> > userspace (NO MEMCPY)
>
> Yes.  But it also means no readahead, because there's nowhere to _put_
> the readahead.  (Unless the application provides such a place, which we

with modern disk (which does readahead in hardware) or/and large blocks
it's no difference.

> agreed it shouldn't need to do.)

application can use O_ASYNC

>
> As for the memcpy, I'd rather memcpy from buffers than wait for a disk
> transfer - even a transfer that hits the disk's track cache.

i prefer waiting (=giving CPU to another process)