| maks blog | |||||
|
Subscribe
Flavours
Blogs |
Fri, 08 Mar 2013
Wrong moves
The new klibc features fixes for latest 3.6 Linux, dash sync and arm fixes. For details see the longer story on: klibc 2.0.2 release announcement or klibc git repository. Thu, 31 May 2012The new klibc features prominently the merged stdio branch (causing the version bump). The buffered I/0 allows a faster fstype and thus faster boot. Plus klibc gained the support of several stream I/O functions, for details see the longer story on: klibc 2.0 release announcement or klibc git repository. The dhcp cmd ipconfig can now generate enough info for a proper lease file too and got fixed concerning the min length of it's replies. Google sponsored the /etc/fstab or cmdline boot mounting in kinit. 2.0 is shipping several arch fixes including i386 signal fixes, m68 support, s390x, sh4 compile fixes. There is still portability work todo: #439181 ia64 shared lib buggy (longdated unresolved ia64 mystery), #634890 amrhf all binaries segfault (maybe signal related), status on parisc is unknown, avr32 patch is untested (might as well just merge) and sh4 SIGILL. We got nice arch testcoverage thanks to mksh testsuite and working mksh klibc support. We are confident of this release on all this archs: alpha, armel, m68k, mips, ppc, ppc64, sparc, s390, s390x and x86. (: Sun, 06 May 2012The Gnome shell is a great step. Finally the Linux default Desktop is not only useful (as Gnome 2.X used to be), but also stylish and modern. It's a different kind of desktop, which seems to irk desktop habits of some peoples. I'm proud to show it off. This happened for the last time years ago while doing some funny config quirks of fvwm. Gnome 3.x is cool. (: Also the customisability via all the extensions is fun. What I do miss is the gnome weather (There used to be an incarnation with plenty of geographical sites, which got somehow got axed later.) and more packaged extensions for out of the box deployments (wish #661782). I seem to be too lunatic to find session settings in "System Settings" for having startup applications run or the used applications reopened (please yes evince show the same pdfs). The other minor critic is that "System Settings" is quite empty, but has on each system a funny advertisement for "Wacom Tablets". Ah and after a myriad of suspends and network-manager stops/restarts gnome-shell likes to segfault, which is no nuisance as it restarts. Tue, 23 Aug 2011Mike Waychison from Google sent fancy costum tailored enhancements to kinit. Thanks to new klibc capget and capset support kinit now understands "drop_capabilities=" bootparam. It specifies a list of capability names that will be dropped before switching over to init. dirent.h saw the addition of scandir() and alphasort(). Gentoo devs pushed a klcc enhancement. For details see klibc 1.5.25 release annoucement. A lot of stuff is in work: current klibc patch queue. Wed, 27 Jul 2011Good pile of fixes made it worth to release current klibc git master. Patches came from Gentoo, Google and Openembedded. For details see klibc 1.5.24 release announcement. This release has several ipconfig (klibc dhclient) enhancements, arm porting fixes and usual cleanups: For details see klibc 1.5.22 release announcements or klibc git. Fri, 13 May 2011
initramfs-tools release 0.99 "scarpe rotte e pur bisogna andar"
This new release features /run usage and xz support. For the details see the release announcement of latest initramfs-tools. The upload itself fixes 18 (19) bugs in the Debian BTS and has also a cute lilo support patch hiding under "initramfs-tools: Fix handling of numeric root= arguments to be udev-friendly" coming from Ubuntu. (; Ben Hutchings revamped the bootloader linux-2.6 hooks in order that update-initramfs no longer calls any bootloader by itself. Thank you for all the contributions. The development docs got nicely refreshed too. Wed, 26 Jan 2011This release is "slightly" delayed due to lots of physics calculations for my final PhD year at the TU Vienna with all fixes that piled up since last summer: Support for newer GNU make 3.82, x86_32 signal fun, the self explantory KBUILD_REPRODUCIBLE and various cleanups: For details see klibc 1.5.21 release announcements or klibc git. Mon, 13 Dec 2010I seem to have a sweet spot for having troubles with printers in stressy times. For restarting a stopped specific queue under CUPS it is easiest to specify: cupsenable <queuename> The symmetric command for stopping is cupsdisable. Sat, 28 Aug 2010This release fixes an important ipconfig regression from Lenny due to a badly tested monster patch 4efbcf90f60. ipconfig should now perform better then ever, thanks to the inflow of fixes