Problems with NVRAM replacement
christop...@gmail.com
I bought a new NVRAM chip as recommended by people here (and the NVRAM
FAQ), with the hope of fixing my Ultra 2's clock. I bought an NVRAM
chip from www.farnell.co.uk. It is exactly the same chip as the
original - same model number and manufacturer. I switched off the
machine at the wall and fitted the new chip using an ESD protection
wristband. I made sure it was the correct way round in the socket with
the pins fully inserted. When I switched the machine back on, I
couldn't get any output to the monitor. All I heard was the power
supply fans whirring. There was no sound from the hard drives. So I put
the old chip back. The machine started to boot, so I switched it off
and tried the new chip again. Same problem. I tried this a couple more
times but with the same result. Eventually I put the old chip back in
so I could use the machine, but it wouldn't boot, saying "Can't open
deblocker package". I assume that the old NVRAM with the failing
battery lost its information while removed. Any suggestions as to what
I should do? Is there anything obvious I've overlooked or should I
return the chip as faulty?
Any advice would be much appreciated.
Best wishes,
Chris Tidy
Dave
Hook up a serial port connection, and see what you get from that.
It is most unlikly a chip from Farnell will be bad, although I know that
Farnell offer good service, so I expect they would courier you a new one
the same day.
It is much more likly to be a problem with the U2, or the way you have
installed it.
--
Dave K
http://www.southminster-branch-line.org.uk/
Please note my email address changes periodically to avoid spam.
It is always of the form: month-year@domain. Hitting reply will work
for a couple of months only. Later set it manually. The month is
always written in 3 letters (e.g. Jan, not January etc)
christop...@gmail.com
Unfortunately I can't hook up a serial port connection as this is my
only Unix box. I have a PC, but I guess I'd need special software to do
it, and I don't have an appropriate serial cable.
I too am doubtful that a chip from Farnell would be bad, but as far as
I can see I did everything correctly. If the new chip is good, it seems
very strange that the machine will not respond with the new chip
fitted, yet it works with the old chip.
Chris
Dave
> Hi Dave,
>
> Unfortunately I can't hook up a serial port connection as this is my
> only Unix box. I have a PC, but I guess I'd need special software to do
> it, and I don't have an appropriate serial cable.
You need nothing special - hyperterminal on a PC will do. Terrterm is
better.
The serial cable is easy to make - a couple of 9 pin connectors and a
bit of wire. Or buy one. You really *must* get a serial cable if you
wnat to do much with a Sun - as you are finding out.
CJT
> christop...@gmail.com wrote:
>
>> Hi Dave,
>>
>> Unfortunately I can't hook up a serial port connection as this is my
>> only Unix box. I have a PC, but I guess I'd need special software to do
>> it, and I don't have an appropriate serial cable.
>
>
> You need nothing special - hyperterminal on a PC will do. Terrterm is
> better.
>
> The serial cable is easy to make - a couple of 9 pin connectors and a
> bit of wire. Or buy one. You really *must* get a serial cable if you
> wnat to do much with a Sun - as you are finding out.
>
>
That seems a bit overstated. I was using Sun stuff for years before I
first truly needed one.
--
The e-mail address in our reply-to line is reversed in an attempt to
minimize spam. Our true address is of the form che...@prodigy.net.
Oscar del Rio
> Dave wrote:
>> The serial cable is easy to make - a couple of 9 pin connectors and a
>> bit of wire. Or buy one. You really *must* get a serial cable if you
>> wnat to do much with a Sun - as you are finding out.
>>
>>
> That seems a bit overstated. I was using Sun stuff for years before I
> first truly needed one.
let's put it this way: sooner or later you will need a serial cable :)
Paul Gress
I believe a "stop-n" is required to set the nvram to default values.
After this you may need to change the mac address to your required value.
Paul
christop...@gmail.com
> I believe a "stop-n" is required to set the nvram to default values.
> After this you may need to change the mac address to your required value.
I tried <Stop-N> with the old NVRAM chip installed. It doesn't appear
to do anything. But <Stop-A> brings me to the "ok" prompt. If I then
type "set-defaults" it appears to set the NVRAM defaults successfully.
But then my computer refuses to boot from the hard disk, saying "Can't
open deblocker package". Does anyone know what this error message means
and what might be a likely cause? It booted fine before I removed the
NVRAM chip, with the only fault being the clock losing time.
As I said, I can't get any response from the machine with the new NVRAM
chip installed. There is no display on the monitor and no sound apart
from the power supply fans. I'll see if I can get a serial cable, but I
was hoping this problem wouldn't be too hard to fix. At the moment the
only other thing I can think of doing is getting another brand new
NVRAM chip.
Any thoughts or suggestions would be very much appreciated.
Chris
Dave
> Paul Gress wrote:
>
>
>>I believe a "stop-n" is required to set the nvram to default values.
>>After this you may need to change the mac address to your required value.
>
>
> I tried <Stop-N> with the old NVRAM chip installed. It doesn't appear
> to do anything.
What stop-n does may not be noticable, especially since your battery was
flat.
But with your new chip, it might be worth holding down stop and N when
powering up - you can't achieve the effect once power is on. Nothing to
loose.
> But <Stop-A> brings me to the "ok" prompt. If I then
> type "set-defaults" it appears to set the NVRAM defaults successfully.
> But then my computer refuses to boot from the hard disk, saying "Can't
> open deblocker package". Does anyone know what this error message means
> and what might be a likely cause? It booted fine before I removed the
> NVRAM chip, with the only fault being the clock losing time.
If stop-n has been run, thinks like 'disk' will have been reset, so may
no longer point to a bootable disk. If the disk is not in the default
location, the Solaris install will change the EEPROM. But stop-n will
clear that. But I guess if the battery has died, it will clear it too.
