Commit 96b5fce introduced regressions, leading to potential misadressing
and a very probable memory leak. This commit fixes the issues and
hardens the module to better detect some error cases in the
future.
It also adds valgrind-based testbench tests which ensure that no
pointer errors exist. If they had been in place, the regressions
would never have been undetected.
Note that commit 96b5fce was introduced in 8.23.0 on Nov, 15th 2016.
Many thanks to Yaroslav Bo for alerting me on the root problem and
providing a very good analysis and guidance.
see also https://github.com/rsyslog/rsyslog/issues/1906
closes https://github.com/rsyslog/rsyslog/issues/1962
see also https://github.com/rsyslog/rsyslog/issues/1964
1) usually, no cancellation happens
2) even if so, there is no cancellation point inside the
destructors, so disabeling cancellation was mood in the first
place...
... plus a fix for a long-time bug in obj-types.h. That lead to
the object pointer only then to become NULL when the object was
actually destructed, I discovered this issue during
introduction of the pRcvFrom property in msg_t, but it potentially had other
effects, too. I am not sure if some experienced instability resulted from this
bug OR if its fix will cause harm to so-far "correctly" running code. The later
may very well be. Thus I will change it only for the current branch and also
the beta, but not in all old builds. Let's see how things evolve.
... this was long overdue, and I finlly tackeld it. It turned out to
be more complex than I initially thought. The next step now probably is
to actually implement multiple rule sets and the beauty that comes
with them.
... at least in some cases ;) I assume there are still a couple
of bugs inside the code. But at least we have something from
where we can continue to work on.
there are still some files left which could go into the
runtime, but I think we will delete most of them once we
are done with the full modularization.
