View previous topic :: View next topic |
Author |
Message |
stef Tux's lil' helper
Joined: 23 Jul 2003 Posts: 93
|
Posted: Sat Oct 08, 2005 10:34 am Post subject: glibc-2.3.5-r2 failed - cannot compute sizeof (long double) |
|
|
I'm using 2005.0 for long time now ... but today when I did a emerge -uDv system, when updating glibc from 2.3.5-r1 to 2.3.5-r2 I got:
>>> Source unpacked.
* Building multilib glibc for ABIs: x86 amd64
* ABI: x86
* CBUILD: x86_64-pc-linux-gnu
* CHOST: x86_64-pc-linux-gnu
* CTARGET: x86_64-pc-linux-gnu
* CBUILD_OPT: i686-pc-linux-gnu
* CTARGET_OPT: i686-pc-linux-gnu
* CC:
* CFLAGS: -march=k8 -pipe -O2
* Configuring GLIBC for linuxthreads with:
--disable-dev-erandom
--with-tls
--without-__thread
--enable-add-ons=linuxthreads,c_stubs,libidn
--enable-kernel=2.4.1
--without-selinux
--without-cvs
--enable-bind-now
--build=i686-pc-linux-gnu
--host=i686-pc-linux-gnu
--disable-profile
--without-gd
--with-headers=//usr/include
--prefix=/usr
--mandir=/usr/share/man
--infodir=/usr/share/info
--libexecdir=/usr/lib/misc/glibc
checking build system type... i686-pc-linux-gnu
<...>
checking for stdint.h... yes
checking for unistd.h... yes
checking for long double... yes
checking size of long double... configure: error: cannot compute sizeof (long double), 77
See `config.log' for more details.
!!! ERROR: sys-libs/glibc-2.3.5-r2 failed.
!!! Function glibc_do_configure, Line 940, Exitcode 1
!!! failed to configure glibc
!!! If you need support, post the topmost build error, NOT this status message.
now all I found regarding this were problems during update from 2004.3 -> 2005.0 but I upgraded long time ago without problems up to now...
ok, so what i found is:
ls -la / |grep lib
drwxr-xr-x 9 root root 4096 Oct 1 13:55 lib
drwxr-xr-x 3 root root 4096 Aug 19 21:33 lib32
lrwxrwxrwx 1 root root 3 Mar 25 2005 lib64 -> lib
is this wrong ... shouldn't it be that lib64 is a dir and lib a link to it?! ... but ... this doesn't do much harm i guess?
now a look to config.log gave:
<....>
configure:7101: checking size of long double
configure:7420: x86_64-pc-linux-gnu-gcc -o conftest -march=k8 -pipe -O2 conftest.c >&5
configure:7423: $? = 0
configure:7425: ./conftest
ACCESS DENIED open_wr: /mounts/gentoo/gentoo/tmp/portage/glibc-2.3.5-r2/work/build-x86-x86_64-pc-linux-gnu-linuxthreads/conftest.val
SECURITY BREACH SANDBOX_LOG /var/log/sandbox/sandbox-sys-libs_-_glibc-2.3.5-r2-16956.log isn't allowed via SANDBOX_WRITE
configure:7428: $? = 1
configure: program exited with status 1
configure: failed program was:
<...>
Ok, ... now it seems as if sandbox doesn't like ./configure to write to the build dir!? I have:
ls -la /var/tmp
lrwxrwxrwx 1 root root 25 Mar 25 2005 /var/tmp -> /mounts/gentoo/gentoo/tmp
Is it a bad idea to have /var/tmp symlink somewhere else? (... didn't have any problems up to now)
... am I overlooking something?! |
|
Back to top |
|
|
Mess n00b
Joined: 09 Oct 2005 Posts: 1 Location: Hungary
|
|
Back to top |
|
|
stef Tux's lil' helper
Joined: 23 Jul 2003 Posts: 93
|
Posted: Sun Oct 09, 2005 4:46 pm Post subject: |
|
|
well my guess was, that I don't have one of the problems described there ... as i get a sandbox access denied in the config.log.
Now I'm not sure if it's because of my symlinking /var/tmp, or really some other things...
