2024-05-06 09:08:15

by Ilpo Järvinen

[permalink] [raw]
Subject: Re: [PATCH] selftests/resctrl: fix clang build warnings related to abs(), labs() calls

On Fri, 3 May 2024, John Hubbard wrote:

> On 5/3/24 11:37 AM, Reinette Chatre wrote:
> > On 5/3/2024 9:52 AM, John Hubbard wrote:
> > > On 5/3/24 1:00 AM, Ilpo J?rvinen wrote:
> > > > On Thu, 2 May 2024, John Hubbard wrote:
> > > ...
> > > > > diff --git a/tools/testing/selftests/resctrl/mbm_test.c
> > > > > b/tools/testing/selftests/resctrl/mbm_test.c
> > > > > index d67ffa3ec63a..c873793d016d 100644
> > > > > --- a/tools/testing/selftests/resctrl/mbm_test.c
> > > > > +++ b/tools/testing/selftests/resctrl/mbm_test.c
> > > > > @@ -33,7 +33,7 @@ show_bw_info(unsigned long *bw_imc, unsigned long
> > > > > *bw_resc, size_t span)
> > > > > ? ????? avg_bw_imc = sum_bw_imc / 4;
> > > > > ????? avg_bw_resc = sum_bw_resc / 4;
> > > > > -??? avg_diff = (float)labs(avg_bw_resc - avg_bw_imc) / avg_bw_imc;
> > > > > +??? avg_diff = (float)(avg_bw_resc - avg_bw_imc) / avg_bw_imc;
> > > > > ????? avg_diff_per = (int)(avg_diff * 100);
> > > > > ? ????? ret = avg_diff_per > MAX_DIFF_PERCENT;
> > > >
> > > > But how are these two cases same after your change when you ended up
> > > > removing taking the absolute value entirely?
> > >
> > > All of the arguments are unsigned integers, so all arithmetic results
> > > are interpreted as unsigned, so taking the absolute value of that is
> > > always a no-op.

(I see there's a better patch posted already but since there are a few
incorrect claims in this discussion, I'll do for the record type of
reply.)

This discussion now went to a tangent about the warning. My main point is
that logic is not correct after removing labs().

I also disagree with the claim that using labs() on unsigned value is
no-op because labs() takes long so unsigned is just forced into signed
when calling which is why the warning triggers but it's very misleading
warning (see below).

> > It does not seem as though clang can see when values have been casted.
> > I tried to do so explicitly with a:
> > avg_diff = labs((long)avg_bw_resc - avg_bw_imc) / (float)avg_bw_imc;
>
> The subtraction result will get promoted to an unsigned long, before being
> passed into labs(3).
>
> > But that still triggers:
> > warning: taking the absolute value of unsigned type 'unsigned long' has no
> > effect [-Wabsolute-value]
>
> As expected, yes.

That error message isn't factually correct:

unsigned long a = LONG_MAX;
long b;

a += 2;
b = (long)a;
printf("%llu %lli %lli\n", a, b, labs(a));

Prints (at least when built with gcc):

9223372036854775809 -9223372036854775807 9223372036854775807

labs(LONG_MAX + 1) won't work though since it's not positively presentable
with long and the value is left untouched.

--
i.


2024-05-06 17:21:23

by John Hubbard

[permalink] [raw]
Subject: Re: [PATCH] selftests/resctrl: fix clang build warnings related to abs(), labs() calls

On 5/6/24 2:07 AM, Ilpo Järvinen wrote:
> On Fri, 3 May 2024, John Hubbard wrote:
>
>> On 5/3/24 11:37 AM, Reinette Chatre wrote:
>>> On 5/3/2024 9:52 AM, John Hubbard wrote:
>>>> On 5/3/24 1:00 AM, Ilpo Järvinen wrote:
>>>>> On Thu, 2 May 2024, John Hubbard wrote:
>>>> ...
>>>>>> diff --git a/tools/testing/selftests/resctrl/mbm_test.c
>>>>>> b/tools/testing/selftests/resctrl/mbm_test.c
>>>>>> index d67ffa3ec63a..c873793d016d 100644
>>>>>> --- a/tools/testing/selftests/resctrl/mbm_test.c
>>>>>> +++ b/tools/testing/selftests/resctrl/mbm_test.c
>>>>>> @@ -33,7 +33,7 @@ show_bw_info(unsigned long *bw_imc, unsigned long
>>>>>> *bw_resc, size_t span)
>>>>>>         avg_bw_imc = sum_bw_imc / 4;
>>>>>>       avg_bw_resc = sum_bw_resc / 4;
>>>>>> -    avg_diff = (float)labs(avg_bw_resc - avg_bw_imc) / avg_bw_imc;
>>>>>> +    avg_diff = (float)(avg_bw_resc - avg_bw_imc) / avg_bw_imc;
>>>>>>       avg_diff_per = (int)(avg_diff * 100);
>>>>>>         ret = avg_diff_per > MAX_DIFF_PERCENT;
>>>>>
>>>>> But how are these two cases same after your change when you ended up
>>>>> removing taking the absolute value entirely?
>>>>
>>>> All of the arguments are unsigned integers, so all arithmetic results
>>>> are interpreted as unsigned, so taking the absolute value of that is
>>>> always a no-op.
>
> (I see there's a better patch posted already but since there are a few
> incorrect claims in this discussion, I'll do for the record type of
> reply.)
>
> This discussion now went to a tangent about the warning. My main point is
> that logic is not correct after removing labs().
>
> I also disagree with the claim that using labs() on unsigned value is
> no-op because labs() takes long so unsigned is just forced into signed
> when calling which is why the warning triggers but it's very misleading
> warning (see below).
>

Yes you are correct.

>>> It does not seem as though clang can see when values have been casted.
>>> I tried to do so explicitly with a:
>>> avg_diff = labs((long)avg_bw_resc - avg_bw_imc) / (float)avg_bw_imc;
>>
>> The subtraction result will get promoted to an unsigned long, before being
>> passed into labs(3).
>>
>>> But that still triggers:
>>> warning: taking the absolute value of unsigned type 'unsigned long' has no
>>> effect [-Wabsolute-value]
>>
>> As expected, yes.
>
> That error message isn't factually correct:
>
> unsigned long a = LONG_MAX;
> long b;
>
> a += 2;
> b = (long)a;
> printf("%llu %lli %lli\n", a, b, labs(a));
>
> Prints (at least when built with gcc):
>
> 9223372036854775809 -9223372036854775807 9223372036854775807
>
> labs(LONG_MAX + 1) won't work though since it's not positively presentable
> with long and the value is left untouched.
>

Thanks for setting the detailed record straight! :)


thanks,
--
John Hubbard
NVIDIA