This fixes a bug in ASN1_TYPE_get. Partly imported from upstream's
261ec72d58af64327214a78ca1c54b169ad93c28, though I don't believe
ASN1_TYPE_set was broken per se. There's also a lot more than in that
commit.
I've added a test to ensure we maintain the unused bits invariant
anyway, in case external code relies on it. (The invariant comes from
the pointer being NULL-initialized and from ASN1_primitive_free zeroing
*pval on free.)
Change-Id: I4c0c57519a7628041d81c26cd850317e01409556
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/46324
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
This API does not come from OpenSSL, but OpenSSL does not appear to have
any way to get this information. There is X509_get0_pubkey_bitstr, but
that only works for X509 objects, not X509_PUBKEY.
Change-Id: Ifc8be554a4d8cbf830c32b95b953f092980804df
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/46304
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
See also 8129ac6ac4c0ca3a488c225cde580ede7dabe874 and
81198bf323ea9deda907714170d329ca7d2ff01f from upstream.
In trying to figure out why ASan (which normally catches overlapping
memcpys) didn't flag this, I noticed that we actually don't have tests
for empty inputs. I've added them to cipher_tests.txt where missing and
fixed a bad assert in ofb.c.
ASan still doesn't flag this because LLVM even requires memcpy handle
dst == src. Still, fixing it is less effort than getting a clear answer
from GCC and MSVC. Though this puts us in the frustrating position of
trying to follow a C rule that our main toolchain and sanitizer disavow.
https://bugs.llvm.org/show_bug.cgi?id=11763https://reviews.llvm.org/D86993
Change-Id: I53c64a84834ddf5cddca0b3d53a29998f666ea2f
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/46285
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
All our EVP_CIPHERs are deterministic, so there's usually no point in
testing only one direction. Some of the ECB tests were missing free
decryption tests. CTR is the same in both directions, but we ought to
test the API agrees. OFB vectors are doubled up, so we can merge them
together. Plus there are typos in the OFB-AES192.Decrypt tests, also
present upstream, so we weren't actually testing everything we should.
(I haven't removed the direction-specific logic altogether since the
tests imported from nist_cavp rely on it. Though there may be something
to be said for running them both ways since they don't actually double
them up...)
Change-Id: I36a77d342afa436e89ad244a87567e1a4c6ee9dc
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/46284
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
In doing so, this switches make_errors.go to take library names as
parameters rather than detecting it from the CWD. (I considered
detecting it, but then we'd need to map evp -> crypto/whatever and
crypto/whatever -> evp in both directions.)
Since crypto/hpke currently sits in the EVP namespace, I've gone ahead
and added that, so it should be easier to define new errors in
crypto/hpke. I've not added crypto/cipher, etc., yet. Moving those will
be a breaking change (consumers that put ERR_LIB_CIPHER and ERR_LIB_EVP
in a switch/case need patches).
Bug: 398
Change-Id: Ibae2afd46e076891fa517c377b540b2e492516f0
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/46264
Reviewed-by: Adam Langley <agl@google.com>
The build scripts distinguish between normal files and bcm.c fragments
based on whether code is in a subdirectory inside crypto/fipsmodule.
Bug: 401
Change-Id: Ieba88178e4f8e19f020e56e2567d5736a34bb43f
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/46224
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
The representation here is a bit more messy than necessary. In doing so,
clean up the variable names and smooth away two rough edges:
- X509_ALGOR_get0 would leave *out_param_value uninitialized if
*out_param_type is V_ASN1_UNDEF. Instead, set it to NULL, so callers
do not accidentally use an uninitialized pointer.
- X509_PUBKEY_set0_param, if key is NULL, would leave the key alone. No
one calls this function externally and none of the (since removed)
callers in OpenSSL rely on this behavior. A NULL check here adds a
discontinuity at the empty string that seems unnecessary here:
changing the algorithm without changing the key isn't useful.
(Note the API doesn't support changing the key without the algorithm.)
