On Fri, Sep 13, 2019 at 1:32 AM Alexandre Oliva <ol...@adacore.com> wrote: > > On Sep 12, 2019, Richard Biener <richard.guent...@gmail.com> wrote: > > > Your new predicate looks a bit excessive here given it builds the type > > again? > > Err, there seems to be some misunderstanding here. The predicate avoids > outputting a type for the definition if it's the same DIE that would go > in the specification. Now, when it's a different DIE, sometimes it > might still refer to the same type, as in array-2.C, but I think that's > not just acceptable, it's desirable, for it reflects the different ways > used to denote the same type.
Yes. > Before posting the patch, I added an inform() to the case in which > completing_type_p would return true (bringing about a debug info > change), and reviewed all of the messages in a bootstrap. The only ones > that weren't just adding array element count were those that inspired > array-2.C and array-3.C. > > > So I'm obviously fine with your patch and I guess if we independently > > arrive at this solution that answers my question what "correct DWARF" > > is by a majority decision ;) > > Good. Once we clear up the misunderstanding (I'm not sure whether you > misunderstood the patch, or I misunderstood your response), I'll be glad > to put it in. > > > So - maybe we can have the patch a bit cleaner by adding > > a flag to add_type_attribute saying we only want it if it's > > different from that already present (on the specification DIE)? > > That's exactly what I meant completing_type_p to check. Do you mean it > should be stricter, do it more cheaply, or what? I meant to do it more cheaply, also the name completing_type_p is misleading IMHO since it simply checks (re-)creating the type DIE will yield a different one. In case the FE would be so stupid to first dwarf2out_early_global_decl with a complete type and then later with an incomplete we'd still place a type DIE with the now incomplete type on the specification DIE ... In my view what the FE does is fishy (re-use the DECL node but replace its type). What we'd need here is probably the merge_decls equivalent in the dwarf2out API, but not exactly sure how that would look like (create a new variable DIE and splice out "common" parts into an abstract DIE used by both the old and the new DIE?) Anyhow, I arrived at the same solution - if the type DIE we created in gen_decl_die is different than the one already present and we're creating a specification, put that type on the new DIE, "shadowing" the declaration ones. Richard. > > -- > Alexandre Oliva, freedom fighter he/him https://FSFLA.org/blogs/lxo > Be the change, be Free! FSF.org & FSF Latin America board member > GNU Toolchain Engineer Free Software Evangelist > Hay que enGNUrecerse, pero sin perder la terGNUra jamás - Che GNUevara
