On Tue, Jul 13, 2004 at 01:58:50AM +0000, Charles M. Hannum wrote:
> On Tuesday 13 July 2004 01:21, Richard Rauch wrote:
> > For some reason, I had the impression that this didn't affect PCMCIA NE2000
> > cards.  Strange brain-glitch on my part, I guess...
> >
> >
> > Anyway, I have a PCMCIA ne0 in my laptop.  It has a small problem which
> > seems to be unaffected by whatever Charles did.  The card works (with a
> > local hack) as before.  If I remove the local hack, the card does not
> > work.
> >
> > The hack is truly a hack: It bypasses a test condition that presumably
> > is useful for some cards, but which my card fails.
> 
> Perhaps you could tell us what that "test condition" is?

The very last if() test of if_ne_pcmcia.c.  About 10 lines
up from the bottom of the file in my present copy.

        if ((last_liobase == psc->sc_pcioh.addr)
            || (last_liobase != new_liobase))
                rv = 0;

Add an (unconditional) {rv = 0;} after that, or fudge the
test to always succeed, and the card works.

See PR kern/18359 for more information, including dmesg for
card details.

The overruns I mentioned in the PR go away if I throttle the card
to 10Mbps (I was advised that PCMCIA can't handle 100MBps).  Other
than that, and the required hack, it works well for me.


-- 
  "I probably don't know what I'm talking about."  http://www.olib.org/~rkr/