Tweaked glibc 2.4 + 2.5 + snapshot ebuilds

Message

wing · Post by **wing** » Thu Apr 27, 2006 8:39 pm

could someone on an amd64 system please quickpkg their glibc and upload it somewhere? I desparately need it, as my multilib is messed up. If you used a newer binutils than 2.16-r2, might as well quickpkg that too.

If hosting's an issue, I can accomidate.

Joffer · Post by **Joffer** » Thu Apr 27, 2006 9:43 pm

wing wrote:could someone on an amd64 system please quickpkg their glibc and upload it somewhere? I desparately need it, as my multilib is messed up. If you used a newer binutils than 2.16-r2, might as well quickpkg that too.

If hosting's an issue, I can accomidate.

Here are mine: http://www.joffer.net/webshare/linux/gentoo/tempfiles/
It's glibc-2.4-r1 and binutils-2.16.91.0.6. They are built with these CFLAGS:

Code: Select all

CFLAGS="-march=k8 -O2 -pipe -mfpmath=387 -fweb -funsafe-math-optimizations -fno-align-labels -fno-align-loops -fno-align-functions -fno-align-jumps -mmmx -m3dnow -msse -msse2 -s"
CXXFLAGS="${CFLAGS}"
LDFLAGS="-Wl,-O1 -Wl,--sort-common -Wl,--enable-new-dtags -Wl,-Bdirect -Wl,-hashvals -Wl,-zdynsort"

. I'll leave the files there for a day or two. Please let me know when you have downloaded them.

wing · Post by **wing** » Fri Apr 28, 2006 4:21 am

Joffer wrote:
wing wrote:could someone on an amd64 system please quickpkg their glibc and upload it somewhere? I desparately need it, as my multilib is messed up. If you used a newer binutils than 2.16-r2, might as well quickpkg that too.

If hosting's an issue, I can accomidate.
Here are mine: http://www.joffer.net/webshare/linux/gentoo/tempfiles/
It's glibc-2.4-r1 and binutils-2.16.91.0.6. They are built with these CFLAGS:
Code: Select all
CFLAGS="-march=k8 -O2 -pipe -mfpmath=387 -fweb -funsafe-math-optimizations -fno-align-labels -fno-align-loops -fno-align-functions -fno-align-jumps -mmmx -m3dnow -msse -msse2 -s"
CXXFLAGS="${CFLAGS}"
LDFLAGS="-Wl,-O1 -Wl,--sort-common -Wl,--enable-new-dtags -Wl,-Bdirect -Wl,-hashvals -Wl,-zdynsort"
. I'll leave the files there for a day or two. Please let me know when you have downloaded them.

You can't believe how grateful I am. Grabbed them!

edit: hmm, by any chance, could you upload your gcc (4.1 I assume)?

Joffer · Post by **Joffer** » Fri Apr 28, 2006 11:39 am

wing wrote:
Joffer wrote:
wing wrote:could someone on an amd64 system please quickpkg their glibc and upload it somewhere? I desparately need it, as my multilib is messed up. If you used a newer binutils than 2.16-r2, might as well quickpkg that too.

If hosting's an issue, I can accomidate.
Here are mine: http://www.joffer.net/webshare/linux/gentoo/tempfiles/
It's glibc-2.4-r1 and binutils-2.16.91.0.6. They are built with these CFLAGS:
Code: Select all
CFLAGS="-march=k8 -O2 -pipe -mfpmath=387 -fweb -funsafe-math-optimizations -fno-align-labels -fno-align-loops -fno-align-functions -fno-align-jumps -mmmx -m3dnow -msse -msse2 -s"
CXXFLAGS="${CFLAGS}"
LDFLAGS="-Wl,-O1 -Wl,--sort-common -Wl,--enable-new-dtags -Wl,-Bdirect -Wl,-hashvals -Wl,-zdynsort"
. I'll leave the files there for a day or two. Please let me know when you have downloaded them.
You can't believe how grateful I am. Grabbed them!

edit: hmm, by any chance, could you upload your gcc (4.1 I assume)?

