1) Could you send-pr this so it will *not* be lost? Preferably with
   patches! 

2) I suspect this might be too late to do this for 1.3 -- the cut is
   mere hours off, and as much as one would like to get every possible
   improvement in there comes the time to "shoot the engineers and
   ship". However, we are planning on doing our next release much
   sooner than the 18 month gap between the branch for 1.2 and 1.3. I
   would guess we'll start in on 1.4 only a few months after 1.3.

Perry

rvb@sicily.odyssey.cs.cmu.edu writes:
> I was just retooling a 1.2 kernel to deal with > 64Meg of memory.
> I took a look at -current and noticed that although its more 
> automatic, the parameters are still too small.  The kernel starts
> at va 0xf8000000 or pde 0x3e0.  This means that the kernel has at
> most 0x20 pde slots or can span 128Meg in va.  I think that this
> value could get quite tight over the next year as people add more
> and more memory to machines.  (I have a server now that has 256Meg
> and I'm planning to use half for buffers.)  I'd like to suggest
> for NetBSD 1.3 that we move the kernel start address back a bit,
> to 0xf00000000 (pde 0x3c0) or maybe even further.