On Wed, Aug 07, 2002 at 08:11:53AM +0100, Joss Winn wrote:
> Hello,
>
> I'm using Debian 3.0 with the stock 2.4.15 deb kernel on a Dell
> Latitude XPi 133. It's a great little console-only machine but for
> one problem. On almost every occasion, 'apm -s' produces a kernel
> panic upon wake up and I have to hard reboot. I get lots of
> messages on the console a couple of which are in English:
>
> <1>Unable to handle kernel pagin request at virtual address dbdf5fdd
> printing eip
>
> and...
>
> <0>Kernel panic: Aiee, killing interrupt handler!
> In interrupt handler - not syncing
>
>
> does this make any sense to anyone? I have attached my dmesg and
> lsmod output in case it might shed light on my set up.
>
> Not being able to put a laptop to sleep is a real hassle. I'd
> appreciate any help.
>
> thanks
>
> Joss
> --
> Linux version 2.4.18-586tsc ([EMAIL PROTECTED]) (gcc version 2.95.4 20011002
> (Debian prerelease)) #1 Sun Apr 14 10:57:57 EST 2002
. . .
> No local APIC present or hardware disabled
> Kernel command line: BOOT_IMAGE=Linux ro root=302
> Initializing CPU#0
> Detected 133.639 MHz processor.
> Console: colour VGA+ 80x25
> Calibrating delay loop... 266.24 BogoMIPS
> Memory: 67984k/73728k available (810k kernel code, 5360k reserved, 229k data,
> 212k init, 0k highmem)
Hmm, not exactly the heftiest machine. Linux won't mind, and it
suggests that it may be old enough, probably you do not have the option
to back off to ACPI.
(on more modern hardware ACPI often works better, or works as well as
can be expected, APM not having been tested by the hardware folks worth
a #*! ... sigh.)
. . .
> apm: BIOS version 1.1 Flags 0x03 (Driver version 1.16)
well, it spotted this, so you probably don't have to force
append="apm=on"
Unfortunately that's bad news; it means it'll be giving it a shot, but
trying obviously gives your poor beastie indigestion.
The next thing I'd suggest is chekcing your CMOS/BIOS entries and see if
any of them affect how APM behaves. e.g. if APM is turned off there, or
something wacky like that.
> Module Size Used by Not tainted
> nfs 69628 1 (autoclean)
> ipt_TOS 1024 0 (autoclean)
> ipt_REJECT 2816 0 (autoclean)
> ipt_state 608 0 (autoclean)
> iptable_mangle 2144 0 (autoclean)
> ip_nat_irc 2336 0 (unused)
> ip_nat_ftp 2944 0 (unused)
> iptable_nat 12980 2 [ip_nat_irc ip_nat_ftp]
> ip_conntrack_irc 2496 0 (unused)
> ip_conntrack_ftp 3232 0 (unused)
> ip_conntrack 12940 4 [ipt_state ip_nat_irc ip_nat_ftp
> iptable_nat ip_conntrack_irc ip_conntrack_ftp]
> iptable_filter 1728 1 (autoclean)
> ip_tables 10528 8 [ipt_TOS ipt_REJECT ipt_state
> iptable_mangle iptable_nat iptable_filter]
> nfsd 65024 8 (autoclean)
> lockd 46816 1 (autoclean) [nfs nfsd]
> sunrpc 58484 1 (autoclean) [nfs nfsd lockd]
> lp 6368 1 (autoclean)
> apm 8956 1 (autoclean)
I've seen it occasionally that modular APM works poorly for people; you
might try building you own kernel with the feature as a built-in.
You might also try the 2.2.20 kernel. Maybe the ipchains isn't as
spiffy, but the defenses are alright, and there are lots of apps to help
you concoct ipchains.
> xirc2ps_cs 11492 1
> serial_cs 4448 0 (unused)
> ds 6464 2 [xirc2ps_cs serial_cs]
> i82365 12608 2
> pcmcia_core 39136 0 [xirc2ps_cs serial_cs ds i82365]
> isa-pnp 28168 0 [i82365]
> parport_pc 22056 2 (autoclean)
Your call whether to build this in, but parport and parport_pc go
together. It's not like your parallel port *itself* is going to detach
and walk away. OTOH you're pretty tight on memory.
> pcd 11072 0 (unused)
> cdrom 27488 0 [pcd]
> bpck 8800 1
> paride 3488 1 [pcd bpck]
> parport 22976 2 [lp parport_pc paride]
> rtc 5528 0 (autoclean)
> unix 13316 8 (autoclean)
> ide-disk 6624 2 (autoclean)
> ide-probe-mod 8064 0 (autoclean)
> ide-mod 130860 2 (autoclean) [ide-disk ide-probe-mod]
shouldn't IDE be built-in? Oh that's right. This is an initrd kernel.
> ext2 30592 0 (autoclean) (unused)
> ext3 56736 1 (autoclean)
> jbd 35048 1 (autoclean) [ext3]
Notice how chubby ext3 is, and jbd using it.
Note that ext3 support on 2.2.x is going to need that support builtin.
You may very well have to build your own to do that so you can choose
your lesser processor.
don't be afraid to
apt-get -b source [packagename]
instead of just rush to install a kernel image. Luckily, installing a
new kernel image does not remove an old one.
* Heather Stern * star@ many places...
