Stage 1/3 Installation Guide for 2005.0 and GCC 3.4.3

[SVN]

@ag_x:

At the time that the author has determined that the Guide is completed and is ready for distirubtion, the author will create an Official PDF for release under a GPL-type copyright. Until that happens, the production of 3rd party PDF documents for distribution is expressly denied.

Read first the topic please.

[kimchi_sg]

[comes out of hiatus]

Marking this for future reference.

@Bob P: You may wish to note that this install method is only useful for 32-bit x86 platforms and any other platform that has gcc 3.3.5 in the stable branch. A check with packages.g.o shows this:

Code: Select all

Releases	alpha	amd64	arm	hppa	ia64	mips	ppc	ppc64	ppc macos	s390	sh	sparc	x86
4.0.0_beta20050402	-	-	-	-	-	-	-	-	-	-	-	-	-
3.4.3.20050110-r2	-	~	-	-	~	~	~	~	-	-	-	~	~
3.4.3.20050110-r1	-	~	-	-	~	~	~	~	-	-	-	~	~
3.4.3-r1	-	+	-	-	-	+	+	+	-	-	-	~	~
3.4.1-r3	-	~	-	-	-	+	+	+	-	+	-	-	~
3.3.5.20050130-r2	~	~	~	~	-	~	-	-	-	-	~	~	~

So looking at the above table, users of the amd64, mips, ppc, ppc64 (gcc 3.4.3-r1 in stable) and ia64 (gcc-3.4.1-r3 in stable) should not waste their time convoluting their installs by using this guide.

[crawls back into his shell]

[atavus]

Hi!

I'd like to add a little tip about userlocales and glibc.
Emerging glibc for both nptl and ithreads does one unwanted thing, it overwrites the locales.build file with a standart one (in the 2nd run), so it is a good thing to do chmod 444 /etc/locales.build, after editing it to match our needs. (or "config protect it")

[keltor]

[comes out of hiatus]

Marking this for future reference.

@Bob P: You may wish to note that this install method is only useful for 32-bit x86 platforms and any other platform that has gcc 3.3.5 in the stable branch. A check with packages.g.o shows this:

Code: Select all

Releases	alpha	amd64	arm	hppa	ia64	mips	ppc	ppc64	ppc macos	s390	sh	sparc	x86
4.0.0_beta20050402	-	-	-	-	-	-	-	-	-	-	-	-	-
3.4.3.20050110-r2	-	~	-	-	~	~	~	~	-	-	-	~	~
3.4.3.20050110-r1	-	~	-	-	~	~	~	~	-	-	-	~	~
3.4.3-r1	-	+	-	-	-	+	+	+	-	-	-	~	~
3.4.1-r3	-	~	-	-	-	+	+	+	-	+	-	-	~
3.3.5.20050130-r2	~	~	~	~	-	~	-	-	-	-	~	~	~

So looking at the above table, users of the amd64, mips, ppc, ppc64 (gcc 3.4.3-r1 in stable) and ia64 (gcc-3.4.1-r3 in stable) should not waste their time convoluting their installs by using this guide.

[crawls back into his shell]

But were these included in the build for 2005.0? Now amd64 users shouldn't use the 2005.0 cd anyway ... there's another cd referenced in the install forum that is based on 3.4 and works much better all around.

[Bob P]

Hi!

I'd like to add a little tip about userlocales and glibc.
Emerging glibc for both nptl and ithreads does one unwanted thing, it overwrites the locales.build file with a standart one (in the 2nd run), so it is a good thing to do chmod 444 /etc/locales.build, after editing it to match our needs. (or "config protect it")

yes, you are right! i have observed that occurrence and i had forgotten to make mention of it. thanks.

EDIT: reviewing this situation, here's the real answer: during the course of the installation, you are NEVER advised to run etc-update. if you don't run etc-update, you will never experience the problems that you've described.

