VirtualBox

Ticket #11938 (new defect)

Opened 8 years ago

Last modified 5 years ago

Failed to open a session for the virtual machine (3D-related)

Reported by: gkholman Owned by:
Component: other Version: VirtualBox 4.2.16
Keywords: save, restore Cc:
Guest type: Windows Host type: Mac OS X

Description

I did a search and did not see anyone post this for V4.2.16. I am getting the following error for *every* attempt to restore a session saved by the "Save the machine state" action. This is for three different Windows virtual machines, and has been happening since at least V4.2.14. There were no problems in V4.1.18 and I didn't do an update between 4.1.18 and 4.2.14. There never is any error information.

===8<---

Failed to open a session for the virtual machine Windows Temp.

No error info.

Result Code: NS_ERROR_CALL_FAILED (0x800706BE) Component: ProgressProxy Interface: IProgress {c20238e4-3221-4d3f-8891-81ce92d9f913}

===8<---

I did see a post where someone first tried killing orphaned processes before trying a restore, and that did not work for me today.

Attachments

VBoxSVC.log Download (1.4 KB) - added by allenb 7 years ago.
Log for crash on launching VM from saved state (v4.3.0)
VBoxSVC.2.log Download (17.9 KB) - added by a.falcon 7 years ago.
VBoxSVC log
VBox.log Download (44.6 KB) - added by a.falcon 7 years ago.
VBoxSVC-bug-report.log Download (5.3 KB) - added by halfer 7 years ago.
VBox log relating to my crashing saved VM issue
VBox-Mint-halfer.log Download (63.7 KB) - added by halfer 7 years ago.
The VM log for a crashing Mint guest
VBox-Mint-halfer.2.log Download (63.7 KB) - added by halfer 7 years ago.
The VM log for a crashing Mint guest
VBox_Error_Restoring.log Download (61.4 KB) - added by Krakel 7 years ago.
Crash when restoring saved state; Ubuntu 32 Guest and 3D Acceleration ON.

Change History

comment:1 Changed 8 years ago by frank

Let me check if I understood you correctly: You are able to start normal VMs if they are not restored from a saved state, is that right? And if you try to start a VM from a saved state, you get the above error and no VBox.log file is created for such a VM session?

A VBoxSVC.log file should be created anyway. You will find it in ~/Library/VirtualBox/VBoxSVC.log.*

Please attach a VBoxSVC log file after you tried to start a VM from a saved state. And if there is a VBox.log file created for such a VM session, attach it as well.

comment:2 Changed 7 years ago by allenb

I am still seeing this same problem on v4.3.0. Launching a VM from a powered-down state works fine. Closing a VM with the "save machine state" option works fine. Launching a VM from a saved state results in a crash once the progress bar gets to about 98%. I am attaching my VBoxSVC.log file which includes the error details. The relevant line appears to be:

ERROR [COM]: aRC=VBOX_E_INVALID_VM_STATE (0x80bb0002) aIID={480cf695-2d8d-4256-9c7c-cce4184fa048} aComponent={Machine} aText={Machine is not locked for session (session state: Unlocked)}, preserve=false

Changed 7 years ago by allenb

Log for crash on launching VM from saved state (v4.3.0)

comment:3 Changed 7 years ago by a.falcon

I have same problem. Running virtual box on

ubuntu12:~$ uname -a Linux ubuntu12 3.8.0-33-generic #48~precise1-Ubuntu SMP Thu Oct 24 16:28:06 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux

@ubuntu12:~$ lsb_release -a No LSB modules are available. Distributor ID: Ubuntu Description: Ubuntu 12.04.3 LTS Release: 12.04 Codename: precise

Virtualbox 4.3.2.r90405

I have Win XP SP3 guest on virtual box, if I take a snapshot while the guest is shutdown I can resume it, but if I take a snapshot from running guest I can't resume.

