https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89501
--- Comment #12 from Linus Torvalds <torva...@linux-foundation.org> --- (In reply to Jeffrey A. Law from comment #11) > > More generally we have considered whether or not we could eliminate the > control dependent path which leads to undefined behavior. But you have to > be careful because statements on the path prior to the actual undefined > behavior may have observable side effects. Note that for the kernel, we consider those kinds of "optimizations" completely and utterly wrong-headed, and will absolutely refuse to use a compiler that does things like that. It's basically the compiler saying "I don't care what you meant, I can do anything I want, and that means I will screw the code up on purpose". I will personally switch the kernel immediately to clang the moment we cannot turn off idiotic broken behavior like that. We currently already disable 'strict-aliasing', 'strict-overflow' and 'delete-null-pointer-checks'. Because compilers that intentionally create garbage code are evil and wrong. Compiler standards bodies that add even more of them (the C aliasing rules being the prime example) should be ashamed of themselves. And compiler people that think it's ok to say "it's undefined, so we do anything we damn well like" shouldn't be working with compilers IMNSHO. If you know something is undefined, you warn about it. You don't silently generate random code that doesn't match the source code just because some paper standard says you "can".
