RFC: What to do about VOP_INACTIVE?

[Poul-Henning Kamp]

--------
In message <CAG6CVpU8J=G8za8W2uan8SAGEbe4PD=SXwMow=E_mkJnMGB96A at mail.gmail.com>, Conrad Meyer writes:

>The nominal return type of the VOP_INACTIVE vnode method is 'int', but
>in practice any error returned is silently discarded.

The return type is an historical accident from John Heidemann's phd.

The way he implemented stackable filesystems cast all VOP's to:

	typedef int vop_t __P((void *));

So that filesystems could have a VOP methodvector which looked like:

	static struct vnodeopv_entry_desc fifo_vnodeop_entries[] = {
		{ &vop_default_desc,            (vop_t *) vn_default_error },
		{ &vop_abortop_desc,            (vop_t *) fifo_badop },
		{ &vop_access_desc,             (vop_t *) fifo_ebadf },
		{ &vop_advlock_desc,            (vop_t *) fifo_advlock },
		[...]

Then on runtime, he built the real vop dispatch tables.

Around 1997 I untangled that so that we got compiler type-checks
all the way though the VOP calls, and there were enough historical
cruft to clean up that I never got around to the return type of
the individual VOPs.

If you want to fix this, the right thing to do is to add return types
in vnode_if.src and teach the tools/vnode_if.awk script about them.

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.