| Message ID | 20221129172912.9576-1-macroalpha82@gmail.com |
|---|---|
| Headers | show |
| Series | drm/panel: Add Samsung AMS495QA01 Panel | expand |
On Tue, 29 Nov 2022 11:29:10 -0600, Chris Morgan wrote: > From: Chris Morgan <macromorgan@hotmail.com> > > Add documentation for the Samsung AMS495QA01 panel. > > Signed-off-by: Chris Morgan <macromorgan@hotmail.com> > Signed-off-by: Maya Matuszczyk <maccraft123mc@gmail.com> > --- > .../display/panel/samsung,ams495qa01.yaml | 57 +++++++++++++++++++ > 1 file changed, 57 insertions(+) > create mode 100644 Documentation/devicetree/bindings/display/panel/samsung,ams495qa01.yaml > Reviewed-by: Rob Herring <robh@kernel.org>
Hi Chris, thanks for your patch! On Tue, Nov 29, 2022 at 6:29 PM Chris Morgan <macroalpha82@gmail.com> wrote: > From: Chris Morgan <macromorgan@hotmail.com> > > Add documentation for the Samsung AMS495QA01 panel. > > Signed-off-by: Chris Morgan <macromorgan@hotmail.com> > Signed-off-by: Maya Matuszczyk <maccraft123mc@gmail.com> > + reset-gpios: true Can you add a description saying that this must always be specified GPIO_ACTIVE_LOW. With that change: Reviewed-by: Linus Walleij <linus.walleij@linaro.org> Yours, Linus Walleij
Hi Chris, thanks for your patch! overall this looks very good. Just some nitpicks. On Tue, Nov 29, 2022 at 6:29 PM Chris Morgan <macroalpha82@gmail.com> wrote: > From: Chris Morgan <macromorgan@hotmail.com> > > Support Samsung AMS495QA01 panel as found on the Anbernic RG503. Note > This panel receives video signals via DSI, however it receives > commands via 3-wire SPI. > > Signed-off-by: Chris Morgan <macromorgan@hotmail.com> > Signed-off-by: Maya Matuszczyk <maccraft123mc@gmail.com> > +config DRM_PANEL_SAMSUNG_AMS495QA01 > + tristate "Samsung AMS495QA01 DSI panel" I am always a bit careful with these "Samsung" panels. Well the panel is surely Samsungs and usually it has that name, but very often it is mainly a display controller, made by someone else, then a physical panel slapped on and then the actual display controller is obscured. Typical example: panel-novatek-nt35560.c handing Samsung ACX424AKP. If you know the actual display controller, then the driver should be renamed after that, but keeping the compatible and DT bindings as is. The reason being that tomorrow there is an LG panel using the same display controller and then we don't get duplicate code. It feels like this is a Novatek controller. Just a feeling I have. Mostly from the way it does brightness using ELVSS and gamma. But who knows. > + depends on OF && SPI > + depends on DRM_MIPI_DSI > + select DRM_MIPI_DBI Nice that you use these helpers! > +#include <linux/media-bus-format.h> (...) > +#include <linux/of_graph.h> Hm. We get to this. > + /** @prepared: value tracking panel prepare status */ > + bool prepared; Drop this and associated code. The framework keeps track of this. > + /* Set gamma change */ > + mipi_dbi_command(dbi, 0xf9, 0x00); For cases like this where you know what the command does, please add: #define MCS_SET_GAMMA 0xf9 (MCS = Manufacturer Command Set, cf drivers/gpu/drm/panel/panel-samsung-s6e63m0.h) and collect them somewhere, then you can drop the comment because it is self-evident what is going on. > + /* Undocumented command */ > + mipi_dbi_command(dbi, 0x26, 0x00); > + /* Set ELVSS value */ > + mipi_dbi_command(dbi, 0xb2, ams495qa01_elvss[tmp]); Same. > +static int ams495qa01_prepare(struct drm_panel *panel) > +{ > + struct ams495qa01 *db = to_ams495qa01(panel); > + struct mipi_dbi *dbi = &db->dbi; > + int ret; > + > + if (db->prepared) > + return 0; As mentioned skip this. > + /* Power up */ > + ret = regulator_enable(db->reg_vdd); > + if (ret) { > + dev_err(db->dev, "failed to enable vdd regulator: %d\n", ret); > + return ret; > + } > + if (db->reg_elvdd) { Do you really need to