2a4039a729151bf84c7
Reviewed-on: https://boringssl-review.googlesource.com/8307
Reviewed-by: Adam Langley <agl@google.com>
The Conscrypt revert cycled in long ago.
Change-Id: If3cdb211d7347dca88bd70bdc643f80b19a7e528
Reviewed-on: https://boringssl-review.googlesource.com/8306
Reviewed-by: Adam Langley <agl@google.com>
(This change will be sent upstream. Since the legacy X.509 stack is just
kept around for compatibility, if they decide to fix it in a different
way, we may wish to revert this and apply their fix.)
Dating back to SSLeay, X509_LOOKUP_METHOD had this X509_LU_RETRY
machinery. But it's not documented and it appears to have never worked.
Problems with the existing logic:
- X509_LU_* is not sure whether it is a type enum (to be passed into
X509_LOOKUP_by_*) or a return enum (to be retained by those same
functions).
- X509_LOOKUP_by_* is not sure whether it returns 0/1 or an X509_LU_*
value. Looking at the functions themselves, one might think it's the
latter, but for X509_LOOKUP_by_subject returning both 0 and
X509_LU_FAIL. But looking at the call sites, some expect 0/1 (such as
X509_STORE_get1_certs) while others expect an X509_LU_* enum (such as
X509_STORE_CTX_get1_issuer). It is very fortunate that FAIL happens to
be 0 and X509 happens to be 1.
These functions primarily call to X509_LOOKUP_METHOD hooks. Looking
through OpenSSL itself and code checked into Google, I found no
evidence that any hooks have been implemented except for
get_by_subject in by_dir.c. We take that one as definitive and observe
it believes it returns 0/1. Notably, it returns 1 on success even if
asked for a type other than X509_LU_X509. (X509_LU_X509 = 1. Others are
different.) I found another piece of third-party software which corroborates
this worldview.
- X509_STORE_get_by_subject's handling of X509_LU_RETRY (it's the j < 0
check) is broken. It saves j into vs->current_method where it probably
meant to save i. (This bug has existed since SSLeay.)
It also returns j (supposedly X509_LU_RETRY) while all callers of
X509_STORE_get_by_subject expect it to return 0/1 by checking with !
instead of <= 0. (Note that all other codepaths return 0 and 1 so this
function did not actually believe it returned X509_LU_* most of the
time.)
This, in turn, gives us a free of uninitialized pointers in
X509_STORE_get1_certs and other functions which expect that *ret is
filled in if X509_STORE_get_by_subject returns success. GCC 4.9 with
optimizations from the Android NDK noticed this, which trigged this
saga.
(It's only reachable if any X509_LOOKUP_METHOD returned
X509_LU_RETRY.)
- Although the code which expects X509_STORE_get_by_subject return 0/1
does not date to SSLeay, the X509_STORE_get_by_subject call in
X509_STORE_CTX_get1_issuer *does* (though, at the time, it was inline
in X509_verify_cert. That code believes X509_STORE_get_by_subject
returns an X509_LU_* enum, but it doesn't work either! It believes
*ret is filled in on X509_LU_RETRY, thus freeing another uninitialized
pointer (GCC noticed this too).
Since this "retry" code has clearly never worked, from SSLeay onwards,
unwind it completely rather than attempt to fix it. No
X509_LOOKUP_METHOD can possibly have depended on it.
Matching all non-broken codepaths X509_LOOKUP_by_* now returns 0/1 and
X509_STORE_get_by_subject returns 0/1. X509_LU_* is purely a type enum
with X509_LU_{REJECT,FAIL} being legacy constants to keep old code
compiling. (Upstream is recommended to remove those values altogether
for 1.1.0.)
On the off chance any get_by_* X509_LOOKUP_METHOD implementations did
not return 0/1 (I have found no evidence anywhere of this, and I believe
it wouldn't have worked anyway), the X509_LOOKUP_by_* wrapper functions
will coerce the return values back to 0/1 before passing up to the
callers which want 0/1. This both avoids the error-prone -1/0/1 calling
convention and, more importantly, avoids problems with third-party
callers which expect a X509_LU_* return code. 0/1 collide with FAIL/X509
while -1 will collide with RETRY and might confuse things.
Change-Id: I98ecf6fa7342866b9124dc6f0b422cb9ce4a1ae7
Reviewed-on: https://boringssl-review.googlesource.com/8303
Reviewed-by: Adam Langley <agl@google.com>
I named the compatibility function wrong.
