[SunHELP] sparc10 fails POST. what is wrong?

[Arthur Wouk]

 my sparc10 went quiet a bit ago, and i am trying to diagnose what is
wrong. 

sparc10, ross hypersparc 142mhz cpu, prom 2.25R, 256 megs ram,
turbogx framebuffer, quantum fireball 2.1gig hard drive.

i have done two types of test on it.

FIRST: i attached a (dumb terminal) sparc2 to port a, with monitor and
keyboard removed. then i powered on and captured the following
information:

># tip cuab
>connected

>           SMCC SPARCstation 10/20 UP/MP POST version VRV3.45 (09/11/95)


>CPU_#0       HyperSPARC ROSS RT620/RT626         0x00100000 Bytes ECache 

>CPU_#1       ******* NOT installed *******
>CPU_#2       ******* NOT installed *******
>CPU_#3       ******* NOT installed *******


>           >>>>> Power On Self Test (POST) is running .... <<<<<



>    <<< CPU_00000000 on MBus Slot_00000000 >>>
>     ERROR : Address = 00000000,  
>     exp[31:00] = aaaaaaaa, obs[31:00] = aaaaaa8a, xor[31:00] = 00000020Can't map device /eccmemctl at f,0

>Can't map device /iommu at f,e0000000

>Device not selected
>#1  hyperSPARC 97/01/28 15:20:13
>ok help
>Illegal Instruction
>ok ?
>108098c8 
>free-memory: bad address.
>ok 

one comment i received  to this is that, if i got this far in POST, then
the mbus module (cpu) is ok. is this plausible? it looks to me as
though the cpu was not reached.

SECOND: i forced a POST using stop-d  on the keyboard while powering
on. this from the sparc10 service manual. the four leds flashed on
twice, there was a delay, and then they flashed on twice again. then
the other three leds went off, but the caps lock led stayed on
permanently.

as i read the service manual, this means that the cpu, main logic
board, and dsimm in slot one are fine.

since it has been suggested that this is a memory problem, i pulled
all the dsimms, and tried several of them in first position (the ss10
has an order of installation). the behavior was unchanged. i pulled
the framebuffer as well - no change. i put them all back. again no
change.

the service manual gives no diagnosis for these events.

the service manual gives no diagnosis for these events.

does anyone have a clue? is this an nvram problem, or a prom problem
possibly? (i recently replaced the prom to upgrade from 2.22 to 2.25R.)