[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [PATCH v1 04/19] include/fpu/softfloat: implement float
From: |
Richard Henderson |
Subject: |
Re: [Qemu-devel] [PATCH v1 04/19] include/fpu/softfloat: implement float16_set_sign helper |
Date: |
Mon, 8 Jan 2018 12:25:59 -0800 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101 Thunderbird/52.5.0 |
On 01/08/2018 04:58 AM, Alex Bennée wrote:
>
> Alex Bennée <address@hidden> writes:
>
>> Richard Henderson <address@hidden> writes:
>>
>>> On 12/11/2017 04:56 AM, Alex Bennée wrote:
>>>> +static inline float16 float16_set_sign(float16 a, int sign)
>>>> +{
>>>> + return make_float16((float16_val(a) & 0x7fff) | (sign << 15));
>>>> +}
>>>> +
>>>
>>> 1) Do we use this anywhere?
>>
>> Yes in the target specific helpers
>>
>>>
>>> 2) While this is probably in line with the other implementations,
>>> but going to a more qemu-ish style this should use deposit32.
>>
>> OK, will do.
>>
>
> It turns out doing this unleashes a weird circular dependency at we need
> qemu/bitops.h but that brings in host-utils.h and bswap.h which tries
> to include softfloat.h again.
Bah.
Just ignore this request for now then.
For future cleanup, I'm sure that bswap.h includes softfloat.h for the
float32/float64 typedefs. We should move those out somewhere else -- probably
qemu/typedefs.h. Which probably drops the number of objects that depend on
softfloat.h by a factor of 100.
r~