Re: CVS commit: src/sys/kern

[Andrew Doran <ad%netbsd.org@localhost>]

On Sun, Dec 07, 2008 at 12:25:58PM +0100, Christoph Egger wrote:

> Andrew Doran wrote:
> > On Sun, Dec 07, 2008 at 02:21:04AM +0000, Christoph Egger wrote:
> > 
> >> Log Message:
> >> Spans may never overlap. Overlapping spans mean memory corruption
> >> when used by kmem(9), for example.
> >> Do sanity checks to detect such spans on DEBUG kernels.
> > 
> > Cool!
> > 
> > One problem: vm->vm_lock isn't held at some points where you call
> > vmem_check_sanity(), so it could crash or return false positives.
> 
> The bug I'm about to track down happens during uao_create() called
> by uvm_init(). Does the vm_lock problem matter in this very early stage?

Probably not, but vmem is used all the time.

Andrew