On Sat, 29 Jan 2000 17:58:21 -0500 (EST) 
 wpaul@ee.columbia.edu (Bill Paul) wrote:

 > They sent me two DM9102A cards recently that took a couple of weeks to
 > arrive. The DM9102A is essentially the same as the DM9102 except it has
 > an external MII interface. The eval boards they sent me have both
 > fast ethernet and HomePNA interfaces, with the HomePNA PHY connected
 > to the external MII interface. However, the way they did it, the
 > external MII management interface is separate from the internal one,
 > meaning that when you flip the port selection bit in CSR6, you can
 > see either the interal PHY or the external one, but not both at the
 > same time. In fact, the internal and external PHYs are wired to MII
 > address 1.

Wow.  That's .... really goofy.  Is the HPNA PHY the 2Mb/s or 10Mb/s
variety?  Just wondering how bizarre the programming interface is.

 > This struck me as a bit goofy since it means you need to do two MII
 > bus probes in order to find both PHYs, and you need two miibus device
 > handles to handle both buses.

Um, yah, wierd.

 > I originally got a preliminary copy of the ST201 datasheet from D-Link,
 > who is apparently using it on a card called the DFE-550TX. I was told this
 > card would appear on the U.S. market eventually, but I haven't seen it
 > yet. D-Link sent me two sample cards. If you can't get one from Sundance,

I might contact D-Link about it, then... thanks for the tip.

 > I can probably part with one of mine. However, don't let Sundance off the
 > hook too easily. I originally had my copy of the datasheet on www.freebsd.org
 > (which I normally do for datasheets from companies who don't already have
 > manuals on their web sites -- since I don't sign NDAs, if they give me
 > something, then I feel justified in doing anything I damn well want with
 > it), however the Sundance people complained and asked me to take it away
 > from there. I told them that I would only do it if they made it readily
 > available on their own server, and if they didn't, then I would put it
 > back.

Heh.  Cool :-)

 > Since they don't seem to have gotten them message, I'm putting it back.
 > You can find it at http://www.freebsd.org/~wpaul/Sundance/st201.pdf.
 > If they don't like it, they can go to hell.

Got it, thanks :-)

I really ought to do something like this with my ... quite insane
collection of docs.

 > Note: there is one gotcha. On pages 50 and 51, the MACCTRL register
 > is described as being a single 32-bit register. However the register
 > summary on page 44 shows it as being two adjacent 16-bit registers.
 > The register summary page is correct: it actually is two 16-bit
 > registers. Trying to to treat it as a single 32-bit register doesn't
 > work. This means you need to divide up the bit definitions into two
 > 16-bit sets instead of one 32-bit set. Not a big deal really, but I got
 > really confused by this initially and it took me a while to figure out
 > where I'd gone wrong. They may have fixed this in later versions of
 > the manual. (Though I wouldn't bet real money on it.)

...thanks again :-)

        -- Jason R. Thorpe <thorpej@nas.nasa.gov>