Change-Id: Idc289c317c5826c338c1daf58a2d3b26b09a7e49
Reviewed-on: https://boringssl-review.googlesource.com/8301
Reviewed-by: Adam Langley <agl@google.com>
SSL_set_bio has some rather complex ownership story because whether rbio/wbio
are both owning depends on whether they are equal. Moreover, whether
SSL_set_bio(ssl, rbio, wbio) frees ssl->rbio depends on whether rbio is the
existing rbio or not. The current logic doesn't even get it right; see tests.
Simplify this. First, rbio and wbio are always owning. All the weird ownership
cases which we're stuck with for compatibility will live in SSL_set_bio. It
will internally BIO_up_ref if necessary and appropriately no-op the left or
right side as needed. It will then call more well-behaved ssl_set_rbio or
ssl_set_wbio functions as necessary.
Change-Id: I6b4b34e23ed01561a8c0aead8bb905363ee413bb
Reviewed-on: https://boringssl-review.googlesource.com/8240
Reviewed-by: Adam Langley <agl@google.com>
These are more remnants of CMS. Nothing uses them directly. Removing them means
more code we don't have to think about when importing upstream patches.
Also take out a bunch of dead prototypes nearby.
Change-Id: Ife094d9d2078570006d1355fa4e3323f435be608
Reviewed-on: https://boringssl-review.googlesource.com/8244
Reviewed-by: David Benjamin <davidben@google.com>
These are more pretty-printers for generic ASN.1 structures. They're never
called externally and otherwise are only used in the X509V3_EXT_PARSE_UNKNOWN
mode for the X509 pretty-print functions. That makes unknown extensions
pretty-print as ASN.1 structures.
This is a rather useless feature, so have that fall through to
X509V3_EXT_DUMP_UNKNOWN which does a hexdump instead.
(The immediate trigger is I don't know what |op| is in upstream's
8c918b7b9c93ba38790ffd1a83e23c3684e66f57 and don't think it is worth the time
to puzzle that out and verify it. Better ditch this code completely.)
Change-Id: I0217906367d83056030aea64ef344d4fedf74763
Reviewed-on: https://boringssl-review.googlesource.com/8243
Reviewed-by: David Benjamin <davidben@google.com>
These functions are never instantiated. (They're a remnant of the PKCS#7 and
CMS bits.) Next time upstream touches this code, we don't have to puzzle
through the diff and import it.
Change-Id: I67c2102ae13e8e0527d858e1c63637dd442a4ffb
Reviewed-on: https://boringssl-review.googlesource.com/8242
Reviewed-by: David Benjamin <davidben@google.com>
There's a __pragma expression which allows this. Android builds us Windows with
MinGW for some reason, so we actually do have to tolerate non-MSVC-compatible
Windows compilers. (Clang for Windows is much more sensible than MinGW and
intentionally mimicks MSVC.)
MinGW doesn't understand MSVC's pragmas and warns a lot. #pragma warning is
safe to suppress, so wrap those to shush them. This also lets us do away with a
few ifdefs.
Change-Id: I1f5a8bec4940d4b2d947c4c1cc9341bc15ec4972
Reviewed-on: https://boringssl-review.googlesource.com/8236
Reviewed-by: Adam Langley <agl@google.com>
While most of OpenSSL's assembly allows out < in too, some of it doesn't.
Upstream seems to not consider this a problem (or, at least, they're failing to
make a decision on whether it is a problem, so we should assume they'll stay
their course). Accordingly, require aliased buffers to exactly align so we
don't have to keep chasing this down.
Change-Id: I00eb3df3e195b249116c68f7272442918d7077eb
Reviewed-on: https://boringssl-review.googlesource.com/8231
Reviewed-by: Adam Langley <agl@google.com>
Set ctx->error = X509_V_ERR_OUT_OF_MEM when verification cannot
continue due to malloc failure. Similarly for issuer lookup failures
and caller errors (bad parameters or invalid state).
Also, when X509_verify_cert() returns <= 0 make sure that the
verification status does not remain X509_V_OK, as a last resort set
it it to X509_V_ERR_UNSPECIFIED, just in case some code path returns
an error without setting an appropriate value of ctx->error.
Add new and some missing error codes to X509 error -> SSL alert switch.
(Imported from upstream's 5553a12735e11bc9aa28727afe721e7236788aab.)
