public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
From: "rguenth at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug ipa/59948] Optimize std::function
Date: Wed, 29 Jan 2014 13:27:00 -0000	[thread overview]
Message-ID: <bug-59948-4-MIzHRcmxIg@http.gcc.gnu.org/bugzilla/> (raw)
In-Reply-To: <bug-59948-4@http.gcc.gnu.org/bugzilla/>

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59948

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2014-01-29
                 CC|                            |hubicka at gcc dot gnu.org
     Ever confirmed|0                           |1

--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
Honza will know the correct condition ...

As for optimization the issue is that

  <bb 2>:
  MEM[(struct _Function_base *)&h]._M_manager = 0B;
  MEM[(int (*<T5869>) (int) *)&h] = f;
  h._M_invoker = _M_invoke;
  h.D.26161._M_manager = _M_manager;
  _4 = std::function<int(int)>::operator() (&h, 1);

  <bb 3>:
  _5 = _4;
  _11 = MEM[(struct _Function_base *)&h]._M_manager;
  if (_11 != 0B)
    goto <bb 4>;
  else
    goto <bb 5>;

  <bb 4>:
  _11 (&MEM[(struct _Function_base *)&h]._M_functor, &MEM[(struct
_Function_base *)&h]._M_functor, 3);

the call to std::function<int(int)>::operator() (&h, 1) may clobber h and
thus the _M_manager field.  This function is not early-inlined at -O[23]
and thus the call is not made direct before IPA inlining (or earlier).

Considering inline candidate _Res std::function<_Res(_ArgTypes
...)>::operator()(_ArgTypes ...) const [with _Res = int; _ArgTypes = {int}].
   Estimating body: _Res std::function<_Res(_ArgTypes
...)>::operator()(_ArgTypes ...) const [with _Res = int; _ArgTypes = {int}]/360
   Known to be false: not inlined
   size:10 time:21
   Estimating body: _Res std::function<_Res(_ArgTypes
...)>::operator()(_ArgTypes ...) const [with _Res = int; _ArgTypes = {int}]/360
   Known to be false: not inlined
   size:10 time:21
  will not early inline: int m()/273->_Res std::function<_Res(_ArgTypes
...)>::operator()(_ArgTypes ...) const [with _Res = int; _ArgTypes =
{int}]/360, growth 6 exceeds --param early-inlining-insns divided by number of
calls

forcing that with --param early-inlining-insns=100 arrives at

  <bb 2>:
  MEM[(struct _Function_base *)&h]._M_manager = 0B;
  MEM[(int (*<T5869>) (int) *)&h] = f;
  h._M_invoker = _M_invoke;
  h.D.26161._M_manager = _M_manager;
  _14 = std::_Function_handler<int(int), int (*)(int)>::_M_invoke
(&h.D.26161._M_functor, 1);

  <bb 3>:
  _15 = MEM[(struct _Function_base *)&h]._M_manager;

which has the same problem - just with another function call which the
early inliner doesn't see.


  parent reply	other threads:[~2014-01-29 13:27 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-01-26 11:31 [Bug ipa/59948] New: " glisse at gcc dot gnu.org
2014-01-26 13:24 ` [Bug ipa/59948] " glisse at gcc dot gnu.org
2014-01-29 13:27 ` rguenth at gcc dot gnu.org [this message]
2014-02-04  2:35 ` hubicka at gcc dot gnu.org
2014-02-04  3:00 ` hubicka at gcc dot gnu.org
2014-05-02 14:06 ` glisse at gcc dot gnu.org
2014-09-26 16:54 ` hubicka at gcc dot gnu.org
2015-02-26  0:09 ` glisse at gcc dot gnu.org
2023-09-25  9:04 ` hubicka 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-59948-4-MIzHRcmxIg@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).