Re: [PATCH v1 5/5] of: property: Delete of_devlink kernel commandline option

From: Saravana Kannan
Date: Tue Mar 31 2020 - 13:10:56 EST


On Tue, Mar 31, 2020 at 5:20 AM Geert Uytterhoeven <geert@xxxxxxxxxxxxxx> wrote:
>
> Hi Saravana,
>
> On Sat, Feb 22, 2020 at 2:41 AM Saravana Kannan <saravanak@xxxxxxxxxx> wrote:
> > With the addition of fw_devlink kernel commandline option, of_devlink is
> > redundant and not useful anymore. So, delete it.
> >
> > Signed-off-by: Saravana Kannan <saravanak@xxxxxxxxxx>
>
> Thanks for your patch!
>
> This is now commit e94f62b7140fa3da ("of: property: Delete of_devlink
> kernel commandline option") upstream.
>
> > --- a/Documentation/admin-guide/kernel-parameters.txt
> > +++ b/Documentation/admin-guide/kernel-parameters.txt
> > @@ -3299,12 +3299,6 @@
> > This can be set from sysctl after boot.
> > See Documentation/admin-guide/sysctl/vm.rst for details.
> >
> > - of_devlink [OF, KNL] Create device links between consumer and
> > - supplier devices by scanning the devictree to infer the
> > - consumer/supplier relationships. A consumer device
> > - will not be probed until all the supplier devices have
> > - probed successfully.
> > -
> > ohci1394_dma=early [HW] enable debugging via the ohci1394 driver.
> > See Documentation/debugging-via-ohci1394.txt for more
> > info.
>
> While I agree with the thunk above...
>
> > diff --git a/drivers/of/property.c b/drivers/of/property.c
> > index 15fc9315f1a7..f104f15b57fb 100644
> > --- a/drivers/of/property.c
> > +++ b/drivers/of/property.c
> > @@ -1299,15 +1299,9 @@ static int of_link_to_suppliers(struct device *dev,
> > return ret;
> > }
> >
> > -static bool of_devlink;
> > -core_param(of_devlink, of_devlink, bool, 0);
> > -
> > static int of_fwnode_add_links(const struct fwnode_handle *fwnode,
> > struct device *dev)
> > {
> > - if (!of_devlink)
> > - return 0;
> > -
> > if (unlikely(!is_of_node(fwnode)))
> > return 0;

Hi Geert,

> ... I have some reservations about removing the actual code.
> The "of_devlink" kernel parameter was supported in v5.5 and v5.6, so
> removing its support may silently break some setups.
>
> Is this likely to happen?

As much as I'd love to see people start using this, I doubt of_devlink
has been significantly adopted outside of Android yet (I'm working on
making that easier :)). I'd be happy to be proven wrong though :)

of_devlink/fw_devlink is mainly useful when module loading is causing
ordering and clean up issues. So if deletion of of_devlink breaks
anything, I'd expect it to break in obvious ways and not silently.

> Do we need a compatibility fallback that warns to user to update his kernel
> command line?

I don't think we need to, but I'm not strongly against your suggestion
either. Let us know what you think.

-Saravana