Re: Resolving open/close, attach/detach races

[Taylor R Campbell <riastradh%NetBSD.org@localhost>]

> Date: Tue, 1 Feb 2022 16:45:29 +0100
> From: "J. Hannken-Illjes" <hannken%mailbox.org@localhost>
> 
> The new KASSERT(devvp->v_specnode->sn_opencnt) from
> spec_node_setmountedfs() fires for unmounts of udf fs as operation
> udf_vfsops.c::udf_unmount() closes the device BEFORE setmountedfs().

Looks like spec_node_getmountedfs/setmountedfs is pretty racy as
currently used, and needs an audit.  I'll remove the assert for now
and instead leave a comment about why these are racy and what needs to
be audited in *_vfsops.c.