PK restartable operations are now implemented using PSA interruptible
ones, so mbedtls_ecp_set_max_ops() can be removed in favor of
psa_interruptible_set_max_ops().
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
The SW implementation is guarded with the `MBEDTLS_PSA_BUILTIN_ALG_SHA3`
macros and not enabled when driver accelaration is set. So disabling
the `PSA_WANT` macros is not needed.
Signed-off-by: Gabor Mezei <gabor.mezei@arm.com>
For each function in `x509_oid.c`, determine where it is used and only
include it in the build if it is needed by the X.509 code. Define the
corresponding internal tables only when they are consumed by a function.
This makes Mbed TLS completely independent of the compilation option
`MBEDTLS_OID_C`. This option remains present only in sample configs for
crypto, where it must stay until TF-PSA-Crypto no longer relies on this
option.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Some code that parses or writes X.509 needs to know OID values. We provide a
convenient list. Don't remove this list from the public interface of the
library.
For user convenience, expose these values in the same header as before and
with the same name as before: `MBEDTLS_OID_xxx` in `<mbedtls/oid.h>`.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Replace the non-X.509-named error code `MBEDTLS_ERR_OID_NOT_FOUND` with
`MBEDTLS_ERR_X509_UNKNOWN_OID`, which already exists and is currently not
used for anything.
Public functions in X.509 propagate this error code, so it needs to have a
public name.
Remove the definition of `MBEDTLS_ERR_OID_NOT_FOUND` in `x509_oid.h`, then
```
git grep -l MBEDTLS_ERR_OID_NOT_FOUND | xargs perl -i -pe 's/\bMBEDTLS_ERR_OID_NOT_FOUND\b/MBEDTLS_ERR_X509_UNKNOWN_OID/g'
```
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
With the removal of MBEDTLS_SHA3_C the test cases with disabled SHA3
dependency are never executed. Adding a temporary `all.sh` component
which disabling the `PSA_WANT_ALG_SHA3_*` macros to cover
these test cases.
Signed-off-by: Gabor Mezei <gabor.mezei@arm.com>
Split `mbedtls_test_ssl_endpoint_init()` into two separate stages:
constructing the SSL configuration, and setting up an SSL session context
with that configuration.
No behavior change.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
This will allow splitting the configuration and setup stages of
`mbedtls_test_ssl_endpoint_init()`, while still checking that the value is
carried over from the configuration to the session context.
No behavior change.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
In `mbedtls_test_ssl_endpoint_init()`, don't change the SSL configuration
object (`mbedtls_ssl_config`) after setting up an SSL context by calling
`mbedtls_ssl_setup()`. This works in practice, but is officially forbidden.
No intended behavior change. The test code calls the library slightly
differently, but this shouldn't make any difference in practice. If it does
make a difference, it fixes a bug in the test code.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
There was a discrepancy between how `mbedtls_test_ssl_endpoint_init()` and
`mbedtls_test_ssl_perform_handshake()` handled client authentication:
`mbedtls_test_ssl_endpoint_init()` defaulted to
`MBEDTLS_SSL_VERIFY_REQUIRED` on both sides, whereas
`mbedtls_test_ssl_perform_handshake()` obeyed `options->srv_auth_mode` which
defaulted to no verification of the client certificate.
Make this more uniform. Now `mbedtls_test_ssl_endpoint_init()` obeys
`options->srv_auth_mode` on servers (still forcing verification on clients,
which is the library default anyway). Also, `options->srv_auth_mode` is now
enabled by default. Thus:
* Tests that call `mbedtls_test_ssl_perform_handshake()` now perform client
certificate verification, unless they disable it explicitly.
* Tests that call `mbedtls_test_ssl_endpoint_init()` on a server are
unchanged. (They would change if they were setting
`options->srv_auth_mode` explicitly, which previously was ignored, but
no test function did this.)
This means that a few test functions now perform client certificate
verification whereas they previously don't. This is harmless except in
`handshake_ciphersuite_select`, where one test case
`Handshake, select ECDH-RSA-WITH-AES-256-CBC-SHA384, opaque` fails with
client authentication because the test code doesn't deal with the weirdness
of static ECDH correctly with respect to client authentication. So keep
the previous behavior in `handshake_ciphersuite_select`, by explicitly
turning off client authentication.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Applying SSL configuration settings recorded in the
`mbedtls_test_handshake_test_options` structure to an
`mbedtls_test_ssl_endpoint` object was split between
`mbedtls_test_ssl_endpoint_init()` and `mbedtls_test_ssl_perform_handshake()`.
