3Com 3CRWE73796B

[David Gibson]

On Fri, Apr 05, 2002 at 12:00:20AM +0930, Benjamin Close wrote:
> David Gibson wrote:
> 
> >On Tue, Apr 02, 2002 at 11:32:04AM +0200, Alex wrote:
> >
> >>Hello.
> >>
> >>I have a problem with a 3com card (3CRWE73796B).
> >>
> >>when I insert it, cardmgr loads the orinoco module and everything seems 
> >>ok,
> >>but when I try to issue a command to the nic I get an error like this:
> >>
> >># iwconfig eth1 essid any
> >>Error for wireless request "Set ESSID" (8B1A) :
> >>   SET failed on device eth1 ; Device or resource busy.
> >>
> >>and in the syslog appears this line:
> >>
> >>kernel: eth1: Error -16 shutting down Hermes chipset
> >>
> >
> >Ah yes, this is a problem we have with Symbol firmware.  We normally
> >issue an 'INIT' command to the firmware to reset the card, but the
> >Symbol firmware falls in a heap if the INIT command is issued more
> >than once.  We had a workaround which used a PCMCIA COR reset on these
> >cards instead but we seem to have accidently lost the bit which
> >activated that automatically in a reshuffle somewhere.  Oops.
> >
> Hmm, there was a commit recently to the FreeBSD tree that used this hack 
> to fix detection of the card. Might help you out...

Detection isn't the issue.  It's just that having detected the Symbol
firmware we forgot to apply the correct workaround.  Anyway this
message doesn't seem to give any information about what or how they're
detecting it.

> >In message: <200204031944.g33JiIr69197 at freefall.freebsd.org>
> >           Warner Losh <imp at FreeBSD.org> writes:
> >:   Modified files:
> >:     sys/dev/wi           if_wi.c if_wi_pci.c if_wireg.h if_wivar.h 
> >:   Log:
> >:   Four fixes from NetBSD:
> >:   1) Properly detect the Symbol based cards (The 3Com Airconnect and 
> >their
> >:      ilk) and only reset them *ONCE* ever.  This appears to make them 
> >work,
> >:      but more testing is needed.  The tests that would wedge up my 
> >machine
> >:      completely now appear to work, but I have not real access points
> >:      handy.
> >
> >I've just confirmed that this makes my 3Com AirConnect (aka 3CRWE737A)
> >work where before it wouldn't work at all with my access point.  Rumor
> >has it that there are a number of minor bugs (ahem compatbility
> >issues) with the Symbol firmware that the Linux driver works around in
> >various ways that aren't in this code (nor in NetBSD's nor OpenBSD's
> >versions).
> >
> >In fact, this message is posted using the '737A card that Bill Paul
> >destroyed at FreeBSDcon '99 and that I pieced back together.

-- 
David Gibson			| For every complex problem there is a
david at gibson.dropbear.id.au	| solution which is simple, neat and
				| wrong.  -- H.L. Mencken
http://www.ozlabs.org/people/dgibson