STM32F7: Add bootloader support (new trial) #5972

bcostm · 2018-01-30T13:42:22Z

Description

First trial done in PR STM32F7: Add bootloader support #5821 but reverted in PR Revert "STM32F7: Add bootloader support" #5932 due to a sync issue with CI tests on ARM side. Not reproduce on my side.
The only thing I changed compared to the previous PR is the Cache initialization which is done first in the mbed_sdk_init function.
Fixes Issue STM32F7 bootloader does not work #5718

Status

IN DEVELOPMENT

Waiting ci-test result to see if sync issue is still present or not.

The mbed_sdk_init can be called either during cold boot or during application boot after bootloader has been executed. In case the bootloader has already enabled the cache, is is needed to not enable it again.

0xc0170 · 2018-01-30T15:44:59Z

First trial done in PR #5821 but reverted in PR #5932 due to a sync issue with CI tests on ARM side. Not reproduce on my side.

As I understood this was related to cache settings as pinpointed in the PR earlier, and confirmed by few other users (their applications broke). Therefore I would like to understand what broke it (here you can see the confirmation it was just one commit #5912 (comment) ), you should be able to reproduce this. 4 lines changes regarding Cache? that from the commit looks like it can affect an app?

bcostm · 2018-01-30T16:02:39Z

Yes I know it is related to this cache settings but I have no clue why this causes sync issue on your side. I have no problem when I run the tests. No clue either how to solve it. I just want to see if the few lines I have modified change something in your tests.

bcostm · 2018-01-31T10:17:26Z

Hi @MikeDK
Would it be possible for you to check this PR and see if you still have a problem with your application as you have noticed in PR #5912 ? On my side I have tested a basic example on a NUCLEO_F746ZG board with ARM, GCC and IAR compilers and all is ok.

MikeDK · 2018-01-31T11:31:02Z

Hi @bcostm
I just did a test with my application (which basically uses EventQueue, SDBlockDevice, FATFileSystem and some own classes), and it seems to work.

The interesting thing is: it even works now without commit 6e71398

I guess I will have to try harder to reproduce the error on my side ;)

MikeDK · 2018-01-31T11:48:42Z

Ok some new info:

When I checkout 7ff1cf5, which is the commit before the revert PR from @0xc0170, build and flash - it is working as long as I do not unpower the board. If I plug out the USB power and replug, then the board is hanging - so error appears only if cold booted ... If only flashing via copy .bin on mass storage device, the problem does not appear.

So this means the error happens somewhere between power up and st-link getting ready I guess.

Now I just checked this PR again ... If I apply all patches but the last one (6e71398), then the same - works as long I do not unpower the board.

If I then apply patch 6e71398, everything works again.

So it seems that last commit really fixes the problem ;)

0xc0170 · 2018-01-31T11:52:36Z

@MikeDK appreciate the extensive testing for this one!

bcostm · 2018-01-31T12:12:20Z

Thanks @MikeDK I appreciate too 👍

When you say

If I then apply patch 6e71398, everything works again.

Is it working when you unplug/replug the board ?

Another question: are you on Linux or Windows ?

MikeDK · 2018-01-31T12:21:03Z

@bcostm yes, with the latest commit, the board is also starting up correctly when unplugged from power and replugged again.

I work under Linux (Debian Stretch) 🥇

bcostm · 2018-01-31T12:22:58Z

OK because I am on Windows and I can't test it on Linux. @0xc0170 if I am not wrong the ci-test which had the sync issue are ran on Linux right ?

0xc0170 · 2018-01-31T12:28:30Z

OK because I am on Windows and I can't test it on Linux. @0xc0170 if I am not wrong the ci-test which had the sync issue are ran on Linux right ?

I believe on both. I could see aborted on linux, but also windows with IAR had issues (what I am not certain is the sequence it has happened if linux was first and started having those issues).

MikeDK · 2018-01-31T12:54:19Z

Maybe I can test this also on Windows ... I'll have to check with my coworkers - most of them use Windows

cmonr · 2018-01-31T17:27:04Z

@studavekar Pinging you just to let you know about the PR.

studavekar · 2018-01-31T18:12:19Z

OK because I am on Windows and I can't test it on Linux. @0xc0170 if I am not wrong the ci-test which had the sync issue are ran on Linux right ?

Interesting if its host OS causing this bug. would be good to know more about what about OS causing this bug ST-link ?

bcostm · 2018-02-01T08:57:15Z

If I then apply patch 6e71398, everything works again.
So it seems that last commit really fixes the problem ;)

Can you please launch the ci-morph tests to see how it is now ?

0xc0170 · 2018-02-01T09:04:15Z

/morph build

mbed-ci · 2018-02-01T09:25:30Z

Build : SUCCESS

Build number : 1035
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/5972/

Triggering tests

/morph test
/morph uvisor-test
/morph export-build

mbed-ci · 2018-02-01T11:12:17Z

Exporter Build : FAILURE

Build number : 715
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/exporter/5972/

mbed-ci · 2018-02-01T11:13:24Z

Test : SUCCESS

Build number : 842
Test logs :http://mbed-os-logs.s3-website-us-west-1.amazonaws.com/?prefix=logs/5972/842

0xc0170 · 2018-02-01T11:51:07Z

/morph uvisor-test

0xc0170 · 2018-02-01T11:51:25Z

/morph export-build

mbed-ci · 2018-02-01T15:03:05Z

Exporter Build : SUCCESS

Build number : 720
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/exporter/5972/

0xc0170 · 2018-02-01T15:12:08Z

/morph uvisor-test

orenc17 · 2018-02-01T16:32:04Z

/morph uvisor-test

cmonr · 2018-02-01T23:08:03Z

Looks like things are still stable. Going to merge this in first thing tomorrow to keep a controlled eye on it.

