Andy Polyakov [Fri, 6 Jul 2018 13:55:34 +0000 (15:55 +0200)]
ec/ecdsa_ossl.c: revert blinding in ECDSA signature.
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>
(Merged from https://github.com/openssl/openssl/pull/6796)
Andy Polyakov [Fri, 6 Jul 2018 13:13:15 +0000 (15:13 +0200)]
bn/bn_{mont|exp}.c: switch to zero-padded intermediate vectors.
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>
(Merged from https://github.com/openssl/openssl/pull/6707)
Andy Polyakov [Fri, 6 Jul 2018 13:02:29 +0000 (15:02 +0200)]
bn/bn_lib.c: add BN_FLG_FIXED_TOP flag.
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>
(Merged from https://github.com/openssl/openssl/pull/6707)
Kurt Roeckx [Wed, 25 Jul 2018 16:55:16 +0000 (18:55 +0200)]
Make number of Miller-Rabin tests for a prime tests depend on the security level of the prime
The old numbers where all generated for an 80 bit security level. But
the number should depend on security level you want to reach. For bigger
primes we want a higher security level and so need to do more tests.
Reviewed-by: Richard Levitte <levitte@openssl.org> Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com> Reviewed-by: Paul Dale <paul.dale@oracle.com>
GH: #6075 Fixes: #6012
(cherry picked from commit feac7a1c8be49fbcb76fcb721ec9f02fdd91030e)
Kurt Roeckx [Wed, 25 Apr 2018 19:47:20 +0000 (21:47 +0200)]
Change the number of Miller-Rabin test for DSA generation to 64
This changes the security level from 100 to 128 bit.
We only have 1 define, this sets it to the highest level supported for
DSA, and needed for keys larger than 3072 bit.
Reviewed-by: Richard Levitte <levitte@openssl.org> Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com> Reviewed-by: Paul Dale <paul.dale@oracle.com>
GH: #6075
(cherry picked from commit 74ee379651fb2bb12c6f7eb9fa10e70be89ac7c8)
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6791)
(cherry picked from commit 037241bf046be8cfc7e9216959393dd20b06fc21)
ecp_nistz256_set_from_affine is called when application attempts to use
custom generator, i.e. rarely. Even though it was wrong, it didn't
affect point operations, they were just not as fast as expected.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6738)
The ecp_nistz256_scatter_w7 function is called when application
attempts to use custom generator, i.e. rarely. Even though non-x86_64
versions were wrong, it didn't affect point operations, they were just
not as fast as expected.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6738)
Andy Polyakov [Sat, 21 Jul 2018 11:50:14 +0000 (13:50 +0200)]
apps/dsaparam.c: fix -C output.
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com> Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
(Merged from https://github.com/openssl/openssl/pull/6758)
Andy Polyakov [Mon, 16 Jul 2018 16:17:44 +0000 (18:17 +0200)]
bn/bn_lib.c address Coverity nit in bn2binpad.
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.
Andy Polyakov [Sun, 4 Feb 2018 14:20:29 +0000 (15:20 +0100)]
bn/bn_lib.c: make BN_bn2binpad computationally constant-time.
"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)
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)
Andy Polyakov [Fri, 6 Jul 2018 12:54:34 +0000 (14:54 +0200)]
bn/bn_mont.c: improve readability of post-condition code.
Reviewed-by: Rich Salz <rsalz@openssl.org> Reviewed-by: David Benjamin <davidben@google.com>
(Merged from https://github.com/openssl/openssl/pull/6662)
Andy Polyakov [Fri, 6 Jul 2018 11:46:07 +0000 (13:46 +0200)]
bn/bn_mont.c: move boundary condition check closer to caller.
Reviewed-by: Rich Salz <rsalz@openssl.org> Reviewed-by: David Benjamin <davidben@google.com>
(Merged from https://github.com/openssl/openssl/pull/6662)
Andy Polyakov [Fri, 6 Jul 2018 11:16:40 +0000 (13:16 +0200)]
bn/bn_lib.c: remove bn_check_top from bn_expand2.
Trouble is that addition is postponing expansion till carry is
calculated, and if addition carries, top word can be zero, which
triggers assertion in bn_check_top.
