MI: *stopped response without frame

MI: *stopped response without frame

[Vladimir Prus]

Hi,
I'm running in a case where *stopped response does not include frame
information, while I think it should. Here's log:

   Writing to gdb: -exec-next
   MI reply: ^running
   MI stop reason running
   MI reply: (gdb)
   MI reply: ~"[Thread -1210950736 (LWP 24385) exited]\n"
   MI reply: ~"Thread Event Breakpoint: gdb should not stop!\n"
   MI reply: *stopped,reason="end-stepping-range",thread-id="2"

I get this by stopping in a threaded program, and then invoking "next"
command several times.

I've two questions:

1. Why is *stopped respose with "end-stepping-range" as reason does not
   include frame information?
2. What does

       Thread Event Breakpoint: gdb should not stop!

   mean?

I've also possible answer to those questions: the situation happens when one
of the thread exits (i.e. it's inside __nptl_death_event). But then,
-stack-list-frames has no problems showing we're inside __nptl_death_event,
so either stop reason should be something different, line "thread-exited",
or *stopped reply should include frame.

Thoughts?


- Volodya

Re: MI: *stopped response without frame

[Daniel Jacobowitz]

On Tue, Apr 25, 2006 at 12:32:07PM +0400, Vladimir Prus wrote:
> I've two questions:
> 
> 1. Why is *stopped respose with "end-stepping-range" as reason does not
>    include frame information?
> 2. What does
> 
>        Thread Event Breakpoint: gdb should not stop!
> 
>    mean?
> 
> I've also possible answer to those questions: the situation happens when one
> of the thread exits (i.e. it's inside __nptl_death_event). But then,
> -stack-list-frames has no problems showing we're inside __nptl_death_event,
> so either stop reason should be something different, line "thread-exited",
> or *stopped reply should include frame.

Or, have you considered, the error message might be correct? :-)  GDB
should not have stopped at all!  It's a bug somewhere.

-- 
Daniel Jacobowitz
CodeSourcery