On Thu, 2018-01-04 at 06:46 -0800, Dave Hansen wrote:
> On 01/04/2018 06:37 AM, David Woodhouse wrote:
> > KPTI complicates this a little; the one in entry_SYSCALL_64_trampoline
> > can't just jump to the thunk because the thunk isn't mapped. So it gets
> > its own copy of the thunk, inline.
> 
> This one call site isn't too painful, of course.
> 
> But, is there anything keeping us from just sticking the thunk in the
> entry text section where it would be available while still in the
> trampoline?

Not really. Since we have a thunk per register and the trampoline is
limited in size, if you wanted to put just the *one* thunk that's
needed into the trampoline then it's slightly icky, but all perfectly
feasible.