Tests: Bool widening and narrowing

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

Author: r41k0u

tests/passing_tests/signedness/bool_int.py

@@ -0,0 +1,24 @@
+# A bool is a 1-bit integer to LLVM, and C's rules for it are not the integer
+# ones: it widens to 0 or 1 (never sign-extends to -1), and an integer narrows
+# to it by comparing with zero (2 is true), not by keeping the low bit.
+from pythonbpf import bpf, section, bpfglobal, compile
+from ctypes import c_void_p, c_int64
+
+
+@bpf
+@section("tracepoint/raw_syscalls/sys_enter")
+def prog(ctx: c_void_p) -> c_int64:
+    t = True
+    n = True
+    n = 2  # bool = 2 is true in C; truncation would make it false
+    s = t + 1  # int + bool promotes the bool to int 1: 2, not 0
+    return c_int64(t + n + s)  # 1 + 1 + 2 = 4, not -1 + 0 + 0
+
+
+@bpf
+@bpfglobal
+def LICENSE() -> str:
+    return "GPL"
+
+
+compile()

tests/test_signedness_ir.py

@@ -50,6 +50,12 @@
         [r"\budiv i64 .*, 4294967294"],
         [r"\bsdiv i64"],
     ),
+    # A bool widens with zext and an integer narrows to it by != 0, never by
+    # trunc; sext of an i1 would return -1 for True.
+    "signedness/bool_int.py": (
+        [r"zext i1 .* to i(32|64)", r"icmp ne i64 .*, 0"],
+        [r"sext i1 ", r"trunc i64 .* to i1"],
+    ),
     # An enum constant is a C `int`, so `XDP_PASS - k` with k a c_uint32 is a
     # u32 operation: the result is cut to 32 bits and zero-extended. Ranked
     # as i64 it would be a signed 64-bit subtraction with no trunc at all.