Crashes with version 2016 update1

[David_Eberly]

Since installing Intel VTune Amplifier XE 2016 update1, I crash regularly when running the tool from within Visual Studio 2013.  I am running on WIndows 10.  I start the profiler, let it run for a while, and then stop it (doesn't matter whether I press the "stop" button or exit the program normally).  The event viewer logs have:

Faulting application name: amplxe-runss.exe, version: 0.0.0.0, time stamp: 0x561d8a1c
Faulting module name: ntdll.dll, version: 10.0.10240.16430, time stamp: 0x55c59f92
Exception code: 0xc0000374
Fault offset: 0x00000000000ea28c
Faulting process id: 0x1368
Faulting application start time: 0x01d10e849c7321c0
Faulting application path: C:\Program Files (x86)\IntelSWTools\VTune Amplifier XE 2016\bin64\amplxe-runss.exe
Faulting module path: C:\WINDOWS\SYSTEM32\ntdll.dll
Report Id: 25e07795-1769-43f7-b72d-07b350b86205
Faulting package full name: 
Faulting package-relative application ID:

I get a crash dump (about 10 MB), but I have little experience working with these to extract more information.  On the latest crash, I get the usual Windows dialog about the program having a problem and "searching for a solution".  When I dismiss the dialog, VTune continues to collect the trace information and finally displayed the results.  However, in other situations the trace is not collected, and in one case the UI where the collected information is displayed showed up as a solid black region.

Anyone else have similar problems?  Thanks.

[Vitaly_S_Intel]

Hi Dave,

I apologize for this issue.

Few questions:

1) Does crash happen if you start analysis not from Visual Studio, but using amplxe-gui ?

2) Which Analysis Type did you try? Does it happen on other analysis types (e.g. Basic Hotspots, Andvanced Hotspots, AH + stacks)?

3) Did you use previous VTune versions on this system? Did you see such crashes?

4) Will you please send me crash dump via private message?

Thanks a lot in advance for your help.

[David_Eberly]

Thank you for responding.

1) I was able to reproduce the crash using amplxe-gui.  It ran successfully when I terminated the program myself (app gets ESC key and terminates normally).  However, when I pressed the "Stop" button from the VTune menu (whether via VS 2013 or standalone amplexe-gui), I get the crash.  The program has a main thread.  It launches a separate thread to manage some stereo image processing.  In this separate thread I launch two threads, one per image of the stereo pair.  The program displays the results of the image processing in the main thread (via Direct3D 11).  After I press the "Stop" button, I can see the displayed images changing for a while (about 5 to 10 seconds), indicating that the program has not yet terminated.  After that, I get the Windows error dialog and the crash dump.

2) Analysis type is Basic Hotspots with the default 10 ms CPU sampling interval.  Since upgrading from Windows 8.1 to Windows 10, I have been unable to run advanced hotspots with stacks, context switches and/or call counts.  When I select Advanced Hotspots, say, with "Hotspots, stacks and context switches", I get an error message "Cannot enable advanced capabilities for Hardware Event-based Sampling: problem with the driver (vtss/vtsspp).  Check that the driver is running and the driver group is in the current user group list. ...".  I had tried suggestions mentioned in the help files but was unable to get this to work.

3) I have not seen the crash when running with VTune XE 2016 pre-update 1 or with VTune XE 2015.

4) I will send you two dumps.  The amplxe-runss.exe.46776.dmp is the one associated with the crash running amplxe-gui stand alone.  The amplxe-runss.exe.4968.dmp is the one associated with the crash running from the VS 2013 environment.

[Vitaly_S_Intel]

Thanks Dave for detailed info and crash dumps.

Can you reproduce this issue with Advanced Hotspots with the first "Hotspots" option (no stacks)?

[David_Eberly]

Hello Vitaly.

No, I cannot run that analysis either.  The Collection Log says "Collection failed."  The critical message is "PMU resources are not available.  Hardware Event-based Sampling is not supported on this operating system.  PMU resources may be made unavailable either by the BIOS or Hypervisor"

Dave

[Vitaly_S_Intel]

Well, this is likely due to Hyper-V enabled on your system - it is not virtualizing PMU.

You can disable Hyper-V and try again, please look in Troubleshooting Help section related to this error:

https://software.intel.com/en-us/node/596775

Scroll page down to the end to find description of this error.

