>>for example, even in 2.4.0, with the low latency patch, it was >>possible to cause scheduling delays of 30-50-1000! msecs by hitting >>the VM and disk subsystems (e.g. a 4-process C++ compile while running >>Ardour). > >Hello. Scheduling delays of what? Soft-realtime processes? User >processes?
scheduling delays of *anything*. the kernel is in the middle of an operation during which is does not reschedule and the operation takes a long time. typical examples involve the VM system and the disk buffer cache. thats the heart of the low latency issue, and its where both the preemptible kernel and low latency patches come in, though in different ways. but i forgot to complete the paragraph above. the problems that were in 2.4.0 were fixed by the time we got to about 2.4.4. that kernel is pretty much free of such delays. --p
