Re: [PATCH v10 0/4] Separate panel orientation property creating and value setting - Hsin-Yi Wang

From: Hsin-Yi Wang <hsinyi@chromium.org>
To: Hans de Goede <hdegoede@redhat.com>
Cc: dri-devel@lists.freedesktop.org, David Airlie <airlied@linux.ie>,
	Daniel Vetter <daniel@ffwll.ch>,
	amd-gfx@lists.freedesktop.org, intel-gfx@lists.freedesktop.org,
	Rob Clark <robdclark@chromium.org>,
	Stephen Boyd <swboyd@chromium.org>,
	Douglas Anderson <dianders@chromium.org>,
	Chun-Kuang Hu <chunkuang.hu@kernel.org>,
	Sean Paul <sean@poorly.run>,
	Maarten Lankhorst <maarten.lankhorst@linux.intel.com>,
	Maxime Ripard <mripard@kernel.org>,
	Thomas Zimmermann <tzimmermann@suse.de>,
	linux-kernel@vger.kernel.org, Rob Herring <robh+dt@kernel.org>,
	Matthias Brugger <matthias.bgg@gmail.com>,
	devicetree@vger.kernel.org, linux-arm-kernel@lists.infradead.org,
	linux-mediatek@lists.infradead.org,
	Simon Ser <contact@emersion.fr>,
	Harry Wentland <harry.wentland@amd.com>,
	Alex Deucher <alexander.deucher@amd.com>,
	Jani Nikula <jani.nikula@linux.intel.com>,
	Emil Velikov <emil.l.velikov@gmail.com>
Subject: Re: [PATCH v10 0/4] Separate panel orientation property creating and value setting
Date: Mon, 30 May 2022 19:34:27 +0800	[thread overview]
Message-ID: <CAJMQK-iM-ip7edA2mBOhp-8maWKG5+kTceZUM5U6BOLLBq1H4Q@mail.gmail.com> (raw)
In-Reply-To: <a8d1fe13-e747-016a-2d45-bfb50f23f2d9@redhat.com>

On Mon, May 30, 2022 at 4:53 PM Hans de Goede <hdegoede@redhat.com> wrote:
>
> Hi,
>
> On 5/30/22 10:19, Hsin-Yi Wang wrote:
> > Some drivers, eg. mtk_drm and msm_drm, rely on the panel to set the
> > orientation. Panel calls drm_connector_set_panel_orientation() to create
> > orientation property and sets the value. However, connector properties
> > can't be created after drm_dev_register() is called. The goal is to
> > separate the orientation property creation, so drm drivers can create it
> > earlier before drm_dev_register().
>
> Sorry for jumping in pretty late in the discussion (based on the v10
> I seem to have missed this before).
>
> This sounds to me like the real issue here is that drm_dev_register()
> is getting called too early?
>
Right.

> To me it seems sensible to delay calling drm_dev_register() and
> thus allowing userspace to start detecting available displays +
> features until after the panel has been probed.
>

Most panels set this value very late, in .get_modes callback (since it
is when the connector is known), though the value was known during
panel probe.

I think we can also let drm check if they have remote panel nodes: If
there is a panel and the panel sets the orientation, let the drm read
this value and set the property. Does this workflow sound reasonable?

The corresponding patch to implement this:
https://patchwork.kernel.org/project/linux-mediatek/patch/20220530113033.124072-1-hsinyi@chromium.org/

Thanks

> I see a devicetree patch in this series, so I guess that the panel
> is described in devicetree. Especially in the case of devicetree
> I would expect the kernel to have enough info to do the right
> thing and make sure the panel is probed before calling
> drm_dev_register() ?
>
> Regards,
>
> Hans
>
>
>
>
> >
> > After this series, drm_connector_set_panel_orientation() works like
> > before. It won't affect existing callers of
> > drm_connector_set_panel_orientation(). The only difference is that
> > some drm drivers can call drm_connector_init_panel_orientation_property()
> > earlier.
> >
> > Hsin-Yi Wang (4):
> >   gpu: drm: separate panel orientation property creating and value
> >     setting
> >   drm/mediatek: init panel orientation property
> >   drm/msm: init panel orientation property
> >   arm64: dts: mt8183: Add panel rotation
> >
> >  .../arm64/boot/dts/mediatek/mt8183-kukui.dtsi |  1 +
> >  drivers/gpu/drm/drm_connector.c               | 58 ++++++++++++++-----
> >  drivers/gpu/drm/mediatek/mtk_dsi.c            |  7 +++
> >  drivers/gpu/drm/msm/dsi/dsi_manager.c         |  4 ++
> >  include/drm/drm_connector.h                   |  2 +
> >  5 files changed, 59 insertions(+), 13 deletions(-)
> >
>