[clang] [analyzer] Refactor CallDescription match mode (NFC) (PR #83432)

[via cfe-commits]

=?utf-8?q?Donát?= Nagy <donat.nagy at ericsson.com>
Message-ID:
In-Reply-To: <llvm.org/llvm/llvm-project/pull/83432 at github.com>


================
@@ -27,20 +27,48 @@ class IdentifierInfo;
 
 namespace clang {
 namespace ento {
-
-enum CallDescriptionFlags : unsigned {
-  CDF_None = 0,
-
-  /// Describes a C standard function that is sometimes implemented as a macro
-  /// that expands to a compiler builtin with some __builtin prefix.
-  /// The builtin may as well have a few extra arguments on top of the requested
-  /// number of arguments.
-  CDF_MaybeBuiltin = 1 << 0,
-};
-
-/// This class represents a description of a function call using the number of
-/// arguments and the name of the function.
+/// A `CallDescription` is a pattern that can be used to _match_ calls
+/// based on the qualified name and the argument/parameter counts.
 class CallDescription {
+public:
+  enum class Mode {
+    /// Match calls to functions from the C standard library. On some platforms
+    /// some functions may be implemented as macros that expand to calls to
+    /// built-in variants of the given functions, so in this mode we use some
+    /// heuristics to recognize these implementation-defined variants:
+    ///  - We also accept calls where the name is derived from the specified
+    ///    name by adding "__builtin" or similar prefixes/suffixes.
+    ///  - We also accept calls where the number of arguments or parameters is
+    ///    greater than the specified value.
+    /// For the exact heuristics, see CheckerContext::isCLibraryFunction().
+    /// Note that functions whose declaration context is not a TU (e.g.
+    /// methods, functions in namespaces) are not accepted as C library
+    /// functions.
+    /// FIXME: If I understand it correctly, this discards calls where C++ code
+    /// refers a C library function through the namespace `std::` via headers
+    /// like <cstdlib>.
+    CLibrary,
----------------
NagyDonat wrote:

> many functions that are "C library" calls are not matched with this mode

I'm planning to ensure that:
- this mode works "out of the box" for matching the C standard library functions and
- the checkers that are looking for a C standard library function do use it.
This is an NFC commit so I didn't change the behavior of this mode yet, but I'm planning to do so in a followup commit. I picked the name `CLibrary` because it reflects the name of the method `isCLibraryFunction()` which is used to implement this mode. I'd say that if `isCLibraryFunction()` returns false for a C library function, then it's buggy and I'll fix its behavior.

Do you see any difficulties that would hinder or block this path?

By the way I dislike the name `MaybeBuiltin` because at first I thought that that it clearly meant "this _may be_ a builtin function (with a modified name), but it _may be_ anything else as well", while in fact it always calls `isCLibraryFunction` (and doesn't accept anything that isn't a C library function).

https://github.com/llvm/llvm-project/pull/83432