Re: [PATCH 1/7] drm: Fix drm_fixp2int_round() making it add 0.5

[Harry Wentland <harry.wentland@amd.com>]

On 2024-03-14 09:31, Melissa Wen wrote:
> On 03/14, Melissa Wen wrote:
>> On 03/13, Arthur Grillo wrote:
>>>
>>>
>>> On 12/03/24 15:27, Melissa Wen wrote:
>>>> On 03/06, Arthur Grillo wrote:
>>>>> As well noted by Pekka[1], the rounding of drm_fixp2int_round is wrong.
>>>>> To round a number, you need to add 0.5 to the number and floor that,
>>>>> drm_fixp2int_round() is adding 0.0000076. Make it add 0.5.
>>>>>
>>>>> [1]: https://lore.kernel.org/all/20240301135327.22efe0dd.pekka.paalanen@collabora.com/
>>>>>
>>>> Hi Arthur,
>>>>
>>>> thanks for addressing this issue.
>>>>
>>>> Please, add a fix tag to the commit that you are fixing, so we can
>>>> easily backport. Might be this commit:
>>>> https://cgit.freedesktop.org/drm/drm-misc/commit/drivers/gpu/drm/vkms?id=ab87f558dcfb2562c3497e89600dec798a446665
>>>>> Suggested-by: Pekka Paalanen <pekka.paalanen@collabora.com>
>>>>> Signed-off-by: Arthur Grillo <arthurgrillo@riseup.net>
>>>>> ---
>>>>>  include/drm/drm_fixed.h | 2 +-
>>>>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>>>>
>>>>> diff --git a/include/drm/drm_fixed.h b/include/drm/drm_fixed.h
>>>>> index 0c9f917a4d4b..de3a79909ac9 100644
>>>>> --- a/include/drm/drm_fixed.h
>>>>> +++ b/include/drm/drm_fixed.h
>>>>> @@ -90,7 +90,7 @@ static inline int drm_fixp2int(s64 a)
>>>>>  
>>>>>  static inline int drm_fixp2int_round(s64 a)
>>>>>  {
>>>>> -	return drm_fixp2int(a + (1 << (DRM_FIXED_POINT_HALF - 1)));
>>>> Also, this is the only usage of DRM_FIXED_POINT_HALF. Can you also
>>>> remove it as it won't be used anymore?
>>>>
>>>>> +	return drm_fixp2int(a + DRM_FIXED_ONE / 2);
>>>> Would this division be equivalent to just shifting 1ULL by 31 instead of
>>>> 32 as done in DRM_FIXED_ONE?
>>>
>>> Yes, but I think the division makes it easier to understand what is
>>> going on.
>>
>> Right. I was thinking about slightly better performance, but I don't
>> have any data. We can go with this since you consider more readable,
>> anyway.
> 
> Just checked that Harry proposed in another patch[1] this:
> `#define DRM_FIXED_HALF		0x80000000ll` for the 0.5 const
> 
> Doesn't it sounds better?
> 

I tend to agree with Arthur and Pekka. DRM_FIXED_ONE / 2 makes it
really clear what's happening here. And I'd imagine compilers would
optimize it out anyways.

Obviously not opposed to a define but if it's only used in one place
I don't think it matters much.

Harry

> [1] https://lore.kernel.org/dri-devel/20240226211100.100108-4-harry.wentland@amd.com/
>>
>> Can you send another version addressing the other comments? Then I can
>> cherry-pick and already apply the fix.
>>
>> Thanks,
>>
>> Melissa
>>
>>>
>>> Best Regards,
>>> ~Arthur Grillo
>>>
>>>>
>>>> Melissa
>>>>
>>>>>  }
>>>>>  
>>>>>  static inline int drm_fixp2int_ceil(s64 a)
>>>>>
>>>>> -- 
>>>>> 2.43.0
>>>>>