Add a test entry point for a handshake with a specially crafted context.
This can be used for defense in depths for scenarios that we think can't
happen, but might actually happen due to a bug, or due to a new feature that
is insufficiently tested. This can also be a proxy for testing aspects of
renegotiation or session resumption in circumstances that are hard to
arrange.
In this commit, use it to test a DTLS handshake with `ssl->badmac_seen`
:(actually `ssl->badmac_seen_or_in_hsfraglen` in the 3.6 branch) set to a
nonzero value. This can't happen, as far as I know (`badmac_seen` can only
be nonzero during renegotiation, not during a nominal handshake or in
session resumption). But it could happen, for example, if
`mbedtls_ssl_session_reset()` failed to reset `badmac_seen`, as was the case
in Mbed TLS 4.0.0 to 4.1.0.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Non-regression test for a bug introduced with TLS handshake defragmentation
support in Mbed TLS 3.6.3, whereby the confusion between the `badmac_seen`
meaning and the `in_hsfraglen` meaning of `ssl->badmac_seen_or_in_hsfraglen`
causes renegotiation to be processed incorrectly after receiving a message
with a bad MAC.
The bug doesn't affect Mbed TLS 4.0.0 and above, but it's still good to have
more testing.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Since Mbed TLS 4.0, this is just the low-level error code. But keep the
combination for uniformity with the rest of the file, and to reduce the
differences with the 3.6 branch.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
In an X.509 certificate, a basicConstraints extension where the cA
field (BOOLEAN) is omitted but the pathLenConstraint field (INTEGER )is
present is syntactically valid, but RFC 5280 states that CAs MUST NOT emit
such a certificate.
Historically, since XySSL 0.9 in 2008, if a basicConstraints extension
started with an INTEGER field, we interpreted that field as a cA value. This
was a deliberate change:
> Fixed x509_get_ext() to accept some rare certificates (like
> www.openssl.org:443) which have an INTEGER instead of a BOOLEAN for
> Extension::BasicConstraints::cA.
However, 18 years later, this does not seem to be relevant, and the details
seem to have been lost. Furthermore, major implementations follow RFC 5280
and interpret `SEQUENCE { INTEGER n }` with cA being set to its default
value (FALSE) and pathLenConstraint set to `n`. This means that such a
certificate with n != 0 would be interpreted as a CA certificate in Mbed TLS
but as a leaf certificate elsewhere, which is dangerous.
Since CAs are not supposed to emit such certificates, stop accepting them
altogether.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
In test cases for pathLenConstraint parsing in a basicConstraints extension
of an X.509 certificate, use the standard type for the leading cA field,
namely BOOLEAN. The test data was encoding that field as INTEGER, which was
accepted as a deviation from the standard since XySSL 0.9.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Don't allow parts of a compound (SEQUENCE or SET) to go beyond the
containing compound.
The inline CSR regression test is derived from the existing "X509 CSR
ASN.1 (OK)" DER test vector. In the extensionRequest attribute, the original
well-formed fragment is:
a029 3027 06092a864886f70d01090e 311a 3018 ...
where 31 1a is the SET containing the Extensions sequence and 30 18 is the
contained Extensions sequence. The malformed test changes only the SET
length byte, from 1a to 19:
a029 3027 06092a864886f70d01090e 3119 3018 ...
The attribute SEQUENCE length and the inner Extensions sequence length are
left unchanged. This makes the SET one byte too short for its containing
attribute, so the parser must reject it with
MBEDTLS_ERR_ASN1_LENGTH_MISMATCH instead of parsing the trailing byte as
data outside the SET.
There is no known security impact, just some risk reduction.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
When parsing the basicConstraints extension, reject junk after the SEQUENCE
inside the extension (which is probably benign), and reject a SEQUENCE that
extends beyond the extension (could be very dangerous).
Add a non-regression test where a certificate that is technically malformed,
but accepted as a leaf certificate by OpenSSL and other X.509
implementations, to be accepted as a CA certificate by Mbed TLS.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
DEBUG_C supposed to have been removed from the test dependencies, still
being there is an oversight. Removing it was the sole purpose of
3e58109fbd.
Signed-off-by: Janos Follath <janos.follath@arm.com>
The unit test framework always loads the client key as well, which
requires a different curve and a hash than the server key.
Signed-off-by: Janos Follath <janos.follath@arm.com>
Add a test case with a successful handshake for each test case that
causes the desired handshake failure, with minimal differences between
the two.
The reason is to have more assurance that the handshake is failing for
the desired reason (as opposed to not having done something correctly in
the test code).
Signed-off-by: Janos Follath <janos.follath@arm.com>
There are other issues that can fail with the same error code. Make sure
that the handshake fails exactly the way we want it to fail by analysing
the client logs.
Signed-off-by: Janos Follath <janos.follath@arm.com>
In `test_suite_debug`, test `mbedtls_debug_snprintf()`, which uses
`mbedtls_vsnprintf()` like `mbedtls_debug_print_msg()`. Do this instead of
testing `mbedtls_snprintf()`, which might be subtly different (older
Windows runtimes had slightly different behavior for vsnprintf() vs
snprintf(); TF-PSA-Crypto might pick up a different function if the
platform configuration is different in TF-PSA-Crypto and Mbed TLS).
