Sticky bit in unsupported AUR packages

Hello,
In my ventures to provide a PKGBUILD for an app called gnump3d, I've come across something with the unsupported (only?) build script packages in AUR that can cause some problems.
Here is an (one) example of such a package:
$ wget --quiet aur.archlinux.org/packages/tomcat/tomcat.tar.gz
$ tar tvzf tomcat.tar.gz
drwxr-sr-x nobody/aur 0 2006-06-27 02:50:53 tomcat/
-rw-r--r-- nobody/aur 938 2006-06-27 02:49:10 tomcat/PKGBUILD
-rw-r--r-- nobody/aur 415 2005-06-16 20:11:45 tomcat/tomcat.conf.d
-rwxr-xr-x nobody/aur 706 2005-06-16 20:11:12 tomcat/tomcat
Notice the owner, group and the sticky bit on the directory. These permissions seem to be standard for the AUR build script packages.
Now, if use aurbuild to build this package, I get this content in the package:
drwxr-sr-x root/root 0 2006-08-08 01:48:05 etc/
drwxr-sr-x root/root 0 2006-08-08 01:48:05 etc/rc.d/
-rwxr-xr-x root/root 706 2006-08-08 01:48:05 etc/rc.d/tomcat
drwxr-sr-x root/root 0 2006-08-08 01:48:05 etc/conf.d/
-rw-r--r-- root/root 415 2006-08-08 01:48:05 etc/conf.d/tomcat
<SNIP>
I'm pretty sure that these sticky bits weren't intended. If I build the package with yaourt, which uses srcpac as a backend, things get more troublesome:
drwxr-sr-x root/549 0 2006-08-08 02:03:00 etc/
drwxr-sr-x root/549 0 2006-08-08 02:03:00 etc/rc.d/
-rwxr-xr-x root/549 706 2006-08-08 02:03:00 etc/rc.d/tomcat
drwxr-sr-x root/549 0 2006-08-08 02:03:00 etc/conf.d/
-rw-r--r-- root/549 415 2006-08-08 02:03:00 etc/conf.d/tomcat
<SNIP>
549 ought to be the group id of aur in the AUR repository.
Careless usage of makepkg can also make this happen. Not building as root, but in a fakeroot environment, seems to prevent it from happen.
From what I can see, the erroneous permissions are created when the install command is used in the PKGBUILD script, within the newly unpacked build script package.
This is certainly not meant to bash the authors of aurbuild or yaourt (or tomcat; many PKGBUILDs use install, mine included), I just wanted to raise the issue. Can something be done about this? Don't know where, though. Removing the sticky bit on the directory in the build script package would be one solution. Or maybe makepkg could warn about erroneous permissions. Or... Or something? Is it a bug or a feature?
Cheers

Ok fixed aurbuild. Now it extracts the tarball from AUR into /tmp then copies it into the build dir which inherets the modes and owenership of the parent directory there (being $HOME), I'm pretty sure this is how tar does it.
As pointed out above this setgid issue only occurred during root user builds.
More info:
-v1.5.1 (August 8, 2006)
*Fixed inhereted setgid bit in the built the package resulting from AUR's set mode of the parent directory in the tarball.
The unwanted inhereted bit occured only under the following circumstances:
- user was root.
- a 'install -m' line was excuted in the PKGBUILD without explicitly setting the first pair of octets, ie 644 instead of 0644
- Any other type of 'chmod' command without the first octet set.
If you have built any packages meeting the criteria list above, rebuild and install offending packages now. Run the update command to get the list of all
packages built from AUR. String them together in a space separated line to aurbuild and use the menu to see if any install/chmod lines were used. You can
use (s) from the menu to skip the current package and move on to the next.

Similar Messages

  • What are the rules for AUR packages that rely on Alien bins [solved]

    Hi,
    I use ArchLinuxArm and want to create an AUR package that gets binaries from, lets say Fedora ARM, extracts them and modifies the config files in order to make it compatible with the Arch way of Linux. Of course, the PKGBUILD I am going to write will limit installation on the appropriate arcitecture, e.g. armv7h, then.
    Is this generally allowed or will my package be deleted after submission to AUR?
    Best,
    RaumZeit
    Last edited by RaumZeit (2013-10-27 01:38:31)

    AUR queries should be posted to the aur-general mailing list.
    As indicated in the thread linked by Karol above, it is permitted to append unsupported architectures to the arch array of a PKGBUILD that builds on officially supported architectures. That does not necessarily mean that we will allow packages that only build on unsupported architectures. This should be discussed on the mailing list. I have a vague memory of this coming up before and I think the consensus was that such packages would not be supported, but that consensus may have changed. Personally, I do not see a problem with such packages right now and I expect official supported to be extended in the future,  so I am in favor of allowing them.
    Packages that install pre-compiled binaries should be distinguished from normal packages with a "-bin" suffix. Such packages are permitted as long as they do not violate applicable licenses.

