Tweaked glibc snapshot ebuilds!

Message

weedy · Post by **weedy** » Mon Mar 12, 2007 6:28 pm

need a new glibc snapshot :/

seren · Post by **seren** » Wed Mar 14, 2007 6:23 am

I did the liberty of creating a new snapshot.

You can view the source here http://code.google.com/p/glibcsnap/

its really just a bumped branch update and I edited the ebuild to use my newly created google code project.

Many many thanks go to pal_gene

Code: Select all

svn checkout http://glibcsnap.googlecode.com/svn/

seren · Post by **seren** » Wed Mar 14, 2007 8:24 pm

Maybe i should have started a new thread?

Code: Select all

Portage 2.1.2.2 (default-linux/amd64/2006.1, gcc-4.1.2, glibc-2.5.90.20070313-r0, 2.6.20-viper4 x86_64)
=================================================================
System uname: 2.6.20-viper4 x86_64 AMD Athlon(tm) 64 Processor 3000+
Gentoo Base System release 1.12.9
Timestamp of tree: Sun, 11 Mar 2007 19:30:05 +0000
ccache version 2.4 [enabled]
dev-java/java-config: 1.3.7, 2.0.31-r4
dev-lang/python:     2.4.4
dev-python/pycrypto: 2.0.1-r5
dev-util/ccache:     2.4-r6
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:  2.17.50.0.12
sys-devel/gcc-config: 1.3.15
sys-devel/libtool:   1.5.23b
virtual/os-headers:  2.6.20-r1
ACCEPT_KEYWORDS="amd64 ~amd64"
AUTOCLEAN="yes"
CBUILD="x86_64-pc-linux-gnu"
CFLAGS="-march=athlon64 -O2 -pipe -fomit-frame-pointer -fno-ident -frename-registers -fweb -fgcse-after-reload -msse3"
CHOST="x86_64-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/java-config/vms/ /etc/php/apache1-php5/ext-active/ /etc/php/apache2-php5/ext-active/ /etc/php/cgi-php5/ext-active/ /etc/php/cli-php5/ext-active/ /etc/revdep-rebuild /etc/splash /etc/terminfo"
CXXFLAGS="-march=athlon64 -O2 -pipe -fomit-frame-pointer -fno-ident -frename-registers -fweb -fgcse-after-reload -msse3"
DISTDIR="/usr/portage/distfiles"
FEATURES="autoconfig ccache distlocks metadata-transfer sandbox sfperms strict"
GENTOO_MIRRORS="http://distfiles.gentoo.org http://distro.ibiblio.org/pub/linux/distributions/gentoo"
LDFLAGS="-Wl,-O1 -Wl,--hash-style=both"
LINGUAS="en"
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/sunrise /usr/portage/local/layman/enlightenment /usr/portage/local/layman/einit /usr/portage/local/layman/xeffects /usr/portage/local/layman/portato /usr/portage/local/layman/custom-kernels /usr/local/portage /usr/portage/local/layman/initng-portage"
SYNC="rsync://rsync.gentoo.org/gentoo-portage"
USE="X aac acpi aiglx alsa amd64 bash-completetion bash-completion berkdb bitmap-fonts cli cracklib crypt cups dbus divx dri dvdread encodedv exo fbcon ffmpeg fortran gdbm gif glitz gpm gtk hal hald iconv imagemagick insecure-savers ipv6 isdnlog java jpeg kde kdeenablefinal kickoff libg++ libnotify midi mozbranding mozdevelop moznocompose moznoirc moznomail moznopango mp3 mpeg mplayer mtp musicbrainz ncurses net nls nptl nptlonly nsplugin offensive opengl pam panel-plugin pcre pdf perl pertty php plugins png ppds pppd python qt qt3 quicktime readline realmedia reflection ruby samba session slp spell spl sse3 ssl svg swat symlink tcl tcpd thumbnail thunar tiff trayicon truetype truetype-fonts type1-fonts unicode vcd visualization visulation wmp xcomposite xine xinerama xml xorg xscreensaver xvid zlib" ALSA_CARDS="ali5451 als4000 atiixp atiixp-modem bt87x ca0106 cmipci emu10k1x ens1370 ens1371 es1938 es1968 fm801 hda-intel intel8x0 intel8x0m maestro3 trident usb-audio via82xx via82xx-modem ymfpci" 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" INITNG_PLUGINS="also bash_launcher chdir chroot conflict cpout critical ctrlaltdel daemon_clean debug_commands envparser find fmon fstat history idleprobe initctl interactive iparser last limit lockfile logfile netdev netprobe ngc4 ngcs nge pause provide reload renice rlparser simple_launcher stcmd stdout suid syncron syslog sysreq unneeded usplash" INPUT_DEVICES="keyboard evdev mouse" KERNEL="linux" LCD_DEVICES="bayrad cfontz cfontz633 glk hd44780 lb216 lcdm001 mtxorb ncurses text" LINGUAS="en" USERLAND="GNU" VIDEO_CARDS="nv nvidia vesa r128"
Unset:  CTARGET, EMERGE_DEFAULT_OPTS, INSTALL_MASK, LANG, LC_ALL, MAKEOPTS, PORTAGE_COMPRESS, PORTAGE_COMPRESS_FLAGS, PORTAGE_RSYNC_EXTRA_OPTS

