[brussels-dev] Re: [driver-discuss] Re: [networking-discuss] Brussels high-level design document

[sowmini.varadhan at sun.com]

Garrett,

let me break up the review comments into smaller chunks (some of which
will be addressed by the co-author, Raymond).

On (06/18/07 11:50), Garrett D'Amore wrote:
> * we still need to work out the details on a few items, including kstat 
> snapshot collection (you indicated it needs consideration), and property 
> registration.

right. I tried to avoid getting into implementation details in the
design doc, partially because the implementation is subject to change
(since I have not worked out all the implementation yet), and because
some reviewers may find this distracting. Do you have any suggestions
for a high-level proposal here?

> * mac_list_private_props ... maybe exposing this as a property might not be 
> the best choice.  its not clear.  i'd love to see this idea more fully 
> explored though.

I kind of liked the idea myself (it was proposed by Huafeng Lv, who
pointed out that QA might need some crutch to figure out syntax/spelling
errors)- what I liked about this, is that it allows a GUI like NWAM 
to get to private property information. I was looking at how the 
gui output changes on my XP control panel as I switch interface cards and 
printing private properties seems like a useful thing to do. 

I understand your concern though, as this exposes something that is
highly unstable, and not something that we want to share with customers.

Maybe a middle ground is to have mac_list_private_props as a private
property itself.

> * I'd avoid making notes/comments about the STABILITY level of individual 
> properties at this point in the games.  We should do that just before the 
> case goes to ARC.

Ok. that's a good point. I was mostly putting it in as a place-holder,
but I agree that it is premature to discuss Stability yet. I'll update
this for the next revision.

> * what about WLAN properties?  It would be nice to consider the current, 
> separate WIFI configuration mechanism to see if we can integrate it into 
> Brussels.  Maybe not initially, but certainly in the mid- term.  (Think 
> SSID, link keys, etc.)

In the case of WLAN properties, my understanding is that we can't integrate
all of them into Brussels, because of the point mentioned in Page 9: some
of those system calls have been written with a desire to retain binary 
compat with code on opensolaris. There may be a way to deal with this
in the same manner as ndd, though: I'd have to think about it..

Thanks for all the comments and input! It's nice to see it all coming
together with different perspectives.

--Sowmini