  • How to inspect AUR package PKBUILD and .install files

    Hi,
    Linux and arch newbie here. I was reading the wiki article about the AUR and noticed this bit:
    Warning: Carefully check all files. cd to the newly created directory and carefully check the PKGBUILD and any .install file for malicious commands. PKGBUILDs are bash scripts containing functions to be executed by makepkg: these functions can contain any valid commands or Bash syntax, so it is totally possible for a PKGBUILD to contain dangerous commands through malice or ignorance on the part of the author. Since makepkg uses fakeroot (and should never be run as root), there is some level of protection but you should never count on it. If in doubt, do not build the package and seek advice on the forums or mailing list.
    This is something that have not been doing at all in the past, but I am trying to improve my practices managing my system.
    The problem is, I do not know what exactly I am looking at or for in these files. If I give these files a look over before installing the package, can I honestly expect to spot something malicious? What would I need to learn to notice if something was fishy?
    Anyway, I am not to worried about this practically, because I only use a handful of AUR packages and I usually install ones based on recommendations, not just at random. But it still seemed interesting for the wiki to stress this so strongly. How important is this guidline anyway?
    Thanks!
    [EDIT: spelling]
    Last edited by supernerd (2014-06-25 10:41:13)

    I scan the whole PKGBUILD. I start by ensuring that the source link to the original source looks accurate. For example, take the source line for gmusicbrowser-git:
    source=("${pkgname}::git+http://github.com/squentin/gmusicbrowser.git")
    I know this is the correct link to the source, and so it passes my check. But suppose it had said:
    source=("${pkgname}::git+http://youvebeenhackedhub.com/1337haxorz/gmusicbrowser.git")
    I would become suspicious. Of course this is an exaggeration, but common sense goes a long ways here. At least check the first time..
    With the source verified, I ensure that the md5sum or sha256sum block has a sum. This way, if a download is compromised at the source, the sha256 or md5sum can catch it before you installed (this assumes that the PKGBUILD is not "bad" and has the sum number of a package that wasnt compromised). Note that with git this isnt necessary (the git process protects against such problems). Anytime a tarball is downloaded and extracted however, the sums should be present in the PKGBUILD. If I go to install an AUR package that has 'SKIP' for the md5sum/sha256sum block, I will double or triple check the source of the tarball (or of the patch files enclosed in the build directory, etc..)
    I also look for any "dangerous" commands in the build and install sections. For example, if I see "rm -rf" I had better see something like $pkgdir to start the directory path or be VERY sure the path is "safe". Since makepkg is not run as root this should theoretically not be a problem, but imagine if someone put "rm -rf /home/*" (warning: do not run that command on your system!) in there! This is mostly common sense; in time as you get more comfortable with bash and various linux commands it will make more and more sense and you will be able to spot mistakes.
    Also, consider the user posting the pkgbuild. "Trusted Users" are selected as trustworthy members of the community, so obviously you can feel much more comfortable with PKGBUILDS they have made (Xyne comes to mind..). For people you may not know, check what other PKGBUILDs they have available. After awhile, you develop a trust for certain people whos PKGBUILDs or software you have used. For example, I wouldnt hesitate to build/install using a PKGBUILD put up by BurntSushi since I use some of his software, have personally corresponded with him, and find him to be responsible. You might "develop" such rapport with other AUR users I dont even know about.
    Consider the vote count of a package as an approximate metric. Dont discount a package because it has 0 votes- it may just be that not many people have use for that particular software. Ive considered hosting a PKGBUILD for "xfce4-terminal-nowindowhints"; consider that tilers generally ignore them anyway, and that my package would only be useful for someone literally using xfce4-terminal with pytyle. How high do you think the vote count would be (even if the PKGBUILD had 0 errors)? On the other hand, you at least have a good chance the PKGBUILD is solid if the package has 354 people voting for it.. That said, the package could have been well-maintained before (when it received a ton of votes), and the quality has dropped since- just be mindful of these trends.
    Finally, adding all of these things together will leave the odds of a malicious PKGBUILD affecting your system pretty slim, though its certainly not impossible. I have never (to my knowledge to be fair) encountered a malicious PKGBUILD, though I have found a few that had errors or outdated sources, etc.
    Last edited by GSF1200S (2014-06-29 10:13:50)

