Discuss the 4.1.2 release

[Perryg]

I am running VB 4.1.2 on two Ubuntu Lucid Linux 10.04.3 hosts with iSCSI shared storage and 26GB RAM per host.
When I try to teleport a VM on one host to the other host, something that worked correctly on 4.0.12, I get a guru meditation error. The screen shot only shows a login prompt. Here is the log
http://pastebin.com/Lj0zzvV6
The forum wouldn't let me paste the log file.
I depend on teleportation for maintenance purposes. I really need to know what is going on, please..

Please report this to bugtracker. You will need to setup an account there as it is on a different system.
It would also help if you can post the ticket number here so other can see the progress, add information, or see the final results.

[Eck]

Another Windows98SE user here to report to mpack.

512MB, VT Nested Paging, the whole enchilada turned on (it zooms with this stuff, glad I got a new laptop so I could use it), 128MB video with the Sci-Tech driver, and the realtek w95 vxd driver for nice AC97 audio.

The screen is set to 1024x768 so I have the black bars on my 1366x768 laptop monitor, but that's the size I used to use on my old HPmx70 monitor anyway so things look fine. Still, I wish the Sci-Tech driver could stretch it out like the VMware guest driver does.

The big glitch only effects startup. I need to open device manager and reinstall the USB 2.0 Intel driver every bootup. It's provided by the Maximus-Decim USB Mass Storage Driver package and it isn't the driver that causes this because it works automatically on VMware and a real machine. Only a VirtualBox Windows 98SE guest needs the reinstall every startup. I've got my printer and scanner on USB 2.0 so it's kind of important, but the problem is just an annoyance as it all works fine as long as I just keep clicking (reinstall driver, select from a list, the driver appears and I hit apply and ok. And bam, my printer and scanner are added to the Device Manager list and work fine. This is only this easy with the 98SE2ME package as the original 98SE drivers need to actually be removed and then after a reboot they are automatically loaded. Next reboot they would be gone. With 98SE2ME, if I ever remove them and restart the system loses them and they appear in no list until the Maximus-Decim package is reinstalled. So after trying that once I wasn't fooled twice, I simply update them on startup, easier eh?

I was excited to discover that my really old programs that were designed for Win3.1, like the older Carmen Sandiego games, and only fill the screen with 256 colors 640x480 set are now working great. I set the Nvidia settings on the host to 640x480 when I want to play the old stuff and then the Windows Display Properties is now actually able to switch to that 256 colors 640x480 mode without rebooting (or even at all in comparison to VMware or the older VirtualBox versions.) This while the host is still set to 24 bit True Color! I don't know why or how that's now possible but congrat's if it was fixed up like that on purpose! Games like these use midi, which doesn't work with the SB16 driver, hence my use of the AC97. And don't even think about letting them install the QuickTime version on these games cdroms. Get the 32 bit 2.1.2.59 from an old version site and the 6.5.2 and install them both, setting the older control panel to optimize using Windows Bitmap and 22050 sound. The versions on the cdroms are buggy older ones and that 16 bit version can even totally fubar Windows!

Next I'll muck about with some AfterDark screensavers. Wasn't old stuff fun?

[Etepetete]

I've noticed that he mouse cursor seems to be more accurate in this version. The issue I experienced about the mouse cursor being too low and to the right as I reported in the 3.0.8 release seems to be fixed. Now the cursor is exactly on track. Maybe the issue with the mouse cursor being too high when it changes from left pointing to right pointing has to do with this fix?

[mpack]

Thanks for all the replies re: Win98 startup. I can confirm that if I leave it running then it eventually boots up, and once it reaches the desktop the performance is about the same as ever. However since I mainly use Win98 as a test environment (including checking out the observations of people here), and that means a lot of resets, this behaviour makes the VM just about unusable.

Ok, so something during boot is taking a helluva time, but it isn't hanging. I've tried it with VT-x on and off - no difference. I would not expect nested paging and other similar suggestions to make any difference to Win98. My own gut feeling is that it's a networking issue, but no real evidence for that yet.

Oh, and I guess I should mention that I'm using an XP Pro host. If networking really is the issue then the choice of host OS could matter I guess.

Maybe I should split this discussion off to a separate support thread...

