virtualbox.org

Hi all,

I am running Windows 7 x64 as a guest within VirtualBox 4.0.10 with host OS openSUSE 11.4 x86-64 (Core 2 duo, 12 GB RAM).

Whilst running a memory intensive application in Windows (5GB), the application crashed and I could see the messages below in the host OS and in dmesg. The guest has 8GB allocated to it and there is headroom at all times in the guest and host.

What can I do to avoid this and make things run stably?

Many thanks!

Message from syslogd@spartan at Jul 19 06:54:55 ...
kernel:[36871.710638] general protection fault: 0000 [#1] PREEMPT SMP

Message from syslogd@spartan at Jul 19 06:54:55 ...
kernel:[36871.710653] last sysfs file: /sys/devices/system/cpu/cpu1/cache/index2/shared_cpu_map


[36871.710659] CPU 0
[36871.710660] Modules linked in: mmc_block nls_iso8859_1 nls_cp437 vfat fat autofs4 ip6t_LOG xt_tcpudp xt_pkttype ipt_LOG xt_limit af_packet nfsd lockd nfs_acl auth_rpcgss rfcomm sunrpc exportfs sco coretemp bnep snd_pcm_oss snd_mixer_oss l2cap snd_seq snd_seq_device edd ip6t_REJECT nf_conntrack_ipv6 nf_defrag_ipv6 ip6table_raw xt_NOTRACK vmnet vboxnetadp vboxnetflt ipt_REJECT vmblock iptable_raw vsock vmci vmmon iptable_filter ip6table_mangle nf_conntrack_netbios_ns nf_conntrack_ipv4 nf_defrag_ipv4 ip_tables xt_conntrack vboxdrv nf_conntrack ip6table_filter ip6_tables x_tables cpufreq_conservative cpufreq_userspace cpufreq_powersave acpi_cpufreq mperf microcode fuse dm_mod usblp nvidia(P) arc4 btusb bluetooth ecb snd_hda_codec_idt sg sr_mod cdrom pcmcia iwlagn snd_hda_intel iwlcore snd_hda_codec snd_hwdep mac80211 dell_wmi sparse_keymap dell_laptop yenta_socket snd_pcm tg3 firewire_ohci cfg80211 r852 sdhci_pci ppdev firewire_core sm_common nand pcmcia_rsrc i2c_i801 sdhci crc_itu_t pcmcia_core dcdbas rfkill nand_ids nand_ecc snd_timer snd soundcore iTCO_wdt video pcspkr mmc_core parport_pc mtd joydev shpchp iTCO_vendor_support snd_page_alloc wmi pci_hotplug battery button parport ac ext4 jbd2 crc16 fan processor thermal thermal_sys [last unloaded: preloadtrace]
[36871.710766]
[36871.710771] Pid: 7801, comm: VirtualBox Tainted: P I 2.6.37.6-0.5-desktop #1 Dell Inc. Precision M6400 /0D138P
[36871.710780] RIP: 0010:[<ffffffffa100f41d>] [<ffffffffa100f41d>] rtR0MemObjLinuxFreePages+0x1d/0xc0 [vboxdrv]
[36871.710800] RSP: 0018:ffff8802be9d5b98 EFLAGS: 00010206
[36871.710805] RAX: 00000000000001e7 RBX: ffffc9000e4f8010 RCX: ffffea00051ddce0
[36871.710811] RDX: ffdfea00051d65f0 RSI: 0000000000000000 RDI: ffffc9000e4f8010
[36871.710816] RBP: ffff8802be9d5c28 R08: 0000000000000000 R09: ffff88031ffebaf0
[36871.710821] R10: 0000000000000093 R11: 000000000000000a R12: ffffc900074b6010
[36871.710826] R13: ffffc9000e4f4010 R14: 0000000000000001 R15: ffff8802be9d22c0
[36871.710832] FS: 00007f2688780700(0000) GS:ffff8800df000000(0000) knlGS:0000000000000000
[36871.710837] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b
[36871.710842] CR2: 00007f9a4075f000 CR3: 00000002a99e6000 CR4: 00000000000426f0
[36871.710847] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[36871.710853] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
[36871.710858] Process VirtualBox (pid: 7801, threadinfo ffff8802be9d4000, task ffff8802be9d22c0)
[36871.710864] Stack:
[36871.710867] ffff880176dce4c0 ffffc9000e4f8010 ffff8802be9d5c28 ffffffffa100fb26
[36871.710875] ffff880176bdfd40 0000000000000001 ffffc9000e4f8010 ffff8802be9d5c28
[36871.710881] ffffc900074b6010 ffffffffa100e08c ffffc9000e4fc010 0000000000000001
[36871.710888] Call Trace:
[36871.710917] [<ffffffffa100fb26>] rtR0MemObjNativeFree+0x1a6/0x1d0 [vboxdrv]
[36871.710940] [<ffffffffa100e08c>] VBoxHost_RTR0MemObjFree+0x7c/0x310 [vboxdrv]
[36871.710959] [<ffffffffa1021a93>] g_abExecMemory+0x1853/0x180000 [vboxdrv]
[36871.710994] DWARF2 unwinder stuck at g_abExecMemory+0x1853/0x180000 [vboxdrv]
[36871.710999]
[36871.711003] Leftover inexact backtrace:
[36871.711003]
[36871.711004] [<ffffffffa1022973>] g_abExecMemory+0x2733/0x180000 [vboxdrv]
[36871.711004] [<ffffffffa10263e1>] g_abExecMemory+0x61a1/0x180000 [vboxdrv]
[36871.711004] [<ffffffffa1006046>] SUPR0ObjRelease+0x1e6/0x210 [vboxdrv]
[36871.711004] [<ffffffff8107f73d>] ? up+0xd/0x40
[36871.711004] [<ffffffffa1027b67>] ? g_abExecMemory+0x7927/0x180000 [vboxdrv]
[36871.711004] [<ffffffffa10359c6>] ? g_abExecMemory+0x15786/0x180000 [vboxdrv]
[36871.711004] [<ffffffff81042e6c>] ? finish_task_switch+0x3c/0xe0
[36871.711004] [<ffffffffa1035fc0>] ? g_abExecMemory+0x15d80/0x180000 [vboxdrv]
[36871.711004] [<ffffffffa100a7f5>] ? supdrvIOCtl+0x1485/0x2860 [vboxdrv]
[36871.711004] [<ffffffffa100f16b>] ? rtR0MemAllocEx+0x7b/0x130 [vboxdrv]
[36871.711004] [<ffffffffa1005438>] ? VBoxDrvLinuxIOCtl+0x128/0x1e0 [vboxdrv]
[36871.711004] [<ffffffffa1005310>] ? VBoxDrvLinuxIOCtl+0x0/0x1e0 [vboxdrv]
[36871.711004] [<ffffffff811637a8>] ? do_vfs_ioctl+0x88/0x300
[36871.711004] [<ffffffff8108eaaa>] ? sys_futex+0x7a/0x190
[36871.711004] [<ffffffff81153d0c>] ? fget_light+0x9c/0xb0
[36871.711004] [<ffffffff81163aa0>] ? sys_ioctl+0x80/0xa0
[36871.711004] [<ffffffff81152a62>] ? sys_write+0x82/0x90
[36871.711004] [<ffffffff81002f8b>] ? system_call_fastpath+0x16/0x1b
[36871.711004] Code: 66 66 66 66 66 2e 0f 1f 84 00 00 00 00 00 55 53 48 89 fb 48 83 ec 08 48 8b 47 48 48 85 c0 74 48 66 0f 1f 44 00 00 48 8b 54 c3 48 <f0> 80 62 01 fb 48 83 e8 01 75 f0 80 7b 40 00 75 3a 48 8b 6b 48
[36871.711004] RIP [<ffffffffa100f41d>] rtR0MemObjLinuxFreePages+0x1d/0xc0 [vboxdrv]
[36871.711004] RSP <ffff8802be9d5b98>
[36871.711253] ---[ end trace 4eaa2a86a8e2da24 ]---