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
The generated unit tests have the input parameters in large stack
buffers and therefore ASan doesn't notice under or overflows in them.
Copy the input parameter into a locally allocated buffer to trigger ASan
if something goes wrong.
Signed-off-by: Janos Follath <janos.follath@arm.com>
Initialize the verify_result field in mbedtls_ssl_session_free().
Previously we were just zeroising the entire session object, which would
yield a default 'success' value if the same object were reused.
Test that this initialisation is actually happening by setting
verify_result manually to zero and calling mbedtls_ssl_session_free() on
the session before checking its value.
Signed-off-by: David Horstmann <david.horstmann@arm.com>
Since we explicitly document the value 0xFFFFFFFF or -1u as representing
'result not available', we can use it as a sensible default value
without creating an API change. Use this value instead of introducing a
new verification result value.
Signed-off-by: David Horstmann <david.horstmann@arm.com>
Write a testcase to get verify_result before we have performed a
handshake and make sure that it is initialised to a failure value.
Signed-off-by: David Horstmann <david.horstmann@arm.com>
Add new field that tells if the corresponding group is supported or not
in the current build.
Test function "test_mbedtls_ssl_get_supported_group_list" is extended
to verify this new feature.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
This is already indirectly checked in 'test_mbedtls_tls_id_group_name_table'
because it's the last item of the list.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
Check provided group_name also against the value returned from
mbedtls_ssl_get_curve_name_from_tls_id().
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
- let the macro be an initializer for the array of known TLS IDs, not
a variable declarator;
- last item's group name is NULL, not an empty string
- change then name of the macro from MBEDTLS_TLS_ID_GROUP_NAME_TABLE to
MBEDTLS_SSL_IANA_TLS_GROUPS_INFO
- define a new public structure "mbedtls_ssl_iana_tls_group_info_t" to
hold each element of the table and that can be used the go over the
list from user code.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
- adjust function name to mbedtls_ssl_get_fatal_alert
- fix missing property name changes for mbedtls_ssl_context
Signed-off-by: Nico Geyso <ng@gsmk.de>
Keep MBEDTLS_PK_WRITE_C as guard only for "debug_print_pk" but let
"mbedtls_debug_print_crt" to work also when MBEDTLS_PK_WRITE_C is disabled.
In this case the only public key won't be printed, but the rest of the
certificate will be.
This commit also updates test coverage by duplicating test cases: now there
will be one case for when MBEDTLS_PK_WRITE_C is enabled and another one
for !MBEDTLS_PK_WRITE_C.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
In tf-psa-crypto "mbedtls_pk_write_pubkey_psa()" is only available when
MBEDTLS_PK_WRITE_C is defined. Therefore we need to add this guard also
in mbedtls to "debug_print_pk" (and indirectly to
"mbedtls_debug_print_crt") and the corresponding tests using it.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
Adjust dumping format of public keys following recent updates to
mbedtls_debug_print_crt() and debug_print_pk()
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
Even though the TLS RFCs do not mandate libraries to expose *Error
Alerts* (as defined in RFC8446 6.2 for TLS 1.3 and in RFC5246 7.2.2 for
TLS 1.2) to the user, there are use cases when it is handy to get the
actual last received fatal error instead of a generic one. For instance
this enables the user to differ between received fatal errors in case
`mbedtls_ssl_handshake()`, `mbedtls_ssl_handshake_step()` or
`mbedtls_ssl_read()` returned `MBEDTLS_ERR_SSL_FATAL_ALERT_MESSAGE`.
This changesets stores the last incoming fatal alert in
`mbedtls_ssl_context` and provides `mbedtls_ssl_get_alert()` as a getter
for retrieving it. Another option would be to provide a callback
mechanisms for all kinds of alerts (not only fatals) but for simplicity
I discarded this option.
Signed-off-by: Nico Geyso <ng@gsmk.de>
This replacement is either:
- "server5-rsa-signed.crt": if a generic secp256r1 EC key is enough, i.e.
any EC key is fine as it's not secp192 since this support is being
removed from TF-PSA-Crypto.
- "server11-rsa-signed.crt": if an EC key which does not belong to "suite-b"
is required. For this case "secp256r1" wouldn't be good, so we use
a "secp256k1" key.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
After some analysis search it was determined that previous test data seem
not to belong to the "framework/data_files" certificate files. Therefore
new test data has been generated from scratch.
The improvement compared to the previous situation is that comments has
been added on top of each test in order to explain how to recreate new test
data.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
The support for TLS ciphersuites without
encryption does not rely anymore on the
MBEDTLS_CIPHER_NULL_CIPHER feature of
the cipher module. Introduce a specific
config option to enable these ciphersuites
and use it instead of MBEDTLS_CIPHER_NULL_CIPHER.
Signed-off-by: Ronald Cron <ronald.cron@arm.com>
Previously these tests used values that will become PSA aliases,
and so the tests will fail once they're changed.
Signed-off-by: Felix Conway <felix.conway@arm.com>
Return a const char* instead of taking a char* as an argument.
This aligns us with the interface used in TF PSA Crypto.
Signed-off-by: Bence Szépkúti <bence.szepkuti@arm.com>
