Re: tulip.c modification changes got rid of kernel unaligned access
On Thu, 23 Oct 1997, Lars Wirzenius wrote:
> I compiled a fresh 2.0.31 kernel yesterday. I changed a
> "static const" instead of a define, but from 200 to 16384.
> The machine was up for 13 hours (while I was home sleeping)
> without giving single unaligned access complaint (as opposed
> to 20 in a minute, as before). So far so good.
Glad to hear it :) It's a start and should definitely improve
performance.
> Yesterday, I was checking the swap partition, using
> badblocks, and that crashes the system. Under Red Hat and
> Debian, using 2.0.27, 2.0.30 and my newly compiled 2.0.31
> (under Debian only). The error message on the console is
> (under 2.0.31):
> Device busy for revalidation (usage=3)
Just curious, did you run "swapoff" before trying to check it? If not,
that may be why it freaked out on you. Playing with swap space while the
"filesystem" is hot is akin to randomly playing with things in RAM (or at
least, it should be like that).
> CIA machine check NOT expected
> (that last error message is amusing, but it might eventually
> cause confusion similar to the classic "Intruder alert" and
> "you don't exist, go away"; one of my co-workers thought I'd
> played a joke and didn't believe it was a real error
> message).
Hehehehe...yeah, this was being discussed on the RedHat AXP list the other
day. I found it kinda funny myself.
> (BTW, sorry for flooding the list with my problems. I try to
> report only relevant stuff. I'm rather unqualified to
> operate an Alpha, since I don't know anything about it.)
Don't worry about it. Solving problems and putting out fires are what we
do best around here (incidentally, I am actually a trained firefighter).
Your posts have been very good and will probably help others in the future
:) Plus, you got me thinking about how to permanently fix that stupid
tulip driver once and for all :)
Chris
--
TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
debian-alpha-request@lists.debian.org . Trouble?
e-mail to templin@bucknell.edu .
Reply to: