YAMAMOTO Takashi wrote:

> i'd vote to have a separate "passthru" kauth request.
> it's different enough from normal raw device i/o, IMO.

I'm somewhat concerned we are providing two requests that can end up
implying more or less the same. If we do that, it'll have to be
documented properly.

How will the policy look like? do we always assume worst-case and
act as if "passthru" is raw i/o to mounted disk or memory?

-e.

-- 
Elad Efrat