Date: Tue, 28 Nov 1995 11:21:47 +0100 From: se@zpr.uni-koeln.de (Stefan Esser) To: "Marc A. Mercier" <mercierm@bose.com> Cc: hackers@freebsd.org Subject: Re: Better scsi and e-net support... Message-ID: <199511281021.AA14947@Sysiphos> In-Reply-To: Michael Smith <msmith@atrad.adelaide.edu.au> "Re: Better scsi and e-net support..." (Nov 28, 6:05)
next in thread | previous in thread | raw e-mail | index | archive | help
On Nov 28, 6:05, Michael Smith wrote: } Subject: Re: Better scsi and e-net support... } Julian Elischer stands accused of saying:> } The PCnet chip is a LANCE, and should be supported by the lnc driver. There } was some talk a while ago about the Compaq PCI chipset - something about } "fundamental brokenness", but I'm also led to believe that a solution } was finally reached. (Check with the PCI gurus on that one) Booting with "-v" will let you see the Ethernet chip's I/O address. It is 0x7000, if I remeber right. If you boot with "-c", enter the port address for "lnc0" and assign the correct IRQ, then you should be able to use the Lance part of the Combo chip. } IIRC, Walnut Creek will upgrade you to the latest version (2.1) free of } charge, especially given the circumstances. You may also need a custom } boot floppy and kernel to deal with the PCI unhappiness if you're on a } Compaq. The boot probe got more and more pedantic over time, to make sure that non-PCI boards are dealt with correctly (i.e. that PCI probes are not performed on non-PCI motherboards, since they might become upset :). But then the Compaq's PCI bus was not detected anymore, since they violate certain rules of the PCI spec, which are supposed to let a system, probe for PCI and configuration details. The current code contains special cases for the COmpaq chip sets, and you should be able to access the PCI bus from both 2.1R and FreeBSD-current. } As for the SCSI side of things; I'd suggest investing another $40 or so } on an NCR-based PCI SCSI controller. I can't see the AMD part being } supported until someone comes up with a card with it on, and someone esle } gives such a card to the PCI driver developers. Wolfgang Stanglmeier had the AMD Combo Chip data book, and considered writing a driver for it at some time. But when he found, that the AMD requires a host interrupt handler to deal with disconnect, we decided to not put any effort into this. The AMD offers far inferior performance at much higher CPU load than the NCR, and it would have been silly to invest hundreds of hours of work into support of that controller chip, which could be replaced by a superior solution at $50 at that time ... BTW: AMD themselves tell you not to use that Combo chip in new designs ... Regards, STefan -- Stefan Esser, Zentrum fuer Paralleles Rechnen Tel: +49 221 4706021 Universitaet zu Koeln, Weyertal 80, 50931 Koeln FAX: +49 221 4705160 ============================================================================== http://www.zpr.uni-koeln.de/~se <se@ZPR.Uni-Koeln.DE>
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199511281021.AA14947>