Note for reviewing: the representation of ASN1_TYPE is specified
somewhat indirectly. ASN1_TYPE uses the ASN1_ANY ASN1_ITEM, which has
utype V_ASN1_ANY. Then you look at asn1_d2i_ex_primitive and asn1_ex_c2i
which peel off the ASN1_TYPE layer and parse directly into the value
field, with a fixup for NULL. Hopefully we can rework this someday...
Change-Id: I628c4e20f8ea2fd036132242337f4dcac5ba5015
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/46165
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
In order to provide evidence to auditors that high-level functions end
up calling into the FIPS module, provide counters that allow for such
monitoring.
Change-Id: I55d45299f3050bf58077715ffa280210db156116
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/46124
Commit-Queue: Adam Langley <agl@google.com>
Reviewed-by: David Benjamin <davidben@google.com>
ASN1_OBJECTs are awkward. Sometimes they are static, when returned from
OBJ_nid2obj, and sometimes they are dynamic, when parsed from
crypto/asn1.
Most structures in crypto/asn1 need to support unknown OIDs and thus
must own their ASN1_OBJECTs. But they also may be initialized with
static ones in various APIs, such as X509_ALGOR_set0. To make that work,
ASN1_OBJECT_free detects static ASN1_OBJECTs and is a no-op.
Functions like X509_ALGOR_set0 take ownership, so OpenSSL has them take
a non-const ASN1_OBJECT*. To match, OBJ_nid2obj then returns a non-const
ASN1_OBJECT*, to signal that it is freeable.
However, this means OBJ_nid2obj's mutability doesn't match its return
type. In the fork, we switched OBJ_nid2obj to return const. But, in
doing so, we had to make X509_ALGOR_set0 and X509_PUBKEY_set0_param take
const ASN1_OBJECT, even though they would actually take ownership of
dynamic ASN1_OBJECTs. There are also a few internal casts with a TODO to
be const-correct.
Neither situation is ideal. (Perhaps a more sound model would be to copy
static ASN1_OBJECTs before putting them in most structs. But that would
not match current usage.) But I think aligning with OpenSSL is the
lesser evil here, since it avoids misleading set0 functions. Managing
ownership of ASN1_OBJECTs is much more common than mutating them. To
that end, I've added a note that ASN1_OBJECTs you didn't create must be
assumed immutable[*].
Update-Note: The change to OBJ_nid2obj should be compatible. The changes
to X509_PUBKEY_set0_param and X509_ALGOR_set0 may require fixing some
pointer types.
[*] This is *almost* honored by all of our functions. The exception is
c2i_ASN1_OBJECT, which instead checks the DYNAMIC flag as part of the
object reuse business. This would come up if we ever embedded
ASN1_OBJECTs directly in structs.
Change-Id: I1e6c700645c12b43323dd3887adb74e795c285b9
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/46164
Commit-Queue: David Benjamin <davidben@google.com>
Commit-Queue: Adam Langley <agl@google.com>
Reviewed-by: Adam Langley <agl@google.com>
OpenSSL classified their behavior as a bug and are fixing it for the
next release. In principle it'd be more compatible to emulate OpenSSL's
bug and undo it when we update OPENSSL_VERSION_NUMBER, but use of
PKCS12_parse is rare and this behavior is confusing, so let's leave it
as-is.
Bug: 250
Change-Id: I5f9825490a8afde67272dfaf476b35dbde94b59c
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/46064
Reviewed-by: Adam Langley <agl@google.com>
We aim to eventually make the entire X509 structure opaque, but let's
start small.
Update-Note: I believe this is now safe to do. If there are compile
failures, switch to X509_get0_notBefore, X509_getm_notBefore, and
X509_set1_notBefore, or revert this if I'm wrong and too many callers
still need updating.
Change-Id: I6e9d91630a10ac777e13ebcdeb543b3cbeea6383
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/45965
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
PKCS#12 is overly general and, among other things, supports disabling
encryption. In practice, the unencrypted form is not widely implemented.
Moreover, even in contexts where cleartext is fine, an unencrypted
PKCS#12 file still requires a password for the mandatory MAC component.
