Bug 910962

Summary:	python3 /usr/lib/python3.3/site-packages/scipy/linalg/tests/test_decomp.py crashes
Product:	[Fedora] Fedora	Reporter:	Orion Poplawski <orion>
Component:	scipy	Assignee:	Orion Poplawski <orion>
Status:	CLOSED ERRATA	QA Contact:	Fedora Extras Quality Assurance <extras-qa>
Severity:	high	Docs Contact:
Priority:	unspecified
Version:	18	CC:	gwync, jspaleta, orion, tomspur, ttomecek
Target Milestone:	---
Target Release:	---
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2013-02-27 02:39:51 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Orion Poplawski 2013-02-14 03:06:56 UTC

Description of problem:

The python3-scipy tests are crashing here:

python3 /usr/lib/python3.3/site-packages/scipy/linalg/tests/test_decomp.py

generally with a segfault in some malloc related routine that points to memory corruption before that point.


Version-Release number of selected component (if applicable):
python3-scipy-0.11.0-3.fc18.i686


First valgrind error:

valgrind python3 /usr/lib/python3.3/site-packages/scipy/linalg/tests/test_decomp.py
==17599== Memcheck, a memory error detector
==17599== Copyright (C) 2002-2012, and GNU GPL'd, by Julian Seward et al.
==17599== Using Valgrind-3.8.1 and LibVEX; rerun with -h for copyright info
==17599== Command: python3 /usr/lib/python3.3/site-packages/scipy/linalg/tests/test_decomp.py
==17599== 
..............==17599== Invalid write of size 8
==17599==    at 0x4F0E4913: dlacpy_ (in /usr/lib/atlas/liblapack.so.3.0)
==17599==    by 0x4F15FD06: dsbevx_ (in /usr/lib/atlas/liblapack.so.3.0)
==17599==    by 0x77D064E: f2py_rout_flapack_dsbevx (flapackmodule.c:25876)
==17599==    by 0x77D8789: fortran_call (fortranobject.c:346)
==17599==    by 0x4067C10: PyObject_Call (in /usr/lib/libpython3.3m.so.1.0)
==17599==    by 0x411BC7C: PyEval_EvalFrameEx (in /usr/lib/libpython3.3m.so.1.0)
==17599==    by 0x411E9F4: PyEval_EvalFrameEx (in /usr/lib/libpython3.3m.so.1.0)
==17599==    by 0x4120A23: PyEval_EvalCodeEx (in /usr/lib/libpython3.3m.so.1.0)
==17599==    by 0x411E6B1: PyEval_EvalFrameEx (in /usr/lib/libpython3.3m.so.1.0)
==17599==    by 0x4120A23: PyEval_EvalCodeEx (in /usr/lib/libpython3.3m.so.1.0)
==17599==    by 0x4091442: ??? (in /usr/lib/libpython3.3m.so.1.0)
==17599==    by 0x4067C10: PyObject_Call (in /usr/lib/libpython3.3m.so.1.0)
==17599==  Address 0x68afda8 is 0 bytes after a block of size 80 alloc'd
==17599==    at 0x4008F6F: malloc (vg_replace_malloc.c:270)
==17599==    by 0x70CFF7A: ??? (in /usr/lib/python3.3/site-packages/numpy/core/multiarray.cpython-33m.so)
==17599==    by 0x70F8B1D: ??? (in /usr/lib/python3.3/site-packages/numpy/core/multiarray.cpython-33m.so)
==17599==    by 0x77D8D08: array_from_pyobj (fortranobject.c:630)
==17599==    by 0x77D04A8: f2py_rout_flapack_dsbevx (flapackmodule.c:25854)
==17599==    by 0x77D8789: fortran_call (fortranobject.c:346)
==17599==    by 0x4067C10: PyObject_Call (in /usr/lib/libpython3.3m.so.1.0)
==17599==    by 0x411BC7C: PyEval_EvalFrameEx (in /usr/lib/libpython3.3m.so.1.0)
==17599==    by 0x411E9F4: PyEval_EvalFrameEx (in /usr/lib/libpython3.3m.so.1.0)
==17599==    by 0x4120A23: PyEval_EvalCodeEx (in /usr/lib/libpython3.3m.so.1.0)
==17599==    by 0x411E6B1: PyEval_EvalFrameEx (in /usr/lib/libpython3.3m.so.1.0)
==17599==    by 0x4120A23: PyEval_EvalCodeEx (in /usr/lib/libpython3.3m.so.1.0)

points to interaction with atlas.

Comment 1 Orion Poplawski 2013-02-14 04:12:23 UTC

Commenting out test_dsbevx in that file gets us farther.   Now valgrind complains about:

==20207== Invalid write of size 8
==20207==    at 0x4F39BBD5: zlacpy_ (in /usr/lib/atlas/liblapack.so.3.0)
==20207==    by 0x4F367812: zhbevx_ (in /usr/lib/atlas/liblapack.so.3.0)
==20207==    by 0x77CD0ED: f2py_rout_flapack_zhbevx (flapackmodule.c:27055)

Hmm, really doesn't like *bevx.  Looks like these tests have some questions about the arguments?

        ## Achtung: Argumente 0.0,0.0,range?
        w, evec, num, ifail, info = zhbevx(self.bandmat_herm, 0.0, 0.0, 1, N,
                                       compute_v=1, range=2)

Comment 2 Orion Poplawski 2013-02-14 04:30:41 UTC

This looks to be the fix: https://github.com/scipy/scipy/pull/404/commits

Building now...

Comment 3 Fedora Update System 2013-02-15 17:18:40 UTC

scipy-0.11.0-4.fc18 has been submitted as an update for Fedora 18.
https://admin.fedoraproject.org/updates/FEDORA-2013-2421/scipy-0.11.0-4.fc18

Comment 4 Fedora Update System 2013-02-17 03:34:17 UTC

Package scipy-0.11.0-4.fc18:
* should fix your issue,
* was pushed to the Fedora 18 testing repository,
* should be available at your local mirror within two days.
Update it with:
# su -c 'yum update --enablerepo=updates-testing scipy-0.11.0-4.fc18'
as soon as you are able to.
Please go to the following url:
https://admin.fedoraproject.org/updates/FEDORA-2013-2421/scipy-0.11.0-4.fc18
then log in and leave karma (feedback).

Comment 5 Fedora Update System 2013-02-27 02:39:57 UTC

scipy-0.11.0-4.fc18 has been pushed to the Fedora 18 stable repository.  If problems still persist, please make note of it in this bug report.