This was surprising, and made it harder to use
`mbedtls_test_ssl_endpoint_init()` for custom behavior. It also meant some
code duplication in `mbedtls_test_ssl_perform_handshake()`.
Move most configuration setup from `mbedtls_test_ssl_perform_handshake()` to
`mbedtls_test_ssl_endpoint_init()`. This changes the behavior in two ways:
* `mbedtls_test_ssl_endpoint_init()` now takes some options into account
that it previously ignored. This is ok because we don't set these options
in any of the existing tests.
* When calling `mbedtls_test_ssl_perform_handshake()`, some SSL configuration
settings are now set (calls to `mbedtls_ssl_conf_xxx()`) before the call
to `mbedtls_ssl_setup()` instead of after. This should be ok since it is
forbidden to change the configuration after `mbedtls_ssl_setup()`,
although the previous test code was getting away with it. This commit does
not move all configuration before `mbedtls_ssl_setup()`, that would be out
of scope of the current series of patches.
Thus there are some internal behavior changes, but they should not affect
any relevant aspect of the tests' behavior.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Split mbedtls_test_ssl_perform_connection() out of
mbedtls_test_ssl_perform_handshake().
No behavior change.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Stop calling mbedtls_test_set_step() in
mbedtls_test_ssl_perform_handshake(). This leaves the caller free to use the
test step as they wish.
No behavior change.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
This will facilitate future refactoring that breaks out code into auxiliary
functions.
No behavior change.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
The DTLS context and the queues now conveyed inside the endpoint object.
Remove the unused parameters.
No behavior change.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
This allows mbedtls_test_ssl_endpoint_init() to no longer interact with the
other endpoint.
No behavior change.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Create an auxiliary function to perform some endpoint setup that involves
both the client and the server. This is only needed for DTLS.
The code that will eventually be in this function is currently mostly in
mbedtls_test_ssl_endpoint_init(). This commit adds the new function to the
control flow; a subsequent commit will move the relevant code.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
This is a step towards making mbedtls_test_ssl_endpoint_init() and
mbedtls_test_ssl_endpoint_free() more self-contained.
No behavior change.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Link the ciphersuite list that's passed to mbedtls_ssl_conf_ciphersuites(),
and needs to survive in memory as long as the configuration object is live,
in the endpoint structure. This way it doesn't have to be a local variable
in mbedtls_test_ssl_do_handshake_with_endpoints().
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Both PKCS7 and X509 rely on PK module under the hood and the latter can
use PSA to store keys and perform operations. Therefore psa_crypto_init()
must be called before any operation can be done with PKCS7 and X509.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
The commented out tests cause crashes (in different ways) until the bug
is fixed; the first two test are passing already and are here mostly to
provide a reference point.
The bug report was using programs/x509/cert_write, but string_to_names()
is what it was really targetting, which is better for automated tests.
The strings used are a minor adapation of those from the report.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
* public/development: (23 commits)
tests: suite_x509parse: set PSA max operations in x509_verify_restart()
library: debug: remove mbedtls_debug_printf_ecdh()
library: debug: make mbedtls_debug_print_psa_ec() static
Remove call to pk_decrypt() in ssl_server2
Change hardcoded error values in ssl-opt to take in the PSA error alias
Test with GCC 15 with sloppy union initialization
Update crypto with the union initialization fixes
Mark ssl_tls12_preset_suiteb_sig_algs const
Mark ssl_tls12_preset_default_sig_algs const
Use PSA macros for the `pkalgs` domain
reverted compat-2.x.h removal from psa-transition.md
Correct ChangeLog file extension
Add ChangeLog
remove compat-2.x.h
Remove trace of secp224k1
Update submodules
Improve comments
Allow gcc-15 to be in $PATH
Enable drivers when testing with GCC 15
GCC 15: Silence -Wunterminated-string-initialization
...
Set also psa_interruptible_set_max_ops() when mbedtls_ecp_set_max_ops()
is set so that the same amount of operations will be used both if legacy
ECDSA_C or PSA is used under the hood to perform the operation.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
ssl-opt checks for specific error code values in the output, but as
MBEDTLS_ERR_ECP_IN_PROGRESS is becoming an alias of PSA_OPERATION_INCOMPLETE
then this hardcoded value will change.
Therefore allow the result to be either the old mbedtls error, or the new PSA
error, as not to break the CI.
