All,
Thanks to a email suggestion from an ipfilter list
reader named Noel (email text appended below), I was
able to locate the bug that was creating a conflict
between ipfilter 4.1.3 and VMWare on my Gentoo Linux
system.
Background: My previous posting described a conflict
that I was experiencing between VMWare and ipfilter on
one of my Gentoo Linux boxes. While running singly,
each program worked fine. However, when run
concurrently with ipfilter, VMWare would run normally,
but would lock up my machine during a shutdown or
reboot. The shutdown process would proceed normally
until VMWare began to shut down. At that time, a
series of "unregister_netdevice: waiting for vmnetx to
become free. Usage count = n" (x = 1 or 8, n =
1,2,3,4,5) messages would begin to appear on the
console, and the shutdown process would stop. Once in
this state, the only option for recovery was to
physically power off the machine.
I'm not sure if this is valid for all Linux distros,
but on my Gentoo box, VMWare is started and stopped
using two init scripts. The first is a "wrapper"
script (for lack of a better word) that is located in
/etc/init.d/. This script conforms to the Gentoo style
of init scripts, but all it does is call the real
VMWare init script and clean up its output.
The real VMWare init script appears to be included
with the VMWare distribution, and is located in
/etc/vmware/init.d/. This script handles the actual
startup/shutdown process of VMWare when called by the
OS specific "wrapper" init script. After reading
Noel's email, I spent some time checking for errors in
the script, and found the one that was causing the
hang at shutdown. There is a function called
vmware_product which defines the product type being
run on the local machine. On my machine, this was
defined as 'ws'. Further down in the script, in the
case statements for the start) and stop) cases, this
value is checked. However, the test in these cases was
checking for the string 'wgs'. I changed the two case
statement string comparison tests to 'ws', and that
solved the problem.
Cheers,
=-Rob
-- BEGIN PASTED EMAIL TEST --
Hi Rob,
had the same issue on Fedora Core 3 after upgrading
from FC2 and found
that it was because the init 0 level wasn't stopping
vmware in the way
it should have been. Don't know how (or if) that
translates to gentoo
and you seem to have checked a lot of those areas
already, but for what
it's worth, that's what I found was the issue.
cheers,
noel.
-- END PASTED EMAIL TEXT --
__________________________________
Do you Yahoo!?
Meet the all-new My Yahoo! - Try it today!
http://my.yahoo.com
