Subj : Re: Challenge: Multithreading & Synchronization
To   : comp.programming.threads
From : Chris Thomasson
Date : Fri May 27 2005 04:12 pm

>> How many existing processors today could trip the assertion?
>
>    In principle, any of them. There is nothing illegal about the compiler 
> implementing the CPU A loop as, in pseudo-code:
>
> shared+=5;
> usleep(1);
> shared=1;
>
>    It would violate no standard at all.

Agreed.




>    In practice, it is hard to imagine any existing processor compiler 
> could actually trip that assertion.

OK. Lets take "any compiler" out of situation, and just use a pseudo 
assembly language:



word_sized_var_t shared = 0
; assume the variable shared located at address 0x12345678
; and "fully initalized to zero" before any CPU executes the loops:


CPU A
--------

1: move $0x12345678 into generic register %r1
2: loop:
3: move $1 into variable that (%r1) points to
4: jump to loop




CPU B
--------

1: move $0x12345678 intp generic register %r1
2: loop:
3: move contents of (%r1) into generic register %r2
4: compare %r2 with $0
5: jump to loop if equal
6: compare %r2 with $1
7: jump to loop if equal
8: crash!




Lets say that the very generic instructions ( move, jump, compare ) and 
registers ( r1, r2 ) represented assembly language from any processors that 
exist today, how many of them could actually reach "crash!" in CPU B?

.