arm64: lib: memory utilities optimization

Optimize memcpy and memmove, to prefetch several cache lines.
We can achieve 15% memcpy speed improvement with the preload method.

Signed-off-by: Hong-Mei Li <a21834@motorola.com>
Reviewed-on: http://gerrit.mot.com/740766


SLTApproved: Slta Waiver <sltawvr@motorola.com>
SME-Granted: SME Approvals Granted
Tested-by: Jira Key <jirakey@motorola.com>
Reviewed-by: Zhi-Ming Yuan <a14194@motorola.com>
Submit-Approved: Jira Key <jirakey@motorola.com>
Signed-off-by: Jesse Chan <jc@linux.com>
 - [Adapt for Exynos 8890 and copy_template]

Original commit: https://github.com/jesec/android_kernel_samsung_universal8890/commit/33b0b18b22798487e7e81ad1f00e57590c00b363