Subject: Booting A3000 with Apollo/68060 fails
To: None <port-amiga@NetBSD.ORG>
From: Michael Shuldman <michaels@ifi.uio.no>
List: port-amiga
Date: 01/04/1997 04:58:17
After switching of some 68060 spesific things (not neccesary
in -current it seems) it now actually boots the kernel, but times out
at the scsi bus (note, the scsi setup and everything else is as it has
been (and worked) for years, part from the Apollo 68060 card (cpu rev 1 according
to dmesg) and the lack of memory on the motherboard (all is on the
apollo card) everything is the same.)  The Apollo also comes with a scsi
controller, nothing is connected to that, everything is on the A3000 mb scsi.

my config is

ECS NTSC Agnus (id=$0030), ECS Denise (id=$00FC)
VERS:   Kickstart version 40.68, Exec version 40.10, Disk version 40.29
RAM:    Node type $A, Attributes $5 (FAST), at $C000000-$DFFFFFF (32.0 meg)
        Node type $A, Attributes $703 (CHIP), at $1000-$1FFFFF (~2.0 meg)
BOARDS:

Board (unidentified):   Prod=8512/34($2140/$22) (@$40000000, size 64meg,
       subsize same)
#this is a cv64 with 4MB of memory

Board + ROM (HD?) (unidentified):   Prod=8738/35($2222/$23) (@$EA0000 128K)
#this is the apollo card, 32MB of memory.


RC:0 5> loadbsd-2.14 -t netbsd
Translated 0c000020 sz 01ffffe0 to 0c000020 sz 01ffffe0
Translated 00001020 sz 001fefe0 to 00001020 sz 001fefe0
mem segment 1: start=0c000000 size=02000000 attribute=0005 pri=64
mem segment 2: start=00000000 size=00200000 attribute=0703 pri=-10
kernel size: 1226352
Using 32M FASTMEM at 0xc000000, 2M CHIPMEM
*** Loading from 0dd91f10 to Fastmem 0c000000 ***
RC:0 5>                                                                        


Using something about -current I get now;

Amiga 3000 (68060 rev 1 (cpu/mmu/fpu))
real mem = 33554432 (4096 pages)
avail mem ) 30285824 (3697)
mem seg 0 at c000000 suze 2000000
mem seg 1 at 0 size 200000
...
zbus0 at mainbus0: i/o size 0x4000000
...
board at zbuus0:... #apollo                                                                  
(it also finds my cybervision card though I didnt write that down)


Then it times out in sbicwait() at the first probe after approximately 5 seconds;

sbicwait TIMEO @913 with asr=x0 csr=x0

(this goes into an endless (it seems) loop, @line changes from 913 to something
else, that "something else" remains constant for as long as I could see, so
do asr and csr.)

the top of stack looks like this after the first timeout (modulo my errors
on writing this down on paper):

sbicwait(2e50041, ffffff80, 0, 394, 19) + 5c
sbicreset(581c00, 581c00, b1d84, 576340, 581c00) +5c
sbicinit(581c00, 581c00, aa5a8, fbe68, 2a3f8) + 126                       
ahscattach(576340, 581, 00, ad3b3, 9d363, b1dbc) +9a


I have various things connected to my scsi bus and I've tried disconnecting
everything but two devices, my HD's, with no visible effect (and as said, the
scsi setup worked flawless for years with previosuly, a 68030 and then a 68040,
neither of which I have access to any more (that is, the 68030 is still there
but I have only chip ram on the mb now).) 

I experience no scsi related problems on the AmigaOS side, my wd chip is revision
-08, anything else I can say ?


Help/tips is greately appreciated.

-- 
  _ // 
  \X/ -- Michael Shuldman <michaels@ifi.uio.no>