GeezerVison

Without my PR, test on FreeBSD 13.2 with clang version 14.0.5: FAILURE

• git clean -fdx; ./configure --with-pydebug && make clean && make && ./python -m test -v test_gdb -u all
• Total tests: run=46 failures=4 skipped=9

Tests with my PR.

Linux, clang 16

Test on Linux, Fedora 38 with clang version 16.0.6 (Fedora 16.0.6-2.fc38):

• clang -Og: SUCCESS with... 41 skipped tests on 46 :-)

  • git clean -fdx; ./configure --with-pydebug && make clean && make && ./python -m test -v test_gdb -u all
  • Total tests: run=46 skipped=41

• clang -O0: SUCCESS with only 1 test skipped!

  • git clean -fdx; ./configure --with-pydebug CFLAGS="-O0" && make clean && make && ./python -m test -v test_gdb -u all
  • Total tests: run=46 skipped=1
  • test_print_after_up() is skipped because of: "Unable to read information on python frame"

FreeBSD, clang 14

Test on FreeBSD 13.2 with clang version 14.0.5:

• clang -Og: SUCCESS with... 41 skipped tests on 46 :-)

  • git clean -fdx; ./configure --with-pydebug && make clean && make && ./python -m test -v test_gdb -u all
  • Total tests: run=46 skipped=41

• clang -O0: SUCCESS with only 1 test skipped!

• git clean -fdx; ./configure --with-pydebug CFLAGS="-O0" && make clean && make && ./python -m test -v test_gdb -u all

• Total tests: run=46 skipped=1

UPDATE: Moreover, test_gdb is no longer skipped on macOS if Python is built with Clang.

diff --git a/Lib/test/test_gdb.py b/Lib/test/test_gdb.py
index c05a2d387c..ca50574e46 100644
--- a/Lib/test/test_gdb.py
+++ b/Lib/test/test_gdb.py
@@ -55,10 +55,6 @@ def get_gdb_version():
 if not sysconfig.is_python_build():
     raise unittest.SkipTest("test_gdb only works on source builds at the moment.")
 
-if 'Clang' in platform.python_compiler() and sys.platform == 'darwin':
-    raise unittest.SkipTest("test_gdb doesn't work correctly when python is"
-                            " built with LLVM clang")
-
 if ((sysconfig.get_config_var('PGO_PROF_USE_FLAG') or 'xxx') in
     (sysconfig.get_config_var('PY_CORE_CFLAGS') or '')):

Test with my PR and clang -O3.

Test on Linux: SUCCESS with only 18 skipped tests on 46, nice!

• git clean -fdx; ./configure --with-pydebug CC=clang && make clean && make && ./python -m test -v test_gdb -u all
• Total tests: run=46 skipped=18

Test on FreeBSD: SUCCESS with only 18 skipped tests on 46, nice!

• git clean -fdx; ./configure && make clean && make && ./python -m test -v test_gdb -u all
• Total tests: run=46 skipped=18

Now I have doubts about my tests on Linux :-( I'm not sure that I passed CC=clang to configure!

I redone my tests on Linux:

• clang -O0: SUCCESS

  • git clean -fdx; ./configure CC=clang -with-pydebug CFLAGS="-O0" && make clean && make && ./python -m test -v test_gdb -u all
  • Total tests: run=46 skipped=1
  • grep -E '^(CC=|OPT=)' Makefile: CC=clang and OPT=-g -Og -Wall

• clang -Og: SUCCESS

  • git clean -fdx; ./configure CC=clang -with-pydebug && make clean && make && ./python -m test -v test_gdb -u all
  • Total tests: run=46 skipped=41
  • grep -E '^(CC=|OPT=)' Makefile: CC=clang and OPT=-g -Og -Wall

• clang -O3: SUCCESS

  • git clean -fdx; ./configure CC=clang && make clean && make && ./python -m test -v test_gdb -u all
  • Total tests: run=46 skipped=18
  • grep -E '^(CC=|OPT=)' Makefile: CC=clang and OPT=-DNDEBUG -g -O3 -Wall

Oh ok, we are safe :-)

GitHub Action jobs:

• Ubuntu: test_gdb passed, Python is built with gcc -g -Og
• Address sanitizer: "test_gdb passed", Python is built with gcc -DNDEBUG -g -O3
• macOS: test_gdb skipped -- Couldn't find gdb on the path

Clang just treats -Og as an alias for -O1, so it's not surprising that the debuginfo quality is not as good as GCC's.
llvm/llvm-project@d340ccc

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

Thanks @vstinner for the PR 🌮🎉.. I'm working now to backport this PR to: 3.11.
🐍🍒⛏🤖

GH-109010 is a backport of this pull request to the 3.12 branch.

GH-109011 is a backport of this pull request to the 3.11 branch.

Sadly, this change wasn't enough to fix PPC64LE Fedora Stable Clang 3.x: I created issue #109053.