Subject: Re: pmap_zero_page problem
To: Jason Thorpe <thorpej@nas.nasa.gov>
From: Manuel Bouyer <bouyer@antioche.lip6.fr>
List: port-i386
Date: 03/31/1999 14:38:06
On Mar 30, Jason Thorpe wrote
> If you're using the nointr allocator, it should never be called in interrupt
> context.

I could't get a stack trace from an irq handler. This one was just used as
a base for source analisys.
> 
> If you're using the default allocator, kmem_map is used instead.  This
> should NOT call pmap_growkernel() to be called, since the KVA space is
> already reserved for kmem_map.

Ok, that's what I missed. I mixed kmem_map and kernel_map:
uvm_km_alloc_poolpage1() is called with kmem_map, and uvm_map() explicitely
checks for map == kernel_map before calling pmap_growkernel().

I'll try something else this evening.

--
Manuel Bouyer, LIP6, Universite Paris VI.           Manuel.Bouyer@lip6.fr
     {Net,Free}BSD: 22 ans d'experience feront toujours la difference
--