SIGSEGV: segmentation violation

Message boards : Number crunching : SIGSEGV: segmentation violation
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile moi

Send message
Joined: 25 Dec 14
Posts: 11
Credit: 433,196
RAC: 0
France
Message 1909569 - Posted: 30 Dec 2017, 10:39:19 UTC
Last modified: 30 Dec 2017, 10:57:11 UTC

Hello,

I have this on my clear linux 4.14.9

https://setiathome.berkeley.edu/result.php?resultid=6264400750
<core_client_version>7.2.42</core_client_version>
<![CDATA[
<message>
process got signal 11
</message>
<stderr_txt>
SIGSEGV: segmentation violation

</stderr_txt>
]]>


What i could do ?
I try to add vsyscall emulation, may be i fail :s

Other ideas ?

Thx guy
ID: 1909569 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 9992
Credit: 130,964,051
RAC: 83,469
Australia
Message 1909573 - Posted: 30 Dec 2017, 11:08:54 UTC - in response to Message 1909569.  
Last modified: 30 Dec 2017, 11:11:04 UTC

May be of use.
https://github.com/BOINC/boinc/issues/2120
https://stackoverflow.com/questions/29377245/how-to-check-the-vsyscall-mode
"Your analysis is correct. For kernel 4.11 and newer you need to set the kernel option vsyscall=emulate in order to run applications compiled with older glibc."
Grant
Darwin NT
ID: 1909573 · Report as offensive
TBar
Volunteer tester

Send message
Joined: 22 May 99
Posts: 4078
Credit: 239,849,595
RAC: 225,982
United States
Message 1909598 - Posted: 30 Dec 2017, 16:15:18 UTC - in response to Message 1909569.  

Hello,
I have this on my clear linux 4.14.9
https://setiathome.berkeley.edu/result.php?resultid=6264400750
<core_client_version>7.2.42</core_client_version>
<![CDATA[
<message>
process got signal 11
</message>
<stderr_txt>
SIGSEGV: segmentation violation
</stderr_txt>
]]>


What i could do ?
The Stock Apps are built with an older OS to ensure compatibility. Since you are running a New OS, with a New Processor, you should try a New CPU App built with a Newer OS. There are a few around, some here, http://lunatics.kwsn.info/index.php?action=downloads;cat=48 If you are able to manually configure an app_info.xml you can even try the Latest AVX2 CPU App from here, TEST for Linux AVX2 CPU. I suppose I should change that post at some point, the 'Test' is over, the Apps work very well on the machines running them. Here's one machine running the Intel version, AVX2jf Linux64 Build 3712
ID: 1909598 · Report as offensive
Profile moi

Send message
Joined: 25 Dec 14
Posts: 11
Credit: 433,196
RAC: 0
France
Message 1909645 - Posted: 30 Dec 2017, 19:56:56 UTC - in response to Message 1909598.  
Last modified: 30 Dec 2017, 20:35:18 UTC

I dont fix my "SIGSEGV: segmentation violation" .
I rebuilt the kerne with intel clear linux dev aids :s but it dont work

I going to try this TBar
ID: 1909645 · Report as offensive
Profile moi

Send message
Joined: 25 Dec 14
Posts: 11
Credit: 433,196
RAC: 0
France
Message 1909680 - Posted: 30 Dec 2017, 22:27:23 UTC - in response to Message 1909645.  

This working thank with AVX,
Next step try AVX2 or SE4.2
ID: 1909680 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 4616
Credit: 297,377,037
RAC: 617,646
United States
Message 1909688 - Posted: 30 Dec 2017, 22:47:15 UTC - in response to Message 1909680.  
Last modified: 30 Dec 2017, 22:50:23 UTC

Great news. I've done some benchmark testing for TBar with the AVX2 app and compared it to the AVX and SSE4.1 app. Marginally faster on AMD hardware. Probably much faster on Intel hardware since Intel kinda has driven all the SSE improvements over the years. I also believe, may be mistaken, that Intel engineers specific hardware paths for AVX2 which should prove faster.
Seti@Home classic workunits:20,676 CPU time:74,226 hours
ID: 1909688 · Report as offensive
TBar
Volunteer tester

Send message
Joined: 22 May 99
Posts: 4078
Credit: 239,849,595
RAC: 225,982
United States
Message 1909699 - Posted: 30 Dec 2017, 23:23:06 UTC - in response to Message 1909688.  

