Thanks for the reply. I have made progress. I don't know why my initial google searches did not turn up a solution, because when I tried again I found something that seems to work. Unfortunately, I lost the site when I put in a slightly different query. Here is what I learned:
Intel D945GCLF2, with Atom 330 processor
The error [from /var/log/syslog]:
kernel: [ 5817.708236] ACPI Exception: AE_AML_INFINITE_LOOP, while evaluating GPE method [_L00] (20101013/evgpe-503)
The solution?
Quote:
$cat /sys/devices/system/clocksource/clocksource0/available_clocksource
hpet acpi_pm
|
append a clock= to the kernel parameters line in lilo.conf
as root, edit /etc/lilo.conf
Quote:
# Start LILO global section
# Append any additional kernel parameters:
append="video=intelfb vt.default_utf8=1 raid=noautodetect clock=acpi_pm"
|
run the command: lilo
reboot
See /usr/src/linux/Documentation/kernel-parameters.txt [clock (deprecated), clocksource]
I should have added "clocksource=hpet" instead of "clock=acpi_pm" because syslog now reports:
Override clocksource acpi_pm is not HRT compatible. Cannot switch while in HRT/NOHZ mode
and dmesg says:
[ 0.490224] HPET: 3 timers in total, 0 timers will be used for per-cpu timer
[ 0.490224] hpet0: at MMIO 0xfed00000, IRQs 2, 8, 0
[ 0.490224] hpet0: 3 comparators, 64-bit 14.318180 MHz counter
[ 0.496034] Switching to clocksource hpet
However, syslog is not adding 37,000+ lines and the system seems to be working much better.
I will change the append line in lilo.conf with clocksource=hpet, reboot, and see what happens.
If that does not work I'll post again, otherwise, problem appears solved.
Added: changed "clock=acpi_pm" to "clocksource=hpet" which seems to work fine (could not resist updating the post).
=== Well, if did not quite work, but I did some more. The problem would seem solved, but then reappear. Since it involves ACPI, I thought perhaps something was looking for things that were not there or otherwise getting wrong responses from the motherboard bios.
I checked the kernel situation and discovered that I did not know what I was running. I thought it was the generic smp kernel, but it wasn't. It was the huge-smp kernel
Quote:
$ uname -r
2.6.37.6-smp
$ ls -l /boot/vmlinuz*
vmlinuz -> vmlinuz-huge-smp-2.6.37.6-smp
vmlinuz-generic-2.6.37.6
vmlinuz-generic-smp-2.6.37.6-smp
vmlinuz-huge-2.6.37.6
vmlinuz-huge-smp-2.6.37.6-smp
|
So I added a symlink to the vmlinuz-generic-smp-2.6.37.6-smp (as root). I recall an error in the past when my image name was too long, so I make symlinks. I could have changed the existing link, but I just added another one.
Quote:
$ cd /boot
$ ln -s /boot/vmlinuz-generic-smp-2.6.37.6-smp vmlinuz-generic-smp
|
I thought I needed to get the /boot/initrd.gz situation straightened out, because my /etc/lilo.conf did not mention an initrd.
I am using /usr/share/mkinitrd/mkinitrd_command_generator.sh to create a /etc/mkinitrd.conf file -- use -c option to generate output for a config file and -k [kernel]
Quote:
# for Slackware
sh /usr/share/mkinitrd/mkinitrd_command_generator.sh -c -k 2.6.37.6-smp
SOURCE_TREE="/boot/initrd-tree"
CLEAR_TREE="1"
OUTPUT_IMAGE="/boot/initrd.gz"
KERNEL_VERSION="2.6.37.6-smp"
KEYMAP="us"
MODULE_LIST="jbd2:mbcache:ext4"
LUKSDEV=""
ROOTDEV="/dev/sda2"
ROOTFS="ext4"
RESUMEDEV=""
RAID=""
LVM=""
UDEV="1"
WAIT="1"
|
Copy the output to /etc/mkinitrd.conf, perhaps changing the name of the OUTPUT_IMAGE="/boot/initrd.gz" to OUTPUT_IMAGE="/boot/initrd-slack.gz" so the existing /boot/initrd is not overwritten.
Then, also as root (the -F option tells mkinird to use the config file:
Quote:
$ mkinitrd -F
$ ls /boot/initrd*
/boot/initrd-slack.gz
|
Then I edited /etc/lilo.conf to change the "stanza" for booting slackware by changing the image, adding the initrd line, and change label to Slackware.
Quote:
# End LILO global section
# Linux bootable partition config begins
image = /boot/vmlinuz-generic-smp
root = /dev/sda2
initrd = /boot/initrd-slack.gz
label = Slackware
read-only
|
Run lilo
OK, now after rebooting it seems the problem is gone and it has not come back. I guess this means I missed something in the installation, since I would not want to run the huge-smp kernel ordinarily.
I also compiled a custom kernel for this motherboard, since I have a few, but it doesn't make much difference in performance.
Thanks