Re: 4400/200: (1) update, (2) "RESETing to change Configuration!"?

>> Sometimes - usually in response to a reset-all or boot command - it
>> will print

>> RESETing to change Configuration!

> This is the expected consequence of losing PRAM battery power (same as losin$

Curious.  This has happened even when the machine has been plugged in
for many minutes, so the battery power line should have long ago
stabilized at one 1N914 drop below the standby 5V line.  (There are
also two resistors in series with it, but if it's pulling enough
current for them to drop significant voltage, the time it would take to
drain dry cells would be measured in days, not years.)

It also has happened multiple times in a row, with the machine fully
powered up all the time except long enough to power-cycle it in
response to that failure mode.

Maybe the power supply is starting to fail and +5 standby is driving a
significantly lower voltage.  I should probably measure it sometime.

> No idea why it doesn't reset load-base, unless the load-base is something th$

load-base           600000              4000 

> Note that replacing the battery will probably also help with point (1) (trim$

That _is_ something I want to do.

It seems to be behaving itself now.  I'm now trying to get it brought
up to my latest source tree, but there's a major flag day issue in
doing so.  The biggest other issue is that I don't really understand
the booting process, so I'm not sure how to create a new bootable disk;
I'm just copying the one I've got.  It has a 2G ATA drive in there (a
real 2G drive, predating the now-ubiquitous labeling of drives in
artifically-shrunk units) which has been in there since I set the
machine up, probably about two decades ago (see the 2001 timestamp
below).  I also eventually figured out how to get it to use an external
SCSI drive (appears it _must_ be ID 0; I've been unable to convince it
to boot off any other SCSI ID). I just copied the ATA drive to a >2G
SCSI drive and it "works fine" (but of course the label thinks it's a
2G drive).  I've been unable to find any NetBSD tools to work with the
Apple label in the NetBSD that's on there (1.4T); I may have to create
one, because I really would like to be able to label drives properly
instead of just copying the working drive I have.  I no longer recall
how I got it labeled in the first place.  The first 2K of the drive
appears to contain labels; it's mostly NULs but I see strings like
"QUANTUM FIREBALLfictitious", "Apple_partition_map", "NetBSD",
"NetBSD/macppc", and "PowerPC".  The next 1K is probably first-stage
booter.  It's got what looks like machine code (though I haven't tried
disassembling it) and contains OF-relevant strings like "finddevice",
"getprop", "open", "/chosen", "boot-device" (that's just a sampling).
8K in, the root filesystem's FFS superblock begins - convenient,
because it means I can mount the raw device as a filesystem on other
machines (big-endian or with FFS_EI) without needing that machine to
understand the partitioning.

It may be relevant that, when booting, it prints "bad partition number,
using 0" before printing the NetbSD/macppc OF booter banner.  After
manually cleaning up the turds from OF's mania for blindly assuming
X3.64 sequences work, here's what's left:

0 > boot scsi/sd@0:,\ofwboot.xcf bad partition number, using 0
>> NetBSD/macppc OpenFirmware Boot, Revision 1.2
>> ([email protected], Sat Apr  7 23:16:40 EDT 2001)
1767548+270356 [93+67424+53577]=0x20f400
 start=0x100000
Copyright (c) 1996, 1997, ...

I do note that that timestamp - the string "Sat Apr  7 23:16:40 EDT
2001" - occurs at exactly one place on the whole drive; this is part of
what, in NetBSD, is /ofwboot.  I haven't dug enough to know much more,
though I do _not_ see any "ofwboot" string in the first 8K of the
drive, and the boot command I've used always specifies ofwboot.xcf, not
just ofwboot.  I clearly need to investigate this more.

/~\ The ASCII				  Mouse
\ / Ribbon Campaign
 X  Against HTML		[email protected]
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B