I get the error message Waiting for VM "XP_SP3" to power on... VBoxManage: error: No error info VBoxManage: error: Details: code NS_ERROR_CALL_FAILED (0x800706be), component ProgressProxy, interface IProgress

I am attaching the VBoxSVC log

Changed 7 years ago by a.falcon

VBoxSVC log

Changed 7 years ago by a.falcon

comment:4 Changed 7 years ago by a.falcon

any updates

comment:5 Changed 7 years ago by frank

Could you test if this problem is still relevant with VBox 4.3.6?

comment:6 Changed 7 years ago by halfer

I am getting the same error for quite a different configuration. Here's the error, to show it is the same:

Failed to open a session for the virtual machine Development.

No error info.

Result Code: NS_ERROR_CALL_FAILED (0x800706BE) Component: ProgressProxy Interface: IProgress {c20238e4-3221-4d3f-8891-81ce92d9f913}

I'm on the latest version, 4.3.8. My host and guest are both Ubuntu 13.10, and I am using the purpose-made file direct from the VB site (virtualbox-4.3_4.3.8-92456~Ubuntu~raring_amd64.deb).

This only pops up for starting a saved machine, and the machine does not successfully start (it is marked as Aborted in the GUI window). Starting a machine from cold is fine, as per the OP's issue.

I had some errors pertaining to my user account (on the host) being unable to access the USB system, for which the solution was to add my user to the "vboxusers" group. This has made the log file much less noisy, but has not affected the problem, so I assume it is irrelevant.

I will try to attach my log file for analysis. If you need any further info, please let me know and I will try to help.

Changed 7 years ago by halfer

VBox log relating to my crashing saved VM issue

comment:7 Changed 7 years ago by halfer

I can't seem to do it, but if someone would remove "V4.2.16 - " from the title, that would be great. Since this problem is exactly duplicated with 4.3.x, this information is probably misleading.

comment:8 Changed 7 years ago by frank

  • Summary changed from V4.2.16 - Failed to open a session for the virtual machine to Failed to open a session for the virtual machine

comment:9 Changed 7 years ago by frank

halfer, first of all, which machine is it which does not start (VM name)? Also it looks to me like the VM you are trying to start aborts. In that case I would be interested to see a core dump. Please do also attach the latest VBox.log file of that VM to this ticket.

comment:10 Changed 7 years ago by halfer

Thanks Frank.

The machine in that case is called "Development", although I can replicate it reliably with a Mint guest called "Linux Mint". I simply start a guest, close and save it, then start it. It appears to be starting, and shows an image on screen of what the guest looked like when I saved it, but then gets to 100% and then aborts. The guest window closes and I am left the above error in a dialogue box.

I'll attach a VBox log for the Mint machine here, and will submit the crash file via FTP as per the instructions in the link.

Changed 7 years ago by halfer

The VM log for a crashing Mint guest

Changed 7 years ago by halfer

The VM log for a crashing Mint guest

comment:11 Changed 7 years ago by halfer

(Apologies, I seem to have uploaded the log file twice, due to a connection error; feel free to delete that, and this message).

comment:12 Changed 7 years ago by halfer

I had trouble sending the crash log, so have sent it via email. Let me know if you didn't receive it!

comment:13 Changed 7 years ago by frank

Thanks halfer for the core dump!

Looking at it your problem seems to be very easy: The /dev/vboxdrv device isn't accessible or does not exist. I assume that you have to recompile your kernel modules for your current host. This problem should be shown as a GUI error message. Why this doesn't work and the GUI crashes on your system is another question.

comment:14 Changed 7 years ago by halfer

Seems okay to me on the host:

jon@xps:~$ ls -l /dev/vboxdrv
crw------- 1 root root 10, 56 Mar  9 16:42 /dev/vboxdrv

If your theory was correct though, would guests also not start up from cold?

Just a thought - are the permissions and owners okay on that device? I've not changed them, but wonder if they might be askew.

