It it only a chrM problem? Quick-n-dirty, modify the chromsizes file and set chrM to 16580, see if that works. Maybe some oddity with the chrM in that particular reference fasta they used? See if things make sense on genome browser of hg19.
Use the bedClip UCSC utility to cut the features down to chrom sizes. But double check before that that you're really on the correct assembly and that you have the correct file.
For hg19, for example. the UCSC browser now has both the original chrM sequence (which they picked way back before NCBI had picked a different reference chrM sequence) and the official mitochondria reference called "chrMT" that was selected by the GRCh. So maybe replacing "chrM" with "chrMT" will fix your problem here.
It's possible that this particular file was made with a genome assembly downloaded from e.g. NCBI or Ensembl and someone added the "chr" to the file, which could lead to this particular problem, a length difference on the mitochondrial sequence.
For many applications, you can also remove all the chrM annotations. For a lot of assays, e.g. RNA-seq and most Chip-seq experiments, the biologists don't care about the mitochondrial genome results so they can be removed.
Perhaps the data uses hg38 not hg19?
The hg19 is correct. In addition, I have used hg38, but showing the same problem.
It it only a chrM problem? Quick-n-dirty, modify the chromsizes file and set chrM to 16580, see if that works. Maybe some oddity with the chrM in that particular reference fasta they used? See if things make sense on genome browser of hg19.