Re: SparcStation IPC: Data Access Exception

[Erik Fair <fair%netbsd.org@localhost>]

On Nov 25, 2014, at 08:04, Mouse <mouse%Rodents-Montreal.ORG@localhost> wrote:

>> The other thing that sparc firmware is picky about is the block/frag sizes f$
> 
> That's weird.  Do you have any idea what the mechanism for that is?  I thought the firmware was used only as a dumb "read this block” driver, and I have trouble imagining how that could care about filesystem frag or block size.

Block size exceeds expected/allocated buffer size, perhaps?

We now do things with these old systems that were not expected or insanely expensive when they were designed/built. A terabyte of disk for USD$50?

The other thing I’d try if the boot problem persists: zero the first part of the disk before install, i.e.

dd if=/dev/zero of=/dev/rsd0c bs=1m count=1

I’ve had this measure to ensure tabula rasa make the difference between inexplicable failure and boot success with sparcs - my uninvestigated presumption is the NetBSD installation procedure doesn’t always stomp completely on what was there before, and remnants can leave a mess the firmware doesn’t like.

Our lot in having to deal with the quirks and vagaries of extant system firmware that we can’t change is not a happy one. Worse, as a subsystem that is seldom used and not generally “user visible”, the Q/A and implementation quality of system firmware tends to be “just enough to get the system to boot” rather than as fully robust as we might prefer.

	Erik <fair%netbsd.org@localhost>