Feature #11788
Updated by Evgeny Novikov over 2 years ago
Lack of these models results in about 6% of all false alarms when checking memory safety.
BTW, I have already made the preliminary version of models and tests for the _kref*_ API in branch _kref-model_. It is necessary to brush up them. It is worth noting that my work was based on contribution from Anton Vasiliev and Vasiliy Naumushkin. It turned out that too many things should be changed due to various reasons.