Frequent BSOD's on Windows guests since 4.3.x

[mjdbb1]

I've been running everything from 4.3.10 to 4.3.15 and I'm getting constant bluescreens in Windows 7 guests. They are fresh instals as well as older guests. The timing seems to be random, but always within a few minutes. When I can catch the BSOD information, it's regarding "REFERENCE_BY_POINTER". The bugcheck is 0x00000018.

At first I thought it was a memory issue, but it occurs even when there is very little memory load on the host. Host is XP SP3, Intel Core2Duo, 2.93Ghz, 3GB Ram. I ran a full memory diagnostic which found no problems.

I've done some experimenting, and I can recreate this issue even with barely any memory load on the host. I rolled back to one of the earlier versions that I knew had worked, 4.2.6, and I cannot recreate a single crash, even with an intentionally heavy memory load on the host. I tried several other 4.2.x builds and stopped at 4.2.16. (Newer builds lead to the well-known security issues that also affect the latest 4.3.x branch). On all of these 4.2.x builds, I just can't get a crash to happen, no matter how hard I try.

Because this didn't happen at all until I upgrade to 4.3.x, I tend to think there's a regression of some kind in 4.3.x. If anyone is so kind as to take a look at this, I am more than happy to provide additional logs, testing, or information.

Thanks for reading!

[Frank Mehnert]

Right now I have no idea what this could be, especially as I'm not aware of similar reports. Your CPU does not support nested paging, otherwise I would suggest to toggle that setting (does not make sense in your case). It is true that there were several changes in the VMM code with VBox 4.3.x but I would still appreciate if you could try the latest 4.2.x release. If 4.2.26 does not work, please try 4.2.24.

And no, newer builds do NOT have well-known security issues. If you are talking about the recent problems on Windows hosts: The newest builds make executing VirtualBox on Windows more secure than it was before.

If your guest works with 4.2.24 it would still make sense to try the latest 4.3.17 test build from #13187.

[mjdbb1]

4.2.26 crash

[mjdbb1]

Thanks for the info. The 'security issues' I was referring to are the issues documented here: ​https://forums.virtualbox.org/viewtopic.php?f=6&t=63556&start=135 I wasn't suggesting that VirtualBox is not secure, I was only referring to the issues that were caused by the security changes. Sorry for the confusion on that.

I was able to run 4.2.24 and it has been stable so far, but on 4.2.26 I encounter the attached error. It seems to be the same issue documented in the thread above for the latest 4.3.x builds. I think it's unrelated, but because of this 4.2.24 is as far as I can go.

The crashes I was originally describing are only happening with the 4.3.x builds, so I'm suspecting there's a regression somewhere.

Thanks for taking the time to look into this. I'll help however I can!

[mjdbb1]

I hate to bump my own topic but I wanted to add that this behavior continues with the most current versions. Can anybody detail any of the VMM changes in 4.3.x that could be responsible for this issue? I hate to be stuck on 4.2.24 but that's the last version that doesn't give me this problem.

[Frank Mehnert]

If this problem still happens with VBox 4.3.20, could you attach a VBoxStartup.log + VBox.log of such a VM session? Thank you!

[mjdbb1]

Startup error

[mjdbb1]

Ok, I just installed 4.3.20 (and also tried 4.3.18) but I recieve the attached error when launching (error01.jpg). I also tried to install these builds on two other XP systems and I recieve the same error.

I haven't been able to run any build later than 4.2.24, because of the issues described here: ​https://forums.virtualbox.org/viewtopic.php?f=6&t=63556&start=135 This new error may be a new issue. But either way I can't proceed to test the original problem on anything later than 4.2.24.

Is there anything else I can do here? Or am I just having a bad run of luck with 4.3.x?