Re: [RFC][CFT][PATCHSET v1] uaccess unification

[Linus Torvalds]

On Thu, Mar 30, 2017 at 11:54 AM, Al Viro <viro@xxxxxxxxxxxxxxxxxx> wrote:
>
> Not even that - again, it will happily trigger page faults unless the
> caller disables those.  __copy_from_user_I_know_what_I_am_doing()?

That's a horrible name. Everybody always thinks they know what they are doing.

There's a reason I called the new odd user access functions
"unsafe_get/put_user()"

But regardless of that, I think you're being silly to even look at the
iovec code. That code simply *isn't* critical enough that one or two
extra instructions matter.

Show me profiles to the contrary. I dare you.

Those things shouldn't be using *anything* odd at all. They should be
using "copy_from_user()". Nothing else.

                   Linus