Merge of #13227 to trunk cause deadlocks on parallel Sys tests #13713
Description
The merge of #13227 seems to have introduced a regression causing a deadlock on parallel multicoretests of Sys.
$ dune exec src/sys/stm_tests.exe -- -v
random seed: 437123728
generated error fail pass / total time test name
[✓] 1000 0 0 1000 / 1000 1.9s STM Sys test sequential
[ ] 60 0 0 60 / 200 5.0s STM Sys test parallel
This test worked fine up until yesterday's previous merge commit 86470c2 - Merge pull request #13694 from mndrix/manual-hash-variant.
Attaching gdb to the above reveals the following stack trace:
Stack trace:
Thread 4 (Thread 0x7c874f400640 (LWP 31382) "stm_tests.exe"):
#0 futex_wait (private=0, expected=2, futex_word=0x569f07463f30) at ../sysdeps/nptl/futex-internal.h:146
#1 __GI___lll_lock_wait (futex=futex@entry=0x569f07463f30, private=0) at ./nptl/lowlevellock.c:49
#2 0x00007c8765a98002 in lll_mutex_lock_optimized (mutex=0x569f07463f30) at ./nptl/pthread_mutex_lock.c:48
#3 ___pthread_mutex_lock (mutex=mutex@entry=0x569f07463f30) at ./nptl/pthread_mutex_lock.c:93
#4 0x0000569eedb75841 in caml_plat_lock_blocking (m=0x569f07463f30) at runtime/caml/platform.h:455
#5 backup_thread_func (v=0x569f07463e90) at runtime/domain.c:1088
#6 0x00007c8765a94ac3 in start_thread (arg=<optimized out>) at ./nptl/pthread_create.c:442
#7 0x00007c8765b26850 in clone3 () at ../sysdeps/unix/sysv/linux/x86_64/clone3.S:81
Thread 3 (Thread 0x7c8754e00640 (LWP 31381) "stm_tests.exe"):
#0 syscall () at ../sysdeps/unix/sysv/linux/x86_64/syscall.S:38
#1 0x0000569eedb8f0e5 in caml_plat_futex_wait (undesired=2, ftx=0x569eedc72ae0 <stw_request>) at runtime/platform.c:307
#2 latchlike_wait (contested=2, unreleased=1, ftx=0x569eedc72ae0 <stw_request>) at runtime/platform.c:347
#3 caml_plat_latch_wait (latch=latch@entry=0x569eedc72ae0 <stw_request>) at runtime/platform.c:351
#4 0x0000569eedb7441e in caml_plat_barrier_wait (barrier=0x569eedc72ae0 <stw_request>) at runtime/caml/platform.h:320
#5 stw_wait_for_running (domain=0x7c8750002b80) at runtime/domain.c:1482
#6 stw_api_barrier (domain=domain@entry=0x7c8750002b80) at runtime/domain.c:1492
#7 0x0000569eedb75b68 in caml_try_run_on_all_domains_with_spin_work (sync=sync@entry=1, handler=handler@entry=0x569eedb8dbb0 <caml_stw_empty_minor_heap>, data=data@entry=0x0, leader_setup=leader_setup@entry=0x569eedb8c760 <caml_empty_minor_heap_setup>, enter_spin_callback=enter_spin_callback@entry=0x569eedb8c790 <caml_do_opportunistic_major_slice>, enter_spin_data=enter_spin_data@entry=0x0) at runtime/domain.c:1710
#8 0x0000569eedb8dd02 in caml_try_empty_minor_heap_on_all_domains () at runtime/minor_gc.c:859
#9 caml_empty_minor_heaps_once () at runtime/minor_gc.c:882
#10 0x0000569eedb762c2 in caml_domain_terminate (last=last@entry=false) at runtime/domain.c:2033
#11 0x0000569eedb76754 in domain_thread_func (v=<optimized out>) at runtime/domain.c:1267
#12 0x00007c8765a94ac3 in start_thread (arg=<optimized out>) at ./nptl/pthread_create.c:442
#13 0x00007c8765b26850 in clone3 () at ../sysdeps/unix/sysv/linux/x86_64/clone3.S:81
Thread 2 (Thread 0x7c874fe00640 (LWP 22201) "stm_tests.exe"):
#0 __futex_abstimed_wait_common64 (private=0, cancel=true, abstime=0x0, op=393, expected=0, futex_word=0x569f07463e74) at ./nptl/futex-internal.c:57
#1 __futex_abstimed_wait_common (cancel=true, private=0, abstime=0x0, clockid=0, expected=0, futex_word=0x569f07463e74) at ./nptl/futex-internal.c:87
#2 __GI___futex_abstimed_wait_cancelable64 (futex_word=futex_word@entry=0x569f07463e74, expected=expected@entry=0, clockid=clockid@entry=0, abstime=abstime@entry=0x0, private=private@entry=0) at ./nptl/futex-internal.c:139
