Closed gabor-mezei-arm closed 1 month ago
In addition to that I think we should think about moving a decent chunk of the code into the framework repository.
Due to dependencies a separate issue (#9325) is created for it.
We have some CI issues which I have posted here:
[2024-07-01T12:58:13.652Z] ******************************************************************
[2024-07-01T12:58:13.652Z] * check_python_files: Lint: Python scripts
[2024-07-01T12:58:13.652Z] * Mon Jul 1 12:58:13 UTC 2024
[2024-07-01T12:58:13.652Z] ******************************************************************
[2024-07-01T12:58:13.652Z] Running pylint ...
[2024-07-01T12:58:45.710Z] ************* Module config
[2024-07-01T12:58:45.710Z] scripts/config.py:483:40: W0613: Unused argument 'name' (unused-argument)
[2024-07-01T12:58:45.710Z] scripts/config.py:483:4: R0201: Method could be a function (no-self-use)
[2024-07-01T12:58:45.710Z]
Just a note on the testing I have been doing on the latest commits. I have been invoking the various CLI arguments such as full
, realfull
etc and this appears to behave as I would expect which is great. I have also been using options like set-all/unset-all
and then giving a list of options from both configs and they all seem to get set correctly and seamlessly.
Does this sound like sufficient testing @gabor-mezei-arm ? Or do you have any suggestions for additional things that I should try?
We don't have tests for config.py
as such. But for what it's worth, there's a script tests/scripts/test_config_script.py
that tests whether two implementations of config.py
have the same behavior. You can use it to validate refactorings (with a copy of the old version of config.py
: the test script doesn't access git). Note that we don't run test_config_script.py
often, so it may have bitrotted.
Thanks for the info, will take a look. I know we don't test config.py
in the usual way, I was wondering if there was maybe further testing I could do, that Gabor has tried while developing this, which might be worth me doing as well. But other than what you mentioned I think I have probably covered enough bases now.
Does this sound like sufficient testing @gabor-mezei-arm ? Or do you have any suggestions for additional things that I should try?
There is test script (test_config_script.py
) written for config.py
but is not updated for the crypto config.
Rebased due to merge conflict
Looks like the latest changes have caused issues in generate_config_tests.py
:
[2024-07-04T14:16:53.179Z] Traceback (most recent call last):
[2024-07-04T14:16:53.179Z] File "../framework/scripts/generate_config_tests.py", line 179, in <module>
[2024-07-04T14:16:53.179Z] test_data_generation.main(sys.argv[1:], __doc__, ConfigTestGenerator)
[204-07-04T14:16:53.179Z] File "/var/lib/build/framework/scripts/mbedtls_framework/test_data_generation.py", line 204, in main
[2024-07-04T14:16:53.179Z] generator = generator_class(options)
[2024-07-04T14:16:53.179Z] File "../framework/scripts/generate_config_tests.py", line 162, in __init__
[2024-07-04T14:16:53.179Z] self.mbedtls_config = config.ConfigFile()
[2024-07-04T14:16:53.179Z] TypeError: __init__() missing 2 required positional arguments: 'default_path' and 'name'
[2024-07-04T14:16:53.179Z] Makefile:34: *** "python3 ../framework/scripts/generate_config_tests.py --list" failed. Stop.
[2024-07-04T14:16:53.179Z] make: *** [clean] Error 2
[2024-07-04T14:16:53.179Z] Makefile:154: recipe for target 'clean' failed
Rebased due to merge conflict
@tom-daubney-arm please have another look
Description
For the repo split, we plan to move all configuration options but the TLS and x509 ones from mbedtls_config.h to crypto_config.h that will become the TF-PSA-Crypto configuration file and probably be renamed tf_psa_crypto_config.h.
We need to adapt config.py. Same options, same interface but handles both mbedtls_config.h and tf_psa_crypto_config.h.
Resolve #9158 Depends on Mbed-TLS/mbedtls-framework#35
PR checklist
Please tick as appropriate and edit the reasons (e.g.: "backport: not needed because this is a new feature")