suspend to ram fails wenn the vbox modul is loaded

Postings relating to old VirtualBox pre-releases
CME
Posts: 49
Joined: 28. Aug 2009, 00:24
Primary OS: Debian other
VBox Version: PUEL
Guest OSses: windows 98se, 2k, xp, 7, solaris, opensolaris, reactos, ubuntu, debian,...
Contact:

suspend to ram fails wenn the vbox modul is loaded

Post by CME »

Hi,

since beta1 suspend to ram stopped working on my netbook (eeepc 1000he) on a debian sid host, kernel 2.6.39, when the vbox modul is loaded (no vm running), when i try to suspend it, the machine will lockup with a cursor displayed in the upper left corner, when i unload the modules, s2r works fine.
pico_c_nickel
Posts: 22
Joined: 22. Jan 2009, 19:16

Re: suspend to ram fails wenn the vbox modul is loaded

Post by pico_c_nickel »

I'm suffering from the same issue. Additionaly machine doesn't switch off after init 0, I have to do it manually with the power switch. Instead a kernel panic appears regarding vboxdrv (sorry, don't have time to post exact error message).
ingo2
Posts: 192
Joined: 28. Feb 2008, 14:15
Primary OS: Debian other
VBox Version: OSE Debian
Guest OSses: OS/2, WinXP-10, Linux
Location: Germany

Re: suspend to ram fails wenn the vbox modul is loaded

Post by ingo2 »

I think that the s2ram problems have silently started with VBox 4.0.6. I did try 4.0.6, 4.0.8, 4.0.10 and all three result in distorted guest after several s2ram cycles of the host. I am back on 4.0.4 where all is perfect and I can suspend and resume as often as I want while guest is up and running (20 -40 times have been tested at least). Host is Debian-Squeeze-amd64.

With best regards,
Ingo

Right now I was able to catch such a situation where the guest (OS/2 Warp4.5) completely froze shortly after resume from s2ram in VBox 4.0.10.

I do report it here, because I doubt it will be fixed in 4.0.x branch.

Setup was done on July 2nd with:
* Upgrading VBox from 4.0.4 on Lucid-amd64 host
* Using a known good HD-image attached as "immutable" to avoid any corruption.
(this image has proven reliable for many months since VBox 3.x withstanding hundreds of s2ram)
* Now in VBox 4.0.10 I first took a snapshot in saved state of the VM just after booting up.
* Then I did nothing but play the Klondike game (which requires continous mouse clicks to play).
* I performed several s2ram+resume on July 2nd and 3rd and finally terminated the VM by "saving machine state" in running condition.

* Today, after the 1st s2ram the VM beheaved normally after resume, but about 10 minutes later (playing Klondike) it completely froze:

CPU-load on the host on the one core the VM is using has rose to 100% ('top' indicates VirtualBox is causing this)
CPU-load in the guest is at 100% as well
VBox-window shows continous read from the harddisk (green dot always on)
Host shows zero disk activity.

The VM stayed unresponsive for some 10 minutes, then I decided to shut it down, "saving" its state - this went smoothly without any problems.

In this situation (VM saved) and VBox terminated I packed all 4 log-files of that VM which start from VBox-upgrade to 4.0.10 into a tar archive and attach it here.
This VM is now unusable, I have to return either to the good snapshot taken at the beginning or revert to the immutable HD-image.

Hope this helps to find the root cause and fix it in 4.1.0,
Ingo
Attachments
Logs.tar.gz
4 log files from VBox 4.0.10 from beginning to corruption of the VM after s2ram
(43.28 KiB) Downloaded 22 times
frank
Oracle Corporation
Posts: 3362
Joined: 7. Jun 2007, 09:11
Primary OS: Debian Sid
VBox Version: PUEL
Guest OSses: Linux, Windows
Location: Dresden, Germany
Contact:

Re: suspend to ram fails wenn the vbox modul is loaded

Post by frank »

Did you try to further isolate the problem? I think one of the following modules must be responsible for the problem:
  • the vboxdrv module (VBox support module)
  • the vboxnetflt module (used for bridged networking and host-only networking)
  • the VMM itself
To reproduce the problem, is it sufficient that the VBox modules are loaded or is it necessary that one VM has to be executed at least once? The original reporter said that the VBox modules must be loaded but ingo2 experienced distorted guests after suspending/resuming the host. ingo2, can you reproduce the problem if you use NAT instead of bridged networking?
ingo2
Posts: 192
Joined: 28. Feb 2008, 14:15
Primary OS: Debian other
VBox Version: OSE Debian
Guest OSses: OS/2, WinXP-10, Linux
Location: Germany

Re: suspend to ram fails wenn the vbox modul is loaded

Post by ingo2 »

Frank Mehnert wrote: .... but ingo2 experienced distorted guests after suspending/resuming the host. ingo2, can you reproduce the problem if you use NAT instead of bridged networking?
I have now set up the the VM to use NAT + DHCP (instead of "bridged" with fix IP).
I Performed several s2ram + resume cycles and did not see any trouble for now. It however will need further tests and I will update the information here latest Thursday.

One question with regard to this:

in former times I had configured VBox to use "bridged" networking, but had configured a bridge with means of the host. Bridge was configured static in /etc/network/interfaces with fixed IP on br0 und no IP assigned to eth0 and the tap-devices. VM was attached to a tap-device. This was required in early VBox-releases and I kept that configuration also for a while in VBox 3.x. Do you think that (in case NAT finally turns out stable) this could also help here? Network module on the host is "forcedeth".

I do not want to change 2 things at the same time to get reliable test results, so I will update results as soon as possible here.

Best regards,
Ingo
CME
Posts: 49
Joined: 28. Aug 2009, 00:24
Primary OS: Debian other
VBox Version: PUEL
Guest OSses: windows 98se, 2k, xp, 7, solaris, opensolaris, reactos, ubuntu, debian,...
Contact:

Re: suspend to ram fails wenn the vbox modul is loaded

Post by CME »

Frank Mehnert wrote:Did you try to further isolate the problem? I think one of the following modules must be responsible for the problem:
  • the vboxdrv module (VBox support module)
  • the vboxnetflt module (used for bridged networking and host-only networking)
  • the VMM itself
To reproduce the problem, is it sufficient that the VBox modules are loaded or is it necessary that one VM has to be executed at least once? The original reporter said that the VBox modules must be loaded but ingo2 experienced distorted guests after suspending/resuming the host. ingo2, can you reproduce the problem if you use NAT instead of bridged networking?
vboxdrv in my case and it's sufficient if the module is loaded.
ingo2
Posts: 192
Joined: 28. Feb 2008, 14:15
Primary OS: Debian other
VBox Version: OSE Debian
Guest OSses: OS/2, WinXP-10, Linux
Location: Germany

Re: suspend to ram fails wenn the vbox modul is loaded

Post by ingo2 »

CME wrote: vboxdrv in my case and it's sufficient if the module is loaded.
As a workaround you could unload the module before s2ram and relod after resume:

Edit /etc/pm/config.d/unload_modules
and put this line

Code: Select all

SUSPEND_MODULES="$SUSPEND_MODULES vboxdrv"
It helps for modules which don't survive s2ram, but I don't know whether this will break some VBox funcitionality.

With best regards,
Ingo
Sasquatch
Volunteer
Posts: 17798
Joined: 17. Mar 2008, 13:41
Primary OS: Debian other
VBox Version: PUEL
Guest OSses: Windows XP, Windows 7, Linux
Location: /dev/random

Re: suspend to ram fails wenn the vbox modul is loaded

Post by Sasquatch »

ingo2 wrote:One question with regard to this:

in former times I had configured VBox to use "bridged" networking, but had configured a bridge with means of the host. Bridge was configured static in /etc/network/interfaces with fixed IP on br0 und no IP assigned to eth0 and the tap-devices. VM was attached to a tap-device. This was required in early VBox-releases and I kept that configuration also for a while in VBox 3.x. Do you think that (in case NAT finally turns out stable) this could also help here? Network module on the host is "forcedeth".

I do not want to change 2 things at the same time to get reliable test results, so I will update results as soon as possible here.

