Mbed-os: Watchdog issue with timeout

Created on 7 Aug 2019  路  6Comments  路  Source: ARMmbed/mbed-os

Description

Currently calculated timout is rounded down on STM devices, but in this test:
https://github.com/ARMmbed/mbed-os/blob/431c4c19ae2056785c85e2a119ed51dd35f79f3c/TESTS/mbed_drivers/watchdog/main.cpp#L223-L230
It is checked if calculated is bigger or equal to the value passed via init function. This test passes for most of the targets, because of their LSI frequency value which after division throgh it gives good integer value so that reload value can be calculated without any rounding. But not in the case of STM32-L0xx boards where the LSI frequency value is set to 37 kHz.

So should we change the test so we accept rounding down, or should we round the values up (if it doesn't exceed the upper limit of timeout)?

@maciejbocianski @mprse @fkjagodzinski @jamesbeyond

Issue request type


[ ] Question
[ ] Enhancement
[x] Bug

CLOSED mirrored bug

All 6 comments

So should we change the test so we accept rounding down, or should we round the values up (if it doesn't exceed the upper limit of timeout)?

The test verifies our timing requirements for watchdog timeout. Has to stay as it is. It is the implementation, that has to be updated, if it doesn't comply with requirements.

cc @ARMmbed/team-st-mcd

Issue confirmed, you can st st device label

11178 has been merged. @Tharazi97 could you confirm it fixed this issue?

Tested, it fixes the issue. Closing.

Was this page helpful?
0 / 5 - 0 ratings

Related issues

cesarvandevelde picture cesarvandevelde  路  4Comments

pilotak picture pilotak  路  3Comments

ashok-rao picture ashok-rao  路  4Comments

davidantaki picture davidantaki  路  3Comments

bulislaw picture bulislaw  路  3Comments