*BSD News Article 16024


Return to BSD News archive

Newsgroups: comp.os.386bsd.questions
Path: sserve!newshost.anu.edu.au!munnari.oz.au!constellation!osuunx.ucc.okstate.edu!moe.ksu.ksu.edu!zaphod.mps.ohio-state.edu!math.ohio-state.edu!wupost!bigboy.sbc.com!news.mtholyoke.edu!news.smith.edu!sophia.smith.edu!jfieber
From: jfieber@sophia.smith.edu (J Fieber)
Subject: Re: Elusive SCSI problems (Adaptec 1542C)  <snarl!>
Message-ID: <1993May13.132848.18516@sophia.smith.edu>
Keywords: Adaptec 1542C, Quantum, SCSI, 386bsd
Sender: root@sophia.smith.edu (Operator)
Organization: Smith College
References: <1993May12.023959.20121@sophia.smith.edu> <1993May12.205857.8857@fcom.cc.utah.edu>
Date: Thu, 13 May 1993 13:28:48 GMT
Lines: 88

In article <1993May12.205857.8857@fcom.cc.utah.edu> terry@cs.weber.edu (A Wizard of Earth C) writes:
>In article <1993May12.023959.20121@sophia.smith.edu> jfieber@sophia.smith.edu (J Fieber) writes:
>>SETTING:
>>  A UM486V AIO motherboard --
>>   - 33MHz 80486dx
>>   - 6 ISA, 2 VL-bus slots
>>   - 256k external cache
>>   - local-bus IDE controller w/ connor 240 meg drive
>>   - floppy controller
>>   - 2 serial, 1 parallel, 1 game port
>>   - AMI bios
>
>HOW MUCH RAM?  You imply 8M; if >16, there will be problems.  Reduce to
>16M or below and retry.

That would be 8 meg.  Sorry I forgot to mention it.

>[ ... ]
>[ BUS ON TIMES ]
>>  Would some kernel hacker tell me how to adjust this properly?)
>[ ... ]
>>ADAPTEC CONFIG:
>
>This is where you diddle tyhe bus-on times.  Unless you have recompiled
>the kernel, you will have a hard time reducing bus-on time.  I believe
>that someone else had this problem (Nate?) and it's now working for him
>with about a 50% reduction in on time.

No.  The 1542C has no "user" adjustment for the bus on/off time.
DMA speed, yes.  There are no jumpers on the board aside from
disabling the floppy and setting the address.  I stuck a couple
lines of code in the aha1542.c file just before the bus speed
that *supposedly* set to set the bus on/off time.  The default
values are (according to the adaptec tech support person) 11 and
4 respectively.  I reduced the on time and increased the off time
but observed no change.  I certainly could be doing something
wrong though; I'm far from being a seasoned kernel hacker.

Also, if the problem is memory, rather than taking a performance
hit I'd squeal in the ear of the dealer and demand decent memory
since the machine is new.

>The problem in his case was that the DMA refresh wasn't getting done.
>You may want to jumper more wait states onto the motherboard to see if
>this will help.  If the wait states do the trick, you should be able
>to load and recompile with your now heavily-sedated machine.  If adding
>wait states doesn't help, then it'd not the bus-on time.

The wait-states are set in via software.  I've tried cranking
them up to the max setting.  No help.  :(  

Rebuilding a kernel isn't a problem as I have a functioning IDE
drive.

>Also: who is the manufacturer of your extern cache silicon?  I probably

I don't know without taking apart the machine but since I still
had the problem with it disabled I would suspect the cache would
not be the problem.  (Feel free to disprove me though...)

******

I've a question about SCSI geometry and disklabels.  When 386bsd
boots up, the aha1542 code reports a geometry.  Is that the
geometry I should use for the disklabel?  It certainly is
different that the geometry I used for the same drive when it
lived in my Amiga.

I also just noticed something strange in the disklabel.  If I
have partitions `a', `c', and `d', it reports that there are 4
partitions.  Is it counting the `b' that I didn't put in?  Must I
put a `b' partition in?   The only way I could label the Quantum
PD210S was to edit the one that it invented when it found none.
newfs most definately did not like what it found there.

I think I'll go through the comp.os.386bsd.* archives looking for
info on SCSI problems...

Finally, I thought that I was having problems with the tape drive
as well.  Pending further testing, that *may* prove to be false.
This makes me even more suspicious about geometry and disklabels.


-john

-- 
=== jfieber@sophia.smith.edu ================================================
======================================= Come up and be a kite!  --K. Bush ===