Stage 1/3 Installation Support - Gentoo 2004.3 & GCC 343

[kimchi_sg]

FWICT it seems that the vast majority (if not all) of the complaints that have been solved by the /proc/mounts fix have been encountered by people who were attempting to adapt the Guide to install Gentoo on hardware that the Guide does not support.

That is not true! Please read my PM.

Also, with this post the support thread is now 1/3 the length of the original thread.

EDIT: Just finished a rebuild on the same machine, and it seems that I can't reproduce the "won't boot" problem - there's only a scary-looking error message to the effect of

Code: Select all

Can't create /etc/mtab~526: Read-only file or directory

but bootup continues as per normal.

The only difference is that on my previous install, I did modify /etc/conf.d/rc to accomodate udev according to the tutorial before rebooting, whereas this time around I had forgotten to do so.

[kimchi_sg]

To those whose emerge -e system halted on dev-lang/perl, like mine just did:

Code: Select all

emerge --sync && emerge --resume

[Bob P]

your box halted on perl? interesting. IIRC i had a box do that yesterday, and it seemed that GCC was using the old CFLAGS instead of the 3.4.3-compliant CFLAGS -- as if the modification of the CFLAGS didn't take. could we both have made typos in our one-liners? or is there some other problem with ebuilds that's popping-up?

fwiw i've just checked on the P3 that I had running the modified big one-liner -- it took 17:30 on a P3-800/256MB.

[Bob P]

FWICT it seems that the vast majority (if not all) of the complaints that have been solved by the /proc/mounts fix have been encountered by people who were attempting to adapt the Guide to install Gentoo on hardware that the Guide does not support.

That is not true! Please read my PM.

i have never been able to reproduce the problem. based on what i've been able to find in the forums, that problem seems to be occurring with SCSI and SATA installations and with some EIDE installations, but not on EIDE installations where the user remembers to process the necessary files in need of updating in /etc.

[kimchi_sg]

your box halted on perl? interesting. IIRC i had a box do that yesterday, and it seemed that GCC was using the old CFLAGS instead of the 3.4.3-compliant CFLAGS -- as if the modification of the CFLAGS didn't take. could we both have made typos in our one-liners? or is there some other problem with ebuilds that's popping-up?

There was breakage to the perl ebuild yesterday, and the perl package maintainer admitted as much on planet Gentoo. It is bug no. 84868 (which has a nice to remember number for it. )

[opensourceguy]

You know, this has to be the most frustrating Gentoo install I have had. I have had a few [re-installing it on the same box a few times]. However, I couldn't get it to install per reg. install guides on this box. So, I went to stage 1, couldn't get to bootstrap. Then found the stage 1 on 3 install. Started that last week, and I am still no further ahead then where I started. I have started with a clean slate about 3 times on just the 1 on 3 install. This has to be some of the most frustrating stuff here. Anyways, now that I have vented a little, onto my problems. . .

I get to step 7.1 [emerge gcc-config glibc binutils gcc], and that goes smoothly. Then, I get to [emerge -e system] and it fails at the app 'db' with the error:

Code: Select all

Checking to see if C Compiler can create Executables:  Error: C Comiler cannot create Executables

I have my make.conf the exact same as per guide, except my march=athlonxp. At the time, I also had 'nptlonly' in my USE flags, but now I am starting 7.1 over, without nptlonly, and see if that helps any [doubtful].

I'll post my emerge info when step 7.1 finishes. I know I can start a new session in the LiveCD, but my screen becomes all off-center, and I don't like messing with my Monitor settings.

Also a little off-topic question doing "hdparm -tT /dev/sda" gives me 7xx mb/s on the timing cached reads. Is that normal? It seems so high. And on this box [500mhz, ATA100 hdd] I get 350mb/s on the timing cached read. I would hardly call this a performance machine, btw.


