Re: [RFCv1 32/42] drm/omap: dsi: convert to drm_panel
Hi Sebastian, > Am 18.11.2019 um 15:51 schrieb H. Nikolaus Schaller : > >> Ok, I tried not to break video mode support, but I do not have any >> hardware. Make sure to set the MIPI_DSI_MODE_VIDEO flag in the panel >> driver. > > Indeed, this may be missing (can't look into the code at the moment)... > Or I did something wrong when refactoring the driver. > We will find out. Yes, MIPI_DSI_MODE_VIDEO was indeed missing/wrongly applied and some more bugs. But I still wasn't able to make it work. I also tried to fake the panel-orisetech-otm8009a.c DSI driver into my setup. It should not properly program the panel by DCS command but it should try to. Result is the same: I can see it being probed and calling get_modes but then: [drm] Cannot find any crtc or sizes And I don't see calls to .enable or .prepare where DCS commands would be sent. BR and thanks, Nikolaus ___ dri-devel mailing list dri-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/dri-devel
Re: [RFCv1 32/42] drm/omap: dsi: convert to drm_panel
Hi Nikolaus, On Tue, Nov 19, 2019 at 10:42:55AM +0100, H. Nikolaus Schaller wrote: > > Am 18.11.2019 um 15:51 schrieb H. Nikolaus Schaller : > > > >> Ok, I tried not to break video mode support, but I do not have any > >> hardware. Make sure to set the MIPI_DSI_MODE_VIDEO flag in the panel > >> driver. > > > > Indeed, this may be missing (can't look into the code at the moment)... > > Or I did something wrong when refactoring the driver. > > We will find out. > > Yes, MIPI_DSI_MODE_VIDEO was indeed missing/wrongly applied and some > more bugs. But I still wasn't able to make it work. > > I also tried to fake the panel-orisetech-otm8009a.c DSI driver into > my setup. It should not properly program the panel by DCS command > but it should try to. > > Result is the same: I can see it being probed and calling get_modes > but then: > > [drm] Cannot find any crtc or sizes > > And I don't see calls to .enable or .prepare where DCS commands would > be sent. You probably want to enable all kind of drm debugging to get more information. I guess, that some timing check fails, e.g. dsi_check_timings(). If the timing checks fail, the mode will not be added as valid mode. When no valid mode is found, the panel will not be enabled. -- Sebastian signature.asc Description: PGP signature ___ dri-devel mailing list dri-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/dri-devel
Re: [RFCv1 32/42] drm/omap: dsi: convert to drm_panel
> Am 18.11.2019 um 15:45 schrieb Sebastian Reichel > : > > Hi, > > On Sun, Nov 17, 2019 at 08:23:05PM +0100, H. Nikolaus Schaller wrote: >>> [...] >> >>> + drm_panel_init(&ddata->panel, dev, &dsicm_panel_funcs, >>> + DRM_MODE_CONNECTOR_DSI); >>> + >> >> This leads to >> >> drivers/gpu/drm/panel/panel-dsi-cm.c: In function 'dsicm_probe': >> drivers/gpu/drm/panel/panel-dsi-cm.c:552:2: error: too many arguments to >> function 'drm_panel_init' >> drm_panel_init(&ddata->panel, dev, &dsicm_panel_funcs, >> ^ >> In file included from drivers/gpu/drm/panel/panel-dsi-cm.c:22:0: >> ./include/drm/drm_panel.h:150:6: note: declared here >> void drm_panel_init(struct drm_panel *panel); >> ^ >> >> (when applied to v5.4-rc7) > > The patchset is based on drm-next from > https://anongit.freedesktop.org/git/drm/drm.git > > For v5.4-rc7 