updated auto-generated files

Signed-off-by: Nikos Mavrogiannopoulos <nmav@redhat.com>

gnutls_priority_init2,gnutls_set_default_priority_append: introduced

This allows enhancing the default priority with additional
options, allowing an application to introduce stricter (or weaker)
settings without requiring it to override all settings.

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

doc update [ci skip]

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

doc update [ci skip]

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

configure: Fix progress message for --enable-tls13-support

Signed-off-by: Andreas Metzler <ametzler@bebt.de>

tests: tls-fuzzer-alpn: operate on random port

This allows parallel run of the test with other tlsfuzzer tests.

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

configure: added option --enable-tls13-support

The new option enables TLS1.3 draft-28 support unconditionally.
Updated the test suite to run when TLS1.3 is enabled by default,
and added a CI run with TLS1.3 enabled.

Resolves #424

Signed-off-by: Nikos Mavrogiannopoulos <nmav@redhat.com>

_gnutls_figure_common_ciphersuite: apply rfc7919 requirements only under TLS1.2

Under TLS1.3 there is no requirement to return insufficient security depending
on the FFDHE group negotiation.

Signed-off-by: Nikos Mavrogiannopoulos <nmav@redhat.com>

supported_versions: do not parse in server side when TLS1.3 is disabled

This allows a server to negotiate older versions using the previous TLS
negotiation scheme.

Signed-off-by: Nikos Mavrogiannopoulos <nmav@redhat.com>

protocols: bumped TLS1.3 protocol to draft-28

Resolves #506

Signed-off-by: Nikos Mavrogiannopoulos <nmav@redhat.com>

tests: mini-record-timing: avoid warning for too large stack usage

Signed-off-by: Nikos Mavrogiannopoulos <nmav@redhat.com>

Merge branch 'tmp-fix-rsae-negotiation' into 'master'

sign_supports_cert_pk_algorithm: corrected check for RSAE-PSS

Closes #500

See merge request gnutls/gnutls!693

tlsfuzzer: updated to include RSA and RSA-PSS related tests

Signed-off-by: Nikos Mavrogiannopoulos <nmav@redhat.com>

sign_supports_cert_pk_algorithm: corrected check for RSAE-PSS

If the signature algorithm sets the `cert_pk` field, ignore the
`pk` field completely. Not doing that would make the RSAE signature
algorithms with RSA-PSS certificates which is against the intended
use of `cert_pk`.

Resolves #500

Signed-off-by: Nikos Mavrogiannopoulos <nmav@redhat.com>

tlsproxy: included but not as submodule

This allows updating the example when necessary within the repository
and reduces the amount of external dependencies for CI.

Signed-off-by: Nikos Mavrogiannopoulos <nmav@redhat.com>

tlsproxy: removed submodule

Signed-off-by: Nikos Mavrogiannopoulos <nmav@redhat.com>

tests: introduced tests about crypto API failures on illegal use

This ensures that any mistakes in using the crypto API are propagated
to the higher level calls, or result to an abort().

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

gnutls_aead_cipher_encryptv: eliminate signed/unsigned warnings under x86

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

accelerated: error on the cases where the nettle API would have errored

This ensures that illegal uses of the API would be propagated to
the higher levels.

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

gnutls_cipher_add_auth: propagate error codes

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

Merge branch 'tmp-32-bit-time_t' into 'master'

tests: verify that certtool will fail with a 32-bit time_t

Closes #370

See merge request gnutls/gnutls!691

certtool: properly print an int64_t value

Also included the gnulib inttype module for portability.

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

certtool: print information on time_t restrictions on failure

This informs the user of the tool why dates after 2038 cannot
be expressed on systems with a 32-bit time_t.

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

tests: verify that certtool operates as expected with dates after 2038

That is, whether it works with a time_t of 64-bit size, and fails
with a time_t of 32-bit size.

Resolves #370

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

tests: check explicitly the size of time_t

Previously we were disabling the 2038 tests on 32-bit systems,
but there can be 32-bit systems with a 64-bit time_t. Ensure
that we run the right tests.

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

tests: better guarding of variable SKIP_DATEFUDGE_CHECK

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

Merge branch 'tmp-post-handshake-auth' into 'master'

post-handshake authentication improvements

Closes #490 and #489

See merge request gnutls/gnutls!687

tests: ignore PIPE signal on TLS1.3-related tests

This was inadvertently omitted and that could cause unexpected
issues when one of the peers would close the connection earlier
than expected.