OTOH, if you do run etc-update, you are responsible for knowing what you are doing. if you choose to deviate from the guide by running etc-update, and you choose to allow my configuration files to become overwritten, then you are responsible for the outcome.

[kimchi_sg]

@Bob P: You may wish to note that this install method is only useful for 32-bit x86 platforms and any other platform that has gcc 3.3.5 in the stable branch....
So looking at the above table, users of the amd64, mips, ppc, ppc64 (gcc 3.4.3-r1 in stable) and ia64 (gcc-3.4.1-r3 in stable) should not waste their time convoluting their installs by using this guide.

But were these included in the build for 2005.0?

Critical Package Updates
Important package-versions in this release:

• gcc 3.4.3-r1

Critical Package Updates
Important package-versions in this release:

• gcc 3.4.1-r3

Now amd64 users shouldn't use the 2005.0 cd anyway ...

And why would that be so? The versions of any software on the install CD has no bearing whatsoever on whether the toolchain in the stageball is the latest one or not, if you follow the instructions in this tutorial.

there's another cd referenced in the install forum that is based on 3.4 and works much better all around.

Please email the gentoo.org developers so that they can provide a link to this wonderful and safe install CD.

Also, I believe the original poster of this thread has good feelings about people installing their systems from unofficial installation media. Especially this liveDVD.

[keltor]

[SNIP]

Also, I believe the original poster of this thread has good feelings about people installing their systems from unofficial installation media. Especially this liveDVD. [/quote]

Well I don't know anything about those liveCDs, but I do know quite a few x86_64 users that have to use the ahorn liveCDs. There's been lots of fun with liveCDs lately ... x86_64 2005.0 cds originally had a bad bootsector and lots of x86 users cannot mount partitions unless they disable lvm2.

[Bob P]

[comes out of hiatus]

Marking this for future reference.

@Bob P: You may wish to note that this install method is only useful for 32-bit x86 platforms and any other platform that has gcc 3.3.5 in the stable branch.

making the changes that you have suggested are not at all necessary. this installation method has always been exclusively for x86 architectures. only a brain dead idot would assume that its suitable for AMD64, mips, ppc, or ppc64. they are totally different architectures and are not based on x86, so no revisions to the Guide are warranted.

Guys, if we could please keep this thread focused on the Stage 1/3 Installation Guide, the x86 Architecture its written for, and the Gentoo 2005.0 Installation Media that is required for the install. Discussion about alternative architectures and alternative installation media belongs somewhere else.

[Sade]

Bob, i had this problem about /dev/null permissions, and while searching to find a fix, i found this thread:
Permission Denied on /dev/null ? [SOLVED]
i noticed that all the people in that thread used your 2005.0 1on3 install method.

I don't know, but there might be a relation. I thought u would be interrested.

[mesut]

I fallowed the guide, and everything went fine, thanks. But i 've some questions.
Since we didnt use nptlonly, glibc have both nptl and linuxthreads, but i thought nptl'ld be the default one.
But;

Code: Select all

 $ /lib/libc.so.6
GNU C Library stable release version 2.3.4, by Roland McGrath et al.
Copyright (C) 2004 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.
There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A
PARTICULAR PURPOSE.
Compiled by GNU CC version 3.4.3-20050110 (Gentoo Linux 3.4.3.20050110-r1, ssp-3.4.3.20050110-0, pie-8.7.7).
Compiled on a Linux 2.6.8 system on 2005-04-01.
Available extensions:
        GNU libio by Per Bothner
        crypt add-on version 2.1 by Michael Glad and others
        linuxthreads-0.10 by Xavier Leroy
        The C stubs add-on version 2.1.2.
        BIND-8.2.3-T5B
        NIS(YP)/NIS+ NSS modules 0.19 by Thorsten Kukuk
        Glibc-2.0 compatibility add-on by Cristian Gafton
        GNU Libidn by Simon Josefsson
        libthread_db work sponsored by Alpha Processor Inc
Thread-local storage support included.
For bug reporting instructions, please see:
<http://www.gnu.org/software/libc/bugs.html>.
 $

and here is my USE ;

