View previous topic :: View next topic |
Author |
Message |
jarealist Apprentice
Joined: 07 Oct 2002 Posts: 228
|
Posted: Thu May 24, 2007 6:58 am Post subject: |
|
|
Just a note for everyone trying this:
The patchsets for glibc-2.6 and gcc-4.2.0 were just bumped to 1.1.
Easy to miss without a revision change. |
|
Back to top |
|
|
loki_val Retired Dev
Joined: 13 Nov 2006 Posts: 418 Location: Denmark
|
Posted: Thu May 24, 2007 8:37 am Post subject: |
|
|
drwook wrote: |
peteralf doesn't say what arch he's on | ~x86 |
|
Back to top |
|
|
irondog l33t
Joined: 07 Jul 2003 Posts: 715 Location: Voor mijn TV. Achter mijn pc.
|
Posted: Thu May 24, 2007 2:36 pm Post subject: |
|
|
Is there any specific CORE 2 DUO -march flag yet in gcc 4.2 or do I have to wait for gcc 4.3? _________________ Alle dingen moeten onzin zijn. |
|
Back to top |
|
|
drwook Veteran
Joined: 30 Mar 2005 Posts: 1324 Location: London
|
|
Back to top |
|
|
Genewb Apprentice
Joined: 09 Jan 2007 Posts: 165
|
Posted: Thu May 24, 2007 6:42 pm Post subject: |
|
|
jarealist wrote: | Just a note for everyone trying this:
The patchsets for glibc-2.6 and gcc-4.2.0 were just bumped to 1.1.
Easy to miss without a revision change. |
Does this effect the ebuilds in Portage? Glancing at the changelog there's no mention (I thought that every time a change is made, it is mentioned there, even if there is no revision bump)... |
|
Back to top |
|
|
irondog l33t
Joined: 07 Jul 2003 Posts: 715 Location: Voor mijn TV. Achter mijn pc.
|
Posted: Thu May 24, 2007 8:26 pm Post subject: |
|
|
Take a diff from the ebuild in /var/db/pkg and the one in portage. If you want to be sure you're up to date you have to do that.
Gentoo's toolchain guys have been doing updates without revision bumps for quite some time. Sometimes it looks like a match between developpers to have as little revision bumps as possible. I am sure that some of the changes should have a bump for tracability, but they might consider anything in -* or ~arch as 'free for changes'.
Let's hope they don't have the impression to be magicans who can fix anything without breaking anything. _________________ Alle dingen moeten onzin zijn. |
|
Back to top |
|
|
Vlad.Sharp Guru
Joined: 08 Dec 2004 Posts: 337 Location: Cambridgeshire, UK
|
Posted: Thu May 24, 2007 9:28 pm Post subject: |
|
|
I can report success with compilation of a minimal Xorg/Xfce4 system (with Qt4 applications working), so far. Kde seems to be compiling fine.
Here's my emerge --info:
Code: | Portage 2.1.2.7 (hardened/x86/2.6, gcc-4.2.0, glibc-2.5-r2, 2.6.21-ck2-rsbac i686)
=================================================================
System uname: 2.6.21-ck2-rsbac i686 Intel(R) Pentium(R) M processor 2.00GHz
Gentoo Base System release 2.0.0_alpha3
Timestamp of tree: Thu, 24 May 2007 20:00:01 +0000
ccache version 2.4 [enabled]
dev-java/java-config: 1.3.7, 2.0.32
dev-lang/python: 2.5.1-r2
dev-python/pycrypto: 2.0.1-r5
dev-util/ccache: 2.4-r7
sys-apps/sandbox: 1.2.18.1
sys-devel/autoconf: 2.13, 2.61
sys-devel/automake: 1.4_p6, 1.5, 1.6.3, 1.7.9-r1, 1.8.5-r3, 1.9.6-r2, 1.10
sys-devel/binutils: 2.17.50.0.16
sys-devel/gcc-config: 1.3.16
sys-devel/libtool: 1.5.23b
virtual/os-headers: 2.6.21
ACCEPT_KEYWORDS="x86 ~x86"
AUTOCLEAN="yes"
