public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
From: "jari.helaakoski at qt dot io" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug c++/113478] New: -Os does not inline single instruction function
Date: Thu, 18 Jan 2024 12:56:28 +0000	[thread overview]
Message-ID: <bug-113478-4@http.gcc.gnu.org/bugzilla/> (raw)

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113478

            Bug ID: 113478
           Summary: -Os does not inline single instruction function
           Product: gcc
           Version: 13.2.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: jari.helaakoski at qt dot io
  Target Milestone: ---

GCC -Os option does not inline single instruction std::atomic function when
accessed via template class. Inlining works ok with -O2. Inlining is also
working if std::atomic is used directly. 

Test case and compilation result found from godbolt:
https://godbolt.org/z/nh758aT3K

Problem reproduces all GCC versions with x86 and aarch64 target and with and
without specific -march and -mtune options.

And here's test case pasted in case of godbolt is not accessable:
template <typename T>
class D {
    public:
#if 1
    inline T test() { return i.load(std::memory_order_relaxed); }
#else
    __attribute__((always_inline)) T test() { return
i.load(std::memory_order_relaxed); }
#endif
    std::atomic<T> i {0};
};

D<char> a1;
D<long> a2;
D<int> a3;
D<short> a4;
D<long long> a5;
D<wchar_t> a6;

extern void mod();

int main()
{
    int ret = 0;
    mod();
    ret |= a1.test();
    ret |= a1.test();
    ret |= a1.test();
    ret |= a1.test();
    ret |= a1.test();
    ret |= a1.test();
    ret |= a1.test();
    ret |= a1.test();
    ret |= a1.test();
    ret |= a1.test();
    ret |= a2.test();
    ret |= a3.test();
    ret |= a4.test();
    ret |= a5.test();
    ret |= a6.test();
    return ret;
}

Function that should be inlined
D<char>::test():
        ldrb    w0, [x0]
        ret

             reply	other threads:[~2024-01-18 12:56 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-01-18 12:56 jari.helaakoski at qt dot io [this message]
2024-01-18 13:12 ` [Bug ipa/113478] " rguenth at gcc dot gnu.org
2024-01-19 10:51 ` hubicka at ucw dot cz
2024-01-19 11:09 ` rguenth at gcc dot gnu.org
2024-01-19 16:43 ` hubicka at ucw dot cz

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-113478-4@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).