SGI: Hardware

Can't get into POD on a Fuel - Revisited

JohnK
From The Swamp
Who joined July 2, 2015, 12:10 p.m.

Wrote the following at March 20, 2016, 6:55 a.m...

After my failure of trying to get into POD on my Fuel, I eventually tried the same procedure on a totally different Fuel (my second one).
See topic http://forums.nekochan.net/viewtopic.php?f=3&t=16730201

Code: Select all


SGI SN1 L1 Controller


Firmware Image B: Rev. 1.24.8, Built 09/15/2003 17:08:18






001a01-L1>debug


debug switches set to 0x0000


001a01-L1>debug 0x10d


debug switches set to 0x010d


001a01-L1>power u


001a01-L1>leds


CPU  A: 0x77: PLED_BIST_BASE:  Running BIST on bank 7


0x5c: PLED_TEST_INTS


0x70: PLED_BIST_BASE:  Running BIST on bank 0


0x71: PLED_BIST_BASE:  Running BIST on bank 1


0x3c: PLED_JUMPRAMU




001a01-L1>leds


CPU  A: 0x80:   POD Mode (0x80/0xBC=okay, solid 0x80=possibly hung polling UART)


0x8c: FLED_GENERAL:  General exception.




001a01-L1>


entering console mode  001a01 CPU0, <CTRL_T> to escape to L1 <<<<<-------------------------JUST HANGS FROM HERE

At this point there is nothing. I can go back and forth between console and L1 modes but that's it. When I try to type in console mode, I get this:

Code: Select all


entering console mode  001a01 CPU0, <CTRL_T> to escape to L1


no response from  001a01 CPU0, system not responding


no response from  001a01 CPU0, system not responding

I am assuming that once the debug switch is set to 0x010d that it should boot into POD directly.

Based on another thread about a Tezro having one of it's CPUs disabled, the same procedure is tried:

Code: Select all


001c01-L1>debug 0x10d (<-Setting DEBUG mode)


debug switches set to 0x010d


001c01-L1>pwr u


(wait about 15sec)


001c01-L1>leds


CPU A: 0xbc: POD Mode (0x80/0xBC=okay, solid 0xBC=possibly hung polling UART) (<- Seems bypassed CPU env check!!)


0x80: POD Mode (0x80/0xBC=okay, solid 0x80=possibly hung polling UART)


CPU B: < CPU not present >


CPU C: 0x45: PLED_LAUNCHLOOP: Slave loop (0x00/0x45=okay, solid 0x45=possibly hung)


0x00: PLED_RESET: Slave loop (0x00/0x45=okay, solid 0x00=possibly hung)


CPU D: < CPU not present >




001c01-L1> (<-Ctrl + D)


entering console mode 001c01 CPU0, <CTRL_T> to escape to L1




A 000 001c01: POD SysCt Cac> go cac <<<<<<---------------------------------------------------SO HOW DO I GET HERE?


A 000 001c01: Must be in Dex mode before switching to Cac or Unc.


A 000 001c01: POD SysCt Cac> clearalllogs


A 000 001c01: *** This must be run only after NUMAlink discovery is complete.


A 000 001c01: *** This will clear all previous log variables such as:


A 000 001c01: *** moduleids, nodeids, etc. for all nodes.


A 000 001c01: Clear all logs? [n] y


A 000 001c01: Checking 1 entries for promlogs


A 000 001c01: .DONE


A 000 001c01: All PROM logs cleared!

What am I missing here? Or is it that on a Fuel, its simply just "broken"? Has anyone ever been successful in doing this on a FUEL and if so, how did you do it?

I'm also trying to get some documentation on the use of the debug switches used on the L1 controller but the "L1 and L2 Controller Software User's Guide" doesn't venture into that territory. Is there a document that exists that does explain the debug switches?

SGI:

SUN: T2000 (x2], SunBlade 1500 Red (x2), SPARCstation 5 (x2) - Nextstep/Openstep, SPARCstation IPX
Alpha: DEC AXP 3000/300LX, API CS20 (Dual 833 Mhz CPUs)
Cobalt RaQ 550

nyef
Who joined April 28, 2015, 7:54 p.m.

Wrote the following at March 20, 2016, 10:16 a.m...

Fuel? IIRC, the serial console is routed to one of the serial ports on the back panel, NOT to the L1. Somewhat of a nasty surprise, really, and I don't know how to change it, but there it is. Power up via the L1, then use a second serial connection or unplug/re-plug to the other port to get at the console.

JohnK
From The Swamp
Who joined July 2, 2015, 12:10 p.m.

Wrote the following at March 21, 2016, 5:29 a.m...

Fuel? IIRC, the serial console is routed to one of the serial ports on the back panel, NOT to the L1. Somewhat of a nasty surprise, really, and I don't know how to change it, but there it is. Power up via the L1, then use a second serial connection or unplug/re-plug to the other port to get at the console.

Yep, there it is!
I thought that by doing a CTRL-D, the L1 would route me into the console mode but your right, it doesn't. It never occurred for me to physically switch to the serial port.Odd that it doesn't though since L1 says it can switch back and forth by using CTRL-D and CTRL-T.
Most of the time its the simple things isn't it?

Thank you much nyef! May the SGI gods be with you and I am forever grateful for your reply!

SGI:

SUN: T2000 (x2], SunBlade 1500 Red (x2), SPARCstation 5 (x2) - Nextstep/Openstep, SPARCstation IPX
Alpha: DEC AXP 3000/300LX, API CS20 (Dual 833 Mhz CPUs)
Cobalt RaQ 550