Kernel panic - not syncing: IO-APIC + timer doesn't work!

I'm having the same issue..

I using noapic a valid workaround here? Or will this make some other features stop working? ( Just so i know how to avoid it for the time being :)

Thanks!

You need to confirm that this actually fixes your problem. The problem I see described in the above links are related to crashdump (kexec/kdump interaction), which seems nothing like what you are describing. Also, the fix is in x86_64, which wont affect the -686 kernel image, nor would it fix a CoreDuo system.

Did you try booting with "noapic"?

I'm seeing this exact problem as well, on the same hardware. The noapic boot option does seem to solve the problem for me, but unfortunately it also makes the mouse pointer very sluggish (I imagine it doesn't get polled often enough by the kernel, or something). The contrast with os X (or, for that matter, linux on the few occasions I do manage to boot it withot noapic) is striking.

I tried attaching a usb mouse, but it has the same problem as the trackpad.

Sorry about posting the irrelevant information.

I booted with the "noapic" and it fixed the kernel panic but, like Karl stated, made the mouse very sluggish. Almost unusable. I tried using the "apic=debug" boot argument but that did not seem to do anything. I can also confirm that the kernel panic still occurs with linux-image-2.6.15-26-686.

Ben, do you have any other suggestions for debug information?

With the -26 kernel image, I've only seen the panic once -- the very first time I booted it. Every time since then (over a dozen boots by now) it boots OK even without noapic.

Every time I try to boot (even with -26), I get the error message at least 5 reboots before it finally boots normally. It seems that if I've been running a while and then reboot, I never get the message. Like the message only occurs if my MacBook is not warmed up.

I also suffer from this bug. It persists into 2.6.15-26.

Not only does 'noapic' make the mouse sluggish, it misses keys if I type at full speed. This is unusably bad.

This might be related to Bug #54621.

OK, I just got my second panic with the -26 kernel. When I rebooted it worked again, but obviously the problem is still there; I'm just a little more lucky than the rest of you. :-)

Me too! Just got a new white macbook and was hit with this problem. the -23 version doesn't seem to have it. It seems to sometimes boot and sometimes not. Disabling noapic has the same sluggish mouse efect. It would be great to solve this.

Unfortunately, I can confirm that this same problem exists in Edgy with the 2.6.17 kernel. Can anyone give any more debug output that has not already been posted yet? Just confirming wont be enough to track the bug down.

It seems to be some kind of race condition in the ACPI startup since sometimes the boot works. Any idea on what would be needed in terms of debug output to track this down?

Confirm this on fully updated Dapper with new MacBook.

"Hi,

This seems to be te cause of the “APIC kernel panics”:

http://llistes.bulma.net/pipermail/bulmailing/Week-of-Mon-20060703/077657.html

(it’s in spanish)

According to the above link, recompiling the kernel to use a timing frequency of 1000 Hz instead of the 250 Hz solves the “panic” problem.

You can install the “linux-source-2.6.15″ package and follow the instructions here to build the kernel:

http://ubuntuforums.org/archive/index.php/t-24853.html

(Note that you have to add a “–initrd” switch in the “make-kpkg” command to make it generate an “initd.img” and make the kernel boot properly)

The problem is that the restricted modules won`t be installed with this new kernel, and this includes the “madwifi” driver for the wifi…

So you have to add it.

Greetings,
Marco. "

This is a comment on the site: http://bin-false.org/?p=17

If you search the page for "APIC," many many people are having this problem? Is there any way we can get the status of this bug updated?

Also see bug #54621 for linux-source-2.6.17 (Ubuntu)
(which is in turn marked as duplicate of bug #53985)

Today I stumbled upon a possible quickfix (didn't try it myself yet, I'm happy enough I got lilo booting again after a kernel update, but that's a different story). It's found at:
www.xs4all.nl/~hajk/mbook-install.pdf

It adds a parameter to the lilo.conf "Append" value:
append="quiet splash lpj=8000000"

Excerpt:
"The boot option lpj=8000000 (eight million) fixes the kernel panics frequently occurring when booting the current Ubuntu kernels (including those in Edgy), see bug #54621 in Launchpad/Malone. This option is quite harmless and sets BogoMips (an approximate speed measure) for the 1.83GHz Core Duo T2400 processor in the MacBook to a reasonable value (about twice the average BogoMips per core followed by three zeros), rather than rely on a kernel estimate during booting (see the output from dmesg) that may be wildly off. Users with a 2.0GHz processor should use a value that’s 10-20% higher, the exact value isn’t that important."

In bug #53985 Ben Collins says:
(kernel) 2.6.17-8.22 should fix this.

Well I've dealt with this bug for 3 months I guess I can wait a few more weeks till Edgy. Lets just hope its really fixed.

Mr. Collins, can we get a little more info than "2.6.17-8.22 should fix this?"

This does not appear to be fixed with linux-image-2.6.17-10-generic (2.6.17-10.26).

I had the same problem with a macbook on an up-to-date edgy system, but I haven't had any kernel panic since I appended "noapic acpi=force irqpoll" to the kernel cmdline.

Edgy Beta (Kernel 2.6.17-10-generic) still suffers from this bug!

"noapic" isn't a good solution since it causes other problems. Does the addition of "acpi=force irqpoll" make a difference in that regard?

Still not fixed with Edgy Beta (Kernel 2.6.17-10-generic)!

Edgy Beta (Kernel 2.6.17-10-generic)

"acpi=force irqpoll" does not seem to make a difference. I removed "noapic" and rebooted, and I needed to retry five times before it wanted to boot...

But I don't have the reported problems with "noapic" : my mouse speed is normal, and it doesn't miss any key.

Problem STILL not fixed. I'm quite sure this is a frequency problem as described above by Rob van den Bogaard.

No. This problem is really the problem in bug 54621. Everyone please look there for the real cause of the problem and a better workaround.

appending lpj=10000000 at boot fixes this 100% of the time for me.

10mil for 2Ghz
8mil for the 1.8_hz (apparently)

Ryan Lortie said:
"No. This problem is really the problem in bug 54621. Everyone please look there for the real cause of the problem and a better workaround."

Didn't I cite Rob van den Boraard who cited the same bug that you just did? And doesn't bug 54621 deal with a frequency-related problem? Perhaps I'm missing something here.

Yup. Missed something.

Actually, 8mil is the correct value for my 2.0GHz. The author of that PDF must have misinterpreted my bug report. In reality, the number for the 1.83 should be lower.

There are tollerances so it doesn't really have to be exact. The further off it is, though, the more likely it is for odd bugs to creep in from other places.

from Ryan Lortie at 2006-10-14 21:58:07 CEST:

"No. This problem is really the problem in bug 54621. Everyone please look there for the real cause of the problem and a better workaround."

I've made both this bug and its other duplicate (Bug #53985) duplicates of Bug #54621 since that describes the actual problem best, according to Ryan Lortie (thanks for the helpful information)!

Beginning with the Hardy Heron 8.04 development cycle, all open Ubuntu kernel bugs need to be reported against the "linux" kernel package. We are automatically migrating this linux-source-2.6.15 kernel bug to the new "linux" package. We appreciate your patience and understanding as we make this transition. Also, if you would be interested in testing the upcoming Intrepid Ibex 8.10 release, it is available at http://www.ubuntu.com/testing . Please let us know your results. Thanks!