[David_Eberly]

The machine is a new Dell workstation which appears to have shipped with Hyper-V enabled.  I disabled Hyper-V and then had to repair the network driver which was somehow tied to using Hyper-V.  I ran VTune using Basic Hotspots, stopped the program via the Stop button on the VTune interface and did not see a crash dialog.

However, I tried Advanced Hotspots with stacks and context switches.  In a few seconds, my machine locked up and then rebooted.  The Event Viewer showed quite a few error messages and there was a minidump.  The first two errors were about trying to open a logfile: C:<path>AppData\Local\Microsoft\Windows\SettingSync\metastore\edb.log ("An attempt to open the file for read only access failed with system error 32. The process cannot access the file because it is being used by another process."  Many errors after that were about failure to create C:\Windows\system32\edbtmp.log ("Access is denied." and "Unable to create a new logfile because the database cannot write to the log drive.  The drive may be read-only, out of disk space, misconfigured, or corrupted."  If I try to view "Administrative Events", I get a dialog entitled "Query Error" with "One or more logs in the query have errors".  The "Log" entries start with Microsoft-Windows-HyperV- but I don't know whether these are from when Hyper-V was enabled.

Any suggestions on how to fix this new problem?  Sorry, virtualization has not been part of my development world...

Thanks.

[David_A_Intel1]

Hi Dave:

Check out this page in the product help and follow the instructions to ensure that the drivers for Advanced Hotspots are loaded.

[David_Eberly]

MrAnderson (Intel) wrote:

Hi Dave:

Check out this page in the product help and follow the instructions to ensure that the drivers for Advanced Hotspots are loaded.

Thanks.  Yes, the drivers are loaded.  As indicated by Vitaly, I had to turn off Hyper-V.  But now the system lock and reboot is a problem...

[David_A_Intel1]

but after the reboot, the drivers check out okay?

The best thing for you to do is to submit an issue at Intel® Premier Support and provide the output of the 'amplxe-feedback -create-bug-report report.zip' command.  If your copy is registered, then you should have an account on Intel Premier Support.  We basically need all your system info to troubleshoot and it is preferred that you don't post it to this public, unsecured forum.

[David_Eberly]

I ran the commands mentioned in your link after your posted your response, and this was indeed after the reboot. he errors about edb.log occur regularly in Event Viewer, so probably irrelevant.  There was a minidmp (375K), but in fact there was also a full Memory.dmp (1 GB) with WER information.  Thanks for mentioning about the feedback report.

[David_Eberly]

MrAnderson (Intel) wrote:

but after the reboot, the drivers check out okay?

The best thing for you to do is to submit an issue at Intel® Premier Support and provide the output of the 'amplxe-feedback -create-bug-report report.zip' command.  If your copy is registered, then you should have an account on Intel Premier Support.  We basically need all your system info to troubleshoot and it is preferred that you don't post it to this public, unsecured forum.

I clicked on your Intel Premier Support link but get a message about not allowed to access.  According to my records I have such support.  I'll try to go through a different path, thanks.  I don't see amplxe-feedback in my installed distribution, so I imagine that is something I download from IPS.

[David_A_Intel1]

Hi Dave:

First, regarding Premier Support, please try scrolling to the bottom right of the page that says no access and see if there is an "Accept" button.  If so, press it.

Second, amplxe-feedback.exe should be in both the bin32 and bin64 folders of your VTune Amplifier installation.

[David_Eberly]

Clicking on the link in the copy of your post sent to me by email launches a login dialog.  Regarding the feedback program, I had sorted by name.  I failed to notice that the DLLs use an underscore after amplxe but the EXEs use a dash.  Just had to scroll down further to find it.

My workstation is a Dell T1700 (has BIOS A16).  I tried the same experiment on another Dell T1700 (has BIOS A15) but with VTune XE 2015 that shipped with Parallel Studio XE 2015 Update 4.  Both machines have Windows 10 (10.0.10240).  The Advanced Hot Spot profiling on this other machine worked fine.

Thanks for your help.  I will contact Premier Support to resolve the issue.

[David_Eberly]

Spoke too soon about that Intel Premier Support link.  It is the standard login, after which I get the "not allowed to access" error.  There is no "accept" at the bottom of the page.  I've sent an email asking what I need to do to access IPS.