Fix crash capture feature for nuvoton

[studavekar]

Description

This should fix #9069, updated to created a memory region instead of block.

Tested on a set of devices, instead of just one like #10311

Device	Test	Results
K64F	tests-mbed_platform-crash_reporting	Pass
NRF52840_DK	tests-mbed_platform-crash_reporting	Pass
DISCO_L475VG_IOT01A	tests-mbed_platform-crash_reporting	Pass
NUMAKER_PFM_NUC472	tests-mbed_platform-crash_reporting	Pass
NUMAKER_PFM_M487	tests-mbed_platform-crash_reporting	Pass

Pull request type

[x] Fix
[ ] Refactor
[ ] Target update
[ ] Functionality change
[ ] Docs update
[ ] Test update
[ ] Breaking change

Reviewers

@SenRamakri @ccli8

Release Notes

[ccli8]

On Nuvoton targets, CSTACK is placed at start of IRAM region to detect stack overflow by hardware error. Would you change crash data to other location like:

M487.icf:

define symbol __ICFEDIT_region_IRAM_start__ = 0x20000000;
define symbol __ICFEDIT_region_IRAM_end__   = 0x20027F00 - 1;
define symbol __region_CRASH_DATA_RAM_start__  = 0x20027F00;
define symbol __region_CRASH_DATA_RAM_end__  = 0x20028000 - 1;

TARGET_NU_XRAM_SUPPORTED/NUC472_442.icf:
TARGET_NU_XRAM_UNSUPPORTED/NUC472_442.icf:

define symbol __ICFEDIT_region_IRAM_start__ = 0x20000000;
define symbol __ICFEDIT_region_IRAM_end__   = 0x2000FF00 - 1;
define symbol __region_CRASH_DATA_RAM_start__  = 0x2000FF00;
define symbol __region_CRASH_DATA_RAM_end__  = 0x20010000 - 1;

[jeromecoutant]

Hi

I still think that this crash feature support should be defined for ALL targets.
Then customer is enabled it or not only with json configuration.

So CRASH_DATA_SIZE should not be part of icf, ld and sct files,
this should be defined at platform side: 0x100 when feature is enabled, 0 when disabled.

Regards,

[studavekar]

I still think that this crash feature support should be defined for ALL targets.

Ideally, that should be the case. However, it's not in the scope of current PR

So CRASH_DATA_SIZE should not be part of icf, ld and sct files,
this should be defined at platform side: 0x100 when feature is enabled, 0 when disabled.

we need to carve out memory to store the crash context and report it after reset. @jeromecoutant can you please elaborate on suggested improvement.

[studavekar]

@ccli8 have made the changes suggested.

[jeromecoutant]

@jeromecoutant can you please elaborate on suggested improvement.

There should not be any impact on RAM when crash feature is disabled.
That's why I suggested a 0 size.

[0xc0170]

I still think that this crash feature support should be defined for ALL targets.
Then customer is enabled it or not only with json configuration.

@jeromecoutant To keep this moving, can you create an issue about this?

[0xc0170]

CI started

[mbed-ci]

Test run: FAILED

Summary: 2 of 7 test jobs failed
Build number : 1
Build artifacts

Failed test jobs:

• jenkins-ci/mbed-os-ci_build-ARM
• jenkins-ci/mbed-os-ci_build-GCC_ARM

[0xc0170]

Failures are related to the changes - linker error about undefined crash symbols

[studavekar]

Failures are related to the changes - linker error about undefined crash symbols

Fixed the CI failure for GCC_ARM and ARM toolchain

[cmonr]

CI started

[mbed-ci]

Test run: FAILED

Summary: 1 of 11 test jobs failed
Build number : 2
Build artifacts

Failed test jobs:

• jenkins-ci/mbed-os-ci_exporter

[0xc0170]

Exporters restarted - internal symlink error