[comp.sys.tahoe] my machine crashes.

phillips@CS.BUFFALO.EDU (Gretchen Phillips) (10/27/87)

I have isolated a problem with my machine that I would appreciate
some thoughts on fixing. I have a simple program that should get
a segmentation fault when it accesses reserved memory. However,
instead of performing the error detection and handling, the machine
just crashes. I would appreciate some advice about how to isolate
the real culprit. This same program works fine (segmentation faults)
on the tahoe at albany so I think it is a local problem. 

I compiled the program on the albany machine and brought the binary
back and that still crashed my machine so it seems to be in the
error handling.

I am including the program here but am omitting the reserved address
for security reasons. If you want an address to put in there that
crashes my machine you can contact me directly.

	main ()
	{
	asm ("movl  $0x????????,r0");
	asm ("cvtbl 4(r0),r0");
	}

Thanks in advance,
	Gretchen

james@CSD4.MILW.WISC.EDU (Jim Lowe) (10/27/87)

> From: Gretchen Phillips <phillips%cs.Buffalo.EDU@csd1.milw.wisc.edu>
> To: info-tahoe@gort
> Subject: my machine crashes.
> 
> I have isolated a problem with my machine that I would appreciate
> some thoughts on fixing. I have a simple program that should get
> a segmentation fault when it accesses reserved memory. However,
> instead of performing the error detection and handling, the machine
> just crashes. I would appreciate some advice about how to isolate
> the real culprit. This same program works fine (segmentation faults)
> on the tahoe at albany so I think it is a local problem. 
> 
> I am including the program here but am omitting the reserved address
> for security reasons. If you want an address to put in there that
> crashes my machine you can contact me directly.
> 
> 	main ()
> 	{
> 	asm ("movl  $0x????????,r0");
> 	asm ("cvtbl 4(r0),r0");
> 	}
> 
We had a simular problem about 3 weeks ago.  We sent a tape into Unisys
and to CCI.  CCI responded with a new version of the microcode.  I can
dig up the CCI HRR number for you if you want it.  Unisys hasn't
responded yet so I can't give you a Unisys SUR number.

What I currently have is 3 versions of the microcode.  One with the EMUL
bug in it, one that crashes with segmentation fault (without the EMUL
bug), and one that seems to work fine.

My guess is that albany is still running the microcode with the broken
EMUL instruction.  We had to run with that microcode until CCI sent us
the new stuff last week.

I am amazed that nobody else has run into this problem.  Anyways, CCI
has a fix for it.


	- Jim 

Internet:	james@csd4.milw.wisc.edu	Uucp:	ihnp4!uwmcsd1!james
Csnet:		james@uwm			Phone:	+1 (414) 229-6431