================
@@ -1165,7 +1165,7 @@ ProgramStateRef
StdLibraryFunctionsChecker::NotNullConstraint::apply(
return State->assume(L, CannotBeNull);
}
-void StdLibraryFunctionsChecker::NotNullConstraint::describe(
+void StdLibraryFunctionsChecker::NullnessConstraint::describe(
DescriptionKind DK, const CallEvent &Call, ProgramStateRef State,
const Summary &Summary, llvm::raw_ostream &Out) const {
assert(CannotBeNull &&
----------------
balazske wrote:
Currently the `NotNullConstraint` is used in two ways:
- Argument preconditions where only a non-null condition is used. When
`describe` is called here the `CannotBeNull` should be true. The
`describeArgumentValue` is called on the negated version of it when
`CannotBeNull` is false.
- Return value conditions at "cases" where it can be null or non-null. Here
always a custom description text is used and `describe` or
`describeArgumentValue` is not called.
Purpose of the assertion seems to be only to make a simplification (not
implement `describe` in the null and `describeArgumentValue` in the non-null
case). The assertion could be extended with something like `getArgNo() != Ret
&&` (verify the first use case above).
https://github.com/llvm/llvm-project/pull/131374
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
