-
Notifications
You must be signed in to change notification settings - Fork 2k
New issue
Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? # to your account
tests/sys/psa_crytpo_se: disable test on esp32-wroom-32 #20150
Conversation
Why? To little memory? |
So that we can get PRs merged again ;-) This test fails in
This effectively prevents merging any PR. |
Thx :-) |
Do we have any idea why all the PSA tests fail on ESP32x SoCs? |
@Einhornhool guessed it could be a stack overflow, but didn't have the toolchain/hardware at hand to reproduce. I was able to reproduce with my ESP32 Ethernet Kit, but when stepping through the code with GDB I got
I updated the Espressif fork of OpenOCD afterwards, but still the exact same issue. I could help to debug, but I really want to get GDB working again to be more productive at debugging. |
🤔 But wouldn't a stack overflow cause a crash? According to the
|
Actually, these tests need a secure element to perform those operations, so they should actually fail on all boards. |
Yes, the can be disabled on a per-board basis as done in this PR. Alternativly, the python test script could be placed in a There was a way to tell to selectively enable the |
If the test needs a secure element, a whitelist might be a better approach than a blacklist as only few boards come with one (I can only think of But those affect the entire build when we indeed just want to whitelist the automatic test. |
Can't this be expressed as a feature? Then we could just set |
Otherwise, there is always |
If I understood @Einhornhool correctly, it was a deliberate design choice to allow compile-testing it for all platforms regardless of whether a secure element is present or not. But personally, I like it if users get an error message at build time for configurations that are known to not work, rather than letting them debug runtime errors. So strong +1 for a feature. |
PSA Crypto only supports one type of secure element so far and it needs a specific configuration for this test. So it can even fail on boards with a builtin SE. This means it's probably best to not run these at all. It is not really necessary to compile-test for all boards, one should be enough. |
Contribution description
As the title says
Testing procedure
This PR should pass the CI, unlike most CI runs recently.
Issues/PRs references
None