OpenSolaris guest boot failure "int 15h function AX=ec00, BX=0003"

[Peter Dunlap]

I recently upgraded my system with a new motherboard (DFI LanParty X48) and Intel Core 2 Q9400 processor. Host operating system is OpenSolaris build 133. Prior to the upgrade I had three guest OSes that were working fine: OpenSolaris build 132, Windows XP and Ubuntu. After the upgrade the OpenSolaris guest will no longer boot but the Windows XP and Ubuntu guests still work fine. I also tried booting an OpenSolaris live CD image which also failed. In all cases the guest appears to hang immediately after the OS is selected at the grub menu. It never actually gets to the point where it prints the Solaris banner.

I'm running VirtualBox 3.1.4 which was freshly installed from the "extra" repository after the hardware upgrade.

According to the log it looks like the last thing that happens is always this unsupported BIOS function:

00:02:33.031 Guest Log: BIOS: * int 15h function AX=ec00, BX=0003 not yet supported!

The Solaris GRUB boot loader does use that function which would explain why it doesn't actually appear to load the Solaris kernel:

​http://src.opensolaris.org/source/xref/ ... asm.S#2659

But if that is really the problem I don't understand how the guest ever worked. I will attach the logfile and the output from "VBoxManage showvminfo" for the guest.

[Peter Dunlap]

Logfile for "Work VM" OpenSolaris guest boot failure

[Peter Dunlap]

Output from "VBoxManage showvminfo" for "Work VM"

[Leif Hedstrom]

I'm having the exact same problem, happened after I upgraded to a new mobo with an i7 920. I also tried reinstalling OpenSolaris, but same problem. None of my other VMs have any problems, and I have a lot of them (about 15 different Linux distros, Windows and FreeBSD).

[michaln]

Correct ​link.

That function has never been supported by the VirtualBox BIOS and I can't find any documentation for it.

[michaln]

The documentation is ​here (section 12.21). Based on the available information, the BIOS service would be a no-op for VirtualBox anyway, so it's highly unlikely that it has anything to do with the hang.

As usual, please do not jump to conclusions and assume that just because something is the last message in the log, it has any relevance to what happened next.

Was the previous processor/OS 32-bit or 64-bit? GRUB is clearly taking a different code path, but that could be because of different bitness or because of other different CPU features.

[Leif Hedstrom]

In my case, the previous host was 32-bit, but I made a complete reinstall of OpenSolaris guest after upgrading to FC12 64-bit, and I still have the same problem. It boots fine from the CD image, but not from the installed OS disk. I'm using the 32-bit Solaris guest on it, I can try the 64-bit version as well if that'll help?

[Peter Dunlap]

Replying to michaln:

The documentation is ​here (section 12.21). Based on the available information, the BIOS service would be a no-op for VirtualBox anyway, so it's highly unlikely that it has anything to do with the hang.

As usual, please do not jump to conclusions and assume that just because something is the last message in the log, it has any relevance to what happened next.

Was the previous processor/OS 32-bit or 64-bit? GRUB is clearly taking a different code path, but that could be because of different bitness or because of other different CPU features.

The old system was AMD64, specifically it was an Athlon64 X2 3600+ "Brisbane" CPU with ASUS M2N nForce430 motherboard.

I am suitably chastened about jumping to conclusions -- I just tried booting an OpenSolaris ISO within VirtualBox on another unrelated system with a Core2 Duo processor and it booted successfully even though the same message appeared in the log. So my defect synopsis is misleading and the real problem is just that OpenSolaris always hangs on my new system just after booting from GRUB. I would be happy to collect any required data or run some experiments to help isolate this problem.

[michaln]

We've updated our BIOS so that the message about an unsupported function won't be printed. That unfortunately won't really do anything for you, because the failure to boot on your CPU is completely unrelated (with 99.99% probability).

You could try changing the guest's RAM size and varying other settings. Would be also interesting to turn on VT-x and see if OpenSolaris boots up that way (32-bit kernel of course).

Could be also worthwhile checking OpenSolaris 0906, just to see if that has the same problem or not.

[aeichner]

Please reopen if still relevant with a recent VirtualBox release.