> As I said, I can't get any response from the machine with the new NVRAM
> chip installed. There is no display on the monitor and no sound apart
> from the power supply fans. I'll see if I can get a serial cable, but I
> was hoping this problem wouldn't be too hard to fix.
That will make your life so much easier, since you will see any error
messages generated.
> At the moment the
> only other thing I can think of doing is getting another brand new
> NVRAM chip.
I think the probababiliy of a new chip from Farnel being dead is close
to zero.
Without wishing to insult your intelligence, are you sure you have the
cip the right way around? The notch will be the end where pin 1 is.
(I've put chips in back to front before. Some like a 741 op-amp will
survive, but others do not like it)
christop...@gmail.com
> But with your new chip, it might be worth holding down stop and N when
> powering up - you can't achieve the effect once power is on. Nothing to
> loose.
Okay, I thought I was supposed to press <Stop-N> after the memory test.
I'll try holding them down when I switch on the power instead.
> If stop-n has been run, thinks like 'disk' will have been reset, so may
> no longer point to a bootable disk. If the disk is not in the default
> location, the Solaris install will change the EEPROM. But stop-n will
> clear that. But I guess if the battery has died, it will clear it too.
The disk is in the default location.
> > As I said, I can't get any response from the machine with the new NVRAM
> > chip installed. There is no display on the monitor and no sound apart
> > from the power supply fans. I'll see if I can get a serial cable, but I
> > was hoping this problem wouldn't be too hard to fix.
>
> That will make your life so much easier, since you will see any error
> messages generated.
I'll have a look on eBay and see what they cost.
> > At the moment the
> > only other thing I can think of doing is getting another brand new
> > NVRAM chip.
>
> I think the probababiliy of a new chip from Farnel being dead is close
> to zero.
>
> Without wishing to insult your intelligence, are you sure you have the
> cip the right way around? The notch will be the end where pin 1 is.
No problem. I understand that it's easy to insert chips the wrong way
round, so I checked that the notch on the new chip was in the same
position as the notch on the old chip. I'm sure I didn't insert the
chip the wrong way round at any time. I also checked the manufacturer's
data sheets when it didn't work, just in case they've reversed the pin
numbering, but they haven't.
Best wishes,
Chris
Dave
> Thanks for the suggestions.
>
>
>>But with your new chip, it might be worth holding down stop and N when
>>powering up - you can't achieve the effect once power is on. Nothing to
>>loose.
>
>
> Okay, I thought I was supposed to press <Stop-N> after the memory test.
> I'll try holding them down when I switch on the power instead.
No, press the keys *before* powering up and keep them pressed for a few
seconds after power up. I've had to do it a couple of times myself - not
sure why.
>>That will make your life so much easier, since you will see any error
>>messages generated.
>
>
> I'll have a look on eBay and see what they cost.
It's a null modem cable you want, but there are some reports of not all
null modem cables working. I built my own from a site that reports them
working with Suns - stokely being one obvious one.
I assume you are in the UK. Maplin should have the plugs and a bit of
wire. I forget what Sun this is on, but I guess it is a fairly old one,
so probably has 25-pin serial connectors. So you are likely to need 25
one end and 9 the other.
I don't know if you are anywhere near Essex, but if you are, and want to
drop in, I'll connect it with my serial cable to my Sun and see if there
are any messages during power up. Drop me a private email if you want to
do that - the reply address is valid.
christop...@gmail.com
> It's a null modem cable you want, but there are some reports of not all
> null modem cables working. I built my own from a site that reports them
> working with Suns - stokely being one obvious one.
>
> I assume you are in the UK. Maplin should have the plugs and a bit of
> wire. I forget what Sun this is on, but I guess it is a fairly old one,
> so probably has 25-pin serial connectors. So you are likely to need 25
> one end and 9 the other.
Thought so. A null modem cable is wired straight through, so that the
wires are all parallel, as opposed to having the wires cross to go pin
1 -> pin 1, pin 2 -> pin 2, etc. - is that right? I've made a serial
cable for this machine before (to connect it to a US Robotics modem),
so if I can find the connectors I'd be happy to do it again. This
machine does indeed have 25-pin connectors.
> I don't know if you are anywhere near Essex, but if you are, and want to
> drop in, I'll connect it with my serial cable to my Sun and see if there
> are any messages during power up. Drop me a private email if you want to
> do that - the reply address is valid.
Thanks very much for the offer. I'm in Shropshire, which would be a way
to travel!
I've also had a little success now. I held down <Stop-N> when I
switched on the power with the old chip, and got a message saying that
the NVRAM parameters were being reset to the defaults. Then it said
"Can't open deblocker package". I tried the same with the new chip.
This time I got an output on the monitor, which confirmed that the
NVRAM parameters were being reset to the defaults, and then said "The
IDPROM contents are invalid" followed by "Can't open deblocker
package". It is just possible that the monitor lead was loose before,
which meant that I wasn't getting a display on the monitor. I tried
resetting the IDPROM contents at the "ok" prompt, but got the same
message when I switched on the machine again. What do you think about
messages? Any thoughts?
Best wishes,
Chris
Sunny
The next step is to program your hostid into the new chip:
http://www.squirrel.com/squirrel/sun-nvram-hostid.faq.html
Sunny
christop...@gmail.com
archive explaining that the "IDPROM contents are invalid" message was
the result on an invalid checksum. So I fixed this and have checked
that the IDPROM contents match what was on the old NVRAM chip. But I
still can't boot from the disk and am getting the "Can't open deblocker
package" message. Does anyone know what the message means and how to
fix it? I'm sure there must be a simple solution to this.
Many thanks,
Chris
Dave (from the UK)
> Dave wrote:
>
>>It's a null modem cable you want, but there are some reports of not all
>>null modem cables working. I built my own from a site that reports them
>>working with Suns - stokely being one obvious one.