Reviewed-by: Rich Salz <rsalz@openssl.org> Reviewed-by: David Benjamin <davidben@google.com>
(Merged from https://github.com/openssl/openssl/pull/6662)
Richard Levitte [Wed, 11 Jul 2018 09:05:15 +0000 (11:05 +0200)]
Windows: avoid using 'rem' in the nmake makefile
To avoid the possibility that someone creates rem.exe, rem.bat or
rem.cmd, simply don't use it. In the cases it was used, it was to
avoid empty lines, but it turns out that nmake handles those fine, so
no harm done.
Reviewed-by: Andy Polyakov <appro@openssl.org> Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/6686)
Richard Levitte [Tue, 10 Jul 2018 12:12:33 +0000 (14:12 +0200)]
Windows: fix echo for nmake
It seems that nmake first tries to run executables on its own, and
only pass commands to cmd if that fails. That means it's possible to
have nmake run something like 'echo.exe' when the builtin 'echo'
command was expected, which might give us unexpected results.
To get around this, we create our own echoing script and call it
explicitly from the nmake makefile.
Fixes #6670
Reviewed-by: Andy Polyakov <appro@openssl.org> Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/6686)
Rich Salz [Thu, 28 Jun 2018 22:13:54 +0000 (18:13 -0400)]
Zero-fill IV by default.
Fixes uninitialized memory read reported by Nick Mathewson
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6603)
(cherry picked from commit 10c3c1c1ec41ce16e51b92bb18fab92d1a42b49c)
Richard Levitte [Fri, 22 Jun 2018 07:33:29 +0000 (09:33 +0200)]
OpenSSL-II style for emacs: don't indent because of extern block
We don't want an indentation step inside a 'extern "C" {' .. '}'
block. Apparently, cc-mode has a c-offsets-alist keyword to allow
exactly this.
Reviewed-by: Rich Salz <rsalz@openssl.org> Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/6557)
David von Oheimb [Sat, 10 Feb 2018 14:45:11 +0000 (15:45 +0100)]
add documentation for OCSP_basic_verify()
Reviewed-by: Rich Salz <rsalz@openssl.org> Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6227)
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de> Reviewed-by: Paul Dale <paul.dale@oracle.com> Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6549)
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de> Reviewed-by: Paul Dale <paul.dale@oracle.com> Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6549)
Matt Caswell [Tue, 19 Jun 2018 14:07:02 +0000 (15:07 +0100)]
Add blinding to a DSA signature
This extends the recently added ECDSA signature blinding to blind DSA too.
This is based on side channel attacks demonstrated by Keegan Ryan (NCC
Group) for ECDSA which are likely to be able to be applied to DSA.
Normally, as in ECDSA, during signing the signer calculates:
s:= k^-1 * (m + r * priv_key) mod order
In ECDSA, the addition operation above provides a sufficient signal for a
flush+reload attack to derive the private key given sufficient signature
operations.
As a mitigation (based on a suggestion from Keegan) we add blinding to
the operation so that:
s := k^-1 * blind^-1 (blind * m + blind * r * priv_key) mod order
Since this attack is a localhost side channel only no CVE is assigned.
This commit also tweaks the previous ECDSA blinding so that blinding is
only removed at the last possible step.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6523)
Jack Bates [Thu, 5 Jan 2017 16:58:18 +0000 (09:58 -0700)]
Convert _meth_get_ functions to const getters
Reviewed-by: Richard Levitte <levitte@openssl.org> Reviewed-by: Matt Caswell <matt@openssl.org>
(cherry picked from commit 693be9a2cb0fc79fe856259feea54772c18a3637)
(Merged from https://github.com/openssl/openssl/pull/5750)
Andy Polyakov [Fri, 8 Jun 2018 13:02:39 +0000 (15:02 +0200)]
bn/asm/sparcv9-mont.pl: iron another glitch in squaring code path.
This module is used only with odd input lengths, i.e. not used in normal
PKI cases, on contemporary processors. The problem was "illuminated" by
fuzzing tests.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6440)
Mingtao Yang [Wed, 6 Jun 2018 16:34:18 +0000 (09:34 -0700)]
modes/ocb128.c: Reset nonce-dependent variables on setiv
Upon a call to CRYPTO_ocb128_setiv, either directly on an OCB_CTX or
indirectly with EVP_CTRL_AEAD_SET_IVLEN, reset the nonce-dependent
variables in the OCB_CTX.
