[DaveFlater]

AFAIK, the optimization undermined the only use case that -fno-omit-frame-pointer actually had on x86_64. Is there a real use case that benefitted from allowing the frame pointer push to wander? Why why why

[sitkack]

That is too bad, it does mention `-fno-schedule-insns2` in the comments. I'll have to see how well that works in practice.

Having high quality low cost stack straces is important for continuous profiling. ref Knuth.