Re: Intel EtherExpress PRO/100 Server Adapter

Torbjorn Lindgren (tl@fairplay.no)
Fri, 5 Mar 1999 11:58:46 +0100 (CET)


On Fri, 5 Mar 1999, Marek Habersack wrote:
> On Thu, 4 Mar 1999, Deepak Saxena wrote:
>
> > The Pro/100 Server Adapter is not an I2O device. The 960 is just used
> > to run the firmware for the card and do buffer mnagement functions
> > under NT and Netware BorderManager. The Pro100 driver won't work since
> > the driver has to go through the firmware and doesn't talk directly
> > to the 82558 MAC chips is my guess.

> But the card works under M$ Winblows with the standard EEPRO/100 driver,
> so I presume it should work with the standard Linux EEPRO100 driver.

They could easily have stuffed drivers for *ten* wildly different cards
into the driver, if they wished. In fact, given the way their marketng
handles the different cards they are probably better of using a single
driver (so you can switch/upgrade card more easily).

It's probably not a big difference in Win95/98 (it notices that something
has changed and reinstalls everything anyway), but handled correctly it
could simplify it for WinNT, which is their target anyway for the high-end
cards!

Given their claims it has to be something more in it than in the ordinary
cards, the question is if you can ignore it completely and use it as an
ordinary card. It would be a minor benefit for their own driver, but on
the other hand they would risk that people used OLD drivers (that perhaps
doesn't support the new features) and didn't get any of the benefits, and
then complained that the card wasn't helping!

Not god PR....

To take another example I'm more familiar with, the 3Com Etherlink III/
IIIXL cards (3c50x, 3c59x, 3c90x & 3c90xB). Windows has three different
driver sets for these, while Linux has two and Solaris has one IIRC.

(AFAIK there's a total of four different, somewhat related, operating
modes on these cards, but the fourth is only supported by 3c90xB cards
and not that much different from the third mode, and I don't think anyone
uses that one. So to get good performance for all cards you need to
support all three (four?) modes, which is done with one, two or three
driver depending on which OS you are using).

More to the point, the 3c90x has a mode where it's "mostly" compatible
with the 3c59x, but making that work on Linux AFAIK required documentation
on the 3c90x (nowadays the driver uses the native mode instead).

And the 3c90xB dropped that backwards support, and added a new mode, but
was supposed to be fully compatible with the 3c90x apart from that...
Fine, except that it didn't work that way, it's only the latest 3c59x.c
drivers that NWay speed negotiation works for example (0.99H and later
IIRC). Not to mention the debacle with Win98 drivers powering down the
3c90xB cards so the driver couldn't revive them, this also requires a
fairly recent driver to handle (2.2.x has 0.99H which should handle both
these).

In short, how the Windows driver looks and works doesn't have to have any
bearings on how other drivers works. Especially when the driver is made by
the manufacturer, which might know a LOT more than what they are telling
others, they could easily have advance warnings on things that MIGHT
change, and have workarounds implemented in the shipping driver just in
case...

-- 
Torbjörn Lindgren
Network Manager, FairPlay International AS
E-mail: tl@fairplay.no

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/