tuneups on previous fix (from torek)
yet more cluster checks
merge vfs_lock and vfs_busy; interlock mount points properly
vget no longer returns a reference on failure
missing vrele in ffs_sync; get rid of hung lock heuristic
If we have failed part way through block allocation, wehave to deallocate any indirect blocks that we have allocated
convert VOP_LOCK to vn_lock; add parameters to VOP_UNLOCK and vget
ufs_reclaim gets proc ptr
mountroot should fail, not panic
use vfs_rootmountalloc
check for forcible unmount at syscall level; better panic check in ffs_sync
change mountlist to CIRCLEQ so we can do unmountall (from mycroft)
lint
give up on rotational layout if there is only one rotational position
have to do vnode_pager_setsize after new length is set (from karels)
have to initialize mp->mnt_vfc before calling ffs_mount (from ralph)
mountroot now uses vfsconf structure; ffs_init moves in from ffs_inode.c;add ffs_sysctl code
doclusterread and doclusterwrite become FFS sysctl variables
add definitions for ffs_sysctl values and names
ffs_init moves to vfs_vfsops.c
doasyncfree and prtrealloc move FFS sysctl variables
have to reinitialize the fs_maxcluster array as it may be wrong
avoid double definition of ufs_daddr_t
daddr_t => ufs_daddr_t
have to save old clsutering info buffer when reloading
12345678910>>...48