Re: Removal of some KAUTH_GENERIC_ISSUSER (pass 1)

[christos%astron.com@localhost (Christos Zoulas)]

In article 
<4f68037e0904111009l4848990dkf22322129f90ae05%mail.gmail.com@localhost>,
Elad Efrat  <elad%NetBSD.org@localhost> wrote:
>On Sat, Apr 11, 2009 at 7:40 PM, Izumi Tsutsui 
><tsutsui%ceres.dti.ne.jp@localhost> wrote:
>
>>> Could you give an example for other ports that may use it? it might be
>>> that some are miscategorized (perhaps as KAUTH_MACHDEP_UNMANAGEDMEM) or
>>> missing.
>>
>> sys/arch/prep/pnpbus/nvram_pnpbus.c
>> sys/arch/sparc/sparc/openprom.c
>> sys/arch/sun3/dev/mem.c
>> sys/dev/ofw/openfirmio.c
>> etc?
>
>Hm, these look like they're subject to the permissions on the device
>file. We'll tackle adding kauth(9) calls there some other time.
>
>> I'm not sure if they should have the same semantics with
>> i386/isa/cmos.c, though..
>
>Nothing prevents us from adding several KAUTH_MACHDEP_FOO actions to
>handle different semantics. :) Would you say it's sensible to add
>KAUTH_MACHDEP_CMOS for the cmos.c case and KAUTH_MACHDEP_NVRAM (perhaps
>with read/write granularity) in the future for different
>cases/semantics?

I think that the cases are going to be similar and I don't really want
to add more KAUTH_MACHDEP_FOO if I don't have to. I like KAUTH_MACHDEP_NVRAM
over KAUTH_MACHDEP_CMOS because it indicates the function not the underlying
technology.

christos