Right, the bug in the kernel code snippet is like GP describes, but the gcc bug isn't that the compiled code is incorrect, it's that people expect an "uninitial

Right, the bug in the kernel code snippet is like GP describes, but the gcc bug isn't that the compiled code is incorrect, it's that people expect an "uninitialized variable" warning and the compiler doesn't produce one.