Code: Select all

 # emerge --info | grep USE
USE="x86 3dnow X acpi alsa apache2 avi berkdb bitmap-fonts cdr crypt curl divx4linux dvd encode esd fam flac gdbm gif gnome gpm gtk gtk2 imagemagick imlib ithreads java jpeg libwww mad mikmod mmx motif mp3 mpeg mysql ncurses nls nptl ogg oggvorbis opengl pam pcmcia pdflib perl php png pthreads python quicktime readline samba sdl sse svga tcpd tiff truetype truetype-fonts type1-fonts unicode usb vhosts vorbis wifi win32codecs xml2 xmms xosd xv xvid zlib"
#

Doesnt that mean, nptl is not the default ?

[moocha]

Bob, i had this problem about /dev/null permissions, and while searching to find a fix, i found this thread:
Permission Denied on /dev/null ? [SOLVED]
i noticed that all the people in that thread used your 2005.0 1on3 install method.

I don't know, but there might be a relation. I thought u would be interrested.

That is a known Gentoo gcc / binutils bug that has been solved in gcc-3.4.3.20050110-r2.

Code: Select all

*gcc-3.4.3.20050110-r2 (08 Apr 2005)

  08 Apr 2005; Mike Frysinger <vapier@gentoo.org>
  +gcc-3.4.3.20050110-r2.ebuild:
  Add fix for PR/16625 from upstream #88022 by Ed Catmur. Patch to prevent
  /dev/null from being deleted #79836 by David Wood.

For more info see http://bugs.gentoo.org/show_bug.cgi?id=79836.

[Bob P]

I fallowed the guide, and everything went fine, thanks. But i 've some questions.
Since we didnt use nptlonly, glibc have both nptl and linuxthreads, but i thought nptl'ld be the default one.
But; ...

this question was been asked and answered as it pertains to the Stage 1/3 install method back in January. the answer is also available in the message bases if you're willing to search for it.

rather than reposting the same information twice, i'll refer you to the search feature or to the Stage 1/3 Install thread for 2004.3.

HINT: you're looking in the wrong place.

[juazp]

it works fine on 64 platforms as long as u know wich steps to skip and what to switch around , this install method allways resulted in great performance / stability in the end so i will keep using it untill i see something better

[Bob P]

it works fine on 64 platforms as long as u know wich steps to skip and what to switch around , this install method allways resulted in great performance / stability in the end so i will keep using it untill i see something better

well, if you know which steps to skip and what to switch around, any installation method will work on any platform. the bottom line is that this install guide is not written for AMD64 systems.

[juazp]

yeah i got it , but i love this install metod and just reporting it works good as long as u got common sense , and thanks alot for this wonderful guide =)

[coffeeman]

Love this howto... my pc rox now...

[Bob P]

PDF version of the Guide now available.

Click Here.

[coffeeman]

Hello Bob P,

Thank you very mutch for this great howto. I´m wrinting "Howto install Gentoo on Toshiba Satellite P10-804 Notebook". Can I use your howto on my??? It will be Creative Commons as well, but i need to insert some Kernel Configs and Xorg Configs. I need your permission to do this because your work is a "No Derivative Works" copyright. You will recive all the credit too.

Txs again

[roki942]

Hi Bob, This rocks!

Following your pdf version and came across something I wasn't sure about so I checked it against here and found a few differences.

6.5 Configuring the USE Flags, Portage Options, and Compile Options: /etc/make.conf

PDF
CFLAGS="-O2 -march=pentium -pipe"
here
CFLAGS="-O2 -march=pentium -fomit-frame-pointer -pipe"

PDF is missing
PORTAGE_TMPFS="/dev/shm"

PDF
FEATURES="candy ccache distlocks sandbox"
here
FEATURES="ccache distlocks sandbox userpriv usersandbox"
It was the "candy" that got me looking.

