Re: [RFC 1/3] [gdb/dap] Fix exit race

[Tom de Vries <tdevries@suse.de>]

On 2/13/24 19:04, Tom Tromey wrote:
>>>>>> "Tom" == Tom de Vries <tdevries@suse.de> writes:
> 
>>> One way to fix this would be to have start_json_writer return the
>>> thread
>>> object, and then have main_loop join the thread to terminate after
>>> writing the None.  This would mean making this particular thread
>>> non-daemon though.
> 
> Tom> And as I understand it, the downside of that is that it could possibly
> Tom> hang the gdb process.
> 
> How so?
> 

I don't have a concrete worry if that is what you mean.  I'm just trying 
to reverse-engineer the decision making the dap threads daemon, and this 
is what I could come up with.  I hope this answers your question.

> Tom> +                queue.task_done()
> 
> I think SimpleQueue doesn't have task_done, so you have to change this
> code to use Queue and instead.
> 

Ack, understood.

Anyway, I've now filed a separate PR for this issue, which I hope makes 
the discussion a bit clearer, so we have:
- a PR for the assertion failure (PR31306)
- a PR for ensuring responses are flushed to client before exiting
   (PR31380)

> Tom>          # JSON-writing thread, so that we can ensure that all
> Tom>          # responses are flushed to the client before exiting.
> Tom>          self.write_queue.put(None)
> Tom> +        self.write_queue.join()
> Tom> +        send_gdb("quit")
> 
> I suspect this isn't needed.

I assume you specifically mean the last line.  Removing that line is the 
proposed fix for PR31306, the patch quoted above is for PR31380, which 
we somehow ended up discussing in this thread on PR31306.

Thanks,
- Tom