Re: [PATCH 22/29] drm/omap: Move DISPC timing checks to CRTC .mode_valid() operation
Hi, ping & actually add the cc :) -- Sebastian On Tue, Dec 11, 2018 at 10:06:07AM +0200, Laurent Pinchart wrote: >> The problem is, that DISPC timings for DSI are not directly >> calculated from mode. Instead DSI code does some modifications >> before applying the DISPC settings. > > I wonder what the best way to fix this would be. DSI isn't so special in this > regard, any encoder can modify modes, resulting in the CRTC outputting a mode > significantly different than the mode being displayed. This is handled by the > DRM helpers through the .mode_fixup() operations, which can mangle modes > along > the pipeline. > > At atomic commit time the atomic helpers call .mode_valid() first, followed > by > .mode_fixup(). An easy fix would be to remove our .mode_valid() handlers > completely and perform both fixup and validation in the .mode_fixup() > handlers. > > However, the .mode_valid() operations are also used in > drm_helper_probe_single_connector_modes(), where no mode fixup is performed. > We can't drop .mode_valid() for that code path (at least not as anything else > than a hack). Daniel, what's the expected way to handle this ? > > -- > Regards, > > Laurent Pinchart signature.asc Description: PGP signature ___ dri-devel mailing list dri-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/dri-devel
Re: [PATCH 22/29] drm/omap: Move DISPC timing checks to CRTC .mode_valid() operation
Hi, On Mon, Dec 10, 2018 at 02:14:01PM +0100, Sebastian Reichel wrote: > On Mon, Dec 10, 2018 at 09:50:08AM +0200, Laurent Pinchart wrote: > > Hi Sebastian, > > > > On Monday, 10 December 2018 00:07:55 EET Sebastian Reichel wrote: > > > On Wed, Dec 05, 2018 at 05:00:15PM +0200, Laurent Pinchart wrote: > > > > The DISPC timings checks relate to the CRTC, but they're performed in > > > > the encoder and connector .atomic_check() and .mode_valid() operations. > > > > Move them to the CRTC .mode_valid() operation. > > > > > > > > Signed-off-by: Laurent Pinchart > > > > --- > > > > > > Reviewed-by: Sebastian Reichel > > > > > > This should also fix the issue with DSI in a less ugly way: > > > > > > https://lists.freedesktop.org/archives/dri-devel/2018-November/196622.html > > > > Should I add the > > > > Fixes: 7c27fa57ef31 ("drm/omap: Call dispc timings check operation > > directly") > > > > tag to this patch ? > > I plan to test this series in combination with my DSI patches today > or tomorrow evening. If it works, probably the following should be > added to the patch description: > > > As a side-effect this fixes DSI, which uses slightly different mode > configuration for the DSI encoder and the DISPC. > > Fixes: 7c27fa57ef31 ("drm/omap: Call dispc timings check operation directly") > I merged this series with my remaining DSI command mode patches (btw. this series does not apply cleanly to master due to a recent patch from Tomi). Unfortunately this patch does not fix the issue. After this patch the mgr_check_timings() call in omap_crtc_mode_valid() will fail. The good news: The Droid 4 display worked again after dropping the mgr_check_timings call as a quick hack. So the other patches from this set are Tested-by: Sebastian Reichel The problem is, that DISPC timings for DSI are not directly calculated from mode. Instead DSI code does some modifications before applying the DISPC settings. -- Sebastian signature.asc Description: PGP signature ___ dri-devel mailing list dri-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/dri-devel
Re: [PATCH 22/29] drm/omap: Move DISPC timing checks to CRTC .mode_valid() operation
