Analyzers
Talk to fellow users of Intel Analyzer tools (Intel VTune™ Profiler, Intel Advisor)

amplxe-gui : segmentation fault on startup

Alexis_R_
New Contributor I
1,926 Views

First time user of the amplifier here. Apologies if this is a FAQ.

When running amplxe-gui, the splash screen comes up for a split second, and I then get Segmentation fault.

$ amplxe-gui
Segmentation fault

This is a fresh install without root privileges. Here are some details about the install and the system (edited hostname):

$ uname -a
Linux host.name.edu 2.6.18-308.16.1.el5 #1 SMP Tue Sep 18 07:21:07 EDT 2012 x86_64 x86_64 x86_64 GNU/Linux

$ cat /etc/redhat-release
Red Hat Enterprise Linux Client release 5.8 (Tikanga)

$ cat /proc/cpuinfo | grep Xeon | tail -1
model name      : Intel(R) Xeon(R) CPU           X5690  @ 3.47GHz

$ amplxe-cl --version
Intel(R) VTune(TM) Amplifier XE 2013 Update 7 (build 288061) Command Line Tool

Please advise how to proceed from there or if you need more info. Thanks.

0 Kudos
54 Replies
Bernard
Valued Contributor I
1,058 Views

Can you post call stack info?

0 Kudos
Peter_W_Intel
Employee
1,058 Views

You need to use root privilege to install the tool.

0 Kudos
Alexis_R_
New Contributor I
1,058 Views

Peter Wang (Intel) wrote:

You need to use root privilege to install the tool.

Hi Peter,

The release notes say:

The install can be run as a non-root user, but in this case not all collectors will be available to the user.

Also, the installer allows for non-root installations.

I don't have root access & I would like to try out whatever features of Amplifier are usable. Is this not possible? If so, you have bugs in the documentation & the installer script.

0 Kudos
Alexis_R_
New Contributor I
1,058 Views

iliyapolak wrote:

Can you post call stack info?

Here is what I get when running amplxe-gui from gdb (I have replaced my install dir with ${INSTALL_DIR} in the output):

(gdb) run
Starting program: ${INSTALL_DIR}/vtune_amplifier_xe_2013/bin64/amplxe-gui
warning: no loadable sections found in added symbol-file system-supplied DSO at 0x2aaaaaaab000
[Thread debugging using libthread_db enabled]

Program received signal SIGSEGV, Segmentation fault.
0x00002aaaac536996 in cfgmgr2::ConfigDescriptorRegistry::ConfigDescriptorRegistry(cfgmgr2::config_type_t, char const*) () from ${INSTALL_DIR}/vtune_amplifier_xe_2013/bin64/../lib64/libamplxe_cfgmgr_2.15.so
(gdb) backtrace
#0  0x00002aaaac536996 in cfgmgr2::ConfigDescriptorRegistry::ConfigDescriptorRegistry(cfgmgr2::config_type_t, char const*) () from ${INSTALL_DIR}/vtune_amplifier_xe_2013/bin64/../lib64/libamplxe_cfgmgr_2.15.so
#1  0x00002aaaac536698 in cfgmgr2::IConfigDescriptorRegistry::create(cfgmgr2::config_type_t, gen_helpers2::sptr_t<cfgmgr2::IContextValueMap const>) ()
   from ${INSTALL_DIR}/vtune_amplifier_xe_2013/bin64/../lib64/libamplxe_cfgmgr_2.15.so
#2  0x00002aaab4288cea in perfclient_1_3::AnalysisTypeHolder::fill() () from ${INSTALL_DIR}/vtune_amplifier_xe_2013/lib64/libamplxe_client_stub_1.3.so
#3  0x00002aaab42825e9 in perfclient_1_3::IdeCallback::initialize(eil1::IDE_Env*, char const*) () from ${INSTALL_DIR}/vtune_amplifier_xe_2013/lib64/libamplxe_client_stub_1.3.so
#4  0x00002aaab4286c11 in StartUp () from ${INSTALL_DIR}/vtune_amplifier_xe_2013/lib64/libamplxe_client_stub_1.3.so
#5  0x0000000000467480 in ?? ()
#6  0x0000000000433abc in ?? ()
#7  0x0000000000450c56 in ?? ()
#8  0x000000000045146a in wxAppConsole::CallOnInit() ()
#9  0x00002aaaab3571b4 in wxEntry(int&, char**) () from ${INSTALL_DIR}/vtune_amplifier_xe_2013/bin64/../lib64/libwx_gtk2-2.8.so.0
#10 0x000000000044dc1b in ?? ()
#11 0x0000003822a1d994 in __libc_start_main () from /lib64/libc.so.6
#12 0x000000000041b3ea in wxHtmlHelpController::wxHtmlHelpController(int, wxWindow*) ()
#13 0x00007fffffffbb18 in ?? ()
#14 0x0000000000000000 in ?? ()

0 Kudos
Bernard
Valued Contributor I
1,058 Views

Do you have registers context?