since Lenny release. This RC fixes are scheduled for Squeeze and it already landed in Sid. 1.5.19 had no release announcement, but fixed compilation on x86_32, the syscall handling on sh4 (initramfs-tools is said to boot fine with it), valgrind ipconfig warnings and added getrusage() for the mksh port. Thanks to hpa for giving me the official co-maintenance of klibc. Thanks to all contributors. P.S.: See klibc git repo. Thu, 22 Apr 2010The overdue dash sync from 0.5.3 took a month to be done, but now klibc is shipping newer dash 0.5.6 then actual unstable dash. Fixes for this release include fstype support for btrfs and ext4 without journal. Moving README's around so that they can be shipped for avid readers: README.ipconfig go :) The goodie from this time is the sh4 build fix form the very active Debian sh4 porters. ipconfig, nfsmount and kinit have now simpler DEBUG build. ipconfig build warnings got shot by a Google patch. If your patch hasn't made it yet, please ping me for next queue: klibc git repo, Unofficial patch queue. P.S.: Ubuntu Lucid pushes out klibc 1.5.17 thanks to Colin Watson. Sat, 20 Mar 2010Not only fixes ipconfig regressions due to fixes in 1.5.16, but ipconfig should no longer discard useful packages. We also fixed a long standing klibc sparc specific socket bug (#444087): sparc lists socket system calls, but does not provide all of them natively. So one is better off on sparc to use sys_socketcall. Thanks to Jan Hauke Rahm the packaging switched to modern Source Format 3.0 (quilt) with debhelper 7 usage reducing cdbs overhead on build. This is a big switch and makes me very happy. New addition include a $(make help) target in Makefile to ease klibc build. A small losetup got added to klibc-utils. i386 and sparc build fine against current linux-libc-dev: klibc-1.5.17 released
New release klibc 1.5.16 uploaded
The upload reduced RC count by one as klibc builds against recent linux-libc-dev including 2.6.34-rc1. Also libklic-dev uses them directly once installed thanks to a patch from Ben Hutchings. The klibc build saw several fixes from a big and refined Google patch queue. The klibc-utils mount grew several useful features and ipconfig saw lots of bug fixes (send requested optional hostname, raise field length for rootpath DHCP option, ..): hpa klibc 1.5.16 release announce, Git klibc repo. Update: According to build log there is still work todo for i386 and sparc. :) Tue, 27 Oct 2009
HTC Magic 90 days Android usage
I love the fact that phones are ready to get a decent OS. I bought the phone for better connectivity and to get GPS goodness. 90 days later seems a good time to review the pros and cons:
Of course Android still seems like an alien sandboxed Linux, but I must say that the average usage capability of the mobile phone is very nicely enhanced. P.S. Using allmost daily also Google Talk, Ultimate Stopwatch & Timer, Finance, GPS Status and the ultra cool Google Sky Map. Mon, 09 Mar 2009
git: Throw away first X commits of a repository
Today got asked by a colleague, who wanted to throw away the early history of his repo. git allows you easily to rewrite the history start with: echo $FIRST_SHA >> .git/info/graft git filter-branch -f Then with interactive git rebase you can rewrite this important initial commit. Of course you have to define $FIRST_SHA and your repo shouldn't be dirty. Wed, 25 Jun 2008
Open Source Graphic Drivers - oops less often
If you haven't read yet: Linux Graphics, a Tale of Three Drivers. The hard data is based on reports thanks to kerneloops.org. It matches quite evenly our bug experience on the Debian Linux images. As already stated "Open Source Graphic Drivers - They don't kill kittens". The unnamed company claims no need for open drivers. Hardware vendors need not to forget that they don't get revenues from the drivers side. The conclusion is to buy hardware, where the vendor cares about Linux support. That means documents the hardware and enhances current free xorg with all needed goodies of enhanced power support and last but not least 3D acceleration support. Fri, 23 May 2008
"Open Source Graphic Drivers - They don't kill kittens"
