Hi all, I've recently joined this list.
It's my expectation that "BP6 lockup" problems do not have one single
cause, and that's a big part of why it's so hard to nail it down. In my
own case I seem to have relatively stable operation (no crashes in 5
days) only with this combination of actions:
(1) append="noapic"; and
(2) creating a Linux kernel with power management completely removed,
and turning off PM and ACPI in the BIOS.
This is with Red Hat 6.0, KDE 1.1.2, one IDE drive, Riva TNT video, and
kernel compiles in an endless loop in an xterm window.
Another part of it is that many of us are overclocking; some folks who
do this may be getting failures because of pushing the limits too far,
yet seeing the reports of failures from other causes and saying "yeah I
get that too". Wrong.
Yet another obstacle is the long test cycle times. You run for two
days, then it croaks, you change something, then it runs for five days
and croaks. You figure what you changed made a difference. Wrong! You
haven't tested enough, and the result is just not statistically
significant.
My point is this: if you just want to have a reasonably reliable system
and get on with life, then you do what you can to make the problem
happen less frequently. However any of us who really want to nail down
the hardware/software bugs involved should probably try to make the
problem happen *more* frequently, and try to work with the kernel
developers to gather the post-mortem data that is most useful to them
(whatever that is!).
Just my .02.
-- Rod
Chris Chiappa wrote:
>
> The fact remains that for hundred of other motherboards out there, Linux is
> completely stable - the BP6 is unique in the number of people reporting
> problems. I think pointing fingers at other hardware (ie the Matrox cards)
> is missing the point - maybe there's some bad interaction between the BP6
> and some other hardware, but it still comes down to being a BP6 bug of some
> sort if no other motherboard exhibit the same problem.
--
=- To unsubscribe, email [EMAIL PROTECTED] with the -=
=- body of "unsubscribe linux-abit". -=
