Announcement

Collapse
No announcement yet.

Startup Error Number: 4 on qemu + BSOD

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Startup Error Number: 4 on qemu + BSOD

    I'm trying to get PerformanceTest 8 working in a win7 64bit VM. The first attempt hit a BSOD and since then I can no longer even start the test. Each attempt tells me that a problem has occurred, Startup Error Number: 4. It also suggests I may need to uninstall and re-install the software, which I've done several times, including reboots, but it doesn't help. I'm attempting to use the free 30-day license as my goal it to simply make it work in the VM. Is there a corrupt license file hidden somewhere that I need to delete? Thanks

  • #2
    Running with DEBUGMODE doesn't tell me much:

    0.000s - DEBUG: Starting...
    0.000s - DEBUG: PerformanceTest 8.0 build 1031 64-Bit
    0.000s - DEBUG OS: Windows 7 Ultimate Edition Service Pack 1 build 7601 (64-bit)
    0.000s - DEBUG Path: C:\Users\Alex\Documents\PassMark\PerformanceTest\P erfTestLog.txt
    0.000s - DEBUG Command line: "C:\Program Files\PerformanceTest\PerformanceTest64.exe" DEBUGMODE
    0.000s - Date: 03/19/14 10:09:24
    0.000s - DEBUG: Opening splash window
    0.125s - DEBUG: Init Preferences
    0.172s - DEBUG: UTF8StringToWChar string truncated
    0.172s - DEBUG: UTF8StringToWChar string truncated
    0.203s - DEBUG: UTF8StringToWChar string truncated
    0.203s - DEBUG: UTF8StringToWChar string truncated
    0.203s - DEBUG: UTF8StringToWChar string truncated
    0.234s - DEBUG: UTF8StringToWChar string truncated
    0.234s - DEBUG: UTF8StringToWChar string truncated
    0.266s - DEBUG: UTF8StringToWChar string truncated
    0.266s - DEBUG: UTF8StringToWChar string truncated
    0.297s - DEBUG: UTF8StringToWChar string truncated
    0.297s - DEBUG: UTF8StringToWChar string truncated
    0.313s - DEBUG: UTF8StringToWChar string truncated
    0.328s - DEBUG: UTF8StringToWChar string truncated
    0.328s - DEBUG: UTF8StringToWChar string truncated
    0.344s - DEBUG: Init CD Burner
    0.531s - DEBUG: Retrieving Computer Name
    0.531s - DEBUG: Retrieving OS Name
    0.531s - DEBUG: Retrieving Graphics Info
    0.531s - DEBUG: Retrieving USB Info
    0.922s - DEBUG: Creating main window
    1.125s - DEBUG: Updating Baseline List
    1.141s - DEBUG: Finished updating Baseline List
    1.891s - DEBUG PERF: Initialize3DEnvironment
    3.859s - DEBUG: Checking language and local
    3.859s - DEBUG: Got language and local
    3.859s - DEBUG: Creating performance stats thread
    3.859s - DEBUG: Performance stats thread created
    3.859s - Initialize CD/DVD Burn
    3.875s - DEBUG: Initialising bitmap resources
    3.906s - DEBUG: Closing splash window
    3.969s - DEBUG PERF: Loading Perf DLL
    3.984s - DEBUG PERF: Found perf functions in DLL
    3.984s - DEBUG PERF: Loaded Perf DLL
    4.438s - DEBUG PERF: Host "ALEX-PC", Index 238, Counter 1850
    4.578s - DEBUG PERF: Assembling counter string1 ""
    4.578s - DEBUG PERF: Counter string1 "\Processor(_Total)\"
    4.578s - DEBUG PERF: Counter string2 "\Processor(_Total)\% Processor Time"
    4.578s - DEBUG PERF: Got counter string DLL
    4.578s - DEBUG PERF: Finished query
    4.609s - DEBUG PERF: Added counter

    Comment


    • #3
      There should be a SysInfoLog.txt file as well, could you please post or email that to us.

