Certain firmwares will cause it to get to a state where flashing fails and the reset button must be pushed at the correct time to recover.

This happens mainly when the firmware uses sleep pm modes (off or standby). I agree that this PR is useful for such cases.

@aabadie Is there anything more that this needs?

I just realized that this change is also affecting other nucleo boards: nucleo-l053r8 and nucleo-l031k8, because this openocd configuration file is shared between them.

Since you are reusing the same openocd configuration than the one use by the b-l072z-lrwan1 board, maybe you could use the shared openocd configuration with this board as well ?

I think we should may even expand this, the reason why the assert was left out seems to be so one could debug without resetting it catching a failure state. I am not sure if this should be the default action though, maybe it would be better to just set a flag to stop the reset assert from happening? If that is the case than many should be changed.

so, how to move this forward?

It would be good to see if the problem occurs on the other nucleo boards and if this fix is valid.

It would be good to see if the problem occurs on the other nucleo boards and if this fix is valid.

I think it is. Some months ago, in a previous PR (maybe #8475), I already tried to pass this change. IT was working on all L0 nucleo boards (for sure). But this is not the question.
This is the question:

the reason why the assert was left out seems to be so one could debug without resetting it catching a failure state. I am not sure if this should be the default action though, maybe it would be better to just set a flag to stop the reset assert from happening?

It is also related to #10479 and #8976 I think.

Since you are reusing the same openocd configuration than the one use by the b-l072z-lrwan1 board, maybe you could use the shared openocd configuration with this board as well ?

Done, and it makes everything slightly cleaner!

so @aabadie is this mergeable? seems to work with the L0s a we have...

I think there was a PR that changed the reset assert to default and allows you to select if you want to flash without reset (so you can get into debug mode)... I don't know if that takes care of this.

I have had the same problem with an f401re and appling the change to
boards/common/stm32/dist/stm32f4.cfg
resolved the problem. Might it be a good change for all stm32 boards?

Ya, I think the issue is that one cannot enter into debug mode without it resetting. I will ask around, I thought there was a pr that would give the option but I couldn't find it.

I would be nice to have this in for everything, it seems to make the flasher more robust.

ping @kaspar030 @gebart @smlng

I'm fine with this PR, but once it's merged we'll also have to merge #10479 rather soonish, otherwise make debug won't work on stm32l0 (and possibly other stm32).

I retriggered Murdock just in case and it's still green. Let's go!

Thanks @aabadie! I forgot about this one.