When initialize_password_db() is called with reload=True, it's assumed that
the free_private_data member of pdb_methods is non-null. This is not
necessarily the case, as the tdb backend has no private data and therefore
no free function. Check to see if we have private data that needs freed
(this text originally written by Steve Langasek)
Created attachment 7533 [details]
Patch used in Debian and Ubuntu
I would use a different patch. In the if clause I would check for free_private_data instead of private_data because that is what we dereference. Would that work for you as well?
The two checks seem to be equivalent, yes. It fits my mental model better to think about whether there's any data to be freed, rather than whether there's a method to do the freeing, but I think both ways will fix the bug. :)
Created attachment 7539 [details]
that's what I am about to push to master, ok?
Seems OK for me and I guess STeve will be OK too.
Could that be a candidate for 3.6.6 and therefore save us one patch in debian/patches?
Comment on attachment 7539 [details]
Pushed patch to v3-6-test.
Will be included in the next bugfix release.
Closing out bug report.