I'm going to skip this one too.
On Tue, Sep 7, 2021 at 7:58 PM Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> wrote:
>
> From: Matteo Croce <mcroce@xxxxxxxxxxxxx>
> Subject: lib/string: optimized memcpy
>
> Patch series "lib/string: optimized mem* functions", v2.
Honestly, if we change the fallback memcpy(), I think the change
should be to remove it.
This is a core architecture thing, and every architecture does their
own. And pretty much every architecture has their own optimizations
for memcpy.
Yes, the byte-at-a-time default implementation is bad. But it's
_intentionally_ bad. It's only meant for initial bringup. No
architecture should actually end up using this in the long run, and if
you see it in profiles it should make you go "Ahh" instead.
Linus
