[rescue] Amiga videos online for your enjoyment

Charles Shannon Hendrix shannon at widomaker.com
Thu May 24 18:02:44 CDT 2007


Sridhar Ayengar wrote:
> Charles Shannon Hendrix wrote:
>>>> True, but that is due to the fact that with Sun Systems you get 
>>>> diagnostics that actually say something useful ... it's just that sun 
>>>> systems are build to a different requirement they are mostly build to 
>>>> run the same applications for years on end without being shutdown at all.
>>> Hearty agreement on the diags. I have had Suns tell me that the memory 
>>> in a specific slot is bad. Never seen that on any PeeCee.
>> I've also had to replace tons of RAM in Suns that their diagnostics never flagged.
>>
>> There is no way to test RAM reliably like Sun does it in their bootup code.
>>
>> They use a roll forward test that virtually guarantees it will cover up at
>> least half of the common RAM errors.
>>
>> Things like memtest86 are the only way to really see most RAM errors.
>>
>> Otherwise, yes, the Sun diagnostics are nice, but I have to say that I'm even
>> disappointed in most UNIX machine ROMs too.
>>
>> None of the ROMs in modern systems really does the kinds of things I would
>> like to see, and some of the UNIX stuff is 1970s cryptic, and honestly... why?
> 
> What sort of diags do you mean, specifically?  Which would you have 
> expected to see ubiquitously by now?

I don't just mean diags.

However, I would like to see diagnostics that just did a quick run through
devices to see if they respond to at least a minimal self-test, and it would
also be nice if the diagnostics were not so hard to use.

Sun's are funky, and some others are all-or-nothing affairs, when I often only
want to know one tiny thing.

Other improvements I'd like to see:

Not all of PC BIOS is bad.  It does some things very well, and makes most
things pretty easy.  UNIX vendors could learn a lot from it, if they'll
promise to ignore the mistakes.

Likewise, PC BIOS makers should look at UNIX ROM code and realize there are
better ways of doing things, and they also should be willing to support serial
terminals.  It's 2007, and there is more than enough ROM space to handle
several input methods easily, along with usable diagnostics.  memtest86 and
prime95 test sequences would be an excellent free addition to modern PC BIOS
for system testing.  In fact, if *I* were a PC maker like Dell, I would have
insisted on this stuff years ago.

Both of them should consider also having a command protocol so you can easily
write your own remote admin, and I don't mean the plethora of ill-concieved
and incompatible remote management stuff we have now.

Aside: I gave up on that stuff some time back... has any of that improved at
all?  Are we anywhere on the road to universal protocol to talk to hardware
both during run times and also boot?

Sun has some particular brain damage in their ROM and the whole boot and
install process.

Just for starters... if you configure a Sun PROM for 38400bps on the console,
why does it revert to 9600bps in the OS?  Can they not come out of the 1970s
and detect the right speed?  While I'm at it, why does an install on a
*HEADLESS* Sun still result in the system defaulting to a graphics console setup?

The biggest problem is probably user interface design, and inflexibility.

Most ROM monitors violate the principle of discovery rather badly.





-- 
shannon           | There is no such thing as security.  Life is either
                  | bold adventure, or it is nothing.
                  |         -- Helen Keller



More information about the rescue mailing list