Hardware or software?

[Mouse <mouse%Rodents-Montreal.ORG@localhost>]

I don't know whether anyone still remembers enough to say anything
useful about this, but, well, if I don't ask I definitely won't get
anything helpful back. :)

I'm trying to run my 5.2-derived NetBSD on an HP 9000/712/80 (A2615A,
the back-panel sticker calls it).  It's not working, and I'm wondering
if anyone remembers enough to be able to tell me whether this is broken
hardware or broken software (incomplete support, bugs, whatever).

I cross-built from i386.  Building the release left me with two files
in installation/, SYSNBSD and netinstall.lif.  I've tried each one, and
I get basically the same behaviour from each.  The machine DHCPs for an
address; the relevant dhcpd.conf clause says

host HP700.Rodents-Montreal.ORG {
        hardware ethernet 08:00:09:7d:2c:a2;
        fixed-address 10.0.1.31;
        option root-path "/nfs/HP700";
        filename "hp700.netinstall.lif";
}

and I've put netinstall.lif in place as hp700.netinstall.lif; when that
didn't work, I put SYSNBSD in there instead - same symptom.  The ROMs
put up a pseudo-GUI display telling me to press ESC to interrupt the
boot, which I did.  This left me at a BOOT_ADMIN, I think it was,
prompt.  I got the screen set to 1280x1024, which is why I have as much
of the output as I do.  If I try to just "boot lan", console output
gets shoehorned into a tiny little piece of the screen, small enough to
be useless; what I can see is consistent with it being the same failure
mode as below.  I got something more useful when I "boot lan isl"; the
whole screen gets used for console output.  (Well, all lines.  It wraps
a column or two before 80 columns, even though there's room for at
least twice that.)

Ten-finger copy:

NetBSD 5.2 (GENERIC) #0: Sun Oct 18 19:51:26 2020
mouse%Stone.Rodents-Montreal.ORG@localhost:/home/mouse/hp700/OBJDIR/sys/arch/hp700/compil
e/GENERIC
HP9000/712/80 (King Gecko)
real mem = 65536 KB (73720 reserved for PROM, 52056 KB used by NetBSD)
avail mem = 51972 KB
mainbus0 (root) [flex fff80000]
pdc0 at mainbus0
cpu0 at mainbus0 hpa fffbe000 path 8 irq 31 ipl 0: PA7100LC (Hummingbird) rev
 5
cpu0: PCX-L, PA-RISC 1.1c, lev 1, cet A, 80 MHz clk
cpu0: shadows, 128K/128K D/I caches, 120 shared TLB, 8 shared BTLB
cpu0: PCX-L (CMOS-26B) floating point, rev 1
mem0 at mainbus0 hpa fffbf000 path 9: viper rev 0, ctrl 40400102 <eisa_prf> si
ze 64MB

machine check: HPMC - WARNING: could not transfer PIM info (-5)

General Registers:
0x00000000 0x00000000 0x00000000 0x00000000
0x00000000 0x00000000 0x00000000 0x00000000
0x00000000 0x00000000 0x00000000 0x00000000
0x00000000 0x00000000 0x00000000 0x00000000
0x00000000 0x00000000 0x00000000 0x00000000
0x00000000 0x00000000 0x00000000 0x00000000
0x00000000 0x00000000 0x00000000 0x00000000
0x00000000 0x00000000 0x00000000 0x00000000

Control Registers:
0x00000000 0x00000000 0x00000000 0x00000000
0x00000000 0x00000000 0x00000000 0x00000000
0x00000000 0x00000000 0x00000000 0x00000000
0x00000000 0x00000000 0x00000000 0x00000000
0x00000000 0x00000000 0x00000000 0x00000000
0x00000000 0x00000000 0x00000000 0x00000000
0x00000000 0x00000000 0x00000000 0x00000000
0x00000000 0x00000000 0x00000000 0x00000000

Space Registers:
0x00000000 0x00000000 0x00000000 0x00000000
0x00000000 0x00000000 0x00000000 0x00000000

IIA 0x0:0x00000000 0x0:0x00000000
IPSW 0
SP 0x0:0x00000000 FP 0x0:0x00000000

Check Type 0
CPU State 0
Cache Check 0
TLB Check 0
Bus Check 0
Assist Check 0Assist State 0
System Responder 0x00000000
System Requester 0x00000000
Path Info 0x00000000
panic: machine check
Stopped in pid 0,1 (system) at  netbsd:cpu_Debugger+0x4:       bv,n    %r0(%r2
)

db>

There is a pause - long enough to be human-perceptible, perhaps two or
three seconds - between the mem0 line and the machine check line.

Any thoughts?

/~\ The ASCII				  Mouse
\ / Ribbon Campaign
 X  Against HTML		mouse%rodents-montreal.org@localhost
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B