On Fri, 21 Nov 2014, Josh Poimboeuf wrote:
> On Fri, Nov 21, 2014 at 03:44:35PM +0100, Miroslav Benes wrote:
> > On Fri, 21 Nov 2014, Jiri Kosina wrote:
> >
> > [...]
> >
> > > [ ... snip ... ]
> > > > +static int klp_init_patch(struct klp_patch *patch)
> > > > +{
> > > > + int ret;
> > > > +
> > > > + mutex_lock(&klp_mutex);
> > > > +
> > > > + /* init */
> > > > + patch->state = LPC_DISABLED;
> > > > +
> > > > + /* sysfs */
> > > > + ret = kobject_init_and_add(&patch->kobj, &klp_ktype_patch,
> > > > + klp_root_kobj, patch->mod->name);
> > > > + if (ret)
> > > > + return ret;
> > >
> > > klp_mutex is leaked locked here.
> > >
> > > > +
> > > > + /* create objects */
> > > > + ret = klp_init_objects(patch);
> > > > + if (ret) {
> > > > + kobject_put(&patch->kobj);
> > > > + return ret;
> > >
> > > And here as well.
> > >
> > > All in all, this is looking very good to me. I think we are really close
> > > to having a code that all the parties would agree with. Thanks everybody,
> >
> > The leaking is my fault. I missed that somehow during rebasing.
> >
> > Seth, could you please fix it in v4?
>
> Is it necessary to grab the mutex at the beginning of klp_init_patch? I
> think we only need it when adding it to the global list at the end of
> the function.
I think it's not necessary now after thinking about that. It could happen
that init values could be written twice to some patch structure if
klp_register_patch would be called twice. But it should not corrupt
anything and adding to the global list is protected. However I think we
should define what is protected by klp_mutex and comment it somewhere near
the mutex definition (if only the klp_patches list is protected or
something more (in the future)).
Mira
--
To unsubscribe from this list: send the line "unsubscribe live-patching" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
