The API used to set what SNI value to send in the ClientHello
can also be used on server SSL objects, with undocumented and
un-useful behavior. Unfortunately, when generic SSL_METHODs
are used, s->server is still set, prior to the start of the
handshake, so we cannot prevent this nonsensical usage at the
present time. Leave a note to revisit this when ABI-breaking
changes are permitted.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6378)
In particular, adhere to the rule that we must not modify any
property of an SSL_SESSION object once it is (or might be) in
a session cache. Such modifications are thread-unsafe and have
been observed to cause crashes at runtime.
To effect this change, standardize on the property that
SSL_SESSION->ext.hostname is set only when that SNI value
has been negotiated by both parties for use with that session.
For session resumption this is trivially the case, so only new
handshakes are affected.
On the client, the new semantics are that the SSL->ext.hostname is
for storing the value configured by the caller, and this value is
used when constructing the ClientHello. On the server, SSL->ext.hostname
is used to hold the value received from the client. Only if the
SNI negotiation is successful will the hostname be stored into the
session object; the server can do this after it sends the ServerHello,
and the client after it has received and processed the ServerHello.
This obviates the need to remove the hostname from the session object
in case of failed negotiation (a change that was introduced in commit
9fb6cb810b in order to allow TLS 1.3
early data when SNI was present in the ClientHello but not the session
being resumed), which was modifying cached sessions in certain cases.
(In TLS 1.3 we always produce a new SSL_SESSION object for new
connections, even in the case of resumption, so no TLS 1.3 handshakes
were affected.)
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6378)
These tiny functions only read from the input SSL, and we are
about to use them from functions that only have a const SSL* available,
so propagate const a bit further.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6378)
The spec says that a client MUST set legacy_version to TLSv1.2, and
requires servers to verify that it isn't SSLv3.
Fixes#6600
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6747)
Originally suggested solution for "Return Of the Hidden Number Problem"
is arguably too expensive. While it has marginal impact on slower
curves, none to ~6%, optimized implementations suffer real penalties.
Most notably sign with P-256 went more than 2 times[!] slower. Instead,
just implement constant-time BN_mod_add_quick.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: David Benjamin <davidben@google.com>
(Merged from https://github.com/openssl/openssl/pull/6664)
It was false positive, but one can as well view it as readability issue.
Switch even to unsigned indices because % BN_BYTES takes 4-6 instructions
with signed dividend vs. 1 (one) with unsigned.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Using the rsa_pss_rsae_sha256 sig alg should imply that the key OID is
rsaEncryption. Similarly rsa_pss_pss_sha256 implies the key OID is
rsassaPss. However we did not check this and incorrectly tolerated a key
OID that did not match the sig alg sent by the peer.
Fixes#6611
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6732)
The GOST ciphers are dynamically loaded via the GOST engine, so we must
be able to support that. The engine also uses DSA and CMS symbols, so we
skip the test on no-dsa or no-cms.
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6730)
If we issue new tickets due to post-handshake authentication there is no
reason to remove previous tickets from the cache. The code that did that
only removed the last session anyway - so if more than one ticket got
issued then those other tickets are still valid.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6722)
The config file can override it.
In case of the server, it needs to be set on the ctx or some of the
other functions on the ctx might file.
Reviewed-by: Rich Salz <rsalz@openssl.org>
DH: #6718
By default `ec_scalar_mul_ladder` (which uses the Lopez-Dahab ladder
implementation) is used only for (k * Generator) or (k * VariablePoint).
ECDSA verification uses (a * Generator + b * VariablePoint): this commit
forces the use of `ec_scalar_mul_ladder` also for the ECDSA verification
path, while using the default wNAF implementation for any other case.
With this commit `ec_scalar_mul_ladder` loses the static attribute, and
is added to ec_lcl.h so EC_METHODs can directly use it.
While working on a new custom EC_POINTs_mul implementation, I realized
that many checks (e.g. all the points being compatible with the given
EC_GROUP, creating a temporary BN_CTX if `ctx == NULL`, check for the
corner case `scalar == NULL && num == 0`) were duplicated again and
again in every single implementation (and actually some
implementations lacked some of the tests).
I thought that it makes way more sense for those checks that are
independent from the actual implementation and should always be done, to
be moved in the EC_POINTs_mul wrapper: so this commit also includes
these changes.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6690)
This commit uses the new ladder scaffold to implement a specialized
ladder step based on differential addition-and-doubling in mixed
Lopez-Dahab projective coordinates, modified to independently blind the
operands.
