Some VirtualBox guests show black cursor

caj2arch · 2016-09-13 21:28:13

I had the same fat-bitted mouse pointer show up in the VirtualBox 5.1.0 upgrade cycle. Thought of it just as a cosmetic irritation and eventually solved (most of) it by fiddling with the mouse pointer options for each Windows user. Only aspect that isn't corrected is the mouse pointer on the pick-a-user-for-login screen.

Summary details first: Host OS is 4.7.2-1-ARCH #1 SMP PREEMPT Sat Aug 20 23:02:56 CEST 2016 x86_64 GNU/Linux, UI is xfce 4.12, VirtualBox is 5.1.4, guest OS is Windows 7-64 Pro. All are current through pacman and Windows Update.

This thread is of GREAT interest to me because of TobalJackson's 2016-08-31 posting. YES, I am experiencing sporadic, spontaneous crashes in the Windows guest. No error message, no BSOD, just the abrupt disappearance of the guest window, after which the VBox Manager shows its status as Aborted. Some restarts have been seamless, some have gone through the "Windows did not shut down successfully" screen. Most of the crashes have occurred while doing nothing exotic in QuickBooks 2016, but most recently the crashes have occurred immediately after logging into Windows. At this point I probably have a corrupted Windows .vdi and will retreat to what I think is a previous good copy. ("Good" is relative here.)

I posted a query on virtualbox.org with exhaustive detail. See https://forums.virtualbox.org/viewtopic.php?f=2&t=79693 which recommends installing Oracle 5.1.6 (available in the AUR), though only to bring VBox up to the current release as a starter.

I will try disabling mouse integration and would be VERY happy to hear the recommendation for Terrowin's question on determining whether Windows or the VBox process is crashing.

Many thanks.

TobalJackson · 2016-09-14 14:48:06

Terrowin: If you haven't enabled windows crash dumps, refer to this article here: https://msdn.microsoft.com/en-us/librar … s.85).aspx

In case this link breaks, I'll copy the relevant info here:

To change these settings, go to Control Panel > System and Security > System. Click Advanced system settings. Under Startup and Recovery, click Settings. Under Write Debugging Information, you can specify a kernel-mode dump file setting. Only one dump file can be created for any given crash. Choose Complete Memory Dump

Once you have this setting on, a windows crash will generate a file at

%SystemRoot%\MEMORY.DMP

. You don't really have to do anything with this file (although you could analyze the circumstances of the crash using WinDbg) other than to note its creation date/time. If the setting I've described above is turned on, and windows does crash, this file will be created with a date/timestamp of the crash.

If, however, Virtualbox is what is crashing, then the windows guest will only see it as a sudden power loss event, and the MEMORY.DMP file will not be created. This is a method you can use to determine if windows or VBox is crashing.

=====================================================

caj2arch: After disabling mouse pointer integration (every boot of my guest - it doesn't seem to be a setting that can be made permanent >_>), the sporadic-turned-frequent crashing of my win7 guest has stopped entirely. In the few instances where I haven't disabled mouse pointer integration before getting to the log in screen I've had the win7 guest crash about 20% of the time when logging in. Now, as long as I disable the integration before "pressing ctrl + alt + del to log in", I avoid VBox crashing entirely.

Hope this helps, and also that eventually this problem will be fixed.

Terrowin · 2016-09-14 22:53:06

TobalJackson wrote:

Once you have this setting on, a windows crash will generate a file at
%SystemRoot%\MEMORY.DMP
. You don't really have to do anything with this file (although you could analyze the circumstances of the crash using WinDbg) other than to note its creation date/time. If the setting I've described above is turned on, and windows does crash, this file will be created with a date/timestamp of the crash.
If, however, Virtualbox is what is crashing, then the windows guest will only see it as a sudden power loss event, and the MEMORY.DMP file will not be created. This is a method you can use to determine if windows or VBox is crashing.
Hope this helps, and also that eventually this problem will be fixed.

Hi TobalJackson, thank you for the explanation. I have the settings enabled as suggested for recording a Complete Memory Dump. As I expected after a crash, there is no MEMORY.DMP file and it must be VirtualBox crashing. Doing some testing I noted that the majority of the time it is random crashes, however having Guest Additions installed and navigating within a shared folder between the host and guest while in the guest Windows OS can often trigger a crash. On newer Windows OS (8.1+) I disable mouse integration and it will still crash. I'm not sure on my end if Windows 7 would still crash eventually. It has been suggested that it is a defective version of VirtualBox in the Arch repos as apparently some other Arch Linux user had installed the version from the Oracle website and their problem was solved. hb860 also touched on this point. It is a long time now that these problems have been occurring, would really like to see a fix for these two issues affecting our version in the repos. I never had problems with VirtualBox aborts before the 5.x series.

caj2arch · 2016-09-15 01:43:05

With cautious optimism and thanks:

I disabled mouse integration and have had two very short Windows sessions with no crash. They were limited to running Windows Update and rebooting to finish the update installation.

