test: some cases are failing with GCC_ARM for Toshiba's TMPM3H6

[ghost]

Description

Target:
Toshiba's TMPM3H6

TMPM3H6 Memory:
RAM - 16K and Flash - 128K

ToolChain:
GCC_ARM

Issue:
Following test cases failed on TMPM3H6 with GCC_ARM - failed for memory allocation on each case. ARM or IAR had passed with no error.

• features-tests-filesystem-buffered_block_device
• features-tests-filesystem-heap_block_device
• features-tests-filesystem-mbr_block_device
• features-tests-filesystem-util_block_device
• tests-mbedmicro-rtos-mbed-threads

Note: If main stack memory is reduced to 2K for GCC_ARM in mbed_rtx.h, then Greentea tests are passed.

Expected behavior
Above mentioned Greentea test cases should pass without change in main stack.

Actual behavior
Above mentioned Greentea test cases fails due to not enough memory (Operator new[] out of memory).

Steps to reproduce

1. Download master from https://github.com/ganesh-ramachandran/mbed-os (since Toshiba TMPM3H6 is yet to be merged)

2. Build command

mbed test -t GCC_ARM -m TMPM3H6 -v -n features-tests-filesystem-buffered_block_device,features-tests-filesystem-heap_block_device,features-tests-filesystem-mbr_block_device,features-tests-filesystem-util_block_device,tests-mbedmicro-rtos-mbed-threads

Logs
Attached below the GCC_ARM logs with 4K and 2K main stack size.
M3H6_GCC_ARM_GreenTea_MainStack_4K.txt
M3H6_GCC_ARM_GreenTea_MainStack_2K.txt

Related Issues
Related issues are in #7115 and #4901

Issue request type

[ ] Question
[ ] Enhancement
[X] Bug

[ciarmcom]

ARM Internal Ref: MBOTRIAGE-1301

[davidsaada]

@ganesh-ramachandran Can you check if #7465 resolves the first 4 tests? 5th test should be handled separately.

[ghost]

@ganesh-ramachandran Can you check if #7465 resolves the first 4 tests? 5th test should be handled separately.

@davidsaada, Thank you for your inputs. By referring #7465, the first 4 tests are resolved.

For 5th test case, observed that test case is passed by reducing 'CHILD_THREAD_STACK_SIZE' from 384 to 223. Please confirm.

[davidsaada]

Thanks @ganesh-ramachandran. Gonna apply a similar fix as in #7465 to tests-mbedmicro-rtos-mbed-threads as well. Either in the same PR or in a separate one.

[0xc0170]

Thanks @ganesh-ramachandran. Gonna apply a similar fix as in #7465 to tests-mbedmicro-rtos-mbed-threads as well. Either in the same PR or in a separate one.

@davidsaada Can you send the PR to fix also threads test?

[davidsaada]

@davidsaada Can you send the PR to fix also threads test?

It's on my todo list, but will take time to resolve (haven't started working on it yet).

[davidsaada]

@ganesh-ramachandran As for the tests-mbedmicro-rtos-mbed-threads test: Do you build it with -DMBED_STACK_STATS_ENABLED=1 flag? Alternatively, does this fail as part of our CI run?
If any of these is true, can you check whether #7730 resolves this problem?
EDIT: Please hold with this check, probably didn't finish my work there.

[davidsaada]

@ganesh-ramachandran Please check whether #7744 solves the tests-mbedmicro-rtos-mbed-threads test problem.

[ghost]

Hi @davidsaada, Thank you for your inputs. By referring #7744 PR, tests-mbedmicro-rtos-mbed-threads test-case is passed.