Date: Wed, 14 Aug 1996 09:00:35 +0200 (MET DST) From: sos@freebsd.org To: dfr@render.com (Doug Rabson) Cc: jmacd@CS.Berkeley.EDU, jkh@time.cdrom.com, jlemon@americantv.com, freebsd-current@freebsd.org, sos@freebsd.org Subject: Re: locking up Message-ID: <199608140700.JAA11609@ra.dkuug.dk> In-Reply-To: <Pine.BSI.3.95.960813111641.8304D-100000@minnow.render.com> from "Doug Rabson" at Aug 13, 96 11:20:58 am
next in thread | previous in thread | raw e-mail | index | archive | help
In reply to Doug Rabson who wrote:
>
> On Mon, 12 Aug 1996, Josh MacDonald wrote:
>
> > > > I get this a lot. It always happens when I'm pressing the CAPS LOCK key
> > > > at the same time as some other key. I HATE IT because I end up having to
> > > > kill about 30 vi sessions, etc., etc...
> > > >
> > > > This is on 2.1.0-R with XFree86 S3 server.
> > >
> > > Does the gdb trick work to get you out? Looks like this is a bug in
> > > our console code then, and not the X server itself.
> > >
> > > > By the way, after killing the X server, you can find the "lost" character
> > > > typed (the one that caused the hang) on the virtual console you started
> > > > X from (eg, at the login prompt)...
> > >
> > > Yep!!! Exact same symptoms here. Oh Soooooooreeeeeen.. We have
> > > another nice little bug for you! :-)
> > >
> > > Jordan
Oh well...
> >
> > Hey yeah me too, cause I use caps-lock for my control key. Three or four
> > times a day I'll have this happen. I'm running XInside.
I have tried this for hours on all of my machines, no lockup :(
Even tried it whith a real beast of a kbd controller I happend
to find some month ago...
> I remember seeing some comments in the pcvt source code that some laptop
> keyboards can lose interrupts when updating the keyboard LEDs. This seems
> suspicious since most of the X keyboard lockups seem to involve the caps
> lock key.
Hmm, this one I know about, the suggested fix is to disable the
LED update sequence (or get another KBD controller)...
This will hang the keyboard both under X and in text mode...
> Maybe we should poll the keyboard sometimes just in case an interrupt has
> been lost. Calling scintr from a timeout 5 times a second would do the
> trick I suppose. Makes me feel slightly ill though. There must be a
> better way...
Hmm, if it really is a lost int, I'm not really sure where to look
for it.
Anyways I'll keep pounding on my machinery see if I can replicate
the bug here...
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Soren Schmidt (sos@FreeBSD.org) FreeBSD Core Team
So much code to hack -- so little time.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199608140700.JAA11609>