SUN Blade 100 - how to make it work again: NVRAM etc

[mar...@feldtmann.online]

Hello,

after some years I started my Blade 100 again and well it did not book any more. I bought a new NVRAM chip (via Farnell) and replace the chip on that motherboard.

But the system is still not booting: what is missing is the Mac-address and the hostid value ... perhpas more, but I do not know that.

On my Blade 100 I have OpenBoot 4.5 and all the papers I read on the Internet uses commads like mkp or stuff like this. My OpenBoot 4.5 do not know thiese commands at all.

How can I fix all these NVRAM issues on a Blade 100 ????

And the next question for the next step is: Is there a Solaris 8 or 9 available for download ? I was able tp download Solaris 10 - but will this work on that machine ?


Thanks from the community.

A Solaris -try-to-get- newbee ...

Marten

[Andrew Gabriel]

In article <635a7c52-03d0-4298...@googlegroups.com>,

Solaris 10 should work, but not in the smallest memory configurations.
I don't know what the minimum is, but you might struggle with anything
less than 1Gbyte. System max is 2Gbytes IIRC.
Also, some of these older systems are extremely slow to boot using the
boot framework in Solaris 10 (Ultra 5/10 take about 5 mins to read
in the RAMdisk image used in Solaris 10 - their boot prom was never
designed for reading in more than a few disk blocks originally, and is
very slow).

--
Andrew Gabriel
[email address is not usable -- followup in the newsgroup]

[Andreas F. Borchert]

On 2017-04-14, mar...@feldtmann.online <mar...@feldtmann.online> wrote:
> after some years I started my Blade 100 again and well it did not book
> any more. I bought a new NVRAM chip (via Farnell) and replace the chip
> on that motherboard.

[..]

> And the next question for the next step is: Is there a Solaris 8 or 9
> available for download ? I was able tp download Solaris 10 - but will
> this work on that machine ?

I have a Sun Blade 100 with 1,25 GB main memory which is running
Solaris 10 without any issues.

(Unfortunately I have no experience how to proceed best here with
the new NVRAM chip.)

Andreas.

[rshe...@gmail.com]

On Friday, April 14, 2017 at 3:35:16 AM UTC-4, mar...@feldtmann.online wrote:
...

> On my Blade 100 I have OpenBoot 4.5 and all the papers I read on the Internet uses commads like mkp or stuff like this. My OpenBoot 4.5 do not know thiese commands at all.
>
> How can I fix all these NVRAM issues on a Blade 100 ????
>

I <think> you can reprogram the chip using the c! command at OBP,
e.g. "<val> <addr> c!"
First you need to find the address of the NVRAM chip and then, of course, you need to know what values to put in.

[mar...@feldtmann.online]

Ok, perhaps I should pull my Blade 100 out of the trash-bin again ... :-)

[rshe...@gmail.com]

On Wednesday, April 26, 2017 at 3:04:31 AM UTC-4, mar...@feldtmann.online wrote:
> Ok, perhaps I should pull my Blade 100 out of the trash-bin again ... :-)
>

If you do pull it out I should be able to walk you thru finding the ram locations, at least I have a process which works on the SB2500. The MAC address may have been on a yellow sticker on, or near, the original NVRAM chip, but most of the values in the chip aren't <terribly> important.

[mar...@feldtmann.online]

The MAC address is indeed on a sticker on the original chip ...

[rshe...@gmail.com]

1) Do a `show-devs' to locate the path to the nvram or eeprom
device node. You get something like "/i2c@1f,464000/idprom@0,ae"

2) Then: " /i2c@1f,464000/idprom@0,ae" select-dev
***NOTE the space at the beginning

3) Then either ".properties" or ".attributes" depending on your version of OB - look for the "address" line

4) Dump 32 bytes to see if it looks right - "<address> 20 dump"

5) If you feel comfortable you can enter the new values - "<val> <address> c!"
***NOTE if there was "FF" or "00" there before it's pretty safe, if it looks like there's other data you may want to confirm the address first.

6) If you do enter all 15 values the 16th is the XOR checksum of them

[rshe...@gmail.com]

On Thursday, May 4, 2017 at 8:13:30 PM UTC-4, rshe...@gmail.com wrote:
> On Thursday, May 4, 2017 at 1:18:56 PM UTC-4, mar...@feldtmann.online wrote:
> 1) Do a `show-devs' to locate the path to the nvram or eeprom
> device node. You get something like "/i2c@1f,464000/idprom@0,ae"
>

> ...

Also I think you may need to add an offset of 1fd8 to the initial address,
on my Ultra 5 it was:

fffba000 + 1fd8 = FFFBBFD8

(I'm piecing this together from old scattered notes)

[rshe...@gmail.com]

On Thursday, May 4, 2017 at 8:40:00 PM UTC-4, rshe...@gmail.com wrote:
> On Thursday, May 4, 2017 at 8:13:30 PM UTC-4, rshe...@gmail.com wrote:

(some incorrect info)

Okay, hold the presses ... I tried this on my SB2500 and the .properties command didn't give an address. So here is an alternate method using debug:

{1} ok debug .idprom
{1} ok .idprom
Press space to step thru the program all the way back to "ok"
Note the address repeating on the left in brackets
e.g. (f0054a8c)

{1} ok f0054a8c (debug
{1} ok .idprom
Press space to step thru the program all the way back to "ok"
Note the new address repeating on the left in brackets
e.g. (f0054a08)

{1} ok f0054a08 (debug
{1} ok .idprom
Press space to step thru the program
Note the address repeating on the RIGHT in brackets
e.g. ( 0 fedc8d28 )
( 1 fedc8d28 )
( 2 fedc8d28 )
etc.

{1} ok fedc8d28 20 dump
The pointer "\/" should be pointing at the first byte of your
idprom data (e.g. at address fedc8d28)

{1} ok debug-off

Enter addresses & values very carefully (a byte in the wrong place could do bad things) using <value> <address> c!
You can confirm your data by repeating the dump command occasionally
(e.g. fedc8d28 20 dump)