Message ID | 20220601094637.1200634-1-hsinyi@chromium.org (mailing list archive) |
---|---|
Headers | show |
Series | Add a panel API to return panel orientation | expand |
Hi, Hsin-Yi: I'm curious that panel driver setting orientation in get_modes() would work or not. It it works, why not just set orientation in get_modes()? I find that exynos dsi driver has implement the get_modes() [1] but Mediatek dsi does not implement. Would you try this? [1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/gpu/drm/exynos/exynos_drm_dsi.c?h=v5.18#n1483 Regards, CK On Wed, 2022-06-01 at 17:46 +0800, Hsin-Yi Wang wrote: > Panels usually call drm_connector_set_panel_orientation(), which is > later than drm/kms driver calling drm_dev_register(). This leads to a > WARN()[1]. > > The orientation property is known earlier. For example, some panels > parse the property through device tree during probe. > > The series add a panel API drm_panel_get_orientation() for drm/kms > drivers. The drivers can use the API to get panel's orientation, so > they > can call drm_connector_set_panel_orientation() before > drm_dev_register(). > > Panel needs to implement .get_orientation callback to return the > property. > > [1] > https://urldefense.com/v3/__https://patchwork.kernel.org/project/linux-mediatek/patch/20220530081910.3947168-2-hsinyi@chromium.org/__;!!CTRNKA9wMg0ARbw!0ytf4wPxKANnYbw_fgq-eqQARnAXIHv4jwu04ZW6X6dAxZDFsJ0CltY_PvCBAQ$ > > > Hsin-Yi Wang (8): > drm/panel: Add an API drm_panel_get_orientation() to return panel > orientation > drm/panel: boe-tv101wum-nl6: Implement .get_orientation callback > drm/panel: panel-edp: Implement .get_orientation callback > drm/panel: lvds: Implement .get_orientation callback > drm/panel: panel-simple: Implement .get_orientation callback > drm/panel: ili9881c: Implement .get_orientation callback > drm/panel: elida-kd35t133: Implement .get_orientation callback > drm/mediatek: Config orientation property if panel provides it > > drivers/gpu/drm/drm_panel.c | 8 ++++++++ > drivers/gpu/drm/mediatek/mtk_dsi.c | 10 ++++++++++ > drivers/gpu/drm/panel/panel-boe-tv101wum-nl6.c | 8 ++++++++ > drivers/gpu/drm/panel/panel-edp.c | 8 ++++++++ > drivers/gpu/drm/panel/panel-elida-kd35t133.c | 8 ++++++++ > drivers/gpu/drm/panel/panel-ilitek-ili9881c.c | 8 ++++++++ > drivers/gpu/drm/panel/panel-lvds.c | 8 ++++++++ > drivers/gpu/drm/panel/panel-simple.c | 9 +++++++++ > include/drm/drm_panel.h | 10 ++++++++++ > 9 files changed, 77 insertions(+) >
On Thu, Jun 2, 2022 at 4:03 PM CK Hu <ck.hu@mediatek.com> wrote: > > Hi, Hsin-Yi: > > I'm curious that panel driver setting orientation in get_modes() would > work or not. It it works, why not just set orientation in get_modes()? > I find that exynos dsi driver has implement the get_modes() [1] but > Mediatek dsi does not implement. Would you try this? It's explained in the cover letter: ``` Panels usually call drm_connector_set_panel_orientation(), which is later than drm/kms driver calling drm_dev_register(). This leads to a WARN()[1]. ``` If the get_modes() in dsi is still being called later than bind(), then the same issue still appears. The purpose is to set the property before drm_dev_register(). > > [1] > https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/gpu/drm/exynos/exynos_drm_dsi.c?h=v5.18#n1483 > > Regards, > CK > > On Wed, 2022-06-01 at 17:46 +0800, Hsin-Yi Wang wrote: > > Panels usually call drm_connector_set_panel_orientation(), which