Page 1 of 2
SIGSEGVProgram Crash with FTABLE Element
Posted: 05 Dec 2017, 16:43
by Roussel
Hi,
I am incurring SIGSEGVProgram issues when using the FTABLE element for tracking. I have some success resolving this issue when altering the number of particles in the bunch but this is very inconsistent and the issue is sporadic.
I have attached the relevant files. Please let me know what you guys think,
Thanks!
Re: SIGSEGVProgram Crash with FTABLE Element
Posted: 05 Dec 2017, 17:14
by michael_borland
I wasn't able to duplicate the problem. Can you verify that you are using the latest release, which is 34.0.1?
--Michael
Re: SIGSEGVProgram Crash with FTABLE Element
Posted: 05 Dec 2017, 17:20
by Roussel
Thanks for the response, I am currently running 34.0.0. This issue is intermittent so it is difficult to reproduce. I initially thought it was an issue with processes on my computer outside of elegant, but the problem has persisted across several computers.
Re: SIGSEGVProgram Crash with FTABLE Element
Posted: 05 Dec 2017, 17:22
by Roussel
I just now tried to reproduce the issue at 10001 particles but it went fine. I changed the number of particles to 50001 and it reproduced the error.
Re: SIGSEGVProgram Crash with FTABLE Element
Posted: 05 Dec 2017, 17:41
by michael_borland
I still don't see the crash, even with 50001 particles. What platform are you using?
--Michael
Re: SIGSEGVProgram Crash with FTABLE Element
Posted: 06 Dec 2017, 04:28
by petrenko
Dear Michael,
We are also encountering a very similar problem!:
> (Windows 10)
> C:\tmp>elegant awake.ele > elegant.txt
> ...
> This is elegant 34.0.0, Oct 31 2017, by M. Borland, J. Calvey, ...
> ...
> Link date: Oct 31 2017 16:49:48, SVN revision: 24662M
> ...
> generating bunch 1
> tracking 500000 particles
> 6 Dec 17 10:13:13: This step establishes energy profile vs s (fiducial beam).
> 6 Dec 17 10:13:13: Rf phases/references reset.
>
> Terminated by SIGSEGVProgram trace-back:
It seems to be triggered by a watchpoint output even in the simplest beamline (for number of particles ~50000-1000000):
W1: WATCH,FILENAME="w1.sdds"
AWAKE: line=(W1)
(the w1.sdds is saved correctly though)
For the number of particles below 10000 this error also occurs from time to time (every 2nd-3rd time, not regularly).
The files are attached.
Best regards,
/Alexey
Re: SIGSEGVProgram Crash with FTABLE Element
Posted: 06 Dec 2017, 11:51
by michael_borland
Alexey,
Thanks for the additional clue. We'll try to hunt this down.
--Michael
Re: SIGSEGVProgram Crash with FTABLE Element
Posted: 06 Dec 2017, 12:42
by Roussel
If it helps at all I'm running elegant on windows 7 using the most recent release of cygwin. It also might be the case that it has nothing to do with the FTABLE element as the error happens after a watch point.
Re: SIGSEGVProgram Crash with FTABLE Element
Posted: 06 Dec 2017, 12:43
by soliday
The problem has been found. It is now just a matter of recompiling for all the distributions.
Re: SIGSEGVProgram Crash with FTABLE Element
Posted: 06 Dec 2017, 16:55
by soliday
The crash was from a bug in the SDDS libraries.
I posted a new elegant release (34.0.1) which I built against a newer version of SDDS.