Signed-off-by: Felix Conway <felix.conway@arm.com>
The goal of testing with GCC 15 is to validate fixes for
https://github.com/Mbed-TLS/mbedtls/issues/9814 . The bug is present in
multiple places, and some of them affect third-party drivers but not our
built-in implementation. (The bug is that driver contexts might not be
zero-initialized, but some of our built-in implementations happen not to
care about this.) Thus, enable the test drivers in the test component that
uses GCC 15, to gain the extra checks performed in the driver wrappers.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
This is a new warning in GCC 15 that our code base triggers in many places.
Silence it for the time being.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
The OID module is used by both crypto and X.509. It has moved to the
`tf-psa-crypto` subdirectory, and the sibling commit
08d8cc57dbe7be54fe3f88ecbc2729300c48d450 removes this subdirectory from the
`mbedtls` repository in order to make `tf-psa-crypto` a submodule. We want
to access the relevant parts directly from X.509 rather than go through the
crypto repository, because OID functions are only accessible as private
interfaces, and crypto doesn't know when a particular OID function is needed
in the build since it depends on X.509 configuration options.
Make a copy of the OID module and its unit tests. In a follow-up, the X.509
module will switch to consuming this copy rather than the one that went into
TF-PSA-Crypto.
Rename the files from `*oid*` to `*x509_oid*` to follow the naming
convention that submodules of X.509 are prefixed with `x509`. This also
avoids file name clashes with TF-PSA-Crypto.
Since OID is not a public interface of Mbed TLS 4.x, move the header file
into `library`.
This commit only makes the files available. Subsequent commits will take
care of making these files used in the build.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Parsing of CRT files with message's hash alg different from the MGF1 was
allowed in the past, but now it fails. So we need to move/adapt tests
relying on this feature, from a "verify" scope to a "parse" one.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
Import component_test_platform_get_entropy_alt() from its counterpart
in TF-PSA-Crypto repo.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
Use the proper Clang C++ compiler to build C++ code otherwise the C
compiler will fail because std::cout() is unknown in
"cpp_dummy_build.cpp".
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
Use alternative implementation of mbedtls_platform_get_entropy() since
the default one lives in "platform.c" and that one is excluded in
this test component.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
By default C++ code would be compiled with GNU while C with Clang and
this can create problems at link time. In order to prevent this we
use Clang for both.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
This is necessary to let entropy being able to gather entropy data from
the native platform source.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
We were not making enough room. We want to move everything from the
place where we are going to insert the new record.
This was not causing failures because the code does not look at the
content after the inserted record, because it correctly returns an error
when seeing the inserted record. But as a matter on principle, the test
code should be doing what it says: just insert a new record but leave
a valid fragment after it.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
This one is overly tight: TLS 1.3 actually only depends on
ChachaPoly || (AES && (GCM || CCM))
Furthermore, this should really be reflected in check_config.h.
Individual test cases should be able to just request PROTO_TLS1_3 and
know that there is ciphersuite that works.
However, resolving that seems out of scope for this PR. (It would also
involve updating depends.py for example.)
So, use a dependency that's stricted than necessary. IMO it's still good
enough as most configs we test will have ChachaPoly. However it would be
good to revisit this when a cleaner solution is implemented.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
The function depends on MBEDTLS_SSL_HANDSHAKE_WITH_CERT_ENABLED which is
basically
MBEDTLS_SSL_TLS1_3_KEY_EXCHANGE_MODE_EPHEMERAL_ENABLED ||
MBEDTLS_KEY_EXCHANGE_WITH_CERT_ENABLED
The individual test cases depend on the specific TLS version.
This is not precise enough. In a build with both TLS versions enabled,
we could have cert-based key exchange in one version but not in the
other. So, we need the 1.3 tests to depend on the 1.3 cert-based key
exchange and similarly for 1.2.
For 1.2, cert-based key exchange means ECDHE-{RSA,ECDSA} or
ECDH-{RSA,ECDSA}. Since the test function sets an ECC cert for the
server, we want one of the ECDSA ones. So, the minimal dependency would
be ECDH_ECDSA || ECDHE_ECDSA. Since dependencies with || are
inconvenient to express, and anyway ECDH_ECDSA (static ECDH) is
something we'd like to remove in 4.0 if we can find the time, I chose to
just depend on ECDHE_ECDSA.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
These are not optimal. For example, the tests should in principle be
able to run in builds without ECDSA, by using RSA certs instead. Ideally
PSK should work too.
