Re: [PATCH] selftests/kcmp: exit with non-zero code in a fail case

[Andrey Wagin <avagin@gmail.com>]

2015-03-13 13:37 GMT+03:00 Michael Ellerman <mpe@ellerman.id.au>:
> On Fri, 2015-03-13 at 12:27 +0300, Andrey Vagin wrote:
>> diff --git a/tools/testing/selftests/kselftest.h b/tools/testing/selftests/kselftest.h
>> index 572c888..a0ec8b8 100644
>> --- a/tools/testing/selftests/kselftest.h
>> +++ b/tools/testing/selftests/kselftest.h
>> @@ -58,5 +58,17 @@ static inline int ksft_exit_skip(void)
>>  {
>>       exit(4);
>>  }
>> +static inline int ksft_exit(void)
>> +{
>> +     if (ksft_cnt.ksft_fail)
>> +             return ksft_exit_fail();
>> +     if (ksft_cnt.ksft_xpass)
>> +             return ksft_exit_xpass();
>> +     if (ksft_cnt.ksft_xskip)
>> +             return ksft_exit_skip();
>> +     if (ksft_cnt.ksft_xfail)
>> +             return ksft_exit_xfail();
>> +     ksft_exit_pass();
>> +}
>
> This function claims to return 'int', but doesn't. So do all the others.

I agree with this. I haven't seen any warning, because exit() never
returns back.

>
> It could be as simple as:

I think it isn't a good idea. In my version we can be sure that
ksft_exit_fail() and ksft_exit() return the same code in a fail case.

>
> static inline void ksft_exit(void)
> {
>         if (ksft_cnt.ksft_fail)
>                 exit(1);
>
>         if (ksft_cnt.ksft_xpass)
>                 exit(3);
>
>         if (ksft_cnt.ksft_xskip)
>                 exit(4);
>
>         if (ksft_cnt.ksft_xfail)
>                 exit(2);
>
>         exit(0);
> }
>
> cheers
>
>
>