Klever

Too complicated rule specification producing too many false alarms while requiring too much resources. We will not fix everything required for it in the near future.

BTW, there is a model for dev_set_drvdata() in linux/drivers/base/dd.* but it isn't compatible with old API, since there dev_set_drvdata() returns nothing while the model assumes it returns int.

It is an important model for many rules.

Possibly, the problem is in changes between versions in interface and in implementation:
3.0-3.15 int dev_set_drvdata(struct device *dev, void *data);
before 3.0 and after 3.16: void dev_set_drvdata(struct device *dev, void *data);

while our pointcut is:
around: call(int dev_set_drvdata(struct device *dev, void *data))

Indeed we would be able to detect these issues easily if #7424 will be implemented.

Since the suggested fix is too broad and can affect everybody it should be tested separately from the main workflow. Taking into account that there is the only false alarm due to this issue and even it is for the rule specification to be marked incomplete, I am unsure that somebody will be able to do that soon.

We have various issues with that model from its absence for old versions of the Linux kernel to its imprecision for the newer ones. Actually it is extremely important interface to be modeled very accurately for many versions of the Linux kernel.

Let's fix and improve specifications after we will have good tests and a testing infrastructure (version:0.3) and likely after we will complete a considerable refactoring of Core (1.0).