Random crash of `test_signal` on macos sonoma

[sobolevn]

Crash report

I've experience a randome crash while fixing #109981
Why random? Because I was not able to reproduce it ever again.

My command:

» ./python.exe -m test -f tests.txt
0:00:00 load avg: 4.78 Run 31 tests sequentially
0:00:00 load avg: 4.78 [ 1/31] test_shelve
0:00:00 load avg: 4.78 [ 2/31] test_shlex
0:00:00 load avg: 4.78 [ 3/31] test_signal
Fatal Python error: Bus error

Current thread 0x000000016c947000 (most recent call first):
  File "/Users/sobolev/Desktop/cpython/Lib/test/test_signal.py", line 1338 in set_interrupts
  File "/Users/sobolev/Desktop/cpython/Lib/threading.py", line 1003 in run
  File "/Users/sobolev/Desktop/cpython/Lib/threading.py", line 1066 in _bootstrap_inner
  File "/Users/sobolev/Desktop/cpython/Lib/threading.py", line 1023 in _bootstrap

Thread 0x00000001de555300 (most recent call first):
  File "/Users/sobolev/Desktop/cpython/Lib/signal.py", line 29 in _int_to_enum
  File "/Users/sobolev/Desktop/cpython/Lib/signal.py", line 57 in signal
  File "/Users/sobolev/Desktop/cpython/Lib/test/test_signal.py", line 1346 in cycle_handlers
  File "/Users/sobolev/Desktop/cpython/Lib/test/test_signal.py", line 1356 in test_stress_modifying_handlers
  File "/Users/sobolev/Desktop/cpython/Lib/unittest/case.py", line 589 in _callTestMethod
  File "/Users/sobolev/Desktop/cpython/Lib/unittest/case.py", line 636 in run
  File "/Users/sobolev/Desktop/cpython/Lib/unittest/case.py", line 692 in __call__
  File "/Users/sobolev/Desktop/cpython/Lib/unittest/suite.py", line 122 in run
  File "/Users/sobolev/Desktop/cpython/Lib/unittest/suite.py", line 84 in __call__
  File "/Users/sobolev/Desktop/cpython/Lib/unittest/suite.py", line 122 in run
  File "/Users/sobolev/Desktop/cpython/Lib/unittest/suite.py", line 84 in __call__
  File "/Users/sobolev/Desktop/cpython/Lib/unittest/suite.py", line 122 in run
  File "/Users/sobolev/Desktop/cpython/Lib/unittest/suite.py", line 84 in __call__
  File "/Users/sobolev/Desktop/cpython/Lib/test/support/testresult.py", line 146 in run
  File "/Users/sobolev/Desktop/cpython/Lib/test/support/__init__.py", line 1152 in _run_suite
  File "/Users/sobolev/Desktop/cpython/Lib/test/support/__init__.py", line 1279 in run_unittest
  File "/Users/sobolev/Desktop/cpython/Lib/test/libregrtest/single.py", line 36 in run_unittest
  File "/Users/sobolev/Desktop/cpython/Lib/test/libregrtest/single.py", line 92 in test_func
  File "/Users/sobolev/Desktop/cpython/Lib/test/libregrtest/single.py", line 48 in regrtest_runner
  File "/Users/sobolev/Desktop/cpython/Lib/test/libregrtest/single.py", line 95 in _load_run_test
  File "/Users/sobolev/Desktop/cpython/Lib/test/libregrtest/single.py", line 138 in _runtest_env_changed_exc
  File "/Users/sobolev/Desktop/cpython/Lib/test/libregrtest/single.py", line 238 in _runtest
  File "/Users/sobolev/Desktop/cpython/Lib/test/libregrtest/single.py", line 266 in run_single_test
  File "/Users/sobolev/Desktop/cpython/Lib/test/libregrtest/main.py", line 290 in run_test
  File "/Users/sobolev/Desktop/cpython/Lib/test/libregrtest/main.py", line 325 in run_tests_sequentially
  File "/Users/sobolev/Desktop/cpython/Lib/test/libregrtest/main.py", line 459 in _run_tests
  File "/Users/sobolev/Desktop/cpython/Lib/test/libregrtest/main.py", line 490 in run_tests
  File "/Users/sobolev/Desktop/cpython/Lib/test/libregrtest/main.py", line 576 in main
  File "/Users/sobolev/Desktop/cpython/Lib/test/libregrtest/main.py", line 584 in main
  File "/Users/sobolev/Desktop/cpython/Lib/test/__main__.py", line 2 in <module>
  File "/Users/sobolev/Desktop/cpython/Lib/runpy.py", line 88 in _run_code
  File "/Users/sobolev/Desktop/cpython/Lib/runpy.py", line 198 in _run_module_as_main

