Re: latest 'guaranteed low latency' patch against 2.2.14

[John Regehr]

> how do you improve task-execution latency without rescheduling actually?
> Short of adding new hardware, or without adding a new (limited) API to
> execute user-specified code. Ie. how do you reschedule to a generic
> user-task without actually rescheduling?

Well, in NT, IRIX, BeOS, and (I think) Solaris tasks running in kernel
mode can be preempted any time as long as they're not holding a spinlock.

Certainly making a kernel preemptible increases complexity.  I'd be
interested in hearing about the tradeoffs here - hasn't Larry McVoy talked
about this in the past on this list?

> any device interrupt can be turned into an NMI interrupt - of course
> special handlers have to be written, and careful coding is needed as no
> other kernel facilities can be relied on.

Won't this hose RT-Linux?  If random device drivers start using NMI (as
Alan Cox suggested in a different mail) then it would be really nice if
there was a way to disable this behavior at compile time.

John Regehr