The arithmetic in `ladder_pre`, `ladder_step` and `ladder_post` is
auto generated with tooling:
- see, e.g., "Guide to ECC" Alg 3.40 for reference about the
`ladder_pre` implementation;
- see https://www.hyperelliptic.org/EFD/g12o/auto-code/shortw/xz/ladder/mladd-2003-s.op3
for the differential addition-and-doubling formulas implemented in
`ladder_step`;
- see, e.g., "Fast Multiplication on Elliptic Curves over GF(2**m)
without Precomputation" (Lopez and Dahab, CHES 1999) Appendix Alg Mxy
for the `ladder_post` implementation to recover the `(x,y)` result in
affine coordinates.
Co-authored-by: Billy Brumley <bbrumley@gmail.com>
Co-authored-by: Sohaib ul Hassan <soh.19.hassan@gmail.com>
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6690)
and catch corner cases better and earlier
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6690)
for specialized Montgomery ladder implementations
PR #6009 and #6070 replaced the default EC point multiplication path for
prime and binary curves with a unified Montgomery ladder implementation
with various timing attack defenses (for the common paths when a secret
scalar is feed to the point multiplication).
The newly introduced default implementation directly used
EC_POINT_add/dbl in the main loop.
The scaffolding introduced by this commit allows EC_METHODs to define a
specialized `ladder_step` function to improve performances by taking
advantage of efficient formulas for differential addition-and-doubling
and different coordinate systems.
- `ladder_pre` is executed before the main loop of the ladder: by
default it copies the input point P into S, and doubles it into R.
Specialized implementations could, e.g., use this hook to transition
to different coordinate systems before copying and doubling;
- `ladder_step` is the core of the Montgomery ladder loop: by default it
computes `S := R+S; R := 2R;`, but specific implementations could,
e.g., implement a more efficient formula for differential
addition-and-doubling;
- `ladder_post` is executed after the Montgomery ladder loop: by default
it's a noop, but specialized implementations could, e.g., use this
hook to transition back from the coordinate system used for optimizing
the differential addition-and-doubling or recover the y coordinate of
the result point.
This commit also renames `ec_mul_consttime` to `ec_scalar_mul_ladder`,
as it better corresponds to what this function does: nothing can be
truly said about the constant-timeness of the overall execution of this
function, given that the underlying operations are not necessarily
constant-time themselves.
What this implementation ensures is that the same fixed sequence of
operations is executed for each scalar multiplication (for a given
EC_GROUP), with no dependency on the value of the input scalar.
Co-authored-by: Sohaib ul Hassan <soh.19.hassan@gmail.com>
Co-authored-by: Billy Brumley <bbrumley@gmail.com>
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6690)
Run `make update ERROR_REBUILD=-rebuild` to remove some stale error
codes for SM2 (which is now using its own submodule for error codes,
i.e., `SM2_*`).
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6690)
Move base 2^64 code to own #if section. It was nested in base 2^51 section,
which arguably might have been tricky to follow.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6699)
Base 2^64 addition/subtraction and final reduction failed to treat
partially reduced values correctly.
Thanks to Wycheproof Project for vectors and Paul Kehrer for report.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6699)
"Computationally constant-time" means that it might still leak
information about input's length, but only in cases when input
is missing complete BN_ULONG limbs. But even then leak is possible
only if attacker can observe memory access pattern with limb
granularity.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5254)
Check that we are either configured for PSK, or that we have a TLSv1.3
capable certificate type. DSA certs can't be used in TLSv1.3 and we
don't (currently) allow GOST ones either (owing to the lack of standard
sig algs).
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6650)
Change the description for BN_hex2bn() so that it uses the same BIGNUM argument name as its prototype.
CLA: trivial
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6712)
Note that exported functions maintain original behaviour, so that
external callers won't observe difference. While internally we can
now perform Montogomery multiplication on fixed-length vectors, fixed
at modulus size. The new functions, bn_to_mont_fixed_top and
bn_mul_mont_fixed_top, are declared in bn_int.h, because one can use
them even outside bn, e.g. in RSA, DSA, ECDSA...
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: David Benjamin <davidben@google.com>
(Merged from https://github.com/openssl/openssl/pull/6662)
The new flag marks vectors that were not treated with bn_correct_top,
in other words such vectors are permitted to be zero padded. For now
it's BN_DEBUG-only flag, as initial use case for zero-padded vectors
would be controlled Montgomery multiplication/exponentiation, not
general purpose. For general purpose use another type might be more
appropriate. Advantage of this suggestion is that it's possible to
back-port it...
bn/bn_div.c: fix memory sanitizer problem.
bn/bn_sqr.c: harmonize with BN_mul.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: David Benjamin <davidben@google.com>
(Merged from https://github.com/openssl/openssl/pull/6662)