0 Kudos
Alexis_R_
New Contributor I
1,058 Views

iliyapolak wrote:
Do you have registers context?

I'm sorry I don't know what you mean - could you explain what this is and how I get it?

0 Kudos
robert-reed
Valued Contributor II
1,058 Views

Very strange.  The stack trace you provided suggests that the blow-up you saw was in some sort of configuration manager,  cfgmgr2::ConfigDescriptorRegistry::ConfigDescriptorRegistry, which to me suggests the possibility that not all went right with the install.  And I realized that I had never installed VTune as a non-root user.  So I picked up a copy of the new Update 8 and attempted to install it on a virgin system, newly installed with RHEL 6.2 (not 5.8 but close).  I had to select the non-root install option and take care to move the installation directory to a writeable place with sufficient space (which on this machine is NOT my home directory), but everything repeated smoothly and I was able to run amplxe-gui without crashing.  

So I can't really conclude much from this exercise.  I was not able to reproduce the symptoms you describe but I don't have an equivalent machine to experiment with.  The only suggestion I can offer at this point is you try to uninstall and reinstall the package.  But I do have some further questions:  Are you able to run amplxe-cl for more rigorous activities than just getting the version number?  Is it possible that your RHEL 5.8 machine is missing some basic support libraries, like Xwindows stuff (needed for amplxe-gui)?  Are there any installation logs still in /tmp that might provide clues about a failed installation?  (Should be some time-stamped files prefaced with "intel." in /tmp that might provide some clues.)

0 Kudos
Alexis_R_
New Contributor I
1,058 Views

I uninstalled update 7 and downloaded & installed update 8.

During installation, I saw the following messages which may be relevant:

/tmp/install.KglE2O/li_plugin_bash.mnfXJV: line 431: ./insmod-sep3: Permission denied

and

Prerequisite(s)
-- Linux* OS kernel version is not suitable for Power analysis

Otherwise, things seemed to me to go smoothly during install. I have checked diskspace is available.

I think the machine has basic support libraries, especially when it comes to X since I run a whole bunch of graphical things on this workstation. But if you have specifics I could check, let me know.

I attach log files from the /tmp partition corresponding to the install of update 8. I notice in there that it says users should be part of vtune group, wich I haven't set up yet - but I guess that should not crash the gui. I attach the logs in case you can see something there.

I've not had a chance to test amplxe-cl yet.

0 Kudos
robert-reed
Valued Contributor II
1,058 Views

I've seen the insmod error on recent installs but it hasn't seemed to cause me any problems, so I think it's innocuous.  The power analysis warning is just because of the age of the kernel you're running on and shouldn't lead to the behavior that you're experiencing.  The Xlib stuff I was thinking about is all pretty basic and generic, so if you're used to running apps that use X graphics, that's probably not the issue.  It still would be interesting to know whether you can use amplxe-cl to collect data despite the condition of your installation, a useful exercise to narrow the cone of suspicion.  

I didn't see anything in the files you uploaded that raised any concern.  It looks like the install was successful (though I'm no expert at reading these files).  You might also try to generate a report using amplxe-feedback (in the bin32 directory), something like:

[bash]

<VTune install directory>/bin32/amplxe-feedback -create-bug-report=<your bug report name>

[/bash]

Upload that report and I can pass it along to the development team. Perhaps they can identify a cause from evidence I'm failing to grasp. 

The only other thing that pops into my head is licensing.  I once saw a problem in an older version of VTune Amplifier because of issues with the license files, either an inadequate one or one found in a directory among a bunch of expired licenses (details fuzzy now), which caused a crash.  But that bug should be long dead.