Change-Id: I3231a6b2e72a3914cb9316b8e90ebaee009a1c5f
Reviewed-on: https://boringssl-review.googlesource.com/8170
Reviewed-by: David Benjamin <davidben@google.com>
This callback is used by BoringSSL tests in order to simulate the time,
so that the tests have repeatable results. This API will allow consumers
of BoringSSL to write the same sort of tests.
Change-Id: I79d72bce5510bbd83c307915cd2cc937579ce948
Reviewed-on: https://boringssl-review.googlesource.com/8200
Reviewed-by: David Benjamin <davidben@google.com>
Match the actual name of the type.
Change-Id: I0ad27196ee2876ce0690d13068fa95f68b05b0da
Reviewed-on: https://boringssl-review.googlesource.com/8187
Reviewed-by: David Benjamin <davidben@google.com>
There's only one thing under "SNI Extension".
Change-Id: I8d8c54c286cb5775a20c4e2623896eb9be2f0009
Reviewed-on: https://boringssl-review.googlesource.com/8181
Reviewed-by: David Benjamin <davidben@google.com>
The business with ssl_record_prefix_len is rather a hassle. Instead, have
tls_open_record always decrypt in-place and give back a CBS to where the body
is.
This way the caller doesn't need to do an extra check all to avoid creating an
invalid pointer and underflow in subtraction.
Change-Id: I4e12b25a760870d8f8a503673ab00a2d774fc9ee
Reviewed-on: https://boringssl-review.googlesource.com/8173
Reviewed-by: Steven Valdez <svaldez@google.com>
Reviewed-by: David Benjamin <davidben@google.com>
The two modes are quite different. One of them requires the BIO honor an
extra BIO_ctrl. Also add an explanation at the top of
addDTLSRetransmitTests for how these tests work. The description is
scattered across many different places.
BUG=63
Change-Id: Iff4cdd1fbf4f4439ae0c293f565eb6780c7c84f9
Reviewed-on: https://boringssl-review.googlesource.com/8121
Reviewed-by: David Benjamin <davidben@google.com>
More spring-cleaning of unnecessary incompatibilities. Since
OpenSSL_add_all_algorithms_conf doesn't specify a configuration file, it's
perfectly sound to have such a function.
Dear BoringSSL, please add all algorithms.
Uh, sure. They were already all there, but I have added them!
PS: Could you also load all your configuration files while you're at it.
...I don't have any. Fine. I have loaded all configuration files which I
recognize. *mutters under breath* why does everyone ask all these strange
questions...
Change-Id: I57f956933d9e519445bf22f89853bd5f56904172
Reviewed-on: https://boringssl-review.googlesource.com/8160
Reviewed-by: Steven Valdez <svaldez@google.com>
Reviewed-by: David Benjamin <davidben@google.com>
The fake numbers collide with other numbers defined below. Also PUSH and POP
are actually used. DUP legitimately isn't though.
Change-Id: Iaa15a065d846b89b9b7958b78068393cfee2bd6f
Reviewed-on: https://boringssl-review.googlesource.com/8143
Reviewed-by: Steven Valdez <svaldez@google.com>
Reviewed-by: David Benjamin <davidben@google.com>
Some OpenSSL consumers use them, so provide no-op versions to make porting code
easier.
Change-Id: I4348568c1cb08d2b2c0a9ec9a17e2c0449260965
Reviewed-on: https://boringssl-review.googlesource.com/8142
Reviewed-by: David Benjamin <davidben@google.com>
Make building against software that expects OpenSSL easier.
Change-Id: I1af090ae8208218d6e226ee0baf51053699d85cc
Reviewed-on: https://boringssl-review.googlesource.com/8141
Reviewed-by: Steven Valdez <svaldez@google.com>
Reviewed-by: David Benjamin <davidben@google.com>
doc.go is still a little unhappy.
Change-Id: I5a8f3da91dabb45d29d0e08f13b7dabdcd521c38
Reviewed-on: https://boringssl-review.googlesource.com/8145
Reviewed-by: David Benjamin <davidben@google.com>
One of these tests the distribution of noise polynomials; the other
tests that that agreed-upon keys (prior to whitening) have roughly equal
numbers of 0s and 1s.
Along the way, expose a few more API bits.
