Comment (by Zach Etienne):

A single timelevel of storage is allocated for Bvec in Convert_to_HydroBase & GiRaFFE_to_HydroBase, which are the only thorns that write to it.

Regarding ADMBase variables, they must first be set in CCTK_INITIAL outside IllinoisGRMHD, otherwise it doesn’t make sense to be using IllinoisGRMHD. The story is the same for vel[], rho, press, and eps gridfunctions – these must be set within CCTK_INITIAL or it doesn’t make sense. I.e., ID_converter_* assumes that ID already exist.

There was a similar issue in ID_converter_GiRaFFE, in which there was the same parameter (lazily copied from ILGRMHD) “pure_hydro_run”, but this is nonsense in the case of GRFFE (which assumes a magnetically dominated plasma), so I removed it.

Should be fixed in WVUThorns git hash:

6313730fe1a0588fcaeddfe1552f776ab9c7929a

--
Ticket URL: https://bitbucket.org/einsteintoolkit/tickets/issues/2394/segfault-in-id_converter_ilgrmhd-when-avec