* Thomas Neumann: > Hi, > >>>>> When dynamically linking a fast enough machine hides the latency, but when >>>>> Statically linking or on slower devices this change caused a 5x increase >>>>> in >>>>> Instruction count and 2x increase in cycle count before getting to main. > > I have looked at ways to fix that. The problem is that with static > linking unwinding tables are registered dynamically, and with my patch > that registration triggers an eager sort of fde lists. While > previously the lists were sorted when the first exception was > thrown. If an application throws at least one exception there is no > downside in eager sorting, but if the application never throws there > is overhead.
Would it be possible to trigger lazy registration if the version is read as a zero? This would not introduce any additional atomic instructions on the fast path. Thanks, Florian