Sure. gcc-4.1.0 is uploaded to the same place.

energyman76b · Post by **energyman76b** » Fri Apr 28, 2006 3:33 pm

Joffer wrote: Here are mine: http://www.joffer.net/webshare/linux/gentoo/tempfiles/
It's glibc-2.4-r1 and binutils-2.16.91.0.6. They are built with these CFLAGS:
Code: Select all
CFLAGS="-march=k8 -O2 -pipe -mfpmath=387

stupid. mfpmath=387? you want to slow down math, do you?

Joffer wrote:

Code: Select all

 -fweb -funsafe-math-optimizations -fno-align-labels -fno-align-loops -fno-align-functions -fno-align-jumps -mmmx -m3dnow -msse -msse2 -s"
CXXFLAGS="${CFLAGS}"

.

-funsafe-math-optimizations extrem stupid, no-align-loops, stupd, no-align-* all stupid, mmx, m3dnow superflous and stupid

You haven NEVER read man gcc, do you?

Joffer wrote:
Code: Select all
LDFLAGS="-Wl,-O1 -Wl,--sort-common -Wl,--enable-new-dtags -Wl,-Bdirect -Wl,-hashvals -Wl,-zdynsort"
. I'll leave the files there for a day or two. Please let me know when you have downloaded them.

and even more stupid flags. I am really surprised, that ANYTHINGS works.

Please, everybody, don't be stupid, don't use such carzy flags and don't download his glibc - it is just a big pile of madness.

Btw, you can take the glibc from the livecd, if nothing works. At least, it should be SAFE.

FireBurn · Post by **FireBurn** » Fri Apr 28, 2006 4:18 pm

CFLAGS="-march=athlon-xp -Os -fomit-frame-pointer -mfpmath=sse,387 -frename-registers"
CHOST="i686-pc-linux-gnu"
CXXFLAGS="${CFLAGS} -fvisibility-inlines-hidden"
LDFLAGS="-Wl,-O1 -Wl,--sort-common -Wl,--as-needed"

wing · Post by **wing** » Fri Apr 28, 2006 4:51 pm

^^^ -fvisibility-inlines-hidden is bound to run into some headaches.
Got it too. It turned out the problem I was happening is that /lib was its own directory and not symlinked to /lib64, doh. Thanks a lot anyways

By the way, I think his glibc would still be safe as the ebuilds only keep the sanest cflags, this is also the case with gcc.

I might as well post my flags:

Code: Select all

CFLAGS="-march=opteron -Os -pipe -ftree-vectorize -fweb"
LDFLAGS="-Wl,-O1 -Wl,--sort-common -Wl,-Bdirect -Wl,-hashvals -Wl,-zdynsort"

Joffer · Post by **Joffer** » Sat Apr 29, 2006 2:53 pm

energyman76b wrote:and even more stupid flags. I am really surprised, that ANYTHINGS works.

Please, everybody, don't be stupid, don't use such carzy flags and don't download his glibc - it is just a big pile of madness.

Btw, you can take the glibc from the livecd, if nothing works. At least, it should be SAFE.

Well, all works, even Xgl. But I know I've got some flags to clean.. My files are removed now. Just wanted to help the guy out. He should recompile his own when he get the system up..

energyman76b wrote:[mfpmath=387? you want to slow down math, do you?
-funsafe-math-optimizations extrem stupid, no-align-loops, stupd, no-align-* all stupid, mmx, m3dnow superflous and stupid

Could you please also elaborate why they are stupid flags, for us who don't know? The LDFLAGS are taken from some of these posts about glibc and binutils overlay, to have Bdirect and hashval enabled..

energyman76b · Post by **energyman76b** » Sun Apr 30, 2006 1:29 am

Directly from man gcc:

-funsafe-math-optimizations
Allow optimizations for floating-point arithmetic that (a) assume
that arguments and results are valid and (b) may violate IEEE or
ANSI standards. When used at link-time, it may include libraries
or startup files that change the default FPU control word or other
similar optimizations.