Extension modules: _testcapi (total: 1)
[1]    3657 bus error  ./python.exe -m test -f tests.txt

Contents of tests.txt:

test_shelve
test_shlex
test_signal
test_site
test_slice
test_smtplib
test_smtpnet
test_socket
test_socketserver
test_sort
test_source_encoding
test_sqlite3
test_ssl
test_stable_abi_ctypes
test_startfile
test_stat
test_statistics
test_str
test_strftime
test_string
test_string_literals
test_stringprep
test_strptime
test_strtod
test_struct
test_structseq
test_subclassinit
test_subprocess
test_sundry
test_super
test_support

But, note that only 3 tests were run before: test_shelve test_shlex test_signal

I've also tried to run ./python.exe -m test -j4 --forever test_shelve test_shlex test_signal, but no luck for now.

Env:

• 98c0c1d (main)
• macos sonoma, apple m2

Linked PRs

• gh-110017: Disable test_signal.test_stress_modifying_handlers on macOS #112834
• [3.12] gh-110017: Disable test_signal.test_stress_modifying_handlers on macOS (GH-112834) #112851
• [3.11] gh-110017: Disable test_signal.test_stress_modifying_handlers on macOS (GH-112834) #112852
• [3.12] gh-110820: Make sure processor specific defines are correct for Universal 2 build on macOS (gh-112828) #112864

[sobolevn]

I got it again from --forever:

0:05:31 load avg: 1.86 [172/1] test_signal process crashed (Exit code -10)
Fatal Python error: Bus error

Current thread 0x000000016c56f000 (most recent call first):
  File "/Users/sobolev/Desktop/cpython/Lib/test/test_signal.py", line 1338 in set_interrupts
  File "/Users/sobolev/Desktop/cpython/Lib/threading.py", line 1003 in run
  File "/Users/sobolev/Desktop/cpython/Lib/threading.py", line 1066 in _bootstrap_inner
  File "/Users/sobolev/Desktop/cpython/Lib/threading.py", line 1023 in _bootstrap

