================
@@ -0,0 +1,48 @@
+// RUN: %clang_analyze_cc1 -triple x86_64-unknown-linux-gnu -std=c++17 \
+// RUN: -analyzer-checker=core,debug.ExprInspection \
+// RUN: -analyzer-config eagerly-assume=false -verify %s
+
+void clang_analyzer_dump_float(float);
+void clang_analyzer_dump_double(double);
+void clang_analyzer_eval(int);
+
+template <typename T> T twice(T x) { return x + x; }
+
+struct Vec {
+ float v;
+ Vec operator+(Vec o) const { return Vec{v + o.v}; }
+};
+
+constexpr double half(double x) { return x / 2.0; }
+
+double defaultArg(double i = 42) { return i; }
+float narrowingDefaultArg(float i = 1.5) { return i; }
+
+void testTemplate() {
+ clang_analyzer_dump_float(twice(1.5f)); // expected-warning{{3 IEEEsingle}}
+ clang_analyzer_dump_double(twice(2.5)); // expected-warning{{5 IEEEdouble}}
+}
+
+void testOverloadedOperator() {
+ Vec a{1.5f}, b{2.5f};
+ clang_analyzer_dump_float((a + b).v); // expected-warning{{4 IEEEsingle}}
+}
+
+void testConstexpr() {
+ clang_analyzer_dump_double(half(3.0)); // expected-warning{{1.5 IEEEdouble}}
+ constexpr double c = 1.25;
+ clang_analyzer_dump_double(c); // expected-warning{{1.25
IEEEdouble}}
+}
+
+// A default argument is not evaluated through the CFG, rather it is folded by
+// getConstantVal.
+void testDefaultArgument() {
+ clang_analyzer_dump_double(defaultArg()); // expected-warning{{42
IEEEdouble}}
+ clang_analyzer_dump_float(narrowingDefaultArg()); // expected-warning{{1.5
IEEEsingle}}
+}
+
+// A negative value does not fit an unsigned integer; the conversion should be
+// unmodeled.
+void testUnsignedFromNegative() {
+ clang_analyzer_eval((unsigned)-1.5f == 0); // expected-warning{{UNKNOWN}}
+}
----------------
jpjepko wrote:
Added positive and negative test cases for this in commit c48451a.
https://github.com/llvm/llvm-project/pull/214098
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
