Subject: Re: storing nvaliases in NVRAM ('NVRAMRC not valid')
To: None <port-macppc@netbsd.org>
From: zeta.orionis <zeta.orionis@gmail.com>
List: port-macppc
Date: 04/16/2005 16:39:16
So is there anything I can do to get past the 'not valid' message?


On 16 Apr 2005, at 07:35, Jeff Walther wrote:

> At 14:12 +0200 04/15/2005, Schwerzmann, Stephan wrote:
>>>  Is resetting PRAM the same as resetting NVRAM, or are they two
>>>  different things?
>>
>> PRAM (=Parameter RAM) is apple-speak for the small memory built
>> using an NVRAM (=Non Volatile RAM) technology chip for the
>> purpose of storing a couple of parameters e.g. the ones relevant
>> for booting and a few more
>>
>> NVRAM is actually a CMOS/SRAM (Static RAM) technology flavor:
>> it is particularily low-power and suits applications where a
>> little bunch of data is to be kept over periods of no ac power
>>
>
> In the older machines, Beige G3 and earlier, the PRAM is stored in the 
> CUDA controller chip which also handles the ADB functions and contains 
> the real time clock.  Certain system settings are stored there.  This 
> chip is typically labeled 341S0788, although there is also a 341S0760 
> (I think) version.
>
> The NVRAM is a separate SRAM chip which has an 8KB capacity.  Apple 
> used several different pin compatible SRAM chips for this application, 
> from such manufactures as Sony, Samsung and Winbond. The NVRAMRC file 
> is stored in NVRAM and some other things.
>
> So PRAM and NVRAM reside in physically separate chips, and they are 
> used to store different items, although I don't know which items go 
> where in any kind of detail.
>
> I've tried replacing the 8K chip with a 32KB pin compatible chip, and 
> the machine works as before, but it still only sees 8KB of NVRAM.  I 
> guess the address space is built into the firmware or the wireing of 
> the interface to the NVRAM chip.
>
> I don't know how the later machines (Blue & White and later) handle 
> this storage.   Since they lack an ADB bus, (except the B&W) the PRAM 
> probably isn't in a CUDA chip any more.
>
> Jeff Walther
>