GCC with TSan + O3 causes an error where it claims key_len and iv_len
may be used uninitialised. This is, as far as I can tell incorrect (the
only way it could not be set is in the error case, and then it is not
used), however the simplest option seemed to be just to fix it.
Signed-off-by: Paul Elliott <paul.elliott@arm.com>
Move the mbedtls_ecc helper functions from psa_core to psa_util.
These files are not implemented as part of the PSA API and should not
be part of the PSA crypto implementation.
Signed-off-by: Joakim Andersson <joakim.andersson@nordicsemi.no>
Remove unused psa_crypto_core.h include.
The PSA util file provides helper functions when using the PSA API.
It should not rely on PSA internal headers, and instead only use
public headers.
Signed-off-by: Joakim Andersson <joakim.andersson@nordicsemi.no>
From time being, domain_parameters could not be extracted
from driver. We need to return error to indicate this
situation. This is temporary and would be fixed after #6494.
Signed-off-by: Pengyu Lv <pengyu.lv@arm.com>
When calculating a result to go into an mbedtls_ms_time_t, make sure
that arithmetic is performed at the final size to prevent overflow.
Signed-off-by: David Horstmann <david.horstmann@arm.com>
Use key enrollment function only when MBEDTLS_PSA_CRYPTO_CLIENT
is enabled, i.e. when the Mbed TLS implementation of PSA Crypto
is being used.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
- move early data check to `prepare`
- avoid `((void) output_len)
- replace check with `session_ticket_allow` in 2nd place
Signed-off-by: Jerry Yu <jerry.h.yu@arm.com>
- get ticket_flags with function.
- improve output message and check it.
- improve `ssl_server2` help message
Signed-off-by: Jerry Yu <jerry.h.yu@arm.com>
Instead of setting both algorithm and enrollement algorithm in the
PK module when parsing the key:
- for Weierstrass keys we only set ECDSA algorithm,
- for Montgomery keys we don't set any algorithm.
Reasons:
- PK module can only do ECDSA and not ECDH
- ECDH is only used in TLS
- Montgomery keys cannot be used to do ECDSA, while Weierstrass ones
can do both ECDSA and ECDH.
So the idea is that once TLS needs the key to do ECDH (either Weierstrass
and Montgomery), it exports the one parsed from the PK module and then
re-imports it setting proper algorithm and flags. In this way the TLS
module will own the new key so it will be its duty to clear it on
exit.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
Instead of extracting key's properties in order to check whether it
supports deterministic or non-deterministic ECDSA, we try both.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
"skip parse certificate verify" can not be changed.
It is used in `Authentication: client badcert, server none`
test.
Signed-off-by: Jerry Yu <jerry.h.yu@arm.com>
