| View previous topic :: View next topic |
| Author |
Message |
Ard Righ
Guru
Joined: 24 Jun 2002
Posts: 337
Location: Wellington, NZ
|
 Posted: Sat Aug 17, 2002 8:01 am Post subject: 1.4 Beta using 3.2 final release |
 |
|
I have the 1.4 beta stage 1 file on my HDD, waiting to install Gentoo using 3.2 final release.
What do I need to edit in the 1.4 beta to stop it from using 3.2_pre, and use 3.2 final instead ?
Also, is someone going to update the 1.4 beta files, to use 3.2 final directly ?  |
|
| Back to top |
|
 |
aardvark
Guru
Joined: 30 Jun 2002
Posts: 576
|
| Posted: Sat Aug 17, 2002 9:18 am Post subject: |
|
|
it will use the final automatically , but in my case I wasn't able to finish the bootstrap on multiple tries.. It got stuck on the last emerge of gcc
BTW. emerge -p gcc will tell you whicht version |
|
| Back to top |
|
|
ialtywu
n00b
Joined: 16 May 2002
Posts: 25
|
| Posted: Sat Aug 17, 2002 9:23 am Post subject: |
|
|
Nothing . . . the 1.4beta stage 1 tarball should use 3.2 final automatically. When you bootstrap it looks for >=3.2_pre and since 3.2 (final) is newer it will be compiled along the way. Of course, you will initially start off with whatever gcc is part of the tarball, but you will end up with the latest gcc 3.2 once your bootstrap is finished.
I am currently building 1.4beta inside VMWare Workstation, and have bootstrapped successfully with no problems. I am now getting ready to emerge system. |
|
| Back to top |
|
|
aardvark
Guru
Joined: 30 Jun 2002
Posts: 576
|
| Posted: Sat Aug 17, 2002 9:49 am Post subject: |
|
|
hmm maybe they tweaked the ebuild,
I was NOT able to bootstrap in the day of yester  |
|
| Back to top |
|
|
ialtywu
n00b
Joined: 16 May 2002
Posts: 25
|
| Posted: Sat Aug 17, 2002 10:03 am Post subject: |
|
|
| aardvark wrote: | hmm maybe they tweaked the ebuild,
I was NOT able to bootstrap in the day of yester |
I dont know, perhaps it was your compilation flags . . . are you using anything exotic? Athlon opts tend to do weird things. I havent tried building on an Athlon yet. Maybe later after I get some sleep.
I am currently building on a PII 650mhz Dell Laptop using i686 generic flags. Note though, that this is under VMWare on Windows 2000 so who knows what the heck is really going on.
So far the emerge system is progressing nicely . . . If this works out ok, I will try building a 'real' gentoo system both on my laptop and on my Athlon. |
|
| Back to top |
|
|
aardvark
Guru
Joined: 30 Jun 2002
Posts: 576
|
| Posted: Sat Aug 17, 2002 11:53 am Post subject: |
|
|
| ialtywu wrote: | | aardvark wrote: | hmm maybe they tweaked the ebuild,
I was NOT able to bootstrap in the day of yester |
I dont know, perhaps it was your compilation flags . . . are you using anything exotic? Athlon opts tend to do weird things. I havent tried building on an Athlon yet. Maybe later after I get some sleep.
|
I tried even with the most basic setting.
Perhaps I will try again some time. It's too hot for me now, as well as my CPU |
|
| Back to top |
|
|
endoalpha
Tux's lil' helper
Joined: 29 Jun 2002
Posts: 101
|
| Posted: Sat Aug 17, 2002 12:05 pm Post subject: |
|
|
I cannot bootstrap a stage1 1.4beta either (after emerge rsync). I can bootstrap if I do NOT emerge rsync before bootstap. but... this will use gcc 3.2pre and not gcc 3.2 final...
my error is:
| Code: |
checking size of char... configure: error: can not run test program while cross compiling
Configure in /var/tmp/portage/gcc-3.2/work/build/fastjar failed, exiting.
|
Any one? i am baffled.... |
|
| Back to top |
|
|
AnimalMachine
Tux's lil' helper
Joined: 27 Apr 2002
Posts: 106
Location: Milwaukee, WI USA
|
| Posted: Sat Aug 17, 2002 1:42 pm Post subject: |
|
|
I tried 1.4_beta yesterday with gcc3.2 [final] as well. Also, I could not get past the last strapping of gcc.