0 Kudos
Alexis_R_
New Contributor I
1,058 Views

Hi Robert,

So I tested amplxe-cl on the Fortran sample test which comes with the distribution (nqueens_parallel). Here's what happened:

$ amplxe-cl -collect hotspots nqueens_parallel
 Usage: threading_issues.exe boardSize
 Using default size of 10
Starting nqueens solver for size 10 with 4 thread(s)
Number of solutions: 724
Correct Result!
Calculations took 4ms.
amplxe: Using result path `/scratch/alr99/130606_ampl_test/nqueens_fortran/linux/r000hs'
Intel(R) VTune(TM) Amplifier XE 2013 (build 290588) feedback tool
Copyright (C) 2009-2013 Intel Corporation. All rights reserved.

Intel(R) VTune(TM) Amplifier XE 2013 Update 8 experienced an unexpected error.

Please send a problem report:
amplxe-feedback --send-crash-report "/tmp/amplxe-log-alr99/2013-06-06-Thu-09-32-24-766441.amplxe-cl/crash_info.txt"

Report data may be used to improve product stability. Our apologies for the inconvenience and thank you for your assistance.

Exception: 0xb, Segmentation&nbsp;fault
Module: libamplxe_cfgmgr_2.15.so
Collecting system information...
Collecting modules information...
Abort


I then ran the amplxe-feedback command as suggested - the txt files are attached to this post.

I do have a bunch of expired licenses in my license directory, but at least the ifort compiler doesn't complain about anything... Do you think it's worth attempting a cleanup of my license folder?

0 Kudos
robert-reed
Valued Contributor II
1,058 Views

It can't hurt to experiment with deleting expired licenses, though I have not seen this bug recently and so hold little hope that the solution will be that simple.  As long as you have the time (It will take a little time for me to file an internal escalation and get someone from the development team to look at this), it's one more possiblity to eliminate.

0 Kudos
David_A_Intel1
Employee
1,058 Views

Hi:

Did you install in your $HOME dir?  Do you have $TMP defined?  Are you executing the commands in a directory that is writeable?  I know you don't want to give personal info away, but without some details it's difficult to determine what is wrong.  BTW, for secure communications with the VTune Amplifier XE support staff, you can submit your problem to Intel(R) Premier Support.

0 Kudos
Alexis_R_
New Contributor I
1,058 Views

MrAnderson (Intel) wrote:

Hi:

Did you install in your $HOME dir?  Do you have $TMP defined?  Are you executing the commands in a directory that is writeable?  I know you don't want to give personal info away, but without some details it's difficult to determine what is wrong.  BTW, for secure communications with the VTune Amplifier XE support staff, you can submit your problem to Intel(R) Premier Support.

Yes, the install is under my $HOME directory.

I have $TMP defined

Yes, I'm running this in a directory that is writable.

I know you don't want to give personal info away, but without some details it's difficult to determine what is wrong.

I have given lots of details - what more do you need?

I don't mind about the personal information. The logs I attached in previous message are full of it.

BTW, for secure communications with the VTune Amplifier XE support staff, you can submit your problem to Intel(R) Premier Support.

Thanks. Where possible I like using the forums rather than Premier because that way the solution is shared with future visitors.

0 Kudos
David_A_Intel1
Employee
1,058 Views

Thanks!  So, when you ran the 'amplxe-feedback -create-bug-report <name>.zip' command, it should have created a zip file with lots of files in it.  One, in particular, woudl be the qfagent<suffix>.log file.  Can you either upload the full zip file or the qfagent log file?  Thanks.

0 Kudos
Alexis_R_
New Contributor I
1,058 Views

MrAnderson (Intel) wrote:

Thanks!  So, when you ran the 'amplxe-feedback -create-bug-report <name>.zip' command, it should have created a zip file with lots of files in it.  One, in particular, woudl be the qfagent<suffix>.log file.  Can you either upload the full zip file or the qfagent log file?  Thanks.

Unfortunately, here's what happens when I try to create a bug report:

$ amplxe-feedback --create-bug-report 130606_1537.zip
Intel(R) VTune(TM) Amplifier XE 2013 (build 290588) feedback tool
Copyright (C) 2009-2013 Intel Corporation. All rights reserved.
Collecting pids tree
sh: /tmp/amplxe-log-alr99/2013-06-06-Thu-15-40-51-190263.amplxe-feedback/qfagent.cmd: /usr/bin/perl^M: bad interpreter: No such file or directory

Of course, I do have perl there:

$ perl --version

This is perl, v5.8.8 built for x86_64-linux-thread-multi

Copyright 1987-2006, Larry Wall

Perl may be copied only under the terms of either the Artistic License or the
GNU General Public License, which may be found in the Perl 5 source kit.

