[weblacky]

Hi John,
OK, How about MORE left field (to the point we aren't even in the field anymore...like neighboring property, someone's calling the police because I'm so far on their property).

Does anyone have good pics of the inside of an L2...I ask because of this little item: https://irix7.com/techpubs/007-4364-006.pdf

This L2 Firmware update claims they NOW USE the RTC in this thing...wait RTC...wait...can't be...dead RTC causing NVRAM values possibly used at kernel startup? I mean it might make sense, they might store the startup location of the firmware for the kernel to boot in an NVRAM location instead of flash or RAM? I'm way out of guessing here, but I have issues explaining why you can start a Linux kernel lunch but cannot actually launch any initrd or anything. Wrong location would make sense?

I'm grasping at straws, is there a RTC IC in this thing that needs a battery hack to boot/reset correctly (must hold values using internal BATTERY during system, reset)? Or perhaps the internal board has a soldered-in CMOS battery kind of thing? Anyway, battery?

I'm throwing stuff out and seeing what sticks? Pics anyone?


BINGO; BATTERY!!!! https://bukosek.si/hardware/collection/s...nside3.jpg

Claims it's a: CR 1/2AA 3V lithium battery for L2 controller. Perhaps the wonky behavior is a symptom of dead battery? Worth a shot to me.

https://bukosek.si/hardware/collection/s...n3400.html

[jan-jaap]

You beat me to it. When I took my O350 home not so long ago I had a look inside the L2 and bought a replacement just in case :-)



This is the exact brand that was in my L2 already. CR14250 = CR1/2AA

[jwhat]

Hi Jan-Jaap & Weblacky,

again many thanks for tip, you will like my answer on this one....

The battery in the L2 is the same as that in some old Macs so I had a few that I had ordered for a Mac.

Having gone to the trouble of re-capping the Mac I was excited to turn the power on.... only to find that the Power Supply had died and also needing re-capping , AAHHH !!!

These old SGIs and Macs have more in common than you might initially think, when it comes to repairs.

So I have already done a swap of battery on the dead L2 so that is not cause of the problem in this case.

I have read Andrej Bukosek's previously and I am sure I read somewhere, that he undertook a major recap activity on 3400, but I cannot seem to find this again (so it maybe it was someone else).

It would be worthwhile to reach out to him on the subject of re-capping SGIs as he might have some valuable information.

Cheers from Oz,

John.

[weblacky]

Well, to be fair, no. The L2 is Linux PowerPC board so it's like the old Macs that were also PowerPC. That's it's not an SGI machine, per say.

Well that sucks, if you're sure that battery is good, then I got nothing. Tapping out...

[jwhat]

Hi Weblacky,

I found the wording in Andrej Bukosek blog, I was referring but it relates to Dallas DS1780E ICs not capacitors:


“The proper way to fix this is to replace all the DS1780E ICs with new ones (they're still being made). However, in an Origin, each brick has multiple of these (e.g. a C-brick has three on the motherboard plus two on each PIMM, an I-brick has three on the removable power section), so in a larger system, replacing about 65 surface-mount ICs is an incredibly tedious task.
Nevertheless, I have managed to replace all the faulty DS1780E ICs in my Origin (around 50 of them!) and have re-enabled environmental monitoring. I left the working DS1780E ICs alone, as it will be interesting to see if they also develop defects with time or not.”

So might be good lesson in dealing with surface mount replacements on SGI.

Cheers from Oz,

john.

[weblacky]

Oh yeah, I bought like 20 of those chips for future use when I get a fuel. I even have the right IC removal iron tip to pop them right off.

They are the fan and environment monitors for the L1. Yeah, I’ve never heard of a Tezro having those issues but you heard a lot from Fuels. I hadn’t remembered anyone had monitoring issues outside a fuel.

It’s not the worst job, the worst job would be trying to replace that megarray connector, I’d likely never get that right and aligned.

[weblacky]

jwhat - Did you ever get farther on this "boot loop" L2 issue? Seems like we never resolved this.

[jwhat]

jwhat - Did you ever get farther on this "boot loop" L2 issue?  Seems like we never resolved this.

Hi Weblacky,

as per last post I did a chip reprogram on dead L2 by copying contents of AMD AM29LV040B EEPROM chip from good L2 to bad L2

>> I opened up the L2 and it has a Motorola XPC860TZP50D3 32 Bit PowerPC chip and an AMD AM29LV040B EEPROM chip.
>> 
>> So I copied a good L2 EEPROM content onto the EEPROM from the bad (and non "flashsc --l2recover .." recoverable) L2, but even with this reflashed EEPROM it still fails to load kernel, so maybe the L2 has got bad RAM or something...

This failed to revive the bad L2, so I have put this L2 into the "to hard" basket.
So currently there is no known way to recover an L2 that is unable to boot into Linux BusyBox kernel.

I did also get DB-50 shells and plugs from Jan-Jaap, to use with my spare 50 Volt DC power adaptor, but have not had time to solder up cable yet.

Cheers from Oz,

jwhat/John

[nevyn]

Update: might as well post a pic of the connector so I won't have to dig this thing up later ;-)



I assume red is +48VDC and black GND, but double check. I'm not responsible.

Sorry for reanimating an old thread.

Since I'm perhaps about to go down this route myself, anybody got a PDF or pinout for the L2 controller power connector so I can double check before I let all the magic smoke escape? I did find some "preliminary" docs for the Origin 3000 series cabling but about the only useful info was the part number for the power brick to L2 controller (018-0858-001) - however without physically having that part and a pinout for the power brick itself it doesn't really tell me much...

[weblacky]

Hi Jan-Jaap, Weblacky & others,

I have dead L2 that is stuck in bootloader loop:

>> Linux/PPC load:
>> Uncompressing Linux...done.
>> Now booting the kernel
>> 
>> 
>> Validating L2 Controller Flash image....OK
>> Booting...
>> 
>> 
>> Ethernet address from Motorola VPD EEPROM is 08:00:69:11:B5:38
>> 
>> Linux/PPC load: ?                   <<=== see if there is some simple help built in....
>> Uncompressing Linux...done.
>> Now booting the kernel
>> 
>> 
>> Validating L2 Controller Flash image....OK
>> Booting...
>> 
>> 
>> Ethernet address from Motorola VPD EEPROM is 08:00:69:11:B5:38
>> 
>> Linux/PPC load:

I have tried the enter: "?" "help" "dir" "ls" and even "init=/bin/sh" at the load prompt but it just ignores the entry and continues on it cycle, entering anything stops the cycle temporarily until you press the return key.

So it is obviously going to some bootloader interpreter but I am not entering a valid command...

Any tips or guidance on this one ?

Cheers from Oz,

John

Jwhat,
So I thought about this a little, outside of just saying bad ram, I'm wondering if you can fool with the boot params here.  Can you check if "loglevel=7" at the Linux/PPC load prompt (no quotes) does anything?  Perhaps we can get some more output (assuming that kernel option is available).