2014-05-24 10:28:56

by dev

[permalink] [raw]
Subject: [refpolicy] Associate attribute with another attribute?

Can attribute be associated with another attribute, the same way that is
done with type(s)?

eg. If we associate attributes with types using: "type mytype1_t,
my_attribute1, my_attribute2...;",

can we associate attributes with attributes using: "attribute
my_attribute0, my_attribute1, my_attribute2...;", or something similar?

--
Mladen Sekara <[email protected]>


2014-05-24 10:29:53

by Daniel Walsh

[permalink] [raw]
Subject: [refpolicy] Associate attribute with another attribute?


On 05/24/2014 06:28 AM, Mladen Sekara wrote:
> Can attribute be associated with another attribute, the same way that is
> done with type(s)?
>
> eg. If we associate attributes with types using: "type mytype1_t,
> my_attribute1, my_attribute2...;",
>
> can we associate attributes with attributes using: "attribute
> my_attribute0, my_attribute1, my_attribute2...;", or something similar?
>
No, although it would be great if it could.

2014-05-24 11:14:34

by dominick.grift

[permalink] [raw]
Subject: [refpolicy] Associate attribute with another attribute?

On Sat, 2014-05-24 at 20:28 +1000, Mladen Sekara wrote:
> Can attribute be associated with another attribute, the same way that is
> done with type(s)?
>
> eg. If we associate attributes with types using: "type mytype1_t,
> my_attribute1, my_attribute2...;",
>
> can we associate attributes with attributes using: "attribute
> my_attribute0, my_attribute1, my_attribute2...;", or something similar?
>

Not with reference policy but it is possible with CIL policy.

Do not ask me how they achieve that though because i do not know.

I suppose that they expand the attributes before the resulting policy
gets translated to policy the kernel understands because i think it is a
limitation is the kernel policy language.

Not that it matters much though, it is handy nevertheless.

2014-05-25 04:10:39

by dev

[permalink] [raw]
Subject: [refpolicy] Associate attribute with another attribute?

Ah, OK.
Probably not needed very often, but in same cases it could save some
time.

Thanks.
--
Mladen Sekara <[email protected]>


On Sat, 2014-05-24 at 06:29 -0400, Daniel J Walsh wrote:
> On 05/24/2014 06:28 AM, Mladen Sekara wrote:
> > Can attribute be associated with another attribute, the same way that is
> > done with type(s)?
> >
> > eg. If we associate attributes with types using: "type mytype1_t,
> > my_attribute1, my_attribute2...;",
> >
> > can we associate attributes with attributes using: "attribute
> > my_attribute0, my_attribute1, my_attribute2...;", or something similar?
> >
> No, although it would be great if it could.

2014-05-27 12:40:52

by cpebenito

[permalink] [raw]
Subject: [refpolicy] Associate attribute with another attribute?

On 05/24/2014 07:14 AM, Dominick Grift wrote:
> On Sat, 2014-05-24 at 20:28 +1000, Mladen Sekara wrote:
>> Can attribute be associated with another attribute, the same way that is
>> done with type(s)?
>>
>> eg. If we associate attributes with types using: "type mytype1_t,
>> my_attribute1, my_attribute2...;",
>>
>> can we associate attributes with attributes using: "attribute
>> my_attribute0, my_attribute1, my_attribute2...;", or something similar?
>>
>
> Not with reference policy but it is possible with CIL policy.
>
> Do not ask me how they achieve that though because i do not know.
>
> I suppose that they expand the attributes before the resulting policy
> gets translated to policy the kernel understands because i think it is a
> limitation is the kernel policy language.
>
> Not that it matters much though, it is handy nevertheless.

Eventually I'd like to make a proper refpolicy high level language on top of CIL, when CIL gets merged. Then it would allow all of the nice features in refpolicy that we all want.

--
Chris PeBenito
Tresys Technology, LLC
http://www.tresys.com | oss.tresys.com