Edit:
changed PORTAGE_TMPDIR in make.conf to:
PORTAGE_TMPDIR=/mounts/gentoo/gentoo/tmp
*just trying, if it works (came past the sizeof long double at least
... but then ... is symlinking /var/tmp considered a bad idea? ... |
|
Back to top |
|
|
Bluekkis n00b
Joined: 09 Oct 2004 Posts: 33
|
Posted: Tue Oct 11, 2005 6:28 pm Post subject: |
|
|
I got this same problem. Can't emerge glibc, gcc and sandbox and all 32bit programs fail to run with "accessing a corrupted shared library". I've searched forums and tryed half a dozen possible fixes I've found including some random voodoo, none have worked. I would really appriciate if someone could tell me where the problem lies... If I remember correctly this started few days ago after update that included glibc, been trying to fix this ever since. _________________ [Gentoo 64b][ASRock ALIVENF7G-HDReady][AMD Athlon64 X2 2,5GHz@1,35V][Corsair XMS2 2x 1GB@800MHz][WD Raptor 150GB][Samsung T166 500GB]
[Ubuntu LTS][VIA EPIA EN-1200EG][VIA Eden C7 1,2GHz][Kingston 1GB@533MHz][4x Samsung T166 500GB] |
|
Back to top |
|
|
petrjanda Veteran
Joined: 05 Sep 2003 Posts: 1557 Location: Brno, Czech Republic
|
Posted: Thu Oct 13, 2005 1:44 pm Post subject: |
|
|
Bluekkis wrote: | I got this same problem. Can't emerge glibc, gcc and sandbox and all 32bit programs fail to run with "accessing a corrupted shared library". I've searched forums and tryed half a dozen possible fixes I've found including some random voodoo, none have worked. I would really appriciate if someone could tell me where the problem lies... If I remember correctly this started few days ago after update that included glibc, been trying to fix this ever since. |
the problem started with r1, once i emerged it i did not have 32bit compatibility anymore, then i found out gcc had been recompiled with the -m64 cflag. it seems that all these problems are related to some kind of a bug in r1 that loses 32bit mode. i my self had a living hell fixing it. i resulted in quickpkging glibc and gcc from stage1 on the livecd and then it worked. since then r1 is hardmasked(manually) on my computer. I am afraid to upgrade to r2. _________________ There is, a not-born, a not-become, a not-made, a not-compounded. If that unborn, not-become, not-made, not-compounded were not, there would be no escape from this here that is born, become, made and compounded. - Gautama Siddharta |
|
Back to top |
|
|
stef Tux's lil' helper
Joined: 23 Jul 2003 Posts: 93
|
Posted: Fri Oct 14, 2005 11:44 am Post subject: |
|
|
seems to work fine for me now (guess it was really because of my symlinks), so maybe a slightly different problem at yours... what exactly the error saying in glibc config.log ? |
|
Back to top |
|
|
sulu Guru
Joined: 21 May 2002 Posts: 399 Location: Dornbirn/Austria
|
Posted: Sat Oct 29, 2005 3:47 am Post subject: |
|
|
Another victim here.
Build of glibc dies with the cannot compute size of long double error.
Quote: | checking size of long double... configure: error: cannot compute sizeof (long double), 77 |
It cannot build the conftest program which is used to get the size of long doubles
It dies at
Quote: | /usr/lib/gcc/x86_64-pc-linux-gnu/3.4.4/../../../../x86_64-pc-linux-gnu/bin/ld: skipping incompatible /usr/lib/gcc/x86_64-pc-linux-gnu/3.4.4/./libgcc.a when searching for -lgcc
/usr/lib/gcc/x86_64-pc-linux-gnu/3.4.4/../../../../x86_64-pc-linux-gnu/bin/ld: skipping incompatible /usr/lib/gcc/x86_64-pc-linux-gnu/3.4.4/libgcc.a when searching for -lgcc
/usr/lib/gcc/x86_64-pc-linux-gnu/3.4.4/../../../../x86_64-pc-linux-gnu/bin/ld: skipping incompatible /usr/lib/gcc/x86_64-pc-linux-gnu/3.4.4/./libgcc.a when searching for -lgcc
/usr/lib/gcc/x86_64-pc-linux-gnu/3.4.4/../../../../x86_64-pc-linux-gnu/bin/ld: skipping incompatible /usr/lib/gcc/x86_64-pc-linux-gnu/3.4.4/libgcc.a when searching for -lgcc
/usr/lib/gcc/x86_64-pc-linux-gnu/3.4.4/../../../../x86_64-pc-linux-gnu/bin/ld: cannot find -lgc |
I tried the
Quote: | emerge -v --oneshot emul-linux-x86-glibc |
No go.
I had a look at
Quote: | 2005.0 Upgrade guide |
But i allready had done this some times ago.
I extractetd the C-Source-Code from the config.log and compiled it manually
Quote: | x86_64-pc-linux-gnu-gcc -o conftest -march=athlon64 -pipe -O2 myconftest.c |
This works without a problem and returns the correct size of 16.