if() this? I thought the regulator framework would just ignore the calls for an optional regulator. > + /* Password to start command sequence */ > + mipi_dbi_command(dbi, 0xf0, 0x5a, 0x5a); > + mipi_dbi_command(dbi, 0xf1, 0x5a, 0x5a); Use the #defines from: drivers/gpu/drm/panel/panel-samsung-s6e63m0.h I'm pretty sure they mean the same thing: #define MCS_LEVEL_2_KEY 0xf0 #define MCS_MTP_KEY 0xf1 > + /* Analog Power condition set */ > + mipi_dbi_command(dbi, 0xf4, 0x33, 0x42, 0x00, 0x08); > + mipi_dbi_command(dbi, 0xf5, 0x00, 0x06, 0x26, 0x35, 0x03); Define MCS_ commands. > + /* GTCON set */ > + mipi_dbi_command(dbi, 0xf7, 0x20); > + > + /* TEMP_SWIRE set */ > + mipi_dbi_command(dbi, 0xb2, 0x06, 0x06, 0x06, 0x06); > + > + /* ELVSS_CON set */ > + mipi_dbi_command(dbi, 0xb1, 0x07, 0x00, 0x10); > + > + /* Gateless signal set */ > + mipi_dbi_command(dbi, 0xf8, 0x7f, 0x7a, 0x89, 0x67, 0x26, 0x38, > + 0x00, 0x00, 0x09, 0x67, 0x70, 0x88, 0x7a, > + 0x76, 0x05, 0x09, 0x23, 0x23, 0x23); Dito > + db->prepared = true; Drop. > +static int ams495qa01_unprepare(struct drm_panel *panel) > +{ > + struct ams495qa01 *db = to_ams495qa01(panel); > + struct mipi_dbi *dbi = &db->dbi; > + > + if (!db->prepared) > + return 0; Drop. > +static int ams495qa01_get_modes(struct drm_panel *panel, > + struct drm_connector *connector) > +{ > + struct ams495qa01 *db = to_ams495qa01(panel); > + const struct ams495qa01_panel_info *panel_info = db->panel_info; > + struct drm_display_mode *mode; > + static const u32 bus_format = MEDIA_BUS_FMT_RGB888_1X24; > + unsigned int i; > + > + for (i = 0; i < panel_info->num_modes; i++) { > + mode = drm_mode_duplicate(connector->dev, > + &panel_info->display_modes[i]); > + if (!mode) > + return -ENOMEM; > + > + drm_mode_set_name(mode); > + > + mode->type = DRM_MODE_TYPE_DRIVER; > + if (panel_info->num_modes == 1) > + mode->type |= DRM_MODE_TYPE_PREFERRED; I think you should probably set the preferred mode even if there are several of them? But maybe just on the first or something. (A bit unsure here, Sam?) > +static int ams495qa01_probe(struct spi_device *spi) > +{ > + struct device *dev = &spi->dev; > + struct device_node *endpoint, *dsi_host_node; (...) > + db->reset = devm_gpiod_get(dev, "reset", GPIOD_OUT_LOW); I would request this GPIOD_OUT_HIGH so reset is asserted on probe. > + db->enable = devm_gpiod_get_optional(dev, "enable", GPIOD_OUT_HIGH); And this GPIOD_OUT_LOW so panel is disabled after probe. > + /* > + * Get the DSI controller that is supplying data for this display > + * which is controlled via SPI 3-wire. > + */ > + endpoint = of_graph_get_next_endpoint(dev->of_node, NULL); > + if (!endpoint) { > + dev_err(dev, "failed to get endpoint\n"); > + return -ENODEV; > + } > + dsi_host_node = of_graph_get_remote_port_parent(endpoint); > + if (!dsi_host_node) { > + dev_err(dev, "failed to get remote port parent\n"); > + goto put_endpoint; > + } > + dsi_host = of_find_mipi_dsi_host_by_node(dsi_host_node); > + if (!dsi_host) { > + dev_err(dev, "failed to find dsi host\n"); > + goto put_host; > + } > + info.node = of_graph_get_remote_port(endpoint); > + if (!info.node) { > + dev_err(dev, "failed to get remote port node\n"); > + ret = -ENODEV; > + goto put_host; > + } > + > + db->dsi_dev = devm_mipi_dsi_device_register_full(dev, dsi_host, &info); I don't get this part. Why do you have to go through all this trouble when no other panel drivers, not even the other SPI DBI panel drivers, does this? drm_of_find_panel_or_bridge() should typically result in the core doing all the lookup and connecting for you from the host side should it not? At the very least this requires a big comment about what is going on and why and what makes this necessary. I can only guess the following: > + db->dsi_dev->lanes = 2; > + db->dsi_dev->format = MIPI_DSI_FMT_RGB888; > + db->dsi_dev->mode_flags = MIPI_DSI_MODE_VIDEO | MIPI_DSI_MODE_VIDEO_BURST | > + MIPI_DSI_MODE_LPM | MIPI_DSI_MODE_NO_EOT_PACKET; > + > + drm_panel_init(&db->panel, dev, &ams495qa01_drm_funcs, > + DRM_MODE_CONNECTOR_DSI); Pixel data passes to the display using DSI but all display control is done over SPI, and the core will not help with this. So from the display controller POV this is a DSI display and from the display POV this is an SPI-controlled display. So it sits on two buses. Data path is on DSI, control path is on SPI. This would be kind of odd actually, normally DSI displays do the display control over DSI as well. SPI control is usually used on DPI displays. But I'm not surprised. If this is necessary, isn't this something we need to teach the core to handle instead of adding quirks like this to all drivers that have this characteristic? Yours, Linus Walleij
Hi Linus, Am Samstag, 3. Dezember 2022, 10:03:42 CET schrieb Linus Walleij: > On Tue, Nov 29, 2022 at 6:29 PM Chris Morgan <macroalpha82@gmail.com> wrote: > > > From: Chris Morgan <macromorgan@hotmail.com> > > > > Support Samsung AMS495QA01 panel as found on the Anbernic RG503. Note > > This panel receives video signals via DSI, however it receives > > commands via 3-wire SPI. > > > > Signed-off-by: Chris Morgan <macromorgan@hotmail.com> > > Signed-off-by: Maya Matuszczyk <maccraft123mc@gmail.com> > > > +config DRM_PANEL_SAMSUNG_AMS495QA01 > > + tristate "Samsung AMS495QA01 DSI panel" > > I am always a bit careful with these "Samsung" panels. Well the > panel is surely Samsungs and usually it has that name, but very often > it is mainly a display controller, made by someone else, then a physical > panel slapped on and then the actual display controller is obscured. > > Typical example: > panel-novatek-nt35560.c handing Samsung ACX424AKP. > > If you know the actual display controller, then the driver should be > renamed after that, but keeping the compatible and DT bindings > as is. The reason being that tomorrow there is an LG panel > using the same display controller and then we don't get duplicate > code. > > It feels like this is a Novatek controller. Just a feeling I have. > Mostly from the way it does brightness using ELVSS and > gamma. But who knows. Though in past projects I've seen the same display-controller used with different panels (and different dsi-init-sequences). In one project the display manufacturer even EOL'ed the first panel and provided a replacement with said same display controller (and a different init) - but the datasheets for the display-controller were for the same chip still. So while in my experience the actual display name from the manufacturer identifies the display + controller combo, I don't really think you can go the other way with the controller name identifying the display+controller combination. But that whole display business is arcane anyway, so this also only stems from past projects and not factual knowledge ;-) . Heiko
On Sat, Dec 03, 2022 at 10:03:42AM +0100, Linus Walleij wrote: > Hi Chris, > > thanks for your patch! > > overall this looks very good. Just some nitpicks. > > On Tue, Nov 29, 2022 at 6:29 PM Chris Morgan <macroalpha82@gmail.com> wrote: > > > From: Chris Morgan <macromorgan@hotmail.com> > > > > Support Samsung AMS495QA01 panel as found on the Anbernic RG503. Note > > This panel receives video signals via DSI, however it receives > > commands via 3-wire SPI. > > > > Signed-off-by: Chris Morgan <macromorgan@hotmail.com> > > Signed-off-by: Maya Matuszczyk <maccraft123mc@gmail.com> > > > +config DRM_PANEL_SAMSUNG_AMS495QA01 > > + tristate "Samsung AMS495QA01 DSI panel" > > I am always a bit careful with these "Samsung" panels. Well the > panel is surely Samsungs and usually it has that name, but very often > it is mainly a display controller, made