[RFC 00/14] Add DT support to OMAPDSS
Hi, This is an RFC for OMAPDSS DT support. I've only added support for a few boards and a few DSS outputs, but they should give quite a good range of different use cases. If these work well, I think the rest of the outputs and panels will be ok too. The purpose of this series is to get comments about the dts changes. There are still work to be done, like adding DT binding documentation. Some notes: * DSS Submodules The DSS submodules are children of the dss_core node. This is done because the submodules require the dss_core to be alive and initialized to work, even if the submodules are not really behind dss_core. Having the submodules as children will make runtime PM automatically handle the dependency to dss_core. I think usually a node being a child means that it's on the parent's bus, which is not the case here. I'm not sure if that's an issue or not. * ti,dsi-module-id There's a ti,dsi-module-id property in the dsi node. The reason for this module-id property is that we have muxes in the dss_core that route clocks to/from a particular DSI module. So we need some way to index the DSI modules. Another option would be to have a list of DSI modules (phandles) in the dss_core. Both options feel a bit wrong to me, but I went for the module-id approach as that is already used when not using DT. * Display chains Currently omapdss driver can handle only one display device connected to one OMAP SoC output. This is not right in many cases, as there may be multiple display devices in a chain, like first an external encoder and then a panel. However, I tried to model this right in the dts. So, for example, for Panda HDMI output we have the following display "entities" in a chain: OMAP SoC HDMI output, TPD12S015 ESD/level shifter, and HDMI connector. These are connected using the "video-source" property, which tells where the component in question gets its video data. You could ask why there's a separate node for HDMI connector. It's true it's not really a proper device, but we need some kind of terminator for the display chains. For HDMI we could as well have an embedded solution, having a chain like this: SoC HDMI, TPD12S015, hardwired embedded HDMI panel. So the connector marks the end of the chain, and acts as a panel in a sense. * DSI lanes The DSI panels contain "lanes" property, which tells how the SoCs DSI pins are connected to the DSI panel. I'm not sure if the panel data is the proper place for this, or should the data be separately for the OMAP DSI node. Tomi Tomi Valkeinen (14): ARM: OMAP: remove DSS DT hack ARM: OMAP2+: add omapdss_init_of() OMAPDSS: Add DT support to DSS, DISPC, DPI OMAPDSS: Add DT support to DSI OMAPDSS: Add DT support to HDMI OMAPDSS: Taal: Add DT support OMAPDSS: TFP410: Add DT support OMAPDSS: panel-generic-dpi: add DT support ARM: omap3.dtsi: add omapdss information ARM: omap4.dtsi: add omapdss information ARM: omap4-panda.dts: add display information ARM: omap4-sdp.dts: add display information ARM: omap3-tobi.dts: add lcd (TEST) ARM: omap3-beagle.dts: add TFP410 arch/arm/boot/dts/omap3-beagle.dts | 20 + arch/arm/boot/dts/omap3-tobi.dts | 13 arch/arm/boot/dts/omap3.dtsi | 49 arch/arm/boot/dts/omap4-panda.dts| 44 +++ arch/arm/boot/dts/omap4-sdp.dts | 64 arch/arm/boot/dts/omap4.dtsi | 71 + arch/arm/mach-omap2/board-generic.c |9 +-- arch/arm/mach-omap2/common.h |2 + arch/arm/mach-omap2/display.c| 34 + arch/arm/mach-omap2/dss-common.c | 24 -- arch/arm/mach-omap2/dss-common.h |2 - drivers/video/omap2/displays/panel-generic-dpi.c | 47 drivers/video/omap2/displays/panel-taal.c| 89 ++ drivers/video/omap2/displays/panel-tfp410.c | 71 + drivers/video/omap2/dss/dispc.c |7 ++ drivers/video/omap2/dss/dpi.c|8 ++ drivers/video/omap2/dss/dsi.c| 24 +- drivers/video/omap2/dss/dss.c|7 ++ drivers/video/omap2/dss/hdmi.c |7 ++ drivers/video/omap2/dss/hdmi_panel.c | 86 - 20 files changed, 641 insertions(+), 37 deletions(-) -- 1.7.10.4 -- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [RFC 00/14] Add DT support to OMAPDSS
