An empty AuthorityKeyIdentifier SEQUENCE must parse. A [0] tag without
a length must still be rejected; that case already passed before the
fix and guards against loosening the check too much.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Christopher Müller <chris@mechatronikmonkey.de>
All fields of AuthorityKeyIdentifier are OPTIONAL, so an empty SEQUENCE
(30 00) is syntactically valid. It was rejected with
MBEDTLS_ERR_X509_INVALID_EXTENSIONS + MBEDTLS_ERR_ASN1_OUT_OF_DATA,
because mbedtls_asn1_get_tag() reports OUT_OF_DATA rather than
UNEXPECTED_TAG when there is nothing left to parse.
Check for the empty sequence explicitly instead of treating OUT_OF_DATA
as an absent keyIdentifier: OUT_OF_DATA is also returned for a [0] tag
with a missing or truncated length, which must still be rejected.
Fixes#10253.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Christopher Müller <chris@mechatronikmonkey.de>
- mbedtls_endpoint_sanity() is only used for TLS 1.2 tests, where RSA
signatures are limited to PKCS#1 v1.5. Moreover the test does not
perform a full handshake, so only being able to hold an RSA public key
is required here: PSA_WANT_KEY_TYPE_RSA_PUBLIC_KEY.
- move_handshake_to_state() is used for both TLS 1.2 and TLS 1.3 tests
and the RSA capabilities it needs depend on the handshake state being
reached, so the RSA dependency is set on test data instead.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
- Mbed TLS only implements TLS 1.2 and DTLS 1.2 (not 1.0/1.1) and there
RSA signatures are limited to PKCS#1 v1.5, so replace
PSA_HAVE_ALG_SOME_RSA_SIGN with PSA_HAVE_ALG_RSA_PKCS1V15_SIGN.
- move_handshake_to_state() sets options.pk_alg = MBEDTLS_PK_RSA, so some
minimal support for RSA keys is always needed: at least an RSA public
key. Stricter requirements only apply depending on the protocol version
(TLS 1.2 vs 1.3) and on the handshake state reached by the test (ex:
initial phase vs ending phase), so they are set per test case.
- Test cases already requiring MBEDTLS_X509_RSASSA_PSS_SUPPORT don't need
PSA_WANT_ALG_RSA_PSS too, since the former depends on the latter.
- Add a comment to the handshake_psk_cipher test cases to explain why a
test which is ideally about PSK has requirements on RSA and EC keys.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
PKCS7 supports ECC as well so the test function should not be guarded or
limited to RSA build symbols. Those guards are now set on the test data
as required by each test itself. This allows for test extension to also
include ECC testing (not present now).
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
pkcs7_verify() is about to lose its blanket RSA guard, so each test case
must declare what it actually needs:
- cases that really validate an RSA signature require
PSA_HAVE_ALG_RSA_PKCS1V15_VERIFY. This includes the cases which used
to pass accidentally because the certificate is invalid and the code
never gets to verify the signature: using the same dependency as the
valid cases is more uniform;
- cases which are expected to fail before reaching signature
verification only require PSA_WANT_KEY_TYPE_RSA_PUBLIC_KEY.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
- Tests that only parse a certificate/CRL/CSR without verifying any
signature only need to be able to hold an RSA public key, so replace
PSA_HAVE_ALG_SOME_RSA_VERIFY with PSA_WANT_KEY_TYPE_RSA_PUBLIC_KEY.
PSA_HAVE_ALG_SOME_RSA_VERIFY is kept in x509_verify tests, which do
verify signatures.
- x509_parse_rsassa_pss_params tests don't need any explicit RSA
dependency: the test function already depends on
MBEDTLS_X509_RSASSA_PSS_SUPPORT which itself depends on
PSA_WANT_KEY_TYPE_RSA_KEY_PAIR_BASIC, i.e. something "stronger" than
PSA_WANT_KEY_TYPE_RSA_PUBLIC_KEY.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
MBEDTLS_X509_RSASSA_PSS_SUPPORT depends on PSA_WANT_ALG_RSA_PSS, so add
the corresponding reverse dependency and include PSA_WANT_ALG_RSA_PSS in
the 'pkalgs' ComplementaryDomain in order to also test builds where
RSA-PSS is disabled.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
OAEP is for RSA encryption and therefore is not the correct check when
we want to test for signatures with PSS.
Additionally, we can use RSA PSS when we only have public key support
(not key pairs) so check for PSA_WANT_KEY_TYPE_RSA_PUBLIC_KEY (which is
auto-enabled when we have key pairs enabled anyway).
Signed-off-by: David Horstmann <david.horstmann@arm.com>
This is useful for debugging purposes because, for example, the user
only want to build a specific suite and/or only run it instead of
building and testing everything for all the jobs.
For this scope the custom command can be a bash script that's executed
instead of 'make' + 'make test'.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
Following recent removal for '_reference' and '_use_psa' functions we
ended up having 2 very similar test components named
'component_test_psa_crypto_config_accel_hash', one using 'default'
configuration and the other one using the 'full' one.
Let's keep the 'full' configuration version and get rid of the 'default'
case.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
'ssl-opt.sh' and 'compat.sh' in accel components were mostly useful back
when the accelerated components were to be compared against the reference
ones to ensure proper test coverage. Now that the reference components
have been removed it doesn't make sense to keep 'ssl-opt.sh' and
'compat.sh' around because they take a lot of simulation time for almost
no benefit. From TLS point of view we can rely on normal 'test_suite_'
testing to be sure that the crypto mechanism works as expected.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
Nowadays USE_PSA_CRYPTO is always enabled and there is no way to disable
it. It doesn't give more information to know that we're setting that
build symbol. Remove that.
While at this remove also '_use_psa' from some test component still using
it in the name for the same reason.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
These were useful in the 3.6 days when 'analyze_outcomes.py' was used to
check that driver acceleration was providing the same coverage as the
reference legacy modules. Since this check has recently been removed
from 'analyze_outcomes.py' all the '_reference' component can be
removed as they are just wasting CI time.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
