> I'm not suggesting that anybody drop what they're doing and
> make this happen, but perhaps it would be a useful optimization to not
> do the fixups a second time if an executable has already been loaded
> and fixed up.

"Right, but how do you do that?"

I mean, the fixups are logically 'local' (MAP_PRIVATE) to the process
that's been fixed up.

If you have a more complicated shared library scheme that does object
caching (some of the work at Utah comes to mind), then you can do
something more intelligent.  But it's not a simple hack.



cgd