From: Paul Eggert <eggert@cs.ucla.edu>
To: Pip Cet <pipcet@gmail.com>
Cc: 36370@debbugs.gnu.org, Gnulib bugs <bug-gnulib@gnu.org>
Subject: Re: bug#36370: 27.0.50; XFIXNAT called on negative numbers
Date: Thu, 27 Jun 2019 14:13:14 -0700 [thread overview]
Message-ID: <5284eb58-3560-da42-d1d1-3bdb930eae49@cs.ucla.edu> (raw)
In-Reply-To: <CAOqdjBcyT17XDSAEm2NVtFbJLyEc4m9jj_9sX-nyOUKca2aUwA@mail.gmail.com>
[-- Attachment #1: Type: text/plain, Size: 770 bytes --]
On 6/27/19 12:56 PM, Pip Cet wrote:
> The eassume tells GCC i is nonnegative, since (!(i >= 0) == !(i >= 0))
> is indeed a constant.
Ah! Thanks, I didn't catch that subtle point. Would the attached patch
to verify.h address that problem? This patch is for Gnulib, but would
propagate into Emacs.
I tried this out with Emacs master and although it did change the
machine code subtly I didn't have the patience to see whether the
changes were likely to improve performance. The changes did grow the
Emacs text segment from 2556193 to 2557657 bytes (a 0.06% growth), which
is not a good sign. This was on Fedora 30 x86-64 with a default Emacs build.
I'll CC: this to bug-gnulib since it's a Gnulib issue. I have not
installed this patch into Gnulib on savannah.
[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: 0001-verify-tweak-assume-performance.patch --]
[-- Type: text/x-patch; name="0001-verify-tweak-assume-performance.patch", Size: 1822 bytes --]
From 9a5a83937544e7c127026fcf32030f7dbaa5766c Mon Sep 17 00:00:00 2001
From: Paul Eggert <eggert@cs.ucla.edu>
Date: Thu, 27 Jun 2019 14:01:53 -0700
Subject: [PATCH] =?UTF-8?q?verify:=20tweak=20=E2=80=98assume=E2=80=99=20pe?=
=?UTF-8?q?rformance?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Suggested by Pip Cet (Bug#36370#30).
* lib/verify.h (assume): Use __builtin_constant_p to generate
better code in recent GCC.
---
ChangeLog | 7 +++++++
lib/verify.h | 6 ++++--
2 files changed, 11 insertions(+), 2 deletions(-)
diff --git a/ChangeLog b/ChangeLog
index 5ae108e25..7059f4f2b 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,10 @@
+2019-06-27 Paul Eggert <eggert@cs.ucla.edu>
+
+ verify: tweak ‘assume’ performance
+ Suggested by Pip Cet (Bug#36370#30).
+ * lib/verify.h (assume): Use __builtin_constant_p to generate
+ better code in recent GCC.
+
2019-06-26 Paul Eggert <eggert@cs.ucla.edu>
strverscmp: sync from glibc
diff --git a/lib/verify.h b/lib/verify.h
index f8e4eff02..9b015c693 100644
--- a/lib/verify.h
+++ b/lib/verify.h
@@ -263,9 +263,11 @@ template <int w>
accordingly. R should not have side-effects; it may or may not be
evaluated. Behavior is undefined if R is false. */
-#if (__has_builtin (__builtin_unreachable) \
+#if ((__has_builtin (__builtin_constant_p) \
+ && __has_builtin (__builtin_unreachable)) \
|| 4 < __GNUC__ + (5 <= __GNUC_MINOR__))
-# define assume(R) ((R) ? (void) 0 : __builtin_unreachable ())
+# define assume(R) (!__builtin_constant_p (!(R) == !(R)) || (R) \
+ ? (void) 0 : __builtin_unreachable ())
#elif 1200 <= _MSC_VER
# define assume(R) __assume (R)
#elif ((defined GCC_LINT || defined lint) \
--
2.21.0
next parent reply other threads:[~2019-06-27 21:17 UTC|newest]
Thread overview: 28+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <CAOqdjBcM09RbDv19xNF7HxmykU2oAJ4Vsm45Y65aYXZbOO9u3g@mail.gmail.com>
[not found] ` <e7d67132-4c2e-5c3a-74ae-78c8d67b8132@cs.ucla.edu>
[not found] ` <CAOqdjBct1qJ43dAL5642B52ZXH9M1x_qYOZX3GzJi6YvckoS7Q@mail.gmail.com>
[not found] ` <de8a8fa5-176c-f22a-fa56-c5d54fd42352@cs.ucla.edu>
[not found] ` <CAOqdjBd7FXkSd=brysRa8bc+o5uHTBshQ2XxQ2ZSyt=naJgp0g@mail.gmail.com>
[not found] ` <7ef599ae-0a1d-e86f-2bed-a1503455833f@cs.ucla.edu>
[not found] ` <CAOqdjBcyT17XDSAEm2NVtFbJLyEc4m9jj_9sX-nyOUKca2aUwA@mail.gmail.com>
2019-06-27 21:13 ` Paul Eggert [this message]
2019-06-27 21:37 ` bug#36370: 27.0.50; XFIXNAT called on negative numbers Pip Cet
2019-06-27 23:45 ` Bruno Haible
2019-06-28 0:04 ` Paul Eggert
2019-06-28 11:06 ` Pip Cet
2019-06-28 12:14 ` Bruno Haible
2019-06-28 12:29 ` Bruno Haible
2019-06-28 13:51 ` Pip Cet
2019-06-28 17:46 ` Paul Eggert
2019-06-28 19:15 ` Pip Cet
2019-06-28 19:56 ` Bruno Haible
2019-06-28 21:08 ` Pip Cet
2019-06-29 5:41 ` Paul Eggert
2019-06-29 6:48 ` Pip Cet
2019-06-29 17:31 ` Paul Eggert
2019-06-30 9:21 ` Pip Cet
2019-06-28 19:11 ` Bruno Haible
2019-06-28 21:07 ` Pip Cet
2019-06-28 23:30 ` Bruno Haible
2019-06-29 5:40 ` Paul Eggert
2019-06-29 5:44 ` Pip Cet
2019-06-29 10:31 ` Bruno Haible
2019-06-29 17:11 ` Paul Eggert
2019-06-29 17:48 ` Bruno Haible
2019-06-30 15:30 ` Pip Cet
2019-06-30 15:45 ` Bruno Haible
2019-07-02 23:39 ` Paul Eggert
2019-07-01 1:46 ` Richard Stallman
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
List information: https://lists.gnu.org/mailman/listinfo/bug-gnulib
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=5284eb58-3560-da42-d1d1-3bdb930eae49@cs.ucla.edu \
--to=eggert@cs.ucla.edu \
--cc=36370@debbugs.gnu.org \
--cc=bug-gnulib@gnu.org \
--cc=pipcet@gmail.com \
/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).