I stumbled across it when changing the nic settings in the boot.sh. I wanted to have vmware run on nic 2 on my mac pro as opposed to the default nic. In doing so I was constantly restarting the vmware services with the boot.sh. One evening I was playing around, went to my mac pro, woke it from sleep, editted the boot.sh file to mess with something. ran the "boot.sh --restart" command and it did not spontaneously reboot. That's when it clicked in my head.

I tested a few more time by putting the mac to sleep and waking it, then starting a VM (in my case any 64-bit vm). It spontaneously rebooted. When it came back up I put it to sleep, woke it, ran the "boot.sh --restart" and started the VM and there was no reboot.

So now whenever I wake the machine from sleep I run the "boot.sh --restart" command and it doesnt reboot the mac (that is when I remember to ;)). I suppose it could be automated so that whenever my mac wakes from sleep it runs the "boot.sh --restart", but i havent gotten to it yet.

So as you can see I stumbled upon it by chance. (For reference I am running a 2008 mac pro, leopard 10.5.4, blah blah blah.) Let me know if you need more info. Thanks!

Ian

2x4-core 2.8Ghz. 8GB of RAM.

Guest:

64-bit XP (fresh install).

Steps to reproduce:

Put system to sleep.

Wake system up.

Attempt to open XP 64-bit.

Note: Since this is a fresh install, I get a warning from VMware that I should install the VMWare tools. The reboot doesn't occur until after I click Ok. Specifically, after this: 8/8/08 8:43:06 PM kernel vmci: Open socket 0xbda2c44.

See the attached 64-bit crash log (yes, I was lucky enough to get logs).

This log is basically from the time I woke up, until I started the VM. The vmci: Open socket message mentioned above is the last thing I see before I click "Ok" in the guest's alert about wanting me to install the VMWare Tools.

I think that the reason some people aren't seeing logs is because the crash can occur before syslogd can flush the logging db to disk. The first time I triggered this I saw messages in console.app that weren't in my log after I rebooted. The second time around I had the trusty old command-option-shift-4-spacebar handy, though I double checked my logs again, and it caught some of the messages (I'm not sure if it caught them all). Hopefully this helps some.

One thing I noticed is the odd vmmon garbled garbage. I saw something similar (and have posted separately about it) when installing XP 64-bit. It almost looks like some weird memory corruption is going on, unless this is some wierd obfuscated blob of gunk, but generally when I see repeating patterns and nonsensical things surrounding it I get suspicious.

• Terry

Finally, I unloaded/loaded ONLY vmmon, to see if the reboot would occur. It did NOT.

So it seems that vmmon is our culprit. Next, I'll test 32-bit systems.

Update: Yes, it affects 32-bit systems as well. I tested XP 32-bit and Win2k3 Server Standard w/SP1.

All of the systems I tested were using the default NAT'd interface. I didn't try bridged mode.

• Terry

I can confirm Terry's findings. Loading
and unloading all modules except for vmmon will result in a system
rebooting. If vmmon is reloaded after a system wakes from
sleep you will not get the reboot.

Ian

Ian