Change-Id: I6b04708d41590de45d82ea95bae1033cfccd5d67
Reviewed-on: https://boringssl-review.googlesource.com/8130
Reviewed-by: Adam Langley <agl@google.com>
This commit adds coverage of the "offer" (first) step, as well as
testing all outputs of the "accept" (second) step, not just the shared
key.
Change-Id: Id11fe24029abc302442484a6c01fa496a1578b3a
Reviewed-on: https://boringssl-review.googlesource.com/8100
Reviewed-by: Adam Langley <agl@google.com>
This keeps the naming convention in line with the actual spec.
Change-Id: I34673f78dbc29c1659b4da0e49677ebe9b79636b
Reviewed-on: https://boringssl-review.googlesource.com/8090
Reviewed-by: David Benjamin <davidben@google.com>
The test vectors are taken from the reference implementation, modified
to output the results of its random-number generator, and the results of
key generation prior to SHA3. This allows the interoperability of the
two implementations to be tested somewhat.
To accomplish the testing, this commit creates a new, lower-level API
that leaves the generation of random numbers and all wire encoding and
decoding up to the caller.
Change-Id: Ifae3517696dde4be4a0b7c1998bdefb789bac599
Reviewed-on: https://boringssl-review.googlesource.com/8070
Reviewed-by: Adam Langley <agl@google.com>
Now that we no longer support Windows XP, this is available.
Unfortunately, the public header version of CRYPTO_MUTEX means we
still can't easily merge CRYPTO_MUTEX and CRYPTO_STATIC_MUTEX.
BUG=37
Change-Id: If309de3f06e0854c505083b72fd64d1dbb3f4563
Reviewed-on: https://boringssl-review.googlesource.com/8081
Reviewed-by: Adam Langley <agl@google.com>
This is easier to deploy, and more obvious. This commit reverts a few
pieces of e25775bc, but keeps most of it.
Change-Id: If8d657a4221c665349c06041bb12fffca1527a2c
Reviewed-on: https://boringssl-review.googlesource.com/8061
Reviewed-by: Adam Langley <agl@google.com>
This code has caused a long history of problems. This change rewrites it
completely with something that is, hopefully, much simplier and robust
and adds more testing.
Change-Id: Ibeef51f9386afd95d5b73316e451eb3a2d7ec4e0
Reviewed-on: https://boringssl-review.googlesource.com/8033
Reviewed-by: Adam Langley <agl@google.com>
Constants representing TLS 1.3 are added to allow for future work to be
flagged on TLS1_3_VERSION. To prevent BoringSSL from negotiating the
non-existent TLS 1.3 version, it is explicitly disabled using
SSL_OP_NO_TLSv1_3.
Change-Id: Ie5258a916f4c19ef21646c4073d5b4a7974d6f3f
Reviewed-on: https://boringssl-review.googlesource.com/8041
Reviewed-by: Adam Langley <agl@google.com>
This renames the Channel ID EncryptedExtensions message to allow for
compatibility with TLS 1.3 EncryptedExtensions.
Change-Id: I5b67d00d548518045554becb1b7213fba86731f2
Reviewed-on: https://boringssl-review.googlesource.com/8040
Reviewed-by: Adam Langley <agl@google.com>
OpenSSL's bbio logic is kind of crazy. It would be good to eventually do the
buffering in a better way (notably, bbio is fragile, if not outright broken,
for DTLS). In the meantime, this fixes a number of bugs where the existence of
bbio was leaked in the public API and broke things.
- SSL_get_wbio returned the bbio during the handshake. It must always return
the BIO the consumer configured. In doing so, internal accesses of
SSL_get_wbio should be switched to ssl->wbio since those want to see bbio.
For consistency, do the same with rbio.
- The logic in SSL_set_rfd, etc. (which I doubt is quite right since
SSL_set_bio's lifetime is unclear) would get confused once wbio got wrapped.
Those want to compare to SSL_get_wbio.
- If SSL_set_bio was called mid-handshake, bbio would get disconnected and lose
state. It forgets to reattach the bbio afterwards. Unfortunately, Conscrypt
does this a lot. It just never ended up calling it at a point where the bbio
would cause problems.
- Make more explicit the invariant that any bbio's which exist are always
attached. Simplify a few things as part of that.
Change-Id: Ia02d6bdfb9aeb1e3021a8f82dcbd0629f5c7fb8d
Reviewed-on: https://boringssl-review.googlesource.com/8023
Reviewed-by: Kenny Root <kroot@google.com>
Reviewed-by: Adam Langley <agl@google.com>
The 'elliptic_curves' extension is being renamed to 'supported_groups'
in the TLS 1.3 draft, and most of the curve-specific methods are
generalized to groups/group IDs.
