Bug #17882: bootstraptest/test_ractor.rb:224 segfaults on Cygwin

• Status: Assigned
• Assignee: cygwin
• Target version: -
• ruby -v: ruby 3.1.0dev (2021-05-21T09:28:24Z master 50a534a152) [x86_64-cygwin]
• Backport: 2.6: UNKNOWN, 2.7: UNKNOWN, 3.0: UNKNOWN

[xtkoba (Tee KOBAYASHI)]

The attached test code is excerpted from bootstraptest/test_ractor.rb:224. This code causes a segmentation fault every time when run on x86_64-cygwin. There are at least 3 types of dying messages, as shown below.

I have no idea whether this is relevant to #17878, which is an issue with the very same test code.

Type 1 (null pointer dereference):

Thread 6 received signal SIGSEGV, Segmentation fault.
[Switching to Thread 5368]
VM_CF_BLOCK_HANDLER (cfp=0x0) at ../vm.c:115
115         const VALUE *ep = VM_CF_LEP(cfp);
(gdb) bt
#0  VM_CF_BLOCK_HANDLER (cfp=0x0) at ../vm.c:115
#1  0x00007ff6acedb495 in rb_vm_frame_block_handler (cfp=<optimized out>) at ../vm.c:128
#2  0x00007ff6acdd954e in pass_passed_block_handler (ec=0x80012bba0) at ../eval_intern.h:17
#3  rb_obj_call_init_kw (obj=obj@entry=123145240968920, argc=argc@entry=1, argv=argv@entry=0xffd0ca08, kw_splat=kw_splat@entry=0) at ../eval.c:1724
#4  0x00007ff6ace3efc2 in rb_class_new_instance (argc=argc@entry=1, argv=argv@entry=0xffd0ca08, klass=klass@entry=123145300575160) at ../object.c:2192
#5  0x00007ff6acdd1c30 in rb_exc_new_str (etype=etype@entry=123145300575160, str=<optimized out>) at ../error.c:1123
#6  0x00007ff6acdd29ad in rb_vraise (exc=123145300575160, fmt=<optimized out>, ap=<optimized out>) at ../error.c:2922
#7  0x00007ff6acdd29e5 in rb_raise (exc=0, fmt=0x0) at ../error.c:2930
#8  0x00007ff6acdf90e8 in rb_io_check_initialized (fptr=0x0) at ../io.c:767
#9  rb_io_check_initialized (fptr=<optimized out>) at ../io.c:764
#10 0x00007ff6acdf90fb in rb_io_check_closed (fptr=0x0) at ../io.c:774
#11 0x00007ff6ace011ec in prep_stdio (f=0x18023acb8 <reent_data+1336>, fmode=fmode@entry=1, klass=123145300573360, klass@entry=140697440105184, path=path@entry=0x7ff6acf158e0 <prelude_table+2944> "<STDIN>") at ../io.c:8239
#12 0x00007ff6ace0122f in rb_io_prep_stdin () at ../io.c:8255
#13 0x00007ff6acebc980 in thread_start_func_2 (th=0x0, th@entry=0x80011cbf0, stack_start=stack_start@entry=0xffd0ccf8) at ../thread.c:801
#14 0x00007ff6acebd032 in thread_start_func_1 (th_ptr=<optimized out>) at ../thread_pthread.c:1035
#15 0x000000018016d45f in pthread::thread_init_wrapper(void*) () from target:/usr/bin/cygwin1.dll
#16 0x00000001800ddbba in pthread_wrapper () from target:/usr/bin/cygwin1.dll
#17 0x0000000000000000 in ?? ()
Backtrace stopped: previous frame inner to this frame (corrupt stack?)

Type 2 (rb_gc_mark(): <address> is T_ZOMBIE):

