As I noted in a previous post I was recently the victim of a Random Kernel Upgrade Hell (RKUH). Maybe I should trademark that acronym, although its not particularly prounounceable like SNAFU, or PEBCAK. Anyway … the fact was that I was spending several hours trying to fix various problems with wifi drivers, VMware server, truecrypt and the sound in Skype, when it suddenly occured to me that a re-install was probably quicker. The double edged sword of Linux: quick to reinstall, but then again why should you need to do it so often? Well I guess in my case I push the OS pretty hard with some esoteric applications, but even so …
So, a few quick notes. The re-installation process was painless. I have my data on a separate /home partion, so I just backed that up to a separate USB hard disk (using partimage on a Gparted boot CD), and installed Ubuntu over the top of my Mandriva partition.I have to say the Ubuntu installer is a bit simpler than the Mandriva one when it comes to the business of disk partitioning.
It took a few minutes of jiggling to re-arrange the partitions once the install had finished, but then I was good to go. I took out some bits of software I didn't need, installed a few must-have applications so I could get on with my work, and finshed the rest of the day happy. In this case the fact that truecrypt had suddenly broken in Mandriva was the deciding factor, as I needed data on that partition to work: once I had that working on Ubuntu, which I have to say was seamless, then I was good to go.
Overnight I installed some more software and did an update to get the software up to date. I have a very slow connection so this kind of thing takes a few hours. I rebooted and everything was fine. For a few days.
But then last night after having worked all day, I switched my computer on again to play a DVD which was skipping on my standard DVD player. It took an age to boot up and wireless didn't work. Also everything was sluggish, taking sometimes two or three seconds to respond to a keypress. I played the DVD and went to bed, sulking.
I really couldn't understand why it was behaving like this all of a sudden. I hadn't done any kernel updates (RKUH!) in the last few days, which is my normal source of instability, and consequent insanity. I checked dmesg, and tracked it down to this section:
[ 33.957128] Probing IDE interface ide0...
[ 34.915447] hda: FUJITSU MHV2060AH, ATA DISK drive
[ 34.915591] hda: host max PIO4 wanted PIO255(auto-tune) selected PIO4
[ 34.919261] hda: UDMA/100 mode selected
[ 34.922862] ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
[ 34.992563] Probing IDE interface ide1...
[ 35.302438] Clocksource tsc unstable (delta = 136417566 ns)
[ 35.306438] Time: acpi_pm clocksource has been installed.
[ 35.319400] hdc: UJDA750 DVD/CDRW, ATAPI CD/DVD-ROM drive
[ 35.319603] hdc: host max PIO4 wanted PIO255(auto-tune) selected PIO4
[ 35.319706] hdc: UDMA/33 mode selected
After searching for a bit, I thought the "Clocksource tsc unstable" looked likely and it seemed that a few other people had been experiencing long boot times due to it. I added clocksource=hpet to my boot stanza, and bingo, boot is once again fast and wireless works like a dream.
So, although the fix was fast, once again I'm forced to ponder the question "Why did this suddenly happen?" My Thinkpad R51e has always caused me problems with Linux due to its 'special' hardware, but really, if I didn't change anything on the system, why should it suddenly start doing this to me? And if I was a normal guy-in-the-street user, how long do you think it might take me to fix this? Or indeed, to run scuttling back to Windows?