Kernel panic: Freeing swap cache pages [Was: Re: Kernel 2.2.1-pl6 problems on Ruffian]
Brederlow <email@example.com> writes:
> I managed to compile and run a 2.2.1 kernel with patch 6 from
> alan. The problem i had before with kernels not booting at all seems
> to be a gcc problem, after I updated my gcc to the current the
> compiled kernels boot.
> But some Problems are left:
> After freeing kernel memory (and only at that time) the kernel hangs
> quite regulary (50% chance?) with an error. Is that known? I will
> write down the exact message next time I boot.
Well, its more like 80-99% of the time. The strange thing is that it
seems to depend on the speed I specify for the ncr driver.
The full message is:
Kernel panic: Freeing swap cache pages
In swapper task - not syncing
I circumvented this by changing the code causing the panic. Now
instead of a panic I just return without freeing the pages. Seems to
> I got a Oops from the spinlock code when using mc to copy a bunch of
> files. I got it only once so far and it might be scsi related.
None have appeared so far. Lets see if anotherone appears sometime.
> I have a Ruffian, which has an onboard 53875 scsi controler. I
> compiled the kernel with the 538xx bsd driver. When I use anything
> else but asynchronus transfer I get scsi errors all the time,
> resulting in a crash every so often. I managed to use a setting of 5
> MHz synchronus, but that seems to be slightly falkey as well. I'm
> wondering whether this is a termination problem or not, so I'm taking
> my alpha apart. Might also be a problem with scsi cable length, since
> I have UW-scsi on one side and U-scsi on the other. Both have rather
> long cables.
I changed the order of my drives and rechecked the termination. I now
have a different drive (my zip drive) terminating the fast scsi
bus. The problem is mostly gone, except one reset at the boot. I seem
to get one timeout during the boot, but not again after that.
> Has anybody else other or same experiences? Especially the onboard
> scsi controler gets me, because asynchon scsi is like seeing a movie
> in slow motion.
Much better now. :) hdparm tells me I get 14 MB/s from my raid.
May the Source be with you.