4.3.22 Windows 10 TP - 3D driver fail

[beikaigun]

All -

I'm running a near new instance of Fedora 21 as the host OS and installed VB 4.3.22 via the virtualbox repository. While still using VB 4.3.20, I have created a couple of VMs based on the Windows Tech Preview 64-bit .iso using the Windows 8.1 configuration as the base for the new VMs. At first, I noticed that the interface would not auto-resize the guest OS, even after GA was installed. However, after an update to VB 4.3.22, this functionality appeared to work on one of the test VMs. I cloned this to start installing some software to test, and it seemed to be going good.

Then, it failed. Something appears to have changed, possibly with the TP, that caused the video driver to no longer work. When booted, the system would reach login. After logging in, the system would freeze and the display stop updating. This would affect the VB screen as well. Killing the VM would restore use of the main VB window. Through trial and error, I discovered that by disabling 3D acceleration, I was able to boot the VM, but the guest display would not auto-resize anymore. Looking in the windows settings, there appears to be an issue with the GA driver. It is loading the Windows 8 version, but it simply isn't working right. I also discovered that this issue exists with my "gold" copy from which I cloned.

Since it had worked, I looked through the update log on the windows guest. The only update I've found in the last couple of days was a virus definition update to windows defender.

Any ideas what could have happened and why?

[loukingjr]

please attach the vbox.log (compressed) for your W10 guest.

[beikaigun]

VBox.log.1.tar.gz

(19.28 KiB) Downloaded 12 times

VBox.log.tar.gz

(18.97 KiB) Downloaded 9 times

VBox.log

Log file from test VM

(111.04 KiB) Downloaded 17 times

EDIT: added compressed versions of both log files for the test VM (new clone)

[beikaigun]

I also tried to post from the "gold" copy, but the file was to big. One thing I'm noticing is lines like this at the end:

Guest Log: VBoxMP::DxgkDdiQueryCurrentFenceNew: WARNING! :=>DxgkDdiQueryCurrentFenceNew
00:03:13.904766 Guest Log: VBoxMP::DxgkDdiQueryCurrentFenceNew: WARNING! :uncompleted fences, u32FenceSubmitted(1909), u32FenceCompleted(1907) u32FenceProcessed(1907)
00:03:14.920600 Guest Log: VBoxMP::DxgkDdiQueryCurrentFenceNew: WARNING! :uncompleted fences, u32FenceSubmitted(1909), u32FenceCompleted(1907) u32FenceProcessed(1907)
00:03:15.936278 Guest Log: VBoxMP::DxgkDdiQueryCurrentFenceNew: WARNING! :uncompleted fences, u32FenceSubmitted(1909), u32FenceCompleted(1907) u32FenceProcessed(1907)

Any idea what this means?

[beikaigun]

VBox.log.tar.gz

Okay, I just noticed your word "Compressed" - sorry, here is the gold copy last run log.

(21.57 KiB) Downloaded 6 times

[loukingjr]

try the fixed GAs… https://www.virtualbox.org/download/tes ... -98282.iso

[loukingjr]

It looks like you are using snapshots?

