Feature #11771
closedFix model for struct_size()
0%
Description
Linux 5.18 changed the definition of macro struct_size(), so that the model for internal __ab_c_size() stopped working (#11624). Moreover, now there is a bunch of false alarms where another similar API, array3_size() hits. The latter changed from the function to the macro in Linux 5.18. It would be better to bind models with the top-level API rather than with underlying helper functions that do not seem to be stable.
Updated by Evgeny Novikov over 2 years ago
- Status changed from New to Resolved
I did all the best in branch overflow-models. Preliminary testing demonstrated that all related false alarms either become safes/unknowns or there are unsafes that demonstrate clearly that some extra models for other APIs should be developed to avoid excessive nondeterminism. Let's see on complete test results.
Updated by Evgeny Novikov over 2 years ago
- Status changed from Resolved to Closed
As expected complete testing demonstrated awesome results, so I merged the branch to master in db65d7feb.