[Bug c++/102015] New: [missed optimization] Small memory overhead in _Rb_tree_impl (fix would require ABI break)

[Bug c++/102015] New: [missed optimization] Small memory overhead in _Rb_tree_impl (fix would require ABI break)

[kamkaz at windowslive dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102015

            Bug ID: 102015
           Summary: [missed optimization] Small memory overhead in
                    _Rb_tree_impl (fix would require ABI break)
           Product: gcc
           Version: 11.2.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: kamkaz at windowslive dot com
  Target Milestone: ---

Current definition of _Rb_tree_key_compare causes size overhead for all
std::(multi)set/map-s:

  template<typename _Key_compare>
    struct _Rb_tree_key_compare
    {
      _Key_compare              _M_key_compare;
      ...
    };

If it were possible to change the ABI in the future, I think it can be improved
by empty-base-optimization for comparators that are not final classes - or by
adding  [[no_unique_address]] to _M_key_compare.

[Bug libstdc++/102015] [missed optimization] Small memory overhead in _Rb_tree_impl (fix would require ABI break)

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102015

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
https://stackoverflow.com/questions/66573773/is-there-a-reason-for-8-bytes-of-size-overhead-in-libstdc-stdmultiset-map

[Bug libstdc++/102015] [missed optimization] Small memory overhead in _Rb_tree_impl (fix would require ABI break)

[kamkaz at windowslive dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102015

--- Comment #2 from Kamil Kaznowski <kamkaz at windowslive dot com> ---
(In reply to Andrew Pinski from comment #1)
> https://stackoverflow.com/questions/66573773/is-there-a-reason-for-8-bytes-
> of-size-overhead-in-libstdc-stdmultiset-map

This is my post, I forgot to post a bug that day.

There is also a mistake by me in the comments - MSVC _Compressed_pair works by
template specialization, not SFINAE.

[Bug libstdc++/102015] [missed optimization] Small memory overhead in _Rb_tree_impl (fix would require ABI break)

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102015

Jonathan Wakely <redi at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |enhancement
   Last reconfirmed|                            |2021-08-23
     Ever confirmed|0                           |1
             Status|UNCONFIRMED                 |NEW

--- Comment #3 from Jonathan Wakely <redi at gcc dot gnu.org> ---
There's no way to use [[no_unique_address]] in C++98 mode, and making it use
EBO would be a pain.

Confirmed as a possible future enhancement.

[Bug libstdc++/102015] [missed optimization] Small memory overhead in _Rb_tree_impl (fix would require ABI break)

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102015

--- Comment #4 from Jonathan Wakely <redi at gcc dot gnu.org> ---
In https://gcc.gnu.org/pipermail/libstdc++/2021-August/053108.html I proposed
dropping C++98 support for the gnu-versioned-namespace, which would allow us to
fix this by using [[__no_unique_address__]].

N.B. I reported this same issue as PR 77402 five years ago.

[Bug libstdc++/102015] [missed optimization] Small memory overhead in _Rb_tree_impl (fix would require ABI break)

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102015

Jonathan Wakely <redi at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |redi at gcc dot gnu.org

--- Comment #5 from Jonathan Wakely <redi at gcc dot gnu.org> ---
*** Bug 77402 has been marked as a duplicate of this bug. ***