Hi, On Mon, Dec 10, 2018 at 09:50:08AM +0200, Laurent Pinchart wrote: > Hi Sebastian, > > On Monday, 10 December 2018 00:07:55 EET Sebastian Reichel wrote: > > On Wed, Dec 05, 2018 at 05:00:15PM +0200, Laurent Pinchart wrote: > > > The DISPC timings checks relate to the CRTC, but they're performed in > > > the encoder and connector .atomic_check() and .mode_valid() operations. > > > Move them to the CRTC .mode_valid() operation. > > > > > > Signed-off-by: Laurent Pinchart > > > --- > > > > Reviewed-by: Sebastian Reichel > > > > This should also fix the issue with DSI in a less ugly way: > > > > https://lists.freedesktop.org/archives/dri-devel/2018-November/196622.html > > Should I add the > > Fixes: 7c27fa57ef31 ("drm/omap: Call dispc timings check operation directly") > > tag to this patch ? I plan to test this series in combination with my DSI patches today or tomorrow evening. If it works, probably the following should be added to the patch description: As a side-effect this fixes DSI, which uses slightly different mode configuration for the DSI encoder and the DISPC. Fixes: 7c27fa57ef31 ("drm/omap: Call dispc timings check operation directly") -- Sebastian > > > drivers/gpu/drm/omapdrm/omap_connector.c | 6 -- > > > drivers/gpu/drm/omapdrm/omap_crtc.c | 9 + > > > drivers/gpu/drm/omapdrm/omap_encoder.c | 6 -- > > > 3 files changed, 9 insertions(+), 12 deletions(-) > > > > > > diff --git a/drivers/gpu/drm/omapdrm/omap_connector.c > > > b/drivers/gpu/drm/omapdrm/omap_connector.c index > > > 8db1f2fbcf43..dd1e0f2e8ffc 100644 > > > --- a/drivers/gpu/drm/omapdrm/omap_connector.c > > > +++ b/drivers/gpu/drm/omapdrm/omap_connector.c > > > @@ -245,8 +245,6 @@ static int omap_connector_mode_valid(struct > > > drm_connector *connector,> > > >struct drm_display_mode *mode) > > > > > > { > > > > > > struct omap_connector *omap_connector = to_omap_connector(connector); > > > > > > - enum omap_channel channel = omap_connector->output->dispc_channel; > > > - struct omap_drm_private *priv = connector->dev->dev_private; > > > > > > struct omap_dss_device *dssdev; > > > struct videomode vm = {0}; > > > struct drm_device *dev = connector->dev; > > > > > > @@ -256,10 +254,6 @@ static int omap_connector_mode_valid(struct > > > drm_connector *connector,> > > > drm_display_mode_to_videomode(mode, ); > > > mode->vrefresh = drm_mode_vrefresh(mode); > > > > > > - r = priv->dispc_ops->mgr_check_timings(priv->dispc, channel, ); > > > - if (r) > > > - goto done; > > > - > > > > > > for (dssdev = omap_connector->output; dssdev; dssdev = dssdev->next) { > > > > > > if (!dssdev->ops->check_timings) > > > > > > continue; > > > > > > diff --git a/drivers/gpu/drm/omapdrm/omap_crtc.c > > > b/drivers/gpu/drm/omapdrm/omap_crtc.c index caffc547ef97..51036e6fca1c > > > 100644 > > > --- a/drivers/gpu/drm/omapdrm/omap_crtc.c > > > +++ b/drivers/gpu/drm/omapdrm/omap_crtc.c > > > @@ -391,6 +391,15 @@ static enum drm_mode_status > > > omap_crtc_mode_valid(struct drm_crtc *crtc,> > > > const struct drm_display_mode *mode) > > > > > > { > > > > > > struct omap_drm_private *priv = crtc->dev->dev_private; > > > > > > + struct omap_crtc *omap_crtc = to_omap_crtc(crtc); > > > + struct videomode vm = {0}; > > > + int r; > > > + > > > + drm_display_mode_to_videomode(mode, ); > > > + r = priv->dispc_ops->mgr_check_timings(priv->dispc, omap_crtc->channel, > > > +); > > > + if (r) > > > + return r; > > > > > > /* Check for bandwidth limit */ > > > if (priv->max_bandwidth) { > > > > > > diff --git a/drivers/gpu/drm/omapdrm/omap_encoder.c > > > b/drivers/gpu/drm/omapdrm/omap_encoder.c index 79ee927cd905..7d01df6fa723 > > > 100644 > > > --- a/drivers/gpu/drm/omapdrm/omap_encoder.c > > > +++ b/drivers/gpu/drm/omapdrm/omap_encoder.c > > > @@ -198,19 +198,13 @@ static int omap_encoder_atomic_check(struct > > > drm_encoder *encoder,> > > >struct drm_connector_state *conn_state) > > > > > > { > > > > > > struct omap_encoder *omap_encoder = to_omap_encoder(encoder); > > > > > > - enum omap_channel channel = omap_encoder->output->dispc_channel; > > > > > > struct drm_device *dev = encoder->dev; > > > > > > - struct omap_drm_private *priv = dev->dev_private; > > > > > > struct omap_dss_device *dssdev; > > > struct videomode vm = { 0 }; > > > int ret; > > > > > > drm_display_mode_to_videomode(_state->mode, ); > > > > > > - ret = priv->dispc_ops->mgr_check_timings(priv->dispc, channel, ); > > > - if (ret) > > > - goto done; > > > - > > > > > > for (dssdev = omap_encoder->output; dssdev; dssdev = dssdev->next) { > > > > > > if (!dssdev->ops->check_timings) > > > > > > continue; >
