YAMAMOTO Takashi wrote:

> IMO, it should
> 	- handle normal filesystem activities including file removal.

I'd go with this one.

> my main concern is, unless metahook handle filesystem activities as its own,
> your new users of it might end up to put their own adhoc hooks into
> vfs code as veriexec does.

Absolutely correct. However, I'm wondering if triggering an event (or
several) when a file is removed shouldn't be implemented as part of a
"fileop" kauth(9) scope?

(and have a "metahook listener", for example)

> yes.  and function names etc.

Okay, so metahook_inode_foo() becomes metahook_file_foo() and "ino_t
inode" becomes "ino_t file"?

-e.

-- 
Elad Efrat