Re: [RFC PATCH 3/3] module: pre-test setting ro_after_init data read-only

[Luis Chamberlain <mcgrof@kernel.org>]

On Wed, Nov 13, 2024 at 07:49:24AM +0100, Christophe Leroy wrote:
> 
> 
> Le 12/11/2024 à 21:28, Luis Chamberlain a écrit :
> > On Sat, Nov 09, 2024 at 11:35:37AM +0100, Christophe Leroy wrote:
> > > diff --git a/kernel/module/main.c b/kernel/module/main.c
> > > index 1bf4b0db291b..b603c9647e73 100644
> > > --- a/kernel/module/main.c
> > > +++ b/kernel/module/main.c
> > > @@ -2582,7 +2582,7 @@ static noinline int do_init_module(struct module *mod)
> > >   	rcu_assign_pointer(mod->kallsyms, &mod->core_kallsyms);
> > >   #endif
> > >   	ret = module_enable_rodata_ro_after_init(mod);
> > > -	if (ret)
> > > +	if (WARN_ON_ONCE(ret))
> > 
> > Do we want panic on warn systems to crash with this?
> > 
> 
> I would say yes, for two reasons:
> 1/ It should never happen
> 2/ Such systems care about security and don't want vulnerable systems

OK thanks for thinking about this.

I think making this clear in the commit log would help as people these
days scream at random WARN_ON_ONCE() sprinkles.

  Luis