They're not very useful.
However, cryptography.io uses them. Previously, we added support for
parsing these. This CL adds support for creating them too, because now
cryptography.io now also depends on that.
Change-Id: Ib7c4e29615047b6c73f887fea7c80f8844999bb7
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/46045
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
pkcs12_test.cc was getting a bit long. Along the way, embed_test_data.go
needed a fix to work around a syntax quirk of C++.
Change-Id: Ic4a19f77d177ebd607918feb253a08f1f9037981
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/46044
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
Update-Note: This removes a function that appears to be unused. It also
hardcodes the use of MD5, so please do not use it.
Change-Id: I67909c6360e4737fc22742592f88b907eb818e96
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/45964
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
I got the values flipped around. Also cryptography.io wants
EC_GROUP_get_asn1_flag to check a curve's encoding. We (mostly) only
support named curves, so just return OPENSSL_EC_NAMED_CURVE.
Change-Id: I544e76b7380ecd8dceb1df3db4dd4cf5cb322352
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/46024
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
This improves compatibility with cryptography.io. cryptography.io
doesn't actually care what we return, since the code won't run, but
feigning success seems better than failure. If some application does try
to run this function and checks, returning an error will probably crash
it.
Change-Id: I7a8164753a2f1a7b31dbeb10c7030c5e5fea2bc9
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/46004
Commit-Queue: David Benjamin <davidben@google.com>
Commit-Queue: Adam Langley <agl@google.com>
Reviewed-by: Adam Langley <agl@google.com>
Per RFC5280, section 5.1.1.2,
[signatureAlgorithm] MUST contain the same algorithm identifier as the
signature field in the sequence tbsCertList (Section 5.1.2.2).
This aligns with a check we already do on the X.509 side.
Update-Note: Invalid CRLs with inconsistent inner and outer signature
algorithms will now be rejected.
Change-Id: I9ef495a9b26779399c932903871391aacd8f2618
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/45946
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
This may as well be computed from block_size. This reduces the
per-EVP_CIPHER_CTX memory usage slightly.
Update-Note: It doesn't look like anyone is reading into this field. If
they are, we can ideally fix it, or revert this if absolutely necessary.
Change-Id: Ieef9177bed1671efca23d4f94d3d528f82568fc6
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/45884
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
The X509 version APIs confuse everyone, including our own tests
(v3name_test.cc was using the wrong value). Introduce constants. Also
document which version to use for each type. It is extra confusing that,
although certificates and CRLs use the same Version enum, RFC5280
defines X.509 v3 certificates with X.509 v2 CRLs. (There are no v3
CRLs.)
I'll send a similar patch to OpenSSL to keep upstream aligned.
Change-Id: If096138eb004156d43df87a6d74f695b04f67480
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/45944
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
This is a no-op, but relying on ASN1_INTEGER_get's internal handling of
NULL is confusing.
Change-Id: I49ab38c170d2f39fd3a4780b8c44b103a2ba4615
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/45924
Reviewed-by: Adam Langley <agl@google.com>
There are a few places where it is useful to run ECDSA with a specified
nonce:
- An ECDSA KAT in the module self-check
- Unit tests for particular test vectors
- Fuzzing the implementation (requested by the cryptofuzz project)
This replaces the fixed_k machinery with a separate function. Although
they are effectively the same, I've used two different functions.
One is internal and only used in the module self-check. The other is
exported for unit tests and cryptofuzz but marked with a for_testing.
(Chromium's presubmits flag uses of "for_testing" functions outside of
unit tests. The KAT version isn't in a test per se, so it's a separate
function.)
Bug: 391
Change-Id: I0f764d89bf0ac2081307e1079623d508fb0f2df7
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/45867
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
The extra computation in the "setup" half is a remnant of
ECDSA_sign_setup, which was designed to amortize the message-independent
portion of the signature. We don't care about this amortization, but we
do want some control over k for testing. Instead, have ecdsa_sign_impl
take k and do the rest.