post a VBox log (as attachment, while posting see below: "Upload attachment")

singletrack wrote:VirtualBox 4.0.10

4.0.12 has been released a few days ago, could you try that?

consider opening a bug report (separate account needed) => http://www.virtualbox.org/wiki/Bugtracker

Log attached. I considered trying 4.0.12 but the changelog makes no mention of anything related to stability fixes for the guest/host configuration I have... Maybe something undocumented has been improved?

singletrack wrote:I considered trying 4.0.12 but the changelog makes no mention of anything related ...

sorry, already outdated: 4.1.0 has just been released ... http://forums.virtualbox.org/viewtopic.php?f=1&t=43215

I was informed today that a bug fix was in version 4.1.0 (released today as fixedwheel states) and to try it to see if it fixed Windows guests with more than 1 processor.
So far all I can get to work reliably is with one processor. I was using 2 just fine until version 4.* came out. Also I believe that you should not use the ICH9 chipset but rather the PIIX3 as this causes strange 10 second hangs.

Perryg wrote:you should not use the ICH9 chipset but rather the PIIX3 as this causes strange 10 second hangs.

as far as i can see in the log: PIIX3 is set, and 1 guest CPU as well

fixedwheel wrote:

Perryg wrote:you should not use the ICH9 chipset but rather the PIIX3 as this causes strange 10 second hangs.

as far as i can see in the log: PIIX3 is set, and 1 guest CPU as well

Yes you are correct, I should have clarified my post better. My intent in posting was to point out that there seems to be serious problems that have been around since the beginning of version 4.*. Reading various posts here and all kinds of various work-a-rounds, it seems to me that the underlining issue is still hidden. Meaning reporting to bugtracker with complete information and log files is probably the only way this is going to be fixed. Actually if the OP looks at the issues in bugtracker they will more than likely find one that matches their issue as well and can simply add their information to it.

Perryg wrote:My intent in posting was to point out that there seems to be serious problems that have been around since the beginning of version 4.*. Reading various posts here and all kinds of various work-a-rounds, it seems to me that the underlining issue is still hidden. (...)

from the log attached by singletrack:

00:00:00.407 [/Devices/e1000/0/] (level 3)
00:00:00.407   PCIBusNo      <integer> = 0x0000000000000000 (0)
00:00:00.407   PCIDeviceNo   <integer> = 0x0000000000000003 (3)
00:00:00.407   PCIFunctionNo <integer> = 0x0000000000000000 (0)
00:00:00.407   Trusted       <integer> = 0x0000000000000001 (1)

i recall a thread about (other) problems with Windows 7 and emulated e1000 NIC (the supposed only possible choice) ... just a shot into the blue, but try virtio-net (see linked thread)

Perryg wrote:I was informed today that a bug fix was in version 4.1.0 (released today as fixedwheel states) and to try it to see if it fixed Windows guests with more than 1 processor.
So far all I can get to work reliably is with one processor. I was using 2 just fine until version 4.* came out.

I tend to come to the same conclusion. I'm not yet 100% sure.
I can say that the last working VB for me is 3.2.10. As soon as I upgraded to a newer version (be it v3 or v4), my Win2008 64 bits guests freeze with the VB process on the host reaching 99% CPU usage.

I am now trying with only one CPU in the guest.

Ok, I have my Win2008 VM running for one week now without a problem. The only thing I did was to select 1 CPU instead of 2.

So there may be a bug when more than one CPU is used, and this was introduced in version 3.2.12