[ingo2]

I still have the annoying beheaviour as already reported for every single VBox-version > 4.0.4, also in 4.1.2:

If the host has been suspended and/or the VM has been closed by saving several times, I first get intermittent mouse hangs for a few seconds. This happenes a few times and eventually ends up in the VM completely frozen (most probably due to the fact it cannot write to its disk). Disk activity LED (the one in VBox window at the bottom) is constantly green, but nothing written/read from host's disk.

I have recently filed a bug http://www.virtualbox.org/ticket/9427 for VM's disk corruption. Probably this bug http://www.virtualbox.org/ticket/9410 is a similar observation. It should not bee too difficult for the developers to pin down the root cause, because they know which "major changes" they made in 4.0.6 compared to 4.0.4 (which runs absolutely reliable with the same *.vdi image). So I am still on 4.0.4

Best regards,
Ingo

[mark rumsey]

My Win98SE VM still frozen stiff with 4.1.2, can't even get it to boot. I don't know that it actually locks up during the boot, but it just sits there on the Win98SE logo and after a few minutes of watching paint dry I get fed up and "power off" the VM. Status line shows no sign of disk or network activity. Host task Manager shows "VirtualBox.exe" has maxed out one core. VBoxSVC.exe is on 0% of CPU.

This VM was fine until I upgraded to 4.1.0. I have a different Win98SE VM on another PC (not a clone), and it does the same thing.

Is anyone able to run a Win98 VM satisfactorally with v4.1.x ? (satisfactorally, meaning as you would expect for an OS that isn't supported with GAs).

Yes, I have Win98SE running quite happily under 4.1.0 complete with the SVGA driver, bridged network and VT-x/AMD-V enabled. One thing it may be worth checking is the memory allocation has not changed. Win9x has a bit of a bug which makes it difficult to run n any machine with more than about 768Mb RAM, and downright impossible on anything with over 2.5GB. I have mine set to 768Mb and it runs nicely.

[abcuser]

RESTORING A SNAPSHOT GETS AN INACCESSIBLE STATE

Hi,
today I have uninstalled VirtualBox 4.0.12 and installed VirtualBox 4.1.2 on Ubuntu 8.04 LTS host. Virtual machine is Windows XP sp3 (guest). I have installed Oralce extension pack and started machine, no problem, installed guest additions, rebooted machine, no problem. Then I shutdown guest and trying to work with snapshots.

I created a snapshot. Then started guest, do some changes and stopped the machine. Created a second snapshot. Then deleted first snapshot created today. Reverted back to a week old snapshot (when asked to create a snapshot of not losing a current state I have unchecked the option). Then created a new snapshot (that is child of a week old snapshot). Reverted back to today's second snapshot and got a Inaccessible virtual machine state (see attached picture of VBox GUI window).

I clicked on Refresh button several times, waited for 15 minutes and clicked on Refresh again. No messages, no errors, just "Inaccessible" state.

Any idea how to get out of inaccessible state?

P.S. I have reported a bug report at: http://www.virtualbox.org/ticket/9457 - it looks like a bug to me. Log files and print-screen of vbox gui when error appears are attached at bug report.
Regards

[mschwartz]

Check for updates on Lion still doesn't tell me there's an upgrade to 4.1.2 (running 4.1.0)

[microapp]

??,
?VirtualBox 4.1.2 for Windows ???,??vbox????oracle linux 5.7, ????FTP, ???????????(? oracle database.iso), ??????. ?????windows xp sp3.
??!

hi, in "VirtualBox 4.1.2 for Windows", i installed a oracle linux 5.7, when i ftp / copy a large file to linux, my system is on breakdown, my platform is windows xp sp3.

best regards.

[appyface]

Running 4.1.2 on Win7 Pro SP1 64-bit host, XP Pro SP3 32-bit guest.

Since upgrading from 4.1.0 to 4.1.2 (Vbox soft and new guest additions) my guest is crashing regularly now. This same guest ran without crashing on all prior versions going back into early 3.x series.

