how to correctly preprocess data in spikeinterface before running kilosort

Hi,

I am confused about the preprocessing chain before running kilosort. In nearly all provided examples, 'highpass_filter', 'phase_shift' and 'common_reference' were run and sorters were called like:
sorting = si.run_sorter('kilosort2_5', rec, output_folder=base_folder / 'kilosort2.5_output', docker_image=True, verbose=True, **params_kilosort2_5)
Here 'car' and 'skip_kilosort_preprocessing' should be True and False by default, which means the filtering and common_reference were calculated again in kilosort. What is the meaning of the preprocessing in spikeinterface?

I am trying to do all preprocessing and drift correction using 'dreg' in spikeinterface before calling kilosort by:
sorting = si.run_sorter(sorter_name='kilosort2_5', recording=raw_rec, folder=r'F:\B12M2S10_g0\B12M2S10_g0_imec0\kilo25-3\sort', skip_kilosort_preprocessing=True, scaleproc=200, do_correction=False, docker_image=False, verbose=True)
The running failed, most likely due to the data not being whitened by setting skip_kilosort_preprocessing as True. Any suggestion about my pipeline?