  • [SOLVED] Archiso : installing AUR packages on a live image

    Hi all,
    I'm quite an arch newbie, I'm trying to setup a live USB stick, with the help of archiso. My goal is to finally get an "audio oriented" system (with jack, ardour, qsampler, and so more...).
    During the setup everything was working very well, until I tried to add some AUR packages to the install.
    On the arch website, I found this tip, which gave me a great hope about this.
    I'm not so familiar with Arch package management, but ok, I try to make a test : adding the "qsampler" AUR package. It needs "linuxsampler", "qt4", and "liblscp" as dependencies. "linuxsampler" and "qt4" are official packages, so I just have to add them to packages.both in the archiso working directory. "liblscp" is an AUR package (with no dependency); so there is 2 AUR packages to install : "liblscp", and "qsampler".
    So I create a directory tree like described in the tip, download the two build packages from AUR, and for each of them I do (something) like described there :
    # tar -xvf tarball_filename.tar.gz
    # cd tarball_filename
    # makepkg --asroot
    # mv *.xz ..
    # cd ..
    # rm -r tarball_filename{,.tar.gz}
    And then:
    # repo-add customrepo.db.tar.gz *.xz
    (I'm staying as root because it's red written to stay as root for the image creation. I think it's stupid, but people make stupid things when they are desesperate. Sorry I didn't take the time to test the code above again, it's only memory, but it was very similar)
    I did the same for both architectures (i686 and x86_64), so that my custom repo looks like this:
    ~/liveusb/customrepo # ls -R
    i686 x86_64
    ./i686:
    customrepo.db customrepo.db.tar.gz liblscp-0.5.6-1-x86_64.pkg.tar.xz qsampler-0.2.3-1-x86_64.pkg.tar.xz
    ./x86_64:
    customrepo.db customrepo.db.tar.gz liblscp-0.5.6-1-x86_64.pkg.tar.xz qsampler-0.2.3-1-x86_64.pkg.tar.xz
    Oops... I just noticed I did wrong for i686 machines, but it doesn't matter for the moment, since I'm working on an x86_64 machine.
    As explained in the tip, I add the following lines to pacman.conf (in the archiso working directory):
    [customrepo]
    SigLevel = Optional TrustAll
    Server = file:///my/path/to/customrepo/$arch
    From my point of view, at this point, the USB stick is ready to be updated:
    ~/liveusb # ./build.sh -v
    and then (with /dev/sdf as my usb stick device):
    ~/liveusb # dd if=out/archlinux-2014.10.01-dual.iso of=/dev/sdf
    But when I boot on the USB stick, there's no trace of qsampler (linuxsampler, however, is present).
    Since it happened, I'm feeling like a lost, lonely man, on a desert island... Thinking about the "why", the "how"..., the meaning of life..., of package management... all this stuff
    I'm sure I did something wrong about the "custom repository", and the main reason is I don't deeply understand all the steps about this; that's why I'm looking for help
    Any idea?
    Many thanks
    Last edited by yolenoyer (2014-10-02 09:16:57)

