NoQ added inline comments.
================
Comment at: clang/lib/StaticAnalyzer/Checkers/SmartPtrModeling.cpp:130
+// Returns empty type if not found valid inner pointer type.
+static QualType getInnerPointerType(const CallEvent &Call, CheckerContext &C) {
+ QualType InnerType{};
----------------
vrnithinkumar wrote:
> It seems like a long shot to me.
> I am not sure is there any direct or easy way to get inner pointer type from
> a smart pointer
That's about right. You're doing exactly what you're asked: grab the template
parameter of the class. The problem is indeed that complicated!
================
Comment at: clang/lib/StaticAnalyzer/Checkers/SmartPtrModeling.cpp:145
+ TemplateArgs.size() > 0 &&
+ "Smart pointer should have specialized with atleast one template
type");
+ auto InnerValueType = TemplateArgs[0].getAsType();
----------------
That's pretty fundamental, right? If it's a specialization, it must have
something specialized. It isn't specific to unique pointers, right?
Because unique pointers aren't special; technically anybody can define an
arbitrary class with name `std::unique_ptr` and any properties they'd like.
It's going to be undefined behavior according to the standard (because
namespace `std` is explicitly reserved for the standard library) but if the
compiler *crashes* it'll still be our fault.
================
Comment at: clang/lib/StaticAnalyzer/Checkers/SmartPtrModeling.cpp:448
+
+ if (NullState) {
+ auto NullVal = C.getSValBuilder().makeNull();
----------------
There's no need to check. You just conjured a brand new symbol out of thin air;
you can be sure that it's completely unconstrained and both states are
feasible. You can instead `assert()` that they're both feasible.
================
Comment at: clang/lib/StaticAnalyzer/Checkers/SmartPtrModeling.cpp:458-460
+ OS << "Assuming smart pointer ";
+ ThisRegion->printPretty(OS);
+ OS << " is null";
----------------
Wait, what happens when the region can't be pretty-printed? Does it print two
spaces between "pointer" and "is"?
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D86027/new/
https://reviews.llvm.org/D86027
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
