I had a problem with booting the netbsd kernel on a 
dell 2600 laptop because it could not recognize the
pcmcia slot. I fixed it by recompiling the kernel so
that it didn't look at the pcmcia slot during bootup.
But IMHO, if the peripheral is non-critical, the
bootup procedure should not be aborted.[that way we
won't be forced to use the tiny image].
thanks
-kamal

--- Richard Rauch <rauch@rice.edu> wrote:
> On Thu, 10 Oct 2002, Richard Rauch wrote:
> 
> > Over the coming weekend, I may have time to fiddle
> with getting 1.6 to
> > work with my laptop's ethernet card.  (1.5 worked
> with a customized kernel
> > config; 1.6 identifies the card but then gives up
> and says it can't match
> > the vender code.  See kern/18359 for more
> details.)
> 
> I put a little time into it, now.
> 
> It appears to fail in
> ne_pcmcia_ax88190_set_iobase(psc), from the
> if_ne_pcmcia.c file (called from ne_pcmcia_attach(),
> as well as from
> ne_pcmcia_enable().  After glancing through
> ne_pcmcia_av88190_set_iobase(), it appeared that
> everything worked with my
> card UP TO the very last check, where for my card
> the check failed and
> {rv} was never cleared to 0.
> 
> So, I made that {rv = 0;} statement (the second to
> last thing before
> either of the fail: labels) unconditional.  Now, my
> PCMCIA MELCO ne0
> ethernet card works with NetBSD 1.6 (slightly
> hacked).
> 
> I do not know why the code does the check; I haven't
> yet determined if the
> check fails on the == relation or on the != check in
> that final if().  I
> know only that, at least for my card, it appears
> that the final check
> prior to setting {rv} is superfluous (and fatally
> wrong).
> 
> 
> I'll switch my laptop over to this pseudo-1.6 kernel
> for a few days, and
> if all goes well (and no one tells me why I really
> want that {rv = 0;}
> line to fail in 1.6 without comparable lossage in
> 1.5.x), maybe I may even
> get my laptop upgraded to ~1.6 this weekend.  (^&
> 
> (If anyone feels like looking, it's less than a
> dozen lines from the
> bottom of /usr/src/sys/dev/pcmcia/if_ne_pcmcia.c. 
> Similarly, if anyone is
> using an AX88190 type PCMCIA ethernet card and can
> conform that 1.6 *does*
> work for them (so the check isn't simply
> reversed/breaking-the-driver), it
> might be nice to hear.  (^&  I don't know anything
> about the hardware, or
> really the framework in which the driver lays, so I
> don't know if my hack
> would help/hurt others.  But, it appears absolutely
> essential to using
> this card in this laptop.)
> 
> 
> If I want to attach a workaround-patch to my PR, how
> do I do that?  (The
> last time I followed someone's advice and got a
> slightly-nastygram from
> someone about not following proper procedure.  Can I
> just reply to the
> message that GNATS sent me for the PR?)
> 
> 
> Thanks in advance.
> 
> 
>   ``I probably don't know what I'm talking about.''
> --rauch@math.rice.edu
> 


__________________________________________________
Do you Yahoo!?
Faith Hill - Exclusive Performances, Videos & More
http://faith.yahoo.com