Re: PATCH: PR target/53416: Wrong code when optimising loop involving _rdrand32_step

["H.J. Lu" <hjl dot tools at gmail dot com>]

On Sun, May 20, 2012 at 10:19 AM, Jakub Jelinek <jakub@redhat.com> wrote:
> On Sun, May 20, 2012 at 10:04:26AM -0700, H.J. Lu wrote:
>> rdrand<mode>_1 must be marked with unspec_volatile since it returns
>> a different value every time. ?OK for trunk, 4.7 and 4.6?
>
> A testcase for this would be nice (runtime is not possible, since the
> RNG in theory could return the same value twice, but scanning assembly
> for a particular number of the rdrand insns would be nice).
>

For

 unsigned int number = 0;
 volatile int result = 0;

 for (register int i = 0; i < 4; ++i) {
  result = _rdrand32_step(&number);
  printf("%d: %d\n", result, number);
 }

the issue isn't about number of rdrand insns.  As long as
it isn't hoisted out of the loop, one rdrand insn is OK.
I don't know how to scan for inside or outside of loop.

-- 
H.J.