[PATCH] of: Add of_platform_depopulate() helper

Pawel Moll pawel.moll at arm.com
Tue May 13 09:08:44 PDT 2014


On Tue, 2014-05-13 at 14:19 +0100, Rob Herring wrote:
> On Tue, May 13, 2014 at 6:50 AM, Pawel Moll <pawel.moll at arm.com> wrote:
> > Drivers can us of_platform_populate() to create
> > platform devices for children of the device
> > main node (this can particularly happen in case
> > of MFD devices). Unfortunately, there was no
> > standard way of removing such sub-devices when
> > the main one is being removed.
> >
> > This patch adds of_platform_depopulate() as
> > a complementary operation for the _populate()
> > one. It removes all platform and amba devices
> > that have been created from the Device Tree,
> > but leaves all other ones untouched (returning
> > -EBUSY in such case).
> >
> > Signed-off-by: Pawel Moll <pawel.moll at arm.com>
> > ---
> >  drivers/of/platform.c       | 52 +++++++++++++++++++++++++++++++++++++++++++++
> >  include/linux/of_platform.h |  5 +++++
> >  2 files changed, 57 insertions(+)
> >
> > diff --git a/drivers/of/platform.c b/drivers/of/platform.c
> > index 2d0c8b7..b5c49c3 100644
> > --- a/drivers/of/platform.c
> > +++ b/drivers/of/platform.c
> > @@ -495,4 +495,56 @@ int of_platform_populate(struct device_node *root,
> >         return rc;
> >  }
> >  EXPORT_SYMBOL_GPL(of_platform_populate);
> > +
> > +static int of_platform_device_destroy(struct device *dev, void *data)
> > +{
> > +       int *parents_children_left = data;
> 
> You are not really tracking how many children are left, so this can
> just be a bool.

That's true.

> > +       int my_children_left = 0;
> > +
> > +       /* Do not touch devices not populated from the device tree */
> > +       if (!dev->of_node || !of_node_check_flag(dev->of_node, OF_POPULATED)) {
> > +               (*parents_children_left)++;
> > +               return 0;
> > +       }
> > +
> > +       device_for_each_child(dev, &my_children_left,
> > +                       of_platform_device_destroy);
> 
> Can't you call of_platform_depopulate here?

... and check its result instead of ? I guess I could. Will have a look
how it looks like.

Paweł




More information about the linux-arm-kernel mailing list