- Arch-general - lists.archlinux.org

[arch-general] What about unused direct libraries in ELF files?
by Gerardo Exequiel Pozzi 10 Apr '09

10 Apr '09

Hi Like my previus message about RPATH in ELF files, I am interested in know opinions of arch-dev-team about ELF files that have direct unused libraries. (I will open a ticket if considered this important in Arch Linux, and gradually helping to fix it for each package that are necessary, like for rpath [#1], also a check for namcap can be implemented like for rpath [#2]) This problem, is in most cases about this issue is the way that libtool or pkg-config is used in the build systems. For example, "pacman" package the pacman binary really only depends on libc.so.6 and libalpm.so.3 all other libs (libdownload.so libarchive.so.2 libacl.so.1 libattr.so.1 libbz2.so.1.0 libz.so.1) are depends of libalpm.so.3 only. Other example are "convert" from imagemagick: It has many unused direct libraries: $ ldd -u -r /usr/bin/convert Unused direct dependencies: /usr/lib/liblcms.so.1 /usr/lib/libtiff.so.3 /usr/lib/libfreetype.so.6 /usr/lib/libjasper.so.1 /usr/lib/libjpeg.so.62 /usr/lib/libgs.so.8 /usr/lib/libpng12.so.0 /usr/lib/libfontconfig.so.1 /usr/lib/libwmflite-0.2.so.7 /usr/lib/libXext.so.6 /usr/lib/libXt.so.6 /lib/libbz2.so.1.0 /usr/lib/libIlmImf.so.6 /usr/lib/libImath.so.6 /usr/lib/libHalf.so.6 /usr/lib/libIex.so.6 /usr/lib/libIlmThread.so.6 /usr/lib/librsvg-2.so.2 /usr/lib/libgdk_pixbuf-2.0.so.0 /usr/lib/libcairo.so.2 /usr/lib/libgobject-2.0.so.0 /usr/lib/libgmodule-2.0.so.0 /usr/lib/libglib-2.0.so.0 /usr/lib/libxml2.so.2 /usr/lib/libz.so.1 /usr/lib/libgomp.so.1 /lib/libpthread.so.0 /usr/lib/libltdl.so.7 /usr/lib/libSM.so.6 /usr/lib/libICE.so.6 /usr/lib/libX11.so.6 This not only affect executables, also libraries. So very complicated dependency "graphs" are here. Derived from this, another issue appears: undefined symbols, because in shared libs this behavior are permited (ldd -u -r /usr/lib/libgdbm_compat.so.3.0.0, but this is a documented [a program that uses it also needs a -lgdbm] ) For example now the KDE packages (from Pierre) [#3] when solved all RPATH also solved this issue, with linking flags "--as-needed" and "--no-undefined" Good Luck! [#1] http://bugs.archlinux.org/task/13890 - Packaging creation and rpath issues [#2] http://bugs.archlinux.org/task/14049 - [namcap] should check about insecure RPATH in ELF files [#3] http://bugs.archlinux.org/task/13965 - [kde] Compile KDE with CMAKE_SKIP_RPATH -- Gerardo Exequiel Pozzi ( djgera ) http://www.djgera.com.ar KeyID: 0x1B8C330D Key fingerprint = 0CAA D5D4 CD85 4434 A219 76ED 39AB 221B 1B8C 330D

2 1

[arch-general] Keyboard repeat in KDE
by Keith 10 Apr '09

10 Apr '09

Leonid wrote: >> Hi, I know I sent this before, but this is a new version of KDE and the problem is becoming really bad. Every once in a while, keyboard repeat will suddenly be unset. I can fix it in systemsettings, only to have it get unset again randomly. Sometimes it will only take a few minutes, sometimes hours. It is very very annoying. Is there anything I can do to fix it? I am running 4.2.2 (not kdemod, vanilla). << +1. It appeared to hit me immediately after either the upgrade to KDE 4.2.2 (or, more likely, I think, the upgrade to xorg-server 1.5.3-5 at 1 AM 4/6). I appear to be having the same problem, and after the same upgrade - to KDE 4.2.2 (not KDEMOD) and xorg-server in extra. Also, my mouse "third-button-wheel" will stop scrolling in a web page, for no reason. It happens so often I'm avoiding using the function - and this effect also seems to be tied to the same time frame. Note also that, in the course of modifying parts of /etc/X11/xorg.conf to get to a usable desktop after this upgrade, I had to (a) remove all my "InputDevice" Sections for a keyboard and two mice (which implicitly sets several possible-gremlin options), and then (b) make a NUMBER of mods, trying to get xorg running stably with a "clean" an Xorg.log.0 as possible. I'm currently trying to figure out how to turn the clues it throws me into a usable conf, but with documentation so blatantly at odds with reality, in at least a couple of interesting places, that's hard. Try 'Option "Accel"' in the Screen Section ('man Xorg', or 'man xorg.conf'). The Xorg (or xorg.conf) man page documents it beautifully; hopefully some day it'll actually OPERATE as documented. This weekend I intend to exercise every known method of automagically generating an xorg.conf for an AMD64-based Arch Linux system, and see if I can either find what I'm missing (in one of them), or manage to get some combination of the suggestions (of all of them) to work. I realise both documentation and KNOWLEDGE are in short supply with the imminent release of xorg-server 1.6, and that it's not supposedly going to be an "easy" or "simple" "migration." I see this as mental prep work... Blue Skies...g

1 0

[arch-general] xf86-video-intel-legacy vs xf86-video-intel
by Damjan Georgievski 09 Apr '09

09 Apr '09

Is it possible to make these packages not conflict each other? So that one could have both installed and alternate when needed? -- damjan

