YAMAMOTO Takashi wrote:
>> YAMAMOTO Takashi wrote:
>>>>> and, if possible, the acutal user request (eg. mly_user_command).
>>>> I assume mly_user_command is a kauth(9) enum? (like KAUTH_REQ_DEVICE_..)
>>>> if yes, do we care about 3rd-party device drivers that can use such
>>>> requests? in such cases we don't have the actual user request.
>>> i meant to pass a driver-specific structure via void *.
>>> bsd44 listeners will just ignore it.
>>> a listener can investigate the arg only when it knows the driver.
>> okay. we have kauth_authorize_device_tty(). do you want to add a
>> kauth_authorize_device() or a different routine? if so, any preference
>> for a prototype?
[...]

> i personally prefer everything typed as far as possible.
> 
> int kauth_authorize_device_passthrough(kauth_cred_t cred, dev_t controller,
>     const void *user_request);

hrm. that seems overkill to me...

> btw, isn't it better to move RAWIO to the device scope?

for both memory and disk?

-e.

-- 
Elad Efrat