Re: [PP?] [PATCH 2/6] Use new/delete for do_module_cleanup

[Simon Marchi <simark@simark.ca>]

On 2020-08-09 9:52 a.m., Tom Tromey wrote:
> This changes do_module_cleanup to use new and delete.  It also removes
> the use of the struct hack from this object -- this requires more
> allocations for now, but this will be removed in a subsequent patch.
> 
> gdb/ChangeLog
> 2020-08-08  Tom Tromey  <tom@tromey.com>
> 
> 	* compile/compile-object-run.c (struct do_module_cleanup): Add
> 	constructor, destructor.
> 	<objfile_name_string>: Don't use struct hack.
> 	(do_module_cleanup): Use delete.
> 	(compile_object_run): Use new.
> ---
>  gdb/ChangeLog                    |  8 ++++++++
>  gdb/compile/compile-object-run.c | 24 ++++++++++++++++--------
>  2 files changed, 24 insertions(+), 8 deletions(-)
> 
> diff --git a/gdb/compile/compile-object-run.c b/gdb/compile/compile-object-run.c
> index a2f39900053..4a18655d488 100644
> --- a/gdb/compile/compile-object-run.c
> +++ b/gdb/compile/compile-object-run.c
> @@ -32,6 +32,17 @@
>  
>  struct do_module_cleanup
>  {
> +  do_module_cleanup () = default;
> +
> +  ~do_module_cleanup ()
> +  {
> +    delete munmap_list_head;
> +    xfree (source_file);
> +    xfree (objfile_name_string);
> +  }
> +
> +  DISABLE_COPY_AND_ASSIGN (do_module_cleanup);
> +

Seeing this makes me a bit nervous, as it easily allows the destructor run on an object
whose fields have not been initialized.  So it will call delete/xfree on uninitialized
pointers.  So I'd rather introduce a constructor before introducing a destructor, or
simultaneously.

The next patches touch that code, so I presume that will change.  But I read patches in
a very linear way, so that is my comment at this intermediary point :).

Simon