Is there a specific reason for using signed ints in the API?

...plus LZ4_MAX_INPUT_SIZE being 2113929216 bytes.

Is there something architecturally crucial that prevents LZ4 from compressing larger (e.g. limited by unsigned 32-bit or even 64-bit) files?

It would be nice to increase this limit for some cases (though I guess I can compress into multiple consecutive chunks limited by max size)