This patchset adds the last bits needed for getting the MSM display bindings in correct shape, and as an example, adds display support for MSM8916. One problem with the MDP5 driver was that device hierarchy didn't match with the hardware. All MDP5 based display blocks contain a top-level MDSS wrapper hardware that manages IRQs, power and some clocks for the sub-blocks (MDP5, DSI, HDMI, eDP etc) within it. The driver stuffs this functionality within the MDP5 driver itself, which results in probably not the cleanest design, and forces us to repeat some resources (like, power domain, some top level clocks) across all the sub-blocks. This is fixed by creating separate MDP5 and MDSS platform devices, and making each piece manage its own resources. MDP4 still continues to have the flat device hierarchy. The second problem was the non-standard connector and gpu DT properites. The former was needed to bind all the external components (DSI, HDMI etc) before we started with intializing modeset. This is fixed by representing the MDP interface outputs as ports, and linking them to the ports of the encoders they are connected to. The 'gpu' property is removed in a hack-ish way. The driver contains a list of all the compatible strings for gpus, and searches the entire OF firmware for a matching node. Once we know what's the right way to link the gpu and display nodes together (if needed at all), we can add the required binding. The device hierarchy for MDP5 platforms fits well for runtime PM adaptation too. Although, for it to work correctly, all the encoder drivers need to adapt to runtime PM too. With that in place, we still hit issues in some usecases where the entire register context isn't correctly restored during resume. It finally boils down to the helpers we use for implementing atomic_commit. This will take some more time to solve. For now, we just enable runtime PM early and leave it enabled. This is necessary for MDP5 based SoCs since Qcom GDSCs are tied to power domains. This will be fixed once we get all paths working properly with runtime PM. This patchset will break bisectability, in the sense that both the downstream and proposed DT bindings won't work if we apply only a partial set of patches. With this series applied, only the proposed bindings will work. Downstream dtsi files from older kernels will have to be adapted slightly to get it running with these changes. This series depends on two patchsets posted before: drm/msm DT prep work: http://www.spinics.net/lists/dri-devel/msg110197.html ADV7533 support+DT bindings: http://www.spinics.net/lists/linux-arm-msm/msg21085.html Archit Taneja (22): drm/msm: Drop the id_table in platform_driver drm/msm: Remove unused fields drm/msm: Get irq number within kms driver itself drm/msm/mdp5: Add MDSS top level driver drm/msm/mdp5: Create a separate MDP5 device drm/msm/mdp5: Prepare new kms_init funcs drm/msm/mdp5: Use the new hierarchy and drop old irq management drm/msm/mdp5: Remove old kms init/destroy funcs drm/msm/mdp5: Use updated MDP5 register names drm/msm/mdp5: Update the register offsets of MDP5 sub-blocks drm/msm: Call pm_runtime_enable/disable for newly created devices drm/msm/mdp5: Add missing mdp5_enable/disable calls drm/msm: Create separate funcs for adding display/gpu components drm/msm: Add display components by parsing MDP ports drm/msm: Add components for MDP5 drm/msm: Drop the gpu binding drm/msm/mdp5: Update compatible strings for MDSS/MDP5 dt-bindings: msm/mdp4: Create a separate binding doc for MDP4 dt-bindings: msm/mdp5: Add MDP5 display bindings dt-bindings: msm/mdp: Provide details on MDP interface ports arm64: dts: msm8916: Add display support arm64: dts: apq8016-sbc: Add HDMI display support .../devicetree/bindings/display/msm/mdp.txt | 57 ---- .../devicetree/bindings/display/msm/mdp4.txt | 112 ++++++++ .../devicetree/bindings/display/msm/mdp5.txt | 160 +++++++++++ arch/arm64/boot/dts/qcom/apq8016-sbc-soc-pins.dtsi | 48 ++++ arch/arm64/boot/dts/qcom/apq8016-sbc.dtsi | 82 ++++++ arch/arm64/boot/dts/qcom/msm8916.dtsi | 120 ++++++++ drivers/gpu/drm/msm/Makefile | 1 + drivers/gpu/drm/msm/mdp/mdp4/mdp4_kms.c | 19 +- drivers/gpu/drm/msm/mdp/mdp4/mdp4_kms.h | 2 + drivers/gpu/drm/msm/mdp/mdp5/mdp5.xml.h | 203 ++++++------- drivers/gpu/drm/msm/mdp/mdp5/mdp5_cfg.c | 113 ++++---- drivers/gpu/drm/msm/mdp/mdp5/mdp5_cmd_encoder.c | 14 +- drivers/gpu/drm/msm/mdp/mdp5/mdp5_ctl.c | 26 +- drivers/gpu/drm/msm/mdp/mdp5/mdp5_encoder.c | 10 +- drivers/gpu/drm/msm/mdp/mdp5/mdp5_irq.c | 125 +------- drivers/gpu/drm/msm/mdp/mdp5/mdp5_kms.c | 320 +++++++++++++-------- drivers/gpu/drm/msm/mdp/mdp5/mdp5_kms.h | 15 +- drivers/gpu/drm/msm/mdp/mdp5/mdp5_mdss.c | 235 +++++++++++++++ drivers/gpu/drm/msm/mdp/mdp5/mdp5_smp.c | 22 +- drivers/gpu/drm/msm/msm_drv.c | 201 ++++++++++--- drivers/gpu/drm/msm/msm_drv.h | 7 + drivers/gpu/drm/msm/msm_kms.h | 8 +- 22 files changed, 1361 insertions(+), 539 deletions(-) delete mode 100644 Documentation/devicetree/bindings/display/msm/mdp.txt create mode 100644 Documentation/devicetree/bindings/display/msm/mdp4.txt create mode 100644 Documentation/devicetree/bindings/display/msm/mdp5.txt create mode 100644 drivers/gpu/drm/msm/mdp/mdp5/mdp5_mdss.c -- The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, hosted by The Linux Foundation _______________________________________________ dri-devel mailing list dri-devel@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/dri-devel