-justin

[kimchi_sg]

I get to step 7.1 [emerge gcc-config glibc binutils gcc], and that goes smoothly. Then, I get to [emerge -e system] and it fails at the app 'db' with the error:

Code: Select all

Checking to see if C Compiler can create Executables:  Error: C Comiler cannot create Executables

I have my make.conf the exact same as per guide, except my march=athlonxp.

http://forums.gentoo.org/viewtopic-t-294109.html

/me hugs Maedhros for writing such a wonderful FAQ.

[Bob P]

your box halted on perl? interesting. IIRC i had a box do that yesterday, and it seemed that GCC was using the old CFLAGS instead of the 3.4.3-compliant CFLAGS -- as if the modification of the CFLAGS didn't take. could we both have made typos in our one-liners? or is there some other problem with ebuilds that's popping-up?

There was breakage to the perl ebuild yesterday, and the perl package maintainer admitted as much on planet Gentoo. It is bug no. 84868 (which has a nice to remember number for it. )

well, i guess that a broken ebuild is more likely than both of us making the same misteak at the same time!

[Bob P]

You know, this has to be the most frustrating Gentoo install I have had. I have had a few [re-installing it on the same box a few times]. However, I couldn't get it to install per reg. install guides on this box. So, I went to stage 1, couldn't get to bootstrap. Then found the stage 1 on 3 install. Started that last week, and I am still no further ahead then where I started. I have started with a clean slate about 3 times on just the 1 on 3 install. This has to be some of the most frustrating stuff here.

Justin -- if you've had critical stops in following a number of different installation methods, it would be helpful to consider what the common denominator in your problem may be. The problem doesn't seem to be in the Stage 1 on 3 installation method if you can't even get a Stage 1 install to bootstrap. It seems that you have a bigger problem that's getting in your way.

As kimchi_sg had mentioned, CFLAG settings are the most common cause of the "C Compiler Cannot Create Executables" error. The Gentoo Fundamentals Page # 20, and the thread that is hyperlinked there, contain lots of helpful information that you should review. You absolutely must learn all of that basic information before you attempt a Stage 1 install or the more advanced Stage 1/3 install using this guide. Insofar as the Stage 1/3 installation method is recommended only for advanced users, I can see how it would be very frustrating for newer Gentoo users who haven't mastered the basics yet.

In addition to CFLAG problems, another problem that has to rate high on your index of suspicion is a typographical error in make.conf. Even if you have all the experience that you need to do this install, the mistake could be as simple as an overlooked typographical error in make.conf that is borking your compiler.

If you are trying to install on Athlon XP, you should review the GCC man page and review the athlon-specific compiler settings for make.conf. you may find something you've missed, and if you have the wrong compiler settings you won't be successful installing gentoo regardless of whether you're trying a Stage 1, a Stage 2, or a Stage 1/3 installtion. if you continue to have this many problems, then in the interim your best bet may be to start with a Stage 3. This would give you a functional system in a minimal amount of time. Then, once you have the system up and running, you can do the system rebuild at your leisure.

[powderific]

Well, I did a normal install to see if I could get it to work with my laptop at all and I figured out what went wrong. On my second install I configured my computer for Pentium M instead of Pentium4/Celeron, which probably did it, and on my first install I had the e1000 drivers compiled into the kernel instead of as a module. For some reason the computer doesn't like this and will fail to boot. (otherwise I used all the same kernel options in this installation and it works perfectly).

I do have windows on my first partition.

So now I have a normal install but I know exactly what I did wrong with the stage 1/3 install so I can do it right next time. (I don't think I have enough time to do it all over right now). Thanks for the help, I'll be doing this install on my desktop as soon as I get back to school (after working with it so much it should be a piece of cake!)

One question though, when I get time to reinstall on my laptop, is it necessary to start over completely or is there an easier way?

[Bob P]

One question though, when I get time to reinstall on my laptop, is it necessary to start over completely or is there an easier way?

if your installation problem was only related to in appropriate compiler parameters, you don't need to do a complete reinstallation. if you boot from the live CD and repeat the necessary mounts and chroot, then all you need to do is fix your make.conf and start over at Section 7. most people have trouble getting restarted, though, and for most people starting over at square one is alot less confusing.