Thread 0x00000001de555300 (most recent call first):
  File "/Users/sobolev/Desktop/cpython/Lib/enum.py", line 1128 in __new__
  File "/Users/sobolev/Desktop/cpython/Lib/enum.py", line 729 in __call__
  File "/Users/sobolev/Desktop/cpython/Lib/signal.py", line 29 in _int_to_enum
  File "/Users/sobolev/Desktop/cpython/Lib/signal.py", line 57 in signal
  File "/Users/sobolev/Desktop/cpython/Lib/test/test_signal.py", line 1346 in cycle_handlers
  File "/Users/sobolev/Desktop/cpython/Lib/test/test_signal.py", line 1356 in test_stress_modifying_handlers
  File "/Users/sobolev/Desktop/cpython/Lib/unittest/case.py", line 589 in _callTestMethod
  File "/Users/sobolev/Desktop/cpython/Lib/unittest/case.py", line 636 in run
  File "/Users/sobolev/Desktop/cpython/Lib/unittest/case.py", line 692 in __call__
  File "/Users/sobolev/Desktop/cpython/Lib/unittest/suite.py", line 122 in run
  File "/Users/sobolev/Desktop/cpython/Lib/unittest/suite.py", line 84 in __call__
  File "/Users/sobolev/Desktop/cpython/Lib/unittest/suite.py", line 122 in run
  File "/Users/sobolev/Desktop/cpython/Lib/unittest/suite.py", line 84 in __call__
  File "/Users/sobolev/Desktop/cpython/Lib/unittest/suite.py", line 122 in run
  File "/Users/sobolev/Desktop/cpython/Lib/unittest/suite.py", line 84 in __call__
  File "/Users/sobolev/Desktop/cpython/Lib/test/support/testresult.py", line 146 in run
  File "/Users/sobolev/Desktop/cpython/Lib/test/support/__init__.py", line 1152 in _run_suite
  File "/Users/sobolev/Desktop/cpython/Lib/test/support/__init__.py", line 1279 in run_unittest
  File "/Users/sobolev/Desktop/cpython/Lib/test/libregrtest/single.py", line 36 in run_unittest
  File "/Users/sobolev/Desktop/cpython/Lib/test/libregrtest/single.py", line 92 in test_func
  File "/Users/sobolev/Desktop/cpython/Lib/test/libregrtest/single.py", line 48 in regrtest_runner
  File "/Users/sobolev/Desktop/cpython/Lib/test/libregrtest/single.py", line 95 in _load_run_test
  File "/Users/sobolev/Desktop/cpython/Lib/test/libregrtest/single.py", line 138 in _runtest_env_changed_exc
  File "/Users/sobolev/Desktop/cpython/Lib/test/libregrtest/single.py", line 238 in _runtest
  File "/Users/sobolev/Desktop/cpython/Lib/test/libregrtest/single.py", line 266 in run_single_test
  File "/Users/sobolev/Desktop/cpython/Lib/test/libregrtest/worker.py", line 86 in worker_process
  File "/Users/sobolev/Desktop/cpython/Lib/test/libregrtest/worker.py", line 109 in main
  File "/Users/sobolev/Desktop/cpython/Lib/test/libregrtest/worker.py", line 113 in <module>
  File "/Users/sobolev/Desktop/cpython/Lib/runpy.py", line 88 in _run_code
  File "/Users/sobolev/Desktop/cpython/Lib/runpy.py", line 198 in _run_module_as_main

Extension modules: _testcapi (total: 1)
Kill <WorkerThread #1 running test=test_signal pid=6276 time=6.6 sec> process group
Kill <WorkerThread #2 running test=test_signal pid=6170 time=23.1 sec> process group
Kill <WorkerThread #3 running test=test_signal pid=6217 time=12.4 sec> process group
0:05:32 load avg: 1.86 Waiting for <WorkerThread #1 running test=test_signal pid=6276 time=7.6 sec> thread for 1.0 sec
0:05:33 load avg: 1.86 Waiting for <WorkerThread #1 running test=test_signal pid=6276 time=8.6 sec> thread for 2.0 sec
0:05:34 load avg: 1.86 Waiting for <WorkerThread #1 running test=test_signal pid=6276 time=9.6 sec> thread for 3.0 sec
0:05:35 load avg: 1.79 Waiting for <WorkerThread #1 running test=test_signal pid=6276 time=10.7 sec> thread for 4.0 sec
0:05:36 load avg: 1.79 Waiting for <WorkerThread #1 running test=test_signal pid=6276 time=11.7 sec> thread for 5.0 sec
0:05:37 load avg: 1.79 Waiting for <WorkerThread #1 running test=test_signal pid=6276 time=12.7 sec> thread for 6.0 sec
0:05:38 load avg: 1.79 Waiting for <WorkerThread #1 running test=test_signal pid=6276 time=13.7 sec> thread for 7.0 sec
0:05:39 load avg: 1.79 Waiting for <WorkerThread #1 running test=test_signal pid=6276 time=14.7 sec> thread for 8.0 sec
0:05:40 load avg: 1.65 Waiting for <WorkerThread #1 running test=test_signal pid=6276 time=15.7 sec> thread for 9.0 sec
0:05:41 load avg: 1.65 Waiting for <WorkerThread #1 running test=test_signal pid=6276 time=16.7 sec> thread for 10.0 sec
0:05:42 load avg: 1.65 Waiting for <WorkerThread #1 running test=test_signal pid=6276 time=17.7 sec> thread for 11.0 sec
0:05:43 load avg: 1.65 Waiting for <WorkerThread #1 running test=test_signal pid=6276 time=18.7 sec> thread for 12.0 sec
0:05:44 load avg: 1.65 Waiting for <WorkerThread #1 running test=test_signal pid=6276 time=19.7 sec> thread for 13.0 sec
0:05:45 load avg: 1.60 Waiting for <WorkerThread #1 running test=test_signal pid=6276 time=20.7 sec> thread for 14.1 sec
0:05:46 load avg: 1.60 Waiting for <WorkerThread #1 running test=test_signal pid=6276 time=21.7 sec> thread for 15.1 sec
0:05:47 load avg: 1.60 Waiting for <WorkerThread #1 running test=test_signal pid=6276 time=22.7 sec> thread for 16.1 sec
0:05:48 load avg: 1.60 Waiting for <WorkerThread #1 running test=test_signal pid=6276 time=23.7 sec> thread for 17.1 sec
0:05:49 load avg: 1.60 Waiting for <WorkerThread #1 running test=test_signal pid=6276 time=24.7 sec> thread for 18.1 sec
0:05:50 load avg: 1.47 Waiting for <WorkerThread #1 running test=test_signal pid=6276 time=25.7 sec> thread for 19.1 sec

