GSI Forum: Bugs, Fixes, Releases » Crash in genfit -> KalmanTask 15/3/2010

Home » PANDA » PandaRoot » Bugs, Fixes, Releases » Crash in genfit -> KalmanTask 15/3/2010

Show: Today's Messages :: Polls :: Message Navigator

Crash in genfit -> KalmanTask 15/3/2010 [message #10399]

Mon, 15 March 2010 10:28

StefanoSpataro
Messages: 2736
Registered: June 2005
Location: Torino

first-grade participant

From: *to.infn.it

Dear all,
I have found that in my dashboard the macro reco_complete_tpc crashes (and I am not the only one).

In particular, the error comes from KalmanTask, therefore tpc genfit. If you execute the script by dashboard, you have an error. If you execute the macro by normal root reco_complete_tpc.C, the program is quitting root.
Is it possible to fix it?
I suppose there is something not protected in the kalman task, which is giving a bad track to genfit and the exception is not correctly handled (trying to guess, I have never used KalmanTask).

The error could not be seen in the past because before there was a gem bug covering it (fixed since a while), then maybe the error exists since a while.

Have a nice fixing, I copy the gdb error trace.

Toggle Spoiler

Report message to a moderator

Re: Crash in genfit -> KalmanTask 15/3/2010 [message #10400 is a reply to message #10399]

Mon, 15 March 2010 11:18

Anonymous Poster

From: *e18.physik.tu-muenchen.de

Hi,

this is a floating point exception in GEANE. Could the GEANE experts please take a look?

Cheers, Christian

Report message to a moderator

Re: Crash in genfit -> KalmanTask 15/3/2010 [message #10401 is a reply to message #10400]

Mon, 15 March 2010 12:53

Lia Lavezzi
Messages: 291
Registered: May 2007
Location: Torino

first-grade participant

From: *pv.infn.it

Ok,
I will try to run the macros and see if I get the crash too. I have to run them without making any change, right?
Ciao,
Lia.

Report message to a moderator

Re: Crash in genfit -> KalmanTask 15/3/2010 [message #10405 is a reply to message #10401]

Mon, 15 March 2010 18:16

Lia Lavezzi
Messages: 291
Registered: May 2007
Location: Torino

first-grade participant

From: *pv.infn.it

Hi Stefano,
the problem was once again an exponent too small (10^-39) in erland.F. Please try the fixed routine I attach here (you should put it in geant3/erdecks/ and compile geant3 again).

The crash GEANE related disappears, but I get this new one Rolling Eyes

Program received signal SIGFPE, Arithmetic exception.
[Switching to Thread -1208796960 (LWP 656)]
0x03f34745 in PndMvdStripClusterTask::CalcMeanCharge (this=0x9efba60, onecluster=@0xbfef2670,
    meanstrip=@0xbfef2518, meanerr=@0xbfef2510, charge=@0xbfef2520)
    at /home/lavezzi/test_dev/original/trunk6/mvd/MvdReco/PndMvdStripClusterTask.cxx:485
485                       meanstrip = meanstrip/charge;

If you confirm that it is ok I will insert it in the new patch we are preparing which should also fix the missing virtual planes problem (it is not ready yet, I am still working on it with Christian).

Please let me know.
Ciao,
Lia.

Attachment: erland.F
(Size: 6.64KB, Downloaded 241 times)

Report message to a moderator

Re: Crash in genfit -> KalmanTask 15/3/2010 [message #10406 is a reply to message #10405]

Mon, 15 March 2010 18:37

StefanoSpataro
Messages: 2736
Registered: June 2005
Location: Torino

first-grade participant

From: *to.infn.it

Hi,
yes it works. The problem has appeared using electrons ad particle hyp for the kalman, also in another macro. Now both are not crashing anymore (neither the mvd crash you are reporting Surprised