In doing so, this lets us compute m and kinv slightly later, which means
fewer temporaries are needed.
Bug: 391
Change-Id: I398004a5388ce5b7e839db6c36edf8464643d16f
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/45866
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
In FIPS mode, we maintain a lock in order to implement clearing DRBG
state on process shutdown. This lock serves two purposes:
1. It protects the linked list of all DRBG states, which needs to be
updated the first time a thread touches RAND_bytes, when a thread
exits, and on process exit.
2. It ensures threads alive during process shutdown do not accidentally
touch DRBGs after they are cleared, by way of taking a ton of read
locks in RAND_bytes across some potentially time-consuming points.
This means that when one of the rare events in (1) happens, it must
contend with the flurry of read locks in (2). Split these uses into two
locks. The second lock now only ever sees read locks until process
shutdown, and the first lock is only accessed in rare cases.
Change-Id: Ib856c7a3bb937bbfa5d08534031dbf4ed3315cab
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/45844
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
This is a very very basic sanity check on k generation, but it helps
make sure we haven't *completely* disconnected the RNG.
Change-Id: If7ae5dd6be3d0866962cd966b8c1ed1cdedffb50
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/45865
Reviewed-by: Adam Langley <agl@google.com>
This reverts commit 4251d0d3f6, except
that the reland dates have been updated to not be in the past.
Change-Id: I7812c0e36d87ed1e049ec0a7d92a23efec881a81
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/45704
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: David Benjamin <davidben@google.com>
While I'm here, rewrite it a bit to align more with our preferred style.
(No assignments in conditions, no need for NULL checks on free
functions.)
See also 64a1b940d2b640e5edf0feae90e81bbb6b4941e7 from upstream.
Change-Id: I99a122343541e89d5950888de2c708cfa3ec45e2
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/45686
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
CVE-2021-23840
(Imported from upstream's 6a51b9e1d0cf0bf8515f7201b68fb0a3482b3dc1.)
This differs slightly from upstream's version:
- EVP_R_OUTPUT_WOULD_OVERFLOW didn't seem necessary when ERR_R_OVERFLOW
already exists. (Also since we use CIPHER_R_*, it wouldn't have helped
with compatibility anyway. Though there's probably something to be
said for us folding CIPHER_R_* back into EVP_R_*.)
- For simplicity, just check in_len + bl at the top, rather than trying
to predict the exact number of bytes written.
Update-Note: Passing extremely large input lengths into EVP_CipherUpdate
will now fail. Use EVP_AEAD instead, which is size_t-based and has more
explicit output bounds.
Change-Id: I31835c89dcdecb6b112828f57deb798dc7187db5
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/45685
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
Update-Note: No one uses this function. It had a NULL dereference in
some error cases. See CVE-2021-23841.
Change-Id: Ie1cc97615ac8b674147715d7d62e62faf218ae65
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/45684
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
This aligns with OpenSSL's behavior. RFC7301 says servers should return
no_application_protocol if the client supported ALPN but no common
protocol was found. We currently interpret all values as
SSL_TLSEXT_ERR_NOACK. Instead, implement both modes and give guidance on
whne to use each. (NOACK is still useful because the callback may be
shared across multiple configurations, some of which don't support ALPN
at all. Those would want to return NOACK to ignore the list.)
To match upstream, I've also switched SSL_R_MISSING_ALPN, added for
QUIC, to SSL_R_NO_APPLICATION_PROTOCOL.
Update-Note: Callers that return SSL_TLSEXT_ERR_ALERT_FATAL from the
ALPN callback will change behavior. The old behavior may be restored by
returning SSL_TLSEXT_ERR_NOACK, though see the documentation for new
recommendations on return values.
Change-Id: Ib7917b5f8a098571bed764c79aa7a4ce0f728297
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/45504
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
This is largely some cleanup so the three features follow the same
patterns and is hopefully cleaner (no more separate static and
non-static paths). The practical impact is probably nil. (Linux-based
ARM builds with crypto extensions as a baseline, if any exist, save
binary size.)
