[clang] nonblocking/nonallocating attributes: 2nd pass caller/callee analysis (PR #99656)

[Doug Wyatt via cfe-commits]

================
@@ -0,0 +1,256 @@
+// RUN: %clang_cc1 -fsyntax-only -fblocks -fcxx-exceptions -std=c++20 -verify %s
+// These are in a separate file because errors (e.g. incompatible attributes) currently prevent
+// the FXAnalysis pass from running at all.
+
+// This diagnostic is re-enabled and exercised in isolation later in this file.
+#pragma clang diagnostic ignored "-Wperf-constraint-implies-noexcept"
+
+// --- CONSTRAINTS ---
+
+void nb1() [[clang::nonblocking]]
+{
+	int *pInt = new int; // expected-warning {{'nonblocking' function must not allocate or deallocate memory}}
+	delete pInt; // expected-warning {{'nonblocking' function must not allocate or deallocate memory}}
+}
+
+void nb2() [[clang::nonblocking]]
+{
+	static int global; // expected-warning {{'nonblocking' function must not have static locals}}
+}
+
+void nb3() [[clang::nonblocking]]
+{
+	try {
+		throw 42; // expected-warning {{'nonblocking' function must not throw or catch exceptions}}
+	}
+	catch (...) { // expected-warning {{'nonblocking' function must not throw or catch exceptions}}
+	}
+}
+
----------------
dougsonos wrote:

This turned out to be a bit of a slippery slope. It is relatively straightforward to describe both a call site and a callee as a function / constructor / destructor / member initializer / lambda / block, but it is painful to repeat that `%select{}` construct across a number of diagnostics, and I don't see ways to reduce the number of diagnostics.

I think I'm inclined to keep the change to make the member initializer / function distinction, since that does seem to benefit from the extra "in constructor here" note, but leave aside the specificity about the kind of function-like object.

https://github.com/llvm/llvm-project/pull/99656