> > Sounds OK to me. But given Jason's semi-recent changes to a more
> > uniform and MI setroot/conf/swapconf, does this even need to be
> > CPU-dependent?  Can it be made MI?
> 
> [snip]
> Perhaps we could insert three functions before the vfs_mountroot
> call, like this:  (instead of just the one shown previously)
> 	cpu_rootconf();
> 	cpu_swapconf();
> 	cpu_dumpconf();
> 
> and put the findroot/setroot stuff in cpu_rootconf().  Whatever.

Looks like swapconf()/dumpconf() usage is the same on every port
(as it should be), so you might as well call those directly from
main().  (I guess this the `MI' thing Jonathan was aiming at..)

-pk