Dan Hollis writes:
> On Wed, 5 Jul 2000, Richard Gooch wrote:
> > RTLinux doesn't have to be conservative this way. It doesn't have to
> > work around broken hardware ("Oh, it hangs? Go use decent hardware").
> > This is a key point I'm finding difficult to get across.
>
> Youre saying we cant do better than 4000ms? Ouch.
I'm saying we can't do better than N ms, where N is an unknown
quantity (because all drivers and core kernel code has not been
audited, and said auditing would take man-years and be an ongoing
project). And people have reported large values of N. Apparently the
PS/2 driver has N >= 50. There may be other stuff that is much worse.
I think Andrew Morton measured something with N = 160.
We should always fix bugs, and fix stupid drivers. No question. We
should always seek to reduce latencies (as long as we don't kill
overall performance).
But people should not fool themselves into thinking that anybody has a
clue how big N could be. Even with a specific hardware combination,
there may always be some kernel slow algorithm that is lurking, ready
to blow out N.
Regards,
Richard....
Permanent: rgooch@atnf.csiro.au
Current: rgooch@ras.ucalgary.ca
-
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/
This archive was generated by hypermail 2b29 : Fri Jul 07 2000 - 21:00:17 EST