CBUILD="i686-pc-linux-gnu"
CFLAGS="-O2 -march=native -pipe -fomit-frame-pointer -fforce-addr -fweb -ftracer -falign-functions=64"
CHOST="i686-pc-linux-gnu"
CONFIG_PROTECT="/etc /usr/kde/3.5/env /usr/kde/3.5/share/config /usr/kde/3.5/shutdown /usr/share/X11/xkb /usr/share/config"
CONFIG_PROTECT_MASK="/etc/env.d /etc/env.d/java/ /etc/gconf /etc/revdep-rebuild /etc/terminfo /etc/texmf/web2c"
CXXFLAGS="-O2 -march=native -pipe -fomit-frame-pointer -fforce-addr -fweb -ftracer -falign-functions=64 -fvisibility-inlines-hidden"
DISTDIR="/usr/portage-distfiles"
FEATURES="ccache distlocks metadata-transfer sandbox sfperms strict"
GENTOO_MIRRORS="http://ftp-stud.fht-esslingen.de/pub/Mirrors/gentoo/ http://ftp.snt.utwente.nl/pub/os/linux/gentoo http://mirror.switch.ch/ftp/mirror/gentoo/ "
LANG="en_GB.UTF-8"
LC_ALL="en_GB.UTF-8"
LDFLAGS="-Wl,-O1 -Wl,--sort-common -s -Wl,--hash-style=both"
LINGUAS="en_GB de ru ja"
MAKEOPTS="-j2"
PKGDIR="/usr/portage/packages"
PORTAGE_RSYNC_OPTS="--recursive --links --safe-links --perms --times --compress --force --whole-file --delete --delete-after --stats --timeout=180 --exclude=/distfiles --exclude=/local --exclude=/packages --filter=H_**/files/digest-*"
PORTAGE_TMPDIR="/var/tmp"
PORTDIR="/usr/portage"
PORTDIR_OVERLAY="/usr/portage/local/layman/d /usr/portage/local/layman/xeffects /usr/local/portage /usr/portage/local/pieworld"
SYNC="rsync://rsync.gentoo.org/gentoo-portage"
USE="X a52 aac acl acpi alsa async avahi berkdb branding bzip2 cairo caps cdr cjk crypt cups d dbus djvu dts dvd dvdr encode exif expat ffmpeg fftw firefox flac gadu gcj gdbm gif glitz gmp gnome gnutls gpm graphviz gs gstreamer gtk hal hardened innodb ipv6 jabber java javascript jce jpeg jpeg2k kde kdeenablefinal kdehiddenvisibility lame lcms libnotify lua mad midi mmap mmx mono mp3 mpeg msn multislot mysql ncurses newspr nls nptl nptlonly nsplugin ogg openexr opengl openmp pam pcre pdf pic pike png postgres ppds python qt qt3 qt4 quicktime readline ruby samba sdl skey socks5 spell sqlite sqlite3 sse sse2 ssl startup-notification svg tcl tcpd tetex threads tiff tk truetype type1 unicode urandom usb vim-syntax vorbis win32codecs x86 xcb xcomposite xine xinerama xml xorg xprint xscreensaver xv yahoo zlib" ALSA_CARDS="hda-intel" ALSA_PCM_PLUGINS="adpcm alaw asym copy dmix dshare dsnoop empty extplug file hooks iec958 ioplug ladspa lfloat linear meter mulaw multi null plug rate route share shm softvol" ELIBC="glibc" INPUT_DEVICES="mouse evdev synaptics vmmouse keyboard" KERNEL="linux" LCD_DEVICES="bayrad cfontz cfontz633 glk hd44780 lb216 lcdm001 mtxorb ncurses text" LINGUAS="en_GB de ru ja" USERLAND="GNU" VIDEO_CARDS="nvidia vesa vmware v4l"
Unset: CTARGET, EMERGE_DEFAULT_OPTS, INSTALL_MASK, PORTAGE_COMPRESS, PORTAGE_COMPRESS_FLAGS, PORTAGE_RSYNC_EXTRA_OPTS
|
That is, with the openssl patch and the patch from the gcc bug that makes xorg-server compile for 8 hours with 800MB memory usage. Oh, and kevquinn's most excellent overlay! |
|
Back to top |
|
|
ppurka Advocate
Joined: 26 Dec 2004 Posts: 3256
|
Posted: Fri May 25, 2007 4:03 am Post subject: |
|
|
Vlad.Sharp wrote: | I can report success with compilation of a minimal Xorg/Xfce4 system (with Qt4 applications working), so far. Kde seems to be compiling fine.