    Thank you for the reply,
    I think I did a more trivial mistake :
    With archiso, the packages are automatically installed, from a package list file called "packages.both", and "packages.x86_64", "packages.i686" for architecture dependent packages. But they only use common repos by default. The "'qsampler" is not in official repos (that's why I choosed this one for my question).
    So, ok, I setup a common repo (with some mistakes but it was working), BUT...
    I just forgot to put the package name in the packages.both file...
    So, now that I did it, I just have an error about the package architecture, which I think possible to fix, just by rebuilding the common repo in a correct manner:
    ~/liveusb # ./build -v
    warning: vlc-2.1.5-3 is up to date -- reinstalling
    warning: mplayer-37224-2 is up to date -- reinstalling
    error: failed to prepare transaction (package architecture is not valid)
    :: package qsampler-0.2.3-1-x86_64 does not have a valid architecture
    ==> ERROR: Failed to install packages to new root
    Trilby wrote:Also, does this need to be a static iso image - is there a reason not to just do a persistent usb install?
    About the static iso image : the idea of building an iso image with all my personnal tools and config already installed on it is very pleasant to me, including the fact that you can burn it on a CD as well. The persistent acpect would be pleasant too, but in a secondary way.
    While I'm writing this message I really understand a bit more about all of this, since yesterday... Sometimes, simply posting a message in forums helps you to understand your own problem, because you have to be clear and concise!
    Thanks

  • [Solved] Cant install\compile certain AUR packages

    This is a brand new 32-bit build on my laptop. Everything is working great, but for some reason I cant get some AUR packages. I have Yaourt, pacman-color, and slock installed from AUR and working fine, but I cant get mpd-git or mplayer-svn installed? I have gcc, python installed and tried both root and user accounts. Not sure what is missing. This is a new dual core laptop so not sure why I am getting the CPU errors below?
    This is for mplayer-svn:
    ==> SVN checkout done or server timeout, updating build dir
    ==> Applying disabled-features patch...
    /var/abs/local/yaourtbuild/mplayer-svn/./PKGBUILD: line 56: patch: command not found
    Detected operating system: Linux
    Detected host architecture: i386
    Checking for host cc ... gcc
    Checking for cross compilation ... yes
    ./configure: line 1605: gcc: command not found
    ./configure: line 1610: gcc: command not found
    Checking for CPU vendor ... GenuineIntel (6:15:11)
    Checking for CPU type ... Intel(R) Core(TM)2 Duo CPU T7500 @ 2.20GHz
    Checking for kernel support of mmx ... failed
    It seems that your kernel does not correctly support mmx.
    To use mmx extensions in MPlayer, you have to upgrade/recompile your kernel!
    Checking for kernel support of mmxext ... failed
    It seems that your kernel does not correctly support mmxext.
    To use mmxext extensions in MPlayer, you have to upgrade/recompile your kernel!
    Checking for kernel support of sse ... failed
    It seems that your kernel does not correctly support sse.
    To use sse extensions in MPlayer, you have to upgrade/recompile your kernel!
    Checking for kernel support of sse2 ... failed
    It seems that your kernel does not correctly support sse2.
    To use sse2 extensions in MPlayer, you have to upgrade/recompile your kernel!
    Checking for kernel support of ssse3 ... failed
    It seems that your kernel does not correctly support ssse3.
    To use ssse3 extensions in MPlayer, you have to upgrade/recompile your kernel!
    Checking for kernel support of cmov ... failed
    It seems that your kernel does not correctly support cmov.
    To use cmov extensions in MPlayer, you have to upgrade/recompile your kernel!
    Checking for mtrr support ... yes
    Checking for GCC & CPU optimization abilities ... CPU optimization disabled. CPU not recognized or your compiler is too old.
    error
    Checking for byte order ... failed to autodetect byte order, defaulting to little-endian
    Checking for extern symbol prefix ...
    Error: Symbol mangling check failed.
    Check "configure.log" if you do not understand why it failed.
    ==> ERROR: Build Failed.
    Aborting...
    Error: Makepkg was unable to build mplayer-svn package.
    This is for mpd-git:
    Looks like I need to set the path for gcc in here, but not sure how?
    configure: error: in `/tmp/yaourt-tmp-banshee/aur-mpd-git/mpd-git/src/mpd-build':
    configure: error: no acceptable C compiler found in $PATH
    See `config.log' for more details.
    make: *** No targets specified and no makefile found. Stop.
    ==> ERROR: Build Failed.
    Aborting...
    Error: Makepkg was unable to build mpd-git package.
    Last edited by banshee28 (2009-09-03 04:15:43)

    Allan wrote:
    You need to install the base-devel group (pacman -S base-devel).
    This is a bit weird though....
    Checking for cross compilation ... yes
    hopefully it goes away!
    Ah, I see all the apps in this group now! Not sure how I missed this? Was this in the wiki somwhere??
    Anyways,its getting further along now ....

  • [Solved] List AUR packages installed and only AUR packages.

