Re: [PATCH] ALSA: Remove subsystem-specific malloc (1/8)

[Dave Jones]

On Wed, Jun 09, 2004 at 10:08:34PM -0700, dean gaudet wrote:

 > > +void *kcalloc(size_t n, size_t size, int flags)
 > > +{
 > > +	if (n != 0 && size > INT_MAX / n)
 > > +		return NULL;
 > 
 > division isn't very efficient :)  it's typically a 40+ cycle operation.
 >
 > there's almost certainly more efficient ways to do this with per-target
 > assembly... but you should probably just crib the code from glibc which
 > avoids division for small allocations:
 > 
 >   /* size_t is unsigned so the behavior on overflow is defined.  */
 >   bytes = n * elem_size;
 > #define HALF_INTERNAL_SIZE_T \
 >   (((INTERNAL_SIZE_T) 1) << (8 * sizeof (INTERNAL_SIZE_T) / 2))
 >   if (__builtin_expect ((n | elem_size) >= HALF_INTERNAL_SIZE_T, 0)) {
 >     if (elem_size != 0 && bytes / elem_size != n) {
 >       MALLOC_FAILURE_ACTION;
 >       return 0;
 >     }
 >   }

Considering this isn't exactly a cycle-critical piece of code,
I'd choose readability over saving 40 cycles which is probably
lost in the noise of every other operation we do in that allocation.

		Dave

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/