Date: Thu, 10 Apr 1997 10:57:04 +0300 (EEST) From: Andrew Stesin <stesin@gu.net> To: Stefan Esser <se@freebsd.org> Cc: hackers@freebsd.org, scsi@freebsd.org Subject: Re: 2 PCI busses, 2 AIC chips, 2.2.1. Howto ? Message-ID: <Pine.BSF.3.95q.970410104207.2999k-100000@trifork.gu.net> In-Reply-To: <19970409130200.12143@x14.mi.uni-koeln.de>
next in thread | previous in thread | raw e-mail | index | archive | help
Hello Stefan, On Wed, 9 Apr 1997, Stefan Esser wrote: > > Sorry for the late reply, I just returned from a vacation ... > Stefan, thanks for replying at all, don't make my cheeks shine red with the above disclaimer :) > > The box: Intel XXpress-<smth>, 2 AIC7870 chips onboard, EISA bus present. > > Is this a Pentium or PPro system ? "Normal" Pentium. 1 CPU module, P5/166. > > What I am missing? Any comments and explanations are appreciated! > > Seems you have some way of obtaining the boot message log. > Could you please provide me with VERBOSE boot messages ? > (Eenter "-v" at the "Boot: " prompt.) I was a bit ill for last few days, wasn't able to do anything creative then... now I feel myself much better, will do this todays evening. > The problem you are facing comes from the fact, that PCI > specifies a single bus directly attached to the CPU, and > a tree structure of seconfary buses connected to it. > Since this limits the number of low-latency slots to at > most 5, server machines often come with more than one bus > directly attached to the CPU. Seems this is exactly what Intel did... > But there is no standard way > of telling, how many buses are actually connected to a CPU > (1 is assumed), and while the bus numbers behing PCI to PCI > bridges can be determined in a chip-independent way, this > is not possible in the case of multiple CPU to PCI chips. > Device specific code has to be added for each of them. What I'm able to tell you for sure is that RedHat Linux 4.1 installation floppies are seeing AICs (and disks) and using them; while FreeBSD-2.2.1 doesn't. > (The information on the number of buses can be queried by > calling the PCI BIOS, but I'd prefer, if I not have to, > for a number of reasons ...) > > I need more information on the chip set used on that mother > board. The Orion should be supported out of the box, but if > this is a system with multiple PCI buses directly attached > to the CPU based on some other chip set, a small addition > to the PCI probe code will be required. I'll set up a serial console and send a boot -v for this box to you this evening. > > Regards, STefan > Best regards, Andrew Stesin nic-hdl: ST73-RIPE
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.3.95q.970410104207.2999k-100000>