Re: gpm/serial bug?

Tyson D Sawyer (tyson@rwii.com)
Sat, 16 Nov 1996 13:55:19 -0500 (EST)

> Date: Sat, 16 Nov 1996 00:40:35 -0500 (EST)
> From: Tyson D Sawyer <tyson@rwii.com>
>
>
> I just spent too many hours figuring out that gpm was running and
> shouldn't have been. I had a serial device attached to /dev/cua0
> on a headless, sort of embedded, Linux PC. A short time after
> starting the program that controls the device on the serial port
> the the kernel would start thinking that bogus data was being typed
> on the console.
>
> This isn't a kernel problem; this isn't even a gpm bug. GPM is a
> program which listens to mouse events (coming from /dev/cua0, in your
> case, because that's how it was started in your /etc/rc scripts) and
> provides cut and paste functionality to the console. Apparently your
> devices was emitting characters which were getting interpreted as mouse
> motions and mouse clicks, which was causing gpm to perform "cut and
> paste" functions.
>
> That's how come you were getting bogus data that was appearing to be
> typed on the console --- that's what cut and paste is all about.
>
> This was a strictly a user configuration error, no more and no less.
> You shouldn't be running gpm on a serial port unless you have a serial
> mouse hooked up to that serial port.
>
> - Ted

OK. I mostly agree with you. I did realize that clearing up the
configuration error eliminates the problem. What I question is
the if reasonable sanity checking on the part of gpm should result
in this:

Nov 15 12:52:54 guest1 /sbin/mingetty[215]: tty1: invalid character for login name
[...]
Nov 15 12:53:45 guest1 init: Id "1" respawning too fast: disabled for 5 minutes

...and more importantly why it should result in a machine that you
can't remotely login to anymore. Also, X clients being remotly
run on the machine would stop, or at least their communications would
stop. This is why I was asking if there might be a kernel problem
that is not normally encountered on a correctly configured machine.

I can't claim that the computer was completely crashing but I was not
able to do anything with the machine other than switch consoles and
toggle keyboard LEDs. <Ctl><Alt><del> didn't work under these
conditions and that doesn't seem right. Does <Ctl><Alt><del> work
on a console with a getty that is disabled by init? Do the other
kernel trapped keys work under these conditions? I was not able
to remember those key combinations at the time so I don't know but
expect they would have worked.

Thanks for pointing out that gpm was pasting. That had not occured
to me.

Ty