Re: kernel libraries and dead code in MODULAR kernels

[David Holland <dholland-tech%netbsd.org@localhost>]

On Mon, Aug 31, 2015 at 04:43:17PM +0000, Eric Haszlakiewicz wrote:
 > On August 30, 2015 11:31:54 PM EDT, Masao Uebayashi <uebayasi%gmail.com@localhost> wrote:
 > >I believe that the exact problem exists in userland's dynamically
 > >linked libraries/programs, right?  If so, how do they deal with this
 > >dead code problem?
 > >
 > In userland the loader maps the library in, and if you happen to
 > have enough adjacent dead code then that memory just doesn't get
 > paged in.  I doubt that's an option for kernel code, and it isn't
 > likely to work particularly well for minimizing memory usage in
 > either case unless you take steps to sort things in some way.

In userland we pretend it isn't important because it allows keeping
one copy of the library code shared across ~all processes, instead of
having one copy each. This doesn't apply in kernel space.

-- 
David A. Holland
dholland%netbsd.org@localhost