Onyx rack start-up problem

[mosiniak]

Some time ago I bought Onyx rack. Now afrer cleaning it up I powerd it up. System started up (log below):

Code: Select all

IP25 SCC(E) SGI Version 6  built 10:11:49 AM May  8, 1996
R10000 2.6 194MHz BE (4-2-2/8) 2MB

Initializing hardware inventory...              ...done.
    CPU 05/00 is bootmaster
Testing Secondary Cache...                      ...passed.
Testing and clearing bus tags...                ...passed.
Configuring memory...
    Using standard interleave algorithm.
Running built-in memory test... 01 03
                                                ...passed.
Writing cfginfo to memory
Initializing MPCONF blocks
Checking slave processor diag results................
    Enabled 640 Megabytes of main memory
    Enabled 4 processors
Downloading PROM header information...
Downloading PROM code...
Jumping into IO4 PROM.

PROM Segment Loader (R10000 IP25) SGI Version 2.1 Rev A MIPS3,   Sep  3, 1996
Loading and executing R10000 boot prom image...

IO4 PROM Monitor SGI Version 4.21 Rev A IP25,   Sep  3, 1996 (BE64)
Sizing caches...
Initializing exception vectors.
Initializing IO4 subsystems.
Fixing vpids...
Initializing environment
sysctlr_getserial: got bad response char: '-1'
Piggyback reads enabled.
Initializing software and devices.
Cannot connect to keyboard -- check the cable.
Cannot open keyboard() for input
Cannot connect to keyboard -- check the cable.
Cannot open keyboard() for inAll initialization and diagnostics completed.
Bootmaster processor already started.
Starting processor #1
Starting processor #2
Starting processor #3
Checking hardware inventory...


System Maintenance Menu

1) Start System
2) Install System Software
3) Run Diagnostics
4) Recover System
5) Enter Command Monitor

Option?


I have three questions:
1) What does it mean "sysctlr_getserial: got bad response char: '-1'" I cannot find nothing in google.
2) There is no echo on serial console and I'm unable to select any option from 1 to 5. Should echo be visible? Are there any special requirements for serial console configuration and/or hardware (serial port configuration is: 9600 8N1; there is RS232-USB serial converter plugged in directly into onyx; PuTTy as a terminal emulator)?
3) Do I need any software on hdd or other media to perform the "Diagnostics"?

[pentium]

I know that in the past there was speculation that problems would start to arise once the Dallas NVRAM module on the system controller had its battery go flat. I've powered my machine up as a paranoid check (it disabled a CPU but that's not a problem right now) and mine still seems to be happy.
If the system controller loses its serial number that should give you the error you are seeing unless someone corrects me. That would mean your NVRAM battery has died.
I also seem to recall the Onyx like a few of the other earlier machines require a specially pinned serial cable for it to work properly. Refer to This Old SGI.
For the IP25 CPU board I seem to recall that yes there is an ide diagnostic for it but I don't remember it being available after a fresh install.

[robespierre]

1) What does it mean "sysctlr_getserial: got bad response char: '-1'"

While I could guess, I don't know what that means. Some problem with the System Controller... does it store a serial number in its Dallas timekeeper?

2) There is no echo on serial console and I'm unable to select any option from 1 to 5. Should echo be visible? Are there any special requirements for serial console configuration and/or hardware (serial port configuration is: 9600 8N1; there is RS232-USB serial converter plugged in directly into onyx; PuTTy as a terminal emulator)?

The Onyx (like all SGI computers before 1996) does not have industry-standard serial ports. The way you can tell is that they are female, whereas standard serial ports on Data Terminal Equipment are always male. If it is possible to directly plug a RS232-USB converter into an Onyx, you are screwed two ways: first, because it is certainly designed for the wrong serial port pinout; and second, because the fact that it is plugged into a female port means it expects the port to be Data Communications Equipment, which it is not.

3) Do I need any software on hdd or other media to perform the "Diagnostics"?

Yes, the diagnostics program is called "ide" and it is loaded from disk, CDROM, tape, or the network via TFTP. The option 3 for Diagnostics is basically a shortcut to try to load ide from the system disk and it is not necessarily the way you want to do it (especially if you don't have an installed system disk). You can run it in more ways from the command prompt (option 5).

[jan-jaap]

1) What does it mean "sysctlr_getserial: got bad response char: '-1'"

While I could guess, I don't know what that means. Some problem with the System Controller... does it store a serial number in its Dallas timekeeper?

Correct.

The system serial number is stored at least twice: once in a Dallas DS1287 in the System Controller and once in the DS1397 on the first IO4.
If one runs out and is replaced, the blank Dallas takes the serial# from the other. If both are valid but different, the System Controller wins.
If both run out, the serial is wiped.

THIS describes (sec. 6.3.3.5) how to set the serial# of an Onyx. In my experience it only works if the serial# is currently blank.

See HERE for an example of how to mod a System Controller DS1287 with a CR2032, and HERE for the same procedure on the IO4.

