Project

General

Profile

Bug #17929

Segmentation fault when MergeWithinTPC "on"

Added by Robert Sulej about 2 years ago. Updated over 1 year ago.

Status:
Assigned
Priority:
Normal
Assignee:
Category:
Reconstruction
Target version:
-
Start date:
10/17/2017
Due date:
% Done:

0%

Estimated time:
Occurs In:
Experiment:
-
Co-Assignees:
Duration:

Description

I need a ticket before it's lost in emails.

Reported by Will:

In larsoft v06_53_00 / lariatsoft v06_53_00, I'm running a simple reco chain (caldata, gaushit, trajcluster, pmtracktc) over LArIAT data files and am noticing occasional seg faults when PMAlgTracking.MergeWithinTPC is turned on.

An example file where this happens is: /pnfs/lariat/data/digit/001/169/lariat_digit_r006245_sr0014_20160203T020152.root (event 539).

The output from the crash is below. Note that a "hit sorting problem" is first reported, and then a segmentation fault occurs shortly after "First hit is projected to inner segment". From looking into the code, I think the crash happens in RecoAlg/ProjectionMatchingAlg.cxx, in the function mergeTracks, line 1330 where ShiftEndsToHits is called.

==========================================================

%MSG-e ChannelFilter: CalWireROIT1034:caldata@ 16-Oct-2017 15:23:48 CDT run: 6245 subRun: 14 event: 539 ChannelFilter.cxx:49
ChannelFilter is now deprecated. Replace it with ChannelStatusService
%MSG
Using views in the following order:
%MSG-i PMAlgTracker: PMAlgTrackMaker:pmtracktc@ 16-Oct-2017 15:23:49 CDT run: 6245 subRun: 14 event: 539
1
%MSG
%MSG-i PMAlgTracker: PMAlgTrackMaker:pmtracktc@ 16-Oct-2017 15:23:49 CDT run: 6245 subRun: 14 event: 539
0
%MSG
Validation mode in config: hits
%MSG-i DataProviderAlg: PMAlgTrackMaker:pmtracktc@ 16-Oct-2017 15:23:49 CDT run: 6245 subRun: 14 event: 539
No plane-to-plane calibration.
%MSG
Downscale mode is: mean
%MSG-i DataProviderAlg: PMAlgTrackMaker:pmtracktc@ 16-Oct-2017 15:23:49 CDT run: 6245 subRun: 14 event: 539
No plane-to-plane calibration.
%MSG
Downscale mode is: mean
Sort hits by clusters...
...done, 5 clusters for 3D tracking.
Reconstruct tracks within Cryo:0 / TPC:0.
Find max cluster...
--- start new candidate ---
use view * 1 plane idx 1 size: 54
--> 0 match with:
cluster in view
0 size: 36
no validation plane

track size: 90
tpc:0
#coll:0 #ind2:54 #ind1:36
initialize trk
Track initialized with hit positions.
optimize trk (7 seg)
nodes done, g = 0.0105512
tune done, g = 0.00924048
good track candidate, MSE = 0.00683616, v = 1
no clusters to extend the track
merge clusters from the validation plane
--> 1 match with:
cluster in view
0 size: 8
no validation plane *
track size: 62
tpc:0
#coll:0 #ind2:54 #ind1:8
initialize trk
Track initialized with hit positions.
clusters do not match, f = 0.0481744
track REJECTED, MSE = 0; v = 0
--> 2 match with:
cluster in view *
0
size: 4
no validation plane

track size: 58
tpc:0
#coll:0 #ind2:54 #ind1:4
initialize trk
Track initialized with hit positions.
clusters do not match, f = 0.0518247
track REJECTED, MSE = 0; v = 0
no matching clusters
Find max cluster...
small clusters only
Find max cluster...
--- start new candidate ---
use view
1 plane idx 1 size: 15
--> 0 match with:
cluster in view
0 size: 8
no validation plane

track size: 23
tpc:0
#coll:0 #ind2:15 #ind1:8
initialize trk
Track initialized with hit positions.
optimize trk (3 seg)
nodes done, g = 0.00268423
tune done, g = 0.00223837
good track candidate, MSE = 0.00202608, v = 1
no clusters to extend the track
merge clusters from the validation plane
no matching clusters
Find max cluster...
--- start new candidate ---
use view
0 plane idx 0 size: 4
no matching clusters
Find max cluster...
small clusters only
Found tracks: 2
--- start new candidate ---
use view
1 plane idx 1 size: 5
--> 0 match with:
cluster in view
0 size: 4
no validation plane *
*
track size: 9
tpc:0
#coll:0 #ind2:5 #ind1:4
initialize trk
Track initialized with hit positions.
optimize trk (2 seg)
nodes done, g = 0.00212643
tune done, g = 0.00218901
good track candidate, MSE = 0.00207651, v = 1
no clusters to extend the track
merge clusters from the validation plane
no matching clusters
Add new track, cut hits from source track.
Merge co-linear tracks within TPC 0.
Merge track (85) with track (9)
%MSG-e pma::Track3D: PMAlgTrackMaker:pmtracktc@ 16-Oct-2017 15:23:49 CDT run: 6245 subRun: 14 event: 539
0 enabled hits in AverageDist2 calculation.
%MSG
%MSG-e pma::Track3D: PMAlgTrackMaker:pmtracktc@ 16-Oct-2017 15:23:49 CDT run: 6245 subRun: 14 event: 539
Track empty.
%MSG
reopt after merging done, g = 1e+10
%MSG-e pma::Track3D: PMAlgTrackMaker:pmtracktc@ 16-Oct-2017 15:23:49 CDT run: 6245 subRun: 14 event: 539
Hit sorting problem.
%MSG
%MSG-w pma::Track3D: PMAlgTrackMaker:pmtracktc@ 16-Oct-2017 15:23:49 CDT run: 6245 subRun: 14 event: 539
First hit is projected to inner segment.
%MSG
Segmentation fault

History

#1 Updated by Katherine Lato over 1 year ago

  • Assignee changed from Robert Sulej to Tracy Usher


Also available in: Atom PDF