it needs to look like this: > > + drm_panel_init(&ddata->panel); > + ddata->panel.dev = dev; > + ddata->panel.funcs = &dsicm_panel_funcs; Ah, ok. The issue with changed parameters seems to already be fixed by two patches in linux-next and therefore soon in v5.5-rc1. > >>> [...] >> >> Otherwise the patches compile fine and my work-in progress >> DSI panel driver is probing, but not initializing. > > Ok, I tried not to break video mode support, but I do not have any > hardware. Make sure to set the MIPI_DSI_MODE_VIDEO flag in the panel > driver. Indeed, this may be missing (can't look into the code at the moment)... Or I did something wrong when refactoring the driver. We will find out. BR and thanks for the great work, Nikolaus ___ dri-devel mailing list dri-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/dri-devel
Re: [RFCv1 32/42] drm/omap: dsi: convert to drm_panel
Hi, On Sun, Nov 17, 2019 at 08:23:05PM +0100, H. Nikolaus Schaller wrote: > > [...] > > > + drm_panel_init(&ddata->panel, dev, &dsicm_panel_funcs, > > + DRM_MODE_CONNECTOR_DSI); > > + > > This leads to > > drivers/gpu/drm/panel/panel-dsi-cm.c: In function 'dsicm_probe': > drivers/gpu/drm/panel/panel-dsi-cm.c:552:2: error: too many arguments to > function 'drm_panel_init' > drm_panel_init(&ddata->panel, dev, &dsicm_panel_funcs, > ^ > In file included from drivers/gpu/drm/panel/panel-dsi-cm.c:22:0: > ./include/drm/drm_panel.h:150:6: note: declared here > void drm_panel_init(struct drm_panel *panel); > ^ > > (when applied to v5.4-rc7) The patchset is based on drm-next from https://anongit.freedesktop.org/git/drm/drm.git For v5.4-rc7 it needs to look like this: + drm_panel_init(&ddata->panel); + ddata->panel.dev = dev; + ddata->panel.funcs = &dsicm_panel_funcs; > > [...] > > Otherwise the patches compile fine and my work-in progress > DSI panel driver is probing, but not initializing. Ok, I tried not to break video mode support, but I do not have any hardware. Make sure to set the MIPI_DSI_MODE_VIDEO flag in the panel driver. -- Sebastian signature.asc Description: PGP signature ___ dri-devel mailing list dri-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/dri-devel
Re: [RFCv1 32/42] drm/omap: dsi: convert to drm_panel
Hi,
> Am 17.11.2019 um 03:40 schrieb Sebastian Reichel
> :
>
> This converts the DSI module to expect common drm_panel display
> drivers instead of dssdev based ones.
>
> This commit is WIP. We somehow need to know the panels resolution
> in omap_dsi_host_attach(), so that we can properly configure the
> DSI bus clock before enabling the bus. The control bus must be
> enabled at this point, so that the panel can use it.
>
> Other drivers run drm_panel_attach() in their dsi host attach,
> which makes it possible to call drm_panel_get_modes() afterwards
> and the get the correct mode from the connector.
>
> Signed-off-by: Sebastian Reichel
> ---
> .../gpu/drm/omapdrm/displays/panel-dsi-cm.c | 217 +++---
> drivers/gpu/drm/omapdrm/dss/dsi.c | 141 ++--
> .../gpu/drm/omapdrm/dss/omapdss-boot-init.c | 1 -
> drivers/gpu/drm/omapdrm/dss/omapdss.h | 22 +-
> drivers/gpu/drm/omapdrm/omap_crtc.c | 17 +-
> 5 files changed, 205 insertions(+), 193 deletions(-)
>
> diff --git a/drivers/gpu/drm/omapdrm/displays/panel-dsi-cm.c
> b/drivers/gpu/drm/omapdrm/displays/panel-dsi-cm.c
> index 18020ac43a83..84b0e79c025e 100644
