[GIT PULL FOR v5.5 - 2nd try] R-Car DU CMM support

[Laurent Pinchart]

Hi Jacopo,

On Thu, Nov 07, 2019 at 12:13:35PM +0100, Jacopo Mondi wrote:
> On Wed, Nov 06, 2019 at 02:56:53PM +0200, Laurent Pinchart wrote:
> > On Wed, Nov 06, 2019 at 02:52:25PM +0200, Laurent Pinchart wrote:
> >> On Wed, Nov 06, 2019 at 11:46:28AM +0100, Jacopo Mondi wrote:
> >>> On Wed, Nov 06, 2019 at 12:02:05PM +0200, Laurent Pinchart wrote:
> >>>> On Wed, Nov 06, 2019 at 12:00:59PM +0200, Laurent Pinchart wrote:
> >>>>> On Wed, Nov 06, 2019 at 01:40:13PM +1000, Dave Airlie wrote:
> >>>>>> On Wed, 6 Nov 2019 at 05:56, Dave Airlie wrote:
> >>>>>>> On Wed, 6 Nov 2019 at 05:49, Laurent Pinchart wrote:
> >>>>>>>>
> >>>>>>>> Hi Dave,
> >>>>>>>>
> >>>>>>>> Has this pull request fallen through the cracks ?
> >>>>>>>
> >>>>>>> It fell down a different crack than usual, it made it from patchwork
> >>>>>>> onto my hard drive, but then I didn't execute the pull.
> >>>>>>>
> >>>>>>> I've pulled it down, thanks for reminder.
> >>>>>>
> >>>>>> Now it failed as I mentioned on irc.
> >>>>>>
> >>>>>> I think the new kconfig option needs to be a tristate, otherwise
> >>>>>> setting it to Y and having rcar-du as M fails to link
> >>>>>
> >>>>> Sorry about that :-S Both I and the 0-day bot failed to catch it.
> >>>>> Jacopo, could you please have a look ? I'm afraid this likely mean a new
> >>>>> version of the series, and thus missing the merge window, as we're
> >>>>> already at -rc6.
> >>>
> >>> I managed to reproduce it by setting DRM=m.
> >>> RCAR_CMM stays indeed =y as it's a bool. Simply setting it to tristate
> >>> is enough to have it set to =m when DRM=m.
> >>>
> >>> Could this be changed when applying the series ?
> >>>
> >>> diff --git a/drivers/gpu/drm/rcar-du/Kconfig b/drivers/gpu/drm/rcar-du/Kconfig
> >>> index 539d232790d1..e6607b5c8bb3 100644
> >>> --- a/drivers/gpu/drm/rcar-du/Kconfig
> >>> +++ b/drivers/gpu/drm/rcar-du/Kconfig
> >>> @@ -14,7 +14,7 @@ config DRM_RCAR_DU
> >>>           If M is selected the module will be called rcar-du-drm.
> >>>
> >>>  config DRM_RCAR_CMM
> >>> -       bool "R-Car DU Color Management Module (CMM) Support"
> >>> +       tristate "R-Car DU Color Management Module (CMM) Support"
> >>>         depends on DRM && OF
> >>>         depends on DRM_RCAR_DU
> >>>         help
> >>
> >> This looks reasonable to me. I'll let Dave decide if he wants to apply
> >> this change himself, or if I should send a new pull request for v5.6.
> >
> > Actually this won't work. With DRM_RCAR_DU=y and DRM_RCAR_CMM=m you'll
> > have a link failure.
> 
> You're right.. Actually just changing the RCAR_CMM option to tristate
> won't be enough. Adding "imply DRM_RCAR_CMM" to the DU option instead
> makes it impossible to select CMM=m when DU=y and viceversa, so it
> should be enough.
> 
> > Let's fix this properly for v5.6. Could you please propose a fix that
> > will work with all combinations of DRM, DRM_RCAR_DU and DRM_RCAR_CMM ?
> 
> Can I send a v6.1 version of the CMM patch only, or would you prefer a
> v7 ? we're anyway late for 5.5 I presume

v6.1 should be enough. Please make sure you test all combinations of
CONFIG_DRM_RCAR_DU and CONFIG_DRM_RCAR_CMM :-)

-- 
Regards,

Laurent Pinchart