This issue tracker has been migrated to GitHub, and is currently read-only.
For more information, see the GitHub FAQs in the Python's Developer Guide.

classification
Title: Valgrind suppressions
Type: Stage: resolved
Components: None Versions: Python 2.7
process
Status: closed Resolution: fixed
Dependencies: Superseder:
Assigned To: Nosy List: Paul.Price, Zhiping.Deng, mgedmin, neologix, python-dev, stutzbach, vstinner
Priority: normal Keywords: patch

Created on 2011-07-22 16:36 by Paul.Price, last changed 2022-04-11 14:57 by admin. This issue is now closed.

Files
File name Uploaded Description Edit
proposed.patch Paul.Price, 2011-12-01 18:05 review
Messages (9)
msg140895 - (view) Author: Paul Price (Paul.Price) Date: 2011-07-22 16:36
I am running Python 2.7.1 under Valgrind 3.6.1 on a x86_64 GNU/Linux box.  Valgrind is flagging errors that are not covered by the standard suppressions file (Misc/valgrind-python.supp; after uncommenting the part I'm supposed to), e.g.:

==5804== Use of uninitialised value of size 8
==5804==    at 0x4EBA956: PyObject_Free (obmalloc.c:958)
==5804==    by 0x4E85D65: code_dealloc (codeobject.c:317)
==5804==    by 0x4ECE23C: tupledealloc (tupleobject.c:220)
==5804==    by 0x4E85E42: code_dealloc (codeobject.c:308)
==5804==    by 0x4F3315D: load_source_module (import.c:1022)
==5804==    by 0x4F33F7E: import_submodule (import.c:2596)
==5804==    by 0x4F3420E: load_next (import.c:2416)
==5804==    by 0x4F3484F: import_module_level (import.c:2137)
==5804==    by 0x4F34DA3: PyImport_ImportModuleLevel (import.c:2189)
==5804==    by 0x4F16ABE: builtin___import__ (bltinmodule.c:49)
==5804==    by 0x4E6DA42: PyObject_Call (abstract.c:2529)
==5804==    by 0x4F16F82: PyEval_CallObjectWithKeywords (ceval.c:3881)

==5804== Use of uninitialised value of size 8
==5804==    at 0x4EBA956: PyObject_Free (obmalloc.c:958)
==5804==    by 0x4E85D65: code_dealloc (codeobject.c:317)
==5804==    by 0x4F3315D: load_source_module (import.c:1022)
==5804==    by 0x4F33F7E: import_submodule (import.c:2596)
==5804==    by 0x4F3420E: load_next (import.c:2416)
==5804==    by 0x4F3484F: import_module_level (import.c:2137)
==5804==    by 0x4F34DA3: PyImport_ImportModuleLevel (import.c:2189)
==5804==    by 0x4F16ABE: builtin___import__ (bltinmodule.c:49)
==5804==    by 0x4E6DA42: PyObject_Call (abstract.c:2529)
==5804==    by 0x4F16F82: PyEval_CallObjectWithKeywords (ceval.c:3881)
==5804==    by 0x4F1ACD3: PyEval_EvalFrameEx (ceval.c:2332)
==5804==    by 0x4F1F347: PyEval_EvalCodeEx (ceval.c:3252)

etc.

Perhaps the suppressions file covers only 32-bit machines, because if I add:

{
   ADDRESS_IN_RANGE/Use of uninitialised value of size 8
   Memcheck:Addr8
   fun:PyObject_Free
}

{
   ADDRESS_IN_RANGE/Use of uninitialised value of size 8
   Memcheck:Value8
   fun:PyObject_Free
}

{
   ADDRESS_IN_RANGE/Use of uninitialised value of size 8
   Memcheck:Addr8
   fun:PyObject_Realloc
}

{
   ADDRESS_IN_RANGE/Use of uninitialised value of size 8
   Memcheck:Value8
   fun:PyObject_Realloc
}

(i.e., "Value8" instead of "Value4") then all is clear.
msg148715 - (view) Author: Zhiping Deng (Zhiping.Deng) Date: 2011-12-01 10:16
It works for me!
msg148730 - (view) Author: Charles-François Natali (neologix) * (Python committer) Date: 2011-12-01 17:50
Could you please provide a diff ?
Also, they should probably be commented by default (as other obmalloc-related calls).
msg148733 - (view) Author: Paul Price (Paul.Price) Date: 2011-12-01 18:05
Here's the diff with the added sections commented out.
msg148800 - (view) Author: Roundup Robot (python-dev) (Python triager) Date: 2011-12-03 13:44
New changeset 3eb73f45a614 by Charles-François Natali in branch 'default':
Issue #12612: Add some Valgrind suppressions for 64-bit machines. Patch by Paul
http://hg.python.org/cpython/rev/3eb73f45a614
msg148801 - (view) Author: Charles-François Natali (neologix) * (Python committer) Date: 2011-12-03 13:45
Committed, thanks for the patch!
msg148809 - (view) Author: Paul Price (Paul.Price) Date: 2011-12-03 15:19
Welcome.  Thank you!
msg254090 - (view) Author: Marius Gedminas (mgedmin) * Date: 2015-11-05 07:23
Could this fix be backported to the 2.7 branch as well?
msg254135 - (view) Author: Roundup Robot (python-dev) (Python triager) Date: 2015-11-05 19:42
New changeset 92dda5f00b0f by doko in branch '2.7':
Issue #12612: Add some Valgrind suppressions for 64-bit machines.
https://hg.python.org/cpython/rev/92dda5f00b0f
History
Date User Action Args
2022-04-11 14:57:20adminsetgithub: 56821
2015-11-05 19:42:18python-devsetmessages: + msg254135
2015-11-05 07:23:16mgedminsetnosy: + mgedmin
messages: + msg254090
2011-12-03 15:19:17Paul.Pricesetmessages: + msg148809
2011-12-03 13:45:55neologixsetstatus: open -> closed
resolution: fixed
messages: + msg148801

stage: resolved
2011-12-03 13:44:40python-devsetnosy: + python-dev
messages: + msg148800
2011-12-01 18:05:17Paul.Pricesetfiles: + proposed.patch
keywords: + patch
messages: + msg148733
2011-12-01 17:50:18neologixsetnosy: + neologix
messages: + msg148730
2011-12-01 10:16:23Zhiping.Dengsetnosy: + Zhiping.Deng
messages: + msg148715
2011-08-01 23:20:50stutzbachsetnosy: + stutzbach
2011-07-22 16:42:42vstinnersetnosy: + vstinner
2011-07-22 16:36:11Paul.Pricecreate