Re: [PATCH 22/29] drm/omap: Move DISPC timing checks to CRTC .mode_valid() operation
Hi Sebastian, (CC'ing Daniel) On Tuesday, 11 December 2018 03:07:45 EET Sebastian Reichel wrote: > On Mon, Dec 10, 2018 at 02:14:01PM +0100, Sebastian Reichel wrote: > > On Mon, Dec 10, 2018 at 09:50:08AM +0200, Laurent Pinchart wrote: > >> On Monday, 10 December 2018 00:07:55 EET Sebastian Reichel wrote: > >>> On Wed, Dec 05, 2018 at 05:00:15PM +0200, Laurent Pinchart wrote: > The DISPC timings checks relate to the CRTC, but they're performed > in the encoder and connector .atomic_check() and .mode_valid() > operations. Move them to the CRTC .mode_valid() operation. > > Signed-off-by: Laurent Pinchart > --- > >>> > >>> Reviewed-by: Sebastian Reichel > >>> > >>> This should also fix the issue with DSI in a less ugly way: > >>> > >>> https://lists.freedesktop.org/archives/dri-devel/2018-November/196622. > >>> html > >> > >> Should I add the > >> > >> Fixes: 7c27fa57ef31 ("drm/omap: Call dispc timings check operation > >> directly") > >> > >> tag to this patch ? > > > > I plan to test this series in combination with my DSI patches today > > or tomorrow evening. If it works, probably the following should be > > added to the patch description: > > > > > > As a side-effect this fixes DSI, which uses slightly different mode > > configuration for the DSI encoder and the DISPC. > > > > Fixes: 7c27fa57ef31 ("drm/omap: Call dispc timings check operation > > directly") > > > > I merged this series with my remaining DSI command mode patches > (btw. this series does not apply cleanly to master due to a recent > patch from Tomi). I'll fix that, thank you for the notice. > Unfortunately this patch does not fix the issue. After this patch the > mgr_check_timings() call in omap_crtc_mode_valid() will fail. > > The good news: The Droid 4 display worked again after dropping the > mgr_check_timings call as a quick hack. So the other patches from > this set are > > Tested-by: Sebastian Reichel Thank you for testing this. > The problem is, that DISPC timings for DSI are not directly > calculated from mode. Instead DSI code does some modifications > before applying the DISPC settings. I wonder what the best way to fix this would be. DSI isn't so special in this regard, any encoder can modify modes, resulting in the CRTC outputting a mode significantly different than the mode being displayed. This is handled by the DRM helpers through the .mode_fixup() operations, which can mangle modes along the pipeline. At atomic commit time the atomic helpers call .mode_valid() first, followed by .mode_fixup(). An easy fix would be to remove our .mode_valid() handlers completely and perform both fixup and validation in the .mode_fixup() handlers. However, the .mode_valid() operations are also used in drm_helper_probe_single_connector_modes(), where no mode fixup is performed. We can't drop .mode_valid() for that code path (at least not as anything else than a hack). Daniel, what's the expected way to handle this ? -- Regards, Laurent Pinchart ___ dri-devel mailing list dri-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/dri-devel
Re: [PATCH 22/29] drm/omap: Move DISPC timing checks to CRTC .mode_valid() operation
Hi, On Wed, Dec 05, 2018 at 05:00:15PM +0200, Laurent Pinchart wrote: > The DISPC timings checks relate to the CRTC, but they're performed in > the encoder and connector .atomic_check() and .mode_valid() operations. > Move them to the CRTC .mode_valid() operation. > > Signed-off-by: Laurent Pinchart > --- Reviewed-by: Sebastian Reichel This should also fix the issue with DSI in a less ugly way: https://lists.freedesktop.org/archives/dri-devel/2018-November/196622.html -- Sebastian > drivers/gpu/drm/omapdrm/omap_connector.c | 6 -- > drivers/gpu/drm/omapdrm/omap_crtc.c | 9 + > drivers/gpu/drm/omapdrm/omap_encoder.c | 6 -- > 3 files changed, 9 insertions(+), 12 deletions(-) > > diff --git a/drivers/gpu/drm/omapdrm/omap_connector.c > b/drivers/gpu/drm/omapdrm/omap_connector.c > index 8db1f2fbcf43..dd1e0f2e8ffc 100644 > --- a/drivers/gpu/drm/omapdrm/omap_connector.c > +++ b/drivers/gpu/drm/omapdrm/omap_connector.c > @@ -245,8 +245,6 @@ static int omap_connector_mode_valid(struct drm_connector > *connector, >struct drm_display_mode *mode) > { > struct omap_connector *omap_connector = to_omap_connector(connector); > - enum omap_channel channel = omap_connector->output->dispc_channel; > - struct omap_drm_private *priv = connector->dev->dev_private; > struct omap_dss_device *dssdev; > struct videomode vm = {0}; > struct drm_device *dev = connector->dev; > @@ -256,10 +254,6 @@ static int omap_connector_mode_valid(struct > drm_connector *connector, > drm_display_mode_to_videomode(mode, ); > mode->vrefresh = drm_mode_vrefresh(mode); > > - r = priv->dispc_ops->mgr_check_timings(priv->dispc, channel, ); > - if (r) > - goto done; > - > for (dssdev = omap_connector->output; dssdev; dssdev = dssdev->next) { > if (!dssdev->ops->check_timings) > continue; > diff --git a/drivers/gpu/drm/omapdrm/omap_crtc.c > b/drivers/gpu/drm/omapdrm/omap_crtc.c > index caffc547ef97..51036e6fca1c 100644 > --- a/drivers/gpu/drm/omapdrm/omap_crtc.c > +++ b/drivers/gpu/drm/omapdrm/omap_crtc.c > @@ -391,6 +391,15 @@ static enum drm_mode_status omap_crtc_mode_valid(struct > drm_crtc *crtc, > const struct drm_display_mode *mode) > { > struct omap_drm_private *priv = crtc->dev->dev_private; > + struct omap_crtc *omap_crtc = to_omap_crtc(crtc); > + struct videomode vm = {0}; > + int r; > + > + drm_display_mode_to_videomode(mode, ); > + r = priv->dispc_ops->mgr_check_timings(priv->dispc, omap_crtc->channel, > +); > + if (r) > + return r; > > /* Check for bandwidth limit */ > if (priv->max_bandwidth) { > diff --git a/drivers/gpu/drm/omapdrm/omap_encoder.c > b/drivers/gpu/drm/omapdrm/omap_encoder.c > index 79ee927cd905..7d01df6fa723 100644 > --- a/drivers/gpu/drm/omapdrm/omap_encoder.c > +++ b/drivers/gpu/drm/omapdrm/omap_encoder.c > @@ -198,19 +198,13 @@ static int omap_encoder_atomic_check(struct drm_encoder > *encoder, >struct drm_connector_state *conn_state) > { > struct omap_encoder *omap_encoder = to_omap_encoder(encoder); > - enum omap_channel channel = omap_encoder->output->dispc_channel; > struct drm_device *dev = encoder->dev; > - struct omap_drm_private *priv = dev->dev_private; > struct omap_dss_device *dssdev; > struct videomode vm = { 0 }; > int ret; > > drm_display_mode_to_videomode(_state->mode, ); > > - ret = priv->dispc_ops->mgr_check_timings(priv->dispc, channel, ); > - if (ret) > - goto done; > - > for (dssdev = omap_encoder->output; dssdev; dssdev = dssdev->next) { > if (!dssdev->ops->check_timings) > continue; > -- > Regards, > > Laurent Pinchart > > ___ > dri-devel mailing list > dri-devel@lists.freedesktop.org > https://lists.freedesktop.org/mailman/listinfo/dri-devel signature.asc Description: PGP signature ___ dri-devel mailing list dri-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/dri-devel
Re: [PATCH 22/29] drm/omap: Move DISPC timing checks to CRTC .mode_valid() operation