If you dump the contents of a Dallas battery backed RAM, the unused space returns 0xff. The MAC address of Indys with empty Dallas chips turns to ff:ff:ff:ff:ff:ff. That's probably happening here too, hence "invalid char: -1".

[mosiniak]

The Onyx (like all SGI computers before 1996) does not have industry-standard serial ports. The way you can tell is that they are female, whereas standard serial ports on Data Terminal Equipment are always male. If it is possible to directly plug a RS232-USB converter into an Onyx, you are screwed two ways: first, because it is certainly designed for the wrong serial port pinout; and second, because the fact that it is plugged into a female port means it expects the port to be Data Communications Equipment, which it is not.

You are absolutely right. I assumed that this is ordinary RS232 and I was wrong. For example: in ordinary RS232 port GND is on pin 5, while in onyx GND is on pin 6. I'll make the adapter cable and we will see if this caused the echo problem.

[jan-jaap]

3) Do I need any software on hdd or other media to perform the "Diagnostics"?

The Onyx performs fairly comprehensive diagnostics when it is powered on.

"Big iron" diagnostics are not included with IRIX, like for the workstations. The diagnostics version and the IRIX version it is installed on must be an exact match.

The diagnostics are split into two sets: the standalone diagnostics and the IRIX based diagnostics. Option #3 from the PROM menu would boot the standalone diagnostics which mostly exercise CPUs, memory etc. If you install the big iron diagnostics, a user 'diag' is added and if you log in as 'diag' you get a menu based interface to the IRIX based diagnostics which exercise the Reality Engine, video and some other option boards. Oh, and the IR diagnostics (irsaudit) are part of IRIX again...

I have some of these Diagnostics discs (for IRIX 5.3, 6.0 and 6.2, but not 6.5 - hint hint).

One thing to be aware of with the Onyx/Challenge series is there are many revisions of each of the boards that make up the system. You need certain revisions of of MC and IO boards to use IP25 (R10K) CPU boards, to use IR graphics etc etc. Some of these incompatibilities are subtle; a certain rev. of a board may work in one system but not the next. It may misbehave but only under certain load patterns. See: http://www.sgistuff.net/mirrors/sgi/Ony ... gine2.html

So, if you find yourself having to replace a board in an Onyx 10000 IR (which needs latest revs of everything), then it would make sense to run the diagnostics afterwards to make sure is healthy.

[jan-jaap]

Oh, since you didn't realize the serial ports are wired in a non-standard way: beware of high voltage differential SCSI inside the Onyx. That is not the same as low voltage differential (LVD) SCSI. If you connect an SE or LVD device to a HVD bus, magic smoke will escape.

Normally, one of the SCSI buses is configured HVD, and one is set up as SE SCSI. It is possible to reconfigure the mode of a SCSI bus by changing a little daughter card on the IO4. Red=HVD, green=SE. You need to change something on the drive box backplane as well, and re-jumper all the drive sleds.

The system disk is normally on the HVD bus, a CDROM or tape drive on the SE bus. Both buses are available in the drive bays, and jumpers on the sled select which bus a device connects to. The Onyx user guide explains all of this.

[mosiniak]

Thanks for all information. Today I built special console cable for Onyx (schematics below) and console is alive.

Code: Select all

DE9M            DE9F
(Onyx side)     (terminal side)

9 (DTR) ---+--- 1 (DCD)
           |--- 6 (DSR)
2 (TXD) ------- 2 (RXD)
3 (RXD) ------- 3 (TXD)
8 (DCD) ------- 4 (DTR)
7 (GND) ------- 5 (GND)
5 (CTS) ------- 7 (RTS)
4 (RTS) ------- 8 (CTS)

case    ------- case


Oh, since you didn't realize the serial ports are wired in a non-standard way

Yesterday I looked for Onyx serial port pinout but I onlny focused on TXD and RXD pins - they were on right positions (for female DCE port), but other pins were scrambled and I didn't notice that...
Now, there is echo and I can enter Command Monitor. Some data below

Code: Select all

System Maintenance Menu

1) Start System
2) Install System Software
3) Run Diagnostics
4) Recover System
5) Enter Command Monitor


Option? 5
Command Monitor.  Type "exit" to return to the menu.

>> printenv
AutoLoad=No
dbgtty=multi(0)serial(0)
root=dks0d1s0
nonstop=0
rbaud=19200
SystemPartition=dksc(0,1,8)
OSLoadPartition=dksc(0,1,0)
OSLoader=sash
OSLoadFilename=unix
TimeZone=MET-1DST
console=G
diskless=0
dbaud=9600
sgilogo=y
netaddr=128.1.1.26
cpufreq=194
NoAutoLoad=CONSOLE OPEN FAILED.
ConsoleOut=multi(0)serial(0)
ConsoleIn=multi(0)serial(0)

>> version
IO4 PROM Monitor SGI Version 4.21 Rev A IP25,   Sep  3, 1996 (BE64)

