As I threatened long ago, and as everyone polled seemed to approve of,
I intend to soon replace all b* functions in the kernel (bcopy, bzero,
etc.) with the correct mem* functions. This will provide a number of
benefits, among others the fact that gcc can inline those functions
on many platforms.

I have several questions.

0) Are there last minute objections?

1) In theory, by being smart about when to use memmove (which handles
overlaps) and when to use memcpy (which does not) we could
potentially, on some platforms, gain performance in cases where paying 
attention to overlap isn't necessary. Does anyone think this is
important, or should I just be "safe" and always replace bcopy with
memmove?

2) Do all our platforms currently have memmove, memcpy, memset, and
memcmp available in libkern, or am I going to have to fix that first?

Perry