More Kernel Bugs than Ports Bugs

Yesterday was a celebration day for the FreeBSD ports people. For the first time since two years, there were more kern/ PRs than ports/ PRs.
Click on the image below to get more details from Brian S. Dean’s PR stats
prstats

XFree86 & VIA KM 400 continued

Yesterday night I retried to get X running on the Acer Aspire 1350.
It looks like the via driver is still under heavy development. Several people have or had problems with it, including Matt Dillon, Alan Cox. Since last week the Unresolved drm Symbols problem has been fixed, but the Server still crashes.
Other XFree86 bugs are reated to the VIA driver:

  • 812 VIA Modeselection problems.
  • 813 Via driver on km400 laptop results in blank lcd

I decided to wait until some of these issues are resolved and the submitted patches found their way into Xfree86 CVS, because with all the posted patches this is a patch merging hell.
In one of the bugs Egbert Eich pointed to a wiki article that explains how to debug the XServer. I will continue reading it when I will enter the next round fighting with the via driver

Recovering Superblocks.

The weekend started with a catastrophe. After the last crash of my workstation
the disklabel was messed up and it didn’t wanted to boot.
I tried to reinstall FreeBSD on the / partition, but that did’t help anything, fsck(8) threw out thousands of error messages and didn’t reconstruct the disklabel.
Being clueless what to do, I first started with my other Computer maintenance.
I extended huckfinn’s too small /var partition from 20MB to 500MB.
I exchanged the SCSI Controller for my CD Burner from the Dawicontrol UW Controller with a broken BIOS to a used Adaptec Controller that was built into Olga. Suprisingly the CD Burner was detected and I didn’t had to deal with Jumpers, Termination and SCSI IDs.
Now I plugged the Dawicontrol Controller into the last-but-one PCI Slot of Sauna and connected it with a spare 6GB UW Seagate HD. Again i jumpered everything right and I installed FreeBSD over an old Windows partition that was on the drive. After reading the fsck_ffs(8) manpage I tried fsck with the -b 32 option and that went well. So only the primary superblock was damaged. Now I just had to find a way to recover the original one. That was difficult, since most postings on the FreeBSD mailinglist said that fsck will do that automagically. But my fsck didn’t!
Since it was already late at night and I went to bed. After sleeping long and an extensive breakfast I tried some different search queries on Google.
And finally I found this thread.

dd if=<disk> skip=32 of=<disk> seek=16 bs=512 count=16

saved my life, everything is mountable again and most data was recovered.
Since I now have two disks in my workstation, I am going to share the diskload between the two drives and perhaps I should think about a better backup solution 🙂

My gf got a new one!

Yesterday after I bought food for the weekend with my gf, I went to the local Computershop. While looking at the Laptops my gf fell in Love with an Acer Aspire.
I have to admit I like riccardo too. 2 GHz are really darn fast. Installing FreeBSD on it was a lot of fun…
As usual the modem didn’t work and additional X seems to require a CVS version. Someone please fix this bug, K THX :).

Mastering fdisk

I just fixed my filesystem problems on polly. The partition table was corrupted,
ad0s1 was moved to ad0s4 and three zerosized partitions were created before.
I moved my partion back to the start and now everything is fine again.
Now I just have to fix my broken cups on sauna to print various university stuff

De-bugging FreeBSD II

Today I tried to reproduce the panics I ran into yesterday with a recent CURRENT.
I could reproduce my UDP locking issue from kern/48560 and the tcp locking issue from yesterday. Truckman confirmed that the NFS via TCP problem is a known problem and pointed to an alpha patch from Jeffrey Hsu from April.
I also ran into another more serious panic while I umount’ed an NFS exported Filesystem. See my mail to freebsd-current for details. phk categorized it as an evil bug, but did not really felt responsible.
At the moment FreeBSD 5.x sux, and I don’t see this changing in the near future….:-(

De-bugging FreeBSD

Today I continued my Ripping project. I managed to panic my System twice,
while I was trying to tune the NFS performance over the wireless link.
See my posting to freebsd-current for details.
Now is a good time to update the world on polly…