>>
>>I assume you are in the UK. Maplin should have the plugs and a bit of
>>wire. I forget what Sun this is on, but I guess it is a fairly old one,
>>so probably has 25-pin serial connectors. So you are likely to need 25
>>one end and 9 the other.
>
>
> Thought so. A null modem cable is wired straight through, so that the
> wires are all parallel, as opposed to having the wires cross to go pin
> 1 -> pin 1, pin 2 -> pin 2, etc. - is that right?
No.
The bottom of this page has all the connections for the 9 and 25 pin
cables. I've made one of each up myself.
http://www.sunhelp.org/unix-serial-port-resources/serial-pinouts/
>>I don't know if you are anywhere near Essex, but if you are, and want to
>>drop in, I'll connect it with my serial cable to my Sun and see if there
>>are any messages during power up. Drop me a private email if you want to
>>do that - the reply address is valid.
>
>
> Thanks very much for the offer. I'm in Shropshire, which would be a way
> to travel!
Sure. The train fare will probably cost more than the Sun!
> I've also had a little success now. I held down <Stop-N> when I
> switched on the power with the old chip, and got a message saying that
> the NVRAM parameters were being reset to the defaults. Then it said
> "Can't open deblocker package". I tried the same with the new chip.
> This time I got an output on the monitor, which confirmed that the
> NVRAM parameters were being reset to the defaults, and then said "The
> IDPROM contents are invalid" followed by "Can't open deblocker
> package". It is just possible that the monitor lead was loose before,
> which meant that I wasn't getting a display on the monitor. I tried
> resetting the IDPROM contents at the "ok" prompt, but got the same
> message when I switched on the machine again. What do you think about
> messages? Any thoughts?
Googling on deblocker seems to show its related to firmware:
http://docs.sun.com/app/docs/doc/806-1379-10/6jalunvei?a=view
http://www.openfirmware.org/1275/mejohnson/
says:
Now many systems will also define a pure software package, called the
"deblocker", whose purpose in life is to "deblock" a block-oriented
device, i.e., provide byte-oriented methods to access the block-oriented
device. This is analogous to the old FORTH trick of implementing virtual
memory using BLOCK, which views the disk as a block-oriented device
(hence the name). The deblocker has its own entry in the device-tree
much like a true physical device, but there are a few significant
differences: it must be a child of the node "/packages" in the
device-tree, it has no address, and it can and should be opened
with"open-package" instead of with "open-device".
But don't ask me what that lot means!! I am out of my depth!
Mike Tremaine
I've never heard of this either but it sounds like something is wrong with
your device tree.
At the "ok" prompt you should be able to issue a "show-devs" this will
list out the device tree. You should have /packages/deblocker. If it is
missing then thats a problem.
You can read more info here.
http://docs.sun.com/app/docs/doc/805-4436/805-4436
-Mike
Colin B.
(tons snipped)
> (I've put chips in back to front before. Some like a 741 op-amp will
> survive, but others do not like it)
It's been a long time since I played with audio circuits, but from what I
remember, a 741 will survive almost anything short of a hammer.
Colin
Darren Dunham
> Thought so. A null modem cable is wired straight through, so that the
> wires are all parallel, as opposed to having the wires cross to go pin
> 1 -> pin 1, pin 2 -> pin 2, etc. - is that right?
No, a "modem cable" would be straight through. A machine like the SUN
is wired DTE while the modem is DCE. A straight through cable connects
them.
A PC would also be DTE, so to connect the two DTE devices, the data
cables (and possibly some others) have to be swapped.
> I've made a serial
> cable for this machine before (to connect it to a US Robotics modem),
> so if I can find the connectors I'd be happy to do it again. This
> machine does indeed have 25-pin connectors.
A cable that connects to a modem will not work by itself to connect to a
PC. You'd need to add or modify something to exchange the data lines.
--
Darren Dunham ddu...@taos.com
Senior Technical Consultant TAOS http://www.taos.com/
Got some Dr Pepper? San Francisco, CA bay area
< This line left intentionally blank to confuse you. >
Darren Dunham
> It is much more likly to be a problem with the U2, or the way you have
> installed it.
I would agree here.
In the few machines I've had run-time clock issues with, it did not
appear to have anything to do with the NVRAM (I kept the old ones and
continued to use them), and the machines usually died within a few days
of other issues.
Motherboard swaps fixed the problem in all cases, but other than the
NVRAM being ruled out, I have no idea which component may have been at
fault.
christop...@gmail.com
<snip>
> I've never heard of this either but it sounds like something is wrong with
> your device tree.
>
> At the "ok" prompt you should be able to issue a "show-devs" this will
> list out the device tree. You should have /packages/deblocker. If it is
> missing then thats a problem.
I checked using "show-devs" and /packages/deblocker is shown in the
device tree. From what I've read /packages/deblocker doesn't point to a
physical device, so it can't be a case of a "broken link". Does anyone
know where deblocker software is stored? Is it on a chip? Is it
possible that my OBP is in some way messed up?
Many thanks,
Chris
Sunny
We're getting in pretty deep, but best I can tell the "deblocker
package" is a collection of routines the OBP needs to access the disk.
Google finds a few examples of device driver code which spits out that
error message, and all of them open the disk-label package followed by
the deblocker package - however none of them appear to pass a
disk-label-dependent argument to the open deblocker routine.
This suggests (to me anyway) that your OBP may be corrupt - but I'm not
sure how you would go about reflashing it when you can't access the disk.
Will it boot cdrom?
Sunny
Mike Tremaine
>> I checked using "show-devs" and /packages/deblocker is shown in the
> device tree. From what I've read /packages/deblocker doesn't point to a
> physical device, so it can't be a case of a "broken link". Does anyone
> know where deblocker software is stored? Is it on a chip? Is it
> possible that my OBP is in some way messed up?
