Subject: Re: vm_fault on i386...
To: Andrew K. Adams <email@example.com>
From: Matthias Drochner <firstname.lastname@example.org>
Date: 06/18/1997 20:31:41
Excerpts from netbsd: 18-Jun-97 Re: vm_fault on i386... Jason
> > root@wraith# vm_fault(0xf880d100,0,1,0) -> 5
> > kernal: page fault trap, code = 0
> > stopped at _pmap_changebit + 0x83: movl 0(%edx),%eax
I've seen similar errors.
The reason was that pde's disappeared magically.
This seemed to be triggered by a combination of heavy DMA
activity (locked pages + tsleep()) and process creation /
Nasty race condition - couldn't track it down at this time.