drm/msm: Add SDM845 DPU support

SDM845 SoC includes the Mobile Display Sub System (MDSS) which is a
top level wrapper consisting of Display Processing Unit (DPU) and
display peripheral modules such as Display Serial Interface (DSI)
and DisplayPort (DP).

MDSS functions essentially as a back-end composition engine. It blends
video and graphic images stored in the frame buffers and scans out the
composed image to a display sink (over DSI/DP).

The following diagram represents hardware blocks for a simple pipeline
(two planes are present on a given crtc which is connected to a DSI
connector):

       MDSS
      +---------------------------------+
      | +-----------------------------+ |
      | | DPU                         | |
      | |  +--------+  +--------+     | |
      | |  |  SSPP  |  |  SSPP  |     | |
      | |  +----+---+  +----+---+     | |
      | |       |           |         | |
      | |  +----v-----------v---+     | |
      | |  |  Layer Mixer (LM)  |     | |
      | |  +--------------------+     | |
      | |  +--------------------+     | |
      | |  |    PingPong (PP)   |     | |
      | |  +--------------------+     | |
      | |  +--------------------+     | |
      | |  |  INTERFACE (VIDEO) |     | |
      | |  +---+----------------+     | |
      | +------|----------------------+ |
      |        |                        |
      | +------|---------------------+  |
      | |      | DISPLAY PERIPHERALS |  |
      | |  +---v-+      +-----+      |  |
      | |  | DSI |      |  DP |      |  |
      | |  +-----+      +-----+      |  |
      | +----------------------------+  |
      +---------------------------------+

The number of DPU sub-blocks (i.e. SSPPs, LMs, PP blocks and INTFs)
depends on SoC capabilities.

Overview of DPU sub-blocks:
---------------------------
* Source Surface Processor (SSPP):
 Refers to any of hardware pipes like ViG, DMA etc. Only ViG pipes are
 capable of performing format conversion, scaling and quality improvement
 for source surfaces.

* Layer Mixer (LM):
 Blend source surfaces together (in requested zorder)

* PingPong (PP):
 This block controls frame done interrupt output, EOL and EOF generation,
 overflow/underflow control.

* Display interface (INTF):
 Timing generator and interface connecting the display peripherals.

DRM components mapping to DPU architecture:
------------------------------------------
PLANEs maps to SSPPs
CRTC maps to LMs
Encoder maps to PPs, INTFs

Data flow setup:
---------------
MDSS hardware can support various data flows (e.g.):
  - Dual pipe: Output from two LMs combined to single display.
  - Split display: Output from two LMs connected to two separate
                   interfaces.

The hardware capabilities determine the number of concurrent data paths
possible. Any control path (i.e. pipeline w/i DPU) can be routed to any
of the hardware data paths. A given control path can be triggered,
flushed and controlled independently.

Changes in v3:
- Move msm_media_info.h from uapi to dpu/ subdir
- Remove preclose callback dpu (it's handled in core)
- Fix kbuild warnings with parent_ops
- Remove unused functions from dpu_core_irq
- Rename mdss_phys to mdss
- Rename mdp_phys address space to mdp
- Drop _phys from vbif and regdma binding names

Signed-off-by: Abhinav Kumar <abhinavk@codeaurora.org>
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Chandan Uddaraju <chandanu@codeaurora.org>
Signed-off-by: Jeykumar Sankaran <jsanka@codeaurora.org>
Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Signed-off-by: Rajesh Yadav <ryadav@codeaurora.org>
Signed-off-by: Sravanthi Kollukuduru <skolluku@codeaurora.org>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
[robclark minor rebase]
Signed-off-by: Rob Clark <robdclark@gmail.com>

1 files changed, 124 insertions, 11 deletions

diff --git a/drivers/gpu/drm/msm/msm_drv.c b/drivers/gpu/drm/msm/msm_drv.c
index e79ad74ca98c..46876bc8b707 100644
--- a/drivers/gpu/drm/msm/msm_drv.c
+++ b/drivers/gpu/drm/msm/msm_drv.c
@@ -1,4 +1,5 @@
 /*
+ * Copyright (c) 2016-2018, The Linux Foundation. All rights reserved.
  * Copyright (C) 2013 Red Hat
  * Author: Rob Clark <robdclark@gmail.com>
  *
@@ -15,6 +16,8 @@
  * this program.  If not, see <http://www.gnu.org/licenses/>.
  */
 
+#include <linux/kthread.h>
+#include <uapi/linux/sched/types.h>
 #include <drm/drm_of.h>
 
 #include "msm_drv.h"
@@ -149,7 +152,7 @@ struct vblank_event {
 	bool enable;
 };
 