The Problem with AVX2 on AMD is only ONE AMD Processor supports AVX2, that being Ryzen. To compile an Optimized App for Ryzen you Must use GCC 6.3 or higher, otherwise you have to compile for Bulldozer with doesn't have AVX2. The current SETI Software/Code doesn't work with any GCC version higher than 4.9.2, meaning the AMD AVX2 App is compiled for Bulldozer, which doesn't have AVX2...the compiler knows this BTW. I'm surprised AVX2 even works in that AMD App, by all rights it shouldn't. Maybe once the SETI Code works with GCC 6.3 it will be possible to compile a True AMD AVX2 App. The Intel App is different, AVX2 has been around for quite a while in Intel Processors and there isn't any problem compiling an Intel App for AVX2. That is the difference.
ID: 1909699 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 4616
Credit: 297,377,037
RAC: 617,646
United States
Message 1909740 - Posted: 31 Dec 2017, 2:53:49 UTC - in response to Message 1909699.  

The Problem with AVX2 on AMD is only ONE AMD Processor supports AVX2, that being Ryzen. To compile an Optimized App for Ryzen you Must use GCC 6.3 or higher, otherwise you have to compile for Bulldozer with doesn't have AVX2. The current SETI Software/Code doesn't work with any GCC version higher than 4.9.2, meaning the AMD AVX2 App is compiled for Bulldozer, which doesn't have AVX2...the compiler knows this BTW. I'm surprised AVX2 even works in that AMD App, by all rights it shouldn't. Maybe once the SETI Code works with GCC 6.3 it will be possible to compile a True AMD AVX2 App. The Intel App is different, AVX2 has been around for quite a while in Intel Processors and there isn't any problem compiling an Intel App for AVX2. That is the difference.

I remember you mentioning this before. Granted, I only did about 8 hours of benchmark testing between the 3 choices, but the AVX2 app came out a few percent faster than the AVX app which which works on Bulldozer and Piledriver. I realize the sample population was way too low, it could have been just a lucky mix of work units. But at least I don't see a PENALTY for running the AVX2 compiled app.

Will just cross my fingers and toes for a BOINC that works with GCC 6.3 in the future and bide my time.
Seti@Home classic workunits:20,676 CPU time:74,226 hours
ID: 1909740 · Report as offensive
Profile MarkJ Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 17 Feb 08
Posts: 1074
Credit: 53,736,336
RAC: 51,648
Australia
Message 1909741 - Posted: 31 Dec 2017, 3:14:05 UTC

If they updated the Seti apps to use GCC 6 then the OP wouldn’t need the the vsyscall=emulate, at least for Seti.

The Einstein developers recently rebuilt one of their apps using libc 2.15 (from around 2012) to work around the issue. I would have thought they would update to a later version rather than going to even older code, but maybe it’s too hard.
BOINC blog
ID: 1909741 · Report as offensive
Profile Karsten Vinding
Volunteer tester

Send message
Joined: 18 May 99
Posts: 233
Credit: 20,842,904
RAC: 1,430
Denmark
Message 1909940 - Posted: 1 Jan 2018, 2:03:20 UTC

I have been trying to run seti on my NAS computer that runs based on Centos/Fedora.

I am able to run e.g. World Community grid and Rosetta, but with Seti, after some time, the crunching starts to fail.

In the results for the WU's I get this:

terminate called after throwing an instance of 'std::bad_alloc'
what(): std::bad_alloc
SIGABRT: abort called
Stack trace (12 frames)

Is this the same problem as described in this thread?
ID: 1909940 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 4616
Credit: 297,377,037
RAC: 617,646
United States
Message 1909951 - Posted: 1 Jan 2018, 2:52:49 UTC

I don't think so. I think your error is a different type of memory allocation error. I'd like others to comment with any insight they might have.
Seti@Home classic workunits:20,676 CPU time:74,226 hours
ID: 1909951 · Report as offensive

Message boards : Number crunching : SIGSEGV: segmentation violation


 
©2018 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.