[prev in list] [next in list] [prev in thread] [next in thread] 

List:       netbsd-port-i386
Subject:    Re: ddb 'step' recovers from a page fault
From:       David Young <dyoung () pobox ! com>
Date:       2008-05-20 15:27:00
Message-ID: 20080520152700.GE28329 () che ! ojctech ! com
[Download RAW message or body]

On Fri, May 16, 2008 at 01:00:04PM -0500, David Young wrote:
> I am seeing "recoverable" page fault traps in ether_output().  What I
> mean by "recoverable" is that if I type 'continue' at the ddb prompt,
> the fault does not re-occur!  In the console log, below, I have typed
> 'step' instead of 'continue' at the debugger prompt.  Note that after
> the 'step', esp and ss registers have changed from outlandish numbers
> (esp 0xc0743240, ss 0x2) to normal numbers (0xc6d427a4, 0x10).

New datapoint: it looks to me like most the vm_map_entry's in the kernel
map are not "wired," and the map itself is marked pageable.  That is, most
of the entries wired counts are 0, and the kernel map flag VM_MAP_PAGEABLE
is set.  The unwired entries include entries covering the kernel text
and data.  Is it possible that the kernel is trying to swap kernel pages?
(Swap is off!)

Dave

-- 
David Young             OJC Technologies
dyoung@ojctech.com      Urbana, IL * (217) 278-3933 ext 24
[prev in list] [next in list] [prev in thread] [next in thread]