This option should never be turned on by any -O option since it can
result in incorrect output for programs which depend on an exact
implementation of IEEE or ISO rules/specifications for math func-
tions.

or with simpler words: don't use it, except you exactly know, what you are doing (for example you wrote a piece of software that benefits from it), because it will break the packages in sublte and non subtle ways. From crash to just incorrect results.

-mfpmath=unit
Generate floating point arithmetics for selected unit unit. The
choices for unit are:

387 Use the standard 387 floating point coprocessor present major-
ity of chips and emulated otherwise. Code compiled with this
option will run almost everywhere. The temporary results are
computed in 80bit precision instead of precision specified by
the type resulting in slightly different results compared to
most of other chips. See -ffloat-store for more detailed
description.

This is the default choice for i386 compiler.

sse Use scalar floating point instructions present in the SSE
instruction set. This instruction set is supported by Pentium3
and newer chips, in the AMD line by Athlon-4, Athlon-xp and
Athlon-mp chips. The earlier version of SSE instruction set
supports only single precision arithmetics, thus the double and
extended precision arithmetics is still done using 387. Later
version, present only in Pentium4 and the future AMD x86-64
chips supports double precision arithmetics too.
For the i386 compiler, you need to use -march=cpu-type, -msse
or -msse2 switches to enable SSE extensions and make this
option effective. For the x86-64 compiler, these extensions
are enabled by default.

The resulting code should be considerably faster in the major-
ity of cases and avoid the numerical instability problems of
387 code, but may break some existing code that expects tempo-
raries to be 80bit.

This is the default choice for the x86-64 compiler.

you see: on non x86-64 cpus the additional correctness may bring problems, but on x86-64 it is default - and sse is faster. So don't switch away from sane, good defaults.

For the rest of your flags: man gcc.

That your complete system is not hosed, has to be credited to 'filter-flags' and 'strip-flags' - some functions added to portage and a lot of ebuilds, because people like YOU are setting dangerous and/or stupid flags, and whined after that, that their system were broken.

Sadly, this two features are stripping almost everything - ruining it for everybody. Thanks again - not!

Joffer · Post by **Joffer** » Sun Apr 30, 2006 5:02 am

energyman76b wrote:...
you see: on non x86-64 cpus the additional correctness may bring problems, but on x86-64 it is default - and sse is faster. So don't switch away from sane, good defaults.

For the rest of your flags: man gcc.

That your complete system is not hosed, has to be credited to 'filter-flags' and 'strip-flags' - some functions added to portage and a lot of ebuilds, because people like YOU are setting dangerous and/or stupid flags, and whined after that, that their system were broken.

Sadly, this two features are stripping almost everything - ruining it for everybody. Thanks again - not!

About the math flag.. I had set it to 387 because I had read several places that AMD was faster with it than sse..
And.. I have not/don't complain about my system being broken since I know I have unusual/crazy flags set.

I do appreciate being told about wrong/stupid flags, but I would like to be told in a civilized way... there is no need to act angry and call people stupid etc.

energyman76b · Post by **energyman76b** » Sun Apr 30, 2006 12:49 pm

You are right, I was rude. Sorry.

But in my defense: I have seen so many posts with bad flags, and so many times someone told this peolpe about them, I got really, really tired to see another set of dangerous flags..

Plus, as I wrote - flags like yours are the reason that filter-flags and strip-flags exists. And this two functions ruin the fun for everybody

anih · Post by **anih** » Sun Apr 30, 2006 7:34 pm

nxsty when we can expect new snapshot?

Joffer · Post by **Joffer** » Sun Apr 30, 2006 11:32 pm

energyman76b wrote:You are right, I was rude. Sorry.

No problem

energyman76b wrote:But in my defense: I have seen so many posts with bad flags, and so many times someone told this peolpe about them, I got really, really tired to see another set of dangerous flags..

Plus, as I wrote - flags like yours are the reason that filter-flags and strip-flags exists. And this two functions ruin the fun for everybody

Well, you are probably very correct. I'm about to recompile my system soon with saner flags, since I've upgraded from a AMD64 3200+ to AMD64 X2 4200+ which has SSE3 instructions, so I want to utilize it where I can.

