After getting really annoyed with all the directories piling up in my /usr/local/xen directory, Benv offered to try and find out why the hell these directories didn’t want to get deleted.
It seemed that the directories had a flag ’schg’ which you can see by doing an ‘ls -lo’ on the directory. The ’schg’ flag means that the directory is system immutable and only the owner and superuser are allowed to remove this flag.
This is done by calling chflags 0 on the directory. This clears all it’s flags.

I fixed my problem by doing:

But when it tried to run configure, there was nothing to configure so I figured it was smart to quit the slackbuild script and check it out myself.
It seems that there was indeed no configure file, but the INSTALL file told me I had to run “make linux install” to build and install the library for my platform. Well that’s nice, but where will it install everything? Benv hinted me that I should probably edit the Makefile and change INSTALL_TOP to INSTALL_TOP= /usr/src/lua-5.1.4/pkg/usr so I could run makepkg after to create my own package. But after creating and installing the lua package, edje still complained that there was no liblua installed on my machine.
Checking my config.log I found out it was searching for liblua in the pkgconfig files. Oddly enough, the lua src did have a etc/lua.pc file, but it does not install it. So I placed it in /usr/src/lua-5.1.4/pkg/usr/share/pkgconfig/ directory and I changed the paths so it would match my installation and I recreated and reinstalled the package.
This time edje did find the library, but crashed with the following error:

After googling a little and checking the error again, I found out that the Makefile needed to be patched so it would also use the -fPIC option. So I edited the /usr/src/lua-5.1.4/src/Makefile and changed

to

, ran the make linux install, created a new package, installed the package and it finally worked \o/

If you wonder what picard is, you can check it out here: PicardTagger

To install Picard, you’ll need the following software

• fftw3
• libofa (0.9.2 or later)
• Libmusicbrainz 2
• mutagen
• ffmpeg
• Picard

Building fftw3

When you build fftw3, make sure you run configure with –enable-shared since it’s default set to ‘no’.

Building libofa

Building libofa is not as trivial as it seems. Since it’s quite old, it’ll need some patches. What I did, was downloading the libofa src from ubuntu and use the ubuntu patches to make it compile.

Building libmusicbrainz

For libmusicbrainz you’ll have to patch some files as well. A nice overview of files that need to be patched can be found here

You’ll also have to install the python library which you can find in the package it self in the python directory. Just run ./setup.py config ./setup.py build and ./setup.py install.

Building mutagen

I downloaded the latest version from mutagen from the google code project site

Just unpack it and run ./setup.py install to install the package

Building ffmpeg

When you build ffmpeg, make sure you run configure with –enable-shared since it’s default set to ‘no’.

Installing Picard

Now you’ve finally installed all the dependencies, you can install Picard. In my case I’m installing Picard 0.11 which expect an older ffmpeg install so it needs to be patched as well to be able to find the avcodec.h and avformat.h files. Edit the setup.py file and search for  cfg.set(’avcodec’, ‘cflags’, cfg.get(’avcodec’, ‘cflags’) + ‘ -DUSE_OLD_FFMPEG_LOCATIONS’). You’ll have to replace it with fg.set(’avcodec’, ‘cflags’, cfg.get(’avcodec’, ‘cflags’)) and all should be working.

So BenV had told me I had to install Slackware 64 on my computer since there’s nothing that should hold me back from it now they got a decent release and I don’t play any games on my Linux anyway.

Of course I should have known from experience that doing something ‘trivial’ as installing an new OS on my computer on BenV’s request, would probably turn out to be a disaster.

I started with figuring out where to put this new installations. I got 4 disks in my computer from which 2 I’m not sure if they’re working properly or not. The other ‘good’ working 2 are a 8 year old WD PATA 120GB disk and an quite new WD SATA 320GB disk. Since the PATA disk is old and slow, I decided to install it on my new disk. Of course this disk already contained 4 primary partitions from which one partition had a FreeBSD installation on it. It looked something like this:

|________|___|________________|________|____|________|

NTFS       free             NTFS                 FFS      EXT3     free

Fortunately our server hosts some quite handy PXE bootable operating systems, so I started with booting my machine from network using the PartedMagic distribution. The EXT3 partition on the disk wasn’t used, so I removed that one first to give me a free slot for a new primary partition. Because the second NTFS partition was only half full, I decided to shrink it a little and move it to the left to give me more space for my new Slackware installation. So after those changes my disk looked like this:

|________|___________|________|________|_____________|

NTFS              NTFS            free          FFS                 free

Now I wanted to move the FFS partition to the left as well, so I had 1 big block of free space. But the partition tool refused to do this because it had no idea what kind of partition it was….. There were 2 options left; just use the biggest part of free space and leave a useless block in the middle, or remove the FFS partition which contained a FreeBSD 7.2 installation which had taken me almost 2 full days to install after BenV told me “FreeBSD 7.2 has been released, you should upgrade to it”, which is a totally different story. I ended up just removing this partition since I’d only booted twice since that installation anyway. I created 1 big extended partition from the free space and added a 80GB logical partition for my Slackware installation which I formatted with EXT4. All together it took about an hour to get all the partitions resized/ moved/ created.

