Re: [gentoo-user] Re: If is doesn't compile with -O3 -pipe is that not a BUG?

2003-08-14 Thread Robert Young


Thorsten Kampe wrote:

 * Robert Young (2003-08-08 18:39 +0200)
  Just a thought

  I submitted a bug against Mozilla
  http://bugs.gentoo.org/show_bug.cgi?id=26068

  It seems to be because of my CFLAGS
  CFLAGS=-march=pentium3 -O3 -pipe -fomit-frame-pointer

 These are *exactly* my flags and Mozilla compiled with those (okay the
 second time - I had to reduce MAKEOPTS from -j(3+2) to -j(3+1)

You have 3 or 4 cpus?

or your not using the recommended values?

Anyway the build finished with -O2 I am wondering if for some reason it
just dose not build the first 2 time.

I am trying CFLAGS=-march=pentium3 -O3 -pipe and see if that works and
if it does I will try
CFLAGS=-march=pentium3 -O3 -pipe  -fomit-frame-pointer and scratch my
head if it works. :)


 while using distcc; where m+n ist the ratio localhost to remote distcc
 host).

 Thorsten

 --
 [EMAIL PROTECTED] mailing list


--
[EMAIL PROTECTED] mailing list



[gentoo-user] Re: If is doesn't compile with -O3 -pipe is that not a BUG?

2003-08-09 Thread Van Gale
Robert Young wrote:
Just a thought

I submitted a bug against Mozilla
http://bugs.gentoo.org/show_bug.cgi?id=26068
It seems to be because of my CFLAGS
CFLAGS=-march=pentium3 -O3 -pipe -fomit-frame-pointer
I changed them to
CFLAGS=-march=pentium3 -O2
and it seems to be working so far.

I have:

CFLAGS=-O3 -march=pentium3 -pipe -fomit-frame-pointer

and the Mozilla build works for me.  But, Mozilla is one of the builds 
that overrides CFLAGS.

Here's a snippet from the 1.4-r3 ebuild:

# Crashes on start when compiled with -fomit-frame-pointer
filter-flags -fomit-frame-pointer
filter-flags -ffast-math
append-flags -s -fforce-addr
# Sparc support ...
replace-sparc64-flags
# Recently there has been a lot of stability problem in Gentoo-land.  Many
# things can be the cause to this, but I believe that it is due to gcc3
# still having issues with optimizations, or with it not filtering bad
# combinations (protecting the user maybe from himeself) yet.
#
# This can clearly be seen in large builds like glibc, where too aggressive
# CFLAGS cause the tests to fail miserbly.
#
# Quote from Nick Jones [EMAIL PROTECTED], who in my opinion
# knows what he is talking about:
#
#   People really shouldn't force code-specific options on... It's a
#   bad idea. The -march options aren't just to look pretty. They enable
#   options that are sensible (and include sse,mmx,3dnow when apropriate).
#
# The next command strips CFLAGS and CXXFLAGS from nearly all flags.  If
# you do not like it, comment it, but do not bugreport if you run into
# problems.
#
# [EMAIL PROTECTED] (13 Oct 2002)
strip-flags
# We set -O in ./configure to -O1, as -O2 cause crashes on startup ...
# (bug #13287)
export CFLAGS=${CFLAGS//-O?}
export CXXFLAGS=${CFLAGS//-O?}
My question is:
Should it not be one of Gentoo's goals to make even optimized
systems stable.  info2flags on my computer would have produced
Probably, and that's probably why some ebuilds override CFLAGS.

CFLAGS=-march=pentium3 -O3 -pipe
I believe.
This goes against what I keep seeing people say that only
CFLAGS=-march=pentium3 -O2  would be stable.
-pipe apparently only affects gcc performance, not generated code.

I've built my entire system with -O3 -march=pentium3 
-fomit-frame-pointer -pipe and don't seem to have any stability 
problems.  Some of the minor problems I have are binary package related 
(like nvidia drivers).

I did have a stability problem with Mozilla when I compile with 
USE=gtk2 but that problem goes away with USE=-gtk, and either way it 
completes the ebuild just fine.



--
[EMAIL PROTECTED] mailing list


[gentoo-user] Re: If is doesn't compile with -O3 -pipe is that not a BUG?

2003-08-08 Thread Thorsten Kampe
* Robert Young (2003-08-08 18:39 +0200)
 Just a thought

 I submitted a bug against Mozilla
 http://bugs.gentoo.org/show_bug.cgi?id=26068

 It seems to be because of my CFLAGS
 CFLAGS=-march=pentium3 -O3 -pipe -fomit-frame-pointer

These are *exactly* my flags and Mozilla compiled with those (okay the
second time - I had to reduce MAKEOPTS from -j(3+2) to -j(3+1)
while using distcc; where m+n ist the ratio localhost to remote distcc
host).

Thorsten


--
[EMAIL PROTECTED] mailing list