>
I googled around like others because this sounds so interesting :). I did
see some cases where hardware failures or lack of support for certain PCI
cards caused this error under Open Firmware on PPC. Seeing that would
suggest that perhaps some piece of hardware is to blame.
First I'd simplify the "variables", by removing all hardware that doesn't
need to be there. That is remove all Sbus cards, check that the CPU{S} are
well seated, check that the memory is well seated, remove the the
hard-drive, and then try to boot from cdrom. If it works start adding
everything back until it fails.
If that doesn't help then I'd probably try to find someone else with an
Ultra 2 so you can compare some of the openboot variables. Using the
Openboot manual you can figure out how to read the values of
/packages/deblocker see if yours seem valid.?? But this just a guess as I
said I never heard of this before. You might post on the
comp.sys.sun.hardware list also just in case there are guru's who don't
bother to look here anymore.
-Mike
Sunny
Mike Tremaine wrote:
I have an Ultra 2, and the problem has similarly piqued my interest :-)
I'm happy to help the OP if I can.
Sunny
ig...@nospam.invalid
>
> We're getting in pretty deep, but best I can tell the "deblocker
> package" is a collection of routines the OBP needs to access the disk.
>
> Google finds a few examples of device driver code which spits out that
> error message, and all of them open the disk-label package followed by
> the deblocker package - however none of them appear to pass a
> disk-label-dependent argument to the open deblocker routine.
>
> This suggests (to me anyway) that your OBP may be corrupt - but I'm not
> sure how you would go about reflashing it when you can't access the disk.
>
> Will it boot cdrom?
That is a very good advice.
If that machine is able to boot from CD-ROM (try a "boot cdrom" with the
"Software 1 of 2" disk into the CD-ROM drive) you will have a chance to
reflash the firmware. Just download the right firmware patch (patch-ID#
104169-08 for the Ultra2 workstations) and make a RockRidge CD with
the firmware update files on it. Just use the "flash-update-Ultra2-latest"
file as boot kernel. (You have an Ultra2, right? Perhaps I am missing
something but I read about that workstation on this thread right now).
If you are able to boot from CD-ROM media and make a RockRidge-compliant
CD with the firmware update files, you can boot it from the "ok" prompt
with a "boot cdrom /flash-update-Ultra2-latest". Good luck!
Any advice against this update method???
I just installed that firmware release on my new Ultra2 workstation,
but I booted from the internal HDD, not CD-ROM. I can try to make
a bootable ISO image and send it to you, if you want. I suppose
that you will be able to copy it to CD media easily.
ig...@nospam.invalid
>
> I just installed that firmware release on my new Ultra2 workstation,
> but I booted from the internal HDD, not CD-ROM. I can try to make
> a bootable ISO image and send it to you, if you want. I suppose
> that you will be able to copy it to CD media easily.
Ok, you will need to change a jumper on the motherboard temporarily
to allow the firmware update too. All the steps are nicely documented
by Sun.
Cheers,
Igor.
Dan Foster
> Sunny <su...@nospam.net> wrote:
>>
>> We're getting in pretty deep, but best I can tell the "deblocker
>> package" is a collection of routines the OBP needs to access the disk.
>>
>> Google finds a few examples of device driver code which spits out that
>> error message, and all of them open the disk-label package followed by
>> the deblocker package - however none of them appear to pass a
>> disk-label-dependent argument to the open deblocker routine.
>>
>> This suggests (to me anyway) that your OBP may be corrupt - but I'm not
>> sure how you would go about reflashing it when you can't access the disk.
>>
>> Will it boot cdrom?
>
> That is a very good advice.
>
> If that machine is able to boot from CD-ROM (try a "boot cdrom" with the
> "Software 1 of 2" disk into the CD-ROM drive) you will have a chance to
> reflash the firmware.
And if need be, or if this is easier:
You can also flash firmware via the network, too. I've done it before.
Why had I done that? Because my Ultra 5 was running only Linux, no
Solaris, and I did have a TFTP server handy... worked like a charm.
'boot net' was all it took once things were set up properly on the server.
-Dan
ig...@nospam.invalid
>
> And if need be, or if this is easier:
>
> You can also flash firmware via the network, too. I've done it before.
>
> Why had I done that? Because my Ultra 5 was running only Linux, no
> Solaris, and I did have a TFTP server handy... worked like a charm.
>
> 'boot net' was all it took once things were set up properly on the server.
Indeed, a firmware upgrade via network! He certainly has a very good
chance to avoid the deblocker related message with a network upgrade.
As you said, it is as easy as setting up a trivial FTP server and
booting the firmware updater from it, in the same way it is possible
booting a small kernel for a diskless computer.
If I remember correctly, Cisco has a free tftp server that runs on
Windows. The original poster says that the SPARC workstation is
the only Unix machine he owns.
I am sure, he will certainly appreciate your suggestion. It *should* work.
Cheers,
Igor.
christop...@gmail.com
> I just installed that firmware release on my new Ultra2 workstation,
> but I booted from the internal HDD, not CD-ROM. I can try to make
> a bootable ISO image and send it to you, if you want. I suppose
> that you will be able to copy it to CD media easily.
Igor,
If you could I'd be very grateful. I just downloaded the PROM update
104169-08 from SunSolve, but it's a .tar.Z file and I can't find
anything on the PC which will open it. I thought WinZip might, but
Windows doesn't associate the file with WinZip, and typing "winzip" in
the Start -> Run box doesn't find the program (it used to with older
versions of Windows). Grrr. I hate Windows. If it's less that 10 MB in
size perhaps you could e-mail it to me at cdt...@yahoo.com?
Thanks to everyone for the advice.
