Re: [for-next][PATCH 12/16] kprobes: Initialize kprobes at postcore_initcall

[Catalin Marinas]

Hi Steve,

On Tue, Jul 09, 2019 at 11:15:20AM -0400, Steven Rostedt wrote:
> On Wed, 3 Jul 2019 15:08:32 +0100
> Catalin Marinas <catalin.marinas@xxxxxxx> wrote:
> > > diff --git a/kernel/kprobes.c b/kernel/kprobes.c
> > > index 5471efbeb937..0ca6f53c8505 100644
> > > --- a/kernel/kprobes.c
> > > +++ b/kernel/kprobes.c
> > > @@ -2235,6 +2235,8 @@ static struct notifier_block kprobe_module_nb = {
> > >  extern unsigned long __start_kprobe_blacklist[];
> > >  extern unsigned long __stop_kprobe_blacklist[];
> > >  
> > > +static bool run_kprobe_tests __initdata;
> > > +
> > >  static int __init init_kprobes(void)
> > >  {
> > >  	int i, err = 0;
> > > @@ -2286,11 +2288,18 @@ static int __init init_kprobes(void)
> > >  	kprobes_initialized = (err == 0);
> > >  
> > >  	if (!err)
> > > -		init_test_probes();
> > > +		run_kprobe_tests = true;
> > >  	return err;
> > >  }
> > >  subsys_initcall(init_kprobes);
> > >  
> > > +static int __init run_init_test_probes(void)
> > > +{
> > > +	if (run_kprobe_tests)
> > > +		init_test_probes();  
> > 
> > A return 0 here.
> > 
> > > +}
> > > +module_init(run_init_test_probes);  
> > 
> > This does the trick. I prefer your fix as it leaves the arch code
> > unchanged. In case you need it:
> 
> And I actually think yours is better for the opposite reason ;-)
> 
> I agree with Masami, that the selftest actually caught a bug here. I
> think the arch code may need to change as the purpose of Masami's
> changes was to enable kprobes earlier in boot. The selftest failing
> means that an early kprobe will fail too.

I just got back from holiday and catching up with emails. Do I still
need to merge the arm64 fix making the debug initialisation a
core_initcall()?

Can we actually get kprobes invoked before init_kprobes() has been
called?

Thanks.

-- 
Catalin