> For this particular case, I'd be fine with adding a special case for `x == y` in the code (suitably commented to explain why the special case is there, and under what conditions it can be removed).

I proposed PR 18094.

Because of your parenthesis, I chose to use "#if defined(_AIX)" with an an explicit comment on AIX version. For example, we may drop this workaround once we would like to drop support for AIX 7.1.


> Actually, the cmath.exp failure looks a little odd to my eyes: it would be a bit surprising to have cmath.exp fail if all of test_math passed. I suspect a dodgy compiler optimisation. But that's another issue.

Michael: would you mind to open a separated issue for "FAIL: test_specific_values (test.test_cmath.CMathTests)"? Is it on a buildbot? Please try to get the machine code of cmath_exp_impl() function. Example:

   objdump -d build/lib.linux-x86_64-3.9-pydebug/cmath.cpython-39d-x86_64-linux-gnu.so

or try gdb:
---
$ gdb ./python
GNU gdb (GDB) Fedora 8.3.50.20190824-26.fc31
(gdb) run
>>> import cmath

<press CTRL+C>

Program received signal SIGINT, Interrupt.
(gdb) set pagination off
(gdb) disassemble cmath_exp_impl 
---