People don't seem to get that once you load a 1MB binary blob into your kernel you are no longer running a free operating system. There is no way to fix it. No way to see what is going wrong. Fedora 9 is shipping the latest xorg Release. It features major improvements for the Intel and Radeon drivers. The well known randr features got pushed even more. Plus gaining better 2D and 3D effects and important power saving improvements. Thanks to all the xorg hackers! Big blame on a unnamed company that can't keep up with latest Linux (Xen gained paravirt_ops support in 2.6.24 - oh yeah but we are still on 2.6.18 state). To truly support Linux you'd have to first release your specs and then help existing xorg hackers. Even Via seem to have got the message lately after years of no transparence. Fri, 28 Sep 2007Apparently Scott James Remnant in his article on version control systems confuses arch and git. One can only speculate that his short git usage stems from the pre 1.0 days, where you had to use higher level tools (called porcelain) to happily work on git. A funny anecdote is that Scott back in his dpkg hacking year promoted arch heavily. Ubuntu^Wbzr propaganda spreads speed gains as big bonus of the last major releases. In order to be able to do that you have to start with a terrible baseline. Testing bzr on middle sized repos is no fun at all. The bzr pain inside launchpad must be beyond imagination. Nowadays it is much easier to hack on mdadm than on lvm2. The reason is that later project uses rusty cvs. With git it is really easy to contribute back. Either you mail the patches or publish your repository. git will help you along on each way. The other very big bonus of git is the big community around git. It is a community excited around building and delivering the best version control system. The git development does regular surveys on git usages and incorporates back the wishlists. The most "funny" way to use git is to run it as cvsserver. You may believe me or not i have seen git cvsserver emulation usage in the wild. Tue, 28 Aug 2007
YAPC::Europe 2007
The usb modem works out of the box with current Linux 2.6.23-rc2 (why would you use older ;-). usb 2-2: new full speed USB device using uhci_hcd and address 8 usb 2-2: configuration #1 chosen from 1 choice option 2-2:1.0: GSM modem (1-port) converter detected usb 2-2: GSM modem (1-port) converter now attached to ttyUSB0 option 2-2:1.1: GSM modem (1-port) converter detected usb 2-2: GSM modem (1-port) converter now attached to ttyUSB1For lazy guys wvdial drei the wvdial.conf for 3 connectivity (hint - fix pin): [Dialer drei] Modem = /dev/ttyUSB0 SetVolume = 0 Dial Command = ATDT Init1 = ATZ ;Init2 = AT+CPIN=1234 Init3 = ATE1V1&D2&C1S0=0+IFC=2,2 Init4 = AT+IPR=115200 Init5 = ATE1 Phone = *99# Stupid Mode = 1 Init8 = AT+CGDCONT=1,"IP","drei.at","0.0.0.0",0,0 Dial Attempts = 2 Username = xx Password = xx Current mood very happy. :) Aboves means connectivity almost everywhere in Austria due to A1 network fallback. Fri, 20 Apr 2007
Fun with Legend Annotations in Graphs
E\sT\N \f{Symbol}t\0 f
Once you start typing such strings blindly, you know that you have been generating lots of graphs with xmgrace. :)LaTeX equivalent would be E_T \tau fTue, 10 Apr 2007
Happy switch from bzr to git
* Push command - Defaults on the non-obvious old sftp protocol. Worked with rsync on some very ancient version. * Repo format - Changes on every major upgrade. * Patchlevel - Probably for svn compat reason the chosen default patchlevel is -p0 not -p1. Latest initramfs-tools repo is available at: Thanks to tailor it is now possible to move from different version control systems. For git as target repository it is recommended to use latest upstream either directly via darcs or at least install the unstable version. The etch version throws strange python backtraces. Wed, 14 Mar 2007
ext[23] online resizing
nancy:/root# egrep usr /proc/mounts /dev/mapper/nancy_vg1-usr /usr ext3 rw,noatime,nodiratime,data=ordered 0 0Now let's extend a bit that logical volume: nancy:/root# lvextend -L+700M /dev/nancy_vg1/usr Extending logical volume usr to 4.68 GB Logical volume usr successfully resizedNext easy step is to invoke resize2fs nancy:/root# resize2fs /dev/nancy_vg1/usr resize2fs 1.40-WIP (14-Nov-2006) Filesystem at /dev/nancy_vg1/usr is mounted on /usr; on-line resizing required old desc_blocks = 1, new_desc_blocks = 1 Performing an on-line resize of /dev/nancy_vg1/usr to 1227776 (4k) blocks. The filesystem on /dev/nancy_vg1/usr is now 1227776 blocks long.Update: Mika of course already blogged about ext3 online resizing. Bug #400797 is a very good reason not to use ext2resize. I remember a post from Theodore Tso on it's bad code quality. Don't trust it. Mon, 28 Aug 2006
network state monitoring
ip monitor allTue, 18 Jul 2006
Set your email per git repo
git repo-config user.email bla@foo.orgMon, 17 Jul 2006
Git server setup
grep git /etc/inetd.conf # git server git stream tcp nowait root /usr/bin/git-daemon git-daemon --inetd --syslog --export-all --base-path=/var/cache/gitAdd the git service: grep git /etc/services git 9418/tcp # gitNow add your users in /var/cache/git/$user and give them write rights. For the fancy web interface you need those settings: cat /etc/gitweb.conf # path to git projects (Latest gitweb has branches for mod_perl usage for now this is enough: egrep git /etc/apache2/sites-enabled/ssl ScriptAlias /git /usr/lib/cgi-bin/gitweb.cgiSmall sample project list: mattems%2Fklibc%2F.git maximilian+attemsMon, 03 Jul 2006
