Date: Wed, 14 Feb 1996 00:14:00 +0100 From: se@zpr.uni-koeln.de (Stefan Esser) To: Richard J Kuhns <rjk@sparcmill.grauel.com> Cc: freebsd-stable@freebsd.org Subject: Re: One problem && one question Message-ID: <199602132314.AA14953@Sysiphos> In-Reply-To: Richard J Kuhns <rjk@sparcmill.grauel.com> "Re: One problem && one question" (Feb 13, 14:27)
next in thread | previous in thread | raw e-mail | index | archive | help
On Feb 13, 14:27, Richard J Kuhns wrote: } Subject: Re: One problem && one question } Justin T. Gibbs writes: } > >Justin T. Gibbs writes: } > ><Buslogic id info> int a irq 10 pci 0:13 } > > mapreg[10] type 1 addr 0x00006000 size ffff0004 } > >bt0: Invalid base address } > } > boot -v with the old kernel and then use dmesg to retrieve the info. } > } } That works fine for any ``working'' kernel; apparently none of the boot } messages from a non-working kernel end up in the circular buffer dmesg } uses. IE, if I boot-working then boot-nonworking then boot-working and run } dmesg, I get the messages from the 2 boot-working kernels only. } } If that's what you were interested in, here's the most recent: } pci0:13: Bus Logic, device=0x1040, class=storage (scsi) [no driver assigned] } map(10): io(6000) Well, seems the PCI code does find the BT946C, but just doesn't know that it knows about it (because a insuitable PCI ID has been specified). } bt0 at 0x330 irq 10 on isa This gives the ISA code a chance to initialize the chip, which it could not have done else :) } > >I'm afraid I won't be able to test quite so often today; the machine has to } > >do some Real Work (TM). } I tried it; I'm afraid it made no difference. By the way, here's one more } line from the (failing) pci-probe: } } bt0: uses 4 bytes of I/O space from 6000 upto 6003 That's not surprising. But that the Buslogic responds to the ISA probe at 0x330 but offers only the 0x6000 port mapping in the PCI config space registers, that's bad ... If you have the correct PCI ID in the bt9xx.c file now, then I'd expect the PCI probe to be able to initialize the card at 0x6000 (I guess it will respond there, too!). There after the ISA probe may still be able to access it at 0x330 (unless the card disables that address, if the PCI region has been accessed at least once ...). If the card actually is found twice at two different port addresses, once as ISA and once as PCI device, then bad things will happen for sure ! Could you please check the bt0 attach message in the PCI scan, and look at the ISA messages whether bt0 appears there, too. I.e. the two lines I did not remove from the quoted boot message log you sent ... Seems the chip (or its BIOS) is severely broken, if it does not announce it's ISA compatibility register range through the mapping registers. It would be hard to compensate for this, unless the ISA compatibility range was in fact disabled after a PCI announced register has been used for the first time. 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?199602132314.AA14953>