by someone else, then a physical > panel slapped on and then the actual display controller is obscured. > > Typical example: > panel-novatek-nt35560.c handing Samsung ACX424AKP. > > If you know the actual display controller, then the driver should be > renamed after that, but keeping the compatible and DT bindings > as is. The reason being that tomorrow there is an LG panel > using the same display controller and then we don't get duplicate > code. > > It feels like this is a Novatek controller. Just a feeling I have. > Mostly from the way it does brightness using ELVSS and > gamma. But who knows. > Will do. I'll make the changes and resubmit. For what it's worth the documentation says this one is a Samsung AMS495QA01 panel on a Magnachip D53E6EA8966 controller IC. > > + depends on OF && SPI > > + depends on DRM_MIPI_DSI > > + select DRM_MIPI_DBI > > Nice that you use these helpers! > > > +#include <linux/media-bus-format.h> > (...) > > +#include <linux/of_graph.h> > > Hm. We get to this. *gulp* > > > + /** @prepared: value tracking panel prepare status */ > > + bool prepared; > > Drop this and associated code. The framework keeps track of this. > Will do. > > + /* Set gamma change */ > > + mipi_dbi_command(dbi, 0xf9, 0x00); > > For cases like this where you know what the command does, > please add: > > #define MCS_SET_GAMMA 0xf9 > > (MCS = Manufacturer Command Set, cf > drivers/gpu/drm/panel/panel-samsung-s6e63m0.h) > > and collect them somewhere, then you can drop the comment > because it is self-evident what is going on. > Gotcha, will do. > > + /* Undocumented command */ > > + mipi_dbi_command(dbi, 0x26, 0x00); > > + /* Set ELVSS value */ > > + mipi_dbi_command(dbi, 0xb2, ams495qa01_elvss[tmp]); > > Same. > > > +static int ams495qa01_prepare(struct drm_panel *panel) > > +{ > > + struct ams495qa01 *db = to_ams495qa01(panel); > > + struct mipi_dbi *dbi = &db->dbi; > > + int ret; > > + > > + if (db->prepared) > > + return 0; > > As mentioned skip this. > > > + /* Power up */ > > + ret = regulator_enable(db->reg_vdd); > > + if (ret) { > > + dev_err(db->dev, "failed to enable vdd regulator: %d\n", ret); > > + return ret; > > + } > > + if (db->reg_elvdd) { > > Do you really need to if() this? I thought the regulator > framework would just ignore the calls for an optional > regulator. I don't know for sure, but I'll make the change if you request it. I think other drivers had an if in this scenario which is why I did it. > > > + /* Password to start command sequence */ > > + mipi_dbi_command(dbi, 0xf0, 0x5a, 0x5a); > > + mipi_dbi_command(dbi, 0xf1, 0x5a, 0x5a); > > Use the #defines from: > drivers/gpu/drm/panel/panel-samsung-s6e63m0.h > I'm pretty sure they mean the same thing: > > #define MCS_LEVEL_2_KEY 0xf0 > #define MCS_MTP_KEY 0xf1 > > > + /* Analog Power condition set */ > > + mipi_dbi_command(dbi, 0xf4, 0x33, 0x42, 0x00, 0x08); > > + mipi_dbi_command(dbi, 0xf5, 0x00, 0x06, 0x26, 0x35, 0x03); > > Define MCS_ commands. > > > + /* GTCON set */ > > + mipi_dbi_command(dbi, 0xf7, 0x20); > > + > > + /* TEMP_SWIRE set */ > > + mipi_dbi_command(dbi, 0xb2, 0x06, 0x06, 0x06, 0x06); > > + > > + /* ELVSS_CON set */ > > + mipi_dbi_command(dbi, 0xb1, 0x07, 0x00, 0x10); > > + > > + /* Gateless signal set */ > > + mipi_dbi_command(dbi, 0xf8, 0x7f, 0x7a, 0x89, 0x67, 0x26, 0x38, > > + 0x00, 0x00, 0x09, 0x67, 0x70, 0x88, 0x7a, > > + 0x76, 0x05, 0x09, 0x23, 0x23, 0x23); > > Dito > > > + db->prepared = true; > > Drop. > > > +static int ams495qa01_unprepare(struct drm_panel *panel) > > +{ > > + struct ams495qa01 *db = to_ams495qa01(panel); > > + struct mipi_dbi *dbi = &db->dbi; > > + > > + if (!db->prepared) > > + return 0; > > Drop. > > > +static int ams495qa01_get_modes(struct drm_panel *panel, > > + struct drm_connector *connector) > > +{ > > + struct ams495qa01 *db = to_ams495qa01(panel); > > + const struct ams495qa01_panel_info *panel_info = db->panel_info; > > + struct drm_display_mode *mode; > > + static const u32 bus_format = MEDIA_BUS_FMT_RGB888_1X24; > > + unsigned int i; > > + > > + for (i = 0; i < panel_info->num_modes; i++) { > > + mode = drm_mode_duplicate(connector->dev, > > + &panel_info->display_modes[i]); > > + if (!mode) > > + return -ENOMEM; > > + > > + drm_mode_set_name(mode); > > + > > + mode->type = DRM_MODE_TYPE_DRIVER; > > + if (panel_info->num_modes == 1) > > + mode->type |= DRM_MODE_TYPE_PREFERRED; > > I think you should probably set the preferred mode even > if there are several of them? But maybe just on the first > or something. (A bit unsure here, Sam?) > I'll keep 60hz as the preferred. 50hz was added at the request of some userspace folks for running PAL based emulators and stuff. > > +static int ams495qa01_probe(struct spi_device *spi) > > +{ > > + struct device *dev = &spi->dev; > > + struct device_node *endpoint, *dsi_host_node; > (...) > > + db->reset = devm_gpiod_get(dev, "reset", GPIOD_OUT_LOW); > > I would request this GPIOD_OUT_HIGH so reset is asserted on probe. > Will do. > > + db->enable = devm_gpiod_get_optional(dev, "enable", GPIOD_OUT_HIGH); > > And this GPIOD_OUT_LOW so panel is disabled after probe. > Will do. > > + /* > > + * Get the DSI controller that is supplying data for this display > > + * which is controlled via SPI 3-wire. > > + */ > > + endpoint = of_graph_get_next_endpoint(dev->of_node, NULL); > > + if (!endpoint) { > > + dev_err(dev, "failed to get endpoint\n"); > > + return -ENODEV; > > + } > > + dsi_host_node = of_graph_get_remote_port_parent(endpoint); > > + if (!dsi_host_node) { > > + dev_err(dev, "failed to get remote port parent\n"); > > + goto put_endpoint; > > + } > > + dsi_host = of_find_mipi_dsi_host_by_node(dsi_host_node); > > + if (!dsi_host) { > > + dev_err(dev, "failed to find dsi host\n"); > > + goto put_host; > > + } > > + info.node = of_graph_get_remote_port(endpoint); > > + if (!info.node) { > > + dev_err(dev, "failed to get remote port node\n"); > > + ret = -ENODEV; > > + goto put_host; > > + } > > + > > + db->dsi_dev = devm_mipi_dsi_device_register_full(dev, dsi_host, &info); > > I don't get this part. > > Why do you have to go through all this trouble when no other panel > drivers, not even the other SPI DBI panel drivers, does this? > > drm_of_find_panel_or_bridge() should typically result in the core > doing all the lookup and connecting for you from the host side > should it not? > > At the very least this requires a big comment about what is > going on and why and what makes this necessary. > > I can only guess the following: > This is the path that "works", but I'll happily change to something else. > > + db->dsi_dev->lanes = 2; > > + db->dsi_dev->format = MIPI_DSI_FMT_RGB888; > > + db->dsi_dev->mode_flags = MIPI_DSI_MODE_VIDEO | MIPI_DSI_MODE_VIDEO_BURST | > > + MIPI_DSI_MODE_LPM | MIPI_DSI_MODE_NO_EOT_PACKET; > > + > > + drm_panel_init(&db->panel, dev, &ams495qa01_drm_funcs, > > + DRM_MODE_CONNECTOR_DSI); > > Pixel data passes to the display using DSI but all display control > is done over SPI, and the core will not help with this. > > So from the display controller POV this is a DSI display > and from the display POV this is an SPI-controlled display. > So it sits on two buses. Data path is on DSI, control path > is on SPI. > > This would be kind of odd actually, normally DSI displays > do the display control over DSI as well. SPI control is usually > used on DPI displays. But I'm not surprised. > > If this is necessary, isn't this something we need to teach the > core to handle instead of adding quirks like this to all drivers that > have this characteristic? > You are correct, this panel is controlled via 3-wire SPI in my example. The