Change-Id: I2214b1a54e2074024b8eeb51799a08b94646cbf3
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/45485
Reviewed-by: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
When a feature is enabled statically in the build config, the compiler
defines __ARM_NEON and also considers itself free to emit NEON code.
In this case, there is no need to check for NEON support at runtime
since the binary will not work without NEON anyway. Moving the check to
compile time lets us drop unused code.
Chrome has required NEON on Android for nearly five years now. However,
historically there was a bad CPU which broke on some NEON code, but not
others. See https://crbug.com/341598 and https://crbug.com/606629. We
worked around that CPU by parsing /proc/cpuinfo and intentionally
dropping the optimization. This is not a stable situation, however, as
we're hoping the compiler is not good enough at emitting NEON to trigger
this bug.
Since then, the number of affected devices has dropped, and Chrome has
raised the minimum Android requirement to L. The Net.HasBrokenNEON
metric from Chrome is now well in the noise.
This CL stops short of removing the workaround altogether because some
consumers of Cronet are unsure whether they needed this workaround.
Those consumers also build without __ARM_NEON, so gating on that works
out. We'll decide what to do with it pending metrics from them.
Update-Note: Builds with __ARM_NEON (-mfpu=neon) will now drop about
30KiB of dead code, but no longer work (if they even did before) on a
particular buggy CPU. Builds without __ARM_NEON are not affected.
Change-Id: Id8f7bccfb75afe0a1594572ea20c51d275b0a256
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/45484
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
test_fips probably needs to exercise everything that we have self-tests
for.
(The following change will eliminate the duplication of the code to
create the FFDH group. For reasons, that can't be done in this change.)
Change-Id: Ia72064db77381e7cf396a34b4723b2607f26f00b
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/45404
Reviewed-by: Adam Langley <alangley@gmail.com>
Reviewed-by: David Benjamin <davidben@google.com>
Commit-Queue: Adam Langley <agl@google.com>
This fixes the build for folks not using bcm.c.
Change-Id: I47935d8af7cb5a12ff2918ee2a8774182681d930
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/45384
Commit-Queue: David Benjamin <davidben@google.com>
Commit-Queue: Adam Langley <agl@google.com>
Reviewed-by: Adam Langley <agl@google.com>
This invovles a |2048|^|225| modexp, which is far from ideal, but is now
required in FIPS mode.
Change-Id: Id7384b4ba92aa74e971231bc44fa0f10434d18e2
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/45085
Commit-Queue: Adam Langley <agl@google.com>
Reviewed-by: David Benjamin <davidben@google.com>
Our use-case for this does not require optimisation at the current time,
so a clean C implementation is fine.
Change-Id: I8f29572c33e8dbcc37961c099c71c14aafc8d0a3
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/45164
Reviewed-by: David Benjamin <davidben@google.com>
In order to support cross-compiling:
1) inject-hash needs to know to use SHA-256 for the hash function
on Android. Since that's a good idea on Aarch64 in general
(due to common hardware support), do it for all Aarch64.
2) We need to use the compiler to run the preprocessor, not plain
cpp, because the compiler will get the built-in #defines right.
Change-Id: Ie00d46e9e6d489fcb9e3f3e5e625aa289c7e0d73
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/45044
Reviewed-by: David Benjamin <davidben@google.com>
We clear all heap memory on free now, thus the difference between these
functions is quite small. There are some differences though:
Firstly, BN_clear_free will attempt to zero out static limb data. But
static data is probably read-only and thus trying to zero it will crash.
Secondly it will try to zero out the BIGNUM structure itself. But either
it's on the heap, and will be zeroed anyway, or else it's on the stack,
and we don't try and clear the stack in general because the compiler is
duplicating bits of it at will anyway.
Change-Id: I8a07385a102cfd308b555432942225c25eb7c12d
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/45084
Reviewed-by: David Benjamin <davidben@google.com>
OpenSSL has a fixed-width version of DH_compute_key nowadays. Searching
around callers of DH_compute_key, many of them go back and re-pad the
secret anyway. Uses of DH should migrate to modern primitives but, in
the meantime, DH_compute_key_padded seems worthwhile for OpenSSL
compatibility and giving fixed-width users a function to avoid the
timing leak.
