The implementation largely follows that of sntrup761x25519-sha512.
Most of the work was done by Sahana with the help of Claude,
Pavol provided fixes to match specs and did a final clean up.
Co-Authored-By: Sahana Prasad <sahana@redhat.com>
Co-Authored-By: Pavol Žáčik <pzacik@redhat.com>
Co-Authored-By: Claude <noreply@anthropic.com>
Signed-off-by: Pavol Žáčik <pzacik@redhat.com>
Reviewed-by: Jakub Jelen <jjelen@redhat.com>
Add a cmake option to enable hermetic-usr, i.e., use of config files in /usr/.
If turned on, GLOBAL_*_CONFIG is prepended with /usr/ and defined as
USR_GLOBAL_*_CONFIG. Config lookup follows this path GLOBAL_*_CONFIG ->
USR_GLOBAL_*_CONFIG.
Introduce a ssh_config_parse primitive. This avoids convoluted checks for file
presence (without modifing the behaviour of ssh_config_parse_file) and allows
marking whether the config is global at the call site.
Signed-off-by: Lucas Mulling <lucas.mulling@suse.com>
Reviewed-by: Jakub Jelen <jjelen@redhat.com>
Some architectures (esp32) might not have this API.
Fixes: #263
Signed-off-by: Jakub Jelen <jjelen@redhat.com>
Reviewed-by: Andreas Schneider <asn@cryptomilk.org>
.. to satisfy restricted environment or fuzzers
We are encountering weird issues in the oss-fuzz that the file disappears during
coverage build so I assume some corpus sneaked in, that contains some commands
that end up being executed as part of the coverage run causing it randomly
failing.
The solution I propose is to build fuzzers without ability to call arbitrary
commands on the filesystem (such as `rm -rf /`) as this is not the point the
fuzzers should be testing.
This is controlled by the WITH_EXEC CMake option (enabled by default).
https://github.com/google/oss-fuzz/issues/10136
Signed-off-by: Jakub Jelen <jjelen@redhat.com>
Reviewed-by: Sahana Prasad <sahana@redhat.com>
Reviewed-by: Eshan Kelkar <eshankelkar@galorithm.com>
The engine API in OpenSSL 3.0 is deprecated so we are in the progress of working
on a PKCS#11 provider for OpenSSL. This commit introduces a conditional build
with the pkcs11-provider support (instead of engines) with all the changes
required for the provider to work with existing code and tests.
The CI modification is only temporary before we will have the real package in
Fedora or somewhere to use.
Signed-off-by: Jakub Jelen <jjelen@redhat.com>
Reviewed-by: Anderson Toshiyuki Sasaki <ansasaki@redhat.com>
Reviewed-by: Norbert Pocs <npocs@redhat.com>
Solving issue #110. The original work is at !231
Some changes were needed because the newly added features in master through time
Signed-off-by: Mohammad Shehar Yaar Tausif <sheharyaar48@gmail.com>
Signed-off-by: Norbert Pocs <npocs@redhat.com>
Reviewed-by: Jakub Jelen <jjelen@redhat.com>
Removing support for older openssl versions than 1.1.1 makes some functions
implicitly included; we do not have to check the availability of these
functions.
Signed-off-by: Norbert Pocs <npocs@redhat.com>
Reviewed-by: Jakub Jelen <jjelen@redhat.com>
Reviewed-by: Andreas Schneider <asn@cryptomilk.org>
The new API does not provide EVP_KDF_CTX_new_id function, insted
it works with EVP_KDF_CTX_new and fetching the algorithm.
Adding a check for both to make it work with the new API too.
Signed-off-by: Norbert Pocs <npocs@redhat.com>
Reviewed-by: Jakub Jelen <jjelen@redhat.com>
OpenSSL 1.0.1 has support for CTR & GCM modes so the checks here are no
longer needed. This allows for a bunch of additional cleanup of the old
code.
As for old MacOS versions etc, LibreSSL is a kind of compatibility layer
there but things already don't work anyway with that, so it doesn't
break anything that isn't already broken. OpenSSL is needed on MacOS
separately anyway (like installed with Homebrew).
Signed-off-by: Dirkjan Bussink <d.bussink@gmail.com>
Reviewed-by: Jakub Jelen <jjelen@redhat.com>
CRYPTO_THREADID_set_callback is available since 1.0.1 which is the
oldest supported version. This means the check and compatibility code
can be removed.
Signed-off-by: Dirkjan Bussink <d.bussink@gmail.com>
Reviewed-by: Jakub Jelen <jjelen@redhat.com>
Since OpenSSL 1.0.1 is the minimum version, this function is always
available so no compatibility check is needed anymore.
Signed-off-by: Dirkjan Bussink <d.bussink@gmail.com>
Reviewed-by: Jakub Jelen <jjelen@redhat.com>
Libgcrypt has supported ChaCha20 and Poly1305 since 1.7.0 version and
provides fast assembler implementations.
v3:
- initialize pointers to NULL
- use 'bool' for chacha20_poly1305_keysched.initialized
- pass error codes from libgcrypt calls to variable
- add SSH_LOG on error paths
v2:
- use braces for one-line blocks
- use UNUSED_PARAM/UNUSED_VAR instead of cast to void
- use calloc instead of malloc+memset
Signed-off-by: Jussi Kivilinna <jussi.kivilinna@iki.fi>
Reviewed-by: Jakub Jelen <jjelen@redhat.com>
Reviewed-by: Andreas Schneider <asn@cryptomilk.org>
Use the newer APIs EVP_DigestSign{Init}() and EVP_DigestVerify{Init}()
to generate and verify signatures instead of the older EVP_Sign{Init,
Update, Final} and EVP_Verify{Init, Update, Final} if supported.
