Jason Thorpe wrote:

> On Mon, 14 Feb 2000 20:29:04 -0500 
>  David Maxwell <david@fundy.ca> wrote:
> 
>  > In the form of a kernel thread? Or were you hinting at some other aspect
>  > that needs to be fixed?
> 
> It needs to be rewritten completely, and not just be a hack underneath
> UFS :-)
> 
> I.e. block allocation should dynamically allocate pages (up to some
> hardlimit, of course), and freeing a block should dynamically free
> pages... it should be using pageable kernel memory, not hanging on a
> user process's address space, etc.

I've been recently thinking about something like this as spare-time
project to learn about both UVM and filesystems, both of which I've
never really played with.  Is there any simple existing examples
of something that allocates pageable kernel memory - does the pool
allocator fall into this category?

Simon.