Invalid nullptr diagnostic

[tlemo]

Bugzilla Link	48534
Version	unspecified
OS	All
CC	@devincoughlin,@haoNoQ

Extended Description

// https://godbolt.org/z/1oze6v

struct IrBuilder;
struct Kernel;

class Passkey {
friend struct IrBuilder;
explicit Passkey(Kernel* kernel) : kernel(kernel) {}

public:
Kernel* const kernel = nullptr;
};

struct Kernel {
void foo() const;
};

struct Node {
// incorrect note: Called C++ object pointer is null
explicit Node(Passkey passkey) { passkey.kernel->foo(); }
};

// enabling the following piece of code (normally in a different translation unit)
// demonstrates that the diagnostic is incorrect
#if 0
struct IrBuilder {
Node* create() {
const Passkey passkey(kernel_);
return new Node(passkey);
}

Kernel* kernel_ = nullptr;

};

Node* test(IrBuilder* ir_builder) {
return ir_builder->create();
}
#endif

:19:38: warning: Called C++ object pointer is null [clang-analyzer-core.CallAndMessage] explicit Node(Passkey passkey) { passkey.kernel->foo(); } ^ :19:38: note: Called C++ object pointer is null 1 warning generated.

[tlemo]

assigned to @haoNoQ

[llvmbot]

Moving to Static-analyzer bugs

[haoNoQ]

Aha, uhm, yeah, i see. The static analyzer indeed thinks that a combination of "const" and a field initializer causes the field to forever stay that way. We'll need to undo this relatively recently added shortcut.

[haoNoQ]

(probably caused by code in RegionStoreManager::getBindingForField())