      After uninstalling could you try deleting the C:\ProgramData\PassMark\PerformanceTest directory to make sure any possible corrupt files have been removed.

      However, I just reproduced the same behavior so it looks like some of the system information collection is causing an issue in virtual machines (as they likely don't always respond as expected when trying to read information directly from the motherboard).

      You should be able to start PerformanceTest using the SAFEMODE command line parameter to skip the information collection phase (originally we couldn't get the tests to run but our VM had the wrong settings and wasn't correctly running as 64 bit).
      Last edited by Tim (PassMark); 03-24-2014, 12:01 AM. Reason: Additional info

      Comment


      • #4
        After uninstall C:\Program Files\PassMark does not exist (assume you mean "Program Files" rather than "ProgramData", which doesn't exist at all). SysInfoLog.txt pasted below. Note that I'm assigning a number of PCI devices including a graphics card, so it has several real devices.

        Regarding the BSOD, I've been able to isolate this to the CPU model reported to the VM. I'm using QEMU/KVM as my hypervisor and if I use either the -cpu options "host" or "Nehalem", which is the closest CPU model to physical hardware, I get a BSOD from the DirectIo64.sys driver provided with PT.

        Click image for larger version

Name:	passmark-performance-test-bsod.png
Views:	1
Size:	31.1 KB
ID:	34874

        Using SAFEMODE does avoid the BSOD. If there's anything I can do in the hypervisor to help avoid this, please let me know. At this point I'm inclined to call it a Passmark bug.

        The startup error is possibly due to file corruption as I installed PT and immediately ran it and encountered a BSOD. On a separate VM used to isolate the option that triggers the BSOD, I have no problem re-running PT and reproducing the BSOD. On the original VM that encountered this problem, I still have no idea how to make PT run-able.

