vboxsf causing kernel panic with mounted loopback file system after restore from suspend

[Greg McLearn]

My CentOS 5.5 guest O/S uses a read/write NTFS shared folder from my Windows XP Pro SP3 host operating system drive. This shared folder is automatically mounted on boot via /etc/fstab to /mnt/host. Within that shared folder is a 100MB file that functions as a loopback filesystem which was created like this:

dd if=/dev/zero of=/mnt/host/my-loopback-fs.img bs=1M count=100 && mke2fs -F -j /mnt/host/my-loopback-fs.img

It can be mounted like:

mount /mnt/host/my-loopback-fs.img -o loop /mnt/lo-fs

Once mounted, I close the VM and save the state.

I then bring the VM back to life. A kernel panic usually sets in seconds after the VM is reloaded. Sometimes it takes 2 minutes, sometimes it is immediate. Odd things appear with the mounted loopback filesystem if I try to use it such as I/O errors. I've managed to unmount it and remount it before the kernel panic sets in, but I cannot seem to get around the fact that it will eventually occur.

Here's uname -a:

Linux ffm-023631-vm1.example.tld 2.6.18-194.32.1.el5 #1 SMP Wed Jan 5 17:53:09 EST 2011 i686 i686 i386 GNU/Linux

I've attached the kernel panic screen and a compiled version of vboxsf.ko.

[Greg McLearn]

Kernel panic for 8664

[Greg McLearn]

Compiled vboxsf.ko kernel module for 8664 (zipped to slide under limit)

[Greg McLearn]

VBox log loading back from suspended state (Kernel panic @ ~2 minutes in is unlogged by vbox, then several more minutes to screencap & process and reset at 4:36)