Hi Sebastian, On Monday, 10 December 2018 00:07:55 EET Sebastian Reichel wrote: > On Wed, Dec 05, 2018 at 05:00:15PM +0200, Laurent Pinchart wrote: > > The DISPC timings checks relate to the CRTC, but they're performed in > > the encoder and connector .atomic_check() and .mode_valid() operations. > > Move them to the CRTC .mode_valid() operation. > > > > Signed-off-by: Laurent Pinchart > > --- > > Reviewed-by: Sebastian Reichel > > This should also fix the issue with DSI in a less ugly way: > > https://lists.freedesktop.org/archives/dri-devel/2018-November/196622.html Should I add the Fixes: 7c27fa57ef31 ("drm/omap: Call dispc timings check operation directly") tag to this patch ? > > drivers/gpu/drm/omapdrm/omap_connector.c | 6 -- > > drivers/gpu/drm/omapdrm/omap_crtc.c | 9 + > > drivers/gpu/drm/omapdrm/omap_encoder.c | 6 -- > > 3 files changed, 9 insertions(+), 12 deletions(-) > > > > diff --git a/drivers/gpu/drm/omapdrm/omap_connector.c > > b/drivers/gpu/drm/omapdrm/omap_connector.c index > > 8db1f2fbcf43..dd1e0f2e8ffc 100644 > > --- a/drivers/gpu/drm/omapdrm/omap_connector.c > > +++ b/drivers/gpu/drm/omapdrm/omap_connector.c > > @@ -245,8 +245,6 @@ static int omap_connector_mode_valid(struct > > drm_connector *connector,> > > struct drm_display_mode *mode) > > > > { > > > > struct omap_connector *omap_connector = to_omap_connector(connector); > > > > - enum omap_channel channel = omap_connector->output->dispc_channel; > > - struct omap_drm_private *priv = connector->dev->dev_private; > > > > struct omap_dss_device *dssdev; > > struct videomode vm = {0}; > > struct drm_device *dev = connector->dev; > > > > @@ -256,10 +254,6 @@ static int omap_connector_mode_valid(struct > > drm_connector *connector,> > > drm_display_mode_to_videomode(mode, ); > > mode->vrefresh = drm_mode_vrefresh(mode); > > > > - r = priv->dispc_ops->mgr_check_timings(priv->dispc, channel, ); > > - if (r) > > - goto done; > > - > > > > for (dssdev = omap_connector->output; dssdev; dssdev = dssdev->next) { > > > > if (!dssdev->ops->check_timings) > > > > continue; > > > > diff --git a/drivers/gpu/drm/omapdrm/omap_crtc.c > > b/drivers/gpu/drm/omapdrm/omap_crtc.c index caffc547ef97..51036e6fca1c > > 100644 > > --- a/drivers/gpu/drm/omapdrm/omap_crtc.c > > +++ b/drivers/gpu/drm/omapdrm/omap_crtc.c > > @@ -391,6 +391,15 @@ static enum drm_mode_status > > omap_crtc_mode_valid(struct drm_crtc *crtc,> > > const struct drm_display_mode *mode) > > > > { > > > > struct omap_drm_private *priv = crtc->dev->dev_private; > > > > + struct omap_crtc *omap_crtc = to_omap_crtc(crtc); > > + struct videomode vm = {0}; > > + int r; > > + > > + drm_display_mode_to_videomode(mode, ); > > + r = priv->dispc_ops->mgr_check_timings(priv->dispc, omap_crtc->channel, > > + ); > > + if (r) > > + return r; > > > > /* Check for bandwidth limit */ > > if (priv->max_bandwidth) { > > > > diff --git a/drivers/gpu/drm/omapdrm/omap_encoder.c > > b/drivers/gpu/drm/omapdrm/omap_encoder.c index 79ee927cd905..7d01df6fa723 > > 100644 > > --- a/drivers/gpu/drm/omapdrm/omap_encoder.c > > +++ b/drivers/gpu/drm/omapdrm/omap_encoder.c > > @@ -198,19 +198,13 @@ static int omap_encoder_atomic_check(struct > > drm_encoder *encoder,> > > struct drm_connector_state *conn_state) > > > > { > > > > struct omap_encoder *omap_encoder = to_omap_encoder(encoder); > > > > - enum omap_channel channel = omap_encoder->output->dispc_channel; > > > > struct drm_device *dev = encoder->dev; > > > > - struct omap_drm_private *priv = dev->dev_private; > > > > struct omap_dss_device *dssdev; > > struct videomode vm = { 0 }; > > int ret; > > > > drm_display_mode_to_videomode(_state->mode, ); > > > > - ret = priv->dispc_ops->mgr_check_timings(priv->dispc, channel, ); > > - if (ret) > > - goto done; > > - > > > > for (dssdev = omap_encoder->output; dssdev; dssdev = dssdev->next) { > > > > if (!dssdev->ops->check_timings) > > > > continue; -- Regards, Laurent Pinchart ___ dri-devel mailing list dri-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/dri-devel