Reviewed-by: Andy Polyakov <appro@openssl.org> Reviewed-by: Rich Salz <rsalz@openssl.org> Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/6420)
Marcus Huewe [Fri, 11 May 2018 10:24:56 +0000 (12:24 +0200)]
Do not free a session before calling the remove_session_cb
If the remove_session_cb accesses the session's data (for instance,
via SSL_SESSION_get_protocol_version), a potential use after free
can occur. For this, consider the following scenario when adding
a new session via SSL_CTX_add_session:
- The session cache is full
(SSL_CTX_sess_number(ctx) > SSL_CTX_sess_get_cache_size(ctx))
- Only the session cache has a reference to ctx->session_cache_tail
(that is, ctx->session_cache_tail->references == 1)
Since the cache is full, remove_session_lock is called to remove
ctx->session_cache_tail from the cache. That is, it
SSL_SESSION_free()s the session, which free()s the data. Afterwards,
the free()d session is passed to the remove_session_cb. If the callback
accesses the session's data, we have a use after free.
The free before calling the callback behavior was introduced in
commit e4612d02c53cccd24fa97b08fc01250d1238cca1 ("Remove sessions
from external cache, even if internal cache not used.").
CLA: trivial
Reviewed-by: Rich Salz <rsalz@openssl.org> Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6222)
Rich Salz [Tue, 5 Jun 2018 15:17:59 +0000 (11:17 -0400)]
Improve wording
Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6413)
Rich Salz [Sat, 2 Jun 2018 18:57:34 +0000 (14:57 -0400)]
Make OS/X more explicit, to avoid questions
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6404)
(cherry picked from commit 886c2e614fc1e78e658122bf6f6bccdd7dd23857)
Ken Goldman [Sat, 2 Jun 2018 20:17:32 +0000 (16:17 -0400)]
Document failure return for ECDSA_SIG_new
ECDSA_SIG_new() returns NULL on error.
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com> Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6398)
Richard Levitte [Thu, 31 May 2018 09:12:34 +0000 (11:12 +0200)]
ENGINE_pkey_asn1_find_str(): don't assume an engine implements ASN1 method
Just because an engine implements algorithm methods, that doesn't mean
it also implements the ASN1 method. Therefore, be careful when looking
for an ASN1 method among all engines, don't try to use one that doesn't
exist.
Fixes #6381
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6383)
Richard Levitte [Thu, 31 May 2018 04:51:25 +0000 (06:51 +0200)]
apps: when the 'compat' nameopt has been set, leave it be
XN_FLAG_COMPAT has a unique property, its zero for value. This means
it needs special treatment; if it has been set (which can only be
determined indirectly) and set alone (*), no other flags should be
set.
(*) if any other nameopt flag has been set by the user, compatibility
mode is blown away.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6382)
Mingtao Yang [Fri, 9 Feb 2018 18:23:18 +0000 (10:23 -0800)]
Add APIs for custom X509_LOOKUP_METHOD creation
OpenSSL 1.1.0 made the X509_LOOKUP_METHOD structure opaque, so
applications that were previously able to define a custom lookup method
are not able to be ported.
This commit adds getters and setters for each of the current fields of
X509_LOOKUP_METHOD, along with getters and setters on several associated
opaque types (such as X509_LOOKUP and X509_OBJECT).
Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6152)
David Benjamin [Sun, 20 May 2018 21:24:30 +0000 (17:24 -0400)]
Save and restore the Windows error around TlsGetValue.
TlsGetValue clears the last error even on success, so that callers may
distinguish it successfully returning NULL or failing. This error-mangling
behavior interferes with the caller's use of GetLastError. In particular
SSL_get_error queries the error queue to determine whether the caller should
look at the OS's errors. To avoid destroying state, save and restore the
Windows error.
Viktor Dukhovni [Tue, 22 May 2018 18:46:02 +0000 (14:46 -0400)]
Skip CN DNS name constraint checks when not needed
Only check the CN against DNS name contraints if the
`X509_CHECK_FLAG_NEVER_CHECK_SUBJECT` flag is not set, and either the
certificate has no DNS subject alternative names or the
`X509_CHECK_FLAG_ALWAYS_CHECK_SUBJECT` flag is set.
Add pertinent documentation, and touch up some stale text about
name checks and DANE.
Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Tim Hudson <tjh@openssl.org>
Viktor Dukhovni [Fri, 18 May 2018 13:09:51 +0000 (09:09 -0400)]
Limit scope of CN name constraints
Don't apply DNS name constraints to the subject CN when there's a
least one DNS-ID subjectAlternativeName.
Don't apply DNS name constraints to subject CN's that are sufficiently
unlike DNS names. Checked name must have at least two labels, with
all labels non-empty, no trailing '.' and all hyphens must be
internal in each label. In addition to the usual LDH characters,
we also allow "_", since some sites use these for hostnames despite
all the standards.
Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Tim Hudson <tjh@openssl.org>
Richard Levitte [Thu, 17 May 2018 07:53:14 +0000 (09:53 +0200)]
Restore check of |*xn| against |name| in X509_NAME_set
A previous change of this function introduced a fragility when the
destination happens to be the same as the source. Such alias isn't
recommended, but could still happen, for example in this kind of code:
X509_NAME *subject = X509_get_issuer_name(x);
/* ... some code passes ... */
X509_set_issuer_name(x, subject);
Fixes #4710
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6280)
Bernd Edlinger [Thu, 29 Mar 2018 09:27:29 +0000 (11:27 +0200)]
Fix a possible crash in BN_from_montgomery_word
Thanks to Darovskikh Andrei for for reporting this issue.
Fixes: #5785 Fixes: #6302
Cherry-picked from f91e026e3832 (without test/bntest.c)
Reviewed-by: Rich Salz <rsalz@openssl.org> Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/6310)
apps/speed: fix possible OOB access in some EC arrays
because there are actually 17 curves defined, but only 16 are plugged for
ecdsa test.
Deduce array size using OSSL_NELEM and so remove various magic numbers,
which required some declarations moving.
Implement OPT_PAIR list search without a null-ending element.
Fix some comparison between signed and unsigned integer expressions.
Matt Caswell [Wed, 16 May 2018 08:58:27 +0000 (09:58 +0100)]
Make BN_GF2m_mod_arr more constant time
Experiments have shown that the lookup table used by BN_GF2m_mod_arr
introduces sufficient timing signal to recover the private key for an
attacker with access to cache timing information on the victim's host.
This only affects binary curves (which are less frequently used).
No CVE is considered necessary for this issue.
The fix is to replace the lookup table with an on-the-fly calculation of
the value from the table instead, which can be performed in constant time.
Thanks to Youngjoo Shin for reporting this issue.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6270)
Richard Levitte [Wed, 16 May 2018 09:12:21 +0000 (11:12 +0200)]
When producing man-pages, ensure NAME section is one line only
There are *roff parsers that are strict about the NAME section being
one line only. The man(7) on Debian GNU/Linux suggests that this is
appropriate, so we compensate our multi-line NAME sections by fixing
the *roff output.
Noted by Eric S. Raymond
Related to #6264
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6268)
Matt Caswell [Wed, 2 May 2018 15:07:13 +0000 (16:07 +0100)]
Mark DTLS records as read when we have finished with them
The TLS code marks records as read when its finished using a record. The DTLS code did
not do that. However SSL_has_pending() relies on it. So we should make DTLS consistent.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6160)
Richard Levitte [Wed, 2 May 2018 12:28:53 +0000 (14:28 +0200)]
UI console: Restore tty settings, do not force ECHO after prompt
The Console UI method always set echo on after prompting without
echo. However, echo might not have been on originally, so just
restore the original TTY settings.
Fixes #2373
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6157)
Reviewed-by: Andy Polyakov <appro@openssl.org> Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6237)
Matt Caswell [Fri, 11 May 2018 09:28:47 +0000 (10:28 +0100)]
Don't memcpy the contents of an empty fragment
In DTLS if we have buffered a fragment for a zero length message (e.g.
ServerHelloDone) then, when we unbuffered the fragment, we were attempting
to memcpy the contents of the fragment which is zero length and a NULL
pointer. This is undefined behaviour. We should check first whether we
have a zero length fragment.
Fixes a travis issue.
[extended tests]
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6224)
Matt Caswell [Thu, 3 May 2018 15:00:51 +0000 (16:00 +0100)]
Keep the DTLS timer running after the end of the handshake if appropriate
During a full handshake the server is the last one to "speak". The timer
should continue to run until we know that the client has received our last
flight (e.g. because we receive some application data).
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6196)
projects / openssl / log