Re: Frequent "unaligned access", an init that dies, ... (fwd)
On Wed, 22 Oct 1997, Lars Wirzenius wrote:
> I get frequent "unaligned access" messages:
>
> Oct 21 19:06:39 zeus kernel: kernel: unaligned trap at fffffc0000363c14: fffffc000443d036 28 2
> Oct 21 19:06:39 zeus kernel: kernel: unaligned trap at fffffc0000363cd0: fffffc000443d036 28 1
> Oct 21 19:06:39 zeus kernel: kernel: unaligned trap at fffffc0000363d70: fffffc000443d046 28 2
> Oct 21 19:06:39 zeus kernel: kernel: unaligned trap at fffffc0000363d78: fffffc000443d046 28 3
> Oct 21 19:06:54 zeus kernel: kernel: unaligned trap at fffffc0000363c14: fffffc0004439036 28 2
> Oct 21 19:06:54 zeus kernel: kernel: unaligned trap at fffffc0000363cd0: fffffc0004439036 28 1
> Oct 21 19:06:54 zeus kernel: kernel: unaligned trap at fffffc0000363d70: fffffc0004439046 28 2
> Oct 21 19:06:54 zeus kernel: kernel: unaligned trap at fffffc0000363d78: fffffc0004439046 28 3
>
> Should I be worried about this?
By any chance, do you have IPX support compiled into your kernel? I
haven't looked up the address listed above yet, but I did notice in the
past that IPX tosses out quite a few unaligned access warnings on active
IPX networks.
As far as being worried, I doubt that you need to be *TOO* worried about
the unaligned traps, but they are an indication that something wasn't
coded properly. From past experience trying to get rid of such problems,
they usually occur when pointers that are not aligned on proper boundaries
are accessed. If you can, let me know exactly what packages are causing
these warnings and/or what commands are issued before you get these. I do
know about man-db's unaligned access problems, not to mention netstat's
(which should be fixed shortly if it hasn't been done already). Since I
only use development packages on my Alpha, I don't often get to see and
test all of the others very well...
> I've had the system die on me a few times (resulting in
> broken filesystems and such), with error messages indicating
> "unable to handle kernel paging request" (or words to that
> effect), "idle task may not sleep", and something that
> indicated that init was behaving badly or crashing.
Eek. Now this is more serious. What kind of system do you have and how
much RAM? Also, have verified that your swap partition didn't get hosed
somehow? (that actually happened to me and caused all kinds of strange and
unexpected problems).
> I use a self-compiled kernel, which seems to work pretty
> well with the Red Hat installation. It's version 2.0.30; I
> guess I should try 2.0.31 as well, but that will have to
> wait a day or two.
FYI, right now I'm flipping between two kernels - a patched 2.0.30
(patched for RTC support and to correct for some Alpha-isms), and a
pristine 2.1.59 kernel, which I'm running right now. Neither has caused
any problems on my UDB. I haven't tested 2.0.31, but probably will by
next week. Forward me a copy of your .config and I'll see if I notice
anything really obvious that might be causing your problems.
If it's any consolation, my UDB has been up for two weeks on my 2.1.59
kernel without one unaligned access....in short, it *IS* possible :P
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: