[Spice-devel] [PATCH] usbredir: fix redirection of user-accesible device nodes.

Michal Suchanek michal.suchanek at ruk.cuni.cz
Mon Jun 29 09:59:26 PDT 2015


Excerpts from Christophe Fergeau's message of Mon Jun 29 17:22:23 +0200 2015:
> On Mon, Jun 29, 2015 at 05:11:36PM +0200, Michal Suchanek wrote:
> > Excerpts from Christophe Fergeau's message of Mon Jun 29 17:01:23 +0200 2015:
> > > On Mon, Jun 29, 2015 at 04:52:31PM +0200, Suchánek Michal wrote:
> > > > Or the other way around compiling in policykit support *should not
> > > > disable* access to already accessible devices.
> > > 
> > > We agree on that, and the log of the commit you revert also agrees
> > > « So this patch changes things to first call the helper and only then
> > > try to open the device node. »
> > > 
> > > > It's indeed the case. However, this is merely a cosmetic issue while
> > > > the fix for the cosmetic issue causes a functional error.
> > > 
> > > But to fix your issue, you don't have to reintroduce this cosmetic
> > > issue, you could just make sure we try to directly open the USB device
> > > if the policykit call failed.
> > 
> > The problem is how do you present the error then?
> > 
> > If policykit call fails and you proceed to open the device directly do
> > you report the policykit error or the libusb error or somehow merge
> > them?
> 
> I'd present the policykit error when the binary is built with policykit
> support.

This gives some interesting results. In spicy I tried to attach a device
which is already attached through qemu usb redirection. This fails and
libusb error 99 (other) is logged, the policykit failure is displayed in
the dialog and the device seems to get upset and has to be reset by the
guest.

It seems to recover reasonably well but the fact that there was 'other
error' rather than the usual 'permission error' is hidden.

Thanks

Michal


More information about the Spice-devel mailing list