Re: [PATCH v2] gdb/jit: add function symbols to global block

["Jan Vraný" <Jan.Vrany@labware.com>]

On Tue, 2026-01-13 at 09:20 -0700, Tom Tromey wrote:
> > > > > > "Jan" == Jan Vrany <jan.vrany@labware.com> writes:
> 
> Jan> +	{
> Jan> +	  new_block = new (&objfile->objfile_obstack) global_block;
> Jan> +	  new_block->set_multidict (
> Jan> +	    mdict_create_linear (&objfile->objfile_obstack, global_symbols));
> Jan> +	}
> Jan>        else
> Jan> -	new_block = new (&objfile->objfile_obstack) block;
> Jan> +	{
> Jan> +	  new_block = new (&objfile->objfile_obstack) block;
> Jan> +	  new_block->set_multidict (
> Jan> +	    mdict_create_linear (&objfile->objfile_obstack, {}));
> Jan> +	}
> 
> Does anything ever get added to the static block?

No (at the moment). 

> If not then making a dictionary doesn't seem necessary?

As the code stands now, I think it has to - the rest of code expects it is
always there. For example, blockvector::~blockvector() just unconditionally
frees the dictionary. Or blpy_repr() that calls mdict_size().

> 
> Also the formatting is wonky here, that "(" should not be at the end of
> the line.  (Two instances of this.)

Sorry, will fix that.

Jan
> 
> Kind of weird that this code doesn't use buildsym, but OTOH buildsym is
> horrible.
> 
> Tom