[clang] [clang][Sema] Add noinline check for __builtin_frame_address and __builtin_return_address (PR #82966)

[Timothy Herchen via cfe-commits]

anematode wrote:

> Although I seem to remember having seen code that uses `always_inline` in order to force `__builtin_return_address` to actually apply to its caller.

Right; that's why I thought just disabling inlining would be a suboptimal choice, although I guess you could have `always_inline` override it.

If the diagnostic is so noisy then maybe I don't see the point of it. Is there any case in Linux where it's a correctness issue to use a caller's address? (Presumably in this case it would just give a confusing call site when you're debugging.)

https://github.com/llvm/llvm-project/pull/82966