        SysInfoLog.txt:
        0.000s - SysInfo: SysInfo v1.0 Build: 1055
        0.531s - SysInfo: OS1: 2 6.1.7601 Service Pack 1
        0.531s - SysInfo: OS2: 2 6.1.7601 Service Pack 1 1 1 0
        0.531s - SysInfo: Operating System: Windows 7 Ultimate Edition Service Pack 1 build 7601 (64-bit)
        0.531s - SysInfo: Video: \\.\DISPLAY1, NVIDIA Quadro K4000, PCI\VEN_10DE&DEV_11FA&SUBSYS_097C10DE&REV_A1
        0.531s - SysInfo: 51 0 75 0 61 0 64 0 72 0 6f 0 20 0 4b 0 34 0 30 0 30 0 30 0 0 0
        0.531s - SysInfo: String: Quadro K4000
        0.531s - SysInfo: 49 0 6e 0 74 0 65 0 67 0 72 0 61 0 74 0 65 0 64 0 20 0 52 0 41 0 4d 0 44 0 41 0 43 0
        0.531s - SysInfo: 0 0
        0.531s - SysInfo: String: Integrated RAMDAC
        0.531s - SysInfo: 56 0 65 0 72 0 73 0 69 0 6f 0 6e 0 20 0 38 0 30 0 2e 0 36 0 2e 0 35 0 33 0 2e 0 30 0
        0.531s - SysInfo: 2e 0 31 0 0 0
        0.531s - SysInfo: String: Version 80.6.53.0.1
        0.531s - SysInfo: Video: NVIDIA Quadro K4000, nvd3dumx,nvwgf2umx,nvwgf2umx, 3221225472, 9.18.13.3250
        0.531s - SysInfo: Video: \\.\DISPLAY2, NVIDIA Quadro K4000, PCI\VEN_10DE&DEV_11FA&SUBSYS_097C10DE&REV_A1
        0.531s - SysInfo: 51 0 75 0 61 0 64 0 72 0 6f 0 20 0 4b 0 34 0 30 0 30 0 30 0 0 0
        0.531s - SysInfo: String: Quadro K4000
        0.531s - SysInfo: 49 0 6e 0 74 0 65 0 67 0 72 0 61 0 74 0 65 0 64 0 20 0 52 0 41 0 4d 0 44 0 41 0 43 0
        0.531s - SysInfo: 0 0
        0.531s - SysInfo: String: Integrated RAMDAC
        0.531s - SysInfo: 56 0 65 0 72 0 73 0 69 0 6f 0 6e 0 20 0 38 0 30 0 2e 0 36 0 2e 0 35 0 33 0 2e 0 30 0
        0.531s - SysInfo: 2e 0 31 0 0 0
        0.531s - SysInfo: String: Version 80.6.53.0.1
        0.531s - SysInfo: Video: NVIDIA Quadro K4000, nvd3dumx,nvwgf2umx,nvwgf2umx, 3221225472, 9.18.13.3250
        0.531s - SysInfo: Video: \\.\DISPLAY3, NVIDIA Quadro K4000, PCI\VEN_10DE&DEV_11FA&SUBSYS_097C10DE&REV_A1
        0.531s - SysInfo: 51 0 75 0 61 0 64 0 72 0 6f 0 20 0 4b 0 34 0 30 0 30 0 30 0 0 0
        0.531s - SysInfo: String: Quadro K4000
        0.531s - SysInfo: 49 0 6e 0 74 0 65 0 67 0 72 0 61 0 74 0 65 0 64 0 20 0 52 0 41 0 4d 0 44 0 41 0 43 0
        0.531s - SysInfo: 0 0
        0.531s - SysInfo: String: Integrated RAMDAC
        0.531s - SysInfo: 56 0 65 0 72 0 73 0 69 0 6f 0 6e 0 20 0 38 0 30 0 2e 0 36 0 2e 0 35 0 33 0 2e 0 30 0
        0.531s - SysInfo: 2e 0 31 0 0 0
        0.531s - SysInfo: String: Version 80.6.53.0.1
        0.531s - SysInfo: Video: NVIDIA Quadro K4000, nvd3dumx,nvwgf2umx,nvwgf2umx, 3221225472, 9.18.13.3250
        0.531s - SysInfo: Video: \\.\DISPLAY4, NVIDIA Quadro K4000, PCI\VEN_10DE&DEV_11FA&SUBSYS_097C10DE&REV_A1
        0.531s - SysInfo: 51 0 75 0 61 0 64 0 72 0 6f 0 20 0 4b 0 34 0 30 0 30 0 30 0 0 0
        0.531s - SysInfo: String: Quadro K4000
