public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
From: "tglek at mozilla dot com" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug other/46770] Replace .ctors/.dtors with .init_array/.fini_array on targets supporting them
Date: Wed, 18 Apr 2012 04:58:00 -0000	[thread overview]
Message-ID: <bug-46770-4-0NcY3VCz7E@http.gcc.gnu.org/bugzilla/> (raw)
In-Reply-To: <bug-46770-4@http.gcc.gnu.org/bugzilla/>

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46770

--- Comment #93 from Taras Glek <tglek at mozilla dot com> 2012-04-18 04:48:18 UTC ---
(In reply to comment #92)
> As I said in comment #47 and elsewhere, you should not confuse the order in
> which entries appear in .ctors or .init_array sections with the order in which
> they appear in the binary.  If you want better layout in the binary, then tell
> the linker to change the layout in the binary.  The order in the .ctors or
> .init_array sections is irrelevant.  The fact that reversing the order of
> constructors happens to give you faster startup for firefox is just a
> coincidence.  Don't let that coincidence drive you toward choices that make no
> sense.

I am not confusing the order in which entries appear. My concerns is the order
in which code gets paged in from disk. Right now that's driven entirely by
translation unit "concatenation" and how that relates to the order of ctor
invocation. Currently the right thing(tm) happens, achieving the same by other
means is considerably more complicated.

There are 2 problems with using section ordering to solve this
a) there is no way to do this kind of section ordering transparently(especially
not by default) with current infrastructure.
b) it doesn't work without lto because initializers pull in other code...
I'll expand on b:
We(C++ code like Firefox/Chrome/etc) have one initializer per TU. The
initializer is often not self-contained and calls other code within the
unit(and rarely in other units...this you can't solve without LTO)
So if you pass a list of init sections to the linker...that list needs to be
transitive and include all of the sections called from inits in order to
achieve a useful level of locality post-reorder.

Or you can keep TUs without any reordering, have decent locality(because
related init code is generally nearby) and a sane page-in pattern if the order
of init executions matches TU layout.


  parent reply	other threads:[~2012-04-18  4:58 UTC|newest]