Chris
Dave (from the UK)
> Igor,
>
> If you could I'd be very grateful. I just downloaded the PROM update
> 104169-08 from SunSolve, but it's a .tar.Z file and I can't find
> anything on the PC which will open it. I thought WinZip might, but
> Windows doesn't associate the file with WinZip, and typing "winzip" in
> the Start -> Run box doesn't find the program (it used to with older
> versions of Windows). Grrr. I hate Windows. If it's less that 10 MB in
> size perhaps you could e-mail it to me at cdt...@yahoo.com?
a bit big to email, but I stuck it here. The server is not fast, but you
will get it.
http://www.althorne.org/104169-08.tar
christop...@gmail.com
cdrom /flash-update-Ultra2-09". The CD-Rom didn't appear to spin up and
I got the message "Can't open boot device". It doesn't seem to be able
to access the CD-Rom and after a few seconds defaults to "boot net".
This doesn't explicitly fail but I get the error message "Timeout
waiting for ARP/RARP packet". It is connected to our ADSL router. I
guess I might be able to flash the PROM over the network if I connect
it to the PC, but I'm not certain this is the problem - I'm starting to
think that I might have a motherboard fault as Darren suggested. I've
tried removing the unnecessary SBus cards and the second processor, and
have exchanged the power supply and graphics card for spares, but all
to no avail. The only thing I haven't touched is the RAM, because I
remember that it was extremely hard to fit in the first place. I've
also noticed some weird faint beeps coming from the speaker while the
machine is trying to boot, but I'm not sure if they're significant.
Looks like I might have to buy a replacement motherboard, or maybe a
whole U2 for spares.
Thanks for all the suggestions. Any more thoughts?
Best wishes,
Chris
christop...@gmail.com
I just tried running "probe-scsi-all" at the "ok" prompt
systematically. The results are really making me concerned about the
motherboard. It's behaving quite erratically. First time I got this:
/sbus@1f,0/SUNW,fas@e,8800000
Target 0
Target 1
Target 2
unexpected message in 13
Next time target 2 was replaced by target 6, the unexpected message bit
was gone and the machine hung, and the third time it didn't find any
devices at all. The fourth time it did the same as the first. There
should actually be targets 0,1,2,5 and 6 on this controller, and they
all have unique IDs. My SBus SCSI controller performs consistently.
I'm getting worried about the health of my motherboard.
Best wishes,
Chris
Sunny
Unfortunately it's not as simple as burning the update to CD - you need
a CD that is both bootable on Sparc and has the OBP update executable.
That's what Igor offered to build for you - very nice of him, since it's
a non-trivial task to build and test such an animal. Perhaps he's an
expert, personally I'd expect a couple of iterations to get it right -
and that's assuming I had known-good hardware to test the CD!
I think we have jumped too far ahead here. IMHO you are correct to
suspect a hardware problem, and have made efforts to isolate it - but
let's step back for a minute...
Your Ultra 2 won't boot from disk - but will it boot from CD? Asked but
not answered :-) Any Solaris installation CD will suffice for testing
this, and there's little point in anyone going to the effort of building
an OBP update CD for you if you can't boot an install CD.
If you still get the deblocker error on attempting CD boot, we'll know
the problem is not related to your disk.
It would also be useful to know if show-devs issued at the OBP prompt
lists /packages/deblocker in the device tree - Mike asked this, again
not answered.
As I said, I have a working Ultra 2 and an interest in getting to the
bottom of your issue because it appears to be quite unique. My U2 is not
running anything important, so if I mess up my hardware in the process,
no big deal :-)
Sunny
christop...@gmail.com
> Unfortunately it's not as simple as burning the update to CD - you need
> a CD that is both bootable on Sparc and has the OBP update executable.
> That's what Igor offered to build for you - very nice of him, since it's
> a non-trivial task to build and test such an animal. Perhaps he's an
> expert, personally I'd expect a couple of iterations to get it right -
> and that's assuming I had known-good hardware to test the CD!
>
> I think we have jumped too far ahead here. IMHO you are correct to
> suspect a hardware problem, and have made efforts to isolate it - but
> let's step back for a minute...
>
> Your Ultra 2 won't boot from disk - but will it boot from CD? Asked but
> not answered :-) Any Solaris installation CD will suffice for testing
> this, and there's little point in anyone going to the effort of building
> an OBP update CD for you if you can't boot an install CD.
Just done that now. It won't boot from the installation CD. Igor,
please don't spend time making the CD image! Sorry I didn't run this
test earlier, but I'm only just getting a grasp of the problem.
> If you still get the deblocker error on attempting CD boot, we'll know
> the problem is not related to your disk.
>
> It would also be useful to know if show-devs issued at the OBP prompt
> lists /packages/deblocker in the device tree - Mike asked this, again
> not answered.
That one I did answer. Maybe the message went astray and didn't reach
your server? /packages/deblocker is shown in the device tree.
Best wishes,
Chris
Dave (from the UK)
> A bit more information...
>
> I just tried running "probe-scsi-all" at the "ok" prompt
> systematically. The results are really making me concerned about the
> motherboard. It's behaving quite erratically. First time I got this:
>
> /sbus@1f,0/SUNW,fas@e,8800000
I have a faint recollection of seeing 'fas' on a Sunswift (SCSI and
100baseT ethernet). Perhaps it stands for fast SCSI.
I have also known a Sunswift card fail and stop a machine booting. They
run very hot (in SS20's anyway), but that does not say much, as SS20s
are not well cooled.
> Target 0
> Target 1
> Target 2
>
> unexpected message in 13
>
> Next time target 2 was replaced by target 6, the unexpected message bit
> was gone and the machine hung, and the third time it didn't find any
> devices at all. The fourth time it did the same as the first. There
> should actually be targets 0,1,2,5 and 6 on this controller, and they
> all have unique IDs. My SBus SCSI controller performs consistently.
