[RFC] free_lang_data disables -g1

[RFC] free_lang_data disables -g1

[Cary Coutant]

This code in free_lang_data():

  /* FIXME.  Remove after save_debug_info is working.  */
  if (!(flag_generate_lto
	|| (!in_lto_p
	    && !flag_gtoggle && debug_info_level <= DINFO_LEVEL_TERSE)))
    return 0;

does not catch the -g1 case, falling on through, freeing the
lang-specific data, and setting debug_info_level to NONE, with the
effect that we get no debug info. Would a patch to change
"debug_info_level <= DINFO_LEVEL_TERSE" to "debug_info_level <
DINFO_LEVEL_TERSE" be reasonable, or is there some reason we want to
proceed with the freeing at -g1?

I understand that this will go away eventually, but in the meantime,
it appears that -g1 is broken.

-cary

Re: [RFC] free_lang_data disables -g1

[Richard Guenther]

On Thu, Nov 26, 2009 at 1:41 AM, Cary Coutant <ccoutant@google.com> wrote:
> This code in free_lang_data():
>
>  /* FIXME.  Remove after save_debug_info is working.  */
>  if (!(flag_generate_lto
>        || (!in_lto_p
>            && !flag_gtoggle && debug_info_level <= DINFO_LEVEL_TERSE)))
>    return 0;
>
> does not catch the -g1 case, falling on through, freeing the
> lang-specific data, and setting debug_info_level to NONE, with the
> effect that we get no debug info. Would a patch to change
> "debug_info_level <= DINFO_LEVEL_TERSE" to "debug_info_level <
> DINFO_LEVEL_TERSE" be reasonable, or is there some reason we want to
> proceed with the freeing at -g1?
>
> I understand that this will go away eventually, but in the meantime,
> it appears that -g1 is broken.

I think that's just a bug and should read debug_info_level <
DINFO_LEVEL_TERSE (or even == DINFO_LEVEL_NONE).

A patch with either change is pre-approved if it bootstraps and
regtests fine.

Thanks,
Richard.

> -cary
>