[lpx]

Hi Bob,

I have follow your guide religiously to turn my PII - 300 mhz into a bullet.

I did no changes in the cflags and use flags the you use because i assume it was a very similar machine.

I'm on the 7.2.4 issue and i get an error and the compiling aborted.

I can't past now the error but i'll do it if it fail again.

i just put it running again but without the mtune flag at cflags

my question is...can i continue with the installation with an error at this stage?

can i use the things that came already compiled with the stage 3 tarball without merging the whole system?

cheers,

[kimchi_sg]

Hi Bob,

I have follow your guide religiously to turn my PII - 300 mhz into a bullet.

I did no changes in the cflags and use flags the you use because i assume it was a very similar machine.

A pentium 2 similar to a pentium classic?

Your safe CFLAGS are here: http://gentoo-wiki.com/Safe_Cflags#Pent ... 28Intel.29

I'm on the 7.2.4 issue and i get an error and the compiling aborted.

I can't past now the error but i'll do it if it fail again.

Well then, what is the error? Show us the last 10 lines of the compiling output before the compile aborted, exactly as was printed on the screen. That will be more helpful in finding out what the ailment is.

i just put it running again but without the mtune flag at cflags

my question is...can i continue with the installation with an error at this stage?

can i use the things that came already compiled with the stage 3 tarball without merging the whole system?

Don't try. It would be the Gentoo equivalent of Russian roulette, if you re-compiled part of your toolchain, then re-omitted to compile the rest of it, as well as all the programs in the system profile.

[lpx]

Hi again,

It seems that I was very wrong... I didn't realized that pentium 2 was so differente in terms of flags from the pentium. Until the 7.2.4 the flags i used were from the guide. After the error i only took out the mtune flag and put it running again the emerge -e system. By now the error didn't appear (It's already in something like 12 hours... more than it was before the error appear). So... if the error appears again the better thing i can do is to take out all the other optimization flags and leave only this ones and try again the emerge -e system?

Pentium II (Intel)

vendor_id  : GenuineIntel
cpu family  : 6
model  : 5
model name  : Pentium II (Deschutes)
stepping  : 2

CHOST="i686-pc-linux-gnu"
CFLAGS="-march=pentium2 -O3 -pipe -fomit-frame-pointer"
CXXFLAGS="-march=pentium2 -O3 -pipe -fomit-frame-pointer"

Many thx,

[kimchi_sg]

After the error i only took out the mtune flag and put it running again the emerge -e system. By now the error didn't appear (It's already in something like 12 hours... more than it was before the error appear). So... if the error appears again the better thing i can do is to take out all the other optimization flags and leave only this ones and try again the emerge -e system?

Because it seems that you like to keep saying "the error, the error, the error" without telling us what exactly your mysterious "error" is, we cannot not comment further until you tell us what the error exactly is.

[lpx]

After the error i only took out the mtune flag and put it running again the emerge -e system. By now the error didn't appear (It's already in something like 12 hours... more than it was before the error appear). So... if the error appears again the better thing i can do is to take out all the other optimization flags and leave only this ones and try again the emerge -e system?

Because it seems that you like to keep saying "the error, the error, the error" without telling us what exactly your mysterious "error" is, we cannot not comment further until you tell us what the error exactly is.

Yeah, you're right and i'm completly aware of that. But yesterday when it occured i couldnt do it because i had to get out, and i didn't want to waste the compute time so i just took out a flag and put it running again.

I'll past the error if it occurs again. thx anyway

[Bob P]

I posted some instructions early in the thread about how to post the relevant information that is required to be submitted for troubleshooting help requests.

Like kimchi_sg said, without telling us the exact error, you're wasting your time asking for help. Its impossible for anyone to help you without detailed information.

To that I would add that you must post the relevant system information referenced in my earlier post. If you don't post the requisite information, don't be surprised when its impossible for anyone to reply with an intelligent answer.

