First time trying anything like this. I'm running a Abit NF7 with a silicon image 3112 that's support was broken in 2.6.35. I get a kernel panic when VFS can't find the root partition, I can't remember the exact line (can't log it, root isn't mounted yet) but it says something like sata1: link down (sstatus 0 scontrol 0). I think there might be a more telling error message before, but it flies by too fast to see or capture with my camera video recorder.

There's a patch that reverts back to when there was a delay in /drivers/ata/libata-sff.c in the function ata_sff_tf_load, by calling ata_wait_idle(ap) which fixes the issue for some people but not my system.

I don't know anything about what I'm doing, but I thought I'd go through and add printk's in each function in sata_sil and see how far the device was initialized. I did this but they all happen early in the boot that flies by. Is there any way to add a pause like the "more" command? Or a suggestion on a better route to take? I don't see myself solving this problem, I just want to provide enough information to the developers to aid in their work. tia

ps these links seem to be related to the issue I'm seeing
http://marc.info/?l=linux-ide&m=128232270016754&w=2
http://thread.gmane.org/gmane.linux.ide/47506
http://forums.gentoo.org/viewtopic-t...2-start-0.html
https://bugzilla.kernel.org/show_bug.cgi?id=16606

Hi,

Welcome to LQ!

You haven't provided enough information to aid in diagnosis of the problem. Which distribution & version?

I suggest that you look at 'How to Ask Questions the Smart Way' so in the future your queries provide information that will aid us in diagnosis of the problem. Specifics since we are not sitting behind you!

You can use 'ctrl-s' to stop the display & 'ctrl-q' to start the display output. Hopefully you can view the errors if quick enough.

You can use your install CD/DVD to boot the system. Your changes in the filesystem devices may have caused this error by the modifications you made. What devices have you added or what was modified? It does sound like volume or device error that causes the current '/etc/fstab' to have improper device filesystem assignments. After booting from the install media you should be able to revert or change the configuration(s) to reflect the desired settings for the next boot. You will have to mount the specific filesystem to edit or revert the changes you have made.

Try the [Scroll Lock] key during boot. It should stop the display - press again to resume, etc.

EDIT: I should have added, you can use SHIFT+PageUp or SHIFT+PageDn to move up/down a paused boot, if you've paused it with Scroll-Lock. I've not tried the CTRL-S/Q method so not sure if it works with that, if you are trying that method.

Hi, Thanks for your reply. I'm running slackware current. I can boot the system with the 2.6.34.1 kernel, but it does not boot with 2.6.35+ with the same options in .config. There was no change to the hardware or any other system, just upgrading to the latest kernel caused a kernel panic.

I'll try the ctrl-s/q. Is there a function call I can add to the sata_sil functions that will pause like the "less" command does? I don't anticipate being able to hit the ctrl-s fast enough, as soon as it uncompresses the kernel the next thing I see is the "sata link down" error, and from viewing the dmesg output when I boot successfully with 2.6.34.1, there's a lot of output earlier that goes that flies by too fast to hit ctrl-s.

I've added printk("__________function_name"); where "function_name" is the name of the procedure to almost every procedure in sata_sil.c in drivers/ata/

I added the underscores so I can differentiate my debugging from standard kernel output. Is there some "pause()" function I can call that waits for a keypress? I have no way to log the output because the panic happens before init is called and before anything is mounted.

Kernel panic happens before root fs is mounted (before init)