[STM32F401xE] Freeze master timer on stop/breakpoint

[betzw]

Description

Freeze master timer when debugging enabled and execution hits a breakpoint or gets stop by user.

Status

READY

[0xc0170]

There are CI failures, please look at them.

Is this intended to stop counters (something that API should specify) ? What are the +/- ?
I have seen this feature in some timers, but not certain all would support it.

cc @c1728p9 @bulislaw

[betzw]

There are CI failures, please look at them.

As most often, my PRs are just proposals to be extended to all ST platforms. In this case only platform TARGET_STM32F401xE is currently supported.
It will be up to a product division (i.e. @bcostm & @LMESTM) together with ARMmbed to decide if and how to proceed.

Is this intended to stop counters (something that API should specify) ? What are the +/- ?
I have seen this feature in some timers, but not certain all would support it.

Well, I won't see it as something an API should specify, it's simply something which makes debugging easier as timers will stop when a debugger stops the exeuction of an application. Typically, having timers continue to run while execution is stopped is an intrusion which often makes it impossible to continue execution afterwards without impacting also the overall behabior (e.g. by running immediately into a timeout).
It's likely that not all platforms support something like this, but the ones which support it would help their programmers when they try to debug their application.

[c1728p9]

Hi @betzw, you might want to leave the RTC counting even when the target is halted. I would expect this to keep time regardless of debugging.

[betzw]

This might be reasonable, I have actually never faced a situation where the RTC created problems/timeouts during debugging, even if I believe you cannot exclude it. It's kind of "there is no right or wrong" ...

[bcostm]

It's ok with me.

I have some remarks:

• Add STM32F401xE in the title
• Add #ifdef TIM_MST_DBGMCU_FREEZE because the hal_tick_xxx.c files are common to all ST devices
• Add the possibility to enable/disable this feature using mbed_app.json file ?

[betzw]

@bcostm
Well, as said before this PR must be extended (in a new PR) to all STM32 platforms by your group, so you can add your modifications there.

[theotherjimmy]

@bcostm Do you have plans to extend this PR?

[bcostm]

I didn't know you were waiting for me... Well, I'd prefer to wait that this PR is accepted and merged first on this target (F401RE). Then sure we'll extend it on all other targets.

[0xc0170]

I didn't know you were waiting for me... Well, I'd prefer to wait that this PR is accepted and merged first on this target (F401RE). Then sure we'll extend it on all other targets.

It needs some work - RTC not freeze, test it (all CI green)

[betzw]

@0xc0170 now all 3 CI test are passing.
Are you ok with merging this PR now?

[0xc0170]

LGTM

@c1728p9 @bulislaw Please review

[c1728p9]

I'm use to having the RTC keep accurate time even when debugging. Its understandable to want to halt it when debugging if it is used for scheduling. As long as everyone is happy with this tradeoff, I'm all for it.

[bcostm]

@betzw I see that RTC freeze has been removed in your commit. Can you please update the PR title and description and remove the RTC also ? Thanks.

[betzw]

@betzw I see that RTC freeze has been removed in your commit. Can you please update the PR title and description and remove the RTC also ? Thanks.

Done.

[studavekar]

/morph test

[mbed-bot]

Result: SUCCESS

Your command has finished executing! Here's what you wrote!

/morph test

Output

mbed Build Number: 1218

All builds and test passed!