On Mon, Aug 15, 2016 at 03:27:08PM +0200, Julien Grall wrote:
> Hi Konrad,
>
> On 12/08/2016 22:50, Konrad Rzeszutek Wilk wrote:
> > On Fri, Aug 12, 2016 at 05:00:47PM +0200, Julien Grall wrote:
> > > > diff --git a/xen/include/asm-arm/current.h
> > > > b/xen/include/asm-arm/current.h
> > > > index 65c0cdf..f4fcfd6 100644
> > > > --- a/xen/include/asm-arm/current.h
> > > > +++ b/xen/include/asm-arm/current.h
> > > > @@ -33,8 +33,15 @@ static inline struct cpu_info *get_cpu_info(void)
> > > >
> > > > #define guest_cpu_user_regs() (&get_cpu_info()->guest_cpu_user_regs)
> > > >
> > > > +#ifdef CONFIG_LIVEPATCH
> > > > +#define switch_stack_and_jump(stack, fn)
> > > > \
> > > > + asm volatile ("mov sp,%0;"
> > > > \
> > > > + "bl check_for_livepatch_work;"
> > > > \
> > >
> > > May I ask why check_for_livepatch_work is called in switch_stack_and_jump?
> >
> > From 29f4ab0b0a4ff62589af35b7cbc2334e1d9acdcd:
> > To perform and action on a payload, the hypercall sets up a data
> > structure to schedule the work. A hook is added in the
> > reset_stack_and_jump
> > to check for work and execute it if needed (specifically we check an
> > per-cpu flag to make this as quick as possible).
> >
> > In this way, patches can be applied with all CPUs idle and without
> > stacks. The first CPU to run check_for_xsplice_work() becomes the
> > master and triggers a reschedule softirq to trigger all the other CPUs
> > to enter check_for_xsplice_work() with no stack. Once all CPUs
> > have rendezvoused, all CPUs disable their IRQs and NMIs are ignored.
> > The system is then quiscient and the master performs the action.
> > After this, all CPUs enable IRQs and NMIs are re-enabled.
>
>
> I am a bit confused, switch_stack_and_jump will only be called on ARM when a
> vCPU is created. So why do you want to check livepatch at that time?
>
> After looking at the x86 version, I noticed that reset_stack_and_jump is
> called every time Xen returns to the guest, correct?
Yes. I assumed ARM was the same.
>
> If so, you may want to call check_for_livepatch_work in
> leave_hypervisor_tail instead. The latter will be call every time Xen
> returns to the guest.
Oooh. Thanks!
>
> Regards,
>
> --
> Julien Grall
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel
