Why ext4?

But that argument even applies to e2fsprogs itself. How do we know there isn't some weird combination of flags you can pass to tune2fs or mke2fs which somehow makes an evil filesystem? At that point, you would have to break e2fsprogs backcompat anyway. This is probably a lot less *likely* than the "some third-party tool generates a weird filesystem image" case, but some of those third-party tools are actually rather popular (e.g. the article mentions FreeBSD), and breaking them would also be unfortunate. So I'm not sure how much sense it makes to draw a bright line around e2fsprogs as the single point of compatibility.