I booted my machine up from network again and started the Slackware 64 installation from our server. Everything went quite well from here. I selected about all available packages for installation (of course I deselected the Emacs package) and I told the setup to use NFS as the package resource so it would install the packages hosted by our own server saving a lot of downloading time.

Because I had used EXT4 for the whole partition, I needed Grub2 to boot it up. So first I booted my old Slackware 32 installation (which was installed on the old 120GB PATA disk), downloaded Grub2, installed it on the new disk, and tried to make a configuration file. Of course this configuration file was useless to boot from, but at least Grub2 was installed on the new disk now. So I changed the disk boot order in my BIOS so it would boot from the new disk, just to find out that installing Grub2 on the new disk wasn’t enough for Grub2 to start up. Of course it needed the files from my Slack32 partition to load up it’s stage 2, so it wouldn’t work.

Since this was all BenV’s idea, I looked at him questioningly for what to do next. Somehow I had to build and install Grub2 on the new partition for it to be able to load completely, but how could I accomplish this without being able to boot from this new partition? My only option seemed to be to boot from network again and use the Slackware64 installer to mount my new partition and chroot to it. This was indeed a good idea. Only about 15 minutes later I was able to boot up the new partition using Grub2.

I’ve been struggling with Grub2 a little more to see if I could make it boot my old Slack32 installation as well, but it required an initrd file since it had a root partition on LVM and somehow it didn’t seem to want to work so eventually I decided I didn’t need to boot from it anymore anyway since I could just mount it in my new Slacware 64 installation and chroot to it if I needed to run anything from it.

The next task was to build E17 so I could load up my X. Because I’m using the Easy_e17.sh script nowadays, this was quite trivial. I just had to install libmpd so mpdule would build as well, but that was about the only thing. I think it took less then 30 mins for E17 with all the modules and extras to build and install. (God I love my new AMD Phenom(tm) II X4 810 Processor). I crossed my fingers and and entered telinit 4 to see if it would work. And there it was, my old trusty entrance login screen, working without any issues (even my mouse and keyboard worked at once this time). I logged in and was happy to find a fully working E17 installation.

So now I figured I should set up my slackpkg for easy updating and upgrading packages. Since we have multiple computers running Slackware around the house, BenV has set up our server as a mirror for slackpkg so we only have to download the latest packages once. I configured my slackpkg to use the server as a mirror and ran the script to see if it all worked. It even found about 4 packages that needed to be updated (this was a surprise since I had just used the same packages for installing my Slackware 64 in the first place).

After that I started downloading VirutalBox so I could use my old virtual Windows installation. But when I tried to run the install, it told me it couldn’t find /bin/sh…. I wondered what it’s problem was and tried to run the install with bash. Now it was telling me it couldn’t find /bin/bash either. Since I was pretty sure both /bin/sh and /bin/bash existed, I wondered what could be the problem. But even strace told me the same. So now I just tried to run bash by it self and it told me the same as well. I closed my Xterm and tried to open a new one, it told me it couldn’t find /bin/bash either. I had no idea what happened in between downloading VirtualBox and trying to install it. Did some permissions change? Did my disk crash somehow? Did my /etc folder get deleted? Either way, I couldn’t even log in on a different TTY either, and a reboot gave me the same problem that it couldn’t find my shell.

So once again, I booted from network and started the Slackware 64 installation distro from there. I mounted my partition and tried to chroot to it, but it gave me the same problem that it couldn’t find my shell. Since /bin/sh was linked to /bin/bash, bash seemed to be the problem. BenV asked me what exactly were the new packaged that I got while running slackpkg, and I remember I got a new bash…. I checked /var/log/packages and it seemed that I indeed had gotten bash 4. Then it struck me, could I really be that stupid? Seems like I was… I had used the Slackware 32 mirror instead of the 64 one for slackpkg. That was the reason I’d gotten new packages while I should have already had the latest ones installed. I must have been lucky that it were only 4 packages, but the fact that one of them was bash had gotten me in this trouble. BenV came with the brilliant idea of linking my /bin/sh with busybox so I was able to chroot to my Slackware 64 partition and scp the right bash package from our server and install it. This fixed the issues and I was able to boot my new installation again.