RE: [fluka-discuss]: Core dump (subscript out of range)

From: Chris Theis <Christian.Theis_at_cern.ch>
Date: Fri, 25 Sep 2015 18:13:45 +0000

Hi Vadim,

from the coredump I deduce that you are using parentheses in your geometry. FLUKA supports parenteses in the input but internally it cannot handle nested structures but needs to "normalize" the boolen expression. This easily results in a huge number of terms. As there is no dynamic memory allocation in FORTRAN77 the statically allocated array dimension of 100.000 is exceeded. See:

http://www.fluka.org/web_archive/earchive/new-fluka-discuss/2701.html

The only way around is to simplify/change the region description yourself.

One general caveat: the available geometry editors in FLAIR or SimpleGeo are not based on the geometry kernel of FLUKA. They have a completely different design and do not require to normalize boolean expressions involving parentheses. Therefore, viewing the geometry with them will allow you to see if the geometry is valid but not if FLUKA can (technically) handle it!

Cheers
Chris


From: owner-fluka-discuss_at_mi.infn.it [owner-fluka-discuss_at_mi.infn.it] on behalf of Vittorio Boccone [dr.vittorio.boccone_at_ieee.org]
Sent: 25 September 2015 19:17
To: Vadim Talanov
Cc: FLUKA discussion list
Subject: Re: [fluka-discuss]: Core dump (subscript out of range)


Can be that in your routine you are not handling correctly large arrays?

I guess you are hitting the well know FORTRAN array limit.

From your Logfile:
Subscript out of range on file line 84, procedure rpnorm.f/rpnorm.
Attempt to access the 100001-th element of variable tx.

On Sep 25, 2015 11:48, "Talanov Vadim" <vadim.talanov_at_psi.ch<mailto:vadim.talanov_at_psi.ch>> wrote:
Dear FLUKA Experts,

Recently an attempt to run one of the FLUKA inputs with the fluka2011.2c-linux-gfor64bit distribution resulted in the segmentation fault and the core dump, with the following first lines in the .log file:

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:
#0 0x7FC051B7BA17
#1 0x7FC051B7C01E
#2 0x35AE43269F
#3 0x35AE4838A0
#4 0x7DCE82 in copytx_
#5 0x5FA9FE in rpnrul_
#6 0x5F90D7 in rpnorm_
#7 0x5E9EE7 in flkcgi_
#8 0x4E78CF in geoinp_
#9 0x408E8B in flukam_

Next attempt to run the same input from FLAIR in flupix-2011.2c-0_flair-2.1-5 resulted in the TIMED-OUT status with the first lines in the .log file:

Subscript out of range on file line 84, procedure rpnorm.f/rpnorm.
Attempt to access the 100001-th element of variable tx.

Two logs are attached. One could suspect that the cause is related to the geometry definition, however the geometry was checked in the FLAIR geometry viewer and no obvious errors were found.

Any hints and ideas are welcome,

Cheers,
Vadim


__________________________________________________________________________
You can manage unsubscription from this mailing list at https://www.fluka.org/fluka.php?id=acc_info
Received on Fri Sep 25 2015 - 20:13:45 CEST

This archive was generated by hypermail 2.3.0 : Fri Sep 25 2015 - 21:30:24 CEST