Quoting Vasant_-_Kanchan@cup.portal.com,

>       Can some one explain to a novice what is the problem
>    with the  BSD ( Mach ) VM system ?. The seperation of the machine
>    dependent parts into the pmap module seems like a good
>    idea. Only recently I have been looking at the BSD lite VM code.
>    So my knowledge is limited.

The main problem is, that the VM system does not know about the existance
of the I/O buffer cache. This can lead to a situation where, for example,
you use read(2) as well as mmap(2) on a file, you end up with 2 copies
of the same data in memory. One in the buffer cache and one as a mapped
range of memory. This is not only a waste, it also has the problem of
changes in one not being reflected in the other. This problem does occur
in practice, as several bug reports show.

Another problem is, that the code just looks too .. complex for its purpose
in NetBSD in general. You say that you have been looking through it, so I
assume you know what I mean here ;-)

There are some other issues that will hopefully be addressed when the VM
system is dealt with.

- Frank