git cheat sheet
# clone the last revision localy git clone /path/.git # clone repo from remote pc git clone git+ssh://user@host/path/.git # create your own branch git branch <username> # set your email and name which will show up in the commits git repo-config user.email "user@host.org" git repo-config user.name "firstname name" # show latest changes git log git log <file> git show git show <commitid> # check which branch currently is active git branch # create and use a test branch git checkout -b test <hackcodehackcompilehackcompile> # commit either all or some parts related to the files # please don't forget to add an meningfull desc # the first line is the subject, all below the body git commit -a git commit file # merge your test branch once your happy back to your branch git checkout username git pull . test # undo all local modification as they b0rk git checkout -f # push your changes to the repo all can have access to git push git+ssh://user@host/path/.git branch # eventually delete the test branch git branch -D test # tag the current revision git revert <commit-id> # change commit message to last edit # (yes that's the only one you can change others are protected) git commit --amend # delete last commit git reset HEAD^ # cherry-pick a change from one local branch to another local branch. git checkout <first local branch> git log path/to/changed/file git checkout <second local branch> git cherry-pick <sha1 refspec from the previous 'git log' output> # merging others changes git fetch git+ssh//user@host/path/.git mater:incoming git log mater..incoming git checkout mater git pull . bob-incomingWed, 15 Mar 2006
Ethtool - query or set your ethernet
$ sudo ethtool -i $iface driver: tg3 version: 3.47 firmware-version: bus-info: 0000:40:00.0Following command shows your NIC statistics: $ sudo ethtool -S $ifaceIf your nic shows some aging signs, your may want to be sure: $ sudo ethtool -t $iface The test extra info: nvram test (online) 0 link test (online) 0 register test (offline) 0 memory test (offline) 0 loopback test (offline) 0 interrupt test (offline) 0A friend had the crazy wish to have an 10MBs connection: sudo ethtool -s $iface speed 10which he could also have had while loading the 8139too media modul param: insmod 8139too media=0x01Disable TCP/UDP checksums ethtool -K $iface tx offTue, 14 Mar 2006
LUKS crypto partition
cryptsetup luksFormat /dev/sda2 WARNING! ======== This will overwrite data on /dev/sda2 irrevocably. Are you sure? (Type uppercase yes): YES Enter LUKS passphrase: Verify passphrase: Command successful.Create mapping between logical and physical partition cryptsetup luksOpen /dev/sda2 sda2 Enter LUKS passphrase: key slot 0 unlocked. Command successful.Now you create your filesystem on top and voila your using an encrypted partition: mkfs.ext3 /dev/mapper/sda2Tue, 24 Jan 2006
boot from usb disc
mkfs.ext3 /dev/sda1 mount /dev/sda1 /mnt find / -xdev | cpio -pm /mnt echo "(hd1) /dev/sdb" >> /mnt/boot/grub/device.map perl -pi -e s/sda1/sdb1/ /mnt/boot/grub/menu.lst perl -pi -e s/sda1/sdb1/ /mnt/etc/fstab grub-install --install-dir /mnt /dev/sdbWed, 18 Jan 2006
install grub
grub> device (hd0) /dev/sda grub> root (hd0,0) grub> setup (hd0)Wed, 14 Dec 2005
Allow other users access to your INBOX
cyradm connect hostname_server sam user.my_username other_user read lam user.my_usernameThe sam command changes the acl's and lam will list the new rights. Mon, 12 Dec 2005
ipw2200 mode monitor
iwconfig eth1 mode monitor Error for wireless request "Set Mode" (8B06) : SET failed on device eth1 ; Invalid argument.The Patch adding IPW2200_MONITOR wasn't enough as it seems: Kconfig options for IPW2200. Firmware 2.4 is said to immediatly restart on monitor mode. Also noticed that ipw2200 doesn't survive kexec, will look into that. Mon, 28 Nov 2005
simple ccache usage
PATH=/usr/lib/ccache:$PATHWhile compiling you want to know your stats: watch ccache -sIf your home is fucked and has not enough place for a big .ccache, you'll see that the cache will not grow big: Free it up and you'll get all the bonus. Mon, 21 Nov 2005
Request-Tracker per commands steuern
sudo apt-get install rt3.4-clients libterm-readkey-perl export RTUSER=user export RTSERVER=https://hostname/ rt ls "Priority > 5 and Status='new'" rt create -t ticket rt edit ticket/XXX rt reply ticket/XXX rt comment ticket/XXX rt logoutThu, 03 Nov 2005
Backup a partition with partimage.
