Description of problem: When trying to share a USB device from a RHEL 6.4 client to a W7 guest you will be promted to enter the root password which most users will not have. Version-Release number of selected component (if applicable): usbredir.x86_64 0.4.3-1.el6 spice-usb-share.x86_64 4.9-9.el6 usbutils.x86_64 003-4.el6 virt-viewer.x86_64 0.5.2-9.el6 How reproducible: Steps to Reproduce: 1.Select native USB for your VM 2. Restart VM 3.Connect to VM 4. Insert USB device in client 5. Select device to share it Actual results: You are promted to enter the root password on the client. Most normal users of RHEL as a desktop will not have the root password. Not they do not need the password to use the USB device on their client, only when they try to share it to their VM. Expected results: The device should be shared without a prompt for the root password just as it was with the previous USB sharing implementation. Additional info:
This behavior is controlled by PolicyKit. The problem is that USB redirection requires raw USB-device access, which can be used to for example circumvent filesystem permissions on a USB attached mass-storage device, something which the user normally cannot do (try attaching an ext2/3/4 formatted usb device and then writing to it as a regular user, or reading a file without read permissions). To avoid causing security issues the default PolicyKit policy for this requires root rights, note that if a user is member of the wheel group he only needs to type his own password. The best we can do is to document how to edit the Policy so that PolicyKit will allow the raw USB device access without prompting including a warning in the documentation about the security implications of this. I've a few short instructions on editing the policy for this here: http://hansdegoede.livejournal.com/11936.html
Can we set this setting as yes by default and document how a user can lock it down. The purpose of the package is to provide USB remoting so I don't think anyone would be surprised by the fact that the package allows raw access to the device. So two actions from my point of view 1. Change default to yes 2. Provide manpage/documentation explaining how to change it
(In reply to comment #2) > Can we set this setting as yes by default and document how a user can lock > it down. > The purpose of the package is to provide USB remoting so I don't think > anyone would be surprised by the fact that the package allows raw access to > the device. > > So two actions from my point of view > > 1. Change default to yes As said before this has security implications. If you can get permission for this from the security team I will happily change this, but until then it stays as it is.
(In reply to comment #3) > (In reply to comment #2) > > Can we set this setting as yes by default and document how a user can lock > > it down. > > The purpose of the package is to provide USB remoting so I don't think > > anyone would be surprised by the fact that the package allows raw access to > > the device. > > > > So two actions from my point of view > > > > 1. Change default to yes > > As said before this has security implications. If you can get permission for > this from the security team I will happily change this, but until then it > stays as it is. This has been in needinfo for 2 months without any response -> closing.
(In reply to comment #3) > (In reply to comment #2) > > Can we set this setting as yes by default and document how a user can lock > > it down. > > The purpose of the package is to provide USB remoting so I don't think > > anyone would be surprised by the fact that the package allows raw access to > > the device. > > > > So two actions from my point of view > > > > 1. Change default to yes > > As said before this has security implications. If you can get permission for > this from the security team I will happily change this, but until then it > stays as it is. Update: we've just gotten permission from the security team (Josh Bressers) to flip the default to yes, so we can go ahead with this. Given that this is just a configuration change, it has a very low chance of causing regressions, so I think we should do this for 6.4, adding devel-ack and proposing as blocker.
Patch send to the list -> POST
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. http://rhn.redhat.com/errata/RHBA-2013-0343.html