RE: [PATCH 1/3] panic: Export panic_cpu and nmi_panic_self_stop

From: æåèå / KAWAIïHIDEHIRO
Date: Tue Mar 01 2016 - 02:48:33 EST


Hi Borislav,

> From: Borislav Petkov [mailto:bp@xxxxxxxxx]
> On Tue, Mar 01, 2016 at 10:50:37AM +0900, Hidehiro Kawai wrote:
> > Export panic_cpu and nmi_panic_self_stop symbols for modules which
> > use nmi_panic() macro.
> >
> > Signed-off-by: Hidehiro Kawai <hidehiro.kawai.ez@xxxxxxxxxxx>
> > Cc: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
> > Cc: Borislav Petkov <bp@xxxxxxx>
> > Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
> > Cc: Michal Hocko <mhocko@xxxxxxxx>
> > Cc: HATAYAMA Daisuke <d.hatayama@xxxxxxxxxxxxxx>
> > Cc: Vitaly Kuznetsov <vkuznets@xxxxxxxxxx>
> > Cc: Tejun Heo <tj@xxxxxxxxxx>
> > ---
> > kernel/panic.c | 2 ++
> > 1 file changed, 2 insertions(+)
> >
> > diff --git a/kernel/panic.c b/kernel/panic.c
> > index d96469d..f4e8035 100644
> > --- a/kernel/panic.c
> > +++ b/kernel/panic.c
> > @@ -69,8 +69,10 @@ void __weak nmi_panic_self_stop(struct pt_regs *regs)
> > {
> > panic_smp_self_stop();
> > }
> > +EXPORT_SYMBOL(nmi_panic_self_stop);
> >
> > atomic_t panic_cpu = ATOMIC_INIT(PANIC_CPU_INVALID);
> > +EXPORT_SYMBOL(panic_cpu);
>
> Can we make nmi_panic() at least a proper function and export that
> instead of exporting all those implementation details...?

The reason I implemented nmi_panic() as a macro is to pass
variable arguments directly to panic(). Fortunately, since all
invocations of nmi_panic() just pass a fixed string, I can change it
to a normal function like:

int nmi_panic(struct pt_regs *regs, const char *msg)
{
...
panic("%s", msg);

If people don't mind if that, I'll change it.

Regards,

--
Hidehiro Kawai
Hitachi, Ltd. Research & Development Group