My first failure had a whole lot of flags. My second failure looked like:
CFLAGS="-march=athlon-xp -O2 -pipe"
my third failure looked like:
CFLAGS="-march=athlon -O2 -pipe"
So being the sour puss that I am, I'm now rebuilding my laptop [again] with 1.2. Not as nifty, but I've learned through the experiment that I just don't have the patience for gcc 3 building yet. Kudos to the dev team for getting this far though ... can't wait for the official release (and hopfully the official upgrade path). |
|
| Back to top |
|
|
Ard Righ
Guru
Joined: 24 Jun 2002
Posts: 337
Location: Wellington, NZ
|
| Posted: Sat Aug 17, 2002 2:08 pm Post subject: |
|
|
Ok, well it always pay to check before I spend an hour or two going through installing Gentoo just to have it die on me during bootstrap :p
I will have a go, I am using -march=pentium4 compile options, so cross fingers it works |
|
| Back to top |
|
|
silvercat
n00b
Joined: 07 Aug 2002
Posts: 54
Location: Canada
|
| Posted: Sat Aug 17, 2002 3:21 pm Post subject: |
|
|
| Code: | checking size of char... configure: error: can not run test program while cross compiling
Configure in /var/tmp/portage/gcc-3.2/work/build/fastjar failed, exiting. |
I think i got this when using the -fomit-frame-pointer flag with a Duron... i did not get this error when just using the flags -march=athlon -O3 -pipe... anyone wanna confirm this ? |
|
| Back to top |
|
|
ialtywu
n00b
Joined: 16 May 2002
Posts: 25
|
| Posted: Sat Aug 17, 2002 6:28 pm Post subject: |
|
|
| silvercat wrote: | | I think i got this when using the -fomit-frame-pointer flag with a Duron... i did not get this error when just using the flags -march=athlon -O3 -pipe... anyone wanna confirm this ? |
I cant confirm this, but I did just successfully bootstrap on my Athlon 850 using the i686 opt flags. No problems whatsoever.
I am beginning to suspect gcc has some kind of bug with the athlon opts.
I am not going to rebuild with athlon opts to test that theory yet though. Anyone else who has failed might want to use the generic i686 opts and let us know what happened (if you have the time) |
|
| Back to top |
|
|
AlterEgo
Veteran
Joined: 25 Apr 2002
Posts: 1619
|
| Posted: Sun Aug 18, 2002 8:21 am Post subject: |
|
|
The problem with the Athlon flags seems to be limited to the bootstrap only.
I had to use i686 to complete the bootstrap without breaking things (the cross-compiler-error).
After the bootstrap, I switched to the following flags:
CHOST="i686-pc-linux-gnu"
CFLAGS="-mcpu=athlon-tbird -O3 -pipe -fomit-frame-pointer -funroll-loops -fforce-addr -frerun-cse-after-loop -frerun-loop-opt -falign-functions=4"
CXXFLAGS=$CFLAGS
and no errors so far (= building system from scratch including KDE)
Athlon 800 |
|
| Back to top |
|
|
klieber
Bodhisattva
Joined: 17 Apr 2002
Posts: 3657
Location: San Francisco, CA
|
| Posted: Mon Aug 19, 2002 12:54 am Post subject: |
|
|
moving from gcc3 to portage.
--kurt
_________________
The problem with political jokes is that they get elected |
|
| Back to top |
|
|
endoalpha
Tux's lil' helper
Joined: 29 Jun 2002
Posts: 101
|
| Posted: Tue Aug 20, 2002 2:16 am Post subject: |
|
|
| AlterEgo wrote: | The problem with the Athlon flags seems to be limited to the bootstrap only.
I had to use i686 to complete the bootstrap without breaking things (the cross-compiler-error).