>> ls
dksc(0,1,8)/: no such device
dksc(0,1,0)/: no such device

>> hinv -v
                   System: IP25
                Processor: 194 Mhz R10000, 2M secondary cache
     Primary I-cache size: 32 Kbytes
     Primary D-cache size: 32 Kbytes
     Secondary cache size: 2 Mbytes
                Processor: 194 Mhz R10000, 2M secondary cache, (cpu 1)
     Primary I-cache size: 32 Kbytes
     Primary D-cache size: 32 Kbytes
     Secondary cache size: 2 Mbytes
                Processor: 194 Mhz R10000, 2M secondary cache, (cpu 2)
     Primary I-cache size: 32 Kbytes
     Primary D-cache size: 32 Kbytes
     Secondary cache size: 2 Mbytes
                Processor: 194 Mhz R10000, 2M secondary cache, (cpu 3)
     Primary I-cache size: 32 Kbytes
     Primary D-cache size: 32 Kbytes
     Secondary cache size: 2 Mbytes
              Memory size: 640 Mbytes
                SCSI Tape: scsi(1)tape(3)
                 Graphics: InfiniteReality Graphics

>> hinv -t
system ARC SGI-IP25 key 0
  processor CPU MIPS-R10000 key 0
    processor FPU MIPS-R10000FPC key 0
    cache primary icache 32 Kbytes (block 1 lines, line 16 bytes)
    cache primary dcache 32 Kbytes (block 1 lines, line 16 bytes)
    cache secondary cache 2048 Kbytes (block 1 lines, line 128 bytes)
  processor CPU MIPS-R10000 key 1
    processor FPU MIPS-R10000FPC key 1
    cache primary icache 32 Kbytes (block 1 lines, line 16 bytes)
    cache primary dcache 32 Kbytes (block 1 lines, line 16 bytes)
    cache secondary cache 2048 Kbytes (block 1 lines, line 128 bytes)
  processor CPU MIPS-R10000 key 2
    processor FPU MIPS-R10000FPC key 2
    cache primary icache 32 Kbytes (block 1 lines, line 16 bytes)
    cache primary dcache 32 Kbytes (block 1 lines, line 16 bytes)
    cache secondary cache 2048 Kbytes (block 1 lines, line 128 bytes)
  processor CPU MIPS-R10000 key 3
    processor FPU MIPS-R10000FPC key 3
    cache primary icache 32 Kbytes (block 1 lines, line 16 bytes)
    cache primary dcache 32 Kbytes (block 1 lines, line 16 bytes)
    cache secondary cache 2048 Kbytes (block 1 lines, line 128 bytes)
  memory main 640 Mbytes
  adapter SCSI WD33C95A key 0
  adapter SCSI WD33C95A key 1
    controller tape ARCHIVE Python 01931-XXX key 3
      peripheral tape unit 0
  controller display SGI-InfiniteReality Graphics key 0
  adapter multi function EPC1.0 key 0
    peripheral serial EVEREST TTY key 6
    controller network et0 key 0
      peripheral network key 0
    controller serial IO4 tty key 0
      peripheral line key 0
    controller serial IO4 tty key 1
      peripheral line key 0
    controller serial IO4 tty key 4
      peripheral line key 0
    controller serial IO4 tty key 5
      peripheral line key 0

>> hinv -b
System Bus Information:
   Slot 1: MC3 memory board with 320 megabytes of memory (Enabled)
   Slot 3: MC3 memory board with 320 megabytes of memory (Enabled)
   Slot 5: IP25 processor board with 4 cpus (Enabled)
   Slot 7: IO4 I/O peripheral controller board (Enabled)
   Slot 9: IO4 I/O peripheral controller board (Enabled)
   Slot 11: IO4 I/O peripheral controller board (Enabled)


I also noticed that after powering the Onyx on and off I can see current date on LCD (23 Nov). Date wasn't set by me so the NVRAM battery must be OK and clock is ticking - right? I cannot see any errors (except "sysctlr_getserial: got bad response char: '-1'") so I think that everything is OK. There is no system drive so I have to install one and I will try with OS.

One thing to be aware of with the Onyx/Challenge series is there are many revisions of each of the boards that make up the system. You need certain revisions of of MC and IO boards to use IP25 (R10K) CPU boards, to use IR graphics etc etc. Some of these incompatibilities are subtle; a certain rev. of a board may work in one system but not the next. It may misbehave but only under certain load patterns. See: http://www.sgistuff.net/mirrors/sgi/Ony ... gine2.html

I think that my configuration is an original working configuration and there will be no problems, but thanks for warning.

One more thing:

Normally, one of the SCSI buses is configured HVD, and one is set up as SE SCSI.

I inspect the SCSI conections and this is what I found:
i) there are three IO4 boards and 6 SCSI buses. Two buses are connected to rear IO panel and 4 other buses are connected internally (I suppose to front disk cages).
ii) One of these four buses is SE (green) and 3 are DIFF (red).
iii) All drive sleds in one box are configured to SE (both channels) and all drive sleds in second box are configured to DIFF (both channels).
If I'm right there is something wrong with this configuration isn't in?

One more time thanks for help. There is a lot of work and fun to get this machine fully operational.