Also use the single shot signature/verification if supported as all the
input is provided at once.
This is a preparation to use Ed25519 implementation from OpenSSL.
Signed-off-by: Anderson Toshiyuki Sasaki <ansasaki@redhat.com>
Reviewed-by: Jakub Jelen <jjelen@redhat.com>
LibreSSL does not support FIPS mode, check for FIPS_mode() in
ConfigureChecks.cmake.
Signed-off-by: Stefan Strogin <steils@gentoo.org>
Reviewed-by: Jakub Jelen <jjelen@redhat.com>
This allows the global client (ssh_session) configuration file path to
be set in configuration time by defining GLOBAL_CLIENT_CONFIG when
calling cmake. If it is not defined, the default path is set as
"/etc/ssh/ssh_config".
usage example:
$ cmake -DGLOBAL_CLIENT_CONFIG=/etc/my/custom/path ..
Signed-off-by: Anderson Toshiyuki Sasaki <ansasaki@redhat.com>
Reviewed-by: Andreas Schneider <asn@cryptomilk.org>
This allows the global bind configuration file path to be set in
configuration time by defining the GLOBAL_BIND_CONFIG when calling
cmake. If no value is defined, the default path is set as
"/etc/ssh/libssh_server_config".
usage example:
$ cmake -DGLOBAL_BIND_CONFIG=/etc/my/custom/path ..
Signed-off-by: Anderson Toshiyuki Sasaki <ansasaki@redhat.com>
Reviewed-by: Andreas Schneider <asn@cryptomilk.org>
Cleanup the KDF function to use only one function per crypto backend.
Improve the KDF function to properly handle requested lenght and to
avoid unnecessarily reallocating buffers.
In OpenSSL use the new EVP_KDF API if available.
Signed-off-by: Simo Sorce <simo@redhat.com>
Reviewed-by: Anderson Toshiyuki Sasaki <ansasaki@redhat.com>
Reviewed-by: Andreas Schneider <asn@cryptomilk.org>
Buffer (un)packing was broken on compilers that are not
gcc-compatible since the checks for an argument count of
-1 have been removed from ssh_buffer_(un)pack(). This
fix no longer uses GCC extensions for the __VA_NARG__
macro, but only plain C99.
Note: The macro can no longer count empty argument lists
(results in compile error) which was not needed anyway.
Signed-off-by: Tilo Eckert <tilo.eckert@flam.de>
Reviewed-by: Andreas Schneider <asn@cryptomilk.org>
Reviewed-by: Jakub Jelen <jjelen@redhat.com>
This completely removes the tests filter code from torture.c and calls
cmocka_set_test_filter() instead, if available. The checks for required
libraries, headers, and the availability of cmocka_set_test_filter()
were added to the cmake configuration.
Signed-off-by: Anderson Toshiyuki Sasaki <ansasaki@redhat.com>
Reviewed-by: Andreas Schneider <asn@cryptomilk.org>
Valgrind detects many uninitialized memory false positives from
libcrypto's AES-NI internals. Roll back to software AES when running
tests.
Signed-off-by: Aris Adamantiadis <aris@0xbadc0de.be>
Reviewed-by: Jakub Jelen <jjelen@redhat.com>
Reviewed-by: Andreas Schneider <asn@cryptomilk.org>
The commit also propares the internals throughout the code base
for the inclusion of a new AEAD cipher, because previously, the
source code counted only with chacha20-poly1305 cipher, which
is very specific in many cases.
The SSH_HMAC_AEAD_GCM mac algorithm is not actually used, but the name
needed to be defined so we can match in the algorithms selection per
OpenSSH specification (MACs are ignored in case GCM is select as a cipher [1]).
If the provided OpenSSL does not provide EVP_aes_128_gcm() function,
the AES-GCM ciphers will not be compiled in.
[1] https://cvsweb.openbsd.org/src/usr.bin/ssh/PROTOCOL.chacha20poly1305?annotate=HEAD
Signed-off-by: Jakub Jelen <jjelen@redhat.com>
Reviewed-by: Andreas Schneider <asn@cryptomilk.org>
Fix the build for mbedTLS:
* set HAVE_CHACHA for non-mbedTLS builds
* only compile chachapoly.c when HAVE_CHACHA
* use empty CHACHA20 in src/kex.c unless HAVE_CHACHA
Signed-off-by: Jon Simons <jon@jonsimons.org>
Reviewed-by: Andreas Schneider <asn@cryptomilk.org>
Useful on Windows platforms where SecureZeroMemory is available.
Signed-off-by: Alberto Aguirre <albaguirre@gmail.com>
Reviewed-by: Andreas Schneider <asn@cryptomilk.org>
