Re: [Spice-devel] [RFC qxl-wddm-dod 0/3] Synchronization of display mode change and pushing drawables

2017-04-27 Thread Yuri Benditovich
On Wed, Apr 26, 2017 at 3:15 PM, Frediano Ziglio wrote: > The warning is mainly there to notify the guest is sending commands for a > not existing surface. > This should not create issues unless the surface is used as source for > other commands. > Practically, Win10 does not produce 'move' oper

Re: [Spice-devel] [RFC qxl-wddm-dod 0/3] Synchronization of display mode change and pushing drawables

2017-04-26 Thread Frediano Ziglio
The warning is mainly there to notify the guest is sending commands for a not existing surface. This should not create issues unless the surface is used as source for other commands. The warning was removed as this allows the guest to fill server logs and some drivers send some spurious comma

Re: [Spice-devel] [RFC qxl-wddm-dod 0/3] Synchronization of display mode change and pushing drawables

2017-04-24 Thread Yuri Benditovich
It would be very helpful to receive a feedback regarding this problem, especially: - Do you see the reason why this erroneous flow on 'old' spice server (0.12.4, for example) can lead to significant device memory leak. - If yes: Is spice server 0.12.4 still actual in the field? - The warning was re

[Spice-devel] [RFC qxl-wddm-dod 0/3] Synchronization of display mode change and pushing drawables

2017-04-16 Thread Yuri Benditovich
This is result of investigation of HCK tests "WHQL FPO optimization check for kernel video drivers" on latest upstream driver. These tests execute in aloop fast change of display mode and full screen rendering between changes. When they run on qxl-wddm-dod driver with separate thread that pushes