Re: [PATCH 2/5] livepatch: Allow architectures to specify an alternate ftrace location

[Miroslav Benes]

On Thu, 14 Apr 2016, Michael Ellerman wrote:

> On Thu, 2016-04-14 at 14:01 +0200, Miroslav Benes wrote:
> > On Wed, 13 Apr 2016, Michael Ellerman wrote:
> 
> > >  static void klp_disable_func(struct klp_func *func)
> > >  {
> > >       struct klp_ops *ops;
> > > @@ -312,8 +325,14 @@ static void klp_disable_func(struct klp_func *func)
> > >               return;
> > >  
> > >       if (list_is_singular(&ops->func_stack)) {
> > > +             unsigned long ftrace_loc;
> > 
> > This is a nit, but could you move the definition up to have them all in 
> > one place to be consistent with the rest of the code? The same applies to 
> > klp_enable_func() below.
> 
> Hmm, actually I moved it in there because you pointed out we only needed it
> inside the if:
> 
> http://lkml.kernel.org/r/alpine.LNX.2.00.1603151113020.20252@xxxxxxxxxxxxx
> 
>     Thinking about it, we need ftrace_loc only in cases where we call 
>     ftrace_set_filter_ip() right? So we can move klp_get_ftrace_location() 
>     call to appropriate if branch both in klp_disable_func() and 
>     klp_enable_func().
> 
> But I guess you meant the function call, not the variable declaration.

Exactly.

> Personally I think it's better this way, as the variable is in scope for the
> shortest possible amount of time, but I can change it if you want me to.

No, it is nothing I would insist on.

Thanks,
Miroslav