2.4.5-ac1 hard disk corruption... acpi responsible?

2.4.5-ac1 hard disk corruption... acpi responsible?

[Cody Gould]

Debian, Intel 440BX2 Seattle, 2x Fujitsu UDMA33 IDE
hard drives.  In my bootscripts, hdparm enables DMA.

I tried most all 2.4.0-acx up to 2.4.4-ac9, and then
plain 2.4.5 today, over the past few months, with no
corruption or instability.

Today I moved to 2.4.5-ac1, the only different thing
than normal was I enabled ACPI instead of APM.  

After 2 minutes in the fresh kernel on the console, I
was make'ing an application, and the kernel gave some
messages about reverting DMA mode, which I have never,
ever seen before, followed by file I/O errors on lots
of source files.

I shutdown and rebooted, ext2fs detected my main disk
as corrupt, and ran a check, spewing off screens and
screens mentioning bad superblocks, incorrect times on
inodes, and corrupt inodes within bad blocks [of which
I have none].

I tar'd up what little I could salvage from my disk in
single user mode where you run the fsck manually, when
the kernel started giving messages about problems
communicating with USB [I compile in the USB for
Intel, and the HID, and have a Sidewinder joystick
plugged in that I wasn't using].  

The system was unuseable past that point, so I wiped
clean and installed Win2K while I get my backups
together.  My hardware is fine, of course, everything
is AOK now.  The kernel had just gone crazy and mad.

That's my experiences.  Sorry it's long and I can't
provide specific error messages, as I couldn't really
do anything with my system in the state it was in.

I hope that helps someone.  Please CC me comments as I
am not on the mailing list.  Thank you.

Cody Gould


_____________________________________________________________________________
http://messenger.yahoo.com.au - Yahoo! Messenger
- Voice chat, mail alerts, stock quotes and favourite news and lots more!

Re: 2.4.5-ac1 hard disk corruption... acpi responsible?

[Alan Cox]

> Today I moved to 2.4.5-ac1, the only different thing
> than normal was I enabled ACPI instead of APM.  

Bad idea. The kernel ACPI is not the most debugged, the ACPI in many BIOSes
is complete garbage and there isnt a lot you can do to debug them either.

APM is a definite better choice, at least in the shorter term

Re: 2.4.5-ac1 hard disk corruption... acpi responsible?

[green]

In article <E1544AG-00026i-00@the-village.bc.nu> you wrote:
> Bad idea. The kernel ACPI is not the most debugged, the ACPI in many BIOSes
> is complete garbage and there isnt a lot you can do to debug them either.
> APM is a definite better choice, at least in the shorter term
Too bad, we cannot get much of APM in SMP, only power-off if one gets lucky.

Bye,
    Oleg