Balloon driver preventing Windows 7 guest from shutting down about kvm-guest-drivers-windows

[JonathonReinhart]

ballon.sys 61.70.104.8100 from virtio-win-0.1.81 does not seem to exhibit this same behavior; it is able to shut down properly.

That version, however crashed with several bug checks: 0xA: IRQL_NOT_LESS_OR_EQUAL once on boot, and 0xC5: DRIVER_CORRUPTED_EXPOOL at shutdown.

[YanVugenfirer]

Hi,

Thanks you for the report. There were recent changes that fixed balloon service misbehaviour that might have cause hang on shutdown.

I will check crashes.

Can you upload mini dump files from the crashes?

Best regards,
Yan.

[JonathonReinhart]

Unfortunately, I do not have minidumps for those crashes. For some reason, Windows never wrote them even though it is configured to do so.

Those crashes though were from 0.1-81, not the latest 0.1-94. If you still want them I can try to reproduce.

Also, if you have any idea what I should look for during the shutdown hang (on 0.1-94), I can try to get this with WinDbg.

Or I can try to get the build environment set up and try the latest from master or stable. Unfortunately there is no correlation with the version numbers and git commits, as there are no tags in git.

[YanVugenfirer]

Hi,

I just pushed tags. Sorry for that.

If you look at the stable branch, before each release status.txt is updated
and there is a commit message that looks like "update status file. mm116
<-->b96
YanVugenfirer/kvm-guest-drivers-windows@16d4b7a
".

b - is the build number that you see in the package name (for
example 0.1-96).
mm - is the tag for this release.

You can also checkout according to the hash of the status.txt update commit.

Best regards,
Yan.

On Mon, Jan 5, 2015 at 10:15 AM, Jonathon Reinhart <[email protected]

wrote:

Unfortunately, I do not have minidumps for those crashes. For some reason,
Windows never wrote them even though it is configured to do so.

Those crashes though were from 0.1-81, not the latest 0.1-94. If you still
want them I can try to reproduce.

Also, if you have any idea what I should look for during the shutdown hang
(on 0.1-94), I can try to get this with WinDbg.

Or I can try to get the build environment set up and try the latest from
master or stable. Unfortunately there is no correlation with the version
numbers and git commits, as there are no tags in git.

On Mon, Jan 5, 2015 at 1:43 AM, Yan Vugenfirer [email protected]
wrote:

Hi,

Thanks you for the report. There were recent changes that fixed balloon
service misbehaviour that might have cause hang on shutdown.

I will check crashes.

Can you upload mini dump files from the crashes?

Best regards,
Yan.

—
Reply to this email directly or view it on GitHub
<
https://github.com/YanVugenfirer/kvm-guest-drivers-windows/issues/26#issuecomment-68673533>

.

Computers are incredibly fast, accurate and stupid. Human beings are
incredibly slow, inaccurate and brilliant. Together they are powerful
beyond imagination.
A. Einstein

—
Reply to this email directly or view it on GitHub
https://github.com/YanVugenfirer/kvm-guest-drivers-windows/issues/26#issuecomment-68678730
.

---

Daynix Computing LTD
Yan Vugenfirer, CEO
Email: [email protected]
Phone (Israel): [phone]
Phone (USA): [phone]
Phone (UK): [phone]
Web: www.daynix.com

[hammerg]

This sound like a bug that was already solved (but no build which include the fix was released yet).

Does the bug reproduce if you disable the balloon service and not the driver?

Thanks, Gal.

[JonathonReinhart]

Manually stopping the BalloonService times out:

Error 1053: The service did not respond to the start or control request in a timely fashion.

However, after that attempt at stopping the service, and then changing the service to Disabled, I was able to shut down the machine successfully (on that same boot).

And unsurprisingly, after the next boot (BallonService still Disabled), I was again able to shut down successfully.

So yes, disabling BootService allows me to shut down successfully, while enabling it prevents a full shutdown.

[hammerg]

I think that commit 5f7a25c is the one that is expcted to resolve this.

[JonathonReinhart]

I'd be happy to try this out tomorrow evening (EST). Would one of you be able to provide an updated build? (.sys and .exe, I'm assuming.) Otherwise it may take me longer to get the build environment set up.

[hammerg]

I can give you an unofficial build of the balloon service (exe).

[JonathonReinhart]

My email address is now on my profile page.

[hammerg]

I've send you an email with the balloon service attached.

[hammerg]

It will probably won't solve your problem. I was able to reproduce this bug on my setup and the bug was in the driver's code. A patch was posted.

[JonathonReinhart]

Edit: Heh.. interesting timing....

I can confirm that the build of balloon service that @hammerg sent me, based on commit 0558b0d, works; I am able to fully shut down the guest. 👍

This is with the latest release of VirtIO Balloon driver 61.71.104.9400.

For reference, this is with amd64\blnsvr.exe, with build date of Mon Jan 5 04:54:53 2015 and SHA1 of
009481149ad127e42c58f005297f91128ddd321b.

The Balloon driver is loaded, BallonService is Started, and blnsvr.exe is running.

Swapping the original blnsvr.exe back in caused the problem to re-appear, as expected.

[hammerg]

Yep. Good timing! :-) Thanks for the verification.

I got consfused between two similar bugs. This bug happens when you shutdown Windows and it was fixed in commit 0558b0d. The second bug which I mentioned in my previous comment occurs when you put the guest to sleep or hibernate.

[JonathonReinhart]

That commit is just debug. I think you meant 5f7a25c. You can close this issue if you'd like.

[hammerg]

Yes, you are right, 5f7a25c. Seems like I need one more cup of coffee.

[JonathonReinhart]

This is fixed by 5f7a25c.

[GioMac]

Still, I confirm exactly same issue for Windows 2008 R2:
balloon driver from virtio-win-0.1-74.iso - works
balloon driver from virtio-win-0.1.96.iso - spinner on shutdown
virtio-win-0.1.96.iso is in "stable" tree.
Not solved?

[hammerg]

That patch which fixed the bug is included from build 99.

[GioMac]

Is there any mailing list for Fedora where stable tree for isos issues are discussed or solved? Looks like there is broken iso shipped as stable.