Return to BSD News archive
Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!hookup!news.moneng.mei.com!howland.reston.ans.net!vixen.cso.uiuc.edu!newsrelay.iastate.edu!news.iastate.edu!ponderous.cc.iastate.edu!michaelv
From: michaelv@iastate.edu (Michael L. VanLoon)
Newsgroups: comp.os.386bsd.questions
Subject: Re: AHA1542 hanging with FreeBSD 1.1beta
Date: 13 Apr 94 01:21:44 GMT
Organization: Iowa State University, Ames, Iowa
Lines: 76
Message-ID: <michaelv.766200104@ponderous.cc.iastate.edu>
References: <Co3MBA.s3@pegasus.com> <Co3zGC.64o@usenet.ucs.indiana.edu> <michaelv.766120335@ponderous.cc.iastate.edu> <Co5pu9.AHG@usenet.ucs.indiana.edu>
NNTP-Posting-Host: ponderous.cc.iastate.edu
In <Co5pu9.AHG@usenet.ucs.indiana.edu> ahabig@bigbang.astro.indiana.edu (Alec Habig) writes:
>michaelv@iastate.edu (Michael L. VanLoon) writes:
>>In ahabig@bigbang.astro.indiana.edu (Alec Habig) writes:
>>>richard@pegasus.com (Richard Foulk) writes:
>>>>I have a 486/66 with an AHA1542CF SCSI card that locks up about
>>>>once a day with the disk activity light on.
>>>My problem is similar to Richard's, but doesn't leave the HD light on.
>Yes. I have an IDE controller.
>>This is a description of *IDE* problems. *IDE* and *SCSI* are
>>completely unrelated. This has no relevance to the above mentioned
>>problem.
>The FAQ description is indeed of an IDE problem, with the wd drivers, not the
>sd drivers. However, the description fits Richard's problem _exactly_. Drive
>light and all. My drive light is not stuck on, even though I have the same
>hardware as the FAQ example talks about. The problems seem related, even
>though the drive light differences and hardware differences. If it looks like
>a duck, walks like a duck, and quacks like a duck, then information on a duck
>is likely to be rather useful stuff.
Well, yes, on the surface they all look like birds. But the
underlying structure is very different.
On IDE, the CPU does busy-loop copies directly to/from the IDE drive
via the IDE "interface". Therefore, if there's a problem, it's
probably related to the software and/or IDE drive. The resultant
problem being that the CPU sits and waits in a software loop for the
drive to become unbusy, but the CPU never receives the interrupt.
This is usually caused by poorly designed or slow IDE drive hardware.
This is fixable by writing software that compensates for poor IDE
designs.
SCSI is different. The CPU simply writes instructions of what the
SCSI device (hard drive) is supposed to do into a block of memory.
Then the controller picks it up, executes the commands without
processor intervention, writes the results back into a memory
structure, and interrupts the CPU when it's done. The CPU never
touches the drive. Hence, if the system hangs with the drive light
stuck on, it is not between the CPU and drive, it has to be between
the drive and SCSI controller; most likely because of bad cables or
termination. If the SCSI-bus hang were to happen while the SCSI
controller had the address/data bus locked doing DMA, the CPU would
never even be able to get access to memory to unfreeze things, because
it would be in a never-ending wait state. This cannot be corrected in
software since the CPU can't run while the SCSI controller is hung in
DMA and the bus is locked; it is a hardware problem.
The symptoms may be similar, but the key components are very
different.
>I repeat my original question :
>Does anyone have the kernal code hack mentioned in the FAQ, so that I can try
>it out without the danger of someone who is not familiar with the code (myself)
>hacking something rather fundamental (disk device drivers in the kernel).
NetBSD-current has the IDE drive problems pretty much defeated. There
are still some IDE drives that refuse to cooperate, I'm told, but I
haven't seen anyone post about them recently. Charles Hannum seems to
have a very good understanding of the problem. I've also heard that
FreeBSD-current has licked most of the IDE problems. You might try
seeing what the wd.c drives look like in one or the other current
distributions.
--
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Michael L. VanLoon Iowa State University Computation Center
michaelv@iastate.edu Project Vincent Systems Staff
Free your mind and your machine -- NetBSD free Un*x for PC/Mac/Amiga/etc.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -