Onyx startup problem. Segment load failed. Solved

[bjornl]

Hello,
I got myself an Onyx. I was told the battery was bad and I had to go into prom and set time and date to get it to boot.
So I set it up with a VT100 emulator on the TTY1 port to get it up and running. However I only got this

Code: Select all

IP19 PROM (BE) SGI Version 13  built 02:03:53 AM Aug 29, 1993
Checking system endianess...                    Big endian
Initializing hardware inventory...              ...done.
    CPU 02/00 is bootmaster
Testing and clearing bus tags...                ...passed.
Configuring memory...
    Using standard interleave algorithm.
Running built-in memory test... 01
                                                ...passed.
Writing cfginfo to memory
Initializing MPCONF blocks
Checking EAROM...                               ...passed.
Testing secondary cache...                      ...passed.
Checking processor diag results...
    Enabled 256 Megabytes of main memory
    Enabled 4 processors
Downloading PROM header information...
Jumping into IO4 PROM.

PROM Segment Loader (R4400 IP19) SGI Version 2.1 Rev A MIPS3,   Sep  3, 1996
Loading and executing R4400 boot prom image...

ERROR: Decompresed length (0x0007a572) != stored length (0x0007a5b0)
Decompression failed!
Segment load failed.  Can't execute image.

Everest PROM Segment Loader

 L) List segments
 P) POD mode
 R) Reset the machine
 S) Select a slot
 X) Load and execute a segment

How serious is this and how do I fix it?
It is maybe time to reseat all boards. The Onyx did travel a 'long' way in the back of my car so something might have come loose.

/Bjorn

[deBug]

Hello,
I got myself an Onyx. I was told the battery was bad and I had to go into prom and set time and date to get it to boot.
So I set it up with a VT100 emulator on the TTY1 port to get it up and running. However I only got this

Code: Select all

IP19 PROM (BE) SGI Version 13  built 02:03:53 AM Aug 29, 1993
Checking system endianess...                    Big endian
Initializing hardware inventory...              ...done.
    CPU 02/00 is bootmaster
Testing and clearing bus tags...                ...passed.
Configuring memory...
    Using standard interleave algorithm.
Running built-in memory test... 01
                                                ...passed.
Writing cfginfo to memory
Initializing MPCONF blocks
Checking EAROM...                               ...passed.
Testing secondary cache...                      ...passed.
Checking processor diag results...
    Enabled 256 Megabytes of main memory
    Enabled 4 processors
Downloading PROM header information...
Jumping into IO4 PROM.

PROM Segment Loader (R4400 IP19) SGI Version 2.1 Rev A MIPS3,   Sep  3, 1996
Loading and executing R4400 boot prom image...

ERROR: Decompresed length (0x0007a572) != stored length (0x0007a5b0)
Decompression failed!
Segment load failed.  Can't execute image.

Everest PROM Segment Loader

 L) List segments
 P) POD mode
 R) Reset the machine
 S) Select a slot
 X) Load and execute a segment

How serious is this and how do I fix it?
It is maybe time to reseat all boards. The Onyx did travel a 'long' way in the back of my car so something might have come loose.

/Bjorn

Congrats to the Onyx Björn!

Well it not much you can do software wise as the boot strap is in the PROM code that cant be loaded.

The easiest thing would probably be to try it with another IP19 board.

[bjornl]

Congrats to the Onyx Björn!

Thank you

Well it not much you can do software wise as the boot strap is in the PROM code that cant be loaded.

The easiest thing would probably be to try it with another IP19 board.

Not the reply I was hoping for but it might be the only solution. It is not a kind of spare card I have lying around

I've been doing some reading and as far as I understand there is one PROM for each CPU. I was thinking 'they can't all be bad' so maybe I could get it running by disabling/enabling CPUs until I found the bad one. If that is possible considering how far I got in the boot process.
I guess it will boot from CPU2 if i disable CPU1 ... or ...?

Do anyone know a better mirror for 'Crimson/Power Series/Onyx RealityEngine/RealityEngine2 Info' at http://www.gisparks.tas.gov.au/sgi/ than the http://www.archive.org? The archive haven't got all pictures.

[jan-jaap]

The problem of your Onyx might be with the IO4, not the CPU card.

As you said, it has one PROM/CPU, but it also has a PROM on the IO4. This is also true for older systems (PowerSeries) btw.

In the log, it seems to pass the CPU diagnostics succesfully, and the bad things happen after it says "Jumping into IO4 PROM."

I think the IO4 is flashed when you upgrade the OS. But I'm afraid you'd need a working system capable of holding multiple IO4's (deskside Challenge, rack Challenge / Onyx) to do it.

Did you check out the "Challeng / Onyx diagnostics roadmap on Ian Mapleson's site?
I think step two in this diagram is where yours goes wrong:

[thegoldbug]

I haven't seen this error message either but I recall some odd ones when I when from an R10K setup Challenge L to an R4400 setup..

[SAQ]

How is the IO4 PROM connected? could it be pulled and programmed out of circuit?

[bjornl]

The problem of your Onyx might be with the IO4, not the CPU card.

As you said, it has one PROM/CPU, but it also has a PROM on the IO4. This is also true for older systems (PowerSeries) btw.