panel can be controlled either by 3-wire SPI or DSI commands depending on whether or not pin 15 is driven high or low. Unfortunately in my case it's hardwired high, so I am forced to do it via 3-wire SPI. I have no way of testing it with pure DSI but that would simplify things quite a bit. Pixel data is transmitted soley through DSI. The way I have it implemented currently is to put the panel on the SPI bus as a DBI panel; traverse through the DT bindings to find the associated DSI controller, then attach it as a DSI device so the DSI bus can transmit the pixel data. I'm absolutely cool with making those functions part of the core and not just specific to this panel, only I might need a bit of help on that part to make sure I do it the right way. I just wasn't sure how often that would be needed since this is the only panel I've ever seen driven this way, especially since it seems like any sane person would just want to do the whole control data/pixel data over DSI to keep things simple. > Yours, > Linus Walleij Thank you for your input.
On Sat, Dec 3, 2022 at 11:18 AM Heiko Stübner <heiko@sntech.de> wrote: > Though in past projects I've seen the same display-controller used with > different panels (and different dsi-init-sequences). In one project the > display manufacturer even EOL'ed the first panel and provided a replacement > with said same display controller (and a different init) - but the > datasheets for the display-controller were for the same chip still. > > So while in my experience the actual display name from the manufacturer > identifies the display + controller combo, I don't really think you can > go the other way with the controller name identifying the display+controller > combination. I don't mean we should do that. What I mean is: - Name the driver after the controller, if we know which one it is, such as panel-novatek-nt35510.c - Provide identifications of the controller+panel combo in e.g. DT compatible strings, provide init data as arrays in the per-variant match data (or similar for ACPI). Yours, Linus Walleij
On Sat, Dec 3, 2022 at 8:01 PM Chris Morgan <macroalpha82@gmail.com> wrote: > Will do. I'll make the changes and resubmit. For what it's worth the > documentation says this one is a Samsung AMS495QA01 panel on a > Magnachip D53E6EA8966 controller IC. I would name the driver panel-magnachip-d53e6ea8966.c and KConfig PANEL_MAGNACHIP_D53E6EA8966 for now but keep the Samsung compatible string & match. Maybe this driver will match more magnachips in the future, maybe not. Sometimes people get hold of datasheets and submit proper code and #defines etc. > > > + if (db->reg_elvdd) { > > > > Do you really need to if() this? I thought the regulator > > framework would just ignore the calls for an optional > > regulator. > > I don't know for sure, but I'll make the change if you request it. I > think other drivers had an if in this scenario which is why I did it. Okay but I don't think that is necessary. > > > + mode->type = DRM_MODE_TYPE_DRIVER; > > > + if (panel_info->num_modes == 1) > > > + mode->type |= DRM_MODE_TYPE_PREFERRED; > > > > I think you should probably set the preferred mode even > > if there are several of them? But maybe just on the first > > or something. (A bit unsure here, Sam?) > > I'll keep 60hz as the preferred. 