Thread overview: 110+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-12-02 19:21 [Bug target/46770] New: " mh+gcc at glandium dot org
2010-12-02 19:24 ` [Bug target/46770] " mh+gcc at glandium dot org
2010-12-02 22:22 ` hubicka at gcc dot gnu.org
2010-12-02 23:01 ` hjl.tools at gmail dot com
2010-12-07 15:45 ` hjl.tools at gmail dot com
2010-12-07 15:45 ` hjl.tools at gmail dot com
2010-12-07 16:45 ` hjl.tools at gmail dot com
2010-12-09 17:55 ` hjl.tools at gmail dot com
2010-12-11 14:15 ` hubicka at gcc dot gnu.org
2010-12-11 14:28 ` hjl.tools at gmail dot com
2010-12-11 14:36 ` mh+gcc at glandium dot org
2010-12-11 15:01 ` hubicka at ucw dot cz
2010-12-11 15:34 ` hjl.tools at gmail dot com
2010-12-11 16:17 ` hubicka at ucw dot cz
2010-12-11 16:54 ` hjl.tools at gmail dot com
2010-12-11 18:33 ` mmitchel at gcc dot gnu.org
2010-12-11 18:47 ` hjl.tools at gmail dot com
2010-12-11 18:50 ` mark at codesourcery dot com
2010-12-11 19:02 ` hjl.tools at gmail dot com
2010-12-11 19:33 ` mark at codesourcery dot com
2010-12-11 19:44 ` hjl.tools at gmail dot com
2010-12-11 19:47 ` hjl.tools at gmail dot com
2010-12-11 19:48 ` mmitchel at gcc dot gnu.org
2010-12-11 19:51 ` hjl.tools at gmail dot com
2010-12-11 19:53 ` hjl.tools at gmail dot com
2010-12-11 19:57 ` mark at codesourcery dot com
2010-12-11 20:04 ` hjl.tools at gmail dot com
2010-12-11 20:17 ` hjl.tools at gmail dot com
2010-12-11 20:19 ` mark at codesourcery dot com
2010-12-11 21:01 ` hubicka at ucw dot cz
2010-12-11 21:07 ` mark at codesourcery dot com
2010-12-11 21:07 ` hjl.tools at gmail dot com
2010-12-11 22:57 ` hjl.tools at gmail dot com
2010-12-11 23:19 ` mark at codesourcery dot com
2010-12-11 23:28 ` hjl.tools at gmail dot com
2010-12-11 23:30 ` mark at codesourcery dot com
2010-12-11 23:48 ` hjl.tools at gmail dot com
2010-12-11 23:55 ` mark at codesourcery dot com
2010-12-12  0:01 ` hjl.tools at gmail dot com
2010-12-12  0:03 ` mark at codesourcery dot com
2010-12-12  0:08 ` hjl.tools at gmail dot com
2010-12-12  0:12 ` mark at codesourcery dot com
2010-12-12  0:20 ` hjl.tools at gmail dot com
2010-12-12  0:25 ` mark at codesourcery dot com
2010-12-12  0:25 ` hjl.tools at gmail dot com
2010-12-12  0:32 ` hjl.tools at gmail dot com
2010-12-12 15:54 ` hjl.tools at gmail dot com
2010-12-12 18:40 ` mark at codesourcery dot com
2010-12-13  2:30 ` ian at airs dot com
2010-12-13 18:08 ` joseph at codesourcery dot com
2010-12-13 19:41 ` hjl.tools at gmail dot com
2010-12-13 20:24 ` ccoutant at gcc dot gnu.org
2010-12-13 20:31 ` hjl.tools at gmail dot com
2010-12-13 20:42 ` ccoutant at gcc dot gnu.org
2010-12-13 20:48 ` hjl.tools at gmail dot com
2010-12-14  0:39 ` ian at airs dot com
2010-12-14  1:14 ` hjl.tools at gmail dot com
2010-12-14  1:24 ` ccoutant at gcc dot gnu.org
2010-12-14  1:32 ` hjl.tools at gmail dot com
2010-12-14 12:52 ` hubicka at gcc dot gnu.org
2010-12-14 13:17 ` paolo.carlini at oracle dot com
2010-12-14 13:26 ` hubicka at ucw dot cz
2010-12-14 13:33 ` paolo.carlini at oracle dot com
2010-12-14 15:17 ` mark at codesourcery dot com
2011-03-25 19:55 ` jakub at gcc dot gnu.org
2011-04-28 16:17 ` rguenth at gcc dot gnu.org
2011-05-09 10:11 ` amodra at gmail dot com
2011-06-23 16:22 ` ian at airs dot com
2011-06-23 23:17 ` hjl.tools at gmail dot com
2011-06-24 13:22 ` ian at airs dot com
2011-06-24 13:35 ` jakub at gcc dot gnu.org
2011-08-20 20:08 ` hjl at gcc dot gnu.org
2011-08-20 20:37 ` [Bug other/46770] " hjl.tools at gmail dot com
2012-02-18  6:05 ` jingyu at gcc dot gnu.org
2012-02-18  7:21 ` jingyu at gcc dot gnu.org
2012-02-22 22:28 ` jingyu at gcc dot gnu.org
2012-04-17  1:22 ` ccoutant at gcc dot gnu.org
2012-04-17 14:59 ` hjl.tools at gmail dot com
2012-04-17 15:30 ` ppluzhnikov at google dot com
2012-04-17 15:49 ` hjl.tools at gmail dot com
2012-04-17 17:18 ` ppluzhnikov at google dot com
2012-04-17 18:04 ` ccoutant at gcc dot gnu.org
2012-04-17 18:20 ` hjl.tools at gmail dot com
2012-04-17 18:54 ` ccoutant at google dot com
2012-04-17 19:04 ` hjl.tools at gmail dot com
2012-04-17 20:25 ` ccoutant at google dot com
2012-04-17 20:39 ` hjl.tools at gmail dot com
2012-04-17 21:12 ` ccoutant at google dot com
2012-04-17 21:34 ` hubicka at ucw dot cz
2012-04-17 22:00 ` ccoutant at google dot com
2012-04-17 22:27 ` hjl.tools at gmail dot com
2012-04-18  0:05 ` tglek at mozilla dot com
2012-04-18  0:51 ` ppluzhnikov at google dot com
2012-04-18  1:33 ` tglek at mozilla dot com
2012-04-18  3:53 ` ian at airs dot com
2012-04-18  4:58 ` tglek at mozilla dot com [this message]
2012-04-19  0:16 ` ian at airs dot com
2012-04-19 15:15 ` hubicka at ucw dot cz
2012-04-22  8:03 ` igodard at pacbell dot net
2012-04-22 17:04 ` ian at airs dot com
2012-04-22 17:47 ` igodard at pacbell dot net
2012-04-22 18:36 ` hubicka at ucw dot cz
2012-04-22 18:45 ` paolo.carlini at oracle dot com
2012-04-22 19:37 ` igodard at pacbell dot net
2012-04-22 21:19 ` ian at airs dot com
2012-04-22 21:53 ` igodard at pacbell dot net
2012-04-22 22:28 ` ian at airs dot com
2014-10-01 16:46 ` marcelo at brs dot ind.br
2014-10-01 18:16 ` marcelo at brs dot ind.br
2014-10-01 18:50 ` ian at airs dot com

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-46770-4-0NcY3VCz7E@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).