        0.531s - SysInfo: 49 0 6e 0 74 0 65 0 67 0 72 0 61 0 74 0 65 0 64 0 20 0 52 0 41 0 4d 0 44 0 41 0 43 0
        0.531s - SysInfo: 0 0
        0.531s - SysInfo: String: Integrated RAMDAC
        0.531s - SysInfo: 56 0 65 0 72 0 73 0 69 0 6f 0 6e 0 20 0 38 0 30 0 2e 0 36 0 2e 0 35 0 33 0 2e 0 30 0
        0.531s - SysInfo: 2e 0 31 0 0 0
        0.531s - SysInfo: String: Version 80.6.53.0.1
        0.531s - SysInfo: Video: NVIDIA Quadro K4000, nvd3dumx,nvwgf2umx,nvwgf2umx, 3221225472, 9.18.13.3250
        0.531s - SysInfo: Video: \\.\DISPLAYV1, RDPDD Chained DD,
        0.531s - SysInfo: Video: \\.\DISPLAYV2, RDP Encoder Mirror Driver,
        0.531s - SysInfo: Video: \\.\DISPLAYV3, RDP Reflector Display Driver,
        0.531s - SysInfo: Video: , ,
        0.531s - SysInfo: USB Devices
        0.547s - SysInfo: USB HC dev: \\.\HCD0
        0.547s - SysInfo: USB HC driver key: {36fc9e60-c465-11cf-8056-444553540000}\0000
        0.563s - SysInfo: USB HC driver key device: Intel(R) ICH9 Family USB Universal Host Controller - 2934
        0.641s - SysInfo: USB Hub: RootHub
        0.641s - SysInfo: USB HC dev: \\.\HCD1
        0.641s - SysInfo: USB HC driver key: {36fc9e60-c465-11cf-8056-444553540000}\0001
        0.641s - SysInfo: USB HC driver key device: Intel(R) ICH9 Family USB Universal Host Controller - 2935
        0.719s - SysInfo: USB Hub: RootHub
        0.719s - SysInfo: USB HC dev: \\.\HCD2
        0.719s - SysInfo: USB HC driver key: {36fc9e60-c465-11cf-8056-444553540000}\0002
        0.719s - SysInfo: USB HC driver key device: Intel(R) ICH9 Family USB Universal Host Controller - 2936
        0.797s - SysInfo: USB Hub: RootHub
        0.797s - SysInfo: USB HC dev: \\.\HCD3
        0.797s - SysInfo: USB HC driver key: {36fc9e60-c465-11cf-8056-444553540000}\0003
        0.797s - SysInfo: USB HC driver key device: Intel(R) ICH9 Family USB2 Enhanced Host Controller - 293A
        0.797s - SysInfo: USB Hub: RootHub
        0.828s - SysInfo: USB Device detected: 3, Intel(R) ICH9 Family USB2 Enhanced Host Controller - 293A, 1, QEMU QEMU USB Tablet (SN: 42)
        0.828s - SysInfo: USB Device: QEMU QEMU USB Tablet (SN: 42)
        0.828s - SysInfo: USB3.0 HC: 0
        0.828s - SysInfo: USB3.0 HC: \\?\pci#ven_8086&dev_2934&subsys_11001af4&rev_03#3 &13c0b0c5&0&20#{3abf6f2d-71c4-462a-8a92-1e6861e6af27}
        0.844s - SysInfo: USB3.0 HC: 1
        0.844s - SysInfo: USB3.0 HC: \\?\pci#ven_8086&dev_2935&subsys_11001af4&rev_03#3 &13c0b0c5&0&21#{3abf6f2d-71c4-462a-8a92-1e6861e6af27}
        0.875s - SysInfo: USB3.0 HC: 2
        0.875s - SysInfo: USB3.0 HC: \\?\pci#ven_8086&dev_2936&subsys_11001af4&rev_03#3 &13c0b0c5&0&22#{3abf6f2d-71c4-462a-8a92-1e6861e6af27}
        0.891s - SysInfo: USB3.0 HC: 3
        0.891s - SysInfo: USB3.0 HC: \\?\pci#ven_8086&dev_293a&subsys_11001af4&rev_03#3 &13c0b0c5&0&27#{3abf6f2d-71c4-462a-8a92-1e6861e6af27}
        0.906s - SysInfo: USB Devices: 1