Thread 7 hit Breakpoint 1, rb_bug (fmt=0x7ff6acf1396a <stat_data_type+8138> "rb_gc_mark(): %p is T_ZOMBIE") at ../error.c:782
782     {
(gdb) bt
#0  rb_bug (fmt=0x7ff6acf1396a <stat_data_type+8138> "rb_gc_mark(): %p is T_ZOMBIE") at ../error.c:782
#1  0x00007ff6acdeae8f in gc_mark_children (objspace=objspace@entry=0x800053970, obj=obj@entry=123145240171400) at ../gc.c:6934
#2  0x00007ff6acdeafb0 in gc_mark_stacked_objects (objspace=0x800053970, incremental=incremental@entry=0, count=count@entry=0) at ../gc.c:6961
#3  0x00007ff6acded415 in gc_mark_stacked_objects_all (objspace=0x800053970) at ../gc.c:7001
#4  gc_marks_rest (objspace=objspace@entry=0x800053970) at ../gc.c:7972
#5  0x00007ff6acdec08c in gc_marks (full_mark=<optimized out>, objspace=0x800053970) at ../gc.c:8028
#6  gc_start (objspace=objspace@entry=0x800053970, reason=<optimized out>, reason@entry=256) at ../gc.c:8862
#7  0x00007ff6acdee522 in heap_prepare (heap=0x800053998, objspace=0x800053970) at ../gc.c:2153
#8  heap_next_freepage (heap=0x800053998, objspace=0x800053970) at ../gc.c:2444
#9  ractor_cache_slots (objspace=objspace@entry=0x800053970, cr=cr@entry=0x800135d60) at ../gc.c:2476
#10 0x00007ff6acdee61a in newobj_slowpath (alloc_size=<optimized out>, wb_protected=0, cr=0x800135d60, objspace=0x800053970, flags=11, klass=123145300573360) at ../gc.c:2517
#11 newobj_slowpath_wb_unprotected (klass=123145300573360, flags=11, objspace=0x800053970, cr=0x800135d60, alloc_size=<optimized out>) at ../gc.c:2547
#12 0x00007ff6acdee815 in newobj_of0 (klass=klass@entry=123145300573360, flags=flags@entry=11, wb_protected=wb_protected@entry=0, cr=<optimized out>, alloc_size=<optimized out>) at ../gc.c:2585
#13 0x00007ff6acdee86d in newobj_of (klass=klass@entry=123145300573360, flags=flags@entry=11, v1=v1@entry=0, v2=v2@entry=0, v3=v3@entry=0, wb_protected=wb_protected@entry=0, alloc_size=40) at ../gc.c:2594
#14 0x00007ff6acdeec23 in rb_wb_unprotected_newobj_of (klass=klass@entry=123145300573360, flags=flags@entry=11, size=40, size@entry=0) at ../gc.c:2610
#15 0x00007ff6acdf666c in io_alloc (klass=klass@entry=123145300573360) at ../io.c:1038
#16 0x00007ff6acdfbca9 in prep_io (fd=2, fmode=fmode@entry=65546, klass=klass@entry=123145300573360, path=path@entry=0x7ff6acf158f1 <prelude_table+2961> "<STDERR>") at ../io.c:8206
#17 0x00007ff6ace011d4 in prep_stdio (f=0x18023ae28 <reent_data+1704>, fmode=fmode@entry=10, klass=123145300573360, klass@entry=34361007456, path=path@entry=0x7ff6acf158f1 <prelude_table+2961> "<STDERR>") at ../io.c:8237
#18 0x00007ff6ace01295 in rb_io_prep_stderr () at ../io.c:8267
#19 0x00007ff6acebc9a6 in thread_start_func_2 (th=0x0, th@entry=0x800134550, stack_start=stack_start@entry=0xffa0ccf8) at ../thread.c:803
#20 0x00007ff6acebd032 in thread_start_func_1 (th_ptr=<optimized out>) at ../thread_pthread.c:1035
#21 0x000000018016d45f in pthread::thread_init_wrapper(void*) () from target:/usr/bin/cygwin1.dll
#22 0x00000001800ddbba in pthread_wrapper () from target:/usr/bin/cygwin1.dll
#23 0x0000000000000000 in ?? ()
Backtrace stopped: previous frame inner to this frame (corrupt stack?)

Type 3 (try to mark T_NONE object):

