+++ This bug was initially created as a clone of Bug #756677 +++

qemu-kvm does have a "scsi" option (to be used like -device
virtio-blk-pci,drive=foo,scsi=off).  However, it only masks the feature
bit, and does not reject the command if a malicious guest disregards
the feature bits and issues a request.

(CVE-2011-4127 mitigation)

--- Additional comment from pmatouse on 2011-11-25 12:56:27 EST ---

How to test:

1) install guest which storage is backed by partition or LV (for example:  -drive file=/dev/VolGroup/bz756677,if=none,id=drive-virt0-0-1,format=raw,cache=none,aio=threads -device virtio-blk-pci,drive=drive-virt0-0-1,id=virt0-0-1)

2) patch and rebuild the guest kernel:
comment out following lines in virtblk_ioctl()@drivers/block/virtio_blk.c

//    if (!virtio_has_feature(vblk->vdev, VIRTIO_BLK_F_SCSI))
//            return -ENOTTY;

3) try sg_dd command in the guest with qemu-kvm command line virt-blk scsi option on / off (...id=virt0-0-1 / ...id=virt0-0-1,scsi=off)

3.1) unfixed qemu-kvm

3.1.1) scsi option on (not off)
  # sg_dd if=/dev/vda blk_sgio=1 bs=512 count=1
  -> ... works ...
3.1.2) scsi option off
  # sg_dd if=/dev/vda blk_sgio=1 bs=512 count=1
  -> ... works ...

-----------------------------------------------

3.2) fixed qemu-kvm

3.2.1) scsi option on (not off)
  # sg_dd if=/dev/vda blk_sgio=1 bs=512 count=1
  -> ... works ...
3.2.2) scsi option off
  # sg_dd if=/dev/vda blk_sgio=1 bs=512 count=1
  INQUIRY failed on /dev/vda
  -> ... doesn't work

If the bug is fixed, you should see the behaviour as outlined in 3.2.