32-bit:

======================================================================
ERROR: test_mktime (test.test_time.TimeTestCase)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/data/prj/python/src/python3-3.7.0/Lib/test/test_time.py", line 446, in test_mktime
    self.assertEqual(time.mktime(tt), t)
OverflowError: mktime argument out of range

======================================================================
FAIL: test_pthread_getcpuclockid (test.test_time.TimeTestCase)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/data/prj/python/src/python3-3.7.0/Lib/test/test_time.py", line 129, in test_pthread_getcpuclockid
    self.assertNotEqual(clk_id, time.CLOCK_THREAD_CPUTIME_ID)
AssertionError: 12 == 12

----------------------------------------------------------------------

64-bit:
======================================================================
ERROR: test_mktime (test.test_time.TimeTestCase)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/opt/lib/python3.7/test/test_time.py", line 446, in test_mktime
    self.assertEqual(time.mktime(tt), t)
OverflowError: mktime argument out of range

======================================================================
ERROR: test_pthread_getcpuclockid (test.test_time.TimeTestCase)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/opt/lib/python3.7/test/test_time.py", line 130, in test_pthread_getcpuclockid
    t1 = time.clock_gettime(clk_id)
OverflowError: signed integer is greater than maximum

The test_mktime failed because
time.mktime(time.localtime(-2)) -- actually any negative value in localtime - fails

The patch in the file Modules/timemodule.c fixes this.

The test_pthread_getcpuclockid is a bit more complex as the result is different
depending on 32 or 64-bit mode.

In 32-bit mode AIX always responds with the constant CLOCK_THREAD_CPUTIME_ID.
The patch in Lib/test/test_time.py makes the test aware of this.

In AIX clockid_t is actually long long. However, changing the type to parse to "L"
broke many other things. It seems to work as a long ("l") when the ABI is 64-bit.

See additional change in Modules/timemodule.c 
static PyObject *
time_clock_gettime

Finally,
added some (additional) range checking for 
_PyTime_localtime for AIX. Not having this complicated testing for above.

Would be very nice to get this to clear on the build bots!

New changeset e2926b72488596f59e43c27f3b7cedf0c5b9e88e by Nick Coghlan (Michael Felt) in branch 'master':
bpo-34373: fix test_mktime and test_pthread_getcpuclickid tests on AIX (GH-8726)
https://github.com/python/cpython/commit/e2926b72488596f59e43c27f3b7cedf0c5b9e88e

Could this alos be backported to Version 3.7 and 3.6 - thx!

On 10/03/2019 19:43, Michael Felt wrote:
> Michael Felt <aixtools@felt.demon.nl> added the comment:
>
> Could this alos be backported to Version 3.7 and 3.6 - thx!
As 3.6 is in security mode - I guess only 3.7 then.
> ----------
> versions: +Python 3.6, Python 3.7
>
> _______________________________________
> Python tracker <report@bugs.python.org>
> <https://bugs.python.org/issue34373>
> _______________________________________
>

New changeset 8709490f48fc27b3dd1a16acb33bea2299c6a575 by Victor Stinner in branch 'master':
bpo-34373: Fix time.mktime() on AIX (GH-12726)
https://github.com/python/cpython/commit/8709490f48fc27b3dd1a16acb33bea2299c6a575