Subject: Re: 7GB memtest causes hard system hang.
To: Frank van der Linden <fvdl@netbsd.org>
From: Greg Oster <oster@cs.usask.ca>
List: port-amd64
Date: 01/16/2006 15:47:30
Frank van der Linden writes:
> On Mon, Jan 16, 2006 at 11:33:08AM -0600, Greg Oster wrote:
> > (Without the "options NKMEMPAGES=150000" tweak, the machine is basically
> > useless to me -- I can easily run "too much stuff" that gets me to the 
> > "no pv entries available" panic :( )
> > 
> > The proper fix is to fix the pmap code.  A short-term workaround is to 
> > do the NKMEMPAGES bump above...
> 
> The pmap code isn't really the problem here. Running out of kernel
> memory (kmem_map in this case) is hard to recover from. With the current
> setup, all that we can do is set nkmempages to a better default value.
> 
> Currently, the default maximum for amd64 is still 128M, which is way
> too low for machines with > 4G of memory. 1G is probably a better
> value for those. Tuning these parameters is somewhat of a black art.

I forget what the "critical point" was, but I just found it very 
annoying that a machine w/ 4GB RAM *free* couldn't automatically find 
a bit more for the kernel to use...  (There's gotta be a way to 
reserve some pages for these sorts of "emergencies" so that 
"something" can grab more memory for the kernel..)  kmem_map should 
automatically expand/shrink as demand warrants...  I'm not sure I 
want to learn that much about that part of the kernel to attempt to
fix that though :) 

Later...

Greg Oster