[plug] Loss of X mouse control

Greg Gamble gregg at maths.uwa.edu.au
Mon Sep 21 14:08:11 WST 1998

Hi Pluggers,

  I'd like to thank all the respondents to my initial message.
In particular, thanks to ...

Chris McDonald who pointed me at the site


Peter Robinson who suggested that one should explicitly set 
the no. of buttons in /etc/XF86Config ... and so I added 

	Buttons    3

to the `Pointer' section of this file (this did no harm but did
not restore mouse control)

Matt Kemner who indicated that gpm should be called with the -R
switch (indeed I found that at bootup the file /etc/rc.d/rc.local
was run and gpm was started up with the -R switch).

  The above may suggest that the suggestions were a dead loss
... but no! ... I learnt a lot and above all after checking various
possibilities that the above suggested I am now convinced that there
is a bug in XFree86 or gpm when run with my video card ... and the
suggestions helped me refine my workaround solution. So for the
record, the bug occurs with 

  XFree86 3.3.2 patch level 3
  gpm-Linux 1.1.3

when one has a

  Leadtek WinFast 3D S680  (AGP - if that matters)

video card which is 4MB, has chipset:

  S3 ViRGE/GX2

and uses the


server. The file /etc/XF86Config has Pointer section consisting

  Protocol  "PS/2"
  Device    "/dev/mouse"
  Buttons   3

Normally /dev/mouse is a symbolic link to /dev/psaux. The problem
eventuates if gpm is started up with this link in place. So the
file /etc/rc.d/rc.local was amended as follows:

rm -f /dev/mouse
touch /dev/mouse

# Running gpm
echo "Running gpm..."
gpm -R -m /dev/mouse -t ms 

rm /dev/mouse
ln -s /dev/psaux /dev/mouse

where the first 2 lines and the last 2 lines are added lines and
the middle 3 lines were in the default /etc/rc.d/rc.local. (I did
experiment with using /dev/pgmdata instead, using the switch -t pnp
for pgm, not using symbolic links etc.) 

  The above amendment to /etc/rc.d/rc.local and others that had the 
same effect were the only solutions I found that restored control to 
the mouse in X ... I hope it helps someone else.
  If I have missed something else I should have tried, do tell.

  Greg Gamble <gregg at maths.uwa.edu.au>

More information about the plug mailing list