[D2Klein]

i am also getting failed to create executable message after switching gcc to 3.4.3
cflags are: -O2 march=pentium-m mtune=pentium-m -fomit-frame-pointer -pipe
any suggestions ?

[Bob P]

i am also getting failed to create executable message after switching gcc to 3.4.3
cflags are: -O2 march=pentium-m mtune=pentium-m -fomit-frame-pointer -pipe
any suggestions ?

yes. re-read the previous post:

To that I would add that you must post the relevant system information referenced in my earlier post. If you don't post the requisite information, don't be surprised when its impossible for anyone to reply with an intelligent answer.

you need to post the contents of emerge info.

[lpx]

Hi again,

After 24 hours compiling the emerge -e system finally finished without the "error" that i can't say what it was.

The only thing i know is that i took off the mtune flag from the cflags and let stay all the others.

i only decided to took that one off because i read somewhere in this thread that it is antagonic.

continuing the guide!!

i just want to ask 2 things:

- i want to run a webserver on this machine. is there any other kernel more apropriated than the one used in this guide? or it is the same thing

- after the emerge of the system there was a lot of things to update with etc-update, but the guides never mention that. i don't know if i should update everything or just let it stay. can anyone give me a tip about this?

cheers and sorry about my lameness

[kimchi_sg]

1. i want to run a webserver on this machine. is there any other kernel more apropriated than the one used in this guide? or it is the same thing
2. after the emerge of the system there was a lot of things to update with etc-update, but the guides never mention that. i don't know if i should update everything or just let it stay. can anyone give me a tip about this?

1. hardened-dev-sources, maybe. But you are on your own there. Please post all your questions about which kernel to use in the "Kernel and Hardware" forum in the future.
2. Use dispatch-conf to update, but if you have edited the particular file already and dispatch-conf shows that the change you made is going to be overwritten, do not let it undo your changes.

[96140]

FWICT it seems that the vast majority (if not all) of the complaints that have been solved by the /proc/mounts fix have been encountered by people who were attempting to adapt the Guide to install Gentoo on hardware that the Guide does not support.

That is not true! Please read my PM.

i have never been able to reproduce the problem. based on what i've been able to find in the forums, that problem seems to be occurring with SCSI and SATA installations and with some EIDE installations, but not on EIDE installations where the user remembers to process the necessary files in need of updating in /etc.

Yes and/or no (YMMV). Recently I had some free time and gave the install another shot, just to see if I could reproduce the error. On my EIDE disk (30GB ATA drive, as non-SATA/SCSI as they come), I only had a working installation after running etc-update and copying /proc/mounts. I was careful to examine the contents of /mounts before and after to compare the two. As it turned out, even with first running etc-update, none of the necessary information on the old /proc/mounts was present in the new /proc/mounts, so it was necessary to copy it.

I don't know how/why Gentoo handles SATA disks differently than the older ATA disks (if it does; seems like it does, at least), but I sure hope that there isn't an underlying hardware problem with the installation--for users of either disk type!

[D2Klein]

@Bob removing the mtune flag did the job

[Bob P]

- after the emerge of the system there was a lot of things to update with etc-update, but the guides never mention that. i don't know if i should update everything or just let it stay. can anyone give me a tip about this?

if you follow the installation guide without making any changes, a number of configuration packages will be changed, and its safe to use etc-update to over-write the config files without worrying about saving the old versions. there's absolutely nothing in those files that will be updated that is worth worrying about, so you can use etc-update wanton disregard for the perils that etc-update usually brings with it.

with that said, after the install has completed, you should not consider etc-update to be your friend. the forum is full of stories about people who hae clobbered their systems with etc-update. for this reason, i think that a program such as dispatch-conf that saves configuration information is a handy program to have around. but its totally unnecessary to use dispatch-conf to save any of the config files during the installation, as long as you're following the guide to the letter. if you change anything, all bets are off.

[racoontje]

I knew this once, but I'm doubting. Is there a difference in the result between bootstrapping && emerging system from stage1 and starting from stage3 and then emerge -e world? AFAIK bootstrap is just a fancy script that downloads and installs the basic packages.