From: Russell King > Sent: 06 March 2024 09:52 > > On Tue, Mar 05, 2024 at 09:58:46AM -0800, Josh Poimboeuf wrote: > > This is an off-by-one bug which is common in unwinders, due to the fact > > that the address on the stack points to the return address rather than > > the call address. > > > > So, for example, when the last instruction of a function is a function > > call (e.g., to a noreturn function), it can cause the unwinder to > > incorrectly try to unwind from the function *after* the callee. > > I suppose this can only happen in __noreturn functions because that > can be: > > foo: > .. > bl bar > .. end of function and thus next function ... > > which results in LR pointing into the next function. > > Would it make better sense to lookup the LR value winding it back by > one instruction like ORC on x86 does (as you mention) rather than > the patch you proposed which looks rather large and complicated? Is it even possible to always reliably get a stack trace from a no-return function on a cpu that uses a 'lr'? If the function doesn't return then the compiler need not save 'lr' on stack and can still use it as a temporary register. Without a valid 'lr' I think all you can do is search the stack for a likely code address? Am I missing something? David - Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)