Hi Tomi, On 03/27/2013 09:45 AM, Tomi Valkeinen wrote: > Hi, > > This is an RFC for OMAPDSS DT support. I've only added support for a few > boards > and a few DSS outputs, but they should give quite a good range of different > use > cases. If these work well, I think the rest of the outputs and panels will be > ok too. > > The purpose of this series is to get comments about the dts changes. There are > still work to be done, like adding DT binding documentation. > > Some notes: > > * DSS Submodules > > The DSS submodules are children of the dss_core node. This is done because the > submodules require the dss_core to be alive and initialized to work, even if > the submodules are not really behind dss_core. Having the submodules as > children will make runtime PM automatically handle the dependency to dss_core. > I think usually a node being a child means that it's on the parent's bus, > which > is not the case here. I'm not sure if that's an issue or not. FWIW, there is a L4_DSS interconnect. It is used internally to connect all the submodules to the DSS L3 port. So this representation is perfectly valid and does represent accurately the HW. Regards, Benoit > > * ti,dsi-module-id > > There's a ti,dsi-module-id property in the dsi node. The reason for this > module-id property is that we have muxes in the dss_core that route clocks > to/from a particular DSI module. So we need some way to index the DSI modules. > Another option would be to have a list of DSI modules (phandles) in the > dss_core. Both options feel a bit wrong to me, but I went for the module-id > approach as that is already used when not using DT. > > * Display chains > > Currently omapdss driver can handle only one display device connected to one > OMAP SoC output. This is not right in many cases, as there may be multiple > display devices in a chain, like first an external encoder and then a panel. > However, I tried to model this right in the dts. > > So, for example, for Panda HDMI output we have the following display > "entities" > in a chain: OMAP SoC HDMI output, TPD12S015 ESD/level shifter, and HDMI > connector. These are connected using the "video-source" property, which tells > where the component in question gets its video data. > > You could ask why there's a separate node for HDMI connector. It's true it's > not really a proper device, but we need some kind of terminator for the > display > chains. For HDMI we could as well have an embedded solution, having a chain > like this: SoC HDMI, TPD12S015, hardwired embedded HDMI panel. So the > connector > marks the end of the chain, and acts as a panel in a sense. > > * DSI lanes > > The DSI panels contain "lanes" property, which tells how the SoCs DSI pins are > connected to the DSI panel. I'm not sure if the panel data is the proper place > for this, or should the data be separately for the OMAP DSI node. > > Tomi > > Tomi Valkeinen (14): > ARM: OMAP: remove DSS DT hack > ARM: OMAP2+: add omapdss_init_of() > OMAPDSS: Add DT support to DSS, DISPC, DPI > OMAPDSS: Add DT support to DSI > OMAPDSS: Add DT support to HDMI > OMAPDSS: Taal: Add DT support > OMAPDSS: TFP410: Add DT support > OMAPDSS: panel-generic-dpi: add DT support > ARM: omap3.dtsi: add omapdss information > ARM: omap4.dtsi: add omapdss information > ARM: omap4-panda.dts: add display information > ARM: omap4-sdp.dts: add display information > ARM: omap3-tobi.dts: add lcd (TEST) > ARM: omap3-beagle.dts: add TFP410 > > arch/arm/boot/dts/omap3-beagle.dts | 20 + > arch/arm/boot/dts/omap3-tobi.dts | 13 > arch/arm/boot/dts/omap3.dtsi | 49 > arch/arm/boot/dts/omap4-panda.dts| 44 +++ > arch/arm/boot/dts/omap4-sdp.dts | 64 > arch/arm/boot/dts/omap4.dtsi | 71 + > arch/arm/mach-omap2/board-generic.c |9 +-- > arch/arm/mach-omap2/common.h |2 + > arch/arm/mach-omap2/display.c| 34 + > arch/arm/mach-omap2/dss-common.c | 24 -- > arch/arm/mach-omap2/dss-common.h |2 - > drivers/video/omap2/displays/panel-generic-dpi.c | 47 > drivers/video/omap2/displays/panel-taal.c| 89 > ++ > drivers/video/omap2/displays/panel-tfp410.c | 71 + > drivers/video/omap2/dss/dispc.c |7 ++ > drivers/video/omap2/dss/dpi.c|8 ++ > drivers/video/omap2/dss/dsi.c| 24 +- > drivers/video/omap2/dss/dss.c|7 ++ > drivers/video/omap2/dss/hdmi.c |7 ++ > drivers/video/omap2/dss/hdmi_panel.c | 86 - > 20 files changed, 641 insertions(+), 37 deletions(-) > -- To unsubscribe from this list: send the line
Re: [RFC 00/14] Add DT support to OMAPDSS
+ mturquette and Rajendra On 03/27/2013 09:45 AM, Tomi Valkeinen wrote: ... > * ti,dsi-module-id > > There's a ti,dsi-module-id property in the dsi node. The reason for this > module-id property is that we have muxes in the dss_core that route clocks > to/from a particular DSI module. So we need some way to index the DSI modules. > Another option would be to have a list of DSI modules (phandles) in the > dss_core. Both options feel a bit wrong to me, but I went for the module-id > approach as that is already used when not using DT. That's not a very nice representation. Ideally you should expose 1 clock node per DSI node, and enabling one will select the proper mux for you. Regards, Benoit > > * Display chains > > Currently omapdss driver can handle only one display device connected to one > OMAP SoC output. This is not right in many cases, as there may be multiple > display devices in a chain, like first an external encoder and then a panel. > However, I tried to model this right in the dts. > > So, for example, for Panda HDMI output we have the following display > "entities" > in a chain: OMAP SoC HDMI output, TPD12S015 ESD/level shifter, and HDMI > connector. These are connected using the "video-source" property, which tells > where the component in question gets its video data. > > You could ask why there's a separate node for HDMI connector. It's true it's > not really a proper device, but we need some kind of terminator for the > display > chains. For HDMI we could as well have an embedded solution, having a chain > like this: SoC HDMI, TPD12S015, hardwired embedded HDMI panel. So the > connector > marks the end of the chain, and acts as a panel in a sense. > > * DSI lanes > > The DSI panels contain "lanes" property, which tells how the SoCs DSI pins are > connected to the DSI panel. I'm not sure if the panel data is the proper place > for this, or should the data be separately for the OMAP DSI node. > > Tomi > > Tomi Valkeinen (14): > ARM: OMAP: remove DSS DT hack > ARM: OMAP2+: add omapdss_init_of() > OMAPDSS: Add DT support to DSS, DISPC, DPI > OMAPDSS: Add DT support to DSI > OMAPDSS: Add DT support to HDMI > OMAPDSS: Taal: Add DT support > OMAPDSS: TFP410: Add DT support > OMAPDSS: panel-generic-dpi: add DT support > ARM: omap3.dtsi: add omapdss information > ARM: omap4.dtsi: add omapdss information > ARM: omap4-panda.dts: add display information > ARM: omap4-sdp.dts: add display information > ARM: omap3-tobi.dts: add lcd (TEST) > ARM: omap3-beagle.dts: add TFP410 > > arch/arm/boot/dts/omap3-beagle.dts | 20 + > arch/arm/boot/dts/omap3-tobi.dts | 13 > arch/arm/boot/dts/omap3.dtsi | 49 > arch/arm/boot/dts/omap4-panda.dts| 44 +++ > arch/arm/boot/dts/omap4-sdp.dts | 64 > arch/arm/boot/dts/omap4.dtsi | 71 + > arch/arm/mach-omap2/board-generic.c |9 +-- > arch/arm/mach-omap2/common.h |2 + > arch/arm/mach-omap2/display.c| 34 + > arch/arm/mach-omap2/dss-common.c | 24 -- > arch/arm/mach-omap2/dss-common.h |2 - > drivers/video/omap2/displays/panel-generic-dpi.c | 47 > drivers/video/omap2/displays/panel-taal.c| 89 > ++ > drivers/video/omap2/displays/panel-tfp410.c | 71 + > drivers/video/omap2/dss/dispc.c |7 ++ > drivers/video/omap2/dss/dpi.c|8 ++ > drivers/video/omap2/dss/dsi.c| 24 +- > drivers/video/omap2/dss/dss.c|7 ++ > drivers/video/omap2/dss/hdmi.c |7 ++ > drivers/video/omap2/dss/hdmi_panel.c | 86 - > 20 files changed, 641 insertions(+), 37 deletions(-) > -- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [RFC 00/14] Add DT support to OMAPDSS
