Re: Optimizing Across && And ||

bart@cs.uoregon.edu (Barton C. Massey) writes:
> I suspect that this sort of thing occurs often in macro expansions.
>
> For example, an implementation of the stdio putc() macro for a
> very popular OS expands to a complicated conditional expression
> including
> ((p)->_flag & _IOLBF) && -(p)->cnt < (p)->_bufsiz
> which is executed once each time an io buffer is output.


True, but at least some of those fields involved in the test are modified
by the I/O operation, so the next test must still reevaluate them. Because
most compilers don't keep aliasing or global data-flow information for
individual fields of structures, once you pass a pointer to the structure
to a function (like _flsbuf), the compiler must assume the whole structure
changed.


> People often write code like
> putc(x);
> putc(y);
> or even
> for(i=0; i<100000; i++) {
> putc(x[i]);
> putc(y[i]);
> }
> in which case there are some serious opportunities for
> optimization here.


Not really, unless your putc differs considerably from every other putc
I've ever examined.


> Of course, you'll now insist that I go find
> examples of this sort of putc() usage, in real programs, in
> places where it matters...


Nope, I wouldn't be that cruel. :-)


--
Bill Leonard
Harris Computer Systems Corporation
2101 W. Cypress Creek Road
Fort Lauderdale, FL 33309
Bill.Leonard@mail.hcsc.com
--

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