Best regards,
Ingo
That network configuration should have been removed a long time ago. Though it should not cause problems in theory, it is obsolete. Please see if removal of this config adds to your system stability.
ingo2 wrote:
CME wrote: vboxdrv in my case and it's sufficient if the module is loaded.
As a workaround you could unload the module before s2ram and relod after resume:

Edit /etc/pm/config.d/unload_modules
and put this line

Code: Select all

SUSPEND_MODULES="$SUSPEND_MODULES vboxdrv"
It helps for modules which don't survive s2ram, but I don't know whether this will break some VBox funcitionality.

With best regards,
Ingo
That will only cause problems if a VM is running. You should not be able to unload the modules when they are in use. If you forcefully do unload them, the VM will most likely crash, or worse, crash your Host.
Read the Forum Posting Guide before opening a topic.
VirtualBox FAQ: Check this before asking questions.
Online User Manual: A must read if you want to know what we're talking about.
Howto: Install Linux Guest Additions
Howto: Use Shared Folders on Linux Guest
See the Tutorials and FAQ section at the top of the Forum for more guides.
Try searching the forums first with Google and add the site filter for this forum.
E.g. install guest additions site:forums.virtualbox.org

Retired from this Forum since OSSO introduction.
ingo2
Posts: 192
Joined: 28. Feb 2008, 14:15
Primary OS: Debian other
VBox Version: OSE Debian
Guest OSses: OS/2, WinXP-10, Linux
Location: Germany

Re: suspend to ram fails wenn the vbox modul is loaded

Post by ingo2 »

Sasquatch wrote:
ingo2 wrote:One question with regard to this:

in former times I had configured VBox to use "bridged" networking, but had configured a bridge with means of the host. Bridge was configured static in /etc/network/interfaces with fixed IP on br0 und no IP assigned to eth0 and the tap-devices. VM was attached to a tap-device. This was required in early VBox-releases and I kept that configuration also for a while in VBox 3.x. Do you think that (in case NAT finally turns out stable) this could also help here? Network module on the host is "forcedeth".

I do not want to change 2 things at the same time to get reliable test results, so I will update results as soon as possible here.

Best regards,
Ingo
That network configuration should have been removed a long time ago. Though it should not cause problems in theory, it is obsolete. Please see if removal of this config adds to your system stability.
Sasquatch you got me wrong:

Frank recommended to check whether my problem goes away when I change networking in the guest from BRIDGE -> NAT. That I did on a test machine and after many tests today untill now all still looks good!

Currently I am using VBox with built in bridging NIC attached to eth0 of the host. With this configuration the problems arose with VBox 4.0.6.

In former times with VBox 2.x and part through 3.x I was using a bridge on the host and the VBox-NIC attached to tap0 (I also needed more tap-devices for other purposes). With that configuration I never had any problems. Please note: the bridge was configured static with the NIC continously in promiscous mode!

That's why I asked for the possibility to try a bridge on the host with current VBox versions and so gain back old stability.

With best regards,
Ingo
ingo2
Posts: 192
Joined: 28. Feb 2008, 14:15
Primary OS: Debian other
VBox Version: OSE Debian
Guest OSses: OS/2, WinXP-10, Linux
Location: Germany

Re: suspend to ram fails wenn the vbox modul is loaded

Post by ingo2 »

The known problem/crash of VM happened again - with NAT networking. VM is totally corupted after some 20 s2ram + resume.

I had to downgrade to 4.0.4 again. As this seems to persist in 4.x series, I consider even to go back to 3.2.12.
ingo2
Posts: 192
Joined: 28. Feb 2008, 14:15
Primary OS: Debian other
VBox Version: OSE Debian
Guest OSses: OS/2, WinXP-10, Linux
Location: Germany

Re: suspend to ram fails wenn the vbox modul is loaded

Post by ingo2 »

My suspicion still presumes it has to do with USB. Somewhere I read that starting with VBox 4.0.6 there was some (not mentioned in change-log) change in the USB subsystem???

