GCC 4.3.0 testing! (part 2)

Message

kernelOfTruth · Post by **kernelOfTruth** » Wed Jun 11, 2008 11:15 am

12345moon12345 wrote:after emerge -e world is approximately 75% ready ( 800 packages) these packages have failed:

libebml libmatroska exiv2 net-toolsf acpid gwenview boost gpm netkit-rsh

thanks

,

please keep this thread updated if you encounter more errors,

thanks

Thargor · Post by **Thargor** » Wed Jun 11, 2008 1:25 pm

As far as I know the boost test are supposed to fail.
One is supposed to determine if it runns ok by the grade of the failure.
At least I read something like that in a boost-1.34 stabilization bug some time ago iirc.

kernelOfTruth · Post by **kernelOfTruth** » Wed Jun 11, 2008 1:37 pm

Thargor wrote:As far as I know the boost test are supposed to fail.
One is supposed to determine if it runns ok by the grade of the failure.
At least I read something like that in a boost-1.34 stabilization bug some time ago iirc.

boost unfortunately didn't even dare to finish (test-suite), it ate all of my RAM (8 GB) and only a reboot via magic sysrq key revived it ...

(I forgot about that, this happened years ago, too -> 1 GB of RAM

)

how long is the test supposed to run on an Core2 Duo E6600 (2.4 GHz) ?

thanks

tallica · Post by **tallica** » Wed Jun 11, 2008 2:22 pm

Hello,

I get:

Code: Select all

Checking for working mktime... no

during emerging bash with GCC-4.3.1

GCC-4.2.3: http://pastebin.com/m62061722 GCC-4.3.1: http://pastebin.com/d2f25c523

Eckos · Post by **Eckos** » Wed Jun 11, 2008 5:41 pm

Michal. wrote:Hello,

I get:
Code: Select all
Checking for working mktime... no
during emerging bash with GCC-4.3.1

GCC-4.2.3: http://pastebin.com/m62061722 GCC-4.3.1: http://pastebin.com/d2f25c523

Did you emerge -C mktemp? If you did, emerge coreutils. Then try again.

tallica · Post by **tallica** » Wed Jun 11, 2008 6:42 pm

Eckos wrote:Did you emerge -C mktemp? If you did, emerge coreutils. Then try again.

No... I don't have mktemp, re-emerge coreutils didn't solve my problem. Will try newer version.

EDIT:
same:

Code: Select all

checking for working mktime... no

SlashBeast · Post by **SlashBeast** » Wed Jun 11, 2008 7:34 pm

Code: Select all

slashbeast@jinchuuriki ~ % qfile /usr/bin/mktemp -v
sys-apps/coreutils-6.12 (/usr/bin/mktemp)

tallica · Post by **tallica** » Wed Jun 11, 2008 8:12 pm

I mean that I don't have package sys-apps/mktemp installed.

Code: Select all

sys-apps/coreutils-6.10-r2 (/usr/bin/mktemp)

12345moon12345 · Post by **12345moon12345** » Thu Jun 12, 2008 2:18 pm

after emerge -e world ( 800 packages ) with gcc 4.3.1 the following packages fail:

[ebuild R ] dev-libs/libebml-0.7.7-r1 0 kB
[ebuild R ] media-gfx/exiv2-0.15-r1 USE="nls unicode zlib -doc" LINGUAS="-de -es -fi -fr -pl -ru" 0 kB
[ebuild R ] sys-apps/net-tools-1.60-r13 USE="nls -static" 0 kB
[ebuild R ] sys-power/acpid-1.0.6-r1 0 kB
[ebuild R ] dev-libs/boost-1.34.1-r2 USE="-debug -doc -icu -pyste -tools" 0 kB
[ebuild R ] sys-libs/gpm-1.20.1-r6 USE="(-selinux)" 0 kB
[ebuild R ] net-misc/netkit-rsh-0.17-r8 USE="pam" 0 kB
[ebuild R ] net-mail/freepops-0.2.5 USE="gnutls" 0 kB
[ebuild R ] dev-libs/elfutils-0.127 0 kB
[ebuild R ] sys-auth/consolekit-0.2.3 USE="pam -debug" 0 kB
[ebuild R ] app-text/libwpd-0.8.9 USE="-doc" 0 kB
[ebuild R ] mail-client/sylpheed-2.4.5 USE="nls ssl -crypt -ipv6 -ldap -pda -spell-xface" 0 kB
[ebuild R ] app-emulation/virtualbox-1.5.6 USE="additions alsa -pulseaudio -sdk" 0kB
[ebuild R ] media-libs/libmatroska-0.8.1 0 kB
[ebuild R ] kde-base/gwenview-4.0.5 USE="-debug -htmlhandbook" 0 kB

pdw_hu · Post by **pdw_hu** » Thu Jun 12, 2008 5:26 pm

Acpid (#225417) and netkit-rsh (#225341) is a glibc 2.8 bug, rather than gcc.

