1 files changed, 90 insertions, 0 deletions
diff --git a/results/classifier/118/graphic/686613 b/results/classifier/118/graphic/686613
new file mode 100644
index 000000000..97b063692
--- /dev/null
+++ b/results/classifier/118/graphic/686613
@@ -0,0 +1,90 @@
+graphic: 0.876
+files: 0.858
+device: 0.848
+permissions: 0.822
+hypervisor: 0.819
+VMM: 0.814
+PID: 0.811
+arm: 0.808
+register: 0.806
+TCG: 0.804
+semantic: 0.799
+vnc: 0.795
+ppc: 0.794
+user-level: 0.793
+network: 0.792
+performance: 0.786
+debug: 0.785
+x86: 0.781
+assembly: 0.771
+socket: 0.769
+architecture: 0.758
+boot: 0.757
+risc-v: 0.752
+mistranslation: 0.749
+virtual: 0.739
+KVM: 0.732
+kernel: 0.706
+i386: 0.667
+peripherals: 0.635
+
+USB MSD are not marked as removable
+
+ Filed from Fedora bug: https://bugzilla.redhat.com/show_bug.cgi?id=589130
+
+Guests can access USB Mass Storage Device, but fail to mark them as removable.
+
+danpb has mentioned that Linux gets this information from SCSI INQUIRY response.  So it's up to the USB Mass Storage Device to decide whether or not it wants to have a removable medium.
+
+The SCSI INQUIRY RMB (removable medium bit) tends to be set on modern USB Mass Storage Devices.  Of course they don't actually have a removable medium.
+
+One side-effect of setting this bit is that Windows Auto-Run will work if RMB=0 but not work if RMB=0.  Also, on RMB=1 devices only the first partition is displayed by Windows - you can't have more than one partition on the device.
+
+For more information, see: http://www.microsoft.com/whdc/connect/usb/usbfaq.mspx
+
+So at the end of the day, it's up to QEMU whether or not to mark USB MSDs as having removable media.  Since modern devices tend to set RMB=1, we probably should too.
+
+"One side-effect of setting this bit is that Windows Auto-Run will work if RMB=0 but not work if RMB=0" should read "but not work if RMB=1".
+
+A sampling of physical USB devices:
+1. Noname 2GB USB stick: RMB=1
+2. Nexus One USB storage: RMB=1
+3. LaCie USB harddisk: RMB=0
+
+I suspect there's no setting that will satisfy everyone here.  It makes sense for a USB harddisk to set RMB=0 because that allows you to put on more than one partition under Windows.
+
+Stefan, you fixed this  in QEMU 24 Jan 2011 so this bug can be marked as Fixed Released?
+
+SEE: http://lists.gnu.org/archive/html/qemu-devel/2011-01/msg02425.html
+Append ",removable=on" to a device definition for USB Mass Storage Devices to override the removable bit (but off by default):
+> qemu -usb
+>     -drive if=none,file=test.img,cache=none,id=disk0
+>     -device usb-storage,drive=disk0,removable=on
+
+However, I cannot see a way to set "removable=on" for usb-storage devices started via libvirt (in 12.04, with libvirt-bin 0.9.8-2ubuntu17.1 and virt-manager 0.9.1-1ubuntu5.1). 
+
+So, is a feature request required for virt-manager to expose this option for virtual disks (default to 'on' for bus='usb') and add support to libvirt's XML to set "removable=on" ?
+
+On Mon, Jul 2, 2012 at 4:04 PM, Sebastian Malcolm
+<email address hidden> wrote:
+> Stefan, you fixed this  in QEMU 24 Jan 2011 so this bug can be marked as
+> Fixed Released?
+
+Yes
+
+> However, I cannot see a way to set "removable=on" for usb-storage
+> devices started via libvirt (in 12.04, with libvirt-bin
+> 0.9.8-2ubuntu17.1 and virt-manager 0.9.1-1ubuntu5.1).
+>
+> So, is a feature request required for virt-manager to expose this option
+> for virtual disks (default to 'on' for bus='usb') and add support to
+> libvirt's XML to set "removable=on" ?
+
+Yes, I just checked libvirt.git/master and cannot see a way to set the
+removable option.  I'm afraid libvirt also needs changes.
+
+Stefan
+
+
+Setting status to "Fix released" according to comment #5 (if there is something left to do for libvirt, please consult their bugtracker instead)
+