Re: [PATCH 10/24] x86/mm/kaiser: Map the entry stack variables

From: Peter Zijlstra
Date: Mon Nov 27 2017 - 16:00:33 EST


On Mon, Nov 27, 2017 at 06:32:11PM +0100, Thomas Gleixner wrote:
> On Mon, 27 Nov 2017, Peter Zijlstra wrote:
>
> > On Mon, Nov 27, 2017 at 11:49:09AM +0100, Ingo Molnar wrote:
> > > diff --git a/arch/x86/kernel/process_64.c b/arch/x86/kernel/process_64.c
> > > index 157f81816915..631e229ab428 100644
> > > --- a/arch/x86/kernel/process_64.c
> > > +++ b/arch/x86/kernel/process_64.c
> > > @@ -59,7 +59,7 @@
> > > #include <asm/unistd_32_ia32.h>
> > > #endif
> > >
> > > -__visible DEFINE_PER_CPU(unsigned long, rsp_scratch);
> > > +__visible DEFINE_PER_CPU_USER_MAPPED(unsigned long, rsp_scratch);
> > >
> > > /* Prints also some state that isn't saved in the pt_regs */
> > > void __show_regs(struct pt_regs *regs, int all)
> >
> > Do we still need this one? It appears to me we'll never actually execute
> > that part of entry_SYSCALL_64 anymore, due to
> > entry_SYSCALL_64_trampoline.
>
> Unless we use entry_SYSCALL_64 when kaiser is disabled at boottime

Right for which you need that patchlet I send earlier today. The patches
as they are will never ever reach that code.