I assume that you have to recompile your kernel modules for your current host

Do you mean with sudo /etc/init.d/vboxdrv setup? I've needed to do that every time the host gets certain updates (I imagine they are new minor versions of the kernel), and I get the GUI warning fine for this.

comment:15 Changed 7 years ago by frank

The file permissions root.root are correct. Yes, I mean '/etc/init.d/vboxdrv setup' but as the device is already there it looks like this is not necessary. Did you install VirtualBox just normally as .deb or did you do any local modification?

comment:16 Changed 7 years ago by halfer

I believe I installed the stock Ubuntu VirtualBox, but I decided it would be good to run the very latest version (possibly because of the crashing issue, though I don't now recall why I did so).

So, I uninstalled VirtualBox in the Software Centre and installed latest VB via the .deb package. I didn't do anything usual as far as I recall. I wonder, could the earlier version have left files behind that have caused a problem?

Last edited 7 years ago by halfer (previous) (diff)

comment:17 Changed 7 years ago by frank

If you always use the .deb package then this should work but make sure that the old packages are really purged (not only removed) from your system. Which package did you install, the most recent on from www.virtualbox.org? What package name?

comment:18 Changed 7 years ago by halfer

As per note 6, I installed "virtualbox-4.3_4.3.8-92456~Ubuntu~raring_amd64.deb".

I've looked into my install history on the host, and this is what I removed prior to installing the deb file:

  • virtualbox (4.2.16-dfsg-3)
  • dkms (2.2.0.3-1.1ubuntu4)
  • virtualbox-dkms (4.2.16-dfsg-3)
  • virtualbox-qt (4.2.16-dfsg-3)

I now forget what dkms does, but I recall that I could not proceed with some part of VB (I think possibly enabling Shared Folders) until it was removed. I think the vboxdrv setup failed a step (or failed completely) until this was removed. To remove it, I simply uninstalled it in the Ubuntu 13.10 host GUI - I don't know if this purges it too. However, the failing step appeared to be happy after I did this.

Since my normal working environment will be entirely in a guest, it will be fairly trivial for me to back up my installers on the host and my VMs, and start from a fresh install, to see if that fixes it. However, I'd be of the view that that is not the most satisfying fix: I think that if there will be a problem saving machine state or restoring it, VirtualBox should refuse to do so and explain clearly what the problem is.

Is there anything more I can try to help pin the problem down? Or does the above help recreate the problem on your machine?

Last edited 7 years ago by halfer (previous) (diff)

comment:19 follow-up: ↓ 20 Changed 7 years ago by halfer

(To clarify, I should think I just uninstalled virtualbox above, and the other three things were dependencies that were removed by the Software Centre automatically).

comment:20 in reply to: ↑ 19 Changed 7 years ago by frank

Replying to halfer:

(To clarify, I should think I just uninstalled virtualbox above, and the other three things were dependencies that were removed by the Software Centre automatically).

The software center does normally only remove the packages but not purge them. Please try

aptitude purge ~c

and see if it helps. I just want to be sure.

comment:21 Changed 7 years ago by halfer

Not a sausage, I'm afraid. I don't have aptitude installed, so did this with apt-get:

jon@xps13:~$ sudo apt-get purge
Reading package lists... Done
Building dependency tree       
Reading state information... Done
The following package was automatically installed and is no longer required:
  kde-l10n-engb
Use 'apt-get autoremove' to remove it.
0 upgraded, 0 newly installed, 0 to remove and 0 not upgraded.
jon@xps13:~$ sudo apt-get autoremove
Reading package lists... Done
Building dependency tree       
Reading state information... Done
The following packages will be REMOVED
  kde-l10n-engb
0 upgraded, 0 newly installed, 1 to remove and 0 not upgraded.
After this operation, 9,990 kB disk space will be freed.
Do you want to continue [Y/n]? y
(Reading database ... 269890 files and directories currently installed.)
Removing kde-l10n-engb ...

