[erlang-questions] : Subtle behaviour of Erlang scheduler

[Rickard Green]

Ulf Wiger (TN/EAB) wrote:
> 
> What would be interesting, though, would be to do
> a deeper analysis of scheduling behaviour in 
> SMP Erlang. As you pointed out before, high prio
> processes don't necessarily starve normal prio
> processes in SMP, even if they busy-loop.
> 

We actually have on the todo list to review the scheduling in the 
runtime system with SMP support.

> 
> (*) Of course, the recommendation has always been to
> regard the scheduler as preemptive, and expect your 
> code to be preemptable whenever possible (that is,
> between any two operations), so in that sense, SMP
> merely makes good on this promise. (:

Yes. We have also introduced the +M command-line flag, in order to try 
to make it easier to find bugs in Erlang code that doesn't follow those 
recommendations.

BR,
Rickard Green, Erlang/OTP, Ericsson AB.