segmentation fault (Trac #197)

@nilswagner01

Original ticket http://projects.scipy.org/numpy/ticket/197 on 2006-07-21 by @nilswagner01, assigned to @teoliphant.

numpy.test(1,10) results in a segfault.
Here is a backtrace

test_boolean (numpy.core.tests.test_numeric.test_index) ... ok
test_divideerr (numpy.core.tests.test_numeric.test_seterr)
Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 16384 (LWP 17793)]
0x00002aaaaadf4cf0 in __pthread_lock () from /lib64/libpthread.so.0
(gdb) bt
#0  0x00002aaaaadf4cf0 in __pthread_lock () from /lib64/libpthread.so.0
#1  0x00002aaaaadf4725 in sem_post@@GLIBC_2.2.5 () from /lib64/libpthread.so.0
#2  0x00002aaaaac76059 in PyThread_release_lock (lock=<value optimized out>) at thread_pthread.h:338
#3  0x00002aaaabff40da in _error_handler (method=2, errobj=0x2aaaadf98cb0, errtype=0x2aaaabfff08f "divide by zero",
    retstatus=1) at ufuncobject.c:519
#4  0x00002aaaabff42ea in PyUFunc_handlefperr (errmask=2, errobj=0x2aaaadf98cb0, retstatus=1) at ufuncobject.c:547
#5  0x00002aaaabffb46d in PyUFunc_GenericFunction (self=0x559710, args=<value optimized out>, mps=0x7fffffc0d2b0)
    at ufuncobject.c:1450
#6  0x00002aaaabffb540 in ufunc_generic_call (self=0x559710, args=0x2aaaadf98710) at ufuncobject.c:2694
#7  0x00002aaaaabfa760 in PyObject_Call (func=<value optimized out>, arg=<value optimized out>,
    kw=<value optimized out>) at abstract.c:1751
#8  0x00002aaaaabfa883 in PyObject_CallFunction (callable=0x559710, format=<value optimized out>) at abstract.c:1793
#9  0x00002aaaaabf8508 in binary_op1 (v=0x888970, w=0x888b40, op_slot=24) at abstract.c:371
#10 0x00002aaaaabf8771 in binary_op (v=0x888970, w=0x888b40, op_slot=<value optimized out>,
    op_name=0x2aaaaac8b2c8 "/") at abstract.c:422
#11 0x00002aaaaac521f1 in PyEval_EvalFrame (f=0x7a9480) at ceval.c:1060
#12 0x00002aaaaac53b97 in PyEval_EvalFrame (f=0x5d5cd0) at ceval.c:3629
#13 0x00002aaaaac55404 in PyEval_EvalCodeEx (co=0x2aaaabbeece0, globals=<value optimized out>,
    locals=<value optimized out>, args=0x2aaaae2bdd10, argcount=2, kws=0x82c7d0, kwcount=0, defs=0x2aaaabbf7b28,
    defcount=1, closure=0x0) at ceval.c:2730
#14 0x00002aaaaac0e9af in function_call (func=0x2aaaabc05758, arg=0x2aaaae2bdcf8, kw=<value optimized out>)
    at funcobject.c:548
#15 0x00002aaaaabfa760 in PyObject_Call (func=<value optimized out>, arg=<value optimized out>,
    kw=<value optimized out>) at abstract.c:1751
#16 0x00002aaaaac532e2 in PyEval_EvalFrame (f=0x5e81c0) at ceval.c:3824
#17 0x00002aaaaac55404 in PyEval_EvalCodeEx (co=0x2aaaabbeed50, globals=<value optimized out>,
    locals=<value optimized out>, args=0x2aaaae1338d8, argcount=2, kws=0x0, kwcount=0, defs=0x0, defcount=0,
    closure=0x0) at ceval.c:2730
#18 0x00002aaaaac0e9af in function_call (func=0x2aaaabc057d0, arg=0x2aaaae1338c0, kw=<value optimized out>)