Code: Select all

CFLAGS="-march=k8 -O2 -pipe -msse3"

I was also thinking about adding (keeping) -fweb. Any experience/knowledge about the flag? It seems to be an ok flag to set, as long as I'm not doing much debugging:

Code: Select all

-fweb

Constructs webs as commonly used for register allocation purposes and assign each web individual pseudo register. This allows the register allocation pass to operate on pseudos directly, but also strengthens several other optimization passes, such as CSE, loop optimizer and trivial dead code remover. It can, however, make debugging impossible, since variables will no longer stay in a "home register". 

Enabled at levels -O3.

[/size]

energyman76b · Post by **energyman76b** » Mon May 01, 2006 4:56 am

Hi,

well, I am using -fweb too, and it did not hurt so far or -ftracer:
-ftracer
Perform tail duplication to enlarge superblock size. This trans-
formation simplifies the control flow of the function allowing
other optimizations to do better job.

I don't know, if it really helps, but I have so far found nothing that implys that this flag is hurtfull. Maybe some flag-cracks can say something about it

wrobel · Post by **wrobel** » Mon May 01, 2006 6:45 am

Hi,

I added the overlay(s) to layman. You can now

Code: Select all

emerge layman
layman -f -a nxsty-binutils -a nxsty-glibc

in order to install the overlay.

Avuton Olrich · Post by **Avuton Olrich** » Mon May 01, 2006 8:53 am

All ebuilds lead to

Code: Select all

http://snigel.no-ip.com/~nxsty/linux/glibc-2.4-patches-1.5.tar.bz2

but that file no longer exists. Any hints?

enderandrew · Post by **enderandrew** » Mon May 01, 2006 12:04 pm

http://enderandrew.com/nxsty/glibc-2.4- ... .5.tar.bz2 should work.

Xake · Post by **Xake** » Mon May 01, 2006 2:15 pm

A comment on the linux-headers-2.6.16 things

https://bugs.gentoo.org/show_bug.cgi?id=130225

Is there anyone more having this problem?

Avuton Olrich · Post by **Avuton Olrich** » Mon May 01, 2006 3:26 pm

enderandrew wrote:http://enderandrew.com/nxsty/glibc-2.4- ... .5.tar.bz2 should work.

No, that also is a broken link. Anyone else?

Raftysworld · Post by **Raftysworld** » Mon May 01, 2006 4:00 pm

Just curious, what happened to the userlocales flag in 2.4-r2?

immudium · Post by **immudium** » Mon May 01, 2006 4:11 pm

Xake wrote:A comment on the linux-headers-2.6.16 things

https://bugs.gentoo.org/show_bug.cgi?id=130225

Is there anyone more having this problem?

You got me curious as to why I was able to emerge mplayer with 2.6.16 headers without a problem. After re-emerging a half dozen times with different use flags, it looks like it's due to the rtc use flag in the ebuild. I guess it's not much of a consolation, but mplayer has always worked just fine for me using the emulated rtc clock. Hopefully the fix isn't too difficult. I don't quite understand the relationship between glibc and linux-headers

nxsty · Post by **nxsty** » Mon May 01, 2006 4:28 pm

wrobel wrote:Hi,

I added the overlay(s) to layman. You can now
Code: Select all
emerge layman
layman -f -a nxsty-binutils -a nxsty-glibc
in order to install the overlay.

Nice!

nxsty · Post by **nxsty** » Mon May 01, 2006 4:29 pm

Raftysworld wrote:Just curious, what happened to the userlocales flag in 2.4-r2?

It was replaced with the debian locale-gen utility. Now you don't have to rebuild glibc anymore to generate locales.

nxsty · Post by **nxsty** » Mon May 01, 2006 4:30 pm

anih wrote:nxsty when we can expect new snapshot?

I have on ready but the latest patch-tarball 1.9 is not up on the mirrors for some reason.

energyman76b · Post by **energyman76b** » Tue May 02, 2006 1:58 am

Hi,

and where do I get/find this 'locale-gen' utility?

Tweaked glibc 2.4 + 2.5 + snapshot ebuilds

Download nogo nomo

Glibc file nogonomo