From mboxrd@z Thu Jan  1 00:00:00 1970
From: jmk@plan9.bell-labs.com
Message-Id: <200006141335.JAA11665@cse.psu.edu>
Date: Wed, 14 Jun 2000 09:34:58 -0400
To: 9fans@cse.psu.edu
Subject: Re: [9fans] Compaq Prof. Wkstn. 5000 and Plan 9
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
Topicbox-Message-UUID: bc0b38de-eac8-11e9-9e20-41e7f4b1d025

	>I have a Compaq 5000 Professional Workstation with dual Pentium
	>Pro II CPUs and internal SCSI and IDE adaptors.

	>Solaris 8 installation proved problematic because Solaris could
	>not see my SCSI drive despite DOS and Windows 98 doing so.

	>Now it seems Plan 9 has the same problem.  Using Compaq's
	>inspect, it appeared the IDE controller was IRQ 15, and
	>the Network, SCSI controller and Matrox all shared IRQ 11.

	>Fiddling with things, I managed to get all devices onto seperate
	>IRQs, but pickdisk still fails with "No disk devices were found
	>on your system."

An aside about IRQs. On a uniprocessor x86 it usually makes sense to
arrange for PCI devices to have separate IRQs for efficiency reasons
but they should work when shared (modulo driver bugs).
However, on a system conforming to Intel's Multiprocessor Specification
it isn't necessary, the hardware and operating system software can and do
arrange for each device to have a separate line to the interrupt controller
and a separate interrupt vector; in other words, the IRQ value the card
reports is irrelevant.

	>Plan 9 appears to recognise the controller because it prints
	>dev A0 config 85C0 capabilities 0F00 mwdma 0103 dma 00000001 rwm 0
	>sd53c8xx: SYM53C875 rev. 0x03 intr=10 command=2000157
	>at boot time, and also "sd53c8xx: differential mode set" 16 times,
	>suggesting once for each target since it is a wide controller.

Do you know if it's really a differential controller and drive? The way
this is determined by the driver is only known to work on cards manufactured
by Symbios themselves (according to the comments in the driver). It's
possible differential mode is being detected incorrectly.

	>I have set debug=1 in the plan9.ini as the troubleshooting manual
	>suggests, but it didn't help much.

	>Pressing Control-R at boot time does not appear to generate
	>any extra debugging information for the SCSI probing.

	>Since my CDROM is IDE and my hard drive is SCSI, is plan 9 perhaps
	>finding the CDROM first and assuming my hard drive must be
	>IDE also?

No, that shouldn't matter.

	>The CDROM can be removed from the system, but there is no way
	>to disable the IDE controller on the motherboard. Setting
	>boot order in the BIOS setup and specifying the SCSI hard drive
	>did not help.

None of those things should make any difference either.


	>Any suggestions other than finding an IDE hard drive and connecting
	>it in place of the CDROM?

It's difficult to debug this at a distance without a working system.
Hooking an ATA drive up to get Plan 9 running might be an easier way to
debug the problem (hook it up to the primary IDE channel to make your
life easier).

Also, as was mentioned in another posting, the distributed bootstraps
don't deal with booting from SCSI discs; that should be in the next update.

--jim