Change-Id: Icd1a1cf7365c8a4a64ae601993dc4273802610fb
Reviewed-on: https://boringssl-review.googlesource.com/7955
Reviewed-by: David Benjamin <davidben@google.com>
open_iscsi assumes that it can get |OPENSSL_malloc| after including only
pem.h and err.h. Since pem.h already includes quite a lot, this change
adds crypto.h to that set so that open_iscsi is happy.
Change-Id: I6dc06c27088ce3ca46c1ab53bb29650033cba267
Reviewed-on: https://boringssl-review.googlesource.com/8031
Reviewed-by: David Benjamin <davidben@google.com>
The key schedule in TLS 1.3 requires a separate Extract and Expand phase
for the cryptographic computations.
Change-Id: Ifdac1237bda5212de5d4f7e8db54e202151d45ec
Reviewed-on: https://boringssl-review.googlesource.com/7983
Reviewed-by: David Benjamin <davidben@google.com>
CECPQ1 is a new key exchange that concatenates the results of an X25519
key agreement and a NEWHOPE key agreement.
Change-Id: Ib919bdc2e1f30f28bf80c4c18f6558017ea386bb
Reviewed-on: https://boringssl-review.googlesource.com/7962
Reviewed-by: David Benjamin <davidben@google.com>
This is consistent with the new convention in ssl_ecdh.c.
Along the way, change newhope_test.c to not iterate 1000 times over each
test.
Change-Id: I7a500f45b838eba8f6df96957891aa8e880ba089
Reviewed-on: https://boringssl-review.googlesource.com/8012
Reviewed-by: David Benjamin <davidben@google.com>
We don't have any of these.
Change-Id: I8d12284fbbab0ff35ac32d35a5f2eba326ab79f8
Reviewed-on: https://boringssl-review.googlesource.com/7981
Reviewed-by: Steven Valdez <svaldez@google.com>
Reviewed-by: David Benjamin <davidben@google.com>
They're completely unused now. The handshake message reassembly logic should
not depend on the state machine. This should partially free it up (ugly as it
is) to be shared with a future TLS 1.3 implementation while, in parallel, it
and the layers below, get reworked. This also cuts down on the number of states
significantly.
Partially because I expect we'd want to get ssl_hash_message_t out of there
too. Having it in common code is fine, but it needs to be in the (supposed to
be) protocol-agnostic handshake state machine, not the protocol-specific
handshake message layer.
Change-Id: I12f9dc57bf433ceead0591106ab165d352ef6ee4
Reviewed-on: https://boringssl-review.googlesource.com/7949
Reviewed-by: Adam Langley <agl@google.com>
Rather than this confusing coordination with the handshake state machine and
init_num changing meaning partway through, use the length field already in
BUF_MEM. Like the new record layer parsing, is no need to keep track of whether
we are reading the header or the body. Simply keep extending the handshake
message until it's far enough along.
ssl3_get_message still needs tons of work, but this allows us to disentangle it
from the handshake state.
Change-Id: Ic2b3e7cfe6152a7e28a04980317d3c7c396d9b08
Reviewed-on: https://boringssl-review.googlesource.com/7948
Reviewed-by: Adam Langley <agl@google.com>
BUF_MEM is actually a rather silly API for the SSL stack. There's separate
length and max fields, but init_buf effectively treats length as max and max as
nothing.
We possibly don't want to be using it long-term anyway (if nothing else, the
char*/uint8_t* thing is irritating), but in the meantime, it'll be easier to
separately fix up get_message's book-keeping and state tracking from where the
handshake gets its messages from.
Change-Id: I9e56ea008173991edc8312ec707505ead410a9ee
Reviewed-on: https://boringssl-review.googlesource.com/7947
Reviewed-by: Adam Langley <agl@google.com>
The existing logic gets confused in a number of cases around close_notify vs.
fatal alert. SSL_shutdown, while still pushing to the error queue, will fail to
notice alerts. We also get confused if we try to send a fatal alert when we've
already sent something else.
Change-Id: I9b1d217fbf1ee8a9c59efbebba60165b7de9689e
Reviewed-on: https://boringssl-review.googlesource.com/7952
Reviewed-by: David Benjamin <davidben@google.com>