    Here's a good one.  Thought this would be easy but thought it over and then looked around a bit and haven't found anything.  There's might just be an easy way to do this that will make me *bonk* my head in the morning but I haven't found it yet.  I'm looking to be able to just list the packages I have installed from AUR and not any that I have gotten from the official repos.  I've checked out some utilities in AUR (like AURcheck) but as far as I can tell they just look for AUR updates.  Anyone know of a way to do this?
    Last edited by Gen2ly (2009-10-30 14:32:22)

    I just got to reinstalling and this was a lifesaver - it worked great.  Thank for the help, brisbin, ghost, Allan...
    @Ghost, I would have used packup but I had a couple downgraded packages and I wanted to be able to troubleshoot it.
    The tip about the grep -v doing 'shortnameplus' was a good tip, Profjim.  I hadn't read this last post before and during the reinstall I was a bit surprised nvidia wasn't installed so... all is good now.
    I created a script to be able to create the backup list and restores from it simliar to ghosts and am able to run it in cron job.  Probably not a big deal, but... phhht.  Here it is for anyone that can use it:
    #!/bin/bash
    # pacbac - Create and restore from list all installed packages
    # Package list locations (official and local)
    pkglsoff=/opt/backup/pc-emach/arch-pkglist-official
    pkglsloc=/opt/backup/pc-emach/arch-pkglist-local
    # Exclude packages
    excldoff=()
    excldloc=()
    # Use filename as program name
    prog=${0##*/}
    # Text color variables
    bldblu='\e[1;34m' # blue
    bldred='\e[1;31m' # red
    bldwht='\e[1;37m' # white
    txtcyn='\e[0;36m' # cyan
    txtund=$(tput sgr 0 1) # underline
    txtrst='\e[0m' # text reset
    info=${bldwht}*${txtrst}
    pass=${bldblu}*${txtrst}
    warn=${bldred}!${txtrst}
    # If restoring, be sure yaourt is installed
    if [[ "$1" == 'r' ]] && [[ -z $(pacman -Qs yaourt) ]]; then
    echo ""
    echo -e "$warn $prog requires ${txtund}}Yaourt${txtrst} to be installed."
    echo -e " ${txtcyn}http://wiki.archlinux.org/index.php/Yaourt${txtrst}"
    echo ""
    exit
    fi
    case $1 in
    b ) # Create list of official repository packages (core, extra, community)
    echo -e "$info Creating list of official (core/extra/community packages) packages installed."
    # Create list, remove local, base
    pacman -Qqe | grep -vx "$(pacman -Qqg base)" | grep -vx "$(pacman -Qqm)" > "$pkglsoff"
    # Create list of local packages (includes the AUR)
    echo -e "$info Creating list of local (includes AUR) packages installed."
    pacman -Qqm > "$pkglsloc"
    echo -e "$pass Official package list saved to ${txtund}"$pkglsoff"${txtrst}"
    echo -e "$pass Local package list saved to ${txtund}"$pkglsloc"${txtrst}" ;;
    r ) # Update repository database, then restore packages from list
    echo -e "$info Installing packages from lists"
    sudo pacman -Sy
    # use -f to overwrite conflicting files
    sudo pacman -S --needed $(cat "$pkglsoff")
    # Yaourt doesn't have --needed check
    yaourt -S $(cat "$pkglsloc" | grep -vx "$(pacman -Qqm)") ;;
    * ) echo -e " pacbac b - build installed packages list. (dir:${txtund}"${pkglsoff%/*}"${txtrst})"
    echo -e " pacbac r - restore installed packages from package list." ;;
    esac
    Last edited by Gen2ly (2009-10-31 14:16:55)

  • [AUR] Packages orphaned?