#3 0x00007c8765a93a41 in __pthread_cond_wait_common (abstime=0x0, clockid=0, mutex=0x569f07463e20, cond=0x569f07463e48) at ./nptl/pthread_cond_wait.c:503
#4 ___pthread_cond_wait (cond=cond@entry=0x569f07463e48, mutex=mutex@entry=0x569f07463e20) at ./nptl/pthread_cond_wait.c:627
#5 0x0000569eedb8ef0d in caml_plat_wait (cond=cond@entry=0x569f07463e48, mut=mut@entry=0x569f07463e20) at runtime/platform.c:146
#6 0x0000569eedb7594f in backup_thread_func (v=0x569f07463d80) at runtime/domain.c:1091
#7 0x00007c8765a94ac3 in start_thread (arg=<optimized out>) at ./nptl/pthread_create.c:442
#8 0x00007c8765b26850 in clone3 () at ../sysdeps/unix/sysv/linux/x86_64/clone3.S:81
Thread 1 (Thread 0x7c8765d03740 (LWP 20146) "stm_tests.exe"):
#0 futex_wait (private=0, expected=2, futex_word=0x569f074bc530) at ../sysdeps/nptl/futex-internal.h:146
#1 __GI___lll_lock_wait (futex=futex@entry=0x569f074bc530, private=0) at ./nptl/lowlevellock.c:49
#2 0x00007c8765a972a3 in __pthread_mutex_cond_lock (mutex=mutex@entry=0x569f074bc530) at ../nptl/pthread_mutex_lock.c:93
#3 0x00007c8765a93934 in __pthread_cond_wait_common (abstime=0x0, clockid=0, mutex=0x569f074bc530, cond=0x569f074cee20) at ./nptl/pthread_cond_wait.c:616
#4 ___pthread_cond_wait (cond=cond@entry=0x569f074cee20, mutex=mutex@entry=0x569f074bc530) at ./nptl/pthread_cond_wait.c:627
#5 0x0000569eedb9539e in sync_condvar_wait (m=0x569f074bc530, c=0x569f074cee20) at runtime/sync_posix.h:116
#6 caml_ml_condition_wait (wcond=<optimized out>, wmut=<optimized out>) at runtime/sync.c:193
#7 <signal handler called>
#8 0x0000569eedb2d3be in camlStdlib__Domain$loop_752 () at domain.ml:292
#9 0x0000569eedb2c4ce in camlStdlib__Mutex$protect_277 () at mutex.ml:28
#10 0x0000569eedb2d35c in camlStdlib__Domain$join_749 () at domain.ml:297
#11 0x0000569eedab3bcf in camlSTM_domain$run_par_660 () at lib/STM_domain.ml:31
#12 0x0000569eedab3e44 in camlSTM_domain$agree_prop_par_715 () at lib/STM_domain.ml:39
#13 0x0000569eedab9b6a in camlUtil$fun_2471 () at lib/util.ml:5
#14 0x0000569eedacf4c0 in camlQCheck2$loop_3994 () at src/core/QCheck2.ml:1644
#15 0x0000569eedacf404 in camlQCheck2$run_law_3989 () at src/core/QCheck2.ml:1649
#16 0x0000569eedad01b1 in camlQCheck2$check_state_input_4057 () at src/core/QCheck2.ml:1771
#17 0x0000569eedad060e in camlQCheck2$check_cell_inner_9734 () at src/core/QCheck2.ml:1841
#18 0x0000569eedac1ad6 in camlQCheck_base_runner$aux_map_1444 () at src/runner/QCheck_base_runner.ml:422
#19 0x0000569eedb09333 in camlStdlib__List$map_334 () at list.ml:87
#20 0x0000569eedac13f3 in camlQCheck_base_runner$run_tests_inner_2444 () at src/runner/QCheck_base_runner.ml:431
#21 0x0000569eedac215d in camlQCheck_base_runner$run_tests_main_inner_3052 () at src/runner/QCheck_base_runner.ml:475
#22 0x0000569eedab348f in camlDune__exe__Stm_tests$entry () at src/sys/stm_tests.ml:329
#23 0x0000569eedaaa3f7 in caml_program ()
#24 <signal handler called>
#25 0x0000569eedb9a196 in caml_startup_common (pooling=<optimized out>, argv=0x7ffe7c62ec68) at runtime/startup_nat.c:127
#26 caml_startup_common (argv=0x7ffe7c62ec68, pooling=<optimized out>) at runtime/startup_nat.c:86
#27 0x0000569eedb9a20f in caml_startup_exn (argv=<optimized out>) at runtime/startup_nat.c:134
#28 caml_startup (argv=<optimized out>) at runtime/startup_nat.c:139
#29 caml_main (argv=<optimized out>) at runtime/startup_nat.c:146
#30 0x0000569eedaa9f72 in main (argc=<optimized out>, argv=<optimized out>) at runtime/main.c:37
To recreate the issue:
$ opam switch create . --empty
$ opam pin add -k path --inplace-build ocaml-variants.5.4.0+trunk-with-13227 .
$ opam install qcheck-core
$ git clone [email protected]:ocaml-multicore/multicoretests.git
$ cd multicoretests
$ dune exec src/sys/stm_tests.exe -- -v