Re: [patch] ep(4) panic

Warner,

> : the way i understand it: this code is in the "probe" routine and when 
> : its called ifp structure was not allocated/setup yet. the "attach" 
> : routine will call "ep_attach" later that will allocate/setup ifp and 
> : read/set mac address (once again). so, the card works just fine.
> : 
> : it is interesting that my 4.x has different comment
> : 
> : $FreeBSD: /repoman/r/ncvs/src/sys/dev/ep/if_ep_pccard.c,v 1.12.2.3 
> : 2003/10/06 02:53:51 imp Exp $
> : 
> : 	/*
> : 	 * For some reason the 3c574 needs this.
> : 	 */
> : 	ep_get_macaddr(sc, (u_char *)&sc->arpcom.ac_enaddr);
> : 
> : 
> : perhaps the comment in -current should be changed as well? can anyone 
> : please shed some light on this?
> 
> The comment in -current is wrong.  The routine doesn't set the MAC
> address, but instead just reads it out of EEPROM.  For reasons I still
> don't know or understand, this still seems to be required.  I'll have
> to look into when I have an hour or two to kill.
> 
> The following patch works on:
> 	3CCFE547BT, 3C574-TX, 3CCE589EC, 3C589D-TP, 3C589C, 3C589B,
> 	3C562D/3C563D, 3CCFEM556
> 
> which covers almost all the cards that are obtainable these days.  It
> doesn't work with:
> 
> 	3C1			(we've never worked with this card)
> 	3C562 or 3C562B/3C563B	I think that there's an address line
> 				work around needed for thse two cards.
> 				5.x doesn't seem to work with them
> 				either.
> 
> So I think it is good/safe to commit.  I'll ask the RE if people here
> test it.

thank you for spending your time and looking into this. your patch is 
essentially the same as my original patch. i expect this patch to work 
just fine. i will double check it on monday.

thanks,
max