From owner-freebsd-bugs Mon Nov 27 13:44:57 1995 Return-Path: owner-bugs Received: (from root@localhost) by freefall.freebsd.org (8.6.12/8.6.6) id NAA01273 for bugs-outgoing; Mon, 27 Nov 1995 13:44:57 -0800 Received: from Sysiphos (Sysiphos.MI.Uni-Koeln.DE [134.95.212.10]) by freefall.freebsd.org (8.6.12/8.6.6) with SMTP id NAA01262 for ; Mon, 27 Nov 1995 13:44:41 -0800 Received: by Sysiphos id AA03231 (5.67b/IDA-1.5 for bugs@freebsd.org); Mon, 27 Nov 1995 22:41:54 +0100 Message-Id: <199511272141.AA03231@Sysiphos> From: se@zpr.uni-koeln.de (Stefan Esser) Date: Mon, 27 Nov 1995 22:41:53 +0100 In-Reply-To: iain@nwpeople.demon.co.uk (Iain Baird) "NCR815 problem" (Nov 27, 18:26) X-Mailer: Mail User's Shell (7.2.6 alpha(2) 7/9/95) To: iain@nwpeople.demon.co.uk Subject: Re: NCR815 problem Cc: bugs@freebsd.org Sender: owner-bugs@freebsd.org Precedence: bulk On Nov 27, 18:26, Iain Baird wrote: } Subject: NCR815 problem } I have just installed an NCR-815 PCI SCSI controller in my PC. } When I boot 2.1.0-951026-SNAP (off IDE) I get the following error: } } pci_map_mem failed: device's memrange 0xc8000-0xc80ff is } incompatible with its bridge's memrange 0x4000000-0xffffffff } } After this the driver still detects sd0 correctly, but I'm unable } to access it in any way. } } Occasionally (with cache enabled) the above error is followed by: } } CACHE TEST FAILED: reg dstat-sstat2 readback ffffffff. } CACHE INCORRECTLY CONFIGURED. } } Disabling the cache doesn't prevent the pci_map_mem error. } } What does all this mean? Do I need a new motherboard? } Have I got a duff NCR815? Thanks in advance for any help. Hmmm, this is a very surprising bug report :) Well, seems your system violates one of the assumptions of the PCI code: The PCI devices are expected to be mapped to addresses beyond the 1st Gigabyte of address space. This used to be treated as fatal, but I choose to make it a warning message, when I found that some Compaq machine behaved just the way your system does ... Your system uses much lower addresses: 0xc0000 is an address below 1MB, i.e. in the ISA peripheral address space. This ought to work, provided your system doesn't try to cache accesses to the 640KB to 1MB address range, and that you didn't turn on copying of BIOS ROMs into RAM at those addresses. Could you please verify the BIOS setup ? } Probing for devices on the PCI bus: } chip0 rev 4 on pci0:0 } chip1 rev 3 on pci0:2 } ncr0 rev 3 int a irq 15 on pci0:3 ^^ I'd rather avoid IRQ 15 in a system that also got IDE drives. Haven't heard that it actually caused problems, but I'd be afraid that IDE might want to have 14 and 15 for its own use (and might for example install interrupt handlers). } pci_map_mem failed: device's memrange 0xcc000-0xcc0ff is incompatible with its bridge's memrange 0x4000000-0xffffffff } ncr0 waiting for scsi devices to settle } (ncr0:0:0): "SEAGATE ST31250N 0006" type 0 fixed SCSI 2 } sd0(ncr0:0:0): Direct-Access } sd0(ncr0:0:0): FAST SCSI-2 100ns (10 Mb/sec) offset 8. } 973MB (1994037 512 byte sectors) } vga0 rev 0 on pci0:4 } pci0:5: vendor=0x1c1c, device=0x0001, class=old (misc) [no driver assigned] ^^^^^^ any idea, what kind of device that is ? } ncr0 rev 3 int a irq 15 on pci0:3 } pci_map_mem failed: device's memrange 0xcc000-0xcc0ff is incompatible with its bridge's memrange 0x4000000-0xffffffff } CACHE TEST FAILED: reg dstat-sstat2 readback ffffffff. The cache test makes the CPU and the NCR write memory, and verifies, that they get the correct results reading back the values written by the other one. If this test fails, then the NCR could not reliably access the command queue written by the host CPU, and the driver aborts at this point, since proceeding might wreck your data. But this is most likely a secondary effect, caused by the use of an unsuitable attach address, which has been choosen by the PCI BIOS. If you have copying of BIOS ROMs to DRAM enabled in the BIOS, then please try without ... If this doesn't help, then send **verbose** boot messages, i.e. enter "-v" at the "Boot: " prompt. This will make the log contain additional PCI setup information ... You will be able to use that 815 controller card, but it might be necessary to build a custom kernel. 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