Does this fix the problem for STM32f4 as well?

yapp. That problem was no real problem, it was just because of my own stupidity...

actually without this PR the stm shows the same behavior than the atmel board with the newest OpenOCD version...

hm, but there seems to be some other trouble - Make is exiting from make flash with some Error.

Like I've written in #2470. ;)

ups, that part I did not read...

ok, here is a small workaround - what do you say?

Can't you figure out why openocd exits with a non-zero exit code?

probably, but not at the moment...

Then at least add a comment in there.

I made a patch for OpenOCD: OlegHahm/openocd@61c1263

To me it looks like that the command value for shutdown is interpreted as an error value without this patch.

The behavior was introduced in openocd-org/openocd@a35712a8

So ... the comment should point to the issue (which has to be created first?) over at openocd?

#2472 should do the trick in a sensible way.

How about we merge this, but leave #2472 open until newer OpenOCD versions are available in most distributions (preferably even in Ubuntu LTS)?

don't really like the idea of keeping #2472 open for (what could be) a long time. I would rather merge this PR and close #2472, and then just revisit this in a year or so. And if we really forget about it by then, it's not like overwriting the openOCD return value does break anything...

We have issues that are open for far more time and I hope that even Ubuntu people aren't that slow. Ignoring the OpenOCD return value is not a clean solution in my opinion.

true

Seems upstream openocd got fixed. See #2897.

But OpenOCD in Ubuntu is still 0.8, right?