    Hello,
    I'm a bit curious about my packages in AUR. I've seen today, that quite a lot (20) packages of mine got orphaned, as "perl-algorithm-annotate". This module is NOT in aur by an other user, and NOT in current/extra/community.
    Yes, it points to an invalid URI, since CPAN once more changed the uris of the "module" pages... Is that the reason why you actually orphan? And why orphan, and not just drop me a message? I'd have worked it off, changing all those packages... but this action by (which TU ever) is quite strange.
    Since it was not me who orphaned them - tell me, why are they orphaned?
    Yours,
    STi
    Last edited by STiAT (2007-05-18 10:23:21)

    Many of them have been moved to [community] and they are automatically orphaned when this happens, they will be adopted by the TU that merged them to [community] shortly.
    Xterminus is the TU in question and he has long maintained a massive perl repo which is automtically updated.  He has now combined this with the AUR/[community] to provide over 100 perl pkgs he can easily provide more too.
    He has apologised for basically muscling in on other peoples perl pkgs but because it was an automated process it would have been very hard for him to contact all you individually.
    Here are links to the two posts he made to the TU list:
    [tur-users] perl stuff
    [tur-users] perl followup

  • Aurebuildcheck - checks if aur packages need rebuild

    Hi everyone, I made a small script which checks locally installed packages for missing dependencies using ldd.
    The script basically gets the list of files contained in local packages and runs ldd on the files to check if dependencies are missing (kinda like findbrokenpkgs or lddd found in "devtools").
    However, since it only checks the files of local packages, it's much faster than lddd or findbrokenpkgs which check all packages.
    The aur package is called aurebuildcheck-git and can be found at https://aur.archlinux.org/packages/aurebuildcheck-git/ .
    Have fun.

    Ok, thank you!
    It's a bit more hacky now, and a bit slower.
    localpackages=`pacman -Qqm`
    localpackagesamount=`echo $localpackages | wc -w`
    echo -e "${localpackagesamount} packages will be checked...\n"
    brokenpkgs=""
    localpackages=`pacman -Qqm`
    for package in $localpackages ; do
    BROKEN="false"
    echo "checking ${package}..."
    packagefiles=`pacman -Qql $package | grep -v "\.a\|\.png\|\.la\|\.ttf\|\.gz\|\.html\|\.css\|\.h\|\.xml\|\.rgb\|\.gif\|\.wav\|\.ogg\|\.mp3\|\.po\|\.txt\|\.jpg\|\.jpeg"`
    IFS=$'\n'
    for file in $packagefiles; do
    if (( $(file $file | grep -c 'ELF') != 0 )); then
    # Is an ELF binary.
    libs=`readelf -d "${file}" | grep "NEEDED.*\[.*\]" | awk '{print $5}' | sed -e "s/\[//" -e "s/\]//"`
    for lib in ${libs} ; do
    # needed libs
    if [ -z `whereis ${lib} | awk '{print $2}'` ] ; then
    # Missing lib.
    echo -e "\t ${RED}${file}${NC} needs ${REDUL}$lib${NC}" # >> $TEMPDIR/raw.txt
    BROKEN="true" # to avoid packages being listed in the brokenpkg array several times
    fi
    done
    fi
    done
    if [[ ${BROKEN} == "true" ]] ; then
    brokenpkgs="${brokenpkgs} ${package}"
    fi
    done

  • Deleted AUR packages

    Hi,
    Is there any policy regarding the deletion of some AUR packages ?
    Recently, I packaged rinse which is similar to debootstrap, but for RPM-based distributions. rinse requires rpm but it seems rpm was deleted. Can anyone tell me why ?
    Thanks.
    Mildred

    brebs wrote:
    Allan wrote:all deletions had to be approved by a TU
    Who writes a web app and doesn't think of the potential ways of abuse, anyway?
    People ...
    ... with less time and energy?
    ... who do believe the honour system works?
    ... who are not as cynical as some seem to be?
    I realise your question is rhetorical but it seemed a bit uncalled for ergo I am answering in an uncalled manner.

  • [SOLVED]Intel XDK AUR package is not running

    Hi,
    I have installed Intel-XDK from AUR:
    https://aur.archlinux.org/packages/intel-xdk/
    The install completed successfully but when i tried to launch the program by typing intel-xdk or intel-xdk.sh nothing happens. And there is no error message. This is freshly installed 64 bit ArchLinux with PekWM.
    Last edited by Paingiver (2015-02-02 22:21:17)

