[gandtimo]

Hi there,
i've got a big problem with my beloved Onyx2.
Last week suddenly when starting the machine, no gfx were output. I connected via serial console and was greeted by this:

Code:

BASEIO PROM Monitor SGI Version 6.156  built 11:26:28 AM Nov 18, 2003 (BE64)
4 CPUs on 2 nodes found.
Installing PROM Device drivers ............
Base I/O Ethernet set to /dev/ethernet/ef0
Installing Graphics Console...
graphics install: searching for pipe 0
Checking out EEPROM
address = 87ff00, limit = 87ff30
EEPROM state = 2020202
Warning: Found Bad KONA eeprom: 0x2020202
       : Not Adding gfx device - gfxpipe
kl_graphics_install: installation failed for /hw/module/1/slot/io4/kona

I booted to single user and tried to /usr/gfx/startgfx. It started and gfxinfo recognised all boards (IR2E). irsaudit went through all tests without error.
I tried to reflash the eeprom and ...

Code:

# ./ireeprom -w -f /usr/gfx/ucode/KONA/tport.bin -F
==== Pipe 0 ====
Reading data from image...
Writing to EEPROM...
........
EEPROM code and data load complete...
Writing EEPROM header...
EEPROM header load complete...
Comparing EEPROM contents with image...
Error reading back EEPROM contents
EEPROM write failed

I tried everything I knew but nothing brought gfx back to live. I switched out the GE16-4 and everything worked fine until yesterday. Same problem again. 
This time, when I boot into single user and tried ireeprom -i after starting gfx, I get a good State, Reading the eeprom again directly after gave me a bad state. I'm confused and hope, anyone got a solution for this!

Code:

# /usr/gfx/KONA/bin/ireeprom -i
==== Pipe 0 ====
Manufacturer code:      0xf3
Device code:            0xb0
State:                  Good
Revision:               131
timestamp:              Fri Jul  1 04:22:03 2005
Code+data length:       0x1e2ec
Checksum:               0x00a38341
# /usr/gfx/KONA/bin/ireeprom -i
==== Pipe 0 ====
Manufacturer code:      0xf3
Device code:            0xb0
State:                  Bad
Revision:               269488275
timestamp:              Tue Jan 14 06:11:07 2014
Code+data length:       0x1011f2fc
Checksum:               0x10b39351
#

Greetings
Timo

[robespierre]

It certainly seems, from all that you have found, that this EEPROM has failed. Maybe you'll be lucky and find it is socketed.

[weblacky]

While I'm not familiar with this particular machine, the fact that you had it working with replacement board and then again had the same problem would lead me to believe that perhaps it's a voltage issue on the EPROM (indirectly) and not the chip itself?

I can't find good technical information on this stuff even in the Onyx2 handbook, I don't even know what ship on the board it is, but oftentimes chips require two different voltages. One voltage to read and operate normally and one voltage for programming mode.

However, I find it extremely odd that another board would suddenly have the same problem so I'm wondering if it uses a voltage from the bus for its programming mode and therefore it's actually a more widespread voltage issue or one of the rails is low and when it goes to use a voltage divider or regulator from another source for programming that maybe it gets the value wrong?

I could've sworn I saw an old post where somebody gently soldered very thin wires to the necessary pins of an installed EPROM and then put the board back in their system and watched the voltage level come up and down when they attempted to program it using the normal commands. That would be the only way I would think to confirm both the chip and that the proper programming voltage is being set. If you confirm the correct voltage is being set then I would say replacement of the chips sounds like a great idea. If you can confirm the voltage isn't correct then obviously replacing the chip isn't going to solve the problem.

[gandtimo]

While I'm not familiar with this particular machine, the fact that you had it working with replacement board and then again had the same problem would lead me to believe that perhaps it's a voltage issue on the EPROM (indirectly) and not the chip itself?

I can't find good technical information on this stuff even in the Onyx2 handbook, I don't even know what ship on the board it is, but oftentimes chips require two different voltages. One voltage to read and operate normally and one voltage for programming mode.

However, I find it extremely odd that another board would suddenly have the same problem so I'm wondering if it uses a voltage from the bus for its programming mode and therefore it's actually a more widespread voltage issue or one of the rails is low and when it goes to use a voltage divider or regulator from another source for programming that maybe it gets the value wrong?

I could've sworn I saw an old post where somebody gently soldered very thin wires to the necessary pins of an installed EPROM and then put the board back in their system and watched the voltage level come up and down when they attempted to program it using the normal commands.  That would be the only way I would think to confirm both the chip and that the proper programming voltage is being set. If you confirm the correct voltage is being set then I would say replacement of the chips sounds like a great idea. If you can confirm the voltage isn't correct then obviously replacing the chip isn't going to solve the problem.

I read about that possibility and if nothing works out, I'm going to try this. But I'm actually too afraid, that the GEs are fine and I don't want to harm them with my soldering skills

[Irinikus]

I've had two of these fail on me (two separate GE16-4's) and will have to look at purchasing new Atmel EEPROM chips as well as the required programming hardware. (It looks like the chips are starting to go bad!)

---

While I'm not familiar with this particular machine, the fact that you had it working with replacement board and then again had the same problem would lead me to believe that perhaps it's a voltage issue on the EPROM (indirectly) and not the chip itself?

I can't find good technical information on this stuff even in the Onyx2 handbook, I don't even know what ship on the board it is, but oftentimes chips require two different voltages. One voltage to read and operate normally and one voltage for programming mode.

However, I find it extremely odd that another board would suddenly have the same problem so I'm wondering if it uses a voltage from the bus for its programming mode and therefore it's actually a more widespread voltage issue or one of the rails is low and when it goes to use a voltage divider or regulator from another source for programming that maybe it gets the value wrong?

I could've sworn I saw an old post where somebody gently soldered very thin wires to the necessary pins of an installed EPROM and then put the board back in their system and watched the voltage level come up and down when they attempted to program it using the normal commands.  That would be the only way I would think to confirm both the chip and that the proper programming voltage is being set. If you confirm the correct voltage is being set then I would say replacement of the chips sounds like a great idea. If you can confirm the voltage isn't correct then obviously replacing the chip isn't going to solve the problem.

I won't rule out the possibility that it could be a problem on the 3V rail or a 3V voltage regulator on the GE16 itself, maybe it's a sign of the Power Supplies in these machines starting to go bad.