-static void vblank_ctrl_worker(struct work_struct *work)
+static void vblank_ctrl_worker(struct kthread_work *work)
 {
 	struct msm_vblank_ctrl *vbl_ctrl = container_of(work,
 						struct msm_vblank_ctrl, work);
@@ -197,7 +200,8 @@ static int vblank_ctrl_queue_work(struct msm_drm_private *priv,
 	list_add_tail(&vbl_ev->node, &vbl_ctrl->event_list);
 	spin_unlock_irqrestore(&vbl_ctrl->lock, flags);
 
-	queue_work(priv->wq, &vbl_ctrl->work);
+	kthread_queue_work(&priv->disp_thread[crtc_id].worker,
+			&vbl_ctrl->work);
 
 	return 0;
 }
@@ -211,17 +215,33 @@ static int msm_drm_uninit(struct device *dev)
 	struct msm_mdss *mdss = priv->mdss;
 	struct msm_vblank_ctrl *vbl_ctrl = &priv->vblank_ctrl;
 	struct vblank_event *vbl_ev, *tmp;
+	int i;
 
 	/* We must cancel and cleanup any pending vblank enable/disable
 	 * work before drm_irq_uninstall() to avoid work re-enabling an
 	 * irq after uninstall has disabled it.
 	 */
-	cancel_work_sync(&vbl_ctrl->work);
+	kthread_flush_work(&vbl_ctrl->work);
 	list_for_each_entry_safe(vbl_ev, tmp, &vbl_ctrl->event_list, node) {
 		list_del(&vbl_ev->node);
 		kfree(vbl_ev);
 	}
 
+	/* clean up display commit/event worker threads */
+	for (i = 0; i < priv->num_crtcs; i++) {
+		if (priv->disp_thread[i].thread) {
+			kthread_flush_worker(&priv->disp_thread[i].worker);
+			kthread_stop(priv->disp_thread[i].thread);
+			priv->disp_thread[i].thread = NULL;
+		}
+
+		if (priv->event_thread[i].thread) {
+			kthread_flush_worker(&priv->event_thread[i].worker);
+			kthread_stop(priv->event_thread[i].thread);
+			priv->event_thread[i].thread = NULL;
+		}
+	}
+
 	msm_gem_shrinker_cleanup(ddev);
 
 	drm_kms_helper_poll_fini(ddev);
@@ -269,6 +289,7 @@ static int msm_drm_uninit(struct device *dev)
 
 #define KMS_MDP4 4
 #define KMS_MDP5 5
+#define KMS_DPU  3
 
 static int get_mdp_ver(struct platform_device *pdev)
 {
@@ -360,7 +381,8 @@ static int msm_drm_init(struct device *dev, struct drm_driver *drv)
 	struct msm_drm_private *priv;
 	struct msm_kms *kms;
 	struct msm_mdss *mdss;
-	int ret;
+	int ret, i;
+	struct sched_param param;
 
 	ddev = drm_dev_alloc(drv, dev);
 	if (IS_ERR(ddev)) {
@@ -379,7 +401,17 @@ static int msm_drm_init(struct device *dev, struct drm_driver *drv)
 	ddev->dev_private = priv;
 	priv->dev = ddev;
 
-	ret = mdp5_mdss_init(ddev);
+	switch (get_mdp_ver(pdev)) {
+	case KMS_MDP5:
+		ret = mdp5_mdss_init(ddev);
+		break;
+	case KMS_DPU:
+		ret = dpu_mdss_init(ddev);
+		break;
+	default:
+		ret = 0;
+		break;
+	}
 	if (ret)
 		goto err_free_priv;
 
@@ -389,7 +421,7 @@ static int msm_drm_init(struct device *dev, struct drm_driver *drv)
 
 	INIT_LIST_HEAD(&priv->inactive_list);
 	INIT_LIST_HEAD(&priv->vblank_ctrl.event_list);
-	INIT_WORK(&priv->vblank_ctrl.work, vblank_ctrl_worker);
+	kthread_init_work(&priv->vblank_ctrl.work, vblank_ctrl_worker);
 	spin_lock_init(&priv->vblank_ctrl.lock);
 
 	drm_mode_config_init(ddev);
@@ -413,6 +445,10 @@ static int msm_drm_init(struct device *dev, struct drm_driver *drv)
 	case KMS_MDP5:
 		kms = mdp5_kms_init(ddev);
 		break;
+	case KMS_DPU:
+		kms = dpu_kms_init(ddev);
+		priv->kms = kms;
+		break;
 	default:
 		kms = ERR_PTR(-ENODEV);
 		break;
@@ -444,6 +480,79 @@ static int msm_drm_init(struct device *dev, struct drm_driver *drv)
 	ddev->mode_config.funcs = &mode_config_funcs;
 	ddev->mode_config.helper_private = &mode_config_helper_funcs;
 
