On Wed, Dec 01, 2021 at 10:24:15AM +0100, Uladzislau Rezki wrote:
> On Tue, Nov 30, 2021 at 12:04:19PM +0100, Sebastian Andrzej Siewior wrote:
> > On 2021-11-30 11:39:09 [+0100], Miroslav Benes wrote:
> > > > --- a/kernel/module.c
> > > > +++ b/kernel/module.c
> > > > @@ -4150,8 +4150,7 @@ static int load_module(struct load_info *info, const char __user *uargs,
> > > > ddebug_cleanup:
> > > > ftrace_release_mod(mod);
> > > > dynamic_debug_remove(mod, info->debug);
> > > > - synchronize_rcu();
> > > > - kfree(mod->args);
> > > > + kvfree_rcu(mod->args);
> > > > free_arch_cleanup:
> > > > cfi_cleanup(mod);
> > > > module_arch_cleanup(mod);
> > >
> > > hm, if I am not missing something, synchronize_rcu() is not really
> > > connected to kfree(mod->args) there. synchronize_rcu() was added a long
> > > time ago when kernel/module.c removed stop_machine() from the code and
> > > replaced it with RCU to protect (at least?) mod->list. You can find
> > > list_del_rcu(&mod->list) a couple of lines below.
> >
> > so instead synchronize_rcu() + kfree() you could do
> > call_rcu(&mod->args->rcu, kfree()) but since you have no RCU-head around
> > in args you wait for the grace period and then invoke kfree.
> >
> > kvfree_rcu() is somehow like call_rcu() + kfree() but without the needed
> > RCU-head.
> > So you avoid waiting for the grace period but mod->args is freed later,
> > after as expected.
> >
> > > And yes, one could ask how this all works. The error/cleanup sequence in
> > > load_module() is a giant mess... well, load_module() is a mess too, but
> > > the error path is really not nice.
> >
> > Well, spring is coming :)
> >
> Indeed that error path sequence is terrible. I will double check if that
> synchronize_rcu() + kfree() is related to any RCU protection and freeing.
>
> If it is not i will drop this patch.
>
>
OK, that kfree has nothing to do with RCU protection:
<snip>
commit 6526c534b2677ca601b7b92851437feb041d02a1
Author: Rusty Russell <rusty@xxxxxxxxxxxxxxx>
Date: Thu Aug 5 12:59:10 2010 -0600
module: move module args strndup_user to just before use
Instead of copying and allocating the args and storing it in
load_info, we can just allocate them right before we need them.
Signed-off-by: Rusty Russell <rusty@xxxxxxxxxxxxxxx>
<snip>
so, i will drop my patch, since the intention is not to free a ptr
after a grace period.
Thank you for the review!
--
Vlad Rezki