However, getting optimal dependencies would be a lot of work that's
largely orthogonal to the purpose of this PR, so we'll settle for good
enough.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
Simulate the server closing the connection after a partial handshake
message.
These test cases don't send a close_notify alert. The test cases
"insert alert record" exercise what happens if the server sends an alert.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Create tests that coalesce the handshake messages in the first flight from
the server. This lets us test the behavior of the library when a handshake
record contains multiple handshake messages.
Only non-protected (non-encrypted, non-authenticated) handshake messages are
supported.
The test code works for all protocol versions, but it is only effective in
TLS 1.2. In TLS 1.3, there is only a single non-encrypted handshake record,
so we can't test records containing more than one handshake message without
a lot more work.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
* development-restricted:
Add missing credit for set_hostname issue
Add changelog entry for TLS 1.2 Finished fix
TLS1.2: Check for failures in Finished calculation
ssl_session_reset: preserve HOSTNAME_SET flag
Document the need to call mbedtls_ssl_set_hostname
Improve documentation of mbedtls_ssl_set_hostname
Changelog entries for requiring mbedls_ssl_set_hostname() in TLS clients
Add a note about calling mbedtls_ssl_set_hostname to mbedtls_ssl_setup
mbedtls_ssl_set_hostname tests: add tests with CA callback
Call mbedtls_ssl_set_hostname in the generic endpoint setup in unit tests
Require calling mbedtls_ssl_set_hostname() for security
Create error code for mbedtls_ssl_set_hostname not called
Keep track of whether mbedtls_ssl_set_hostname() has been called
Access ssl->hostname through abstractions in certificate verification
mbedtls_ssl_set_hostname tests: baseline
Add a flags field to mbedtls_ssl_context
Automate MBEDTLS_X509_TRUSTED_CERTIFICATE_CALLBACK dependency
Make guards more consistent between X.509-has-certs and SSL-has-certs
Fix Doxygen markup
Make ticket_alpn field private
Conflicts:
programs/ssl/ssl_test_common_source.c
Now that in TF-PSA-Crypto CI, the TF-PSA-Crypto
all.sh components are run in pure TF-PSA-Crypto
context, there is no need to run them as part of
mbedtls CI anymore. The all.sh wrapper wrapping
./tests/scripts/mbedtls-all.sh and
./tf-psa-crypto/tests/scripts/all.sh can be
removed.
Signed-off-by: Ronald Cron <ronald.cron@arm.com>
The openssl version in the docker image doesn't support TLS 1.3, so we
can't run the test.
Signed-off-by: Max Fillinger <maximilian.fillinger@foxcrypto.com>
HKDF-Expand can produce at most 255 * hash_size bytes of key material,
so this limit applies to the TLS 1.3 key material exporter.
Signed-off-by: Max Fillinger <maximilian.fillinger@foxcrypto.com>
With the removal of the component_test_psa_inject_entropy test,
MBEDTLS_NO_DEFAULT_ENTROPY_SOURCES is no longer set in any tests, and so
the CI will complain unless it is added to the ignore list.
Signed-off-by: Felix Conway <felix.conway@arm.com>
Adapt builds and check-generated-files.sh to the fact
that generate_config_tests.py does not generate
test_suite_config.psa_boolean.data in Mbed TLS 4.x
context anymore.
Signed-off-by: Ronald Cron <ronald.cron@arm.com>
This is more flexible: the test data gets to decide whether we want to
assert the presence of a pattern or not.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
This should avoid running into a bug with printf format specifiers one
windows.
It's also a logical move for actual tests: I used the highest debug
level for discovery, but we don't need that all the time.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
Declare the same dependencies as for the previous TLS 1.3 tests, except
for part that varies with the cipher suite (ie AES-GCM).
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
We're not sending a signature_algorithm extension, which means SHA-1.
Caught by depends.py hashes
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
In addition to secp256r1 for the handshake, we need secp384r1 as it's
used by the CA certificate.
Caught by depends.py curves
Also, for the "unknown ciphersuite" 1.2 test, use the same key type and
all the same dependencies as of the "good" test above, to avoid having
to determine a second set of correct dependencies just for this one.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
This guards the definition of mbedtls_test_ssl_endpoint which we rely
on, so the function won't compile without it.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
Two of these tests reveal bugs in the code, so they're commented out for
now.
For the other tests, the high-level behaviour is OK (break the
handshake) but the details of why are IMO not good: they should be
rejected because interleaving non-HS record between HS fragments is not
valid according to the spec.
To be fixed in future commits.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
