[Spice-devel] [PATCH 2/5] WinUsbDev: add redirecting state to GUdevClientPrivate

[Dmitry Fleytman]

> On Jul 3, 2015, at 16:24 PM, Christophe Fergeau <cfergeau at redhat.com> wrote:
> 
> And please don't cc: Daynix - RedHat <daynix-rh-patches at daynix.com> in
> your series as this is not publicly reachable, and I keep getting
> bounces.

Sorry for that. It was cc’ed by mistake.

> 
> Christophe
> 
> On Fri, Jul 03, 2015 at 03:18:41PM +0200, Christophe Fergeau wrote:
>> On Fri, Jul 03, 2015 at 04:05:40PM +0300, Dmitry Fleytman wrote:
>>> 
>>>> On Jul 3, 2015, at 15:56 PM, Christophe Fergeau <cfergeau at redhat.com> wrote:
>>>> 
>>>> On Thu, Jul 02, 2015 at 04:41:32PM +0300, Kirill Moizik wrote:
>>>>> we need to add redirecting  state to singleton  since wnd_proc callback should not query usb devices
>>>>> in the middle of redirecting flow.
>>>> 
>>>> Why ?
>>> 
>>> Because in the middle of redirecting flow caller may get inconsistent
>>> enumeration results (due to device resets performed by UsbDk and
>>> Windows mechanisms handling those resets),
>>> UsbDk API intentionally returns error when enumeration is called during start/stop redirection flows.
>> 
>> All of this should be in the commit log.
>> 
>> If enumerating returns an error, maybe the wnd_proc callback can just
>> deal with the error, set a boolean indicating that an enumeration should
>> be performed when possible, and retry the enumeration as soon as the
>> 'redirecting' state changes?

I would say it is more robust do deal with this at the layer that knows what is going on...

>> 
>> Christophe
> 
> 
> 
>> _______________________________________________
>> Spice-devel mailing list
>> Spice-devel at lists.freedesktop.org
>> http://lists.freedesktop.org/mailman/listinfo/spice-devel
>