is > > later than drm/kms driver calling drm_dev_register(). This leads to a > > WARN()[1]. > > > > The orientation property is known earlier. For example, some panels > > parse the property through device tree during probe. > > > > The series add a panel API drm_panel_get_orientation() for drm/kms > > drivers. The drivers can use the API to get panel's orientation, so > > they > > can call drm_connector_set_panel_orientation() before > > drm_dev_register(). > > > > Panel needs to implement .get_orientation callback to return the > > property. > > > > [1] > > https://urldefense.com/v3/__https://patchwork.kernel.org/project/linux-mediatek/patch/20220530081910.3947168-2-hsinyi@chromium.org/__;!!CTRNKA9wMg0ARbw!0ytf4wPxKANnYbw_fgq-eqQARnAXIHv4jwu04ZW6X6dAxZDFsJ0CltY_PvCBAQ$ > > > > > > Hsin-Yi Wang (8): > > drm/panel: Add an API drm_panel_get_orientation() to return panel > > orientation > > drm/panel: boe-tv101wum-nl6: Implement .get_orientation callback > > drm/panel: panel-edp: Implement .get_orientation callback > > drm/panel: lvds: Implement .get_orientation callback > > drm/panel: panel-simple: Implement .get_orientation callback > > drm/panel: ili9881c: Implement .get_orientation callback > > drm/panel: elida-kd35t133: Implement .get_orientation callback > > drm/mediatek: Config orientation property if panel provides it > > > > drivers/gpu/drm/drm_panel.c | 8 ++++++++ > > drivers/gpu/drm/mediatek/mtk_dsi.c | 10 ++++++++++ > > drivers/gpu/drm/panel/panel-boe-tv101wum-nl6.c | 8 ++++++++ > > drivers/gpu/drm/panel/panel-edp.c | 8 ++++++++ > > drivers/gpu/drm/panel/panel-elida-kd35t133.c | 8 ++++++++ > > drivers/gpu/drm/panel/panel-ilitek-ili9881c.c | 8 ++++++++ > > drivers/gpu/drm/panel/panel-lvds.c | 8 ++++++++ > > drivers/gpu/drm/panel/panel-simple.c | 9 +++++++++ > > include/drm/drm_panel.h | 10 ++++++++++ > > 9 files changed, 77 insertions(+) > > >
Hi, On Wed, Jun 1, 2022 at 2:46 AM Hsin-Yi Wang <hsinyi@chromium.org> wrote: > > Panels usually call drm_connector_set_panel_orientation(), which is > later than drm/kms driver calling drm_dev_register(). This leads to a > WARN()[1]. > > The orientation property is known earlier. For example, some panels > parse the property through device tree during probe. > > The series add a panel API drm_panel_get_orientation() for drm/kms > drivers. The drivers can use the API to get panel's orientation, so they > can call drm_connector_set_panel_orientation() before drm_dev_register(). > > Panel needs to implement .get_orientation callback to return the property. > > [1] https://patchwork.kernel.org/project/linux-mediatek/patch/20220530081910.3947168-2-hsinyi@chromium.org/ > > Hsin-Yi Wang (8): > drm/panel: Add an API drm_panel_get_orientation() to return panel > orientation > drm/panel: boe-tv101wum-nl6: Implement .get_orientation callback > drm/panel: panel-edp: Implement .get_orientation callback > drm/panel: lvds: Implement .get_orientation callback > drm/panel: panel-simple: Implement .get_orientation callback > drm/panel: ili9881c: Implement .get_orientation callback > drm/panel: elida-kd35t133: Implement .get_orientation callback > drm/mediatek: Config orientation property if panel provides it > > drivers/gpu/drm/drm_panel.c | 8 ++++++++ > drivers/gpu/drm/mediatek/mtk_dsi.c | 10 ++++++++++ > drivers/gpu/drm/panel/panel-boe-tv101wum-nl6.c | 8 ++++++++ > drivers/gpu/drm/panel/panel-edp.c | 8 ++++++++ > drivers/gpu/drm/panel/panel-elida-kd35t133.c | 8 ++++++++ > drivers/gpu/drm/panel/panel-ilitek-ili9881c.c | 8 ++++++++ > drivers/gpu/drm/panel/panel-lvds.c | 8 ++++++++ > drivers/gpu/drm/panel/panel-simple.c | 9 +++++++++ > include/drm/drm_panel.h | 10 ++++++++++ > 9 files changed, 77 insertions(+) Seems reasonable to me. I added a request to the panel-edp patch that probably applies to all of the panel patches requesting a comment next to the old calls in get_modes(). I assume that all but patch #8 ought to land through drm-misc. I'm happy to help commit them once everyone is happy. If the Mediatek maintainer(s) want to Ack patch #8 I can land that in drm-misc too. If we want something fancier like an immutable branch you'll probably have to get someone involved that's got a higher drm pay grade than me. ;-) -Doug
Hi, On 6/2/22 23:48, Doug Anderson wrote: > Hi, > > On Wed, Jun 1, 2022 at 2:46 AM Hsin-Yi Wang <hsinyi@chromium.org> wrote: >> >> Panels usually call drm_connector_set_panel_orientation(), which is >> later than drm/kms driver calling drm_dev_register(). This leads to a >> WARN()[1]. >> >> The orientation property is known earlier. For example, some panels >> parse the property through device tree during probe. >> >> The series add a panel API drm_panel_get_orientation() for drm/kms >> drivers. The drivers can use the API to get panel's orientation, so they >> can call drm_connector_set_panel_orientation() before drm_dev_register(). >> >> Panel needs to implement .get_orientation callback to return the property. >> >> [1] https://patchwork.kernel.org/project/linux-mediatek/patch/20220530081910.3947168-2-hsinyi@chromium.org/ >> >> Hsin-Yi Wang (8): >> drm/panel: Add an API drm_panel_get_orientation() to return panel >> orientation >> drm/panel: boe-tv101wum-nl6: Implement .get_orientation callback >> drm/panel: panel-edp: Implement .get_orientation callback >> drm/panel: lvds: Implement .get_orientation callback >> drm/panel: panel-simple: Implement .get_orientation callback >> drm/panel: ili9881c: Implement .get_orientation callback >> drm/panel: elida-kd35t133: Implement .get_orientation callback >> drm/mediatek: Config orientation property if panel provides it >> >> drivers/gpu/drm/drm_panel.c | 8 ++++++++ >> drivers/gpu/drm/mediatek/mtk_dsi.c | 10 ++++++++++ >> drivers/gpu/drm/panel/panel-boe-tv101wum-nl6.c | 8 ++++++++ >> drivers/gpu/drm/panel/panel-edp.c | 8 ++++++++ >> drivers/gpu/drm/panel/panel-elida-kd35t133.c | 8 ++++++++ >> drivers/gpu/drm/panel/panel-ilitek-ili9881c.c | 8 ++++++++ >> drivers/gpu/drm/panel/panel-lvds.c | 8 ++++++++ >> drivers/gpu/drm/panel/panel-simple.c | 9 +++++++++ >> include/drm/drm_panel.h | 10 ++++++++++ >> 9 files changed, 77 insertions(+) > > Seems reasonable to me. I added a request to the panel-edp patch that > probably applies to all of the panel patches requesting a comment next > to the old calls in get_modes(). > > I assume that all but patch #8 ought to land through drm-misc. I'm > happy to help commit them once everyone is happy. If the Mediatek > maintainer(s) want to Ack patch #8 I can land that in drm-misc too. If > we want something fancier like an immutable branch you'll probably > have to get someone involved that's got a higher drm pay grade than > me. ;-) Thanks, patch 8/8 looks good to me now too, so: Reviewed-by: Hans de Goede <hdegoede@redhat.com> for the series. Note I'm currently travelling so I cannot push this to drm-mic-next myself. Regards, Hans