I'm surprised this didn't cause any tests to fail. I suppose we need specifically a blocking syscall with a deadline with a condition that becomes true, but is falsified before the event runs...

I'm surprised this didn't cause any tests to fail. I suppose we need specifically a blocking syscall with a deadline with a condition that becomes true, but is falsified before the event runs...

Yeah I think it depends on if the syscall handler believes the value of _syscallcondition_satisfied() (if it believes that the syscall condition trigger wasn't spurious), or if it does its own check to see if it should re-block and create a new syscall condition. I'm not sure what happens though if there become multiple syscall conditions with their own timers.

The tor benchmark networking results change slightly, but not significantly. The runtime and memory results do not change.

https://github.com/shadow/benchmark-results/tree/master/tor/2023-02-22-T15-06-28