[aarch64] use crc32 instructions when available

[sebpop]

The time goes from 0.175s down to 0.007s on a Graviton A1 instance on the
following benchmark:

function simple_crc32() {
  $a = "foo";
  for ($i = 0; $i < 10000; $i++) {
      crc32($a);
      $a .= "o";
  }
}

[peter279k]

Do you have any reference about execution time benchmark?

[sebpop]

I'm not sure what you are asking for, could you please be more specific?

The benchmark is a synthetic one that I run with the timing code from Zend/bench.php.

[peter279k]

@sebpop, thank you for your reply.

I want to know where the reference presents about this performance issue.

[sebpop]

Is there anything that needs to be addressed in this change, or is it ok to commit?
Thanks!

[sebpop]

I rebased the patch on master, and tested with make test.
I removed the alignment code from before the loop to reduce code size and no performance change.
@dstogov the patch is ready to be committed to master and php-7.4, thanks!

[nikic]

Merged as 2a535a9, thanks!