partimaged -D -g 10 -d /var/lib/partimagedThe client side change the device to the needed partition. It will still prompt you for numerous questions like username and password. Also tell partimage not to split the image. partimage -f0 -z0 -s hostname.org save /dev/sda2 image_filenameFri, 21 Oct 2005
Quick implementation guide of a local Debian repository
dpkg-scanpackages sarge override | gzip > sarge/Packages.gz dpkg-scansources sarge | gzip > sarge/Sources.gz Anyway with the time that gets pretty boring and you may already have tasted the sweeet dupload utility, which places build deb packages at their right place. Archives reindexing and optional removing of old packages happening in the background. First set up an unprivileged user with an accessible public_html (you may want to restrict it to your own net). Disallow any password access. Use public_key from your local network. Afterwards you install mini-dinstall. Read /usr/share/doc/mini-dinstall/README. It explains the following steps: No you can add some sections to your dupload.conf file for further convinience, like the one below (just copied the sarge section, but the rest is not hard to guess):
package config;
$cfg{'iw-sarge'}{'fqdn'}='assi';
$cfg{'iw-sarge'}{'method'}='scpb';
$cfg{'iw-sarge'}{'login'}='iw-deb';
$cfg{'iw-sarge'}{'incoming'}='~/public_html/debian/sarge/';
$cfg{'iw-sarge'}{'mailtx'}='bla@foo.org';
$cfg{'iw-sarge'}{'preupload'}{'deb'}='chmod 0644 %1';
$cfg{'iw-sarge'}{'preupload'}{'changes'}='chmod 0644 %1';
$cfg{'iw-sarge'}{'preupload'}{'file'}='chmod 0644 %1';
1;
Now whenever building a package you can use (as bonus you receive nice mails):
dupload -to iw-sarge buchta_0.0.2_i386.changes mini-dinstall allows to sign your archive files. Add the correct sources.list to your clients and be happy. :) mini-dinstall has some downsides: It uses per default a flat archive structure and has not yet intelligence to really setup an proper archive. So it's a quick solution for a small range of packages. The Debian Bug Tracking system has a patch which should improve that handling, but it didn't get applied over the years. Also the usage of Dnotify instead of the newer Inotify shows it's age. The next step will probably feature dak, which has an Postgresql backend and is not so quick on setup, but is used for the Debian archives. Thu, 06 Oct 2005
Debian chroot
apt-get install sbuild /usr/share/sbuild/buildd.chroot sarge /srv/ai/aipc50/sbuild/sarge/ http://http.at.debian.org/debian /usr/share/sbuild/add_sbuild_user /srv/ai/aipc50/sbuild/sarge/ stable attems /usr/share/sbuild/chrapt /srv/ai/aipc50/sbuild/sarge/ apt-get update /usr/share/sbuild/chrapt /srv/ai/aipc50/sbuild/sarge/ apt-get upgrade /usr/share/sbuild/chrapt /srv/ai/aipc50/sbuild/sarge/ apt-get install sudo fakeroot ssmtprebuild your chroot: mkfs.ext3 /dev/mapper/p_brane-sid mount /dev/mapper/p_brane-sid /usr/share/sbuild/buildd.chroot sid /srv/ai/aipc50/sbuild/sid/ http://http.at.debian.org/debian mount -a /usr/share/sbuild/add_sbuild_user /srv/ai/aipc50/sbuild/sid/ unstable attemsFri, 09 Sep 2005
Java deb
apt-get install java-package TMPDIR=~/tmp fakeroot make-jpkg jre-1_5_0_04-linux-i586.bin sudo dpkg -i sun-j2re1.5_1.5.0+update04_i386.deb |
||||