[SunRescue] Problem with Weitek PowerUP

[William Barnett-Lewis]

A few nights ago I had a problem with my IPX after I upgraded it with a
Weitek PowerUP. No one here replied though I did get the answer below.  It
appears that Solaris and NetBSD stress the cache in very different ways...

I'm posting it here as a way to spread the knowledge as it _is_ an unusual
answer and I'd like to share it.

William

> This in fact appears to be the case here as well. I have a second IPX. I
> swapped the cache chips on the two motherboards and reloaded NetBSD 1.5 from
> CDROM. At this time I am running successfully. I will probably begin a make
> install on something larger (Emacs say) from the packages system to give a
> better stress test. However, the fact that X does not cause a panic instantly
> leads me to expect that the difficulty is solved.
> 
> My thanks to all who replied as all of the messages taught me something
> valuable. 
> 
> William
> 
> 
> on 2/19/01 4:14 AM, Petter Skott at lps at ludd.luth.se wrote:
> 
>> 
>> I've seen a similar problem on a IPX when upgrading to a PowerUP chip
>> Solaris worked just fine but NetBSD paniced frequently.
>> It turned out that the cache chip was broken so i replaced it whith a chip
>> from another IPX that had a broken motherboard.
>> 
>> /Petter
>> 
>> On Sun, 18 Feb 2001, William Barnett-Lewis wrote:
>> 
>>> Hello, 
>>> 
>>> I was running NetBSD/sparc 1.5. On Friday I upgraded my IPX with a Wietek
>>> PowerUP chip. Immediately NetBSD stopped working correctly. On attempting
>>> the startx command I received either an illegal instruction or alignment
>>> fault panic. Loading Solaris 7 runs fine, if with the performance penalty
>>> that drove me to BSD in the first place... :'p.
>>> 
>>> In NetBSD I booted up to the console prompt (not running XDM), when I logged
>>> in as my user account and typed startx.  I got an illegal instruction panic
>>> or an alignment fault panic. I've made sure that the chip is seated properly
>>> and nothing else was different from a perfectly running system before. As
>>> root it fired up fine... right up until I tried to open the system console
>>> (which _is_ set to open automatically for my user account...hmm...). After
>>> that, so much for root. It's blew out on all attempts after that.  If I
>>> tried to boot from my Wasabi CDROM, I would still get a panic (of the
>>> aforementioned types) before it finishes loading the system install utility.
>>> 
>>> My understanding is that the PowerUP is completely binary compatible,
>>> however I was running Window Maker as a window manager. Could my local
>>> compile of that have left some odd code that the Weitek was barfing on?
>>> 
>>> I searched the archives somewhat (is there a search utility?) but didn't
>>> find anything on this.
>>> 
>>> Aside from ripping my hair out and sticking to Solaris, I don't see an
>>> answer. Thanks in advance to any help anyone can help.
>>> 
>>> William
>>> 
>>