Re: [PATCH] fix error handling in load_module()

[Rusty Russell <rusty@rustcorp.com.au>]

On Fri, 11 Sep 2009 06:44:30 am Andrew Morton wrote:
> On Mon, 7 Sep 2009 19:45:58 +0530
> Kamalesh Babulal <kamalesh@linux.vnet.ibm.com> wrote:
> > Once the percpu_modalloc fails, percpu_modfree(mod->refptr) is called on a NULL pointer.
> > We try calling it on a NULL pointer. The following patch fixes the problem by introducing 
> > a check for mod->refptr before calling percpu_modfree.
> 
> Where did it crash and why did it crash?  That trace is pretty unclear.
> 
> > diff --git a/kernel/module.c b/kernel/module.c
> > index 2d53718..7f89258 100644
> > --- a/kernel/module.c
> > +++ b/kernel/module.c
> > @@ -2379,7 +2379,8 @@ static noinline struct module *load_module(void __user *umod,
> >  	module_unload_free(mod);
> >  #if defined(CONFIG_MODULE_UNLOAD) && defined(CONFIG_SMP)
> >   free_init:
> > -	percpu_modfree(mod->refptr);
> > +	if (mod->refptr)
> > +		percpu_modfree(mod->refptr);
> >  #endif
> >  	module_free(mod, mod->module_init);
> >   free_core:
> 
> My reverse engineering of the secret, undocumented percpu_modfree()
> indicates that its mad inventor intended that percpu_modfree(NULL) be a
> valid thing to do.

Yes, percpu_modfree() should handle NULL.  If it doesn't, that's the bug.

Confused,
Rusty.