wglDXSetResourceShareHandleNV() fails

[Sergey_A_1]

Hello!

I reported this problem a week ago at https://servicerequestmgmt.intel.com (Service Ticket 8001209460) but got no help, thus trying to re-post it here because the problem looks really easy to fix at the graphic driver level.

We experience troubles with WGL_NV_DX_interop extension. wglDXSetResourceShareHandleNV(void *dxObject, HANDLE shareHandle) call fails on attempts to resize a 3D window, and rendering is broken.

We re-create a Direct3D surface when a window is resized, but its physical address sometimes coincides with the address of the previous instance. wglDXSetResourceShareHandleNV() returns FALSE when dxObject argument value is the same as the value passed to the previous call of this function. NVIDIA and AMD drivers handle this situation without any problems, however Intel graphic drivers seems to fail systematically (checked on Intel HD Graphics 2500 and 4000).

The very first call to wglDXSetResourceShareHandleNV(void *dxObject, HANDLE shareHandle) returns TRUE correctly, and rendering works fine till the first attempt to resize the 3D window.

System Setup Information:
-----------------------------------------

System Used: Fujitsu Lifebook AH532/G52
CPU SKU: Intel(R) Core(TM) i7-3632QM
GPU SKU: Intel(R) HD Graphics 4000
Processor Line: N/A
System BIOS Version: N/A
CMOS settings: N/A
Graphics Driver Version: 10.18.10.4061 (15.33.32.64.4061) 
GOP/VBIOS Version: N/A
Operating System: Windows
OS Version: Windows 7 Professional(6.1.7601)
API: OpenGL, WGL, WGL_NV_DX_interop extension
Occurs on non-Intel GPUs?: NO

Steps to Reproduce:
-------------------------------
I have a C++ code snippet that reproduces the problem, you can see it online: 

http://git.dev.opencascade.org/gitweb/?p=occt.git;a=commitdiff;h=65cd950ae1df04bb80c69fb13e896b2fcfa66080#patch12

The problem occurs in D3DHost_FrameBuffer::Init() call.

I can try to upload a test WPF application that reproduces the problem.

Expected Results:
-------------------------------

wglDXSetResourceShareHandleNV(void *dxObject, HANDLE shareHandle) should return TRUE if a new Direct3D surface is passed to through dxObject argument it even if its value coincides with some previously passed value.

Actual Results:
-------------------------------

wglDXSetResourceShareHandleNV(void *dxObject, HANDLE shareHandle) returns FALSE if dxObject argument value coincides with the value passed to a previous call of this function.

Additional Information:
-------------------------------

The graphic driver was downloaded from https://downloadcenter.intel.com/download/24593/Intel-HD-Graphics-Driver-for-Windows-7-8-8-1-64-bit.

The same problem was reproduced with desktop Intel Core i5 CPU and Intel HD Graphics 2500.

[Michael_C_Intel2]

Hi Sergey,

I have submitted an internal ticket and talked to the driver development team. If you get a response from your ticket please let me know.

-Michael

[Sergey_A_1]

Hello Michael, Thanks a lot for fast reply! Actually, my ticket was closed (I was redirected to Intel software support) so I do not expect any further response from that channel. Now I see that fortunately I might get some real results with your help. Best regards, Sergey

[Sergey_A_1]

Michael Coppock (Intel) wrote:

I have submitted an internal ticket and talked to the driver development team.

Hello Michael,

Is there any news from the driver development team? What is their opinion? Can we expect this bug to be fixed some day? Or maybe there is even some forecast for the correction date/corrected driver release?

Thanks a lot in advance for any information about the progress!

Best regards,

Sergey

[Michael_C_Intel2]

Hi Sergey,

I have no updates yet, the driver team is looking at many issues and have yet to start debug on this one. If it is indeed a driver issue it will be fixed. I cannot forecast when as of yet, we need to find the problem root cause.

-Michael

[Michael_C_Intel2]

Hi Sergey,

Investigation is starting on this issue, the engineer asked if he could a WPF application for the issue to speed up his investigation. Can you try and get one to me? You can send me a private message with a attachment.

Thanks,

-Michael

[Sergey_A_1]

Hello Michael,

Done, check your message box please.

Best regards,

Sergey

[Michael_C_Intel2]

Got it.

Thanks, I will pass this along.

-Michael

[Michael_C_Intel2]

Hi Everyone,

We have confirmed it is a driver issue and the graphics driver development team is finding the root cause. Once we have identified the cause and the required fix I can give a ETA on when a driver will be released with said fix.

-Michael

[Michael_C_Intel2]

Hi,

This issue has been fixed and the next 15.36 or 15.40 driver release should work for you.

-Michael