Sorry i dont get it. _________________ AMD64-Phenom2 940, ASUS M4??-Deluxe-Board , 8 GB-RAM , 1*1TB SATA HDD , 1*750GB SATA Hd, NVIDIA GTX 260 |
|
Back to top |
|
|
sulu Guru
Joined: 21 May 2002 Posts: 399 Location: Dornbirn/Austria
|
Posted: Wed Nov 02, 2005 6:50 am Post subject: |
|
|
Bump.
I am using the 2005.0 profile and had done the emul-library-thingy before i ran into the glibc-problem.
Should i re-emerge the 32-bit-emul-libraries ?
Is this really a problem with 32-bit compatibility ?
Generally this box runs fine. Emerges normally work without a hitch.
I use it for development in C and have zero problems. _________________ AMD64-Phenom2 940, ASUS M4??-Deluxe-Board , 8 GB-RAM , 1*1TB SATA HDD , 1*750GB SATA Hd, NVIDIA GTX 260 |
|
Back to top |
|
|
sulu Guru
Joined: 21 May 2002 Posts: 399 Location: Dornbirn/Austria
|
Posted: Mon Nov 07, 2005 8:58 am Post subject: Solved for me |
|
|
Hmmmm....
Keyword multilib was missing in /etc/make.conf.
I found several references to a missing keyword Mulitilib in several glibc-threads.
I was very sure that i had multilib enabled in my make.conf. (I think it is recommended in the 2004-3 -> 2005 profile update guide)
But as is could not bring glibc to compile i started to check everything i found mentioned in the glibc-threads.
And then i found that no mulitlib was present in my make.conf.
I rebuilt gcc with multilib and reemerged all of the emul-libs and reemerged glibc.
And viola, everything is fine again. _________________ AMD64-Phenom2 940, ASUS M4??-Deluxe-Board , 8 GB-RAM , 1*1TB SATA HDD , 1*750GB SATA Hd, NVIDIA GTX 260 |
|
Back to top |
|
|
petrjanda Veteran
Joined: 05 Sep 2003 Posts: 1557 Location: Brno, Czech Republic
|
Posted: Mon Nov 07, 2005 11:49 am Post subject: Re: Solved for me |
|
|
sulu wrote: | Hmmmm....
Keyword multilib was missing in /etc/make.conf.
I found several references to a missing keyword Mulitilib in several glibc-threads.
I was very sure that i had multilib enabled in my make.conf. (I think it is recommended in the 2004-3 -> 2005 profile update guide)
But as is could not bring glibc to compile i started to check everything i found mentioned in the glibc-threads.
And then i found that no mulitlib was present in my make.conf.
I rebuilt gcc with multilib and reemerged all of the emul-libs and reemerged glibc.
And viola, everything is fine again. |
Isnt multilib USE flag ignored nowadays? _________________ There is, a not-born, a not-become, a not-made, a not-compounded. If that unborn, not-become, not-made, not-compounded were not, there would be no escape from this here that is born, become, made and compounded. - Gautama Siddharta |
|
Back to top |
|
|
sulu Guru
Joined: 21 May 2002 Posts: 399 Location: Dornbirn/Austria
|
Posted: Mon Nov 07, 2005 1:27 pm Post subject: |
|
|
petrjanda wrote
Quote: | Isnt multilib USE flag ignored nowadays? |
Honestly i dont know.
I just waded through several threads on glibc and freqeuently found mentioned that multilib must be set in make.conf order so you dont get any binary compatibilty errors with which causes the glibc-build to die with
Quotation from the 2005.0 Upgrade guide
Quote: |
If you get a problem when ./configure is running on glibc, and it dies at 'checking size of long double... configure: error: cannot compute sizeof (long double), 77', check the config.log, and look for the error. This is the first time your toolchain is used to create and execute a 32bit application. The problems and associated causes you will see are:
Code Listing 9: cannot compute size of long double
Can't find libc.so
|
As i had exactly that problem i assumed something associtated with 32-bit-compatibility should be the problem.
The only thing i found was the missing "mutlilib" in make.conv. So i startet rebuilding everything with use of multilib.
Rigth now i think the key thing was to rebuild gcc with Code: | USE=multilib emerge gcc |
and maybe to re-emerge all of the emul-libs what fixed my box.
If you are interested i could check the bash-history for the commands i used.
Actually the success took me a little bit by surprise. _________________ AMD64-Phenom2 940, ASUS M4??-Deluxe-Board , 8 GB-RAM , 1*1TB SATA HDD , 1*750GB SATA Hd, NVIDIA GTX 260 |
|
Back to top |
|
|
|
|
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
|
|