Subject: VS4k90 firmware damage
To: None <port-vax@netbsd.org>
From: Michael Kukat <michael@unixiron.org>
List: port-vax
Date: 01/26/2003 21:59:07
Hello all 4k90-users,

the mail found here about this VS4k90 firmware update recently was a very good
hint. I looked at my both 90's, both have a 1.0-firmware. For those who want to
know: the byte at 0x200a0009 is 0x01 here, 0x200a0008 is 0xca.
I first followed the hint of someone to just write back the original value, so
i did from firmware, from booter and from kernel, tries in byte, short and
long-access. But no success. This is clear: A Flash is like an EPROM in this
case, you can just reset bits. If you look at the dz-driver, you see, a 0 is
written to the location, the a 1. So why is 0 in the firmware? Because of the
fact that you can just clear bits in EPROM-like devices. To reprogram this byte
with the real value, you have to read out the whole flash, find out where the
control register resides, do the bulk erase (yes, the Flash in the 4k90 isn't
blocked, it can just be bulk-erased according to the datasheet), and reprogram
the whole array.
As this was too much for my little time to implement (i would have to read the
whole datasheet and find the control register), i decided to give the firmware
update a chance. Declared as 1.3 image, this is in fact a 1.4 firmware for
Cougar. It should work on models 90, 91A and 96. And i can confirm: Ir works
perfect, my 4k90 feels well again.

So: son't try around too much, DEC made crap here, the Vpp line is hardwired to
the power supply, write protect can't be disabled, and repairing the sometimes
rather old version is lots of effort. "Upgrade" is easiest thing to fix the
problem.

...Michael

-- 
http://www.unixiron.org/    Home Powered by: (Net|Open|Free)BSD IRIX NonStop-UX
Solaris AIX HP-UX Tru64 MUNIX Ultrix VMS SINIX Dolphin_Unix OpenStep MacOS A/UX