Re: [PATCH v2 0/3] Add tests for abs(), labs(), llabs() (BZ #30263)

[Florian Weimer <fweimer@redhat.com>]

* Siddhesh Poyarekar:

> On 2023-05-04 08:53, Wilco Dijkstra via Libc-alpha wrote:
>> Hi Joe,
>> 
>>> Add testcases for abs(), labs(), and llabs().  Test the bounds of the
>>> given type, zero, and part of the full range of values for labs() and
>>> llabs().  Test the full range for abs().
>>>
>>> Changes in V2:
>>> Reduce number of iterations by using a prime interval.  For llabs() test
>>> the 32bit crossover point.
>> v2 looks good to me.
>> Reviewed-by: Wilco Dijkstra  <Wilco.Dijkstra@arm.com>
>
> I've pushed these now.

This breaks on various 32-bit architectures.  As seen on microblaze:

tst-llabs.c: In function ‘do_test’:
tst-llabs.c:40:12: error: overflow in conversion from ‘long long int’ to ‘long int’ changes value from ‘-9223372036854775807’ to ‘1’ [-Werror=overflow]
   40 |   for (i = LLONG_MIN + 1; i < LLONG_MIN + INT_MAX; i += LARGE_PRIME)
      |            ^~~~~~~~~
tst-llabs.c:43:12: error: overflow in conversion from ‘long long int’ to ‘long int’ changes value from ‘9223372034707292160’ to ‘-2147483648’ [-Werror=overflow]
   43 |   for (i = LLONG_MAX - INT_MAX; i < LLONG_MAX - LARGE_PRIME; i += LARGE_PRIME)
      |            ^~~~~~~~~
cc1: all warnings being treated as errors

Thanks,
Florian