Re: Right way to cut latencies? (was: <no subject>)

Brian Swetland (swetland@be.com)
Sun, 29 Aug 1999 05:50:04 -0700

[David Olofson <audiality@swipnet.se>]
> On Sun, 29 Aug 1999, Benno Senoner wrote:
> > But I'm still not convinced it this is really possible.
> > You have to check rescheduling during
> > kernel memcopy routines , or large mem blocks moves, since at 100MB/sec
> > 1MB of data = 10msec latency.
>
> This doesn't sound too sexy... But it seems pretty obvious to me that
> it can't be solved by speeding things up. This is why RTKs are preemptive,
> or keep that kind of operations in user space...

Pardon the newbie question, but am I right in understanding from this
discussion that preemption doesn't happen unless explicitly asked
for (by schedule())? If the a driver was entered and proceeded to
do some very long term PIO without sleeping, would it prevent other
processes from running?

Or did I just miss something obvious?

Brian

-- 
 Brian Swetland  | Kernel Engineer | "Home is where the dot files are."
 swetland@be.com | Be Incorporated |                       -- Keith Garner


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/