FS#41597 - [virtualbox] kernel panic and screen freezes
Attached to Project:
Community Packages
Opened by patrick (potomac) - Sunday, 17 August 2014, 22:52 GMT
Last edited by Sébastien Luttringer (seblu) - Thursday, 21 August 2014, 15:29 GMT
Opened by patrick (potomac) - Sunday, 17 August 2014, 22:52 GMT
Last edited by Sébastien Luttringer (seblu) - Thursday, 21 August 2014, 15:29 GMT
|
Details
Description:
virtualbox 4.3.14-3 has a weird bug since a recent upgrade of virtualbox-host-modules and kernel 3.16 I have randomly a kernel panic and sometimes a freeze screen ( keyboard and mouse don't react ) if I load the virtualbox-host-modules ( modprobe vboxdrv && modprobe vboxnetadp && modprobe vboxnetflt ) and run virtualbox, the kernel panic : kernel panic - not syncing: fatal exception in interrupt kernel offset : 0x0 from 0xffffffffff panic occured, switching back to text console the problem has began recently since the kernel 3.16 and the upgrade of virtualbox-host-modules ( 4.3.14-3 -> 4.3.14-5 ), one of this 2 packages ( or both ) are probably guilty, I have archlinux 64 bits, I use the radeon driver ( I have an ati HD4650 Pcie graphic card ) Additional info: * package version(s) virtualbox 4.3.14-3 virtualbox-host-modules 4.3.14-5 linux 3.16.1-1 * config and/or log files etc. Steps to reproduce: the bug is easily reproducible : - load manually the modules : modprobe vboxdrv && modprobe vboxnetadp && modprobe vboxnetflt - start virtualbox ( for example an archlinux guest ) - after a few minutes a kernel panic, a freeze or a crash of virtualbox will occur |
This task depends upon
Closed by Sébastien Luttringer (seblu)
Thursday, 21 August 2014, 15:29 GMT
Reason for closing: Upstream
Thursday, 21 August 2014, 15:29 GMT
Reason for closing: Upstream
but it doesn't solve the problem, I get again a kernel panic 5 minutes after running virtualbox,
the kernel panic occurs in host ( archlinux ), so each time I had to do a "reset" in order to restart my PC ( and then fsck runs each time ), it's a very high bug
I downgraded linux 3.16.1-1 to 3.16-2 and the bug is still here,
but if I downgrade linux 3.16.1-1 to linux 3.15.8-1 ( and virtualbox-host-modules to 4.3.14-3 ) then the bug is gone :
pacman -U /var/cache/pacman/pkg/linux-3.15.8-1-x86_64.pkg.tar.xz /var/cache/pacman/pkg/linux-headers-3.15.8-1-x86_64.pkg.tar.xz /var/cache/pacman/pkg/virtualbox-host-modules-4.3.14-3-x86_64.pkg.tar.xz
so there is a bug in linux 3.16.x related to the virtual machines features ( Vt-x ? ), who can trigger a kernel panic,
my cpu is a pentium Dual-Core E6800 3.33 Ghz,
the host is archlinux 64 bits, the guest for virtualbox is archlinux i686 virtual machine
FS#41556the user "Alim Gokkaya" seems to have a similar problem with virtualbox and kernel 3.16.x ( see in the comment section ),his journal.log about the kernel crash seems similar to mine
my virtual machines are only "linux OSes" ( archlinux, debian ) and all can trigger the bug,
on a french forum of archlinux someone told me the same thing, he has the bug with linux guest ( manjaro, kubuntu ),
the bug occurs especially if the virtual machine makes a lot of I/O ( read/write ) on partition
$ uname -a
Linux smartlu 3.16.0-seblu #1 SMP PREEMPT Mon Aug 4 17:46:25 CEST 2014 x86_64 GNU/Linux
$ pacman -Q virtualbox-host-dkms virtualbox
virtualbox-host-dkms 4.3.14-3
virtualbox 4.3.14-3
The traceback occur within the fuse stack, do you experiment traceback without fuse filesystem mounted?
the bug with virtualbox occurs with a random time, in the french forum of archlinux we are 3 to have seen this bug with the last kernel package ( 3.16.1-1 ), the link if you understand french :
https://forums.archlinux.fr/topic15778.html
best chance to trigger this bug is to use a linux guest and start to do a lot of I/O in the virtual machine ( like an update with pacman -Syu inside the virtual machine ),
after doing a lot of I/O inside the virtual machine I notice this error message in the linux guest :
failed command: Write FPDMA QUEUED
status: {DRDY}
then few seconds later the virtual machine freezes and sometimes a kernel panic ( in host ) occurs
The topmost stack information from trace_kernel_vbox file is
kernel: [<ffffffffa0b8fbdb>] ? fuse_copy_do+0xbb/0xd0 [fuse]
kernel: [<ffffffffa0b90181>] fuse_copy_page+0xa1/0x5f0 [fuse]
Fuse: http://en.wikipedia.org/wiki/Filesystem_in_Userspace
Do you use file sharing between you vm and you guest? Could you paste the output of
$ mount|grep fuse
the output of mount | grep fuse :
/dev/sdb1 on /mnt/xp type fuseblk (rw,nosuid,nodev,relatime,user_id=0,group_id=0,allow_other,blksize=4096)
fusectl on /sys/fs/fuse/connections type fusectl (rw,relatime)
/dev/sda5 on /mnt/wdc type fuseblk (rw,nosuid,nodev,relatime,user_id=0,group_id=0,allow_other,blksize=4096)
gvfsd-fuse on /run/user/1000/gvfs type fuse.gvfsd-fuse (rw,nosuid,nodev,relatime,user_id=1000,group_id=100)
my vm ( archlinux i686 ) is located in /mnt/xp ( ntfs partition )
I created a bug report in kernel's bugzilla :
https://bugzilla.kernel.org/show_bug.cgi?id=82951