You are reducing your probabiity of solving this by having lost of
devices on the bus. Far better to *only* have the CD or one disk. Remove
all sbus cards, framebuffer, all disks etc. Just hook up a serial
terminal. It could be one of your cards, scsi devices or similar that is
dead.
> I'm getting worried about the health of my motherboard.
Me too, but you seem to be doing tests with far too many things in the
machine.
christop...@gmail.com
with a floppy and framebuffer installed. Won't boot - still get the
deblocker error message. "probe-scsi-all" is at least consistent, but
still doesn't show the device details like manufacturer, model, etc.,
as it used to. It gives this output:
/sbus@1f,0/SUNW,fas@e,8800000
Target 0
I'll continue sometime tomorrow...
Chris
Dan Foster
file from the network as previously suggested?
It's not guaranteed to work, but avoids any HD or CDROM-related
complications.
All you need is a free TFTP server for Windows running and the firmware
updater file itself named in the proper format. That's it.
For example, a Yahoo search revealed many free TFTP servers... one was:
http://www.winagents.com/downloads/tftpsetup.exe
Install and run that. Then run the installed TFTP server.
Put the Sun firmware updater image file (uncompressed and extracted) in
whatever the TFTP Root path is -- by default, c:\Program
Files\WinAgents\TFTP Server\ (I think)
I don't recall what filename you will need to give to the
extracted+uncompressed firmware image file, but just look at the TFTP
requests from the Sun in the TFTP server log messages after you do 'boot
net' on the Sun.
The exact file size for the Ultra 2 OBP PROM version 3.25.0 update is
847,920 bytes. Make sure your updater image file matches this size
exactly.
Shut down the Sun, open it up, and make sure jumper J2003 is moved to
pins 2+3. (By default, it is at pins 1+2 which prevents writes to the
firmware area.)
Then at the 'ok' prompt, type in 'boot net' and watch the messages on
the TFTP server log window.
It should load the software and then start the PROM updater utility.
If it cannot find the file, rename the file to match what the Sun is
expecting to see according to any one of its TFTP requests.
I've done a PROM update for my Ultra 5 with success this way.
-Dan
Sunny
christop...@gmail.com wrote:
> Sunny wrote:
>
>>Unfortunately it's not as simple as burning the update to CD - you need
>>a CD that is both bootable on Sparc and has the OBP update executable.
>>That's what Igor offered to build for you - very nice of him, since it's
>>a non-trivial task to build and test such an animal. Perhaps he's an
>>expert, personally I'd expect a couple of iterations to get it right -
>>and that's assuming I had known-good hardware to test the CD!
>>
>>I think we have jumped too far ahead here. IMHO you are correct to
>>suspect a hardware problem, and have made efforts to isolate it - but
>>let's step back for a minute...
>>
>>Your Ultra 2 won't boot from disk - but will it boot from CD? Asked but
>>not answered :-) Any Solaris installation CD will suffice for testing
>>this, and there's little point in anyone going to the effort of building
>>an OBP update CD for you if you can't boot an install CD.
>
>
> Just done that now. It won't boot from the installation CD.
What was the error message?
Igor,
> please don't spend time making the CD image! Sorry I didn't run this
> test earlier, but I'm only just getting a grasp of the problem.
>
>
>>If you still get the deblocker error on attempting CD boot, we'll know
>>the problem is not related to your disk.
>>
>>It would also be useful to know if show-devs issued at the OBP prompt
>>lists /packages/deblocker in the device tree - Mike asked this, again
>>not answered.
>
>
> That one I did answer. Maybe the message went astray and didn't reach
> your server? /packages/deblocker is shown in the device tree.
Ahh - still can't find that post here, must have gone astray.
Perhaps we should explore /packages/deblocker as Mike suggested?
Sunny
Igor Sobrado
>
> Unfortunately it's not as simple as burning the update to CD - you need
> a CD that is both bootable on Sparc and has the OBP update executable.
> That's what Igor offered to build for you - very nice of him, since it's
> a non-trivial task to build and test such an animal. Perhaps he's an
> expert, personally I'd expect a couple of iterations to get it right -
> and that's assuming I had known-good hardware to test the CD!
Thanks, but I am not an expert at all. Just a physics graduate student
working on a Ph.D. in Computer Science. I have some knowledge on the
layout of Solaris CDs and I think that I will be able to make a CD-ROM
bootable from the OBP.
I will try it on my Ultra2 Creator. As soon as I get a bootable CD
it will be available from our anonymous FTP server, with a MD5 checksum.
Certainly, I would not advice about transferring big files by email!
About the ARP/RARP messages, I get the same on my Ultra2 Creator.
It is connected to a 3Com ADSL router. Do not worry about these
messages.
But the suggestion of flashing the firmware from a tftp server
is much better. Certainly, it would be a fine workaround for
any disk-related problem. And it seems that the original poster
has a problem reading from disk drives.
> I think we have jumped too far ahead here. IMHO you are correct to
> suspect a hardware problem, and have made efforts to isolate it - but
> let's step back for a minute...
I really would not care about the ARP/RARP timeouts. :-)
> Your Ultra 2 won't boot from disk - but will it boot from CD? Asked but
> not answered :-) Any Solaris installation CD will suffice for testing
> this, and there's little point in anyone going to the effort of building
> an OBP update CD for you if you can't boot an install CD.
Indeed, booting an install CD is the best way to assure that
a firmware upgrade from a CD-ROM will be possible. I do not now
if the problem in the OBP affects both hard disk drives and CD-ROM
drives - it would be nice to know.
> If you still get the deblocker error on attempting CD boot, we'll know
> the problem is not related to your disk.
In this case, I would certainly suggest following the fine advice
about a tftp firmware upgrade. It should be really easy to do, and the
network interface card seems working. These ARP/RARP timeouts happen
to me too, perhaps our ADSL routers are not sending reverse ARPs.
