On Fri, May 19, 2006 at 06:14:43 +1000, elliott-brennan wrote:
> Until recently I was running FC4 on the same
> machine. I installed Ubuntu 5.10 (with KDE) approx
> three weeks ago. Over the last three days, the
> machine has started to freeze. Running at the time
I've had a similar problem since the last Ubuntu kernel update.
I was running Hoary and it had been stable for months, then
after installing the new kernel and rebooting the machine began
randomly freezing solid. It doesn't seem to be particularly
dependent upon what's running -- I've had it lock up overnight
when I've been logged out.
At the same time I started seeing messages like these in the logs:
hda: dma_intr: status=0x51 { DriveReady SeekComplete Error }
hda: dma_intr: error=0x84 { DriveStatusError BadCRC }
ide: failed opcode was: unknown
Thinking it was a failing hard drive, I replaced the drive with a
new one, and at the same time installed Breezy. The dma errors
continued to happen and the machine continues to randomly freeze
solid, more frequently than on Hoary. I'd get two or three days
with Hoary, but I'm lucky to get 24 hours running on Breezy.
I'm sure it *could* be two faulty drives, but it's unlikely, and
especially starting at exactly the time the new kernel was
installed. I can't prove it's the kernel, but given the timing,
I do suspect it. If I knew how to downgrade the kernel package
I'd do it to see if the problems go away.
The 686-smp kernel is much more prone to freezing than the non-smp
386 kernel.
I've found that turning off dma seems to fix the problem
(hdparm -d0 /dev/hda), but it also makes the machine noticeably
slower. Still, a reduction in performance beats random freezing.
One interesting point: "ide=nodma" on the kernel command line
is supposed to turn off dma at boot time. I've checked the
source and code to detect that command is there. It doesn't
work though: I don't see the "Prevented DMA" message that it's
supposed to print and I get some dma errors before hdparm is
run to turn it off. Sometimes the machine even freezes early
in the boot process (after "Creating initial device nodes" but
before "Setting disc parameters").
> Are there any logs I can check to see what was
> happening at the time?
Have a look in /var/log/syslog for dma errors.
And if anyone can tell me how to downgrade my kernel package
I'd be grateful. I'd also like to know how to find the source
difference between any two arbitrary package versions.
Cheers,
John
--
"I wish I could find a one-armed economist, so he couldn't say, 'on the
other hand...'"
-- Harry Truman.
--
SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html
