Re: [PATCH v2 x86 1/2] fix page faults by nmiaction in nmi if kmemcheck is enabled

[Don Zickus <dzickus@redhat.com>]

On Mon, Mar 05, 2012 at 06:49:07PM +0100, Peter Zijlstra wrote:
> On Mon, 2012-03-05 at 18:05 +0800, Li Zhong wrote:
> > +static struct nmiaction arch_trigger_all_cpu_bt_nmiaction = {
> > +       .handler        = arch_trigger_all_cpu_backtrace_handler,
> > +       .name           = "arch_bt",
> > +};
> > +
> >  static int __init register_trigger_all_cpu_backtrace(void)
> >  {
> > -       register_nmi_handler(NMI_LOCAL, arch_trigger_all_cpu_backtrace_handler,
> > -                               0, "arch_bt");
> > +       register_nmi_handler(NMI_LOCAL, &arch_trigger_all_cpu_bt_nmiaction);
> >         return 0;
> >  } 
> 
> If you look at things like cpu_notifier() you can shorten this still:
> 
> #define nmi_notifier(t, fn, n)				\
> do {							\
> 	static struct nmiaction fn##_na = {		\
> 		.handler = (fn),			\
> 		.name = (n),				\
> 	};						\
> 	register_nmi_handler((t), &fn##_na);		\
> } while (0)
> 
> The whole thing then becomes:
> 
>   nmi_notifier(NMI_LOCAL, arch_trigger_all_cpu_backtrace_handler, "arch_bt");

Hi Li,

I would be open to this suggestion.  If you want to modify the patch for
this approach.

Cheers,
Don