PCF data was not being read correctly, but I was incrementing the read pointer by the wrong amount (the offset and not the size).
Parsing some of the files, it just seems that the font data is going to be wasted for the most part. Like there are properties, other things and such. I think it would be best to just load the font, then draw something much simpler like a basic bitmap table of bytes or similar. Since much of the information is going to be dropped by SquirrelJME anyway. I think what is needed is a basic ascent and descent (in pixels). Each glyph would have its width in pixels (just a byte) then there would be the actual pixel data and such. That would be the fastest and simplest to load, would use the least amount of memory. Otherwise weaker systems are just going to stall heavily on font stuff. Everything can be pretty much enforced ISO that I use already. I think it is ISO-8859-15.