Make autounmountd(8) less intrusive.

[trasz]

Summary

Add a new unmount(2) flag, MNT_NONBUSY, to check whether there are any
open vnodes before proceeding. Make autounmound(8) use this flag. Without
it, even an unsuccessfull unmount causes filesystem flush, which interferes
with normal operation.

[kib]

sys/kern/vfs_mount.c

1219

Why do you need to lock the root vnode ? You vput the root vnode immediately after VFS_ROOT(), which allows the memory to be reused. Just checking VV_ROOT is enough for your purposes.

1234

The printf() must be removed.

1287

This only should be done after we set _our_ MNTK_UNMOUNT flag.

[bcr]

typo

sbin/umount/umount.c

131

s/mutualy/mutually/

[trasz]

Misc fixes.

[trasz]

sys/kern/vfs_mount.c

1287

I'd prefer not, to be honest. This is "best effort" functionality, false positives (or negatives) are just fine, autounmountd(8) can cope with them, and the whole point of this is to make it as non-intrusive as possible.

[kib]

sys/kern/vfs_mount.c

1287

Right, so you do not want to race for the full vnode list iteration with the parallel unmount.

[trasz]

Move vfs_check_usecounts() down slightly, as suggested by kib@.