On Wed, 11 Dec 1996, Mark Brinicombe wrote:

> ok leave it long enough and it will continue. This is on my todo list.
> The problem occurs in the keyboard reset routine. There are occasions when the
> keyboard fails to respond correctly on in the expected time to a reset command.
> If so the driver tries again and again .... eventually it gives up but not
> until rather a lot of retries.
> The quick fix is to increase the wait time for each retry or to reduce the
> number of retries.
> 
> Ohhh, just looking at the code. I delays 1ms each retry and will retry 100000
> times so that means if the that after 100secs it should continue the boot.
> I think that retry count could be reduced by a factor of 10 at least.
> 
> If you want to patch it look at kbdreset() in arm32/mainbus/kbd.c

ok. I'll look at it.

Now, looking at the TODO in arm32, item 1 is being done indirectly by Ian
McEwan  - ijm@doc.ic.ac.uk the in_cksum code is writtern but has a bug in it
that he wants to fix before it's distributed the speed difference is:

[from spray localhost]
before:

423 packets/sec, 35.5K bytes/sec

afterwards:

516 packets/sec, 43.4K bytes/sec

	Intrestingly enough spraying my machine from another drops 95.6% of
the packets - tho I suspect the etherH rather than anything else ;-)

(I've known about this for weeks and have been meaning to tell you for about
the same length of time - the road to hell is paved with good intentions ;-)

Printing is mentioned - whats wrong with it since this RiscBSD machine is
printer-serving for our lan and it works ok?

Compiling -current again - in ioconf.c the cfdata structs all have an extra
element at the end (compared with <sys/device.h> - do I need a newer version
of configure?

-- 
Jasper Wallace-Internet Vision | Caving-Acorn-RiscBSD-Chocolate-Banks-Fish
My opinions != IVisions's      | PovRay-FVWM-Perl-C-Mono-Food-Wired-?
E-Mail:jasper@ivision.co.uk    | This is the Internet, untamed and free,
Tel: +44 171 589 4500 (work)   | Where you can be anything you want to be