You may want to copy /afs/cern.ch/user/w/wmurray/public/tb95/car/f128an.car, and change every F128 into the relevant identifier for your data. There will be common blocks (F128CO) which are the wrong sizes for your data, and of course the decoding will be wrong. Then you can optimize the routines for your own data as required.
Any histo's booked in ABCD_SOR will be automatically saved; please do not fiddle with directories; you are put in '/ABCD' so there is no conflict with other histograms.
There is a routine 'F2FHIT', which finds hits and stores them in a common block. This common block is written out as an ntuple, and any hits here will be available for the alignment, so it makes sense to fill this if possible.
Then change and run strpjob.make to make the executable, strpjob.exe.
STRP_SOR, STRP_EOB etc go through the record and call e.g. ABCD_EOB if there is ABCD data present.