[apparmor] [RFC] How to handle multiple opencl implementations?

[Jamie Strandboge]

On Wed, 2018-05-09 at 19:55 +0300, Vincas Dargis wrote:
> On 5/9/18 5:05 PM, Jamie Strandboge wrote:
> > On Tue, 2018-05-08 at 23:09 -0700, John Johansen wrote:
> > > 
> > > On top of each of the opencl-XXX abstractions I think it would
> > > be worth having a generic opencl abstraction that includes
> > > the various sub-abstractions, its wide now but the intent
> > > will be to tighten it up with conditionals once better support
> > > lands.
> > 
> > This could work well.
> > 
> 
> Let me check if I got it right.
> 
> So we have implementation-specific abstractions:
> opencl-pocl,
> opencl-beignet,
> opencl-whatever
> 
> And:
> opencl
> 
> that includes all of above by default? In the future, it will have 
> conditional support to disable/enable specific subset of OpenCL 
> implementations? (should be TODO: implement conditional switches to 
> select OpenCL implementations).
> 
> The only problem I see here is the generic `/etc/OpenCL/** r,` rule
> that 
> is used to figure out what OpenCL implementations are available.
> 
> If profile author includes, let's say, only `opencl-beignet`, it
> will 
> have to write these common rules by himself?
> 
> Maybe we need `opencl-common`, that would be included by all of
> specific 
> abstractions?
> 
> I understand that in the future, with proper conditional support,
> there 
> will be no issue I have mentioned as every profile could just
> include 
> `opencl` with proper conditionals defined, and `opencl` will have
> all 
> these common rules.
> 
> So:
> 
> A. we have additional opencl-common?
> 
> B. we don't care too much yet and expect generic `opencl` abstraction
> to 
> be used with all implementations included by default _and_ common
> rules 
> inline, until we get conditional support?

C. in each opencl-* abstraction, include all the rules it needs to work
standalone. Ie, each has `/etc/OpenCL/** r,`

This is what I was expecting, but I wouldn't be opposed to opencl-
common if it had more than that one rule (I don't care for abstractions
that are too fine-grained).

-- 
Jamie Strandboge             | http://www.canonical.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: This is a digitally signed message part
URL: <https://lists.ubuntu.com/archives/apparmor/attachments/20180509/58e7fea8/attachment.sig>