> It would also be useful to know if show-devs issued at the OBP prompt
> lists /packages/deblocker in the device tree - Mike asked this, again
> not answered.
Indeed, this information would be useful also.
> As I said, I have a working Ultra 2 and an interest in getting to the
> bottom of your issue because it appears to be quite unique. My U2 is not
> running anything important, so if I mess up my hardware in the process,
> no big deal :-)
Well... an Ultra2 is a fine machine, it is better not messing up
the computer. :-)
I will try to make a bootable flash upgrade CD-ROM. It will be a way
to check that I really moved the motherboard to its original position
after upgrading the firmware. ;-)
I will make that CD available on an anonymous FTP server for some days,
as soon as I check that it really boots on my Ultra2.
Cheers,
Igor.
Igor Sobrado
> Sunny wrote:
>>
>> Your Ultra 2 won't boot from disk - but will it boot from CD? Asked but
>> not answered :-) Any Solaris installation CD will suffice for testing
>> this, and there's little point in anyone going to the effort of building
>> an OBP update CD for you if you can't boot an install CD.
>
> Just done that now. It won't boot from the installation CD. Igor,
> please don't spend time making the CD image! Sorry I didn't run this
> test earlier, but I'm only just getting a grasp of the problem.
I am very sorry to read that.
But do not worry, you have a chance to do a network firmware upgrade yet.
Just set up a tftp server on the Windows PC and copy the firmware
upgrade executable to it. Perhaps you will need to rename it if there
are issues with large filenames between the Windows PC and the Sun
workstation. I would suggest renaming it to someone as "u2flash".
Perhaps Dan will provide a better feedback as he currently did a
network firmware upgrade on a Sun workstation.
> That one I did answer. Maybe the message went astray and didn't reach
> your server? /packages/deblocker is shown in the device tree.
What is the output of ".version" in the OBP?
Can you do a full hardware test? On the OBP change "diag-switch?" to true:
ok> setenv diag-switch? true
"diag-level" should be currently set to "max", if not:
ok> setenv diag-level max
(you can check the current settings typing "printenv" in OBP)
Now, you reboot the workstation:
ok> reset
and look at the output (do not worry about the ARP/RARP timeout messages).
Again, that is only to see if there is a hidden problem we do not know
yet. Your best chance is _doing a network firmware upgrade_.
Cheers,
Igor.
Igor Sobrado
>
> Again, that is only to see if there is a hidden problem we do not know
> yet. Your best chance is _doing a network firmware upgrade_.
I see that Dan Foster has provided a very detailed description of
the steps required to do a network firmware upgrade at the end of
this thread. I am sure, with these steps a new firmware upgrade
will be possible.
I wish the original poster the best luck with his Ultra 2 workstation!
Best regards,
Igor.
christop...@gmail.com
errors. I removed all the SCSI devices except the CD-Rom, but now the
machine appears to have died completely. No output on the monitor, and
no exchange of graphics cards, hard drives or power supplies will
resurrect it.
I could try connecting to the serial port or flashing the PROM over the
network, but this involves borrowing my father's PC. I might not be
able to do this, and I also wouldn't want to do anything which could
mess up his system. The U2 seems to be getting steadily worse so I'm
inclined to think flashing the PROM won't fix it.
Looks like I have a few options. Get a second hand U2 motherboard,
although I'd probably get better value buying a whole U2 for spares.
Get an Ultra 60 and populate it with the components from my Ultra 2
(I've always fancied one, and I need SCSI). Or build a PC with SCSI,
some variant of UNIX and possibly Windoze too. My immediate inclination
is to buy another U2 for spares. Any thoughts?
Best wishes,
Chris
Dan Foster
>
> Looks like I have a few options. Get a second hand U2 motherboard,
> although I'd probably get better value buying a whole U2 for spares.
> Get an Ultra 60 and populate it with the components from my Ultra 2
> (I've always fancied one, and I need SCSI). Or build a PC with SCSI,
> some variant of UNIX and possibly Windoze too. My immediate inclination
> is to buy another U2 for spares. Any thoughts?
Why not buy an Ultra 60? They're going for about USD $150-ish on eBay.
Granted, if you're tight on money, the U2 on eBay is going for about USD
$50.
Either will be decent options.
-Dan
Colin B.
> I tried setting "diag-level" and "diag-switch" but it didn't show any
> errors. I removed all the SCSI devices except the CD-Rom, but now the
> machine appears to have died completely. No output on the monitor, and
> no exchange of graphics cards, hard drives or power supplies will
> resurrect it.
If diag-level is set to max, it can spend as long as half an hour doing
testing before putting out anything to the video display. It may just be
a matter of waiting long enough. At any rate, you REALLY REALLY REALLY
need to get a serial console on this thing to do any meaningful diagnostics.
I don't understand why you haven't got a serial cable yet--they're cheap
and available everywhere. This will give you REAMS of information that will
help determine if you actually have a problem.
> Looks like I have a few options. Get a second hand U2 motherboard,
> although I'd probably get better value buying a whole U2 for spares.
> Get an Ultra 60 and populate it with the components from my Ultra 2
> (I've always fancied one, and I need SCSI). Or build a PC with SCSI,
> some variant of UNIX and possibly Windoze too. My immediate inclination
> is to buy another U2 for spares. Any thoughts?
Get a cable first! :-)
The U2 is a decent machine for cheap. Loaded up, it's relatively close to
the Ultra60 in many ways. A 60 or 80 is a really nice box, but if I were to
spend $400+ on a Sun system, I'd probably go with an Ultra20, and buy an
aftermarket SCSI card. That essentially falls into the category of a PC with
SCSI and some variant of Unix (Solaris10).
But get a serial console happening first. Please!
