public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
From: "redbeard0531 at gmail dot com" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug libstdc++/104223] New: GCC unable to inline trivial functions passed to views::filter and transform unless lifted into types
Date: Tue, 25 Jan 2022 13:11:22 +0000	[thread overview]
Message-ID: <bug-104223-4@http.gcc.gnu.org/bugzilla/> (raw)

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104223

            Bug ID: 104223
           Summary: GCC unable to inline trivial functions passed to
                    views::filter and transform unless lifted into types
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: libstdc++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: redbeard0531 at gmail dot com
  Target Milestone: ---

I'm not sure if this is an issue with the optimizer or the way that the library
code is written, or some combination of the two, but the end result seems
unfortunate. with() and without() are logically the same function, the only
difference is that with() lifts the function pointers into types using a
templated lambda variable, while without() just passes the function names
directly to the library. It seems interesting that the optimizer knows that
they are "constant enough" to emit direct rather than indirect calls to t() and
f(), however, it isn't constant enough to inline those calls.


https://godbolt.org/z/EqWzzh916
Flags: -std=c++2b -O3
Reproduces on at least 11.2 and trunk

#include <ranges>

namespace views = std::views;

void trace() noexcept;
inline int f(int i) {
    trace();
    return i;
}

inline bool t(int) { return true; }

// for some reason gcc needs this in order to inline f() and t()
template <auto f>
auto typify = [] (int i) { return f(i); };

void with() {
    for (auto&& x : views::single(1) | views::transform(typify<f>) |
views::filter(typify<t>)) {}
}

void without() {
    for (auto&& x : views::single(1) | views::transform(f) | views::filter(t))
{}
}

with():
        sub     rsp, 8
        call    trace()
        add     rsp, 8
        jmp     trace()
without():
        sub     rsp, 8
        mov     edi, 1
        call    f(int)
        mov     edi, eax
        call    t(int)
        test    al, al
        jne     .L10
        add     rsp, 8
        ret
.L10:
        mov     edi, 1
        add     rsp, 8
        jmp     f(int)

             reply	other threads:[~2022-01-25 13:11 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-01-25 13:11 redbeard0531 at gmail dot com [this message]
2022-01-25 17:22 ` [Bug ipa/104223] " pinskia at gcc dot gnu.org
2022-01-26  7:55 ` rguenth at gcc dot gnu.org

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=bug-104223-4@http.gcc.gnu.org/bugzilla/ \
    --to=gcc-bugzilla@gcc.gnu.org \
    --cc=gcc-bugs@gcc.gnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).