From owner-freebsd-hackers Fri Mar 24 10:16:38 1995 Return-Path: hackers-owner Received: (from majordom@localhost) by freefall.cdrom.com (8.6.10/8.6.6) id KAA14896 for hackers-outgoing; Fri, 24 Mar 1995 10:16:38 -0800 Received: from cs.weber.edu (cs.weber.edu [137.190.16.16]) by freefall.cdrom.com (8.6.10/8.6.6) with SMTP id KAA14890 for ; Fri, 24 Mar 1995 10:16:37 -0800 Received: by cs.weber.edu (4.1/SMI-4.1.1) id AA10007; Fri, 24 Mar 95 11:09:14 MST From: terry@cs.weber.edu (Terry Lambert) Message-Id: <9503241809.AA10007@cs.weber.edu> Subject: Re: Fix for 3C509 (ep) driver To: babkin@hq.icb.chel.su (Serge A. Babkin) Date: Fri, 24 Mar 95 11:09:14 MST Cc: davidg@Root.COM, mmead@goof.com, freebsd-hackers@FreeBSD.org In-Reply-To: <199503242252.RAA01136@hq.icb.chel.su> from "Serge A. Babkin" at Mar 24, 95 05:52:52 pm X-Mailer: ELM [version 2.4dev PL52] Sender: hackers-owner@FreeBSD.org Precedence: bulk > I have fixed another bug in 3C509 driver that prevents FBSD from > network booting through other connectors than BNC. Now it will > use transceiver specified in the EEPROM instead of select based > on interface flags. >From my reading of this, this means that it's better for remote boot from the card, since now it's possible to designate the interface, but it won't resond to specific use of the "link" flags? I think specific use of the link flags is still desirable; I have more than one machine that I switch between nets using the link flags to do the job because I don't want to crap on the main wire with experimental code. Perhaps "link0" should be the card default and "link1", etc. should pick specific card connectors? Also: for card that support dual interfaces, it's frequently true that it's possible for the probe routine to pick the interface correctly and automatically by picking an interface that knows about being disconnected from the net first, followed by the one that doesn't know if it's on the net or not. Clearly, this wouldn't work for my machine with multiple wires hooked up, but for the typical installation, it could cause the ethernet driver to simply "do the right thing". Does the idea of having link type flags need to be revisited? link description ----- ---------------------------------------------- 0* Do the right thing based on connectedness 1 Use the EEPROM settings 2 AUI 3 BNC 4 ... * indicates default behaviour It almost makes sense to go away from the current numeric identifiers, as ugly as that might be... Finally: Many cards support determining the IRQ and memory addresses dynamically based on knowing the I/O address to INB from. This doesn't seem to be in there for a lot of the cards known to do this? Terry Lambert terry@cs.weber.edu --- Any opinions in this posting are my own and not those of my present or previous employers.