    at funcobject.c:548
#19 0x00002aaaaabfa760 in PyObject_Call (func=<value optimized out>, arg=<value optimized out>,
    kw=<value optimized out>) at abstract.c:1751
#20 0x00002aaaaac02131 in instancemethod_call (func=<value optimized out>, arg=0x2aaaae1338c0, kw=0x0)
---Type <return> to continue, or q <return> to quit---
    at classobject.c:2431
#21 0x00002aaaaabfa760 in PyObject_Call (func=<value optimized out>, arg=<value optimized out>,
    kw=<value optimized out>) at abstract.c:1751
#22 0x00002aaaaac5380d in PyEval_EvalFrame (f=0x6f7290) at ceval.c:3755
#23 0x00002aaaaac55404 in PyEval_EvalCodeEx (co=0x2aaaabbde7a0, globals=<value optimized out>,
    locals=<value optimized out>, args=0x2aaaae2bcd10, argcount=2, kws=0x0, kwcount=0, defs=0x2aaaabc09628,
    defcount=1, closure=0x0) at ceval.c:2730
#24 0x00002aaaaac0e9af in function_call (func=0x2aaaabc07d70, arg=0x2aaaae2bccf8, kw=<value optimized out>)
    at funcobject.c:548
#25 0x00002aaaaabfa760 in PyObject_Call (func=<value optimized out>, arg=<value optimized out>,
    kw=<value optimized out>) at abstract.c:1751
#26 0x00002aaaaac02131 in instancemethod_call (func=<value optimized out>, arg=0x2aaaae2bccf8, kw=0x0)
    at classobject.c:2431
#27 0x00002aaaaabfa760 in PyObject_Call (func=<value optimized out>, arg=<value optimized out>,
    kw=<value optimized out>) at abstract.c:1751
#28 0x00002aaaaac33b0a in slot_tp_call (self=<value optimized out>, args=0x2aaaadf0ca90, kwds=0x0)
    at typeobject.c:4526
#29 0x00002aaaaabfa760 in PyObject_Call (func=<value optimized out>, arg=<value optimized out>,
    kw=<value optimized out>) at abstract.c:1751
#30 0x00002aaaaac5380d in PyEval_EvalFrame (f=0x5d5850) at ceval.c:3755
#31 0x00002aaaaac55404 in PyEval_EvalCodeEx (co=0x2aaaabbf27a0, globals=<value optimized out>,
    locals=<value optimized out>, args=0x2aaaae2bd530, argcount=2, kws=0x71e290, kwcount=0, defs=0x0, defcount=0,
    closure=0x0) at ceval.c:2730
#32 0x00002aaaaac0e9af in function_call (func=0x2aaaabc06050, arg=0x2aaaae2bd518, kw=<value optimized out>)
    at funcobject.c:548
#33 0x00002aaaaabfa760 in PyObject_Call (func=<value optimized out>, arg=<value optimized out>,
    kw=<value optimized out>) at abstract.c:1751
#34 0x00002aaaaac532e2 in PyEval_EvalFrame (f=0x5d2c70) at ceval.c:3824
#35 0x00002aaaaac55404 in PyEval_EvalCodeEx (co=0x2aaaabbf2810, globals=<value optimized out>,
    locals=<value optimized out>, args=0x2aaaad88b6e0, argcount=2, kws=0x0, kwcount=0, defs=0x0, defcount=0,
    closure=0x0) at ceval.c:2730
#36 0x00002aaaaac0e9af in function_call (func=0x2aaaabc060c8, arg=0x2aaaad88b6c8, kw=<value optimized out>)
    at funcobject.c:548
---Type <return> to continue, or q <return> to quit---
#37 0x00002aaaaabfa760 in PyObject_Call (func=<value optimized out>, arg=<value optimized out>,
