Re: Kernel BUG at mm/vmscan.c:428 (2.6.17-rc4-git2, Dualcore AMD x86-64)

[Nick Piggin <nickpiggin@yahoo.com.au>]

Antti Salmela wrote:
> My Dualcore AMD x86-64 desktop has been a little bit unstable as long as I
> remember, but at last I had enough willpower to enable netconsole to see
> if anything was reported after a lockup. For some reason these
> BUG messages made now their way to local log file too.
> 
> May 15 06:51:43 enigma kernel: ----------- [cut here ] --------- [please bite here ] ---------
> May 15 06:51:43 enigma kernel: Kernel BUG at mm/vmscan.c:428
> May 15 06:51:43 enigma kernel: invalid opcode: 0000 [1] PREEMPT SMP 

Thanks for reporting.

Either you have an active page on the inactive list, or your hardware has
flipped a bit in page->flags. I was going to say the latter is more likely,
however -- AFAIKS, the first oops should cause that page to be lost from the
LRU list, so the second oops shouldn't happen if the flip a single bad bit,
and should be pretty unlikely if it is a random error.

Still, anything explainable by a single bit flip is worth running memtest86+
overnight for.

Nick

-- 
SUSE Labs, Novell Inc.
Send instant messages to your online friends http://au.messenger.yahoo.com