[PATCH][PR debug/65549] Restore DW_AT_abstract_origin for cross-unit call sites

[Richard Biener]

On Tue, 9 Jun 2015, Pierre-Marie de Rodat wrote:

> Hello,
> 
> With the recent work for PR debug/65549, we observed a regression in the
> generated debugging information. Take the attached reproducer, build it and
> look at the output DWARF:
> 
>      $ gcc -c -O1 -g foo.c
>      $ objdump --dwarf=info foo.o
>      [...]
>       <2><4e>: Abbrev Number: 3 (DW_TAG_GNU_call_site)
>          <4f>   DW_AT_low_pc      : 0x9
>       <2><57>: Abbrev Number: 0
> 
> There is no DW_AT_abstract_origin attribute anymore, whereas there used to be
> one.
> 
> On PowerPC with -mlongcall, for instance, call instructions are indirect and
> we (at AdaCore) rely on this attribute to determine what function is actually
> called (it's easier this way than interpreting machine code...). The DWARF
> proposal for call sites also say debuggers should be able to use this
> attribute (to build virtual call stacks in the presence of tail calls), but I
> could not observe this in practice with GDB.
> 
> The attached patch is an attempt to restore this attribute. The logic behind
> it is: this is what we used to do previously for contexts that are alien
> translation unit (through the call to force_decl_die), so this should not
> harm.
> 
> Bootstrapped and regtested on x86_64-linux. Ok to commit?

Hmm, so the underlying issue is that we don't associate comp_unit_die ()
with any TRANSLATION_UNIT_DECL.  For the LTO early debug work I did
the following at the very beginning of dwarf2out_early_finish:

  /* Pick the first TRANSLATION_UNIT_DECL we didn't create a DIE for
     and equate it with our default CU DIE.  LTO output needs to be
     able to lookup DIEs for translation unit decls.  */
  unsigned i;
  tree decl;
  FOR_EACH_VEC_SAFE_ELT (all_translation_units, i, decl)
    if (!lookup_decl_die (decl))
      equate_decl_number_to_die (decl, comp_unit_die ());

We create some DIEs for builtin types too early before frontends
got a chance to build their TRANSLATION_UNIT_DECL, so comp_unit_die
gets created before there is any TRANSLATION_UNIT_DECL.  Another
approach would of course be that the Frontends register their main
TRANSLATION_UNIT_DECL with dwarf2out via a debug hook.

But - does the above work for you and fix the regression?  If so
adding a debug hook would probably be the cleaner solution still.

Thanks,
Richard.

> Thank you in advance!
> 
> gcc/ChangeLog
>         * dwarf2out.c (lookup_context_die): Return the DIE for the
>         current compilation unit when the input context is a
>         TRANSLATION_UNIT_DECL.
> 
> 

-- 
Richard Biener <rguenther@suse.de>
SUSE LINUX GmbH, GF: Felix Imendoerffer, Jane Smithard, Dilip Upmanyu, Graham Norton, HRB 21284 (AG Nuernberg)