> From: Paulo Alexandre Pinto Pires <pappires@vortex.del.ufrj.br>
> Date: Sat, 26 Apr 1997 01:24:54 -0300 (EST)

> Hi, all.
> 
> > > This is my problem.  The machine halts at multiuser boot, with
> > > the following message:
> > > 
> > > 	pmap_fault: kernel_map & va < avail
> > > 
> > > What does this mean and how can I overcome it?
> > 
> > Can you get DDB to give you a backtrace?  That would be very helpful
> > in determining why the fault occurred. 
> 
> I'll only be able to do it on Monday now, but I'll send the results
> as soon as I can.
> 
> But what does this mean, anyway?

It means the kernel took an MMU fault (page not mapped)
at an address where the kernel does not allows faults.

> From: Paulo Alexandre Pinto Pires <pappires@vortex.del.ufrj.br>
> Date: Mon, 28 Apr 1997 17:19:41 -0300 (EST)
[...]
> Stopped at	_Debugger+0x6: unlk a6
> db> trace
> _Debugger(0,8,e553e00,ee87b38,e075c1a) + 6
> __pmap_fault(e100000,0,1) + 32
> _trap(8,145,8) + 3ca
> sun3_mmu_specific(?)
> _nfs_reply(e528680) + 38
> _nfs_request(e554d80,e554700,7,0,e525a80,ee87d20,ee87d1c,ee87d18) + 31c
> _nfs_writerpc(e554d80,ee87da4,e525a80,ee87d98,ee87d94) + 5c8
> _nfs_doio(e726258,e525a80,0,1,5f) + 3ac
> _nfs_strategy(ee87e10) + 78
> _nfs_writebp(e726258,1,ee87ef0,e05fe50,ee87e98) + 142
> _nfs_bwrite(ee87e98) + 10
> _nfs_flush(e554d80,e525a80,1,e553e00,1) + 3ae
> _nfs_fsync(ee87f24) + 1c
> _sys_fsync(e553e00,ee87f84,ee87f7c) + 86
> _syscall(5f) + 13e

More specifically, the backtrace above shows that there was an
attempt to access VA=8 (null pointer plus struct offset?), and
that access happened at nfs_reply+0x38.  The next step would be
to disassemble that function to see what data structure it was
using, and try to deduce why it ended up using a null pointer.

Are your kernel sources up-to-date with -current?

Gordon