> --- a/drivers/gpu/drm/omapdrm/displays/panel-dsi-cm.c
> +++ b/drivers/gpu/drm/omapdrm/displays/panel-dsi-cm.c
> @@ -6,8 +6,6 @@
> * Author: Tomi Valkeinen
> */
>
> -/* #define DEBUG */
> -
> #include
> #include
> #include
> @@ -20,11 +18,14 @@
> #include
>
> #include
> +#include
> +#include
> +#include
>
> +#include
> #include
> #include
> -
> -#include "../dss/omapdss.h"
> +#include
>
> #define DCS_READ_NUM_ERRORS 0x05
> #define DCS_GET_ID1 0xda
> @@ -35,11 +36,8 @@
>
> struct panel_drv_data {
> struct mipi_dsi_device *dsi;
> -
> - struct omap_dss_device dssdev;
> - struct omap_dss_device *src;
> -
> - struct videomode vm;
> + struct drm_panel panel;
> + struct drm_display_mode mode;
>
> struct mutex lock;
>
> @@ -71,7 +69,11 @@ struct panel_drv_data {
> bool ulps_enabled;
> };
>
> -#define to_panel_data(p) container_of(p, struct panel_drv_data, dssdev)
> +
> +static inline struct panel_drv_data *panel_to_ddata(struct drm_panel *panel)
> +{
> + return container_of(panel, struct panel_drv_data, panel);
> +}
>
> static int _dsicm_enable_te(struct panel_drv_data *ddata, bool enable);
>
> @@ -285,10 +287,15 @@ static void dsicm_hw_reset(struct panel_drv_data *ddata)
>
> static int dsicm_power_on(struct panel_drv_data *ddata)
> {
> - struct omap_dss_device *src = ddata->src;
> u8 id1, id2, id3;
> int r;
>
> + r = regulator_bulk_enable(DCS_REGULATOR_SUPPLY_NUM, ddata->supplies);
> + if (r) {
> + dev_err(&ddata->dsi->dev, "failed to enable supplies: %d\n", r);
> + return r;
> + }
> +
> dsicm_hw_reset(ddata);
>
> ddata->dsi->mode_flags |= MIPI_DSI_MODE_LPM;
> @@ -322,10 +329,6 @@ static int dsicm_power_on(struct panel_drv_data *ddata)
> if (r)
> goto err;
>
> - r = src->ops->dsi.enable_video_output(src, ddata->dsi->channel);
> - if (r)
> - goto err;
> -
> ddata->enabled = 1;
>
> if (!ddata->intro_printed) {
> @@ -342,18 +345,17 @@ static int dsicm_power_on(struct panel_drv_data *ddata)
>
> dsicm_hw_reset(ddata);
>
> + regulator_bulk_disable(DCS_REGULATOR_SUPPLY_NUM, ddata->supplies);
> +
> return r;
> }
>
> -static void dsicm_power_off(struct panel_drv_data *ddata)
> +static int dsicm_power_off(struct panel_drv_data *ddata)
> {
> - struct omap_dss_device *src = ddata->src;
> int r;
>
> ddata->enabled = 0;
>
> - src->ops->dsi.disable_video_output(src, ddata->dsi->channel);
> -
> r = mipi_dsi_dcs_set_display_off(ddata->dsi);
> if (!r)
> r = dsicm_sleep_in(ddata);
> @@ -363,52 +365,17 @@ static void dsicm_power_off(struct panel_drv_data
> *ddata)
> "error disabling panel, issuing HW reset\n");
> dsicm_hw_reset(ddata);
> }
> -}
>
> -static int dsicm_connect(struct omap_dss_device *src,
> - struct omap_dss_device *dst)
> -{
> - struct panel_drv_data *ddata = to_panel_data(dst);
> -
> - ddata->src = src;
> - return 0;
> -}
> -
> -static void dsicm_disconnect(struct omap_dss_device *src,
> - struct omap_dss_device *dst)
> -{
> - struct panel_drv_data *ddata = to_panel_data(dst);
> -
> - ddata->src = NULL;
> -}
> -
> -static void dsicm_pre_enable(struct omap_dss_device *dssdev)
> -{
> - struct panel_drv_data *ddata = to_panel_data(dssdev);
> - struct omap_dss_device *src = ddata->src;
> - int r;
> - struct omap_dss_dsi_config dsi_config = {
> - .vm = &ddata->vm,
> - .hs_clk_min = 15000,
> - .hs_clk_max = 3,
> - .lp_clk_min = 700,
> - .lp_clk_max = 1000,
> - };
> -
> - r = regul
