Re: Optimizing in assembly language

Scottie wrote:


>IIRC DEC/Mips used at least a bit of this approach. There were some
>problems, the most obvious of which were related to use of memory as
>semaphores. Code motion to improve memory bandwith removed full
>protection of critical sections.
> ...
> while (lock != 0)
> ...
> ;
> lock = 1
> ...update data with mutiple instruction...
> lock = 0
> ...
>Dead store elimination, if implemented, can remove locks on
>resources above entirely, leaving code like:
> ...
> while (lock != 0)
> ...
> ;
> ...update data with multiple instruction...
> lock = 0
> ...
>or even (on discovering the protect loop always exits with lock==0):
> ...
> while (lock != 0)
> ...
> ;
> update data with more than one indivisible instruction
> ...


That problem can be overcome easily by qualifying 'lock' as a
'volatile' variable, a variable of which the value can be changed
externally. That's very useful for e.g. memory mapped I/O,
semaphores:-), etc. Anything a program does to a volatile variable
should be left untouched by an optimizer.


        Eric

Return to the comp.compilers page.
Search the comp.compilers archives again.