On 2013-03-27 11:30, Benoit Cousson wrote: > Hi Tomi, > > On 03/27/2013 09:45 AM, Tomi Valkeinen wrote: >> Hi, >> >> This is an RFC for OMAPDSS DT support. I've only added support for a few >> boards >> and a few DSS outputs, but they should give quite a good range of different >> use >> cases. If these work well, I think the rest of the outputs and panels will be >> ok too. >> >> The purpose of this series is to get comments about the dts changes. There >> are >> still work to be done, like adding DT binding documentation. >> >> Some notes: >> >> * DSS Submodules >> >> The DSS submodules are children of the dss_core node. This is done because >> the >> submodules require the dss_core to be alive and initialized to work, even if >> the submodules are not really behind dss_core. Having the submodules as >> children will make runtime PM automatically handle the dependency to >> dss_core. >> I think usually a node being a child means that it's on the parent's bus, >> which >> is not the case here. I'm not sure if that's an issue or not. > > FWIW, there is a L4_DSS interconnect. It is used internally to connect > all the submodules to the DSS L3 port. So this representation is > perfectly valid and does represent accurately the HW. Ah, yes, I can see it mentioned in the OMAP4430 Block Diagram figure in the TRM. No other mentions, though, I guess it's not really relevant =). But good to know that the DT representation is actually correct. Tomi signature.asc Description: OpenPGP digital signature
Re: [RFC 00/14] Add DT support to OMAPDSS
On 2013-03-27 11:41, Benoit Cousson wrote: > + mturquette and Rajendra > > On 03/27/2013 09:45 AM, Tomi Valkeinen wrote: > > ... > >> * ti,dsi-module-id >> >> There's a ti,dsi-module-id property in the dsi node. The reason for this >> module-id property is that we have muxes in the dss_core that route clocks >> to/from a particular DSI module. So we need some way to index the DSI >> modules. >> Another option would be to have a list of DSI modules (phandles) in the >> dss_core. Both options feel a bit wrong to me, but I went for the module-id >> approach as that is already used when not using DT. > > That's not a very nice representation. Ideally you should expose 1 clock > node per DSI node, and enabling one will select the proper mux for you. Afaik, we need the DSI module id for the following things: 1) DSI pad configuration (OMAP4_CTRL_MODULE_PAD_CORE_CONTROL_DSIPHY). 2) Setting clk mux in dss_core to select input clock for DSI protocol engine. The choice is between DSS_CLK from PRCM, or clock from a DSI PLL. DSI1 module can only use DSI1 PLL clock, and DSI2 module can only use DSI2 PLL clock. 3) Setting clk mux in dss_core to select input clock for LCD/DISPC. The choice is between DSS_CLK from PRCM, or a clock from DSI PLL. DISPC can use either DSI1 or DSI2 PLL, LCD1 can use DSI1 PLL clock, LCD2 can use DSI2 PLL clock. 4) Selecting the video output from DISPC which is routed to the DSI module. LCD1 output can go to DSI1, LCD2 output can go to DSI1 or DSI2. All the above of course change between OMAP versions. So managing all the above was easiest with a plain module-id assigned to the DSI device. I guess 1) could be somehow managed by having separate nodes in the dts for the DSIPHY control, and a link from the DSI module to that DSIPHY node. 4) could possibly be managed by having each DSS output node having a list of possible inputs from the DISPC. And, I guess, a link somehow to a mux node in the dss_core. Or something... All in all, I don't have any clear idea how these should be handled. Tomi signature.asc Description: OpenPGP digital signature