This is the first error message I can see in the guest log:

01:46:24.583 Changing the VM state from 'RUNNING' to 'RUNNING_LS'.
01:46:24.696
01:46:24.696 !!R0-Assertion Failed!!
01:46:24.696 Expression: <NULL>
01:46:24.696 Location : e:\tinderbox\win-rel\src\vbox\vmm\vmmall\PGMAllBth.h(1458) pgmR0BthPAEPAESyncPageWorkerTrackDeref
01:46:24.696 HCPhys=00000000520a7000 wasn't found!
01:46:24.696 !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
01:46:24.696 !!
01:46:24.696 !! Guru Meditation -2701 (VERR_VMM_RING0_ASSERTION)
01:46:24.706 !!
01:46:24.707 !!R0-Assertion Failed!!
01:46:24.707 Expression: <NULL>
01:46:24.707 Location : e:\tinderbox\win-rel\src\vbox\vmm\vmmall\PGMAllBth.h(1458) pgmR0BthPAEPAESyncPageWorkerTrackDeref
01:46:24.707 HCPhys=00000000520a7000 wasn't found!
01:46:24.707 !! ACTIVE TRAP=0e ERRCD=0000000000000003 CR2=00000000f4cd6c64 PC=00000000804ee6f6 Type=0 (Guest!)
01:46:24.707 !!
01:46:24.707 !! CallRing3JmpBuf:
01:46:24.707 !!

(ignore source not found messages)

The guest freezes, a message box to acknowledge the crash appears, I OK that and the guest goes down. The guest is now shown as inaccessible in the GUI and the hard disk is detached along with all child snapshots. I'd really like to salvage the guest and its current hard disk with snapshots, but my various attempts with Vboxmanage and/or 'surgeries' on the XML files hasn't met with any success yet. All I can do to recover, is delete the guest and all of its files, then re-import it from a week-old backup and try again.

I'm considering going back to 4.1.0 to see if the guest returns to normal operation.

[appyface]

I found the problem.

I have task scheduler jobs that start some VM's, like this one, to ensure they're running during a particular period of time.

In earlier Vbox versions, the scheduled start command simply failed if the target VM was already running. In 4.1.2 the attempt to start crashes a running VM.

I'm off to search the bug tracker and if needed, submit a ticket.

[mpack]

Some good news on my Win98SE problem.

I recently discovered (after playing with Win8) that VT-x was actually disabled in my host PC BIOS, even though I'd thought it was enabled. Now that I have it working I thought I would try it with a Win98 VM. It makes a difference all right, the VM zooms! It boots up in seconds, compared to the minutes taken previously, and everything inside the VM - particularly screen paints - seem to be much faster. Oh, and the VM when idle seems to take 0-1% of host CPU, instead of the 20-30% seen previously.

I'm happy, but still curious. Did I miss a recent change whereby "difficult" guests like Win98 would rely more on VT-x?

[johnelee]

Re mpack...Good news re win98 - I'd like to run win me on vb but have been deterred by alleged lack off performance - runs ok on vpc. Believe that one needs to change the video driver to get adequate resolution.

Did you find that everything on win98 was faster or just some things? Any before & after using vt-x times?
What h/w are you running on?

John

[mpack]

Did you find that everything on win98 was faster or just some things? Any before & after using vt-x times?

I didn't do any benchmarks, but I just timed it at 18secs to reach the desktop with VT-x enabled, when I tried again with VT-x turned off it took 1m47s to do the same, and that is with the benefit of data probably still in the host disk cache from the first test.

Repaints, e.g. as you drag whole windows around, is noticeably jerky, rippling, stop-start stuff - but with VT-x enabled this seems as smooth as on the host. I had no idea that VT-x would have this effect... I expected VT-x would only help make 16bit code faster, surely the graphics drivers are all 32bit?

As I said, I haven't done formal benchmarks, and TBH I'm not really interested in finding out the full extent of the differences. The improvement on boot times and graphics is very obvious, I don't need to formally benchmark it. I have not measured disk performance, as that is less important to me.

What h/w are you running on?

Ordinary business PC, ordinary onboard graphics chipset (I'm not into PC gaming), Core 2 duo processor with VT-x enabled in host BIOS.

A lengthy discussion of Win98/WinME tidbits would be off topic in this thread. You should read the Win98 Howto(s), and start a new discussion thread if you still have questions.

[EgonF]

I'm looking for documentation/howto what should be done to upgrade from 4.0.8 to 4.1.2.