<OBJ_INFO:gc_mark_ptr@../gc.c:6580> 0x00006fffffe7fb70 [0 M   ] T_NONE

------------------------------------------------------------------------

Thread 7 hit Breakpoint 1, rb_bug (fmt=fmt@entry=0x7ff6acf138a0 <stat_data_type+7936> "try to mark T_NONE object") at ../error.c:782
782     {
(gdb) bt
#0  rb_bug (fmt=fmt@entry=0x7ff6acf138a0 <stat_data_type+7936> "try to mark T_NONE object") at ../error.c:782
#1  0x00007ff6acdea5c9 in gc_mark_ptr (objspace=0x800053970, obj=123145300736880) at ../gc.c:6581
#2  0x00007ff6ace672ea in ractor_mark (ptr=0x800117240) at ../ractor.c:197
#3  0x00007ff6acdeafb0 in gc_mark_stacked_objects (objspace=objspace@entry=0x800053970, incremental=incremental@entry=1, count=count@entry=2147483647) at ../gc.c:6961
#4  0x00007ff6acded3f9 in gc_mark_stacked_objects_incremental (count=2147483647, objspace=0x800053970) at ../gc.c:6995
#5  gc_marks_rest (objspace=objspace@entry=0x800053970) at ../gc.c:7968
#6  0x00007ff6acdee4f1 in gc_marks_continue (heap=0x800053998, objspace=0x800053970) at ../gc.c:8012
#7  heap_prepare (heap=0x800053998, objspace=0x800053970) at ../gc.c:2148
#8  heap_next_freepage (heap=0x800053998, objspace=0x800053970) at ../gc.c:2444
#9  ractor_cache_slots (objspace=objspace@entry=0x800053970, cr=cr@entry=0x800129350) at ../gc.c:2476
#10 0x00007ff6acdee61a in newobj_slowpath (alloc_size=<optimized out>, wb_protected=0, cr=0x800129350, objspace=0x800053970, flags=11, klass=123145300573360) at ../gc.c:2517
#11 newobj_slowpath_wb_unprotected (klass=123145300573360, flags=11, objspace=0x800053970, cr=0x800129350, alloc_size=<optimized out>) at ../gc.c:2547
#12 0x00007ff6acdee815 in newobj_of0 (klass=klass@entry=123145300573360, flags=flags@entry=11, wb_protected=wb_protected@entry=0, cr=<optimized out>, alloc_size=<optimized out>) at ../gc.c:2585
#13 0x00007ff6acdee86d in newobj_of (klass=klass@entry=123145300573360, flags=flags@entry=11, v1=v1@entry=0, v2=v2@entry=0, v3=v3@entry=0, wb_protected=wb_protected@entry=0, alloc_size=40) at ../gc.c:2594
#14 0x00007ff6acdeec23 in rb_wb_unprotected_newobj_of (klass=klass@entry=123145300573360, flags=flags@entry=11, size=40, size@entry=0) at ../gc.c:2610
#15 0x00007ff6acdf666c in io_alloc (klass=klass@entry=123145300573360) at ../io.c:1038
#16 0x00007ff6acdfbca9 in prep_io (fd=0, fmode=fmode@entry=65537, klass=klass@entry=123145300573360, path=path@entry=0x7ff6acf158e0 <prelude_table+2944> "<STDIN>") at ../io.c:8206
#17 0x00007ff6ace011d4 in prep_stdio (f=0x18023acb8 <reent_data+1336>, fmode=fmode@entry=1, klass=123145300573360, klass@entry=140697440105184, path=path@entry=0x7ff6acf158e0 <prelude_table+2944> "<STDIN>") at ../io.c:8237
#18 0x00007ff6ace0122f in rb_io_prep_stdin () at ../io.c:8255
#19 0x00007ff6acebc980 in thread_start_func_2 (th=0x0, th@entry=0x80010dcb0, stack_start=stack_start@entry=0xffa0ccf8) at ../thread.c:801
#20 0x00007ff6acebd032 in thread_start_func_1 (th_ptr=<optimized out>) at ../thread_pthread.c:1035
#21 0x000000018016d45f in pthread::thread_init_wrapper(void*) () from target:/usr/bin/cygwin1.dll
#22 0x00000001800ddbba in pthread_wrapper () from target:/usr/bin/cygwin1.dll
#23 0x0000000000000000 in ?? ()
Backtrace stopped: previous frame inner to this frame (corrupt stack?)

[xtkoba (Tee KOBAYASHI)]

This no longer happens as of 4cc44bd819, though not bisected. Thanks.

[hsbt (Hiroshi SHIBATA)]

• Status changed from Open to Assigned
• Assignee set to ko1 (Koichi Sasada)

[ko1 (Koichi Sasada)]

Sorry now I don't have cygwin.
Should we support it?

[jeremyevans0 (Jeremy Evans)]

ko1 (Koichi Sasada) wrote in #note-3:

Sorry now I don't have cygwin.
Should we support it?

We currently have 16 open bugs related to cygwin: http://tagged-ruby-bugs.jeremyevans.net/#-cygwin. Considering WSL (Windows Subsystem for Linux) has been around for over 5 years, and cygwin has no platform maintainer according to doc/maintainers.rdoc, I think it is no longer necessary to support cygwin. I recommend dropping cygwin support and closing related tickets.

[naruse (Yui NARUSE)]

• Assignee changed from ko1 (Koichi Sasada) to cygwin