vipernicus · Post by **vipernicus** » Wed Mar 14, 2007 9:09 pm

@seren
You don't need --hash-style in your LDFLAGS anymore, hash-style=gnu is now the new default. Remove --hash-style from your ldflags and emerge -e world, and things should shrink considerably without both.

seren · Post by **seren** » Wed Mar 14, 2007 10:01 pm

@seren
You don't need --hash-style in your LDFLAGS anymore, hash-style=gnu is now the new default. Remove --hash-style from your ldflags and emerge -e world, and things should shrink considerably without both.

didn't know that, have any docs referring to this i would like to read up.

mbar · Post by **mbar** » Thu Mar 15, 2007 8:11 am

seren wrote:i would like to read up.

yep, same here

cruzki · Post by **cruzki** » Thu Mar 15, 2007 10:18 am

same here.

and another question, it's the default from what version of glibc and binutils?

seren · Post by **seren** » Fri Mar 16, 2007 8:50 pm

im just wondering how everyone is liking the snapshot i made? a little positive encouragement =P... i intend on doing weekly snapshots, im building a new rig but before i dismantle this pc and wait for the parts to arrive ill do a snapshot from there it will be weekly.

Thanks

Seren

n0rbi666 · Post by **n0rbi666** » Fri Mar 16, 2007 8:59 pm

seren wrote:im just wondering how everyone is liking the snapshot i made? a little positive encouragement =P... i intend on doing weekly snapshots, im building a new rig but before i dismantle this pc and wait for the parts to arrive ill do a snapshot from there it will be weekly.

For me (~x86) it works okay ;] thx for ebuild

nbvcxz · Post by **nbvcxz** » Sun Mar 18, 2007 1:13 pm

@seren: your ebuild is working quite good, but can't apply the gcc-4.3.patch - so this is the second (except of coreutils) package that must be compiled with gcc-4.1 (later I will try with gcc-4.2 which should also work)

mbar · Post by **mbar** » Sun Mar 18, 2007 6:07 pm

nbvcxz wrote:@seren: your ebuild is working quite good, but can't apply the gcc-4.3.patch

same here, this simply fails

nesl247 · Post by **nesl247** » Sun Mar 18, 2007 9:13 pm

seren, would you mind working on the toolchain overlay instead of maintaining your own? I can give you access. I'd prefer it this way so that people don't have to choose between a thousand overlays for similar ebuilds.

seren · Post by **seren** » Mon Mar 19, 2007 1:59 am

No not at all. please pm me with info.

seren · Post by **seren** » Mon Mar 19, 2007 2:10 am

nbvcxz wrote:@seren: your ebuild is working quite good, but can't apply the gcc-4.3.patch - so this is the second (except of coreutils) package that must be compiled with gcc-4.1 (later I will try with gcc-4.2 which should also work)

I haven't tested with gcc 4.3, can someone please paste the output of the compile or the reject log if it fails on patching.

Seren

nesl247 · Post by **nesl247** » Mon Mar 19, 2007 3:29 am

Added your jabber, accept nesl247@gmail.com.. Easier to chat there.

roderick · Post by **roderick** » Mon Mar 19, 2007 12:10 pm

So, previously we were using --hash-style=both. Now the indication is that this can be dropped as the default is gnu which is the new one.

What effect does having both set do now? And what will dropping this do? Any min ver of glibc/binutils when this occurred?

ianegg · Post by **ianegg** » Mon Mar 19, 2007 3:33 pm

roderick wrote:So, previously we were using --hash-style=both. Now the indication is that this can be dropped as the default is gnu which is the new one.

What effect does having both set do now? And what will dropping this do? Any min ver of glibc/binutils when this occurred?

Also, should this show up in the emerge output anywhere, ie. how can I check it's working?

vipernicus · Post by **vipernicus** » Mon Mar 19, 2007 4:10 pm

Gentoo adds a patch to latest binutils snapshots to make hash-style=gnu the default, and no it will not show up in output. When you use hash-style=both, it is like having double the size, because you have double the hash-styles.

