L29Ah wrote on Tue, Apr 07, 2020: > In fact i would prefer disabling the full reads unconditionally, but > AFAIR some userspace programs might interpret a short read as EOF (and > also would need to check the logic that motivated the kernel-side > looping). Willy is correct there we can't just do that, way too many applications would break. I think O_NONBLOCK on regular files is a good compromise, let's not go overboard :) -- Dominique
