[rescue] Need help with SGI serial

[Jonathan C. Patschke]

On Fri, 25 Apr 2003, Joshua Newton wrote:

> "Data path test:	*FAILED*"
> "	RTC path test:	*FAILED*"

Hm.  I haven't seen those before.  Did you try reseating the RTC and
memory to see if it's just an intermittedn contact problem?

> Now, I spent a while rebooting it yesterday as I relearned IRIX -- the
> joys of fx (give me cfdisk! fdisk! even dd! well, maybe not dd)

fx isn't particularly difficult.  I still think OpenBSD's partition
editor (not the disklabel editor--the partition editor) takes the cake
for being the biggest PITA.  It won't even let you know about
overlapption partitions!

> I'm guessing the RTC message means it's time to replace the Dallas clock
> chip. Unfortunately, Dallas no longer seems to offer RAMified
> Timekeepers with 8k and response time of 150ns. The closest they have is
> 8k and 120ns. Any guesses as to what might or might not happen if I swap
> it for a 120ns chip?

I'd assume it'd work just fine.

> Or, just for kicks, a 32k/120ns chip?

Well, it probably won't use the additional 24k.  SGIs don't store a
whole lot in NVRAM.

> If I do have to replace the chip, what are the procedures for restoring
> eaddr and any other critical variables?

I can tell you that it's possible, but I couldn't tell you exactly how,
offhand.

> And, finally, does the "data path" message mean that something else is
> flaking out?

That's what's making me think your memory might not be properly seated.
Bad cache can also cause this.

-- 
Jonathan Patschke *) "who has been feeding my cats crack again??  they
Thorndale, TX     (*  were doing chuck norris shit all over the apartment
                  *)  this morning"                     --alex j avriette