Sebastian Strollo has provided a fix for the "Dead-loop on netdevice
imq"-issue,
new patches are available at http://trash.net/~kaber . The old page at
http://luxik.cdi.cz/~patrick/imq has been updated to redirect there. IMQ
now also
doesn't add the input/output device's link-layer header-sizes to the
packet before
enqueuing them anymore, it is not possible to accurately determine these
at the point
where it hooks.

Thanks again to Jan Rafaj who updated the patch to 2.4.21 and Sebastian
Strollo
who fixed this long-standing problem.

Bye,
Patrick