[PATCH v3 02/17] drm/bridge: Verify lane assignment is going to work during atomic_check

[Stephen Boyd]

Quoting Andy Shevchenko (2024-08-20 10:29:41)
> On Tue, Aug 20, 2024 at 10:24:47AM -0700, Stephen Boyd wrote:
> > Quoting Andy Shevchenko (2024-08-20 10:17:46)
> > > On Tue, Aug 20, 2024 at 10:12:55AM -0700, Stephen Boyd wrote:
> > > > Quoting Andy Shevchenko (2024-08-20 03:09:29)
> > > > > On Mon, Aug 19, 2024 at 03:38:16PM -0700, Stephen Boyd wrote:
>
> ...
>
> > > > > > +     /*
> > > > > > +      * Ensure this bridge is aware that the next bridge wants to
> > > > > > +      * reassign lanes.
> > > > > > +      */
> > > > > > +     for (i = 0; i < num_input_lanes; i++)
> > > > > > +             if (i != input_lanes[i].logical && !num_output_lanes)
> > > > > > +                     return -ENOTSUPP;
> > > > >
> > > > > Besides missing {} this code is internal to the Linux kernel. Is it okay?
> > > >
> > > > ENOTSUPP is used by select_bus_fmt_recursive() so I simply followed that
> > > > style.
> > >
> > > Okay, just be aware of that side effect of that code, also checkpatch may
> > > complain (however it might be false positive).
> >
> > Yes checkpatch complained but didn't enlighten me. Please tell me the
> > side effect as I'm unaware!
>
> I already told you above, if this code ever appears in user space it will be
> printed as a number and very much confuse the user!
>
> That's why usage of this code either has to be documented or be subsystem
> _known_ practice (GPIO library comes to my mind as it uses it internally,\
> but filters for user space).
>

Ok, got it. Thanks!