segmentation violation during DST writing after selection with `DTF_FUN`
This seg fault happens very rarely, and is holding back the current stripping campaigns.
It was noticed in 2016 data with the
Xic2LambdaKPi module, but it's likely that also other lines are affected.
The seg fault occurs when trying to write the DST after a candidate for
StrippingXic2LambdaKPi_XicpLine was found. The code runs fine when both
DTF_FUN cuts of the mother combination are removed or replaced by their non-DTF equivalents.
Possible source of error
It seems as if a vertex has been modified/removed? by the DTF, such that the link provided to the DST writers points nowhere.
How to move forward?
The quick and easy solution is to remove cuts on
DTF_FUN functors. This is not a big issue for the
Xic2LambdaKPi module, but would certainly harm the
XicToSigmaKPi lines which rely heavily on the DTF (see also the discussion in https://groups.cern.ch/group/lhcb-reconstruction/Lists/Archive/Flat.aspx?RootFolder=%2Fgroup%2Flhcb%2Dreconstruction%2FLists%2FArchive%2FReconstruction%20of%20Sigma%2B%20%2D%20p%20pi0&FolderCTID=0x0120020030FF8979BD9D1640BD2549BA2063796A). I would like to keep the Sigma lines, since they give a hint whether or not it is feasible to write trigger lines for them in Run3.
Having said that, it would be much appreciated if @ibelyaev and @wouter have a look, to possibly find and fix the source of the problem.
More (random) information
- The event that causes the trouble is a
\Xi_c^+ \to \Lambda K_S^0 \pi^+candidate. In that event, the
\Lambdais DD and
- Switching off the
dstWritersequence (as it's done in the users tests) doesn't cause a seg fault.
- Log files:
liaisons testing script on the problematic RDST running only
... same in DEBUG mode
... and switching off
dstWritersequence. The selection that passes here is the one triggering the seg fault