Code: Select all

00:00:06.119679 Guest Additions information report: Version 4.3.22 r98236 '4.3.22'
00:00:06.119730 Guest Additions information report: Interface = 0x00010004 osType = 0x0003B100 (64-bit)
00:00:06.119783 Guest Additions capability report: (0x0 -> 0x0) seamless: no, hostWindowMapping: no, graphics: no
00:00:55.267344 Guest Additions capability report: (0x0 -> 0x1) seamless: yes, hostWindowMapping: no, graphics: no
00:00:55.288324 Guest Additions capability report: (0x1 -> 0x5) seamless: yes, hostWindowMapping: no, graphics: yes
00:01:19.009264 Guest Additions capability report: (0x5 -> 0x0) seamless: no, hostWindowMapping: no, graphics: no
00:01:21.835511 Guest Additions capability report: (0x0 -> 0x1) seamless: yes, hostWindowMapping: no, graphics: no
00:01:21.835653 Guest Additions capability report: (0x1 -> 0x5) seamless: yes, hostWindowMapping: no, graphics: yes
00:01:27.651020 Guest Additions capability report: (0x5 -> 0x0) seamless: no, hostWindowMapping: no, graphics: no
00:01:29.372258 Guest Additions capability report: (0x0 -> 0x1) seamless: yes, hostWindowMapping: no, graphics: no
00:01:29.372383 Guest Additions capability report: (0x1 -> 0x5) seamless: yes, hostWindowMapping: no, graphics: yes
00:01:37.760520 Guest Additions capability report: (0x5 -> 0x0) seamless: no, hostWindowMapping: no, graphics: no
00:01:41.071206 Guest Additions capability report: (0x0 -> 0x1) seamless: yes, hostWindowMapping: no, graphics: no
00:01:41.071744 Guest Additions capability report: (0x1 -> 0x5) seamless: yes, hostWindowMapping: no, graphics: yes
00:02:10.387075 Guest Additions capability report: (0x5 -> 0x5) seamless: yes, hostWindowMapping: no, graphics: yes
00:02:11.459458 Guest Additions capability report: (0x5 -> 0x5) seamless: yes, hostWindowMapping: no, graphics: yes
00:02:11.459531 Guest Additions capability report: (0x5 -> 0x0) seamless: no, hostWindowMapping: no, graphics: no
00:02:11.459576 Guest Additions capability report: (0x0 -> 0x0) seamless: no, hostWindowMapping: no, graphics: no
00:02:22.221222 Guest Additions capability report: (0x0 -> 0x0) seamless: no, hostWindowMapping: no, graphics: no
00:02:30.540592 Guest Additions capability report: (0x0 -> 0x0) seamless: no, hostWindowMapping: no, graphics: no
00:02:41.549831 Guest Additions information report: Version 4.3.22 r98236 '4.3.22'
00:02:41.549881 Guest Additions information report: Interface = 0x00010004 osType = 0x0003B100 (64-bit)
00:02:41.550020 Guest Additions capability report: (0x0 -> 0x0) seamless: no, hostWindowMapping: no, graphics: no
00:02:43.573760 Guest Additions capability report: (0x0 -> 0x0) seamless: no, hostWindowMapping: no, graphics: no
 

[beikaigun]

Downloading to install now.

[beikaigun]

No, I was doing full clones. The goal is to build a good "gold" copy, then clone to create special purpose VMs - starting with one to run image scanners.

It looks like you are using snapshots?

Code: Select all

00:00:06.119679 Guest Additions information report: Version 4.3.22 r98236 '4.3.22'
00:00:06.119730 Guest Additions information report: Interface = 0x00010004 osType = 0x0003B100 (64-bit)
00:00:06.119783 Guest Additions capability report: (0x0 -> 0x0) seamless: no, hostWindowMapping: no, graphics: no
00:00:55.267344 Guest Additions capability report: (0x0 -> 0x1) seamless: yes, hostWindowMapping: no, graphics: no
00:00:55.288324 Guest Additions capability report: (0x1 -> 0x5) seamless: yes, hostWindowMapping: no, graphics: yes
00:01:19.009264 Guest Additions capability report: (0x5 -> 0x0) seamless: no, hostWindowMapping: no, graphics: no
00:01:21.835511 Guest Additions capability report: (0x0 -> 0x1) seamless: yes, hostWindowMapping: no, graphics: no
00:01:21.835653 Guest Additions capability report: (0x1 -> 0x5) seamless: yes, hostWindowMapping: no, graphics: yes
00:01:27.651020 Guest Additions capability report: (0x5 -> 0x0) seamless: no, hostWindowMapping: no, graphics: no
00:01:29.372258 Guest Additions capability report: (0x0 -> 0x1) seamless: yes, hostWindowMapping: no, graphics: no
00:01:29.372383 Guest Additions capability report: (0x1 -> 0x5) seamless: yes, hostWindowMapping: no, graphics: yes
00:01:37.760520 Guest Additions capability report: (0x5 -> 0x0) seamless: no, hostWindowMapping: no, graphics: no
00:01:41.071206 Guest Additions capability report: (0x0 -> 0x1) seamless: yes, hostWindowMapping: no, graphics: no
00:01:41.071744 Guest Additions capability report: (0x1 -> 0x5) seamless: yes, hostWindowMapping: no, graphics: yes
00:02:10.387075 Guest Additions capability report: (0x5 -> 0x5) seamless: yes, hostWindowMapping: no, graphics: yes
00:02:11.459458 Guest Additions capability report: (0x5 -> 0x5) seamless: yes, hostWindowMapping: no, graphics: yes
00:02:11.459531 Guest Additions capability report: (0x5 -> 0x0) seamless: no, hostWindowMapping: no, graphics: no
00:02:11.459576 Guest Additions capability report: (0x0 -> 0x0) seamless: no, hostWindowMapping: no, graphics: no
00:02:22.221222 Guest Additions capability report: (0x0 -> 0x0) seamless: no, hostWindowMapping: no, graphics: no
00:02:30.540592 Guest Additions capability report: (0x0 -> 0x0) seamless: no, hostWindowMapping: no, graphics: no
00:02:41.549831 Guest Additions information report: Version 4.3.22 r98236 '4.3.22'
00:02:41.549881 Guest Additions information report: Interface = 0x00010004 osType = 0x0003B100 (64-bit)
00:02:41.550020 Guest Additions capability report: (0x0 -> 0x0) seamless: no, hostWindowMapping: no, graphics: no
00:02:43.573760 Guest Additions capability report: (0x0 -> 0x0) seamless: no, hostWindowMapping: no, graphics: no
 