unK · Post by **unK** » Thu Jun 12, 2008 7:13 pm

Major part of this packages can be emerged without errors, you just have to emerge versions from ~arch.

kernelOfTruth · Post by **kernelOfTruth** » Thu Jun 12, 2008 7:26 pm

unK wrote:Major part of this packages can be emerged without errors, you just have to emerge versions from ~arch.

almost correct:

if you're using ~arch already, glibc-2.8 causes some emerge-errors

search for glibc 2.8 consolekit and you should find a link to flameeyes' blog & a link to the bugs.gentoo.org tracker

this upgrade (or the combination with gcc-4.3.1-hardened upgrade) seemingly b0rked my whole system: currently rebuilding system + world (X almost instantenously crashes + several packages don't build)

unK · Post by **unK** » Thu Jun 12, 2008 7:35 pm

To emerge consolekit with glibc-2.8 it is enough to create file /etc/portage/env/sys-auth/consolekit and put into this:

Code: Select all

CFLAGS="${CFLAGS} -D_GNU_SOURCE"
CXXFLAGS="${CFLAGS}"

Eckos · Post by **Eckos** » Fri Jun 13, 2008 4:22 pm

I wouldn't advise people to use glibc-2.8. It's too buggy right now.

sgarcia · Post by **sgarcia** » Fri Jun 13, 2008 8:57 pm

Um.

And if you've made the mistake of installing it, what then?

Code: Select all

 * Sanity check to keep you from breaking your system:
 *  Downgrading glibc is not supported and a sure way to destruction
 *

If it's *really* that bad, I guess I can install from scratch (this is a fresh install) but I'm three days into this fresh install, so you can understand my reluctance...

kernelOfTruth · Post by **kernelOfTruth** » Fri Jun 13, 2008 9:22 pm

Eckos wrote:I wouldn't advise people to use glibc-2.8. It's too buggy right now.

++

I guess I need to wipe my system, too

will play back the stage4-tarball & back to work

sgarcia · Post by **sgarcia** » Fri Jun 13, 2008 10:08 pm

I think I'm going to take a chance and live with it. I've just gotten compiz-fusion working, and I don't really want to go through that again...

loftwyr · Post by **loftwyr** » Fri Jun 13, 2008 10:19 pm

Read this for pointers

http://blog.flameeyes.eu/articles/2008/ ... 8-failures

SlashBeast · Post by **SlashBeast** » Fri Jun 13, 2008 10:20 pm

Code: Select all

dev-libs/boost-1.35.0-r1
dev-util/boost-build-1.35.0-r1

Boost compiled on gcc 4.3.0 work for me.

kev009 · Post by **kev009** » Sat Jun 14, 2008 6:58 am

And what exactly makes glibc 2.8 buggy? Seriously, a couple of packages need a patch of a few lines at most to compile.. mostly due to CLEANUP.

Please stop spreading FUD.

Thargor · Post by **Thargor** » Sat Jun 14, 2008 7:21 am

The

Code: Select all

checking for working mktime... no

is an autotools bug in older versions that is triggered by gcc-4.3

And glibc-2.8 isn't that buggy, most of the compile errors are due to internal cleanups in glibc, just like the ones gcc-4.3 triggered

tallica · Post by **tallica** » Sat Jun 14, 2008 10:25 am

Thargor wrote:The
Code: Select all
checking for working mktime... no
is an autotools bug in older versions that is triggered by gcc-4.3

After emerge:

Code: Select all

sys-devel/autoconf ~amd64
sys-devel/automake ~amd64
sys-devel/libtool ~amd64
sys-devel/autoconf-wrapper ~amd64

I still get

Code: Select all

checking for working mktime... no

kernelOfTruth · Post by **kernelOfTruth** » Sat Jun 14, 2008 1:02 pm

just a short question:

is it safe to use

Code: Select all

CFLAGS=${CFLAGS} -D_GNU_SOURCE"

and

Code: Select all

CXXFLAGS=${CFLAGS} -D_GNU_SOURCE"

globally ?

Thargor · Post by **Thargor** » Sat Jun 14, 2008 7:34 pm

Michal. wrote:After emerge:
Code: Select all
sys-devel/autoconf ~amd64
sys-devel/automake ~amd64
sys-devel/libtool ~amd64
sys-devel/autoconf-wrapper ~amd64
I still get
Code: Select all
checking for working mktime... no

That's probably because due to incompatibilities between different autotools versions, different packages use older versions.
Could also be that the bug is in older configure.in files, I can't remember it exactly, google could probably help.

kernelOfTruth · Post by **kernelOfTruth** » Sat Jun 14, 2008 8:21 pm

xorg-server / xorg almost instantaneously crashes for me on almost any occasion:

https://hardened.gentooexperimental.org ... /ticket/25

anyone encountered the same ?

or is this likely a hardened specific error

thanks

system: ~amd64, gcc-4.3.1 hardened+pie v7
core2 duo E6600

GCC 4.3.0 testing! (part 2)

Re: list of failed packages

failing packages