> The source is the external timers, that depend on "outer world", e.g., slirp. > These timers are not recorded/replayed, but may affect the icount warping > (which should remain deterministic).
I thought we agreed earlier in this discussion that external timers are deterministic by design, because they are subset of virtual clock timers. Regarding slirp module, although it interfaces "outer world" (except when 'restrict' option is on), it still uses IPv6 timer deterministically. You may check it yourself. Otherwise it shouldn't use timer of 'virtual clock' type for this purpose and this needs to be fixed. So I still don't understand. And I'm not asking you to explain. I'm just hinting you that your fix may finally happen to be incorrect in case if it based on wrong assertion I'm talking about. > Please check the new version of the series. Sorry, I'm too busy right now for deep analysis. I've just tried to briefly understand issue you pointed and provide minimal feedback in order to prevent possible wrong decisions. I'll return to this work later.
