On Sun, Jun 09, 2013 at 03:29:43PM +0200, Dominique Pellé wrote: > Hi > > These valgrind error are annoying but they are not a bug. > They should not cause a crash. See this document that > describes why: > > http://svn.python.org/view/python/trunk/Misc/README.valgrind?view=markup > --without-pymalloc > > That's not a new issue either. I have seen those valgrind > errors with Python in Vim for several years. Of course, they > make it harder to spot other real errors.
Hello Dominique, Thanks for your response. I didn't know that Python manages memory this way. (I have to admit I don't really understand why they think using uninitialized is a good idea.) But that only explains the "uninitialized value" warnings, not the use-after-frees. But looking at Python's suppression file those are ignored there as well. I've no idea what they do to make valgrind think they're using free()d memory when in fact they're using it correctly. The documentation from README.valgrind is not enough to explain that. > If you want to get rid of those valgrind errors, you have to > configure & rebuild the Python lib with --without-pymalloc > as described at the above link. I tried that years ago > and it did cleaned up the valgrind errors. I have not retried > lately. I wonder why valgrind default ignored errors does not > include them by the way. For now I use Python's suppression file. Seems to work fine. No valgrind errors left (with the patch). Regards Simon -- + privacy is necessary + using gnupg http://gnupg.org + public key id: 0x92FEFDB7E44C32F9
signature.asc
Description: Digital signature