+	/**
+	 * this priority was found during empiric testing to have appropriate
+	 * realtime scheduling to process display updates and interact with
+	 * other real time and normal priority task
+	 */
+	param.sched_priority = 16;
+	for (i = 0; i < priv->num_crtcs; i++) {
+
+		/* initialize display thread */
+		priv->disp_thread[i].crtc_id = priv->crtcs[i]->base.id;
+		kthread_init_worker(&priv->disp_thread[i].worker);
+		priv->disp_thread[i].dev = ddev;
+		priv->disp_thread[i].thread =
+			kthread_run(kthread_worker_fn,
+				&priv->disp_thread[i].worker,
+				"crtc_commit:%d", priv->disp_thread[i].crtc_id);
+		ret = sched_setscheduler(priv->disp_thread[i].thread,
+							SCHED_FIFO, &param);
+		if (ret)
+			pr_warn("display thread priority update failed: %d\n",
+									ret);
+
+		if (IS_ERR(priv->disp_thread[i].thread)) {
+			dev_err(dev, "failed to create crtc_commit kthread\n");
+			priv->disp_thread[i].thread = NULL;
+		}
+
+		/* initialize event thread */
+		priv->event_thread[i].crtc_id = priv->crtcs[i]->base.id;
+		kthread_init_worker(&priv->event_thread[i].worker);
+		priv->event_thread[i].dev = ddev;
+		priv->event_thread[i].thread =
+			kthread_run(kthread_worker_fn,
+				&priv->event_thread[i].worker,
+				"crtc_event:%d", priv->event_thread[i].crtc_id);
+		/**
+		 * event thread should also run at same priority as disp_thread
+		 * because it is handling frame_done events. A lower priority
+		 * event thread and higher priority disp_thread can causes
+		 * frame_pending counters beyond 2. This can lead to commit
+		 * failure at crtc commit level.
+		 */
+		ret = sched_setscheduler(priv->event_thread[i].thread,
+							SCHED_FIFO, &param);
+		if (ret)
+			pr_warn("display event thread priority update failed: %d\n",
+									ret);
+
+		if (IS_ERR(priv->event_thread[i].thread)) {
+			dev_err(dev, "failed to create crtc_event kthread\n");
+			priv->event_thread[i].thread = NULL;
+		}
+
+		if ((!priv->disp_thread[i].thread) ||
+				!priv->event_thread[i].thread) {
+			/* clean up previously created threads if any */
+			for ( ; i >= 0; i--) {
+				if (priv->disp_thread[i].thread) {
+					kthread_stop(
+						priv->disp_thread[i].thread);
+					priv->disp_thread[i].thread = NULL;
+				}
+
+				if (priv->event_thread[i].thread) {
+					kthread_stop(
+						priv->event_thread[i].thread);
+					priv->event_thread[i].thread = NULL;
+				}
+			}
+			goto err_msm_uninit;
+		}
+	}
+
 	ret = drm_vblank_init(ddev, priv->num_crtcs);
 	if (ret < 0) {
 		dev_err(dev, "failed to initialize vblank\n");
@@ -1060,12 +1169,13 @@ static int add_display_components(struct device *dev,
 	int ret;
 
 	/*
-	 * MDP5 based devices don't have a flat hierarchy. There is a top level
-	 * parent: MDSS, and children: MDP5, DSI, HDMI, eDP etc. Populate the
-	 * children devices, find the MDP5 node, and then add the interfaces
-	 * to our components list.
+	 * MDP5/DPU based devices don't have a flat hierarchy. There is a top
+	 * level parent: MDSS, and children: MDP5/DPU, DSI, HDMI, eDP etc.
+	 * Populate the children devices, find the MDP5/DPU node, and then add
+	 * the interfaces to our components list.
 	 */
-	if (of_device_is_compatible(dev->of_node, "qcom,mdss")) {
+	if (of_device_is_compatible(dev->of_node, "qcom,mdss") ||
+	    of_device_is_compatible(dev->of_node, "qcom,sdm845-mdss")) {
 		ret = of_platform_populate(dev->of_node, NULL, NULL, dev);
 		if (ret) {
 			dev_err(dev, "failed to populate children devices\n");
@@ -1177,6 +1287,7 @@ static int msm_pdev_remove(struct platform_device *pdev)
 static const struct of_device_id dt_match[] = {
 	{ .compatible = "qcom,mdp4", .data = (void *)KMS_MDP4 },
 	{ .compatible = "qcom,mdss", .data = (void *)KMS_MDP5 },
+	{ .compatible = "qcom,sdm845-mdss", .data = (void *)KMS_DPU },
 	{}
 };
 MODULE_DEVICE_TABLE(of, dt_match);
@@ -1198,6 +1309,7 @@ static int __init msm_drm_register(void)
 
 	DBG("init");
 	msm_mdp_register();
+	msm_dpu_register();
 	msm_dsi_register();
 	msm_edp_register();
 	msm_hdmi_register();
@@ -1214,6 +1326,7 @@ static void __exit msm_drm_unregister(void)
 	msm_edp_unregister();
 	msm_dsi_unregister();
 	msm_mdp_unregister();
+	msm_dpu_unregister();
 }
 
 module_init(msm_drm_register);