Colin
Dave (from the UK)
> The U2 is a decent machine for cheap. Loaded up, it's relatively close to
> the Ultra60 in many ways. A 60 or 80 is a really nice box, but if I were to
> spend $400+ on a Sun system, I'd probably go with an Ultra20, and buy an
> aftermarket SCSI card. That essentially falls into the category of a PC with
> SCSI and some variant of Unix (Solaris10).
Although an Ultra 20 is well over $400 and you don't have to spend
anything like $400 for an Ultra 60 or Ultra 80.
I paid £60 in the UK yesterday (around $107) for an Ultra 60 with:
1 x 450 MHz / 1 GB RAM / 18 GB disk / floppy /
And an Ultra 80 sold on eBay recently for $267, complete with two
processors.
Depressing really. I paid about $1500 for a single CPU once. I can't
recall what my Ultra 80 cost me, but perhaps that is a good thing.
christop...@gmail.com
> christop...@gmail.com wrote:
> > I tried setting "diag-level" and "diag-switch" but it didn't show any
> > errors. I removed all the SCSI devices except the CD-Rom, but now the
> > machine appears to have died completely. No output on the monitor, and
> > no exchange of graphics cards, hard drives or power supplies will
> > resurrect it.
>
> If diag-level is set to max, it can spend as long as half an hour doing
> testing before putting out anything to the video display. It may just be
> a matter of waiting long enough. At any rate, you REALLY REALLY REALLY
> need to get a serial console on this thing to do any meaningful diagnostics.
> I don't understand why you haven't got a serial cable yet--they're cheap
> and available everywhere. This will give you REAMS of information that will
> help determine if you actually have a problem.
Thanks very much for that suggestion, Colin. I left it for half an hour
and it did output some data to the monitor. It said "Power On Self Test
Failed. Cause: SysIO U1001". Opening the case I find that U1001 is a
large, square IC on the motherboard, so at first I assumed I needed a
new motherboard. I'm still not too hopeful, but I found a couple of
interesting posts in the Sun Managers archives:
http://www.sunmanagers.org/pipermail/sunmanagers/2002-February/010443.html
http://www.sunmanagers.org/pipermail/sunmanagers/2003-April/022504.html
It appears that he managed to get rid of the "SysIO U1001" error by
updating the OBP, but still had problems with the system (the messages
are a year apart, too). I e-mailed him to ask if he ever managed to
completely fix the machine, or if it turned out to be a hardware
failure. I don't want to be banging my head against a brick wall trying
to fix a hardware failure :-). My reluctance to get a null modem cable
and attack the problem that way is because I don't have a second Unix
box, and will have to borrow a PC for the experiment. I'm sure I'll end
up with a second Unix box in the future, but I haven't got one right
now. Any thoughts and best guesses as to whether or not I have a
hardware problem?
Many thanks,
Chris
Dave (from the UK)
> My reluctance to get a null modem cable
> and attack the problem that way is because I don't have a second Unix
> box, and will have to borrow a PC for the experiment.
I understand you don't have your own PC so need to borrow one and I can
understand your reluctance of risking messing up someone elses machine.
But you don't even need to install any software on the borrowed PC, as
Winblows has hyperterminal, which will do.
Start -> All programs -> Accessories -> Communications -> Hyperterminal
(on my XP pro PC anyway).
BUT, if the PC is newish, it might not even have a serial port!!
You will probably find a mate or neighbour has a really old PC you can
have. It would not even need a disk - a live version of Linux or Solaris
will be sufficent to read a serial port.
> I'm sure I'll end
> up with a second Unix box in the future, but I haven't got one right
> now.
But it is irrelevant.
If you really want to have your own machine to test it, and not spend
much money, I'll sell you a Sun SPARC 20 for £ 20 + postage. Nothing
special, but good enough for reading the serial port of your Ultra 2.
You could even use it to flash your prom, but someone else said you can
do that from winblows too.
But given the postage costs even on a SPARC 20 are not going to be
cheap, it would make far more sence to buy something like an Ultra 60.
I paid £ 60 + VAT for this one (18 GB disk / 1 GB RAM / 1 x 450 MHz CPU)
As you can see, it failed to attract any bids at £70, despite being on
eBay for a week.
You could use the fact it failed to get a bid at £70 to perhaps get
other dealers with them in the UK to be a bit more realistic on prices.
I believe the company has two more Ultra 60s, but I have no idea of the
cost. I posted their email earlier if you want to contact them.
This company in the UK has some interesting items on eBay, with sensible
starting prices, like an Ultra 10 starting at only £20, a Ultra 30 at
£30 and a Ultra 60 at £50.
http://search.ebay.co.uk/_W0QQsassZend-o-line_services_ltdQQhtZ-1
Of course, what they start at and what they sell for may be different,
but perhaps not.
> Any thoughts and best guesses as to whether or not I have a
> hardware problem?
I suspect you do have a hardware fault, but you are making hard work for
yourself.
Perhaps it is about time to buy yourself another machine and look at
possibly sorting out the Ultra 2 at a later date.
Sunny
Every diagnostic you try seems to lead to an error message - but the
errors point in different directions and confuse the issue :-(
That makes me suspect power supply or memory. You already swapped the
power supply, so I think it's time to rip out most of the memory to see
if the symptoms change.
Sunny
Mr R G Shepherd
My reluctance to get a null modem cable
> and attack the problem that way is because I don't have a second Unix
> box, and will have to borrow a PC for the experiment. I'm sure I'll end
> up with a second Unix box in the future, but I haven't got one right
> now. Any thoughts and best guesses as to whether or not I have a
> hardware problem?
You could throw a KNOPPIX disk in your dad's PC.
That mounts the disks on boot, so if you're fussed you can unmount them manually...
It has minicom on it, which i've used to get a serial console before.
It may also even have a tftp server on it?
If you've got net access then you can dl the obp image and net-boot-update poor
old U2.
Rob
I just checked. KNOPPIX does have in.tftpd on it.