(Ken - the part of this letter that applies to you starts with the
paragraph that reads "I know the fact that...")

On 9/2/97 at 10:46 PM -0400, you wrote:

> Can someone with a Q610 please run the command "dm f9800000 40"
> and send me the output?

 Displaying memory from f9800000
  F9800000  FFFF F008 FFFF F000  FFFF F100 FFFF F71E  ................
  F9800010  FFFF F030 FFFF F030  FFFF F031 FFFF F001  ...0...0...1....
  F9800020  FFFF F000 FFFF F000  FFFF F000 FFFF F600  ................
  F9800030  0000 0000 0000 0000  FFFF FFFF FFFF FFFF  ................

Note: The periods were all non-ASCII bullets, so I replaced them with...
periods.

> Alternatively, could someone with a Q610 please tell me if they've tried
> my Quadra SCSI improvements and whether they've worked.

If your kernel is ftp.netbsd.org/NetBSD/arch/mac68k/kernels/test.gz, then I
gave it a whack yesterday, and it gave me the same error that
GENERIC[SBC]-32 and -38 gave me. I'll retype the relevant bits here:

esp0 at obio0: address 0x81f000: NCR53C96, 16MHz, SCSI ID 7
scsibus0 at esp0: 8 targets
esp0: SCSI bus parity error        (repeats twice)
data interrupt, but no count left.esp0: SCSI bus parity error
(repeats roughly 17 times)

My system has a 68LC040, but I don't think this is the problem, as with
older kernels I was able to boot with a lot more success, even though I got
the expected segfaults and stuff.

FWIW, I'm running a Quantum Fireball 2.1 gig drive. Other items on my SCSI
bus are a Microtek scanner and an NEC CD-ROM drive. Thazzit. :)

I know the fact that I have a 68LC040 makes my system less useful for this,
but I'd be happy to test any and all kernels for you. My SE/30 is my
primary NetBSD machine, so it doesn't matter how often the Quadra gets
trashed. (Well... As long as the trashing stays within the NetBSD
partitions! I'd cry if the Mac side were trashed.)

Once it looks like there's a working kernel again, I'd love to throw myself
at the FPE problem. (Going my the SoftwareFPU docs, my 68LC040 is one of
the later, less buggy variety.) As soon as it looks like I have a stable
kernel source on which to build, I'll hit up Ken Nakata for ideas, since
the general wisdom is that he knows what's wrong. Heck... I have all the
free time I need. I can't seem to find employment, so I might as well do
something useful with my time. I could conceivably put in sixteen hours a
day on the problem and still leave myself abundant time for sleeping.
<maniacal grin> Maybe that's just what I'll do.

Hm. Interesting. I just popped in a new kernel (built yesterday from the
previous night's sup) and the same "data interrupt" error message came up.
This time, however, I became aware of something that was probably happening
before as well - the light on my CD drive was going berserk. I turned off
the CD drive and tried again, and VOILA, I booted. Things even seemed a tad
more stable than last time I tried. For example, I could get through a
session with vipw without getting a segfault. I guess I'd ought to write to
Ken to see if he can give me pointers. BETTER YET... I'll just cc: this to
him.

Ken: Can you point me in the right direction? You've said before that you
know what's wrong with FPE, but you haven't had free time. I, conversely,
have NO CLUE what's wrong with it, but I have more than enough free time to
dedicate to making the beast work, as well as a nearly ideal setup to do so
- a working NetBSD machine on which I can compile kernels, and a
semi-working 68LC040-based machine that's all set up to go, excepting the
funky FPE problem.

Later...

--
        Mason Loring Bliss    /\    mason@acheron.middleboro.ma.us
      PGPKeyID: 0x25B3D5B5   /()\   awake ? sleep() : dream();
<barbaric>YAWP!</barbaric>  /    \  http://www.webtrek.com/mason