support 'west flash' with native_sim, add debugserver support

[nashif]

• west: add native_sim flash command
• west: runner: native_sim: add debugserver support

Fixes #36706

[aescolar]

2 nits:

• About the rename, given that "native_sim" is the name of one of the boards in tree, and that this works with any other "native" (posix arch based) board in tree including native_posix, I would prefer if we did not call the runner also "native_sim". Maybe just "native" or "native_runner" or "native_run"..
• You can add the flash target to the other boards' cmake (native_posix and nrf_bsim)

[alwa-nordic]

Just my two cents: I think the old name native_gdb makes more sense than just native. From what I understand, the 'runners' are meant to be plug-able targets or environments. This particular runner is for setting up a GDB environment. This is why I named it gdb and not native_posix in the first place. One may want to define a coexisting runner for LLDB or even something with GUI like Ozone or Vscode. It makes sense to want to switch debuggers, just like one may want to switch between debugger probes.

[nashif]

Just my two cents: I think the old name native_gdb makes more sense than just native.

This implements debugserver, debug and flash commands, independent of how you debug or flash etc, debug interfaces can be something else other than gdb, still covered by the same runner, i.e.. through command line arguments. We should not have a runner for gdb and another one for something else.. Also, look at scripts/west_commands/runners/, none of the existing runners is that gdb oriented in the name..

[alwa-nordic]

can be something else other than gdb, still covered by the same runner, i.e.. through command line arguments

Sure, 'can'. But it's not. This runner is de-facto GDB-specific. Early abstraction is usually bad abstraction. E.g. You are adding a '-tui' parameter to this runner. This would make no sense in a GUI or server environment.

Btw, it would be nice if we documented how to trigger the '-tui' parameter.

[alwa-nordic]

I just realized that debugserver is not gdbserver. Now I'm not sure about the naming at all any more. I think I don't understand Zephyr's runner concept.

[alwa-nordic]

I think it's weird that runners have a separate command for 'debug' and 'debugserver'. I would think that should be two separate runners. What do you think? This may be off-topic.

[nashif]

Btw, it would be nice if we documented how to trigger the '-tui' parameter.

west debug --tui

[alwa-nordic]

Btw, it would be nice if we documented how to trigger the '-tui' parameter.

west debug --tui

I see you added code to make it work now. Pro tip: Don't rebase before force-pushing changes. Do an extra force-push just for the rebase. This makes the 'Compare'-feature of GitHub work much better.

[nashif]

changing assignee, most related to native targets, no changes to west

[alwa-nordic]

The kwargs are not used. Is it not better to remove them? Same applies to do_debug and do_debugserver.

[nashif]

This is the "interface" afaik, looking at all the runners:

scripts/west_commands/runners/intel_cyclonev.py:            self.do_flash_elf(**kwargs)
scripts/west_commands/runners/intel_cyclonev.py:            self.do_flash_elf(**kwargs)
scripts/west_commands/runners/intel_cyclonev.py:    def do_flash_elf(self, **kwargs):
scripts/west_commands/runners/native.py:            self.do_flash(**kwargs)
scripts/west_commands/runners/native.py:    def do_flash(self, **kwargs):
scripts/west_commands/runners/nsim.py:            self.do_flash(**kwargs)
scripts/west_commands/runners/nsim.py:    def do_flash(self, **kwargs):
scripts/west_commands/runners/openocd.py:            self.do_flash_elf(**kwargs)
scripts/west_commands/runners/openocd.py:            self.do_flash(**kwargs)
scripts/west_commands/runners/openocd.py:    def do_flash(self, **kwargs):
scripts/west_commands/runners/openocd.py:    def do_flash_elf(self, **kwargs):

No harm in leaving this allowing future expansions