New submission from 盛茂家:
This corruption is so strange that I can't even reappear it.
(gdb) bt
#0 0x00000000006d7e9f in func_dealloc.lto_priv () at Objects/funcobject.c:451
#1 0x00000000005c730b in ask (printException=true, okIfFunctionNull=<optimized
out>, errorPrefix=0x884bca "", pArgs=0x300dd440, pFunction=0x1d1f1de8) at
script.cpp:758
#2 Script::call(_object*, _object*, char const*, bool) (pFunction=0x1d1f1de8,
pArgs=0x300dd440, errorPrefix=0x884bca "", okIfFunctionNull=<optimized out>) at
/home/smj/original/bwengine/src/lib/pyscript/script.ipp:25
now I know that in _PyObject_GC_UNTRACK,
(gdb) x/40x op
0x1d1f1da8: 0x6d 0x6f 0x6e 0x5f 0x73 0x65 0x72 0x76
0x1d1f1db0: 0x65 0x72 0x2f 0x72 0x75 0x6e 0x5f 0x73
0x1d1f1db8: 0x63 0x72 0x69 0x70 0x74 0x2e 0x70 0x79
0x1d1f1dc0: 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00
0x1d1f1dc8: 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00
-> mov -0x20(%rdi),%rax -> %rax = 0
0x1d1f1dd0: 0xf8 0x4d 0xc5 0x2a 0x00 0x00 0x00 0x00
-> mov -0x18(%rdi),%rdx -> %rdx = 0x2ac54df8
0x1d1f1dd8: 0xfe 0xff 0xff 0xff 0xff 0xff 0xff 0xff
0x1d1f1de0: 0xff 0xff 0xff 0xff 0xff 0xff 0xff 0xff
0x1d1f1de8: 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00
<--PyFunctionObject ob_refcnt 0 %rdi
0x1d1f1df0: 0x00 0xff 0xba 0x00 0x00 0x00 0x00 0x00
-> _typeobject* -> name function
0x1d1f1df8: 0x30 0x1e 0x78 0x21 0x00 0x00 0x00 0x00
0x1d1f1e00: 0x28 0xaa 0xba 0x24 0x00 0x00 0x00 0x00
0x1d1f1e08: 0x10 0x53 0xfb 0x2f 0x00 0x00 0x00 0x00
When corruption,
the ob_refcnt is 0, the gc_refs is _PyGC_REFS_UNTRACKED, the gc_next is NULL
but gc_prev is not.
How can it be? Is there some unknown bug in GC?
----------
components: Interpreter Core
messages: 297696
nosy: 盛茂家
priority: normal
severity: normal
status: open
title: _PyObject_GC_UNTRACK corruption when call a lambda function with C API
type: crash
versions: Python 2.7
_______________________________________
Python tracker <rep...@bugs.python.org>
<http://bugs.python.org/issue30852>
_______________________________________
_______________________________________________
Python-bugs-list mailing list
Unsubscribe:
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
