GNOME Bugzilla – Bug 625725
memory problem in gst_element_factory_get_static_pad_templates
Last modified: 2010-10-08 16:53:55 UTC
the truth is that i don't know what's the exact problem. but while develop gstreamer-java and try to rebuild on different fedora/redhat release i repeatedly encounter this problem. there was a junit test for ElementFactory. and one specific test for StaticPadTemplates. this test call the native gst_element_factory_get_static_pad_templates and this test during this call used to fail on 64bit... of i know this is not an exact bugreport but i can't tell more just the build logs. search for "ElementFactoryTest" in the logs. eg. on fedora-13 this log show a "No known Enum mapping": http://koji.fedoraproject.org/koji/getfile?taskID=2360036&name=build.log while exactly the same source in another build log crash the java vm: http://koji.fedoraproject.org/koji/getfile?taskID=2369025&name=build.log while on fedora-14 we've got only a -------------------------------- (test:18038): GStreamer-WARNING **: static caps 0x7fa8f8233338 string is NULL -------------------------------- http://koji.fedoraproject.org/koji/getfile?taskID=2360044&name=build.log so i don't know, but these kind of error used to be a memleak or memory overrun or other such kind of memory/pointer problems in the native code. and this bug is not just happened in the latest version but for at least a half year. does anybody can look into the code and try to find the reason? thanks in advance.
one interesting thing it's only happened on 64bit and never on 32 bit!
What exactly makes you think this is a bug in GStreamer and not in the java bindings? Can you provide a small test case in C that reproduces the problem?
it's not possible to crash java vm from java just in case there is a java vm bug. unfortunately we use jna to native access so there is a chance to be a jna bug or we use jna in a wrong way. but many project use jna so i assume this has the smallest chance. in many case when such thing happened with us it was some kind of gstreamer bug. as i wrote i can't write c test for it. if you see our (tibor kocsis and me) previous bug reports, they were all valid. so i hope we've got the respect as you 'believe' us. we've got a 3 line junit test for this elementfactory test and today i try to collect all of those gstreamer functions which called during this test, may be that can help.
This has nothing to do with 'respect' or 'believing' you. A bug is only useful to us if it contains information we can work with. So far you have only provided a link to some build logs and say 'something is wrong in gst_element_factory_get_static_pad_templates()', which seems unlikely seeing that it's a fairly often-used function which only consists of a 'return factory->staticpadtemplates'. Anyway, please provide more information and context. (It would also be nice if you could make a bit more use of newlines and paragraphs, and start new sentences with a capital letter, so that your comments are easier to read.)
i try to create a sample apps. one more warning which i got today: there are two type of warning: GStreamer-WARNING **: static caps 0x7fd39c2882d8 string is NULL GStreamer-CRITICAL **: gst_caps_ref: assertion `GST_CAPS_REFCOUNT_VALUE (caps) > 0' failed
My first guess would be a refcounting bug in the bindings somewhere (unref'ing something that it shouldn't be unrefing).
Closing this bug report as no further information has been provided. Please feel free to reopen this bug if you can provide the information asked for. Thanks!
it seems it was as gstreamer-java bug (at least i hope so)