kernel backtrace pointing at KVM

kernel backtrace pointing at KVM

[Ivelin Ivanov]

Hello KVM team,

I am using KVM for maintaining mixture of Windows and Linux VMs.
recently I reinstalled the system with latest kernel 2.6.30 and
qemu-kvm-0.86.
I kept using the KVM module from the 2.6.30 kernel not from the pack
with the qemu-kvm.
Last night one of the Windows XP SP3 system rebooted and I found the
following trace in my logs:

------------[ cut here ]------------
WARNING: at arch/x86/kvm/x86.c:204 kvm_queue_exception_e+0x24/0x45 [kvm]()
Hardware name: IBM System x3650 -[7979B4G]-
Modules linked in: 8021q garp tun reiserfs st ide_gd_mod ide_cd_mod
bridge stp kvm_intel kvm ipv6 af_packet cpufreq_conservative
cpufreq_userspace cpufreq_powersave acpi_cpufreq speedstep_lib loop
dm_mod i5k_amb ibmpex i5000_edac ibmaem i2c_i801 iTCO_wdt
iTCO_vendor_support rtc_cmos bnx2 sr_mod rtc_core ipmi_msghandler ses
edac_core rtc_lib button pcspkr i2c_core cdrom serio_raw enclosure
shpchp pci_hotplug joydev sg usbhid hid uhci_hcd ehci_hcd usbcore sd_mod
crc_t10dif edd ext3 mbcache jbd fan ide_pci_generic ide_core ata_generic
ata_piix libata thermal processor thermal_sys hwmon aacraid scsi_mod
Pid: 26872, comm: qemu-system-x86 Not tainted 2.6.30-9-pae #1
Call Trace:
 [<c012eda8>] warn_slowpath_common+0x60/0x90
 [<c012ede5>] warn_slowpath_null+0xd/0x10
 [<f7fea423>] kvm_queue_exception_e+0x24/0x45 [kvm]
 [<f7feb509>] kvm_task_switch+0xfb/0xada [kvm]
 [<c03441e2>] ? _spin_unlock+0xf/0x23
 [<f7ff041e>] ? kvm_mmu_page_fault+0x16/0x75 [kvm]
 [<f805dea3>] handle_task_switch+0x6d/0x96 [kvm_intel]
 [<f805e3fd>] kvm_handle_exit+0x1c3/0x1e0 [kvm_intel]
 [<f7feb11a>] kvm_arch_vcpu_ioctl_run+0x89d/0xab6 [kvm]
 [<c01a7cfd>] ? do_sync_readv_writev+0xa1/0xdf
 [<f7fe34ef>] kvm_vcpu_ioctl+0xec/0x602 [kvm]
 [<c03441e2>] ? _spin_unlock+0xf/0x23
 [<f7fe3403>] ? kvm_vcpu_ioctl+0x0/0x602 [kvm]
 [<c01b2fe7>] vfs_ioctl+0x22/0x69
 [<c01b3467>] do_vfs_ioctl+0x439/0x472
 [<c0190277>] ? unmap_region+0x10a/0x124
 [<c01a8e52>] ? fget_light+0x8a/0xb1
 [<c01b34e0>] sys_ioctl+0x40/0x5a
 [<c0102b0c>] sysenter_do_call+0x12/0x28
---[ end trace 9cc5a56b9c7eda31 ]---

Here is some more info:
# uname -a
Linux pc187 2.6.30-9-pae #1 SMP PREEMPT Wed Jun 17 15:29:59 EEST 2009
i686 i686 i386 GNU/Linux
# qemu-system-x86_64  --version
QEMU PC emulator version 0.10.50 (qemu-kvm-devel-86), Copyright (c)
2003-2008 Fabrice Bellard
# modinfo /lib/modules/2.6.30-9-pae/kernel/arch/x86/kvm/kvm.ko
filename:       /lib/modules/2.6.30-9-pae/kernel/arch/x86/kvm/kvm.ko
license:        GPL
author:         Qumranet
srcversion:     934CD9DB264501B0431438A
depends:
vermagic:       2.6.30-9-pae SMP preempt mod_unload modversions CORE2
parm:           oos_shadow:bool
parm:           msi2intx:bool

Do I do something wrong or is it a bug in KVM ?
Thanks for the great work!
Hope to hear from you soon.
Best regards,

Ivelin Ivanov

P.S. Please, don't hesitate to ask me if you need any more info to trace
this issue.

Re: kernel backtrace pointing at KVM

[Avi Kivity]

On 06/20/2009 09:27 AM, Ivelin Ivanov wrote:
> WARNING: at arch/x86/kvm/x86.c:204 kvm_queue_exception_e+0x24/0x45 [kvm]()
> Call Trace:
>   [<c012eda8>] warn_slowpath_common+0x60/0x90
>   [<c012ede5>] warn_slowpath_null+0xd/0x10
>   [<f7fea423>] kvm_queue_exception_e+0x24/0x45 [kvm]
>   [<f7feb509>] kvm_task_switch+0xfb/0xada [kvm]
>   [<c03441e2>] ? _spin_unlock+0xf/0x23
>   [<f7ff041e>] ? kvm_mmu_page_fault+0x16/0x75 [kvm]
>   [<f805dea3>] handle_task_switch+0x6d/0x96 [kvm_intel]
>   [<f805e3fd>] kvm_handle_exit+0x1c3/0x1e0 [kvm_intel]
>   [<f7feb11a>] kvm_arch_vcpu_ioctl_run+0x89d/0xab6 [kvm]
>   [<c01a7cfd>] ? do_sync_readv_writev+0xa1/0xdf
>   [<f7fe34ef>] kvm_vcpu_ioctl+0xec/0x602 [kvm]
>   [<c03441e2>] ? _spin_unlock+0xf/0x23
>   [<f7fe3403>] ? kvm_vcpu_ioctl+0x0/0x602 [kvm]
>   [<c01b2fe7>] vfs_ioctl+0x22/0x69
>   [<c01b3467>] do_vfs_ioctl+0x439/0x472
>   [<c0190277>] ? unmap_region+0x10a/0x124
>   [<c01a8e52>] ? fget_light+0x8a/0xb1
>   [<c01b34e0>] sys_ioctl+0x40/0x5a
>   [<c0102b0c>] sysenter_do_call+0x12/0x28
> ---[ end trace 9cc5a56b9c7eda31 ]---
>
> Here is some more info:
> # uname -a
> Linux pc187 2.6.30-9-pae #1 SMP PREEMPT Wed Jun 17 15:29:59 EEST 2009
> i686 i686 i386 GNU/Linux
> # qemu-system-x86_64  --version
> QEMU PC emulator version 0.10.50 (qemu-kvm-devel-86), Copyright (c)
> 2003-2008 Fabrice Bellard
> # modinfo /lib/modules/2.6.30-9-pae/kernel/arch/x86/kvm/kvm.ko
> filename:       /lib/modules/2.6.30-9-pae/kernel/arch/x86/kvm/kvm.ko
> license:        GPL
> author:         Qumranet
> srcversion:     934CD9DB264501B0431438A
> depends:
> vermagic:       2.6.30-9-pae SMP preempt mod_unload modversions CORE2
> parm:           oos_shadow:bool
> parm:           msi2intx:bool
>
> Do I do something wrong or is it a bug in KVM ?
>    

It's a bug in kvm.  Is there anything in the Windows logs from the time 
of the reboot?  Did Windows create a memory dump?

-- 
error compiling committee.c: too many arguments to function