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

tests: check for GNUTLS_E_GOT_APPLICATION_DATA on post-handshake auth

That is, check whether GNUTLS_E_GOT_APPLICATION_DATA is received as
documented, and whether post-handshake auth can complete while this
is being sent.

Resolves #490

Signed-off-by: Nikos Mavrogiannopoulos <nmav@redhat.com>

post-handshake: return GNUTLS_E_GOT_APPLICATION_DATA as documented to

Relates #490

Signed-off-by: Nikos Mavrogiannopoulos <nmav@redhat.com>

tests: introduced test for post-handshake auth + PSK

Relates #489

Signed-off-by: Nikos Mavrogiannopoulos <nmav@redhat.com>

tls13 handshake: allow certificate messages after handshake

This allows post-handshake authentication even when PSK
is negotiated.

Resolves #489

Signed-off-by: Nikos Mavrogiannopoulos <nmav@redhat.com>

gnutls_session_get_flags: introduced GNUTLS_SFLAGS_POST_HANDSHAKE_AUTH

This allows a server application to detect whether the client
would support post handshake authentication or not without initiating
via gnutls_reauth().

Signed-off-by: Nikos Mavrogiannopoulos <nmav@redhat.com>

gnutls-serv: make --disable-client-cert and --require-client-cert options incompatible

That is refuse to run when both options are specified.

Resolves #502

Signed-off-by: Nikos Mavrogiannopoulos <nmav@redhat.com>

tests: verify whether GNUTLS_TLS_VERSION_MAX is negotiated on default mode

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

Merge branch 'tmp-ci-cleanups' into 'master'

Fixes + cleanups for .gitlab-ci.yml

See merge request gnutls/gnutls!690

Fixes + cleanups for .gitlab-ci.yml

p11tool: remove duplicate branch

The GNUTLS_PKCS11_OBJ_ATTR_MATCH and GNUTLS_PKCS11_OBJ_ATTR_ALL
attributes are the same, so there is no need to handle them
separately.

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

Add strdup-posix gnulib module

Some files in gl/tests won't build in environments without
strdup(), e.g. MinGW on Debian. The gnulib docs advise to
explicitly add the module.

Signed-off-by: Tim Rühsen <tim.ruehsen@gmx.de>

testcompat-tls13-openssl: fix openssl interactions

 * Do not require certificate validation on tests where no certificate is sent
 * Rekey test performs data transfer after re-key

This introduces a dependency on the expect package for testing, and
updates openssl to address an issue in post-handshake auth interop
testing.

Resolves #488

Signed-off-by: Nikos Mavrogiannopoulos <nmav@redhat.com>

gnutls-serv: when post-handshake auth is asked; require a certificate

This allows testing post-handshake authentication using gnutls-serv.

Signed-off-by: Nikos Mavrogiannopoulos <nmav@redhat.com>

key update: corrected generation of keys

Resolves #485

Signed-off-by: Nikos Mavrogiannopoulos <nmav@redhat.com>

gnutls-cli: wait for all server data prior to closing connection

This cleans-up the existing code which was disfunctional and
allows detecting errors which happen after we transmit data
to the server.

Relates #485

Signed-off-by: Nikos Mavrogiannopoulos <nmav@redhat.com>

.gitignore: added new test executables

Signed-off-by: Nikos Mavrogiannopoulos <nmav@redhat.com>

tests: eliminated compiler warnings

Signed-off-by: Nikos Mavrogiannopoulos <nmav@redhat.com>

Update .gitignore files according to bootstrap

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

src: fix regenerating autogen files if builddir = srcdir

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

Merge branch 'timespec-overflow2' into 'master'

avoid more possible overflows when substracting timespecs

See merge request gnutls/gnutls!685

convert from milliseconds to timespec without loop

Signed-off-by: Rolf Eike Beer <eike@sf-mail.de>

use timespec_sub_ms() instead of open coding it

Signed-off-by: Rolf Eike Beer <eike@sf-mail.de>

avoid overflow when substracting timespecs if rdtsc is not available

This may still overflow on platforms where unsigned long is 32 bit (e.g. 32 bit
Un*x, any Windows) when the delta is more than 4 seconds.

Signed-off-by: Rolf Eike Beer <eike@sf-mail.de>

lib/nettle/gost: support building with mini-nettle/mini-gmp

Do not depend directly on gmp.h.

