[PATCH v3 6/9] xen/sys-hypervisor: Add release callback to hyp_sysfs_kobj_type

[Ian Rogers]

On Sun, Nov 8, 2026 at 10:56 PM Juergen Gross <jgross at suse.com> wrote:
>
> On 08.11.26 23:39, Ian Rogers wrote:
> > hypervisor_init() in drivers/base/hypervisor.c creates hypervisor_kobj
> > with kobject_create_and_add() (which allocates a dynamic kobject using
> > dynamic_kobj_ktype), and hypervisor_subsys_init() in
> > drivers/xen/sys-hypervisor.c then overwrites hypervisor_kobj->ktype with
> > &hyp_sysfs_kobj_type, which had no .release callback.
> >
> > Add hyp_sysfs_release() calling kfree(kobj) to preserve the dynamic
> > kobject release behavior.
> >
> > Fixes: cff7e81b3dd7 ("xen: add /sys/hypervisor support")
> > Signed-off-by: Ian Rogers <irogers at google.com>
> > ---
> >   drivers/xen/sys-hypervisor.c | 6 ++++++
> >   1 file changed, 6 insertions(+)
> >
> > diff --git a/drivers/xen/sys-hypervisor.c b/drivers/xen/sys-hypervisor.c
> > index 91923242a5ae..0e8a2cd69fd6 100644
> > --- a/drivers/xen/sys-hypervisor.c
> > +++ b/drivers/xen/sys-hypervisor.c
> > @@ -659,7 +659,13 @@ static const struct sysfs_ops hyp_sysfs_ops = {
> >       .store = hyp_sysfs_store,
> >   };
> >
> > +static void hyp_sysfs_release(struct kobject *kobj)
> > +{
> > +     kfree(kobj);
> > +}
> > +
> >   static const struct kobj_type hyp_sysfs_kobj_type = {
> > +     .release = hyp_sysfs_release,
> >       .sysfs_ops = &hyp_sysfs_ops,
> >   };
> >
>
> Wouldn't it make sense to have a common function for just kfree(kobj)?
>
> In your series there are 3 instances of .release just doing kfree(kobj).
>
> In case this is not wanted or with using the suggested common .release
> helper you can add my
>
> Reviewed-by: Juergen Gross <jgross at suse.com>

Thanks Juergen!

lib/kobject.c already has a static dynamic_kobj_release() that does
this (and there are ~15 other places in the tree open-coding a static
kfree(kobj) release wrapper), so include/linux/kobject.h +
lib/kobject.c would be the natural home if we exported one.

I kept them as local static functions here for two reasons:

1. Patches 2/9, 3/9, and 6/9 carry Fixes: tags, and keeping the small
   local callbacks avoids a dependency on a new driver-core export for
   backports.
2. Since kobjects are normally embedded in a larger struct and freed via
   container_of(), I wasn't sure whether Greg and Danilo would view an
   exported kfree(kobj) release helper (especially alongside
   __must_init) as an invitation for misuse on embedded kobjects.

Unless Greg or Danilo would like dynamic_kobj_release() (or similar)
exported now, I'll plan to keep the local callbacks for this series
and can send a follow-up cleanup for the tree-wide callers if
driver-core maintainers think a shared helper makes sense.

Thanks,
Ian

>
>
> Juergen