Interesting that the only code change that I could tell was the positioning of code in mbed_overrides.c

bcostm · 2018-02-02T08:36:25Z

OK. I don't know exactly what causes the error previously. For the story, I looked in ST CubeF7 examples and found that the cache initialization was done BEFORE the clock setting. This is why I changed it also.

But I also remind the remark from @LMESTM in the previous PR:

As of now, cache is enabled in targets/TARGET_STM/mbed_overrides.c. But shouldn't this be generic ?
Also in case of bootloader, shoudn't mbed_start_application handle the cache cleaning / disabling before jumping into new application ?

cmonr · 2018-02-03T02:07:07Z

@bcostm Interesting. I'm not sure mbed_start_application would be a good place to put the cache-handling code, since it sounds like the function is called after all peripherals are setup (at least, that's the impression I get).

Care to take this comversation to a new issue? I suspect we might have more M7 issues down the line, and it would be good to continue discussion in an issue instead of a completed PR :)

bcostm · 2018-02-05T05:56:23Z

OK I'll open a new issue on this subject to follow-up this.

mbed-ci · 2018-02-09T11:40:17Z

Exporter Build : FAILURE

Build number : 775
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/exporter/5972/

bcostm added 6 commits January 29, 2018 15:52

BL STM32746xG: Update scatter/link files

e10e555

BL NUCLEO_F746ZG: enable bootloader

39cc300

BL NUCLEO_F746ZG: change scb->vector assignment

e933971

NUCLEO_F767ZI: Add bootloader support

5ad0887

Check cache before enabling it

8ed0f26

The mbed_sdk_init can be called either during cold boot or during application boot after bootloader has been executed. In case the bootloader has already enabled the cache, is is needed to not enable it again.

STM32F7: move cache initialization

6e71398

0xc0170 added the needs: review label Jan 30, 2018

cmonr added needs: work and removed needs: review labels Feb 1, 2018

cmonr added ready for merge and removed needs: work labels Feb 1, 2018

cmonr merged commit 1e79439 into ARMmbed:master Feb 2, 2018

cmonr added release-version: 5.7.5 and removed ready for merge labels Feb 2, 2018

bcostm mentioned this pull request Feb 5, 2018

Cortex-M7 cache setting #6010

Closed

bcostm deleted the dev_BL_STM32F7 branch March 21, 2018 09:27

STM32F7: Add bootloader support (new trial) #5972

STM32F7: Add bootloader support (new trial) #5972

Uh oh!

Conversation

bcostm commented Jan 30, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Status

Uh oh!

0xc0170 commented Jan 30, 2018

Uh oh!

bcostm commented Jan 30, 2018

Uh oh!

bcostm commented Jan 31, 2018

Uh oh!

MikeDK commented Jan 31, 2018

Uh oh!

MikeDK commented Jan 31, 2018

Uh oh!

0xc0170 commented Jan 31, 2018

Uh oh!

bcostm commented Jan 31, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

MikeDK commented Jan 31, 2018

Uh oh!

bcostm commented Jan 31, 2018

Uh oh!

0xc0170 commented Jan 31, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

MikeDK commented Jan 31, 2018

Uh oh!

cmonr commented Jan 31, 2018

Uh oh!

studavekar commented Jan 31, 2018

Uh oh!

bcostm commented Feb 1, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

0xc0170 commented Feb 1, 2018

Uh oh!

mbed-ci commented Feb 1, 2018

Build : SUCCESS

Triggering tests

Uh oh!

mbed-ci commented Feb 1, 2018

Exporter Build : FAILURE

Uh oh!

mbed-ci commented Feb 1, 2018

Test : SUCCESS

Uh oh!

0xc0170 commented Feb 1, 2018

Uh oh!

0xc0170 commented Feb 1, 2018

Uh oh!

mbed-ci commented Feb 1, 2018

Exporter Build : SUCCESS

Uh oh!

0xc0170 commented Feb 1, 2018

Uh oh!

orenc17 commented Feb 1, 2018

Uh oh!

cmonr commented Feb 1, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

bcostm commented Feb 2, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

cmonr commented Feb 3, 2018

Uh oh!

bcostm commented Feb 5, 2018

Uh oh!

mbed-ci commented Feb 9, 2018

Exporter Build : FAILURE

Uh oh!

Uh oh!

bcostm commented Jan 30, 2018 •

edited

Loading

bcostm commented Jan 31, 2018 •

edited

Loading

0xc0170 commented Jan 31, 2018 •

edited

Loading

bcostm commented Feb 1, 2018 •

edited

Loading

cmonr commented Feb 1, 2018 •

edited

Loading

bcostm commented Feb 2, 2018 •

edited

Loading