Add uVisor support for the DISCO_F429ZI

[AlessandroA]

Note: Please do not merge yet. Our tests need to be updated to test the new target.

This PR adds support for uVisor on the DISCO_F429ZI target.

Changelog:

• Port of uVisor to DISCO_F429ZI, by @adustm.
• New version of uVisor, v0.26.2, needed to enable the new target.
• Update to other supported linker scripts, to make them consistent with our porting guide.

@0xc0170 @meriac @sg-

@mazimkhan Please use my branch https://github.com/AlessandroA/mbed-os/tree/stm32f4_support to test the new target locally. If those tests succeed you can report back here and I'll remove the WIP tag.

[bridadan]

Went ahead and added the big fat do not merge label to match your description 😄

[mazimkhan]

@AlessandroA I get following error while compiling for profile disco_f429zi_gcc_arm_r_1_1_2

err.txt

[adustm]

Hello @AlessandroA
Unfortunately, mbed-os tests are failing with this PR (or did I launch them wrong ?)
I have done the following :

mbed import https://github.com/AlessandroA/mbed-os
cd mbed-os

(at this point I am on origin/master)
mbed test -m DISCO_F429ZI -t GCC_ARM
It gives 138 tests OK (the complete list ok)

git checkout stm32f4_support
(at this point I am on your Pull Request)
mbed test -m DISCO_F429ZI -t GCC_ARM
It gives TIMEOUT ERROR for every tests

Can you tell me if there is something I did wrong ? and if not, is it possible for you to check that, please ?
Thanks in advance,
Kind regards

[AlessandroA]

Hi @adustm, I will look into that this afternoon and will report back here. Thanks for reporting the issue!

[AlessandroA]

Hi @adustm just a quick update. We found a couple of issues with the tests and we are debugging them at the moment. We plan to release support for the target of this PR with a restricted set of tests, and then work to increase the test coverage. I will update this PR when we are ready to review and merge.

[adustm]

Hi Alessandro,
Happy 2017 to you.
Could you give me some news ? Is there any plan for this ?
Kind regards
Armelle

[sg-]

@AlessandroA any update? If it needs much more work we should close and re-open when ready.

[AlessandroA]

Hi @adustm , @sg- , I updated this PR. The current status is:

• 75% of uVisor tests passing.
• 1 example out of 3 working (the other 2 fail after a couple of seconds).

If all non-uVisor-specific tests succeed for this target, my suggestion is to merge this PR. The uVisor-specific issues will be solved in the next weeks, as some of the tasks we are working on address the underlying problems. That will require a new release of uVisor anyway.

[adustm]

Hi @AlessandroA, @sg-
Thank you for the update.
You did not mention the status of mbed test -m DISCO_F429ZI -t GCC_ARM (regression tests). Did you try it again ?
Cheers,

[AlessandroA]

@adustm those tests fail for me even with the HEAD of master, so I am not sure if the issue is with my board/tools. If they are enabled, I'd rather wait for the results of the CI job which is running in this PR.

[adustm]

Hi @AlessandroA
I had no idea that the master HEAD was wrong at that time... but I won't spend time to confirm it. I passed the tests with your branch and without it, as I previously did in the past.
mbed_lib_rev133 for mbed os 2 tests:
python ./mbed/tools/singletest.py --auto -f DISCO_F429ZI --tc GCC_ARM
[...]Result: 32 OK / 1 NOT_SUPPORTED
mbed-os HEAD of master :
results for mbed test -m DISCO_F429ZI -t GCC_ARM

those tests fail for me even with the HEAD of master, so I am not sure if the issue is with my board/tools.

I just launched mbed test -m DISCO_F429ZI -t GCC_ARM it on HEAD of mbed-os master and it gives :
'[...]mbedgt: test suite results: 48 OK'

Then I did fetch your branch git checkout AlessandroA/stm32f4_support
and lanched again mbed test -m DISCO_F429ZI -t GCC_ARM
'[...]mbedgt: test suite results: 48 OK'

As a conclusion : the porting of uvisor your did for DISCO_F429ZI is now fine for me, causing no dammage. 👍

Your suggestion :

If all non-uVisor-specific tests succeed for this target, my suggestion is to merge this PR. The uVisor-specific issues will be solved in the next weeks, as some of the tasks we are working on address the underlying problems. That will require a new release of uVisor anyway.

is fine for me also. 👍

[adustm]

@bcostm

[adustm]

Hello @sg- , @bridadan
what do you think about @AlessandroA 's idea:

my suggestion is to merge this PR. The uVisor-specific issues will be solved in the next weeks, as some of the tasks we are working on address the underlying problems. That will require a new release of uVisor anyway.

Could the Wip title and the do-not-merge label be removed ?
Kind regards
Armelle

[bridadan]

@adustm We can definitely get a test run on it! The NUCLEO_F429ZI has some failing tests on master when heap and stack stats are enabled (the always are in CI). But I believe this issue is being worked through here: #3319

/morph test

[adustm]

Thanks @bridadan .
You are right about the issue you have with the heap&stack issue. I never see this error on my side.

This PR is for DISCO_F429ZI by the way, not NUCLEO
(I plan the NUCLEO asap after the merge of this one)
cheers

[bridadan]

This PR is for DISCO_F429ZI by the way, not NUCLEO

True, though I think the problem is present in both (as well as a few other ST targets)? I haven't actually tried it on the DISCO though. Shouldn't block this PR from coming in though.

[mbed-bot]

Result: SUCCESS

Your command has finished executing! Here's what you wrote!

/morph test

Output

mbed Build Number: 1389

All builds and test passed!

[bridadan]

CI looks ok, I'll leave the final call up to @0xc0170 or @sg-

[sg-]

LGTM 👍