== Tests result: FAILURE ==

1 test failed:
    test_signal

171 tests OK.

Total duration: 5 min 50 sec
Total tests: run=15,586 skipped=569
Total test files: run=172 failed=1
Result: FAILURE

[vstinner]

Random crash of test_signal on macos sonoma

Maybe the test_signal.test_stress_modifying_handlers() was already unstable before macOS 14: I saw a crash on macOS 12.7 (GHA macOS), see issue gh-110083. Maybe it's just that it was missed before I deplayed --rerun-fail on Python CIs.

[erlend-aasland]

I got a failure-then-success for test_signal recently: https://github.com/python/cpython/actions/runs/6415769021/job/17418272051?pr=110374

[Eclips4]

Random crash of test_signal on macos sonoma

Maybe the test_signal.test_stress_modifying_handlers() was already unstable before macOS 14: I saw a crash on macOS 12.7 (GHA macOS), see issue gh-110083.

I came to the same conclusion. It would be great if someone with macOS 13 (or older) try to reproduce it.
Reproducer: ./python.exe -m test test_signal -m test_stress_modifying_handlers --forever

[ronaldoussoren]

I caught the error in a debugger:

* thread #12, stop reason = EXC_BAD_ACCESS (code=257, address=0x1)
    frame #0: 0x0000000000000001
error: memory read failed for 0x1
Target 0: (python.exe) stopped.
(lldb) thread backtrace
* thread #12, stop reason = EXC_BAD_ACCESS (code=257, address=0x1)
  * frame #0: 0x0000000000000001
    frame #1: 0x0000000189ba3a24 libsystem_platform.dylib`_sigtramp + 56
    frame #2: 0x0000000189b74cc0 libsystem_pthread.dylib`pthread_kill + 288
    frame #3: 0x0000000189a4d540 libsystem_c.dylib`raise + 32
    frame #4: 0x00000001001f1748 python.exe`signal_raise_signal_impl(module=<unavailable>, signalnum=<unavailable>) at signalmodule.c:445:11 [opt]
    frame #5: 0x00000001001528c0 python.exe`_PyEval_EvalFrameDefault(tstate=0x000000014e6e39c0, frame=<unavailable>, throwflag=<unavailable>) at generated_cases.c.h:1134:19 [opt]
    frame #6: 0x0000000100053a98 python.exe`method_vectorcall [inlined] _PyObject_VectorcallTstate(tstate=0x000000014e6e39c0, callable=0x00000001008904a0, args=0x0000000170e06ef8, nargsf=1, kwnames=0x0000000000000000) at pycore_call.h:168:11 [opt]
    frame #7: 0x0000000100053a6c python.exe`method_vectorcall(method=<unavailable>, args=0x0000000100400450, nargsf=<unavailable>, kwnames=0x0000000000000000) at classobject.c:70:20 [opt]
    frame #8: 0x0000000100229588 python.exe`thread_run(boot_raw=0x0000600001790800) at _threadmodule.c:1204:21 [opt]
    frame #9: 0x00000001001c9ed4 python.exe`pythread_wrapper(arg=<unavailable>) at thread_pthread.h:236:5 [opt]
    frame #10: 0x0000000189b75034 libsystem_pthread.dylib`_pthread_start + 136

Note the SEGV on address 0x1 triggered from within libc. A second thread is in the sigaction implementation.