<snip>
That is, with the openssl patch and the patch from the gcc bug that makes xorg-server compile for 8 hours with 800MB memory usage. Oh, and kevquinn's most excellent overlay! | Do you have any reference to the patch for gcc (or is the patch for xorg-server)? _________________ emerge --quiet redefined | E17 vids: I, II | Now using kde5 | e is unstable :-/ |
|
Back to top |
|
|
Vlad.Sharp Guru
Joined: 08 Dec 2004 Posts: 337 Location: Cambridgeshire, UK
|
Posted: Fri May 25, 2007 5:49 am Post subject: |
|
|
ppurka wrote: | Vlad.Sharp wrote: | I can report success with compilation of a minimal Xorg/Xfce4 system (with Qt4 applications working), so far. Kde seems to be compiling fine.
<snip>
That is, with the openssl patch and the patch from the gcc bug that makes xorg-server compile for 8 hours with 800MB memory usage. Oh, and kevquinn's most excellent overlay! | Do you have any reference to the patch for gcc (or is the patch for xorg-server)? |
Yeah, sure - the link to the bug was on the previous page. Here you go: http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30052
First make sure to copy the 4.2.0 ebuild (and files) to your overlay.
Take the patch that is there (if you are using fortran, please change a line as instructed in the bug discussion) and put it into /your/overlay/sys-devel/gcc/files/gcc-4.2.0/gcc-4.2.0-backportalias.diff
Then, after the last instance of "epatch" add the backportalias.diff you just put there (by copying the previous line if you like and replacing the patch name). Note the diff filename
Should apply cleanly |
|
Back to top |
|
|
mbar Veteran
Joined: 19 Jan 2005 Posts: 1990 Location: Poland
|
Posted: Fri May 25, 2007 8:59 am Post subject: |
|
|
Vlad.Sharp wrote: |
Should apply cleanly |
Sorry, this fails for me, could you upload working ebuild for gcc 4.2? Preferably the whole sys-devel/gcc overlay subtree. |
|
Back to top |
|
|
purpler n00b
Joined: 11 Jan 2007 Posts: 38 Location: /v*/l*/p*/world
|
|
Back to top |
|
|
Vlad.Sharp Guru
Joined: 08 Dec 2004 Posts: 337 Location: Cambridgeshire, UK
|
Posted: Fri May 25, 2007 9:25 am Post subject: |
|
|
mbar wrote: | Vlad.Sharp wrote: |
Should apply cleanly |
Sorry, this fails for me, could you upload working ebuild for gcc 4.2? Preferably the whole sys-devel/gcc overlay subtree. |
I think I know what the problem is. I've uploaded the portage version of gcc 4.2 here: www.the-sharp.net/gentoo/gcc-4.2.0_overlay_patched.tbz2 (SHA512 sum: fef561155da60c96168781e10ef55821d27d1150f6347c8639dbbd641b56dd5ff63da7b4976a3a22fe2caf0af31d186cea0b47e79fa1138b48290db3939ec3a2)
I haven't modified it to work with fortran though. |
|
Back to top |
|
|
mbar Veteran
Joined: 19 Jan 2005 Posts: 1990 Location: Poland
|
Posted: Fri May 25, 2007 11:06 am Post subject: |
|
|
thanks, it worked, I don't need fortran |
|
Back to top |
|
|
ppurka Advocate
Joined: 26 Dec 2004 Posts: 3256
|
Posted: Fri May 25, 2007 3:45 pm Post subject: |
|
|
Vlad.Sharp wrote: | Yeah, sure - the link to the bug was on the previous page. Here you go: http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30052
First make sure to copy the 4.2.0 ebuild (and files) to your overlay.
Take the patch that is there (if you are using fortran, please change a line as instructed in the bug discussion) and put it into /your/overlay/sys-devel/gcc/files/gcc-4.2.0/gcc-4.2.0-backportalias.diff
Then, after the last instance of "epatch" add the backportalias.diff you just put there (by copying the previous line if you like and replacing the patch name). Note the diff filename
Should apply cleanly | Thanks for the detailed reply I had somehow missed that problem and the patch, I did collect the openssl patch and I noted the problem of openssh-4.5 with openssl-0.9.8e. I still haven't done the world compilation (trying to upgrade to the same versions in my other machine too, so that I can use distcc ). _________________ emerge --quiet redefined | E17 vids: I, II | Now using kde5 | e is unstable :-/ |
|
Back to top |
|
|
jarealist Apprentice
Joined: 07 Oct 2002 Posts: 228
|
Posted: Fri May 25, 2007 7:57 pm Post subject: |
|
|
Genewb wrote:
Quote: | Does this effect the ebuilds in Portage? Glancing at the changelog there's no mention (I thought that every time a change is made, it is mentioned there, even if there is no revision bump)... |
I noticed the ebuild changed for glibc-2.6.