A restore of a saved machine gives the same results again, even after a reboot (I am assuming the save is fine and the restore is problematic, but I guess the save could be corrupted). Looking at the above console output, it didn't mention the items I listed in note 18, so I expect they were purged in some other way. I didn't do so manually.

If you get a clean Ubuntu install, install stock VB, uninstall without purging, then install the latest deb binary, are you able to replicate?

Of coure, it is possible I have an unreproducible problem, in which case I might uninstall VB via the Software Centre, remove ~/.VirtualBox plus any other manual cleaning you recommend, and then reinstall, to see if that helps.

comment:22 Changed 7 years ago by halfer

(By the way, thanks for your continuing advice! Much appreciated.)

comment:23 Changed 7 years ago by halfer

FWIW, 4.3.10 didn't fix this. I've also found some instability when running two machines simultaneously (2G and 4G in a 8G host) so something is definitely awry with my install. That said, if I just stick to the one 4G installation, and don't try to save its state, it really is rock solid.

So, I'll back up my machines and reinstall the host to see if this fixes things. If you want to close as not reproducible, that's fine with me.

I may not now see updates to this ticket, so if anyone is able to turn on comment notifications for me, please do - I don't know that I can do that myself.

comment:24 Changed 7 years ago by halfer

Aha! A possible clash of versions was a false lead. I've now narrowed this down a bit.

I backed up my VMs and a few things on the host, and installed the latest Ubuntu LTS, 14.04 as host. I then brought all my VMs back and installed the latest version of VirtualBox (virtualbox-4.3_4.3.10-93012~Ubuntu~raring_amd64.deb - this is for Raring as there is not yet a 14.04 build available). I copied my ~/.VirtualBox folder back into place too.

The very same bug was noticed on my old Ubuntu 13.10 and Mint 16 guests: saving machine state appears to work fine, but restore crashes every time. However I now notice that an incomplete install of Arch does not suffer from the same problem.

It turns out that I'd not switched on 3D Acceleration in the Display settings for Arch, so I tried turning this off for the Mint guest, and that fixed the problem immediately. I then tried the same for the 13.10 guest, and guess what? Fixed!

So, my advice to anyone suffering from this problem is to turn off 3D Acceleration in the settings and see if that fixes it. I've turned mine back on, as I just end up sleeping the host in practice anyway.

Frank, I suspect there should be information about my graphics subsystem in previous system files and dumps, but if you need further info, please let me know.

comment:25 Changed 7 years ago by chrisb

I see this bug with the exact same NS_ERROR_CALL_FAILED window when resuming snapshots of Ubuntu 14.04 on a Debian (jessie) host with virtualbox-4.3.10. It only happens if 3D Acceleration is enabled for the host (as stated by user halfer above), and the host is logged in to Unity (trying to resume a snapshot of the login screen works, resuming a snapshot of the desktop does not). Backtrace:

Core was generated by `/usr/lib/virtualbox/VirtualBox -startvm trusty'.
Program terminated with signal 11, Segmentation fault.