    Thank you Lone_Wolf for alerting me about the matter.  I think it is a problem for some other programs too because someone created an AUR package just for this symlink:
    https://aur.archlinux.org/packages/libudev.so.0/
    When the package didn't run without any error message i emailed the AUR maintainer. He quickly replied and said he is a developer at Intel and will look into matter when he got free time. I searched the internet but couldn't find a lot of information about it. Then i stumbled upon a thread on Manjaro(https://forum.manjaro.org/index.php?topic=11695.0) forums. In their system AUR package was giving the error of libudev.so.0 not found and suggested to install AUR package libudev.so.0. Installed that and program worked.
    What can we do about that beside that symlinking?
    Later edit: Also i stumbled upon some post that Intel-Xdk Linux install script looks for a package named gtk2. If your distro's gtk2 package is not named gtk2 it is not installing. Someone created a dummy gtk2 package and fooled the program to run.
    Why install scripts from big companies always have problems on Linux? Until today there isn't even one package that haven't given me problems. On the other side installed a lot of Linux distros and %99 of time it went flawlessly.
    Last edited by Paingiver (2015-02-04 00:33:28)

  • Setting the sticky bit on a file fails in a zone

    On initial inspection, the sticky bit on directories seem to be consistent across both the Global zone and local zone. However, they are inconsistent for files. Is this a bug, and if it is how to I raise a defect report?
    Excerpt from chmod man page:
    If a regular file is not executable and has {stick bit} set, the file is assumed to be a swap file. In this case, the system's page cache will not be used to hold the file's data. If the {stick bit} bit is set on any other file, the results are unspecified.
    Solaris 10 update 4 (Global Zone).
    # uname -a
    SunOS mumble.amtest.com 5.10 Generic_120011-14 sun4v sparc SUNW,Sun-Fire-T1000
    # chmod 1777 testfile
    # ls -l testfile
    -rwxrwxrwt 1 root other 0 Jul 11 14:24 testfile
    # chmod 1770 testfile
    # ls -l testfile
    -rwxrwx--T 1 root other 0 Jul 11 14:24 testfile
    # chmod 1000 testfile
    # ls -l testfile
    ---------T 1 root other 0 Jul 11 14:24 testfile
    Solaris 10 update 4 (Local Zone)
    # uname -a
    SunOS qatamos-z1 5.10 Generic_120011-14 sun4v sparc SUNW,Sun-Fire-T1000
    # chmod 1777 testfile
    # ls -l testfile
    -rwxrwxrwx 1 root other 10 Jul 10 14:59 testfile
    # chmod 1770 testfile
    # ls -l testfile
    -rwxrwx--- 1 root other 10 Jul 10 14:59 testfile
    # chmod 1000 testfile
    # ls -l testfile
    ---------- 1 root other 10 Jul 10 14:59 testfile

    It isn't the fact that the bit isn't set. I guess the biggest complaint is that there is no error when trying to set the bit.
    chmod returns success, even though it failed to set the bit.
    Normally, if someone does not have enough privileges to modify permissions on a file, an error is returned.

  • How to list AUR packages in terminal with yaourt or other helper?

    I can list packages I have installed from AUR, for example to find all the developer components of Xfce that I have installed, I can run the following command:
        pacman -Qim | grep -E "Name           : xf".+devel
    Which outputs:
        Name           : xfce4-appfinder-devel
        Name           : xfce4-dev-tools-devel
        Name           : xfce4-panel-devel
        Name           : xfce4-session-devel
        Name           : xfce4-settings-devel
        Name           : xfdesktop-devel
        Name           : xfwm4-devel
    But, how do I find out if there are other packages that match the same regex pattern available in the AUR, maybe using yaourt? When I try:
        yaourt -Si | grep -E "Name           : xf".+devel
    I get no output. When I don't include .+devel, I find out that none of the AUR packages are included. If I try to force it to search the AUR by including -a, it says that's not a valid option.
    So, how do I search the AUR from the terminal?

    falconindy wrote:
    cower transparently supports regex -- but not thanks to the AUR. In reality, the query "xf.*-devel" is actually asking the AUR to search for "xf". The returned results are filtered against the regex "xf.*-devel" and out pops magic.
    Yaourt doesn't do this.
    I saw that, but Yaourt should be able to simply query the AUR without it trying to install stuff, but how? Does anyone know? I knew yaourt didn't support regex, that's why I was piping the output to grep. It worked for pacman, and works for yaourt when it uses pacman, but I can't figure out how to get it to return the contents of the AUR so I can pipe those contents through grep.

  • AUR package interface - sort criteria out of action

    The "search by" field on the main AUR package list (http://aur.archlinux.org/packages.php) now only contains name and maintainer.  There used to be at least 5 options, one of which was the ability to search by age.  Searches in reverse order are also no longer possible.
    Problem exists regardless of logged in status.
    Can this be corrected please.

