[PATCH v2] aarch64: Optimized memset for Kunpeng processor.

[Zhangxuelei (Derek)]

Hi Wilco,

> This won't work, when you align like that you need an extra 
> unaligned store at the end.
> The easiest solution is not to align and just keep the stp 
> and str using dstin.
>
> +2:     stp     q0, q0, [dst, 32]
> +       stp     q0, q0, [dstend, -32]
> +       ret
>
> Both of these must use dstend since this is the 1-64 bytes for 
> the tail part.

Sorry for my mistakes, and I have corrected them in new patch.

Cheers
Xuelei