5 4

[arch-general] xinput in ArchLinux
by Damjan Georgievski 09 Apr '09

09 Apr '09

There's a PKGBUILD in AUR of xinput, but I think it needs to be with the other Xorg packages in extra, probably even in the "xorg" group. xinput is a utillity that can dynamically change different properties of the X input devices like the ones that otherwise need to be configured in the xorg.conf file or the HAL fdi file. For example, enabling wheel emulation of a trackpoint. You can also use it to enable or disable input devices. -- damjan

4 6

[arch-general] Keyboard repeat in KDE
by Leonid Grinberg 09 Apr '09

09 Apr '09

Hi, I know I sent this before, but this is a new version of KDE and the problem is becoming really bad. Every once in a while, keyboard repeat will suddenly be unset. I can fix it in systemsettings, only to have it get unset again randomly. Sometimes it will only take a few minutes, sometimes hours. It is very very annoying. Is there anything I can do to fix it? I am running 4.2.2 (not kdemod, vanilla). -- Thanks in advance, Leonid

2 1

[arch-general] [signoff] kernel-2.6.29.1-3
by Tobias Powalowski 09 Apr '09

09 Apr '09

Hi Major Upstream Changes: http://kernelnewbies.org/LinuxChanges Archlinux Changes: - added aufs2 to kernel package - removed unionfs support Signoffs: - Please test also with binary modules. - Please report functionality of aufs2. Known issues: - ipw2200 seems to have some issues Other devs / users should report how much impact this have? Pierre reported network scanning is not working correct, anyone else with issues? Thanks for your help. Hope to get kernel .29 series to core in next 2 days. greetings tpowa -- Tobias Powalowski Archlinux Developer & Package Maintainer (tpowa) http://www.archlinux.org tpowa(a)archlinux.org

4 4

[arch-general] libmtp update
by Federico Manganelli 09 Apr '09

09 Apr '09

Here's is a working PKGBUILD for libmtp 0.3.7, that is standing still for a while...It should address bug fs#10583 HTH thanks mangus

3 3

[arch-general] eject: unable to eject, last error: Inappropriate ioctl for device
by Csomay Mihaly 09 Apr '09

09 Apr '09

Hi, I have a script which works fine on Ubuntu but fails on Arch: -------------------------------------------------------- #! /bin/bash -eu # # Burn a DVD from the working directory # # Usage: burn <volume name> # # Creates an .iso file from the working directory in /tmp, # writes the image to /dev/dvdrw, then checks the integrity # of the disc. # # Prerequisites: # An empty DVD in /dev/dvdrw # The size of the working directory fits on a DVD VOLNAME="$1" TMPDIR=`mktemp -d` || exit 1 ISOFILE="$TMPDIR/$VOLNAME.iso" GENLOG="$TMPDIR/genisoimage.log" GROWLOG="$TMPDIR/growisofs.log" BURNER="/dev/dvdrw" echo "Image and logs in $TMPDIR" echo -n "Creating image... " genisoimage -udf -allow-limited-size -input-charset iso8859-2 -iso-level 4 -J -joliet-long -log-file "$GENLOG" -r -V "$VOLNAME" -v -o "$ISOFILE" . &> /dev/null echo "Done." echo -n "Burning image... " growisofs -dvd-compat -speed=16 -Z "$BURNER"="$ISOFILE" &> "$GROWLOG" echo "Done." echo -n "Reloading media... " eject "$BURNER" eject -t "$BURNER" echo "Done." echo -n "Verifying image... " EXTENTS=`egrep -e "^[0-9]+ extents written " "$GENLOG" | egrep -o -e "^[0-9]+"` MD5ISO=`md5sum "$ISOFILE" | egrep -o -i -e "^[0-9a-f]{32}"` MD5DVD=`dd if="$BURNER" bs=2048 count="$EXTENTS" 2> /dev/null | md5sum | egrep -o -i -e "^[0-9a-f]{32}"` if [ "$MD5ISO" == "$MD5DVD" ]; then echo "OK." echo -n "Cleaning up... " rm -r "$TMPDIR" echo "Done." else echo "ERROR!" fi echo "ISO MD5: $MD5ISO" echo "DVD MD5: $MD5DVD" eject "$BURNER" -------------------------------------------------------- When it reaches eject "$BURNER", it gives the error in the subject. If after that I type in "eject /dev/dvdrw" in that terminal, it ejects fine. After install I couldn't even eject by pressing the button on the drive, but the patch in here fixed that: http://bugs.archlinux.org/task/9537 I have xfce4. Where should I look to fix this? Thanks. csm

1 0

[arch-general] After upgrade to KDE 4.2.2 AltGr missing in Konsole and gtk apps
by waldek 08 Apr '09

08 Apr '09

Hi, could somebody have a look at this topic: http://bbs.archlinux.org/viewtopic.php?id=69393 in short, after the upgrade to KDE 4.2.2 AltGr key does not work in Konsole, Yakuake and gtk applications. It works in all other native KDE applications. maybe some KDE users with Gernam keyboard here who could check if they have the same problem before I open a bug report at kde.org. Cheers Waldek

1 0

Re: [arch-general] [arch-dev-public] [signoff] kernel-2.6.29.1-3
by Christopher Rogers 08 Apr '09

08 Apr '09

Aufs2 works fine. Other problem: But I think there is a upload bug with rtl8187 driver. With kernel 2.6.28.8 i can get to my full upload speed at the low end. With kernel 2.6.29 it starts at full upload speed then goes down from there. It bugs me cause i upload livecds and it fells like the new 2.6.29 is kill the connection with upload. kernel 2.6.28.8 upload speed: 500Kbyte/sec goes down to 256Kbyte/sec kernel 2.6.29 upload speed: 256Kbyte/sec goes down to like 90Kbyte/sec or less. Other then this bug it can go in to core. Since i don't think its going to get solved just yet.

1 0