I usually catch changes to patchsets when I run "eclean -di distfiles". The one that thing that bothers me the most is when there are changes made to init scripts without doing a revision change. I run an ~x86 setup on both my laptops so by rights I don't have a really good reason to complain -- so I don't. Not all devs note everything in the changelogs, but most do. |
|
Back to top |
|
|
devsk Advocate
Joined: 24 Oct 2003 Posts: 2995 Location: Bay Area, CA
|
Posted: Fri May 25, 2007 9:07 pm Post subject: |
|
|
did anybody find a fix for kdelibs on amd64? It only compiles without hidden visibility, and the resultant libraries are bigger by at least 5 to 10%. 3.5.7 doesn't have the fix either. |
|
Back to top |
|
|
devsk Advocate
Joined: 24 Oct 2003 Posts: 2995 Location: Bay Area, CA
|
Posted: Sat May 26, 2007 10:49 pm Post subject: |
|
|
Is anybody else noticing the fact that gcc 4.2 created objects are bigger than 4.1.2 for the same set of CFLAGS/CXXFLAGS? |
|
Back to top |
|
|
buddabrod Apprentice
Joined: 15 Oct 2006 Posts: 241 Location: Germany
|
Posted: Sun May 27, 2007 11:27 am Post subject: |
|
|
devsk wrote: | Is anybody else noticing the fact that gcc 4.2 created objects are bigger than 4.1.2 for the same set of CFLAGS/CXXFLAGS? | Yes, i noticed that ,too. |
|
Back to top |
|
|
cneil n00b
Joined: 28 May 2007 Posts: 39 Location: Stillwater, OK
|
Posted: Mon May 28, 2007 4:42 am Post subject: Possible fix for kde-3.5.7 with gcc-4.2 on AMD64 |
|
|
devsk wrote: | did anybody find a fix for kdelibs on amd64? It only compiles without hidden visibility, and the resultant libraries are bigger by at least 5 to 10%. 3.5.7 doesn't have the fix either. |
I just finished installing Gentoo on my AMD64 using gcc-4.2. I emerged all of kde-meta-3.5.7 with kdehiddenvisibility in my USE flags. I got the same errors you mentioned in a previous post (the relocation error with a suggestion to recompile with -fPIC) when I was playing around with some of the gcc-4.2 prereleases. Upgrading to the latest prerelease version of binutils (currently 2.17.50.0.16) fixed the problem for me. However, when I last encountered that error, I scrapped the install and started over with the prerelease binutils, so I don't know if any or all of kde's dependencies will have to be reemerged with the new binutils.
Hope this helps!
P.S. Woohoo! My first post - long time lurker, first time poster. |
|
Back to top |
|
|
devsk Advocate
Joined: 24 Oct 2003 Posts: 2995 Location: Bay Area, CA
|
Posted: Mon May 28, 2007 4:48 am Post subject: Re: Possible fix for kde-3.5.7 with gcc-4.2 on AMD64 |
|
|
cneil wrote: | devsk wrote: | did anybody find a fix for kdelibs on amd64? It only compiles without hidden visibility, and the resultant libraries are bigger by at least 5 to 10%. 3.5.7 doesn't have the fix either. |
I just finished installing Gentoo on my AMD64 using gcc-4.2. I emerged all of kde-meta-3.5.7 with kdehiddenvisibility in my USE flags. I got the same errors you mentioned in a previous post (the relocation error with a suggestion to recompile with -fPIC) when I was playing around with some of the gcc-4.2 prereleases. Upgrading to the latest prerelease version of binutils (currently 2.17.50.0.16) fixed the problem for me. However, when I last encountered that error, I scrapped the install and started over with the prerelease binutils, so I don't know if any or all of kde's dependencies will have to be reemerged with the new binutils.
| does 2.17.50.0.16 binutils work for the whole system completely? Did you get any segv or other issues with that version of binutils? I am hesitant to move to unstable (upstream) version of basic things like gcc and binutils. |
|
Back to top |
|
|
cneil n00b
Joined: 28 May 2007 Posts: 39 Location: Stillwater, OK
|
Posted: Mon May 28, 2007 5:11 am Post subject: Re: Possible fix for kde-3.5.7 with gcc-4.2 on AMD64 |
|
|
Quote: | does 2.17.50.0.16 binutils work for the whole system completely? Did you get any segv or other issues with that version of binutils? I am hesitant to move to unstable (upstream) version of basic things like gcc and binutils. |
Jeez, that was fast...
