* amd/srvr_nfs.c (find_nfs_srvr): don't blindly copy the hostent
IP address. First check if it differs from the existing one of
the fserver, and copy only if it changed. If it did change, flush
the fhandle cache to avoid a stale fhandle being reused. This
allows Amd to detect IP address changes even for an fserver that
lost one or more NFS pings, but not enough to be declared totally
down. We handle the "totally down" fserver case in
check_fs_addr_change().
* amd/ops_nfs.c (nfs_umount), amd/sched.c (sched_task),
amd/nfs_subr.c (nfs_quick_reply): code clarity.
* conf/mount/mount_linux.c (linux_nfs_error): dlog mappings of
errnos to NFS errors.
* conf/umount/umount_linux.c (umount2_fs): cleanup code. Trying
stat() seems doomed to hang at times, so don't try stat at all,
because umount2() appears to be clever enough to never hang.