    Snowman wrote:
    tomk wrote:
    Probably something to do with this.
    Post it on Flyspray.
    No. It's due to an AUR update: http://www.archlinux.org/blog/2005/10/21/aur-127/
    You need to click on the column headers to sort.
    So how are we supposed to sort by age? It seems like that's the only really important sorting need, to see the latest packages in aur (aside from the 10 listed on the front page).

  • Sticky Bit - Problem...

    Hello,
    I have one thing to ask, to clarify... Some weeks ago after a system migration from win/sql to linux/oracle (and after an oracle upgrade from 10.2.0.4 to 11.2.0.1), we faced an error during DB13, the job Scheduled Update Statistics finished with an error! After some research, we found the solution... the reason was the version of oracle instant client that was the version 10.2.0.2 (the first version of this OIC)!
    So after I performed the update of oracle instant client (OIC) to version 10.2.0.4 (V3) the error finally disappeared... and it was the solution, which in that moment I thought it had been responsible for this correction!... but now I'm not so sure! Why?! Because the same problem appear again on another system, I performed the same solution, I updated the oracle instant client to version 10.2.0.4 v3 but unfortunately the problem still continue appear! So what I concluded, was that is not the solution, I realized this is not only because of the old OIC version that is installed in our system... there is other problem which is related to permissions!
    Later I discovered that the STICKY BIT (?) had been applied before I have upgraded the oracle instant client... and in fact in this new system, after the solution of oracle client update the error still continued appear and in fact only after the application of sticky bit in this system the error definitely no longer appear!
    So my question is what is this?... what is Sticky Bit?
    I looked in SAP Marketplace if there were any related sapnotes to sticky bit but I didn´t found anything... this is some mandatory procedure?... if so, sap recommends the application of this where? in which situations?
    I don´t know exactly what is STICKY bit and what he do... but the truth is that from the moment I applied it, with command ./saproot.sh (with root) on /sapmnt/SID/exe, we resolve the other part of this problem... I tried to search sapnotes in SAP marketplace about this sticky bit but I didn´t find anything... Can you explain me better if sap advice to apply this sticky bit... and if it can origin any problem or if there are any conflict with other things when we applied it on system!??
    Kind regards,
    Joao Dimas - Portugal

    Now I think I understand the meaning and what is its functionality... but explain me another thing, the application of sticky bit with ./saproot.sh <SAPSID> changed other four files: brarchive; brbackup; brconnect and icmbnd  (this five files now have a red colour)...
    Yes, and they have an "s" instead of an "x" if you use ls -l.
    The same thing is true for those files. The br*tools must be run as the user running/owning the database software (oraadm during startsap.
    So, if this is so important to run (because if I didn´t ran this command the job throught DB13 can not runs) why there isn´t sap notes or sap documentation, discussing this issue and the need to apply the sticky bit to fix this problem?
    It's documented - but hidden:
    http://help.sap.com/saphelp_nw70ehp1/helpdata/en/43/3753c4b87b6025e10000000a422035/frameset.htm
    Markus

  • [SOLVED] can't build an AUR package while installing Arch

    So I am nearly there installing Arch for the first time. And I will start with saying that I am really sorry if I missed the answer to my problem but I have had a look around the documentation and the general webz for something regarding this problem and haven't found anything.
    I am trying to set my wireless connection up but I apparently need the AUR b43-firmware package installed for my Broadcom BCM4322 [14e4:432b] to be functional.
    So I follow the documentation to install an AUR package, but I am stuck at the building stage as makepkg does not allow me to build if I am root, and I am root by default while installing/configuring Arch...
    What should I do to build and install that package so I have a wireless connection later on?
    Cheers
    Last edited by chtfn (2015-03-21 14:11:12)

    I am connected to the wired network, and I followed the steps to make it persistent for later, but I just wanted to go through the wireless stuff too to be extra sure I will be able to connect to the Internet later on. As it is the first time I install Arch, I am trying to be extra cautious, and having that security would make me feel more comfortable! Plus, I am keen to learn how to do those things from the command line.
    Isn't there a way to switch to a normal user just for a command, and automatically reverting back to root when that action is finished? Just like sudo but the other way round

Maybe you are looking for