I just finished installing about 1350 packages with binutils-2.17.50.0.16 (all ~amd64). The vast majority of those were with gcc-4.2. There were only a handful of packages that failed with gcc-4.2 - many of those were documented earlier in this thread (and possibly other threads). If it helps, here's a list of packages that failed and what I did to work around them:
app-office/openoffice-2.2.0 suffers from numerous crashes and problems at runtime. I compiled it with gcc-4.1 (I also had to compile STLport and hunspell with gcc-4.1 or openoffice would fail to compile due to version mismatches).
dev-libs/g-wrap-1.9.6-r3 fails to compile because gcc-4.2 issues a new warning (something about inlining and gnu89/gnu99) and g-wrap uses -Werror. Adding -fgnu89-inline to CFLAGS suppresses the warning.
dev-libs/openssl-0.9.8e segfaults (as mentioned earlier in this thread). I could have used the patch to get it to work with gcc-4.2, but I just compiled it with gcc-4.1.
dev-libs/pwlib-1.10-7 compiles fine with gcc-4.2, but compiling net-libs/opal fails. Compiling pwlib with gcc-4.1 will allow opal to compile (opal even compiles with gcc-4.2 once pwlib is compiled with gcc-4.1).
dev-utils/boost-build-1.34-pre20061214 fails (segfaults in the middle of the compile) with -O2 and gcc-4.2. Compiling with -fno-strict-aliasing in CFLAGS works fine. (boost-build seems to be the a re-packaging/fork of jam?)
dev-utils/jam-2.5-r3 fails with -O2 and gcc-4.2 (same error in the same file as boost-build). Compiling with -fno-strict-aliasing in CFLAGS also works fine. There's a bug out on this one: bug 173703
There were a handful of other failures (mostly parallel make failures) that I also experienced with gcc-4.1.
Hope this helps,
Neil
Last edited by cneil on Mon May 28, 2007 5:25 am; edited 1 time in total |
|
Back to top |
|
|
cneil n00b
Joined: 28 May 2007 Posts: 39 Location: Stillwater, OK
|
Posted: Mon May 28, 2007 5:15 am Post subject: Re: Possible fix for kde-3.5.7 with gcc-4.2 on AMD64 |
|
|
cneil wrote: | I just finished installing about 1350 packages with binutils-2.17.50.0.16 (all ~amd64). |
Oh, and I also used --as-needed - and I only had two failures because of it (bashmark-0.6.2 and gst-plugins-0.8.12).
Neil |
|
Back to top |
|
|
devsk Advocate
Joined: 24 Oct 2003 Posts: 2995 Location: Bay Area, CA
|
Posted: Mon May 28, 2007 6:09 am Post subject: |
|
|
thanks Neil!
I was more interested in knowing about the stability of the packages built with unstable binutils once they are run. Did you notice any problems in using those packages? |
|
Back to top |
|
|
cneil n00b
Joined: 28 May 2007 Posts: 39 Location: Stillwater, OK
|
Posted: Mon May 28, 2007 7:30 am Post subject: |
|
|
devsk wrote: | thanks Neil!
I was more interested in knowing about the stability of the packages built with unstable binutils once they are run. Did you notice any problems in using those packages? |
So far, no problems, though I haven't necessarily used all ~1350 packages installed on my computer. KDE, kmail, konqueror, konsole, k3b, kmymoney2, gkrellm, openoffice, mplayer, amarok, portage (python, gcc, etc.) all work fine. That's the stuff I use the most.
Neil |
|
Back to top |
|
|
n0rbi666 l33t
Joined: 04 Mar 2005 Posts: 707 Location: \Poland\Krakow
|
Posted: Mon May 28, 2007 8:06 am Post subject: |
|
|
I have binutils-2.17.50.0.15 and everything is ok, no segfaults.
With 2.17.50.0.16 - when emerging mesa, ld segfaults - so I get back to 2.17.50.0.15 |
|
Back to top |
|
|
|