ARM: More intrinsics

[Andrew Haley]

On 03/15/2012 04:09 AM, pcpa at mandriva.com.br wrote:
>> Xerxes noticed an inconsistency in my patch.  I've removed the unused
>> masks, and redefined the condition in the atomic operations to use ETT
>> (else, true, true) rather than the confused encoding I used before.
>>
>> So, it's now
>>
>> +      it(codebuf, COND_EQ, IT_MASK_ETT);
>>
>> instead of
>>
>> -      it(codebuf, COND_NE, IT_MASK_TEE);
> 
>   I feel there is something bogus here, because the first condition
> should be 'T' always, and the mask is based on bit 0 of the test
> condition. Talking this because I wrote support for arm thumb jit
> recently.

I've now got this:

int it(CodeBuf *codebuf, unsigned cond, unsigned mask)
{
  if (cond & 1) {
    // If this is a negated condition, flip all the bits above the
    // least significant bit that is 1.  Note that at least one bit is
    // always 1 in mask
    switch (mask & (-mask)) {
    case 8:
      break;
    case 4:
      mask ^= 8;
      break;
    case 2:
      mask ^= 0x0c;
      break;
    case 1:
      mask ^= 0x0e;
      break;
    default:
      // Impossible unless someone specified an incorrect mask
      longjmp(compiler_error_env, COMPILER_RESULT_FAILED);
    }
  }

  return out_16(codebuf, T_IT(cond, mask));
}

Which does the right thing per the spec and is simpler.

Thanks for the idea,

Andrew.