> nope. i observed latency issues with restart + ksoftirqd as well.
Ingo, what is impact on total performance yet?
Just to remind: simple constantation of the fact that "latency is bad"
still does not imply that dead loops are inevitable.
We really could try to complicate policing of softirqs (f.e. counting cycles
on them and breaking loop with falling to ksoftirqd after it preempts
threads for more than some threshold). But it is still not clear
that it is worth to do. If you see an impact, let's try this way.
Probably, priority on ksoftirqd should be tuned. Seems, it still
should have high priority, a bit less than real-time.
Alexey
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/