(lldb) thread list
Process 18586 stopped
  thread #1: tid = 0xe789b2, 0x0000000189b3aebc libsystem_kernel.dylib`__sigaction + 8, queue = 'com.apple.main-thread'
* thread #12: tid = 0xe7996d, 0x0000000000000001, stop reason = EXC_BAD_ACCESS (code=257, address=0x1)

This looks more and more like a bug in macOS. I'll see if I can whip up a reproducer in C.

[ronaldoussoren]

This is a reproducer in C that crashes almost immediately for me on a 14.1 system. The crash is similar to the one above:

* thread #2, stop reason = EXC_BAD_ACCESS (code=257, address=0x1)
    frame #0: 0x0000000000000001
error: memory read failed for 0x1
Target 1: (repro) stopped.
(lldb) thread backtrace
* thread #2, stop reason = EXC_BAD_ACCESS (code=257, address=0x1)
  * frame #0: 0x0000000000000001
    frame #1: 0x0000000189ba3a24 libsystem_platform.dylib`_sigtramp + 56
    frame #2: 0x0000000189b74cc0 libsystem_pthread.dylib`pthread_kill + 288
    frame #3: 0x0000000189a4d540 libsystem_c.dylib`raise + 32
    frame #4: 0x0000000100003e18 repro`raising_thread(arg=0x0000000000000000) at repro.c:17:9
    frame #5: 0x0000000189b75034 libsystem_pthread.dylib`_pthread_start + 136
(lldb) thread select 1
* thread #1, queue = 'com.apple.main-thread'
    frame #0: 0x0000000189b3aebc libsystem_kernel.dylib`__sigaction + 8
libsystem_kernel.dylib`:
->  0x189b3aebc <+8>:  b.lo   0x189b3aedc               ; <+40>
    0x189b3aec0 <+12>: pacibsp 
    0x189b3aec4 <+16>: stp    x29, x30, [sp, #-0x10]!
    0x189b3aec8 <+20>: mov    x29, sp
(lldb) thread backtrace
* thread #1, queue = 'com.apple.main-thread'
  * frame #0: 0x0000000189b3aebc libsystem_kernel.dylib`__sigaction + 8
    frame #1: 0x0000000189ba2d28 libsystem_platform.dylib`__platform_sigaction + 108
    frame #2: 0x0000000100003ea0 repro`update_signal at repro.c:34:9
    frame #3: 0x0000000100003f38 repro`main at repro.c:54:9
    frame #4: 0x00000001897f90e0 dyld`start + 2360
(lldb) 

Could one of you check my C code to make sure it doesn't contain a glaring issue? I intend to file an issue with Apple about this. I will test on older macOS versions over the weekend.

#include <signal.h>
#include <unistd.h>
#include <pthread.h>
#include <stdio.h>

static int sigcount = 0;
static int update_count = 0;

void sighandler(int sig)
{
    return;
}

void* raising_thread(void* arg)
{
    for (;;) {
        raise(SIGUSR1);
        sigcount += 1;
    }
    return NULL;
}

void update_signal(void)
{
    struct sigaction context, ocontext;

    if (update_count++ % 2 == 0) {
        context.sa_handler = sighandler;
    } else {
        context.sa_handler = SIG_IGN;
    }
    sigemptyset(&context.sa_mask);
    context.sa_flags = SA_ONSTACK;
    if (sigaction(SIGUSR1, &context, &ocontext) == -1) {
        perror("sigaction");
        _exit(1);
    }
}

int main(void)
{
    alarm(30);
    update_signal();

    pthread_t thread;


    if (pthread_create(&thread, NULL, &raising_thread, NULL) !=  0) {
        perror("pthread_create");
        return 1;
    }

    for (;;) {
        update_signal();
    }
}

[ronaldoussoren]

I've filed an issue with Apple about this: FB13453490

[ned-deily]

FWIW, your C test also crashed on macOS 13.5.1 and 12.6.1 (both Apple Silicon).