On Tue, 2007-08-14 at 16:23 +0530, pradeep singh wrote:
> On 8/14/07, arshad hussain <arshad.super@xxxxxxxxx> wrote:
> [snip]
> > > then why is memcpy present in the sources can't we
> > > simply do
> > >
> > > "#define memcpy memmove" in include/linux/string.h
> > >
> > > or am I missing something?
>
> I don't know but memcpy generates better asm code AFAIK.
> Essentially memove is nothing but
> int memmove(void *dest, void *src, size_t n)
> {
> //some checks
> //some adjustments to src and destination as they may overlap as per
> //definition.
>
> memcpy(dest, src, n);
>
> }
It's more like:
if (dest <= src)
do what memcpy does
else
do basically memcpy, just backwards
If you know that you don't have overlap, you can save yourself a
conditional jump. Not the biggest optimization, but it can add up if
you call it often, and it's easy, so you might as well take advantage of
it.
Avishay
--
To unsubscribe from this list: send an email with
"unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx
Please read the FAQ at http://kernelnewbies.org/FAQ