[beikaigun]

No joy - same issue as before. I'm going to try a full removal and reinstall, see how that goes....

[beikaigun]

Here is a screen shot from the VM showing the driver in error.

[beikaigun]

Again, no joy. Rolled back the driver, and then installed the newer 4.3.23 driver and rebooted. I'm seeing those same error messages....

[beikaigun]

I may be on to something, thanks to you.

In looking through the logs and finding that error, I was able to google part of it and found:
https://www.virtualbox.org/ticket/13510
http://ubuntuforums.org/showthread.php?t=2160297
and
https://www.virtualbox.org/ticket/13599

This led me to believe I may have an issue with whatever the current kernel is on my system. I therefore did an update. Kernel mod went from 3.18.6-200.fc21.x86_64 to 3.18.7-200.fc21.x86_64. After a reboot, things worked for a moment with the 4.3.22 GA driver, but it still isn't consistent. I'm wondering if VirtualBox may be having some issues with the kernel that fedora is using at present....

[beikaigun]

So... It now looks like the first Windows 10 VM after a reboot will actually function. After that, all VMs (even a restart of the first) exhibit this behavior. Any thoughts as to what might be changing to cause this?

BTW - error logs for the system do not show any NVRAM issue like in the other posts.

[beikaigun]

First - I'm kind of curious as to why you thought I was using snapshots. I never have, should I be?

Second - and this is for anyone else who wants to jump in - I've been trying to read up on the error and what it is. I've determined the following:
1) It is definitely related to the video driver. The call that is generating the error is part of Virtualbox's WDDM miniport display driver. (http://www.virtualbox.org/svn/vbox/trun ... MPWddm.cpp) Now whether this means there is a problem with the driver or there is a problem with something it relies on is still a mystery.

2) The term "fences" was kind of unusual. It turns out this is a term to refer to a graphical data structure. A few more details can be found through this page (https://msdn.microsoft.com/en-us/librar ... s.85).aspx) - look in the remarks section for a basic description of the data structure.

3) based on the above, the complaint appears to be that two of the submitted jobs using fences are not returning (1909 submitted, 1907 completed). Still no idea why this is, but it suggests that data is getting lost somewhere or that the fences are being rejected by the GPU/kernel in a way that VB is not tolerant of.

4) "NVRM", as seen in the linked ticket above (https://www.virtualbox.org/ticket/13599) seems to be specific to Nvidia GPUs. I'm using the Intel HD GPU on the Core i7 chip, not a separate graphics card. Different card, different drivers. I may have to resurrect one of my old Nvidia cards and see if there is a difference, but I'm a little reluctant to to go that route at this time.

5) Window 10 initially had the WDDM 1.3, but will have a new WDDM 2.0 implementation when shipped. It's supposed to still support anything over 1.0 - which was first released with Windows Vista - however, as Windows 10 is still beta, that may be part of what is changing around. Details of the subsystem can be found here (https://msdn.microsoft.com/en-us/librar ... s.85).aspx)

6) It seems that at present, when I reboot the system, the first Windows 10 VM that comes up does so with no issues. It's only after that (including when I reboot the VM) that the issue happens - although I do have one VM for which it happens regardless. That one has some 3rd party software installed. I plan to investigate further to see if there is any other thing that might be affecting it.

So, hopefully this is enough for someone more familiar with graphics drivers to dive in and take a look. I'll also test with earlier kernels (although I'm still limited to the current series for Fedora 21 without a reinstall to an earlier variant) to see if kernel mods somewhere made a difference.