Steven Stallion wrote:
On Mon, 21 Sep 2009 06:55:55 -0700, Garrett D'Amore <[email protected]>
Some hardware simply *can't* easily do this. Some drivers are really
stale and haven't been updated:
* dnet only does link state reporting on certain cards (this with MII
transceivers)
This one is my fault. Unfortunately the PHYs were so varied on these
devices that adding in link state reporting for all variants was
prohibitively expensive.
Yes. Crufty old tulip cards with SIA based PHYs instead of MII. It
gets hard to do this properly, especially in face of potential
interfaces like thin-net, fiber, and AUI. I'm not even sure it is
*possible* to detect link failure in all those situations.
* spwr doesn't do it all (but spwr could easily be improved, and indeed
it ought to be.)
I've actually started the process of emancipating spwr. Once this
is finished, the 'new' spwr will support all of the modern GLDv3
entry points, including mac_link_update and of course m_setprop/m_getprop.
Good. spwr is actually a fairly simple device -- much simpler than our
current (closed) code would lead one to believe. You can count me in a
sponsor for integrating your version of the driver when its ready. :-)
- Garrett
_______________________________________________
networking-discuss mailing list
[email protected]