I appear to have set mouse integration off and made it permanent by the following steps: In VBox Manager and the Windows guest selected (but not running) choose Settings (click the gear), choose User Interface from the left menu, drop down the Input menu, and uncheck the Mouse Integration box. Clicking OK gives the message "Saving Settings" and the setting do appear to be maintained in subsequent sessions.

Will also configure for Windows crash dumps to help determine the misbehaving culprit.

WFV · 2016-09-15 21:35:21

TobalJackson wrote:

Has anyone else experienced the crashing issue I described above with Win7 x64 (or other) guests with virtualbox >= 5.1.x?

This forum is about the black cursor and might not be related to the guests crashing. There are several bug reports about guests crashing with VBox =>5.1:
https://www.virtualbox.org/ticket/15874

I'm getting frequent crashes (aborted) randomly since updating =>5.1 - maybe run for several days without issue, then go to open a file on the guest and bam vbox guest closes instantly and the Main Menu shows "aborted". Some rolled back to 5.0.28 for stability. From dmesg all relevant instances immediately preceding and following, the rest of dmesg time line is far enough from either end it is unlikely related. The abort is instant:

[30347.872018] usb 6-5: new low-speed USB device number 15 using ohci-pci
[30347.924020] usb 7-1: new low-speed USB device number 14 using ohci-pci
[30348.029064] input: Microsoft Microsoft 3-Button Mouse with IntelliEye(TM) as /devices/pci0000:00/0000:00:12.0/usb6/6-5/6-5:1.0/0003:045E:0040.0018/input/input41
[30348.080169] hid-generic 0003:045E:0040.0018: input,hidraw0: USB HID v1.10 Mouse [Microsoft Microsoft 3-Button Mouse with IntelliEye(TM)] on usb-0000:00:12.0-5/input0
[30348.082570] input: Dell Dell USB Keyboard as /devices/pci0000:00/0000:00:13.0/usb7/7-1/7-1:1.0/0003:413C:2003.0019/input/input42
[30348.134329] hid-generic 0003:413C:2003.0019: input,hidraw1: USB HID v1.10 Keyboard [Dell Dell USB Keyboard] on usb-0000:00:13.0-1/input0
[30378.973065] EMT-3[1293]: segfault at 10 ip 00007f85f07f2202 sp 00007f861e28fbb0 error 4 in VBoxDD.so[7f85f06ee000+191000]
[30379.239177] device enp4s0 left promiscuous mode
[30379.244048] vboxnetflt: 282087 out of 623089 packets were not sent (directed to host)
[30388.110666] vboxdrv: ffffffffa08f1020 VMMR0.r0
[30388.441886] vboxdrv: ffffffffa09f3020 VBoxDDR0.r0
[30388.548472] VBoxNetFlt: attached to 'enp4s0' / c8:60:00:60:b7:ed
[30388.590419] vboxdrv: ffffffffa00c3020 VBoxEhciR0.r0
[30388.595696] device enp4s0 entered promiscuous mode
[31056.094219] usb 6-5: USB disconnect, device number 15
[31056.094308] usb 7-1: USB disconnect, device number 14

EDIT: added dmesg segfault and changed the bug report links.

Last edited by WFV (2016-09-16 06:18:13)

caj2arch · 2016-09-17 14:22:03

WFV: Your experience matches mine, except I don't think I've gone 'several days' without a problem. Thanks for the link to the virtualbox.org ticket. Good to know that side is aware of the issue. I'll watch what goes on there. And my log typically includes the vboxnetflt line about packets not sent.

TobalJackson: It turns out that I have had memory dump set all along. No Memory.DMP file in C:\WINDOWS, which argues on the side of VBox crashing, not Windows.

I agree with WFV that this thread started out as a cursor issue. Does it make sense to open a new post? A more pertinent title might attract more attention.

caj2arch · 2016-09-17 20:52:33

Just had the Windows 7-64 guest disappear as I was logging off my non-admin user account. Confirmed that mouse integration is set to OFF. VBox Manager shows guest status as Aborted. Upon restart, get "Windows did not shut down normally". After starting Windows normally, attempted to log in with admin account and guest disappeared again after showing wallpaper and outlines of a few windows. GRRR!!

Will start a new thread for this issue.

bhrgunatha · 2016-09-18 07:18:06

kwash wrote:

Same here. The issue is present in 5.1.4.

Just updated to 5.1.6.
It's still not fixed.

Last edited by bhrgunatha (2016-09-18 07:22:25)

Terrowin · 2016-09-18 15:15:38

caj2arch wrote:

Just had the Windows 7-64 guest disappear as I was logging off my non-admin user account. Confirmed that mouse integration is set to OFF. VBox Manager shows guest status as Aborted. Upon restart, get "Windows did not shut down normally". After starting Windows normally, attempted to log in with admin account and guest disappeared again after showing wallpaper and outlines of a few windows. GRRR!!
Will start a new thread for this issue.

Yes, please do start a new topic. As you said hopefully it will attract more readers and potentially those who maintain VirtualBox.
I myself tried to downgrade but nothing seemed to work. I got an error everytime I would try to start a VM and I couldn't get the right guest additions.
It crashes (aborts) daily between 5 and 10 times for me.

kwash · 2016-09-21 00:19:50

I can confirm that the wrong mouse cursor issue is still present in 5.1.6 from repository.

I've installed virtualbox-bin from the AUR and the issue is gone! Tested with Windows 7 64-bit and Windows 10 64-bit guests.

This virtualbox-bin AUR package is essentially the binary installation file from Oracle webiste. No compilation is made by makepkg, so the package building process is lightweight, besides the 80+ megabyte source download.

So maybe it's something related to the repository version of VirtualBox, since the issue is not present in the Oracle binary one.

Of course we would like to see a fixed repository version, but I'll stick with virtualbox-bin for a while.

caj2arch · 2016-09-21 00:33:50

I just posted a recap of the spontaneous crash situation on https://bbs.archlinux.org/viewtopic.php?id=217359 if anyone cares to review and add comments.

kwash wrote:

I've installed virtualbox-bin from the AUR and the issue is gone! Tested with Windows 7 64-bit and Windows 10 64-bit guests.

Any idea what the difference is between the repository and AUR versions?

hb860 · 2016-09-21 12:50:31

I only hope that Oracle provided sources which can be compiled properly.
It is a good idea to recompile it ourselve and see if the repo version is somehow built incorrectly.

Terrowin · 2016-09-21 22:17:46

kwash wrote:

I've installed virtualbox-bin from the AUR and the issue is gone! Tested with Windows 7 64-bit and Windows 10 64-bit guests.
This virtualbox-bin AUR package is essentially the binary installation file from Oracle webiste. No compilation is made by makepkg, so the package building process is lightweight, besides the 80+ megabyte source download.
So maybe it's something related to the repository version of VirtualBox, since the issue is not present in the Oracle binary one.
Of course we would like to see a fixed repository version, but I'll stick with virtualbox-bin for a while.

Thanks for the tip kwash, I've installed this version too and everything seems alright now on my end.

@caj2arch thanks for starting the new topic about VirtualBox crashing, looks like you covered all the important points from my perspective as well.

caj2arch · 2016-09-26 00:09:51

I installed virtualbox-bin from the AUR and had 2 good sessions before the Windows guest crashed earlier this evening. For those of you who succeeded, could you post your experience on https://bbs.archlinux.org/viewtopic.php?id=217359 so we're all together? Thanks.

Last edited by caj2arch (2016-09-26 00:13:12)

hb860 · 2016-10-23 22:58:07

VirtualBox 5.1.8
The issue is still not fixed in my setup.
Did the clean install of Windows 10 b14951 in this version, the cursor still looks weird.
Screenshot: http://i.imgur.com/Es51OkH.png

Arch Linux

#26 2016-09-13 21:28:13

Re: Some VirtualBox guests show black cursor

#27 2016-09-14 14:48:06

Re: Some VirtualBox guests show black cursor

#28 2016-09-14 22:53:06

Re: Some VirtualBox guests show black cursor

#29 2016-09-15 01:43:05

Re: Some VirtualBox guests show black cursor

#30 2016-09-15 21:35:21

Re: Some VirtualBox guests show black cursor

#31 2016-09-17 14:22:03

Re: Some VirtualBox guests show black cursor

#32 2016-09-17 20:52:33

Re: Some VirtualBox guests show black cursor

#33 2016-09-18 07:18:06

Re: Some VirtualBox guests show black cursor

#34 2016-09-18 15:15:38

Re: Some VirtualBox guests show black cursor

#35 2016-09-21 00:19:50

Re: Some VirtualBox guests show black cursor

#36 2016-09-21 00:33:50

Re: Some VirtualBox guests show black cursor

#37 2016-09-21 12:50:31

Re: Some VirtualBox guests show black cursor

#38 2016-09-21 22:17:46

Re: Some VirtualBox guests show black cursor

#39 2016-09-26 00:09:51

Re: Some VirtualBox guests show black cursor

#40 2016-10-23 22:58:07

Re: Some VirtualBox guests show black cursor

Board footer