Bump BORINGSSL_API_VERSION since one of the uses is in wpa_supplicant
and they like to compile against a wide range of Android revisions.
Update-Note: No compatibility impact, but callers that use
DH_compute_key and then fix up the removed leading zeros can switch to
this function. Then they should migrate to something else.
Change-Id: Icf8b2ace3972fa174a0f08ece39710f7599f96f2
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/45004
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
Aarch64 perlasm outputs need to be run via the C preprocessor before
being parsed by delocate.
Change-Id: Ie420388e0e1707cb064d696ee8f87728cab9a36e
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/44847
Reviewed-by: David Benjamin <davidben@google.com>
Commit-Queue: Adam Langley <agl@google.com>
Windows on Arm (WoA) builds are currently using the C implementations
of the various functions within BoringSSL. This patch enables feature
detection for the Neon and hardware crypto optimizations, and updates
the perl script to generate AArch64 .S files for WoA.
Note these files use GNU assembler syntax (specifically tested with
Clang assembler), not armasm.
Change-Id: Id8841f4db0498ec16215095a4e6bd60d427cd54b
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/43304
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: David Benjamin <davidben@google.com>
gai_strerror is one of the Windows functions which behaves differently
whether UNICODE is defined. See
https://docs.microsoft.com/en-us/windows/win32/intl/conventions-for-function-prototypes
Call gai_strerrorA so that we behave consistently in both modes. This
fixes the build failure in
https://chromium-review.googlesource.com/c/chromium/src/+/2613519.
It also fixes a type error in the connect BIO (built but not used in
Chromium), which wasn't noticed because ERR_add_error_data is a variadic
function and untyped. (The type error won't go out of bounds because
we're interpreting a NUL-terminated WCHAR* as a NUL-terminated char*.
The string will be misinterpreted, but it still will be terminated
either at the NUL WCHAR or, more likely, the upper zero byte of the
first Latin-1 character in the string.)
The ERR_add_error_data call raises the question of which of our char*
strings are UTF-8 and which are the POSIX locale / Windows code page
(when those are not also UTF-8). This CL doesn't address this and only
fixes the character width error. Realistically, calling code tosses
char* to printf so often that non-UTF-8 locales are probably a lost
cause. (Although right now we do not transform any OS error strings, so
tossing them to printf works fine. The outputs of functions like
ASN1_STRING_to_UTF8, not so much.)
Change-Id: Ie789730658829bde90022605ade2c86b8a65c3de
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/44964
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
Updating HPKE to draft-irtf-cfrg-hpke-07 added a number of tests to
crypto/hpke/hpke_test_vectors.txt. The time to run HPKE crypto tests
increased from 0.09 to 0.56 seconds (averaged over 6 runs). The runtime
for the whole crypto_test suite increased from 86.44 to 88.19 (measured
once).
Profiling revealed an excessive amount of CPU time (~50% for HPKE tests)
spent on std::map lookups in ReadNext(). I found a slow loop that
computes the suffix for a repeated attribute by hammering a std::map
with incremental guesses.
This CL adds a std::map<std::string, size_t> for counting repeated
attributes, eliminating the need for the loop. This reduces the runtime
for HPKE tests from 0.56 to 0.12 seconds (averaged over 6 runs). For the
whole crypto_test suite, runtime is reduced from 88.19 to 86.71
seconds (measured once).
Change-Id: Ie87f4a7f3edc95d434226d2959dcf09974f0656f
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/44905
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: David Benjamin <davidben@google.com>
Change-Id: I814f55742910c519e9b64aca1b15a4d754adc541
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/44944
Commit-Queue: Adam Langley <agl@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: David Benjamin <davidben@google.com>
David Benjamin
Dan McArdle
Bradley Hess
Adam Langley
Anthony Roberts
Adam Langley