#0  draw_depth_stencil_pixels (pixels=0x7f7622977010, unpack=0x7f767c80e558, type=34042, height=954, width=989, y=0, x=0, ctx=0x7f767c7f23e8)
    at ../../../../src/mesa/swrast/s_drawpix.c:575
#1  _swrast_DrawPixels (ctx=ctx@entry=0x7f767c7f23e8, x=x@entry=0, y=y@entry=0, width=width@entry=989, height=height@entry=954, 
    format=format@entry=34041, type=type@entry=34042, unpack=unpack@entry=0x7f767c80e558, pixels=pixels@entry=0x7f7622977010)
    at ../../../../src/mesa/swrast/s_drawpix.c:697
#2  0x00007f76838416f7 in _mesa_meta_DrawPixels (ctx=ctx@entry=0x7f767c7f23e8, x=x@entry=0, y=y@entry=0, width=width@entry=989, 
    height=height@entry=954, format=format@entry=34041, type=type@entry=34042, unpack=unpack@entry=0x7f767c80e558, pixels=pixels@entry=0x7f7622977010)
    at ../../../../src/mesa/drivers/common/meta.c:2789
#3  0x00007f768392e2cc in intelDrawPixels (ctx=0x7f767c7f23e8, x=0, y=0, width=989, height=954, format=34041, type=34042, unpack=0x7f767c80e558, 
    pixels=0x7f7622977010) at ../../../../../../../src/mesa/drivers/dri/i965/intel_pixel_draw.c:171
#4  0x00007f76836f6a00 in _mesa_DrawPixels (width=989, height=954, format=34041, type=34042, pixels=0x7f7622977010)
    at ../../../../src/mesa/main/drawpix.c:162
#5  0x00007f76a8772cc6 in crStateApplyFBImage (to=to@entry=0x7f767cf3daa0, data=data@entry=0x7f76a8abd9e0)
    at /tmp/buildd/virtualbox-4.3.10-dfsg/src/VBox/GuestHost/OpenGL/state_tracker/state_diff.c:351
#6  0x00007f76a873e0b2 in crVBoxServerLoadFBImage (pSSM=pSSM@entry=0x7f76a955ab90, version=version@entry=43, 
    pContextInfo=pContextInfo@entry=0x7f767c5577a0, pMural=pMural@entry=0x7f767d02b2a0)
    at /tmp/buildd/virtualbox-4.3.10-dfsg/src/VBox/HostServices/SharedOpenGL/crserverlib/server_main.c:2263
#7  0x00007f76a874197d in crVBoxServerLoadState (pSSM=0x7f76a955ab90, version=43)
    at /tmp/buildd/virtualbox-4.3.10-dfsg/src/VBox/HostServices/SharedOpenGL/crserverlib/server_main.c:2446
#8  0x00007f76a873b187 in svcLoadState (u32ClientID=<optimized out>, pvClient=<optimized out>, pSSM=0x7f76a955ab90)
    at /tmp/buildd/virtualbox-4.3.10-dfsg/src/VBox/HostServices/SharedOpenGL/crserver/crservice.cpp:364
#9  0x00007f76aa35c969 in hgcmServiceThread (ThreadHandle=2147483665, pvUser=0x7f7690002730)
    at /tmp/buildd/virtualbox-4.3.10-dfsg/src/VBox/Main/src-client/HGCM.cpp:639
#10 0x00007f76aa35b0c2 in hgcmWorkerThreadFunc (ThreadSelf=<optimized out>, pvUser=0x7f7690002810)
    at /tmp/buildd/virtualbox-4.3.10-dfsg/src/VBox/Main/src-client/HGCMThread.cpp:194
#11 0x00007f76e1269fdc in rtThreadMain (pThread=0x7f76900029d0, NativeThread=<optimized out>, pszThreadName=<optimized out>)
    at /tmp/buildd/virtualbox-4.3.10-dfsg/src/VBox/Runtime/common/misc/thread.cpp:712
#12 0x00007f76e12ccb3c in rtThreadNativeMain (pvArgs=0x7f76900029d0)
    at /tmp/buildd/virtualbox-4.3.10-dfsg/src/VBox/Runtime/r3/posix/thread-posix.cpp:324
#13 0x00007f76e1b1d062 in start_thread (arg=0x7f76a8abf700) at pthread_create.c:312
#14 0x00007f76e164cbfd in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:111

draw_depth_stencil_pixels is in  ./src/mesa/swrast/s_drawpix.c. The offending code at line 575 is a dereference of depthRb which is 0x0. depthRb is set as so:

depthRb = ctx->ReadBuffer->Attachment[BUFFER_DEPTH].Renderbuffer;
(gdb) p ctx->ReadBuffer->Attachment[BUFFER_DEPTH]
$16 = {Type = 0, Complete = 0 '\000', Renderbuffer = 0x0, Texture = 0x0, TextureLevel = 0, CubeMapFace = 0, Zoffset = 0, Layered = 0 '\000'}

So it looks like this is a bug in either VirtualBox or the Intel MESA drivers.

