Use a reader count instead of holding lock.

[sson]

Summary

This change uses a reader count instead of holding the mutex for the
interpreter list to avoid the problem of holding a non-sleep lock during
a page fault as reported by witness.  In addition, it consistently uses
memset()/memcpy() instead of bzero()/bcopy() except in the case where
bcopy() is required (i.e. overlapping copy).

[sbruno]

Tested on head with WITNESS/INVARIANTS and all options in GENERIC. This resolves the witness enabled panic on HEAD.

[sbruno]

[sbruno]

https://svnweb.freebsd.org/changeset/base/284029

[sbruno]

This was reverted after review. Will go back to D1971 as per suggestions to really try to use an sx_lock