    kw=<value optimized out>) at abstract.c:1751
#38 0x00002aaaaac02131 in instancemethod_call (func=<value optimized out>, arg=0x2aaaad88b6c8, kw=0x0)
    at classobject.c:2431
#39 0x00002aaaaabfa760 in PyObject_Call (func=<value optimized out>, arg=<value optimized out>,
    kw=<value optimized out>) at abstract.c:1751
#40 0x00002aaaaac33b0a in slot_tp_call (self=<value optimized out>, args=0x2aaaae136f10, kwds=0x0)
    at typeobject.c:4526
#41 0x00002aaaaabfa760 in PyObject_Call (func=<value optimized out>, arg=<value optimized out>,
    kw=<value optimized out>) at abstract.c:1751
#42 0x00002aaaaac5380d in PyEval_EvalFrame (f=0x633460) at ceval.c:3755
#43 0x00002aaaaac53b97 in PyEval_EvalFrame (f=0x67a970) at ceval.c:3629
#44 0x00002aaaaac55404 in PyEval_EvalCodeEx (co=0x2aaaabbe4b20, globals=<value optimized out>,
    locals=<value optimized out>, args=0x62baf8, argcount=3, kws=0x62bb10, kwcount=0, defs=0x2aaaabc0f2a8,
    defcount=2, closure=0x0) at ceval.c:2730
#45 0x00002aaaaac53aba in PyEval_EvalFrame (f=0x62b960) at ceval.c:3640
#46 0x00002aaaaac55404 in PyEval_EvalCodeEx (co=0x2aaaaab24b90, globals=<value optimized out>,
    locals=<value optimized out>, args=0x541af0, argcount=2, kws=0x541b00, kwcount=0, defs=0x2aaaacc9a890,
    defcount=2, closure=0x0) at ceval.c:2730
#47 0x00002aaaaac53aba in PyEval_EvalFrame (f=0x541960) at ceval.c:3640
#48 0x00002aaaaac55404 in PyEval_EvalCodeEx (co=0x2aaaacca33b0, globals=<value optimized out>,
    locals=<value optimized out>, args=0x0, argcount=0, kws=0x0, kwcount=0, defs=0x0, defcount=0, closure=0x0)
    at ceval.c:2730
#49 0x00002aaaaac556d2 in PyEval_EvalCode (co=<value optimized out>, globals=<value optimized out>,
    locals=<value optimized out>) at ceval.c:484
#50 0x00002aaaaac70719 in run_node (n=<value optimized out>, filename=<value optimized out>, globals=0x503b50,
    locals=0x503b50, flags=<value optimized out>) at pythonrun.c:1265
#51 0x00002aaaaac71bc7 in PyRun_InteractiveOneFlags (fp=<value optimized out>, filename=0x2aaaaac95e73 "<stdin>",
    flags=0x7fffffc0faf0) at pythonrun.c:762
#52 0x00002aaaaac71cbe in PyRun_InteractiveLoopFlags (fp=0x2aaaab809e00, filename=0x2aaaaac95e73 "<stdin>",
    flags=0x7fffffc0faf0) at pythonrun.c:695
#53 0x00002aaaaac7221c in PyRun_AnyFileExFlags (fp=0x2aaaab809e00, filename=0x2aaaaac95e73 "<stdin>", closeit=0,
    flags=0x7fffffc0faf0) at pythonrun.c:658
---Type <return> to continue, or q <return> to quit---
#54 0x00002aaaaac77b25 in Py_Main (argc=<value optimized out>, argv=0x7fffffc127bc) at main.c:484
#55 0x00002aaaab603ced in __libc_start_main () from /lib64/libc.so.6
#56 0x00000000004006ea in _start () at start.S:113
#57 0x00007fffffc0fb88 in ?? ()
#58 0x00002aaaaabc19c0 in rtld_errno () from /lib64/ld-linux-x86-64.so.2

