@0xc0170 What we talked about earlier.
@LMESTM I guess you're having the same issues.
Target
EFM32
Toolchain:
GCC_ARM
mbed-os sha:
458e726b1
Expected behavior
Unit tests work with all tickers corresponding to the new requirements (width between 16 bits and 32 bits, frequency between 250kHz and 8MHz).
Actual behavior
At various stages in the unit test (https://github.com/ARMmbed/mbed-os/blob/458e726b1e2c32fb2ea6238771190d6fdd78e18d/TESTS/mbed_hal/common_tickers/main.cpp) there are steps that don't account for getting a potential overflow on a fast-running short-width timer.
Examples:
Indeed - same issues faced, now also with sleep tests, for instance here:
https://github.com/ARMmbed/mbed-os/blob/feature-hal-spec-sleep/TESTS/mbed_hal/sleep/main.cpp#L87
Agree also, seems that some tests need few update to ticker_info_t with frequency other than 1MHz and counter different than 32.
This test assumes that each test case (if needed) is executed before overflow occurs. overflow_protect() function should protect against overflow.
I also had some issues with overflows and had to modify this first test version. Final test version is provided here PR https://github.com/ARMmbed/mbed-os/pull/5629 (https://github.com/ARMmbed/mbed-os/pull/5629/commits/80c55e47f85aaeb8137367dbd962cb47b86e4bf6).
Please check if this test version works.
@mprse @jamesbeyond could you split the test fixes to separate PR please.
PR with test fixes only can be found here PR https://github.com/ARMmbed/mbed-os/pull/6472.
PR with test fixes only can be found here PR #6472.
Integrated, shall this issue be closed?
ARM Internal Ref: MBOTRIAGE-205
Closing since this issue appears resolved.
If it isn't, feel free to open a new issue.
Most helpful comment
@mprse @jamesbeyond could you split the test fixes to separate PR please.