In the log, it seems to pass the CPU diagnostics succesfully, and the bad things happen after it says "Jumping into IO4 PROM."

I think the IO4 is flashed when you upgrade the OS. But I'm afraid you'd need a working system capable of holding multiple IO4's (deskside Challenge, rack Challenge / Onyx) to do it.

Did you check out the "Challeng / Onyx diagnostics roadmap on Ian Mapleson's site?
I think step two in this diagram is where yours goes wrong:

You have a point there. To bad the IO4 is as rare as the IP19, which make them expensive.
The problem is that I have not tried to upgrade the OS, so there have been no flashing on my part. The Onyx still has it's original IRIX6.2.

One way to go would be, finding a rack, insert my IO4 into it and use flash command to restore the PROM to a correct state.
Do you think a 'initalllogs', 'clearalllogs' in the POD could help?

I have taken a look at Ian's roadmap to try and see what could be wrong, but your guess is probably better than mine

/Bjorn

[bjornl]

How is the IO4 PROM connected? could it be pulled and programmed out of circuit?

That is a very interesting thought.

[SAQ]

Depending on how hard toodling over to the southern bit of northern Europe is, Jan-Jaap probably could help you with it somehow.

[jan-jaap]

Depending on how hard toodling over to the southern bit of northern Europe is, Jan-Jaap probably could help you with it somehow.

Well, the flash software and firmware is indeed on the CDROM (IRIX 6.2, disc1, /firmware), but it's an ELF executable, so I assume it has to be run from IRIX. You need a functional IO4 to boot the system into IRIX.

So, what you need is someone with a system that can hold more than one IO4, like a rack or a deskside challenge. Install the misbehaving IO4 as a secondary, and see if you can flash it back to life. That's assuming this IO4 'magically' corrupted it's PROM and isn't just really broken in one way or the other.

All I have is a deskside Onyx, and those take only a single IO4.

[bjornl]

Any idea what can cause this magically corrupted PROM?
Could some static charge have found it's way from outside the chassi into the PROM?

Now I have removed and reinserted IP19 and IO4, just to see if it helped. It didn't, surprise ... NOT.

I tried also L) List segments, and got a list of PROM images.
Then I tried X) Load and execute a segment. First with the appropriate R4400 image, it failed of course.
The other images (R8000 and R10000) failed with the same error message 'ERROR: Decompresed length (0x0007a572) != stored length (0x0007a5b0)' (but different sizes). How can all images lengths be wrong?? I rather exptected some kind of 'Wrong cpu type' message.

I have heard of a university here, that should have an Onyx rack. If they still have it, are willing to help me (risking their own machine), and there is a guru that can do it, it is worth a shot. A real longshot

[bjornl]

Well, no luck in the Onyx rack department apart from they want me to visit and remove it from their storage

Back to the deskside ...
Can anyone tell me where this PROM chip is physically? Is it socketed for easy removal (and replacement)?
There is a nice picture at the SGIstuff homepage
http://storage.g-lenerz.de/images/sgist ... 4+scip.jpg

I haven't tried this yet, but could a 'zap' help? http://www.futuretech.blinkenlights.nl/ ... ampwd.html
There is some interesting VR software in the machine and I'm afraid I'll make it useless.
On the other hand, the entire machine is useless right now.

[deBug]

Well, no luck in the Onyx rack department apart from they want me to visit and remove it from their storage

Back to the deskside ...
Can anyone tell me where this PROM chip is physically? Is it socketed for easy removal (and replacement)?
There is a nice picture at the SGIstuff homepage
http://storage.g-lenerz.de/images/sgist ... 4+scip.jpg

I haven't tried this yet, but could a 'zap' help? http://www.futuretech.blinkenlights.nl/ ... ampwd.html
There is some interesting VR software in the machine and I'm afraid I'll make it useless.
On the other hand, the entire machine is useless right now.

They can not be seen in your photo as they are covered by the two SCSI Mezaine boards.

[bjornl]

I tried the 'zap' command in POd mode but all I got was this

Code: Select all

NVRAM inventory is already invalid

Does anyone know how to interpret that?

Can anyone tell me where this PROM chip is physically? Is it socketed for easy removal (and replacement)?
There is a nice picture at the SGIstuff homepage
http://storage.g-lenerz.de/images/sgist ... 4+scip.jpg

They can not be seen in your photo as they are covered by the two SCSI Mezaine boards.

OK. I haven't got a better picture but I have checked the IO4 and there are a lot of chips with partnumbers beginning with 070-. Could it be one of these?
Nothing shows up when I Google these numbers.

[deBug]

I tried the 'zap' command in POd mode but all I got was this

Code: Select all

NVRAM inventory is already invalid

Does anyone know how to interpret that?

Can anyone tell me where this PROM chip is physically? Is it socketed for easy removal (and replacement)?
There is a nice picture at the SGIstuff homepage
http://storage.g-lenerz.de/images/sgist ... 4+scip.jpg

They can not be seen in your photo as they are covered by the two SCSI Mezaine boards.

OK. I haven't got a better picture but I have checked the IO4 and there are a lot of chips with partnumbers beginning with 070-. Could it be one of these?
Nothing shows up when I Google these numbers.

I don't know witch one is the prom memory but you should be able to figure it out by looking at the circuit numbers and google them and find out witch one is a prom circuit.