C stack frame sizes

[Mike Lutz]

>> ... does the Berkeley RISC require a parallel stack,
>> maintained by software protocols, to hold structures, arrays, and other
>> humongous local variables?
>
>The RISC needs a memory stack anyway, because of the possibility that the
>"register stack" will overflow.

True, but the memory overflow stack is basically a 'shadow' of the
register stack, or, alternatively, one can view the register stack as a
cache for the most recently activated procedures.  The original RISC-I
papers from Berkeley imply this when discussing the software handlers
for register stack overflow/underflow.  Note that the plan to have the
registers addressable as memory locations agrees with the shadow/cache
model.

However, since the idea the register stack is designed for scalars and
pointers, I still think there is a need for a separate stack for large,
automatic, data structures - no?
-- 
Mike Lutz	Rochester Institute of Technology, Rochester NY
UUCP:		{allegra,seismo}!rochester!ritcv!mjl
CSNET:		mjl%rit at csnet-relay.ARPA