Bug 75640 – glibc fails to merge during bootstrap

• Status: RESOLVED FIXED
• Product: Gentoo Linux

[Description (disabled) 2004-12-25 11:35:55 UTC]

When I try to bootstrap, the glibc (2.3.4.20040808-r1) merge fails *after* compilation.

----------------------------
...
>>> /usr/include/features.h
>>> /usr/include/limits.h
>>> /usr/include/regexp.h
>>> /usr/include/memory.h
--- /sbin/
>>> /sbin/sln
>>> /sbin/ldconfig
!!! FAILED postinst: 2816
livecd portage #
----------------------------

I've tried bootstrapping using ACCEPT_KEYWORDS="amd64", but ACCEPT_KEYWORDS="~amd64" gives exactly the same error.

dmesg shows something interesting:
----------------------------
ebuild.sh[25427]: segfault at 0000002a960c58b0 rip 0000002a960c58b0 rsp 0000007fbfffd170 error 14
ebuild.sh[25426]: segfault at 0000002a960c58b0 rip 0000002a960c58b0 rsp 0000007fbfffd860 error 14

Reproducible: Always
Steps to Reproduce:
1. Boot from the AMD64 live-cd.
2. Follow the installation instructions found in the Handbook.
Actual Results:  
The bootstrap failed to merge glibc.

Expected Results:  
The bootstrap should have succeded.

Portage 2.0.51-r3 (default-linux/amd64/2004.3, gcc-3.4.2,
glibc-2.3.4.20040808-r1, 2.6.9-gentoo-r1 x86_64)
=================================================================
System uname: 2.6.9-gentoo-r1 x86_64 AMD Athlon(tm) 64 Processor 3000+
Gentoo Base System version 1.4.16
Autoconf:
Automake:
Binutils: sys-devel/binutils-2.15.90.0.1.1-r3
Headers:  sys-kernel/linux26-headers-2.6.8.1-r1
Libtools:
ACCEPT_KEYWORDS="amd64"
AUTOCLEAN="yes"
CFLAGS="-march=athlon64 -O3 -pipe -g3"
CHOST="x86_64-pc-linux-gnu"
COMPILER=""
CONFIG_PROTECT="/etc /usr/kde/2/share/config /usr/kde/3/share/config
/usr/share/config /var/qmail/control"
CONFIG_PROTECT_MASK="/etc/gconf /etc/terminfo /etc/env.d"
CXXFLAGS="-march=athlon64 -O3 -pipe -g3"
DISTDIR="/usr/portage/distfiles"
FEATURES="autoaddcvs autoconfig distlocks nostrip sandbox"
GENTOO_MIRRORS="http://gentoo.osuosl.org
http://distro.ibiblio.org/pub/Linux/distributions/gentoo"
MAKEOPTS=""
PKGDIR="/usr/portage/packages"
PORTAGE_TMPDIR="/var/tmp"
PORTDIR="/usr/portage"
PORTDIR_OVERLAY=""
SYNC="rsync://rsync.gentoo.org/gentoo-portage"
USE="amd64 acpi alsa berkdb bitmap-fonts crypt f77 fortran gif gpm ipv6 jp2 jpeg
lzw lzw-tiff multilib ncurses nls nptl opengl oss pam perl png python readline
ssl tcpd tiff truetype usb xml2 xpm xrandr xv zlib"

[Comment 1 Jan Kundrát (RETIRED) 2004-12-25 12:57:06 UTC]

http://forums.gentoo.org/viewtopic.php?t=265281

[Comment 2 SpanKY 2004-12-25 15:04:25 UTC]

try with just CFLAGS="-march=athlon64 -O2 -pipe" ?

[Comment 3 Jan Kundrát (RETIRED) 2004-12-25 15:11:29 UTC]

used CFLAGS should be irrelevant as the error occurs before rebuilt packages gets used, AFAIK. or maybe some pakcage merged during bootstrap.sh is broken?

[Comment 4 SpanKY 2004-12-25 22:09:56 UTC]

bootstrap utilizes the user's CFLAGS ... otherwise the bootstrap process would be kind of pointless now wouldnt it

[Comment 5 (disabled) 2004-12-26 01:04:56 UTC]

I tried CFLAGS="-march=athlon64 -pipe", glibc succeeds to merge. However, when I try CFLAGS="-march=athlon64 -O2 -pipe -g3" (-g3 = generate extra debugging info), it fails again. On a sidenote, everything I try to do after the glibc merge failure within the chroot segfaults.
I'm currently bootstrapping with CFLAGS="-march=athlon64 -O2 -pipe", to see if -g3 causes this.

[Comment 6 SpanKY 2004-12-26 01:08:51 UTC]

thats because the glibc you emerged with -g3 was broken ... thus everything done in the chroot after would break :)

[Comment 7 (disabled) 2004-12-26 01:41:54 UTC]

If that's the case, how would -g3 cause this? I mean, it doesn't affect the code, does it?

[Comment 8 SpanKY 2004-12-26 01:45:11 UTC]

of course -g affects the generated code :)

-g controls how much debugging information is added to generated code ... the higher the #, the more debugging information you get

[Comment 9 (disabled) 2004-12-26 01:50:13 UTC]

I thought the debugging info was added to a different section of the ELF file.

The bootstrap using CFLAGS="-march=athlon64 -O2 -pipe" just succeeded merging glibc. So... how would I add debugging information to glibc afterwards? Re-emerge it with -g3 and hope it won't fail? :P

(On a sidenote, bootstrapping with -g3 worked on x86 last time I tried.)

[Comment 10 SpanKY 2004-12-26 01:58:16 UTC]

try just good old fashioned '-g' ?

also, you really cant compare experiences with diff arches and gcc code generation :)

[Comment 11 (disabled) 2004-12-26 03:02:49 UTC]

-g works, so far. At least glibc is merged :)

[Comment 12 Jakub Moc (RETIRED) 2005-09-01 15:08:53 UTC]

*** Bug 104506 has been marked as a duplicate of this bug. ***

[Comment 13 Richard Freeman 2005-10-05 19:19:22 UTC]

I am getting the segfault issues as well.  I am unable to re-emerge glibc or
other large packages in an x86 chroot due to random segfaults.  

Oddly enought I'm running 2.3.5-r1 in my amd64 root and that works just fine. 
It is the x86 chroot that is giving me problems.

I'm trying to rebuild from another chroot and install as a binary.  We'll have
to see how that works out.

[Comment 14 Mark Loeser (RETIRED) 2005-12-15 02:15:37 UTC]

This is fixed for the user now.