Me just came an idea:
If that's true: probaly the nasty beheaviour with s2ram results from my configuration to unload/reload the module usblp for suspend. This configuration has proven reliable since years and is done because "usblp does not have a suspend/resume script".

If it actually is the USB-system, this might also explain the problems reported by CME - in case he is using i.e. snd-usb-audio module?

Best regards,
Ingo
fixedwheel
Volunteer
Posts: 1699
Joined: 13. Sep 2008, 02:18

Re: suspend to ram fails wenn the vbox modul is loaded

Post by fixedwheel »

ingo2 wrote:If that's true: probaly the nasty beheaviour with s2ram results from my configuration to unload/reload the module usblp for suspend. This configuration has proven reliable since years and is done because "usblp does not have a suspend/resume script".
my Broadcom NetXtreme BCM5755 aka "Tigon" NIC needs unload/reload script for suspend (last time i checked on Debian Lenny, anyway the script still is active). When running a virtual machine attached to VBox "bridged" while suspending the hosts network is dead after resume, and with "/etc/init.d/networking restart" i can re-establish ... (edit: this behaviour with VBox 4.0.x and any previous version, did not run 4.1beta on this host)
fixedwheel
Volunteer
Posts: 1699
Joined: 13. Sep 2008, 02:18

Re: suspend to ram fails wenn the vbox modul is loaded

Post by fixedwheel »

fixedwheel wrote:my Broadcom NetXtreme BCM5755 aka "Tigon" NIC needs unload/reload script for suspend (last time i checked on Debian Lenny, anyway the script still is active).
glad i rechecked: now i'm able to suspend/resume w/o unloading tg3 module, running "bridged" guest while suspending host and everything is up again after resume :mrgreen:
frank
Oracle Corporation
Posts: 3362
Joined: 7. Jun 2007, 09:11
Primary OS: Debian Sid
VBox Version: PUEL
Guest OSses: Linux, Windows
Location: Dresden, Germany
Contact:

Re: suspend to ram fails wenn the vbox modul is loaded

Post by frank »

fixedwheel: Can you explain your problem and the solution again? Did I understand you correctly that you unloaded/re-loaded the tg3 module for suspend/resume and that this made the networking stop working and that is actually not necessary?

ingo2, it's a shame. So switching to NAT does not resolve your problem either. Yes, there were a lot of USB changes but USB on Linux hosts should be harmless as we don't fiddle with any kernel code there (this is different for Windows hosts where we require our own kernel driver). Candidates for host crashes is normally kernel code, either in the host modules or in VMM. At least between VBox 4.0.4 and VBox 4.0.6 I did not spot any problematic code after a quick look.
ingo2
Posts: 192
Joined: 28. Feb 2008, 14:15
Primary OS: Debian other
VBox Version: OSE Debian
Guest OSses: OS/2, WinXP-10, Linux
Location: Germany

Re: suspend to ram fails wenn the vbox modul is loaded

Post by ingo2 »

I just checked the device nodes which VBox creates at boot-up:

Code: Select all

~$ ls -l /dev/vboxusb
insgesamt 0
drwxr-x--- 2 root vboxusers  80  8. Jul 11:16 001
drwxr-x--- 2 root vboxusers 120  8. Jul 11:16 002
~$ ls -l /dev/vboxusb/001
insgesamt 0
crw-rw---- 1 root vboxusers 189, 4  8. Jul 11:16 005
crw-rw---- 1 root vboxusers 189, 6  8. Jul 11:16 007
~$ ls -l /dev/vboxusb/002
insgesamt 0
crw-rw---- 1 root vboxusers 189, 129  8. Jul 11:16 002
crw-rw---- 1 root vboxusers 189, 130  8. Jul 11:16 003
crw-rw---- 1 root vboxusers 189, 131  8. Jul 11:16 004
crw-rw---- 1 root vboxusers 189, 132  8. Jul 11:16 005
It appears that device "005" shows up twice - could that matter?
Please note that I have attached alltogether 3 USB hid-devices:
a) standard optical mouse
b) Logitech trackball
c) APC-UPS

However I already tried to remove one mouse which did not change things either.

Best regards,
Ingo
Post Reply