Re: memset.S and PowerPC

[Tom Gall <tom_gall at vnet dot ibm dot com>]

Hi Dave,

That reminds me.....

Actually I had done a rewrite that fixed that. Tho I hadn't convinced
myself
that it was ok from a performance perspective. Got busy with other
things....

You are right tho. the cacheline size is now surfaced by the kernel.

However a 3rd approach and probably an even better one was rolling
around that is exporting from the kernel a set of bla's one of which would

be memset.

Basically when you ran your kernel, the routine is surfaced so that
glibc etc can call it and thus get an optimized routine specific to
that architecture/environment. Good idea and something that's done
on other OS's tho I don't know that anyone actually started working
on it.

"David Müller (ELSOFT AG)" wrote:

> Hello
>
> AFAICS the memset.S in sysdeps/powerpc is broken for PowerPC
> implementations with cache line sizes other than 8 words.
> My current work-around is to delete this specific memset.S file. Is
> there a smarter way of doing it?
>
> I've heard that there are some ideas of using a variable (exported by
> the OS kernel) to give glibc a chance to adapt itself to the cache line
> size of the hardware. What's the status of this idea/project? Anybody
> using it already?
>
> TIA
>
> Dave

--
Tom Gall - PPC64, PPC32, Embedded "Where's the ka-boom? There was
Linux Technology Center           supposed to be an earth
(w) tom_gall@vnet.ibm.com         shattering ka-boom!"
(w) 507-253-4558                 -- Marvin Martian
(h) tgall@rochcivictheatre.org
http://www.ibm.com/linux/ltc