The segfault appears in /var/log/kern.log as:

ShCrOpenGL[5296]: segfault at 54 ip 00007fb2e54d5a30 sp 00007fb2e677c550 error 4 in i965_dri.so[7fb2e535d000+4ed000]}}}

Last edited 7 years ago by chrisb (previous) (diff)

Changed 7 years ago by Krakel

Crash when restoring saved state; Ubuntu 32 Guest and 3D Acceleration ON.

comment:26 Changed 7 years ago by Krakel

I see the same error reproduceably with:

  • VirtualBox 4.3.10 on Debian AMD64
  • Guest is Ubuntu 32 Bit
  • 3D acceleration is enabled

var/log/kern.log reads the same:

May 20 14:14:42 Thur kernel: [  300.677348] ShCrOpenGL[3525]: segfault at 54 ip 00007f615673d5d0 sp 00007f61579e4550 error 4 in i965_dri.so[7f61565c4000+4ee000]

VirtualBox Log file is attached (VBox_Error_Restoring.log)

Three other VMs on the same system work well, but these have 3D Acceleration off.

comment:27 Changed 7 years ago by chrisb

Bug still present in VirtualBox 4.3.14 on Debian Jessie.

comment:29 Changed 7 years ago by frank

  • Summary changed from Failed to open a session for the virtual machine to Failed to open a session for the virtual machine (3D-related)

comment:30 Changed 6 years ago by halfer

I think @chrisb is on to something. After months of putting it off, I downloaded the latest BIOS update for my Dell machine and re-flashed using FreeDOS. I can now run machines in parallel, and restore a saved machine, without any crash at all. The 3D accel flag is enabled. I presume the buggy graphics drivers have been fixed in the newer BIOS image.

I'll update this ticket if I get any further problems.

comment:31 Changed 5 years ago by halfer

My assessment above that I can run VMs in parallel was premature - in fact, I cannot.

I have now run two BIOS updates from Dell; the first one helped somewhat, and the second has made no further difference. I am definitively able to save and restore VMs - I do this often, and the old crash has not once reared its head. So, the issue does look like the Dell BIOS bug.

However, running two machines in parallel results in extremely sluggish mouse pointer movement before the host machine locks up completely. It crashes so substantially that even the LED fails to toggle on the Caps Lock key, and I have to do a hard power-off.

I am using 5.0.2 so I will try to get the time to upgrade to latest (5.0.8 at time of writing). However since this has been a long-standing problem (possibly not of VirtualBox's making) I doubt that it will fix it.

I acknowledge that it is possible that the behaviour I describe is not related to the above. I will see if I can also find the time to replicate the crash (quite easy) and see if there is anything in the log files (it could be a different error, I suppose).

Last edited 5 years ago by halfer (previous) (diff)

comment:32 Changed 5 years ago by halfer

I'm most interested to see that the Guest Additions 5.0.17 (non-stable) release has a fix relating to 3D acceleration problems on Debian systems - sounds like that will fix the problem above.

Unfortunately I prefer not to download it without checksums to ensure it has been downloaded without corruption. I have asked in the "Suggestions" area of the forum for checksums for test releases, to no avail so far - if anyone triaging the bugs queue can provide them here, I will be very happy to test.

comment:33 Changed 5 years ago by halfer

Following on from above, I didn't get to try 5.0.17 but 5.0.18 exhibited some new issues (e.g. #15324) that meant I had to roll back. I am now on 5.0.20 and it seems good presently. I will therefore try some machine saves/restores and some parallel machines to see if the 3D issue (which I believe is responsible for some crashing behaviour) is now resolved.

Note: See TracTickets for help on using tickets.

www.oracle.com
ContactPrivacy policyTerms of Use