6.7.1 Activate the userlocales USE flag for glibc is missing

Code: Select all

sys-devel/libperl ithreads
dev-lang/perl ithreads

7.2.1 Updating make.conf
PDF
CFLAGS="-O3 -march=pentium -mtune=petium -fforce-addr -momit-leaf-frame-pointer -fomit-frame-pointer -ftracer -pipe"
here
CFLAGS="-O3 -march=pentium -fforce-addr -momit-leaf-frame-pointer -fomit-frame-pointer -ftracer -pipe"

and 7.2.5 Summary Step 3 has it as gcc-config 2

PDF is missing: 10.2.5 Check Our Results (the complete section)

This is a really great guide ...... when next I have time to continue I'll be building the world.

[Bob P]

thanks for the feedback. i have to admit, the PDF is not a word for word transcription of the first post on this thread, but the overall installtion procedure remains the same.

because the 2005.0 PDF was updated from the 2004.3 version's PDF, there were a few errors that crept into the 2005.0 revision of the document. thanks for pointing them out!

fwiw, i cannot find "petium" anywhere in the documents. are you sure about that?

section 10.2.5 was deleted on purpose. if you like, perform those steps after rebooting.

most of the discrepancies that you've mentioned are only clerical errors and typos. i'm not too worried asbout them, as i think that most people would be able to recognize them and deal with when they encounter them. the only things that i think are really important to mention here are the FEATURES statements regarding userpriv and usersandbox, and ithreads as a USE flag.

if you read the support thread, some people have had problems with userpriv and usersandbox with certain ebuilds; problematic ebuilds are why i left those statements out of the 2004.3 version of the guide, but as the problem seemed to have been resolved with the offensive ebuilds, i tried putting them back into the 2005.0 version. ebuilds being what they are, some crappy ebuilds started causing problems again. the solution is to just remove userpriv and usersandbox from the features statement if you should encounter problems. for the sake of preventing this problem, i left those statements out of the PDF on purpose.

ithreads is optional, so leaving ithreads out of package.use was a non-critical oversight. the contents of package.use have been fixed in today's revision to the PDF.

clerical errors like these are some of the reasons why i don't want derivative works based upon the Guide. i've mentioned my rationale for this before -- the Stage 1/3 Guide is a document in a state of transition, and i prefer that all of the information related to this installation method be retained in one centralized location, rather than being spread across a number of documents. this makes maintenance of the project alot simpler.

thanks again for your help.

[Bob P]

updated PDF now available.

[coffeeman]

Bob P,

Autho ?

[moocha]

Bob, just found another argument for the two emerge -e system thing. Had the time of my life tracking down an e2fsprogs bug that turned out to be caused by sloppy programming in e2fsprogs. A source file included some package-specific headers with the #include <file.h> (system include path first) form instead of the #include "file.h" (private include path) form. As it happened, a header file with that same name had been left over from a (much) older e2fsprogs install, got picked up in the compile and provided incorrect values for some constants. Two emerge -e system don't allow this to happen since the second emerge would have picked up the current file in this case, because the first pass would have overwritten the old one.
I think it would be wise to switch to the two-pass version.

Edit: s/private/package-specific/

[Bob P]

agreed. all things considered, the two pass version is what we've been doing to build the Jackass! Project, so its probably worth putting into the Stage 1/3 docs as well. thanks again for keeping an eye on this!

until i get a chance to update the Guide: everyone please replace the "emerge -e system" statement in the Guide with "emerge -e system && emerge -e system".

[Hobbit_HK]

Hey Bob! Great guide
But... Don't you think we should do "etc-update" or "dispatch-conf" before we start editing the rc config files? Before we do this they'll be in their old version, and will be replaced later when we WILL do dispatch-conf...