50hz was added at the request of some > userspace folks for running PAL based emulators and stuff. OK > This is the path that "works", but I'll happily change to something > else. > > > > + db->dsi_dev->lanes = 2; > > > + db->dsi_dev->format = MIPI_DSI_FMT_RGB888; > > > + db->dsi_dev->mode_flags = MIPI_DSI_MODE_VIDEO | MIPI_DSI_MODE_VIDEO_BURST | > > > + MIPI_DSI_MODE_LPM | MIPI_DSI_MODE_NO_EOT_PACKET; > > > + > > > + drm_panel_init(&db->panel, dev, &ams495qa01_drm_funcs, > > > + DRM_MODE_CONNECTOR_DSI); > > > > Pixel data passes to the display using DSI but all display control > > is done over SPI, and the core will not help with this. > > > > So from the display controller POV this is a DSI display > > and from the display POV this is an SPI-controlled display. > > So it sits on two buses. Data path is on DSI, control path > > is on SPI. > > > > This would be kind of odd actually, normally DSI displays > > do the display control over DSI as well. SPI control is usually > > used on DPI displays. But I'm not surprised. > > > > If this is necessary, isn't this something we need to teach the > > core to handle instead of adding quirks like this to all drivers that > > have this characteristic? > > > > You are correct, this panel is controlled via 3-wire SPI in my example. > The panel can be controlled either by 3-wire SPI or DSI commands > depending on whether or not pin 15 is driven high or low. Unfortunately > in my case it's hardwired high, so I am forced to do it via 3-wire SPI. > I have no way of testing it with pure DSI but that would simplify > things quite a bit. Pixel data is transmitted soley through DSI. OK > The way I have it implemented currently is to put the panel on the SPI > bus as a DBI panel; traverse through the DT bindings to find the > associated DSI controller, then attach it as a DSI device so the DSI > bus can transmit the pixel data. > > I'm absolutely cool with making those functions part of the core and > not just specific to this panel, only I might need a bit of help on > that part to make sure I do it the right way. I just wasn't sure how > often that would be needed since this is the only panel I've ever seen > driven this way, especially since it seems like any sane person would > just want to do the whole control data/pixel data over DSI to keep > things simple. It doesn't seem all that unique. Can you put some helper in drivers/gpu/drm/drm_of.c with the rest and it will (hopefully) not be linked in unless used anyway. Yours, Linus Walleij
From: Chris Morgan <macromorgan@hotmail.com> Add the Samsung AMS495QA01 panel as found on the Anbernic RG503. This panel uses DSI to receive video signals, but 3-wire SPI to receive command signals. Changes since V3: - Updated documentation to add spi-peripheral-props.yaml per updates made for similar devices. Note that I removed a "Reviewed-by" tag from Rob Herring since this change probably needs to be confirmed. - Added binding for RG503, since this device is now accepted with this request: https://lore.kernel.org/linux-rockchip/166274831283.21181.6861718157177507544.b4-ty@sntech.de/ Changes since V2: - Added 50hz mode at request of userspace devs. - Renamed "dupa" to panel name. Good catch Maya. - Added Maya's Signed-off-by. - Removed check for max backlight, since it is already done by backlight_device_set_brightness. - Fixed minor formatting issues on devicetree binding documentation and added port to provided example. Changes since V1: - Removed errant reference to backlight in documentation. This is an OLED panel. - Made elvss regulator optional. In my case its hard wired and not controllable. - Added "prepared" enum to track panel status to prevent unbalanced regulator enable/disable. Chris Morgan (3): dt-bindings: display: panel: Add Samsung AMS495QA01 drm/panel: Add Samsung AMS495QA01 MIPI-DSI LCD panel arm64: dts: rockchip: add display to RG503 .../display/panel/samsung,ams495qa01.yaml | 57 ++ .../dts/rockchip/rk3566-anbernic-rg503.dts | 61 ++ drivers/gpu/drm/panel/Kconfig | 10 + drivers/gpu/drm/panel/Makefile | 1 + .../gpu/drm/panel/panel-samsung-ams495qa01.c | 547 ++++++++++++++++++ 5 files changed, 676 insertions(+) create mode 100644 Documentation/devicetree/bindings/display/panel/samsung,ams495qa01.yaml create mode 100644 drivers/gpu/drm/panel/panel-samsung-ams495qa01.c