On 28.03.2011, at 19:55, Peter Maydell wrote:
> On 28 March 2011 18:23, Alexander Graf wrote:
>> On 03/24/2011 06:29 PM, Peter Maydell wrote:
+/* condition codes for binary FP ops */
+static uint32_t set_cc_f32(float32 v1, float32 v2)
+{
+if (float32_is_any_nan(v1) || flo
On 28 March 2011 18:23, Alexander Graf wrote:
> On 03/24/2011 06:29 PM, Peter Maydell wrote:
>>> +/* condition codes for binary FP ops */
>>> +static uint32_t set_cc_f32(float32 v1, float32 v2)
>>> +{
>>> + if (float32_is_any_nan(v1) || float32_is_any_nan(v2)) {
>>> + return 3;
>>> +
On 03/28/2011 10:23 AM, Alexander Graf wrote:
> I just checked the macros there and it looks like
> float32_compare_quiet returns eq when both numbers are NaN.
No it doesn't -- a == b will never be true for either operand as NaN.
Have a look at the "real" softfloat version anyway, not the
softflo
On 03/24/2011 06:29 PM, Peter Maydell wrote:
On 24 March 2011 15:58, Alexander Graf wrote:
This is more random comments in passing than a thorough review; sorry.
+#if HOST_LONG_BITS == 64&& defined(__GNUC__)
+/* assuming 64-bit hosts have __uint128_t */
+__uint128_t dividend
Am 24.03.2011 um 18:29 schrieb Peter Maydell :
> On 24 March 2011 15:58, Alexander Graf wrote:
>
> This is more random comments in passing than a thorough review; sorry.
>
>> +#if HOST_LONG_BITS == 64 && defined(__GNUC__)
>> +/* assuming 64-bit hosts have __uint128_t */
>> +__u
Am 24.03.2011 um 18:41 schrieb Richard Henderson :
> On 03/24/2011 10:29 AM, Peter Maydell wrote:
>> On 24 March 2011 15:58, Alexander Graf wrote:
>>
>> This is more random comments in passing than a thorough review; sorry.
>>
>>> +#if HOST_LONG_BITS == 64 && defined(__GNUC__)
>>> +/*
On 24 March 2011 15:58, Alexander Graf wrote:
This is more random comments in passing than a thorough review; sorry.
> +#if HOST_LONG_BITS == 64 && defined(__GNUC__)
> + /* assuming 64-bit hosts have __uint128_t */
> + __uint128_t dividend = (((__uint128_t)env->regs[r1]) << 64) |
>
On 03/24/2011 10:29 AM, Peter Maydell wrote:
> On 24 March 2011 15:58, Alexander Graf wrote:
>
> This is more random comments in passing than a thorough review; sorry.
>
>> +#if HOST_LONG_BITS == 64 && defined(__GNUC__)
>> +/* assuming 64-bit hosts have __uint128_t */
>> +__uint1
