Subject: Re: mc* performance patch
To: Allen Briggs <briggs@NetBSD.org>
From: Chris Tribo <ctribo@college.dtcc.edu>
List: port-macppc
Date: 01/14/2005 23:16:16
On Fri, 14 Jan 2005, Allen Briggs wrote:

> On Fri, Jan 14, 2005 at 10:31:03PM -0500, Tim Kelly wrote:
> > I may have reversed the chips between Beige G3s and AGP G4s.
>
> Yeah.  I think the beige g3s use if_bm.

Correct. Anything gossamer based, B&W G3, PCI G4 and Mainstreet,
Wallstreet, PDQ. I'm not sure on the pismo and lombard, but I assume they
are bmac too, but like the B&W G3 and PCI G4's they support 100MB/sec.

The Broadcom BCM57xx was added in with Sawtooth and later.

> > I did not know that mii needed certain hardware to be present. I thought it
> > was a generic approach to ethernet device drivers, and hadn't been applied
> > yet to those two.
>
> $ wtf mii
> mii: mii (4) - IEEE 802.3 Media Independent Interface network bus
>
> There are a number of ways that the MAC can talk to the physical
> ethernet bus.  MII tries to standardize the interconnection so
> that, in theory, the same PHY can be used with a range of MACs (or
> vice versa).  In many cases these days, even integrated chips will
> use a MAC <-> MII <-> PHY structure on a single core.  There are
> some variants like GMII and RGMII for gig-e, too.
>
> With an MII and PHY attachment, the software interface for probing
> and configuring the media should be the same, even if the hardware
> is different.  But there are some quirks in different PHY
> implementations, so you can't always use 'ukphy' (unknown/generic
> PHY) or the equivalent 'gentbi' (generic 10-bit interface) for fiber.
>
> -allen
>
>

-- 
The nice thing about Windows is - It does not just crash, it displays a
dialog box and lets you press 'OK' first.
						-- Arno Schaefer