@teoliphant

@teoliphant wrote on 2006-07-21

Please tell us the version of NumPy and Python you are running when reporting a segfault. Also use

to enclose text.

@nilswagner01

@nilswagner01 wrote on 2006-07-21

Numpy version 1.1.2878

Scipy version 0.5.0.2115

test_divideerr (numpy.core.tests.test_numeric.test_seterr)Segmentation fault

Python 2.4.1 (#1, Sep 12 2005, 23:33:18)

[GCC 4.0.2 20050901 (prerelease) (SUSE Linux)] on linux2

@alberts

@alberts wrote on 2006-07-21

Tests of !NumPy 1.1.2880 pass on Windows and Linux without problems. I'm using Python 2.4.3 in both cases. I also ran the tests through Valgrind 3.1.0 on Linux, and everything seems to be fine.

Maybe you have a stale build? Maybe Python 2.4.1 also has thread issues?

@nilswagner01

@nilswagner01 wrote on 2006-07-21

I am always doing rm -rf build.
I am using Python 2.4, 2.4.1 on Linux (32 and 64 bit systems). The segfault is present on all
3 machines.

BTW is

test_divideerr (numpy.core.tests.test_numeric.test_seterr)Segmentation fault

a new test ?

I didn't get any segfault one week ago.

@alberts

@alberts wrote on 2006-07-21

No, that test has been there for a while.

That test is pretty simple, so see if you can get it to segfault in a seperate script, and start commenting out lines from the test so that we can see specifically which line is causing it to die.

@nilswagner01

@nilswagner01 wrote on 2006-07-21

from numpy.core import *

from numpy.random import rand, randint

from numpy.testing import *

from numpy.core.multiarray import dot as dot_

class test_seterr(NumpyTestCase):

def test_divideerr(self):

seterr(divide='raise')

    try:

        array([1.]) / array([0.])

    except FloatingPointError:

        pass

    else:

        self.fail()

    seterr(divide='ignore')

    array([1.]) / array([0.])

if name == 'main':

NumpyTest().run()

The problem is seterr(divide='raise')
If that line is active I get a segfault.

@rkern

@rkern wrote on 2006-07-21

I also get a bus error trying to execute that test on PPC OS X with Python 2.4.1 and r2863. I do not get the error on AMD64 Ubuntu, Python 2.4.3. I'll try with the latest SVN in a bit.

And Nils, please read WikiFormatting to properly format code and other such text. Your bug reports are difficult to read.

@rkern

Milestone changed to 1.0 Release by @rkern on 2006-07-21

@alberts

@alberts wrote on 2006-07-21

A common element between all the crashers seems to be Python < 2.4.3.

@teoliphant

@teoliphant wrote on 2006-07-24

This crash also happens with Python 2.3. Actually Python 2.4.2 works for me.

It is definitely a threading issue. If you remove the threads
#define NPY_ALLOW_THREADS 0

This is done for Python 2.3 but perhaps the check needs to be extended to anything less than Python 2.4.2

trac user AndrewJ wrote on 2006-07-24

I can confirm that the problem goes away on upgrading to 2.4.3 on OSX.

Andrew

@teoliphant

@teoliphant wrote on 2006-07-26

I've added a check so that threading is disabled on anything but 2.4.2 and above.

Navigation Menu

Repository navigation

segmentation fault (Trac #197) #795

seterr(divide='raise')

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

segmentation fault (Trac #197) #795

Description

Activity

numpy-gitbot commented on Oct 19, 2012

numpy-gitbot commented on Oct 19, 2012

numpy-gitbot commented on Oct 19, 2012

numpy-gitbot commented on Oct 19, 2012

numpy-gitbot commented on Oct 19, 2012

numpy-gitbot commented on Oct 19, 2012

seterr(divide='raise')

numpy-gitbot commented on Oct 19, 2012

numpy-gitbot commented on Oct 19, 2012

numpy-gitbot commented on Oct 19, 2012

numpy-gitbot commented on Oct 19, 2012

numpy-gitbot commented on Oct 19, 2012

numpy-gitbot commented on Oct 19, 2012

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions