In VirtualBox 5 I cannot enable second monitor

[bmn]

I set in preferences that I want to have 2 displays. However, in Linux I see only one. When I right click on the screen icon in the status bar, I see "Second virtual display" and all the options are greyed out except for "Enable". Unfortunately, nothing happens when I click on "Enable".

Linux says:

$> xrandr 
Screen 0: minimum 64 x 64, current 1920 x 976, maximum 32766 x 32766
VGA-0 connected primary 1920x976+0+0 0mm x 0mm
   1920x976      60.00*+
   2560x1600     60.00  
   2560x1440     60.00  
   2048x1536     60.00  
   1920x1600     60.00  
   1920x1080     60.00  
   1600x1200     60.00  
   1680x1050     60.00  
   1400x1050     60.00  
   1280x1024     60.00  
   1024x768      60.00  
   800x600       60.00  
   640x480       60.00

$> systemctl status vboxadd.service 
vboxadd.service - LSB: VirtualBox Linux Additions kernel modules
   Loaded: loaded (/etc/init.d/vboxadd)
   Active: active (exited) since k 2015-08-25 16:12:09 CEST; 12min ago
  Process: 7224 ExecStop=/etc/init.d/vboxadd stop (code=exited, status=1/FAILURE)
  Process: 7242 ExecStart=/etc/init.d/vboxadd start (code=exited, status=0/SUCCESS)

aug 25 16:12:09 linux-suse.site vboxadd[7224]: Stopping VirtualBox Additions ..failed
aug 25 16:12:09 linux-suse.site vboxadd[7224]: (Cannot unload module vboxguest)
aug 25 16:12:09 linux-suse.site vboxadd[7242]: Starting the VirtualBox Guest Additions ..done

Host: Win7 x64
Guest: openSUSE 13.2 with kernel 3.16

[bmn]

Virtualbox.log

[Michael Thayer]

I'm sorry, but I really can't reproduce this. Can you give a reproduction scenario with a new VM?

[GMC]

Did some substantial diagnosis of this problem. It appears that the VBoxVideo driver version 5.0.4_OSEr102546 has this issue; the following was observed:

Guest	Driver Version	Observed
Arch Linux	5.0.4_OSEr102546	No second screen, issues in seamless mode
(same)	5.0.4r102546	No issues
Ubuntu 15.04	4.3.26_Ubuntu	No issues
(same)	5.0.4r102546	No issues

Workaround therefore is to not use the distribution provided guest additions, but remove them and install from the ISO image provided by VirtualBox itself.

[Michael Thayer]

Would you mind investigating a little bit more? For a start, in the log file you provided I saw lines like "Restoring guest size-hint for screen 1 to 0x0" which looked rather strange to me, and where I do not immediately see the connection to the Additions version. Could you check the log files for the machines you tested to see if there is a correlation with the second screen issue?

[Michael Thayer]

The output of xrandr in the guest when you the screen is not available would be interesting too.

[Michael Thayer]

And /var/log/Xorg.0.log from the guests.

[bmn]

Version 5.0.4 is affected, too. I found out that many other users are experiencing this bug.

Some more information can be found on public forums:

• ​https://forums.virtualbox.org/viewtopic.php?f=6&t=69095
• ​https://bbs.archlinux.org/viewtopic.php?id=200217
• ​https://bugs.archlinux.org/task/45748?project=5

xrandr output:

xrandr program version       1.4.3
Server reports RandR version 1.4

Screen 0: minimum 64 x 64, current 1920 x 976, maximum 32766 x 32766
VGA-0 connected primary 1920x976+0+0 (0x1e2) normal (normal) 0mm x 0mm
        Identifier: 0x3f
        Timestamp:  17535
        Subpixel:   unknown
        Gamma:      1.0:1.0:1.0
        Brightness: 1.0
        Clones:    
        CRTC:       0
        CRTCs:      0
        Transform:  1.000000 0.000000 0.000000
                    0.000000 1.000000 0.000000
                    0.000000 0.000000 1.000000
                   filter: 
  1920x976 (0x1e2) 113.479MHz *current +preferred
        h: width  1920 start 1922 end 1924 total 1926 skew    0 clock  58.92KHz
        v: height  976 start  978 end  980 total  982           clock  60.00Hz
  2560x1600 (0x41) 247.259MHz
        h: width  2560 start 2562 end 2564 total 2566 skew    0 clock  96.36KHz
        v: height 1600 start 1602 end 1604 total 1606           clock  60.00Hz
  2560x1440 (0x42) 222.626MHz
        h: width  2560 start 2562 end 2564 total 2566 skew    0 clock  86.76KHz
        v: height 1440 start 1442 end 1444 total 1446           clock  60.00Hz
  2048x1536 (0x43) 190.036MHz
        h: width  2048 start 2050 end 2052 total 2054 skew    0 clock  92.52KHz
        v: height 1536 start 1538 end 1540 total 1542           clock  60.00Hz
  1920x1600 (0x44) 185.589MHz
        h: width  1920 start 1922 end 1924 total 1926 skew    0 clock  96.36KHz
        v: height 1600 start 1602 end 1604 total 1606           clock  60.00Hz
  1920x1080 (0x45) 125.498MHz
        h: width  1920 start 1922 end 1924 total 1926 skew    0 clock  65.16KHz
        v: height 1080 start 1082 end 1084 total 1086           clock  60.00Hz
  1600x1200 (0x46) 116.210MHz
        h: width  1600 start 1602 end 1604 total 1606 skew    0 clock  72.36KHz
        v: height 1200 start 1202 end 1204 total 1206           clock  60.00Hz
  1680x1050 (0x47) 106.824MHz
        h: width  1680 start 1682 end 1684 total 1686 skew    0 clock  63.36KHz
        v: height 1050 start 1052 end 1054 total 1056           clock  60.00Hz
  1400x1050 (0x48) 89.084MHz
        h: width  1400 start 1402 end 1404 total 1406 skew    0 clock  63.36KHz
        v: height 1050 start 1052 end 1054 total 1056           clock  60.00Hz
  1280x1024 (0x49) 79.474MHz
        h: width  1280 start 1282 end 1284 total 1286 skew    0 clock  61.80KHz
        v: height 1024 start 1026 end 1028 total 1030           clock  60.00Hz
  1024x768 (0x4a) 47.833MHz
        h: width  1024 start 1026 end 1028 total 1030 skew    0 clock  46.44KHz
        v: height  768 start  770 end  772 total  774           clock  60.00Hz
  800x600 (0x40) 29.306MHz
        h: width   800 start  802 end  804 total  806 skew    0 clock  36.36KHz
        v: height  600 start  602 end  604 total  606           clock  60.00Hz
  640x480 (0x4b) 18.837MHz
        h: width   640 start  642 end  644 total  646 skew    0 clock  29.16KHz
        v: height  480 start  482 end  484 total  486           clock  60.00Hz

[bmn]

The following does not help:

• turning on/off the 3D acceleration
• changing monitor numbers from 2 -> 1, then 1 -> 2 (or play this with 3)
  • either from GUI
  • or from VBoxManage
• removing any of the following lines from <machinename>.vbox file:

  <GuestProperty name="/VirtualBox/GuestAdd/Vbgl/Video/0" value="1920x976x32,0x0,1" timestamp="1445263857226959000" flags=""/>
  <GuestProperty name="/VirtualBox/GuestAdd/Vbgl/Video/1" value="800x600x32,0x0,1" timestamp="1445256421803528100" flags=""/>
  <GuestProperty name="/VirtualBox/GuestAdd/Vbgl/Video/2" value="800x600x32,0x0,1" timestamp="1445256426396946600" flags=""/>
  <GuestProperty name="/VirtualBox/GuestAdd/Vbgl/Video/SavedMode" value="1920x976x32" timestamp="1445263857227459100" flags=""/>
  

[Fab123]

Virtualbox 5.0.10 is affected, too.

[Benjamin ROBIN]

I can reproduce this bug, and I think I found the root cause. I did create a quick and dirty patch. Everything is explained here : ​https://bugs.archlinux.org/task/45748

Basically I did add memory barrier inside the VBoxHGSMIBufferSubmit() near VBoxVideoCmnPortWriteUlong()

To properly fix it, I think the pointer returned by VBoxHGSMIBufferAlloc() shall be volatile

[sunlover]

benjarobin, thanks a lot for spotting this! The fix will be included in the next VirtualBox release.

[Frank Mehnert]

The actual changeset is r59179. As you already found out, the I/O instruction for submitting HGSMI commands has side effects, therefore the memory clobber will prevent gcc from doing dead-code elemination.

[Frank Mehnert]

Fix is part of VBox 5.0.12.