[Mat_as] Giovannini wrote:
> Colin Wood wrote:
> > [Mat_as] Giovannini wrote:
> > > Um, I tried GENERIC-5, GENERICSBC-5 (no need actually, but just to be
> > > sure), and MRG-5 on my Q800/20Mb/500Mb.
> > > GENERIC kernels since 1.4 release have this weird effect on me, er, my
> > > machine, that the keyboard (Series II Extended ISO) is initially WITH
> > > CAPS LOCK ON. MRG, on the other hand, correctly boots in caps lock off.
> > 
> > needless to say, this is really, really, weird....i can't think of
> > anything offhand which would cause this to occur.  you can toggle it off,
> > right?
> 
> Yes, I can; but it's somewhat annoying. I gather from your astonishment
> that nobody else is having this problem?

well, not that i've seen, but even if someone else has seen it, it's still
really, really weird since there is certainly nothing in the current code
which should cause us to come up in capslock mode.  of course i should
probably go dig through the code again just to be sure.  what's really
disturbing is that it appears to be the hwdirect code that's doing this
and not my keyboard driver...
 
> > > If things go right, should I install the rest of the snapshot?
> > 
> > if you really want to run -current, sure ;-)
> 
> Um, it's a production server. I'm I safe? ;-)

well....i'd say at this point you probably are, but i'd stick with 1.4 if
you want to be really safe.  in all reality we might have a 1.4.1 in the
next month or so, and that'll have a few bugfixes, etc. that aren't in
1.4.

> > > I "ifconfig sn0 up" a couple of times, but it gets worse and worse until
> > > I better reboot it. I would appreciate ideas about this.
> > 
> > you got me on this one....either you're getting too many packets too
> > quickly or else you ethernet interrupts aren't being serviced fast
> > enough...
> 
> It could be the traffic generated by Retrospect; on the other hand I
> don't see an unusually high packet count with netstat.
> 
> But why doesn't the driver just scratch the buffer and continue instead
> of shutting the interface down?

it's not the driver (i don't think), i believe the buffers are actually on
the chip, so it's the chip hardware that's getting confused, not the
driver (although the driver is obviously not doing something right or we
wouldn't start wedging in the first place).

later.

colin