[Call For Testing] VirtualBox for FreeBSD! take 4

[Juergen Lock]

On Thu, May 28, 2009 at 06:22:19PM +0200, Juergen Lock wrote:
> In article <20090528115754.GA1390 at carrot.geeknest.org> you write:
> >On Wed, May 27, 2009 at 03:43:43PM +0200, Martin Wilke wrote:
> >> 
> >> Howdy,
> >> 
> >> First of all sorry for all unanswered mails, I got a stupid flu,
> >> but now i feel better... ok now back to vbox, time for a new call
> >> for testing :-)
> >> 
> >> Following was added/fixed:
> >> 
> >> - ACPI Support was added
> >> - hostDVD support was added
> >> - Fix startup on HEAD
> >> - Plist problem under AMD64 was fixed
> >> - Qt4 Frontend is now Optional
> >> - Desktop file was added
> >> - Xorg dependencies was fixed
> >> - Guest additions was added (thx to Maho NAKATA <chat95 (at) mac dot com>)
> >> 
> >> Open task:
> >>  We have got 2 patches for nls support and the request
> >>  to make dbus and pulseaudio optional. These both will
> >>  be added with the next run. 
> >> 
> >>  We'd like to say many many thanks for all your feedback.
> >> 
> >> http://people.freebsd.org/~miwi/vbox/virtualbox_5.tgz
> >> 
> >FreeBSD version:
> >FreeBSD carrot.geeknest.org 7.2-STABLE FreeBSD 7.2-STABLE #16 r192900: Wed May 27 19:09:49 UTC 2009     lulf at carrot.geeknest.org:/usr/home/devel/freebsd/obj/usr/src/sys/CARROT  i386
> >Using the svn port: rev 387
> >
> >I get a panic when loading vboxdrv. It seems related to the hz fix-patch that
> >was added, as the assertion Assert(cMillies != RT_INDEFINITE_WAIT); on line
> >	212  in semevent-r0drv-freebsd.c:212 fails.
> >
> >0  doadump () at pcpu.h:196
> >#1  0xc058dd08 in boot (howto=260) at /usr/src/sys/kern/kern_shutdown.c:418
> >#2  0xc058dfe5 in panic (fmt=Variable "fmt" is not available.
> >) at /usr/src/sys/kern/kern_shutdown.c:574
> >#3  0xc07f3ff4 in trap_fatal (frame=0xf442ebc8, eva=0)
> >    at /usr/src/sys/i386/i386/trap.c:939
> >#4  0xc07f4d31 in trap (frame=0xf442ebc8) at /usr/src/sys/i386/i386/trap.c:726
> >#5  0xc07daddb in calltrap () at /usr/src/sys/i386/i386/exception.s:159
> >#6  0xc60e70be in rtSemEventWait (EventSem=0xc6032c90, cMillies=4294967295, 
> >    fInterruptible=false)
> >    at /usr/home/lulf/src/virtualbox/work/virtualbox-2.2.2r19980/src/VBox/Runtime/r0drv/freebsd/semevent-r0drv-freebsd.c:212
> >#7  0xc60e71b0 in RTSemEventWait (EventSem=0xc6032c90, cMillies=4294967295)
> >    at /usr/home/lulf/src/virtualbox/work/virtualbox-2.2.2r19980/src/VBox/Runtime/r0drv/freebsd/semevent-r0drv-freebsd.c:240
> >#8  0xc60e47f1 in rtTimerThread (Thread=0xc6034c90, pvUser=0xc6033390)
> >    at /usr/home/lulf/src/virtualbox/work/virtualbox-2.2.2r19980/src/VBox/Runtime/generic/timer-generic.cpp:238
> >#9  0xc60e96c0 in rtThreadMain (pThread=0xc6034c90, NativeThread=3314350912, 
> >    pszThreadName=0xc6034cd0 "TIMER")
> >    at /usr/home/lulf/src/virtualbox/work/virtualbox-2.2.2r19980/src/VBox/Runtime/common/misc/thread.cpp:635
> >#10 0xc60f5ee7 in rtThreadNativeMain (pvThreadInt=0xc6034c90)
> >    at /usr/home/lulf/src/virtualbox/work/virtualbox-2.2.2r19980/src/VBox/Runtime/r0drv/freebsd/thread2-r0drv-freebsd.c:112
> >#11 0xc05689c1 in fork_exit (callout=0xc60f5ec0 <rtThreadNativeMain>, 
> >    arg=0xc6034c90, frame=0xf442ed38) at /usr/src/sys/kern/kern_fork.c:811
> >#12 0xc07dae50 in fork_trampoline () at /usr/src/sys/i386/i386/exception.s:264
> 
> This is also what I get now (only looked like a `hang' yesterday since
> I was in X and it dropped into ddb, now I did the kldload from console,
> forced a crashdump and the backtrace looks just like yours, except that
> my vboxdrv.ko doesn't have debug symbols.)

Ok I just tried reverting the hz=100 patch (got it from
	http://pastebin.ca/1433127
as mentioned here a while ago), and that fixed the problem for me.
(Of course I'm running with hz=1000, otherwise it probably wouldn't work.)

 So at least I can now confirm that patch seems to be whats causing
the problem...

 HTH,
	Juergen

PS: smp isn't supposed to work yet, right?  At least I get a warning saying
I had only one cpu when trying to give a guest more than one, but in fact
this box has 4 cores...