Strandedness of coverage

I was unable to determine how strandedness information is used in megadepth - like `featureCounts -s 1/2` parameter - to resolve overlapping gene ambiguity when mapping with a stranded RNA-seq protocol when `--coverage` is being used. Does megadepth use strandedness information, or is this all treated as unstranded? If the latter, how should we resolve stranded data for overlapping genes? 