Re: [RFC] Change the_dummy_target to be a global

[Pedro Alves <palves@redhat.com>]

On 03/04/2019 08:41 PM, Tom Tromey wrote:
> While debugging gdb, I printed the target stack and got:
> 
>     (top-gdb) p g_target_stack
>     $10 = {
>       m_top = thread_stratum,
>       m_stack = {0x142b0b0, 0x13da600 <exec_ops>, 0x1c70690, 0x13d63b0 <ravenscar_ops>, 0x0, 0x0, 0x0}
>     }
> 
> (This is clearly from before the change to make ravenscar
> multi-target-capable.)
> 
> Here, 0x142b0b0 is the singleton dummy target.  It seems to me that
> since this is always a singleton, it would be a bit nicer if it were a
> global, so that it would be noted in the above.
> 
> This patch implements this idea, and now I get:
> 
>     (top-gdb) p g_target_stack
>     $2 = {
>       m_top = dummy_stratum,
>       m_stack = {0x1f1b040 <the_dummy_target>, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}
>     }
> 
> I did not do the same for the debug target.  It didn't seem as useful
> to me.
> 
> gdb/ChangeLog
> 2019-03-04  Tom Tromey  <tromey@adacore.com>
> 
> 	* target.c (the_dummy_target): Move later.  Change type to
> 	"dummy_target".
> 	(initialize_targets): Don't initialize the_dummy_target.

Yes, I think that's fine.

At some point in the multi-target branch I made a change like that,
but the current state of the branch it's back to being heap-allocated.
I'm not seeing any reason that requires that, though.  I replicated
the change on top of the branch and ran the new multi-target tests,
and they all still passed.

Thanks,
Pedro Alves