Complete documentation for Perl, including FAQ lists, should be found on
this system using "man perl" or "perldoc perl".  If you have access to the
Internet, point your browser at http://www.perl.org/, the Perl Home Page.

$ which perl
/usr/bin/perl
$ ls -lh /usr/bin/perl
-rwxr-xr-x 2 root root 19K Oct 12  2012 /usr/bin/perl

I'm trying to sort this out now so I can get a log file to you. I suspect the ^M in the error message hints at some text file encoding issue, but no luck so far. Have you seen that issue before with the amplxe-feedback tool?


0 Kudos
David_A_Intel1
Employee
1,058 Views

That reminds me, which shell are you running?  I thought I saw tcsh in one of the previous files you provided.  Can you start up bash and try commands from that shell?

0 Kudos
Alexis_R_
New Contributor I
1,058 Views

MrAnderson (Intel) wrote:

That reminds me, which shell are you running?  I thought I saw tcsh in one of the previous files you provided.  Can you start up bash and try commands from that shell?

I use tcsh normally. Here is what happens when I do things under bash - as far as I can tell, exactly the same:

$ bash
bash-3.2$ source ~/intel/vtune_amplifier_xe/amplxe-vars.sh
Copyright (C) 2009-2013 Intel Corporation. All rights reserved.
Intel(R) VTune(TM) Amplifier XE 2013 (build 290588)
bash-3.2$ amplxe-cl -collect hotspots nqueens_parallel
 Usage: threading_issues.exe boardSize
 Using default size of 10
Starting nqueens solver for size 10 with 4 thread(s)
Number of solutions: 724
Correct Result!
Calculations took 30ms.
amplxe: Using result path `/scratch/alr99/130606_ampl_test/nqueens_fortran/linux/r003hs'
Intel(R) VTune(TM) Amplifier XE 2013 (build 290588) feedback tool
Copyright (C) 2009-2013 Intel Corporation. All rights reserved.

Intel(R) VTune(TM) Amplifier XE 2013 Update 8 experienced an unexpected error.

Please send a problem report:
amplxe-feedback --send-crash-report "/tmp/amplxe-log-alr99/2013-06-06-Thu-15-56-41-235633.amplxe-cl/crash_info.txt"

Report data may be used to improve product stability. Our apologies for the inconvenience and thank you for your assistance.

Exception: 0xb, Segmentation&nbsp;fault
Module: libamplxe_cfgmgr_2.15.so
Collecting system information...
Collecting modules information...
Aborted
bash-3.2$ amplxe-feedback --create-bug-report 130606_1557.zip
Intel(R) VTune(TM) Amplifier XE 2013 (build 290588) feedback tool
Copyright (C) 2009-2013 Intel Corporation. All rights reserved.
Collecting pids tree
sh: /tmp/amplxe-log-alr99/2013-06-06-Thu-15-57-17-153382.amplxe-feedback/qfagent.cmd: /usr/bin/perl^M: bad interpreter: No such file or directory


0 Kudos
Alexis_R_
New Contributor I
1,058 Views

A. Rohou wrote:

I'm trying to sort this out now so I can get a log file to you. I suspect the ^M in the error message hints at some text file encoding issue, but no luck so far. Have you seen that issue before with the amplxe-feedback tool?

I'm having no luck. It seems this .cmd file is being generated on the fly by amplxe-feedback tool, so I have no way of removing special characters from it or changing the encoding (it's encoded for DOS, not UNIX right now, which is why I think the line endings may be wrong).

Is there another way for me to get the log information to you?

EDIT: I was able to run the qfagent.cmd script manually after running it through the dos2unix utility. This means that the amplxe-feedback has a bug when running under UNIX - it generates the .cmd file with the wrong encoding.

I am trying to run this script manually now. WIll upload output in a separate post if I can get it to work.

0 Kudos
David_A_Intel1
Employee
1,058 Views

Thanks.  No, we have never seen this issue before.  These scripts all work fine for us and we haven't had a customer report this problem.  It will be interesting to get to the root cause of your issue. :)

0 Kudos
Bernard
Valued Contributor I
970 Views

A. Rohou wrote:

Quote:

iliyapolak wrote:Do you have registers context?

I'm sorry I don't know what you mean - could you explain what this is and how I get it?

Sorry for late answer.Register context = values contained in gp registers during segfault.

0 Kudos
Reply