ENH: Properly support FreeBSD/powerpc64

[pkubaj]

1. FreeBSD sets machine name (uname -m) differently than Linux and instead uses uname -p for actual architecture name, but this is not present in uname() function. Check just for powerpc or ppc and compiler macros instead.
2. FreeBSD uses elf_aux_info instead of getauxval.

[charris]

Looks like a couple of tests need fixing and/or the detection can be made backwards compatible.

[pkubaj]

Can you point me to what's actually wrong? I'm quite puzzled right now.

[charris]

You can find the log by clicking on the details. For the smoke test it is at https://github.com/numpy/numpy/actions/runs/3678279521/jobs/6221343413, and the error is

_________________ TestCCompilerOpt_ppc64_clang.test_interface __________________

self = <numpy.distutils.tests.test_ccompiler_opt.TestCCompilerOpt_ppc64_clang object at 0x7f198b4b8b80>

    def test_interface(self):
        wrong_arch = "ppc64" if self.arch != "ppc64" else "x86"
        wrong_cc   = "clang" if self.cc   != "clang" else "icc"
        opt = self.opt()
>       assert_(getattr(opt, "cc_on_" + self.arch))
E       AssertionError

[pkubaj]

I added a commit that works around the test failures.

[pkubaj]

I had some issues again with architecture detection so I added '.*' around 'powerpc'. I'd rather prefer just '.powerpc.', since it would be much simpler, but it somehow doesn't work with tests.
Also, commits squashed.

Summary after compilation:

########### EXT COMPILER OPTIMIZATION ###########
INFO: Platform      :
  Architecture: ppc64le
  Compiler    : unix-like

CPU baseline  :
  Requested   : 'min'
  Enabled     : VSX VSX2
  Flags       : -mcpu=power8
  Extra checks: VSX_ASM

CPU dispatch  :
  Requested   : 'max -xop -fma4'
  Enabled     : VSX3 VSX4
  Generated   :
              :
  VSX3        : VSX VSX2
  Flags       : -mcpu=power9 -mtune=power9
  Extra checks: none
  Detect      : VSX3
              : build/src.freebsd-13.1-RELEASE-p5-powerpc-3.9/numpy/core/src/_simd/_simd.dispatch.c
              : build/src.freebsd-13.1-RELEASE-p5-powerpc-3.9/numpy/core/src/umath/loops_trigonometric.dispatch.c
              : numpy/core/src/umath/_umath_tests.dispatch.c
              :
  VSX4        : VSX VSX2 VSX3
  Flags       : -mcpu=power10 -mtune=power10
  Extra checks: VSX4_MMA
  Detect      : VSX4
              : build/src.freebsd-13.1-RELEASE-p5-powerpc-3.9/numpy/core/src/_simd/_simd.dispatch.c
              : build/src.freebsd-13.1-RELEASE-p5-powerpc-3.9/numpy/core/src/umath/loops_arithmetic.dispatch.c
              : build/src.freebsd-13.1-RELEASE-p5-powerpc-3.9/numpy/core/src/umath/loops_hyperbolic.dispatch.c
              : build/src.freebsd-13.1-RELEASE-p5-powerpc-3.9/numpy/core/src/umath/loops_modulo.dispatch.c
              : build/src.freebsd-13.1-RELEASE-p5-powerpc-3.9/numpy/core/src/umath/loops_trigonometric.dispatch.c
INFO: CCompilerOpt.cache_flush[857] : write cache to path -> /wrkdirs/usr/ports/math/py-numpy/work-py39/numpy-1.23.5/build/temp.freebsd-13.1-RELEASE-p5-powerpc-cpython-39/ccompiler_opt_cache_ext.py
INFO:
########### CLIB COMPILER OPTIMIZATION ###########
INFO: Platform      :
  Architecture: ppc64le
  Compiler    : unix-like

CPU baseline  :
  Requested   : 'min'
  Enabled     : VSX VSX2
  Flags       : -mcpu=power8
  Extra checks: VSX_ASM

CPU dispatch  :
  Requested   : 'max -xop -fma4'
  Enabled     : VSX3 VSX4
  Generated   : none

[seiko2plus]

I had some issues again with architecture detection so I added '.*' around 'powerpc'. I'd rather prefer just '.powerpc.', since it would be much simpler, but it somehow doesn't work with tests.

The fake compiler within distutils tests counts on (ppc64|ppc64le) to differentiate between big|little endian, also it is necessary for Linux. The term ppc64(el|le) is been used among the distributions since Power7.

[pkubaj]

I know Linux uses ppc64 / ppc64le names. I just didn't know the test suite relied on them. If it didn't, checking just for ppc* and then correct compiler macros would be also ok.

[seiko2plus]

LGTM, Thank you. The following suggestions aren't necessary but still great to have them.

[seiko2plus]

Thank you @pkubaj.

[InessaPawson]

Hi-five on merging your first pull request to NumPy, @pkubaj! We hope you stick around! Your choices aren’t limited to writing code – you can review pull requests, help us stay on top of new and old issues, develop educational material, work on our website, add or improve graphic design, create marketing materials, translate website content, write grant proposals, and help with other fundraising initiatives. For more info, check out: https://numpy.org/contribute
Also, consider joining our mailing list. This is a great way to connect with other cool people in our community and be part of important conversations that affect the development of NumPy: https://mail.python.org/mailman/listinfo/numpy-discussion