After the bootstrap, I switched to the following flags:
CHOST="i686-pc-linux-gnu"
CFLAGS="-mcpu=athlon-tbird -O3 -pipe -fomit-frame-pointer -funroll-loops -fforce-addr -frerun-cse-after-loop -frerun-loop-opt -falign-functions=4"
CXXFLAGS=$CFLAGS
and no errors so far (= building system from scratch including KDE)
Athlon 800 |
I can confirm the former, however even on my P3 with -march=pentium3 will fail the bootstrap. I guess we will have to bootstrap with -march=i686 as the best optimization for now..
My cxx and cflags for bootstrapping were '-march=i686 -O3 -pipe'
I dont think I will be trying any other optimizations any time soon... |
|
| Back to top |
|
|
eivinn
Apprentice
Joined: 10 Jul 2002
Posts: 219
Location: Norway
|
| Posted: Tue Aug 20, 2002 12:24 pm Post subject: |
|
|
Trying with:
| Code: | | pentium4 -fomit-frame-pointers -msse2 |
|
|
| Back to top |
|
|
AlterEgo
Veteran
Joined: 25 Apr 2002
Posts: 1619
|
| Posted: Tue Aug 20, 2002 7:54 pm Post subject: |
|
|
| AlterEgo wrote: | The problem with the Athlon flags seems to be limited to the bootstrap only.
I had to use i686 to complete the bootstrap without breaking things (the cross-compiler-error).
After the bootstrap, I switched to the following flags:
CHOST="i686-pc-linux-gnu"
CFLAGS="-mcpu=athlon-tbird -O3 -pipe -fomit-frame-pointer -funroll-loops -fforce-addr -frerun-cse-after-loop -frerun-loop-opt -falign-functions=4"
CXXFLAGS=$CFLAGS
and no errors so far (= building system from scratch including KDE)
Athlon 800 |
Quoting myself once more...
I think there's something wrong with the process by which gcc 3.2 is introduced. I just succesfully bootstrapped using -march-athlon-tbird after breaking my previously built system (especially portage and gcc) when emerging the new binutils 2.13...
I then remembered this post about getting gcc3.1 going based on 2.95 tarballs:
and acted accordingly (thanks again Niek)
This is what I did:
> Use the stage 1 1.4 Beta tarball
make.conf:
CHOST="i386-pc-linux-gnu"
CFLAGS="-mcpu=i386 -O3 -pipe"
CXXFLAGS="-mcpu=i386 -O3 -pipe"
(YES i386, because that's the gcc configuration in the tarball)
> emerge gcc (3.2 that is)
> emerge clean gcc (remove 3.2 pre-release)
> check gcc --version
Then change your make-variables: in my case:
CHOST="i686-pc-linux-gnu"
CFLAGS="-march=athlon-tbird -O3 -pipe -fomit-frame-pointer -funroll-loops -fforce-addr -frerun-cse-after-loop -frerun-loop-opt -falign-functions=4"
CXXFLAGS=$CFLAGS
> run bootstrap.sh
and continue the rest of the install without any problem 
It's probably differences in parameters between gcc 3.2-pre and gcc3.2 that are the cause of all these troubles (that's my analysis at least). |
|
| Back to top |
|
|
eivinn
Apprentice
Joined: 10 Jul 2002
Posts: 219
Location: Norway
|
| Posted: Wed Aug 21, 2002 3:00 pm Post subject: |
|
|
| You can also use i686 for bootstrapping. Then you can use pentium4 or whatever for the rest, but kdeaddons... |
|
| Back to top |
|
|
aardvark
Guru
Joined: 30 Jun 2002
Posts: 576
|
| Posted: Wed Aug 21, 2002 3:18 pm Post subject: |
|
|
I manage to bootstrap a gcc3.2 system fram a 1.4 stage1
- Everyting compiled fine afterwards, no hangup whatsowever
- The emerged system however did not "feel" fast though. The gcc.3.1 system was blazingly fast. I have a dual booting gentoo 1.2 install as well and believe it or not, it feels quite a bit faster than the gcc 3.2 system (more responsive). Strange! Anyone else?
- The issues with binary compatibility seem to have vanished. It seems as if gcc3.2 is binary compatible with 2.95 again. I do recall having read something about ABI being compatible in gcc 3.2, but I did not know it would mean this. That is NICE of course!
I hope to resolve the speed issues. |
|
| Back to top |
|
|
|
Portage & Programming