        Comment


        • #5
          Thanks for the log.

          ProgramData is a hidden system folder, you would need to enable the "Show hidden files, folders and drives" option in the windows explorer folder options, though if you paste C:\ProgramData\PassMark into an explorer address bar it should open the hidden folder without having to enable the option.

          Comment


          • #6
            Further testing shows:

            -cpu Penryn: works
            -cpu Nehalem: BSOD

            The feature differences between these are fairly negligible, popcnt & sse4.2. Removing these from the Nehalem model still results in BSOD. The difference seems to be the CPU model. The QEMU Penryn CPU is family 6, model 23, stepping 3 and Nehalem is the same except family 26. I can therefore get the following results:

            -cpu Penryn,model=26: BSOD
            -cpu Nehalem,model=23: works

            So Performance Test is doing something different based on the CPU model within family 6 that results in a BSOD. Nehalem is where Intel first moved the memory controller from the MCH on to the CPU, so perhaps PT is trying to probe this and getting unexpected results since the VM is still based on an i440FX platform model.

            Comment


            • #7
              Originally posted by Tim (PassMark) View Post
              Thanks for the log.

              ProgramData is a hidden system folder, you would need to enable the "Show hidden files, folders and drives" option in the windows explorer folder options, though if you paste C:\ProgramData\PassMark into an explorer address bar it should open the hidden folder without having to enable the option.
              Yes! Removing C:\ProgramData\PassMark allows it to re-install and work! Furthermore I was able to avoid the BSOD on this VM by switching to "-cpu Penryn,+popcnt,+sse4_2". Thanks and please let me know if I can assist further with the BSOD.

              Comment


              • #8
                There is a new test build of PerformanceTest available from http://www.passmark.com/ftp/petst8.0.1032.exe that should fix a crash on startup in VMs.

                Comment


                • #9
                  Originally posted by Tim (PassMark) View Post
                  There is a new test build of PerformanceTest available from http://www.passmark.com/ftp/petst8.0.1032.exe that should fix a crash on startup in VMs.
                  Behaves exactly the same for me, BSOD from the Passmark provided DirectIo64.sys driver if I specify Nehalem as the CPU type, works if I specify Penryn.

                  Comment


                  • #10
                    We don't see any crashes in VirtualBox or VMware but these are all using default settings and not trying to specify different CPU settings (we tried a windows build of Qemu but it didn't run properly).

                    If you are trying to change the CPU type to something other than what is in your system (what is the actual CPU in your system?) then it sounds like the hardware isn't being represented properly in the VM (and then it is more a emulated system than virtual environment ).

                    Comment


                    • #11
                      Originally posted by Tim (PassMark) View Post
                      We don't see any crashes in VirtualBox or VMware but these are all using default settings and not trying to specify different CPU settings (we tried a windows build of Qemu but it didn't run properly).

                      If you are trying to change the CPU type to something other than what is in your system (what is the actual CPU in your system?) then it sounds like the hardware isn't being represented properly in the VM (and then it is more a emulated system than virtual environment ).
                      Nope, my host CPU is a Xeon W3520 which ark identifies as a Bloomfield:

                      http://ark.intel.com/products/39718/...-GTs-Intel-QPI

                      And wikipedia confirms is Nehalem microarchitecture:

                      http://en.wikipedia.org/wiki/Bloomfi...icroprocessor)

                      So I'm trying to expose a CPU to the guest as close to the host CPU as possible for optimal performance. I don't know if you're going to be able to do this with QEMU for Window, but I sure wouldn't want to sit through a Windows guest install without the hardware virtualization provided by KVM. I'd suggest installing a Linux host system; pretty much any distribution has KVM support and virt-manager provides an easy way to install a guest and even configure the CPU type.

                      I think the fact that I can make PT induce a BSOD by changing the CPU family exposed, which means nothing to the hypervisor, strongly implicates PT as the problem. With a default CPU configuration QEMU/KVM also won't reproduce the problem, but that potentially leaves performance on the table if the guest can't see all the features of the processor.

                      Comment


                      • #12
                        We had another user report a similar problem yesterday. This user was also using qemu.
                        (Windows 7 Ultimate x86_64, which is virtualized with qemu-kvm on Gentoo Linux)

                        The solution to getting PerformanceTest working in qemu was,

                        "options kvm ignore_msrs=1" line (without quotes) should be added to
                        /etc/modprobe.d/kvm.conf
                        to prevent BSOD at program startup.

                        Our guess is that there are various unimplemented MSRs in qemu for certain CPU types. As part of the CPU emulation is missing from qemu you get a crash when the execution of the MSR is attempted.

                        Background information: A model-specific register (MSR) is any of various control registers in the x86 instruction set used for debugging, program execution tracing, computer performance monitoring, and toggling certain CPU features. PerformanceTest uses some of these to check for overclocking and similar CPU performance details. Turning all these off in qemu, might result in some of the system information reported by PerformanceTest being wrong.

                        Comment


                        • #13
                          The symptom got reintroduced in Windows 11. On AMD machine, when running PerformanceTest app, Windows 11 crashes after several seconds (the benchmark is not started yet, it is probably during initial machine check) with SYSTEM_SERVICE_EXCEPTION from DirectIo64.sys.

                          Comment

                          Working...
                          X