2.6.9-1.11_FC2 - "Kernel panic

michaelsr · 01-10-2005, 12:10 PM

I have just installed the new kernel: 2.6.9-1.11_FC2 and when I try to boot it produces the following error message (the previous kernel still works without any problems - 2.6.9-1.6_FC2):

kernel panic - not syncing: drivers/ide/pci/piix.c:390:spin-lock(drivers/ide/ide.c:c03712a8) already locked by drivers/ide/ide-iops.c/1234

Anyone have any idea what is wrong and what I should do?

Thanks,

Michael.

tcordes · 01-12-2005, 10:54 AM

Same thing just happened to me when updating FC1 to FC2 using yum. The kernel is 2.6.10-1.8_FC2. The board is a MSI-6565 (Intel E7205) with a DupliDisk1 for hardware RAID1.

I managed to boot into the new kernel by tweaking the grub kernel boot parameters at the grub boot screen to add "ide=nodma" (no quotes) to the linux line and making sure that /etc/sysconfig/harddisks had USE_DMA=0.

Can you let me know what hardware you are using? I think my issue is a conflict between the board and the DD1.

JordanH · 02-08-2005, 04:45 PM

tcordes nailed this one on the head.

I have not been able to boot with the new kernel's that are being distributed with Fedora Core 3. FC3 works just fine with 2.6.9-681 inclusive but fails with the following message for all subsequent kernel's.

Quote:

Kernel panic - not syncing: drivers/ide/pci/piix.c:232: spin_lock(drivers/ideide.c:c036e408) already locked by drivers/ide/ide-iops.c/1222

The above fix seems to work. My question is what changed and why is it important? My drives are new enough, both Western Digital Caviar's (160GB and 200GB) so they should be supported. Why would this break between kernel releases?

Thanks,
Jordan.

JordanH · 02-08-2005, 04:54 PM

The fix above seems to have cut my timing buffer reads to squat!

hdparm -Tt /dev/hda = 2.91MB/s
hdparm -Tt /dev/hdc = 3.06MB/s

Totally not cool. I need another solution fast.

tcordes · 02-08-2005, 10:06 PM

Are you using any sort of RAID device?

If not, perhaps the new kernel has some flaw in DMA handling for certain chipsets/drive combos. That would be bad -- we should look for a RH bugzilla report on this.

And yes, your disk performance will go all to heck with DMA off -- it can be really painful.

JordanH · 03-12-2005, 10:42 AM

I am not using a RAID device. Just a standard, IBM NetVista PC. a couple IDE drives a couple optical drives. Nothing fancy.

I've been on the road for the last 5 weeks and again for the next 3 so I won't be home to play with this box for a while. I will be interested in hearing if anyone else resolved this issue yet...

meyster · 03-26-2005, 03:28 AM

the same problem on my system.

FC3. The 681 is the last kernel that works fine, but all newer kernels have this problem.

I have an older Intel 815 Board and a P3-1GHz CPU with 2 x Samsung 160 GB. It's my file-server so deactivating DMA is no real solution.

Will try another 815 based board soon.

JordanH · 03-26-2005, 11:43 AM

Has anyone else reported this yet?

tcordes · 04-01-2005, 05:09 PM

Quote:

Originally posted by JordanH
Has anyone else reported this yet?

I haven't bugzilla'd it yet, but there may already be some related ones there. The system I run it on is at a different location, and a production machine so I can't easily test new ideas or provide the type of info a bugzilla would require. Someone should make sure it's known in the normal channels.

JordanH · 04-01-2005, 07:32 PM

Yay, I'm finally home now and can play with this all I like (or at least until the GF comes over and complains that I'm on the computer again)

I just did a bugzilla search...
https://bugzilla.redhat.com/bugzilla....cgi?id=144743
https://bugzilla.redhat.com/bugzilla....cgi?id=145097

My problem is the first one. The second one looks related.