Closes: #497
Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

avoid rounding errors and overflows when substracting timespecs

The current Unix time will cause overflows if multiplied with 1000, which could
lead to rounding errors afterwards. Do the substractions first so all numbers
stay small enough to fit into unsigned ints.

Signed-off-by: Rolf Eike Beer <eike@sf-mail.de>

Merge branch 'tmp-fix-fips-generation' into 'master'

wrap_nettle_pk_generate_keys: retry on provable key generation

Closes #283

See merge request gnutls/gnutls!681

wrap_nettle_pk_generate_keys: retry on provable key generation

This resolves issue with occasional failures under RSA key generation
in FIPS140-2 mode.

Resolves #283

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

Merge branch 'auto-sync-translations' into 'master'

Let ./bootstrap sync from translationproject.org

Closes #470

See merge request gnutls/gnutls!678

Let ./bootstrap sync from translationproject.org

This makes manual updating of the translations obsolete.
From now on, builds and tarballs will always have the latest translations
included.

We should not forget to inform translationproject.org to update the
translations before a release. How to do that is described at
https://translationproject.org/html/maintainers.html (6. Announcing).

gnutls_session_get_desc: fixed desc printing of custom groups

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

doc update [ci skip]

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

aarch64: use getauxval() if available to discover cpu caps

This improves CPU detection by avoiding the parsing of
of a human-readable file and allows operation under debian
multilib qemu setup.

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

.gitlab-ci.yml: no need for submodule update on cross-builds

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

.gitlab-ci.yml: use qemu for aarch64 testing

This eliminates the need (and costs) to maintain a separate baremetal
system.

Signed-off-by: Nikos Mavrogiannopoulos <nmav@redhat.com>

.gitlab-ci.yml: corrected typo

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

.gitlab-ci.yml: skip submodule initialization when not necessary

This prevents unnecessary download of submodules on CI.

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

.gitlab-ci.yml: updated x86 CI builds with better datefudge detection

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

.gitlab-ci.yml: debian stretch build replaced by buster

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

doc update [ci skip]

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

doc update [ci skip]

Signed-off-by: Nikos Mavrogiannopoulos <nmav@gnutls.org>

Merge branch 'gost-no-tls' into 'master'

GOST certificates/PKCS#7/PKCS#12 support

See merge request gnutls/gnutls!654

tests: add PKCS#12 test script for GOST 28147-89-encrypted files

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

certtool: honour --hash option when generating PKCS#12 files

Use algorithm specified with --hash option when generating MAC for
PKCS#12 file, allowing user to select algorithms other than SHA-1.

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

tests: expand pkcs7 test to also check GOST files

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

test: test GOST keys import/export

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

certtool: ask if certificate will be used for data encryption

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

tests: add common gost certificates for tests

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

Support key matching with GOST keys

GOST keys do not support signing non-GOST hashes, so use correct digest
algorithm when verifying that GOST public and private keys match.

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

Add generated GOST credentials for tests

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

Use GOST R 34.11-94 when generating key for PKCS data to be encrypted with GOST 28147-89

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

certtool: support generating GOST-encrypted PKCS#8/12 files

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

Add gost certificates to chainverify tests

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

Expand x509 sign/verify test with GOST algorithms

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

oids: expand to include GOST digests/signatures

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

tests: privkey-keygen: adapt to support GOST algorithms

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

Support GOST private keys generation

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

certtool: support dumping GOST private key information

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

Add several DN entry definitions used by qualified GOST signatures

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

certool: export GOST privkeys only in PKCS#8 format

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

Add support for PKCS12 files using GOST MAC

Local PKCS12-based standard derives from RFC 7292 (PKCS #12) in using
PBKDF2 to generate MAC key rather than using PKCS12 scheme.

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

Add support for PBES2/PBKDF2 using GOST algorithms

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

Support PKCS#12 key derivation with GOST digests

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

Add support for importing/exporting GOST private keys

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

Support importing/exporting X.509 GOST public keys

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

Add ASN.1 definitions for GOST keys

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

nettle: add support for GOST 34.10 public keys

There is no support for GOST public keys derivation, as it is used only
for TLS or PKCS#7 with encrypted content.

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

Add few functions to support basic operations with GOST public keys

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

Add declarations for GOST R 34.10 signatures

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

Define GOST R 34.10 curves

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

Add declarations to support GOST public keys

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

Add support for I/O of little-endian MPI

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>

nettle: add support for unsigned LE MPIs

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>