- marked as major
QuadrupoleStepper abort for reference particle in fieldmaps branch
Issue #135
resolved
In the fieldmaps branch the test 1_bunchtest_Reference fails with:
---> Begin of event: 0
Program received signal SIGABRT, Aborted.
0x00007fff861a1286 in __pthread_kill ()
(gdb) bt
#0 0x00007fff861a1286 in __pthread_kill ()
#1 0x00007fff8a6559f9 in pthread_kill ()
#2 0x00007fff8d0c1a4b in __abort ()
#3 0x00007fff8d0c2329 in __stack_chk_fail ()
#4 0x00000001001f42cf in BDSIntegratorQuadrupole::Stepper (this=0x1077488c0, yInput=0x7fff5fbfcd70, dydx=0x7fff5fbfd0d0, h=499.99999999900001, yOut=0x7fff5fbfcd10, yErr=0x7fff5fbfcdd0) at /Users/jsnuverink/Documents/bdsim/src/BDSIntegratorQuadrupole.cc:248
This is an indication of a stack overflow. I couldn't pinpoint it but this happens when the backupstepper is invoked in AdvanceHelix() (since the particle goes through the centre and sees no magnetic field).
Calling the backupstepper here might be a bit of an overkill since this does several steps on top of the steps from BDSIntegratorQuadrupole::Stepper while the advance will just be a linear step.
When replacing the backupstepper with the linear step like is done for kappa==0, there is no crash.
Comments (2)
-
reporter -
- changed status to resolved
Fix quadrupole integrator use of back up stepper, logic and tracking - resolves Issue
#135.→ <<cset 0a4bd33180c2>>
- Log in to comment
This issue currently prevents the reporting back to cdash for the fieldmaps branch. Since the fieldmaps branch is planned to merge soon into develop, I will upgrade its priority.