gh-109972: Split test_gdb into a package of 3 tests

[vstinner]

Split test_gdb.py file into a test_gdb package made of multiple
tests, so tests can now be run in parallel.

• Create Lib/test/test_gdb/ directory.
• Split test_gdb.py into multiple files in Lib/test/test_gdb/
  directory.
• Move Lib/test/gdb_sample.py to Lib/test/test_gdb/ directory.
  Update get_sample_script(): use file to locate gdb_sample.py.
• Move gdb_has_frame_select() and HAS_PYUP_PYDOWN to test_misc.py.
• Explicitly skip test_gdb on Windows. Previously, test_gdb was
  skipped even if gdb was available because of
  gdb_has_frame_select().

• Issue: test_gdb failed with 15 min timeout on PPC64LE RHEL8 3.x (LTO + pydebug) #109972

[vstinner]

Test on PPC64LE RHEL8 3.x buildbot.

When Python is built with ./configure --with-pydebug --with-lto && make -j10, test_gdb takes 19 min 45 sec.

With this change, it takes 12 min 2 sec when tests are run in parallel.

$ ./python -m test test_gdb -u all -j0 --slowest 
0:00:00 load avg: 1.39 Run 3 tests in parallel using 3 worker processes
0:00:30 load avg: 3.01 running (3): test_gdb.test_backtrace (30.0 sec), test_gdb.test_misc (30.0 sec), test_gdb.test_pretty_print (30.0 sec)
0:01:00 load avg: 3.34 running (3): test_gdb.test_backtrace (1 min), test_gdb.test_misc (1 min), test_gdb.test_pretty_print (1 min)
0:01:30 load avg: 3.67 running (3): test_gdb.test_backtrace (1 min 30 sec), test_gdb.test_misc (1 min 30 sec), test_gdb.test_pretty_print (1 min 30 sec)
0:01:41 load avg: 3.72 [1/3] test_gdb.test_misc passed (1 min 41 sec) -- running (2): test_gdb.test_backtrace (1 min 41 sec), test_gdb.test_pretty_print (1 min 41 sec)
0:02:11 load avg: 3.49 running (2): test_gdb.test_backtrace (2 min 11 sec), test_gdb.test_pretty_print (2 min 11 sec)
0:02:41 load avg: 3.37 running (2): test_gdb.test_backtrace (2 min 41 sec), test_gdb.test_pretty_print (2 min 41 sec)
0:03:11 load avg: 3.16 running (2): test_gdb.test_backtrace (3 min 11 sec), test_gdb.test_pretty_print (3 min 11 sec)
0:03:41 load avg: 3.17 running (2): test_gdb.test_backtrace (3 min 41 sec), test_gdb.test_pretty_print (3 min 41 sec)
0:04:11 load avg: 3.47 running (2): test_gdb.test_backtrace (4 min 11 sec), test_gdb.test_pretty_print (4 min 11 sec)
0:04:41 load avg: 3.36 running (2): test_gdb.test_backtrace (4 min 41 sec), test_gdb.test_pretty_print (4 min 41 sec)
0:05:11 load avg: 3.22 running (2): test_gdb.test_backtrace (5 min 11 sec), test_gdb.test_pretty_print (5 min 11 sec)
0:05:41 load avg: 3.04 running (2): test_gdb.test_backtrace (5 min 41 sec), test_gdb.test_pretty_print (5 min 41 sec)
0:06:11 load avg: 3.04 running (2): test_gdb.test_backtrace (6 min 11 sec), test_gdb.test_pretty_print (6 min 11 sec)
0:06:41 load avg: 3.02 running (2): test_gdb.test_backtrace (6 min 41 sec), test_gdb.test_pretty_print (6 min 41 sec)
0:07:11 load avg: 3.01 running (2): test_gdb.test_backtrace (7 min 11 sec), test_gdb.test_pretty_print (7 min 11 sec)
0:07:19 load avg: 3.01 [2/3] test_gdb.test_pretty_print passed (7 min 19 sec) -- running (1): test_gdb.test_backtrace (7 min 19 sec)
0:07:49 load avg: 2.55 running (1): test_gdb.test_backtrace (7 min 49 sec)
0:08:19 load avg: 2.18 running (1): test_gdb.test_backtrace (8 min 19 sec)
0:08:49 load avg: 1.71 running (1): test_gdb.test_backtrace (8 min 49 sec)
0:09:19 load avg: 1.67 running (1): test_gdb.test_backtrace (9 min 19 sec)
0:09:49 load avg: 1.46 running (1): test_gdb.test_backtrace (9 min 49 sec)
0:10:19 load avg: 1.28 running (1): test_gdb.test_backtrace (10 min 19 sec)
0:10:49 load avg: 1.17 running (1): test_gdb.test_backtrace (10 min 49 sec)
0:11:19 load avg: 1.10 running (1): test_gdb.test_backtrace (11 min 19 sec)
0:11:49 load avg: 1.06 running (1): test_gdb.test_backtrace (11 min 49 sec)
0:12:02 load avg: 1.05 [3/3] test_gdb.test_backtrace passed (12 min 2 sec)

== Tests result: SUCCESS ==

10 slowest tests:
- test_gdb.test_backtrace: 12 min 2 sec
- test_gdb.test_pretty_print: 7 min 19 sec
- test_gdb.test_misc: 1 min 41 sec

All 3 tests OK.

Total duration: 12 min 2 sec
Total tests: run=51 skipped=1
Total test files: run=3/3
Result: SUCCESS

[vstinner]

I splitted the test even more, I added test_cfunction and test_cfunction_full.

Timings on my Fedora 38 x86-64:

- test_gdb.test_pretty_print: 13.3 sec
- test_gdb.test_cfunction: 10.5 sec
- test_gdb.test_cfunction_full: 6.5 sec
- test_gdb.test_misc: 3.9 sec
- test_gdb.test_backtrace: 2.8 sec

Total duration: 13.4 sec

Timings on RHEL8 ppc64le, buildbot with slowest test_gdb:

10 slowest tests:
- test_gdb.test_cfunction: 9 min 33 sec
- test_gdb.test_pretty_print: 8 min 36 sec
- test_gdb.test_cfunction_full: 2 min 56 sec
- test_gdb.test_misc: 1 min 49 sec
- test_gdb.test_backtrace: 1 min 45 sec

Total duration: 9 min 33 sec

[vstinner]

Hum, I went too far in terms of refactoring. I made this PR as simple as possible, to only split test_gdb into multiple files.

I will write a follow-up PR for the heavy refactoring.

[vstinner]

I added 5 minutes to buildbot timeout: python/buildmaster-config#408 Before, buildbots always removed 5 minutes to their default timeout of 20 minutes. I fixed this issue to have use more consistent timeouts on all Python CIs.

[miss-islington]

Thanks @vstinner for the PR 🌮🎉.. I'm working now to backport this PR to: 3.12.
🐍🍒⛏🤖

[miss-islington]

Sorry, @vstinner, I could not cleanly backport this to 3.12 due to a conflict.
Please backport using cherry_picker on command line.
cherry_picker 8f324b7ecd2df3036fab098c4c8ac185ac07b277 3.12

[miss-islington]

Thanks @vstinner for the PR 🌮🎉.. I'm working now to backport this PR to: 3.12.
🐍🍒⛏🤖

[miss-islington]

Sorry, @vstinner, I could not cleanly backport this to 3.12 due to a conflict.
Please backport using cherry_picker on command line.

cherry_picker 8f324b7ecd2df3036fab098c4c8ac185ac07b277 3.12

[bedevere-app]

GH-110339 is a backport of this pull request to the 3.12 branch.