Just look into the ftp://gentoo.mirrors.tds.net/gentoo/dis ... .0.tar.bz2

The patch is called: 77_all_generate_gnu-hash.patch

Code: Select all

--- binutils/ld/ldmain.c
+++ binutils/ld/ldmain.c
@@ -306,7 +306,7 @@ main (int argc, char **argv)
   link_info.gc_sym_list = NULL;
   link_info.base_file = NULL;
   link_info.emit_hash = TRUE;
-  link_info.emit_gnu_hash = FALSE;
+  link_info.emit_gnu_hash = TRUE;
   /* SVR4 linkers seem to set DT_INIT and DT_FINI based on magic _init
      and _fini symbols.  We are compatible.  */
   link_info.init_function = "_init";

This is the same thing they did with the dtags support.

I would assume that at some point there will only be hash-style=gnu.

Although, looking again, I'm starting to wonder if this is enabling hash-style=both...

Is this the old hash-style?

Code: Select all

   link_info.emit_hash = TRUE;

roderick · Post by **roderick** » Mon Mar 19, 2007 5:10 pm

Ok, re-compiling without the LDFLAG set... hmmm... I hope it still enables GNU (haha).

octoploid · Post by **octoploid** » Mon Mar 19, 2007 5:19 pm

ianegg wrote:
roderick wrote:So, previously we were using --hash-style=both. Now the indication is that this can be dropped as the default is gnu which is the new one.

What effect does having both set do now? And what will dropping this do? Any min ver of glibc/binutils when this occurred?
Also, should this show up in the emerge output anywhere, ie. how can I check it's working?

Just run:

Code: Select all

readelf -a a.out |  grep GNU_HASH

where a.out is the program or library you want to check.
If it outputs something like:

Code: Select all

  [ 4] .gnu.hash         GNU_HASH         0000000000400390  00000390
 0x000000006ffffef5 (GNU_HASH)           0x400390

then you are using hash-style=gnu or hash-style=both ...

vipernicus wrote: Although, looking again, I'm starting to wonder if this is enabling hash-style=both...

Is this the old hash-style?
Code: Select all
   link_info.emit_hash = TRUE;

Yes it is. It looks like hash-style=both is the default now.
You still have to explicitly specify hash-style=gnu in your LDFLAGS,
if you just want to use hash-style=gnu and not both.

vipernicus · Post by **vipernicus** » Mon Mar 19, 2007 6:37 pm

octoploid wrote:
ianegg wrote:
roderick wrote:So, previously we were using --hash-style=both. Now the indication is that this can be dropped as the default is gnu which is the new one.

What effect does having both set do now? And what will dropping this do? Any min ver of glibc/binutils when this occurred?
Also, should this show up in the emerge output anywhere, ie. how can I check it's working?
Just run:
Code: Select all
readelf -a a.out |  grep GNU_HASH
where a.out is the program or library you want to check.
If it outputs something like:
Code: Select all
  [ 4] .gnu.hash         GNU_HASH         0000000000400390  00000390
 0x000000006ffffef5 (GNU_HASH)           0x400390
then you are using hash-style=gnu or hash-style=both ...
vipernicus wrote: Although, looking again, I'm starting to wonder if this is enabling hash-style=both...

Is this the old hash-style?
Code: Select all
   link_info.emit_hash = TRUE;
Yes it is. It looks like hash-style=both is the default now.
You still have to explicitly specify hash-style=gnu in your LDFLAGS,
if you just want to use hash-style=gnu and not both.

Or, we could just rediff the patch and throw it in an overlay

StringCheesian · Post by **StringCheesian** » Mon Mar 19, 2007 9:05 pm

octoploid wrote:Yes it is. It looks like hash-style=both is the default now.
You still have to explicitly specify hash-style=gnu in your LDFLAGS,
if you just want to use hash-style=gnu and not both.

Is there a reason you'd want both as apposed to just gnu?

Xake · Post by **Xake** » Mon Mar 19, 2007 11:15 pm

StringCheesian wrote:Is there a reason you'd want both as apposed to just gnu?

Nope, not anymore since "hashstyle=gnu" seems to work as it should, "both" is more of a safe way.

seren · Post by **seren** » Tue Mar 20, 2007 1:12 am

Update
03/19/07

Change log:
Updated to 03/19/07 snapshot
Removed Bdirect patch from glibc-2.5-extra_patches-1.4.1 patch failed and I figured bdirect was replaced with hash-style so I didn't attempt to fix.
Migrated overlay to toolchain overlay *thanks nesl247*
Distfiles remain on googlecode

Code: Select all

layman -a toolchain

to add the overlay via layman

Code: Select all

layman -S

to update with layman

Code: Select all

svn co svn://svn.nesl247.org/toolchain_overlay/sys-libs/glibc/

vipernicus · Post by **vipernicus** » Tue Mar 20, 2007 3:51 am

Looks like you still need to do --hash-styles=gnu in order to get only GNU_HASH, if nothing is specified, then default is --hash-style=both.