Merge tag 'drm-msm-next-2017-08-22' of git://people.freedesktop.org/~robclark/linux into drm-next

Updates for 4.14..  I have some further patches from Jordan to add
multiple priority levels and pre-emption, but those will probably be
for 4.15 to give me time for the mesa parts.

* tag 'drm-msm-next-2017-08-22' of git://people.freedesktop.org/~robclark/linux:
  drm/msm/mdp5: mark runtime_pm functions as __maybe_unused
  drm/msm: remove unused variable
  drm/msm/mdp5: make helper function static
  drm/msm: make msm_framebuffer_init() static
  drm/msm: add helper to allocate stolen fb
  drm/msm: don't track fbdev's gem object separately
  drm/msm: add modeset module param
  drm/msm/mdp5: add tracking for clk enable-count
  drm/msm: remove unused define
  drm/msm: Add a helper function for in-kernel buffer allocations
  drm/msm: Attach the GPU MMU when it is created
  drm/msm: Add A5XX hardware fault detection
  drm/msm: Remove uneeded platform dev members
  drm/msm/mdp5: Set up runtime PM for MDSS
  drm/msm/mdp5: Write to SMP registers even if allocations don't change
  drm/msm/mdp5: Don't use mode_set helper funcs for encoders and CRTCs
  drm/msm/dsi: Implement RPM suspend/resume callbacks
  drm/msm/dsi: Set up runtime PM for DSI
  drm/msm/hdmi: Set up runtime PM for HDMI
  drm/msm/mdp5: Use runtime PM get/put API instead of toggling clocks

31 files changed, 574 insertions, 273 deletions

diff --git a/drivers/gpu/drm/msm/adreno/a3xx_gpu.c b/drivers/gpu/drm/msm/adreno/a3xx_gpu.c
index 0e3828ed1e46..7791313405b5 100644
--- a/drivers/gpu/drm/msm/adreno/a3xx_gpu.c
+++ b/drivers/gpu/drm/msm/adreno/a3xx_gpu.c
@@ -486,8 +486,6 @@ struct msm_gpu *a3xx_gpu_init(struct drm_device *dev)
 	adreno_gpu = &a3xx_gpu->base;
 	gpu = &adreno_gpu->base;
 
-	a3xx_gpu->pdev = pdev;
-
 	gpu->perfcntrs = perfcntrs;
 	gpu->num_perfcntrs = ARRAY_SIZE(perfcntrs);
 
diff --git a/drivers/gpu/drm/msm/adreno/a3xx_gpu.h b/drivers/gpu/drm/msm/adreno/a3xx_gpu.h
index 85ff66cbddd6..ab60dc9e344e 100644
--- a/drivers/gpu/drm/msm/adreno/a3xx_gpu.h
+++ b/drivers/gpu/drm/msm/adreno/a3xx_gpu.h
@@ -28,7 +28,6 @@
 
 struct a3xx_gpu {
 	struct adreno_gpu base;
-	struct platform_device *pdev;
 
 	/* if OCMEM is used for GMEM: */
 	uint32_t ocmem_base;
diff --git a/drivers/gpu/drm/msm/adreno/a4xx_gpu.c b/drivers/gpu/drm/msm/adreno/a4xx_gpu.c
index 19abf229b08d..58341ef6f15b 100644
--- a/drivers/gpu/drm/msm/adreno/a4xx_gpu.c
+++ b/drivers/gpu/drm/msm/adreno/a4xx_gpu.c
@@ -568,8 +568,6 @@ struct msm_gpu *a4xx_gpu_init(struct drm_device *dev)
 	adreno_gpu = &a4xx_gpu->base;
 	gpu = &adreno_gpu->base;
 
-	a4xx_gpu->pdev = pdev;
-
 	gpu->perfcntrs = NULL;
 	gpu->num_perfcntrs = 0;
 
diff --git a/drivers/gpu/drm/msm/adreno/a4xx_gpu.h b/drivers/gpu/drm/msm/adreno/a4xx_gpu.h
index 01247204ac92..f757184328a3 100644
--- a/drivers/gpu/drm/msm/adreno/a4xx_gpu.h
+++ b/drivers/gpu/drm/msm/adreno/a4xx_gpu.h
@@ -23,7 +23,6 @@
 
 struct a4xx_gpu {
 	struct adreno_gpu base;
-	struct platform_device *pdev;
 
 	/* if OCMEM is used for GMEM: */
 	uint32_t ocmem_base;
diff --git a/drivers/gpu/drm/msm/adreno/a5xx_gpu.c b/drivers/gpu/drm/msm/adreno/a5xx_gpu.c
index f9eae03aa1dc..17c59d839e6f 100644
--- a/drivers/gpu/drm/msm/adreno/a5xx_gpu.c
+++ b/drivers/gpu/drm/msm/adreno/a5xx_gpu.c
@@ -284,28 +284,14 @@ static int a5xx_me_init(struct msm_gpu *gpu)
 static struct drm_gem_object *a5xx_ucode_load_bo(struct msm_gpu *gpu,
 		const struct firmware *fw, u64 *iova)
 {
-	struct drm_device *drm = gpu->dev;
 	struct drm_gem_object *bo;
 	void *ptr;
 
-	bo = msm_gem_new_locked(drm, fw->size - 4, MSM_BO_UNCACHED);
-	if (IS_ERR(bo))
-		return bo;
+	ptr = msm_gem_kernel_new_locked(gpu->dev, fw->size - 4,
+		MSM_BO_UNCACHED | MSM_BO_GPU_READONLY, gpu->aspace, &bo, iova);
 
-	ptr = msm_gem_get_vaddr(bo);
-	if (!ptr) {
-		drm_gem_object_unreference(bo);
-		return ERR_PTR(-ENOMEM);
-	}
-
-	if (iova) {
-		int ret = msm_gem_get_iova(bo, gpu->aspace, iova);
-
-		if (ret) {
-			drm_gem_object_unreference(bo);
-			return ERR_PTR(ret);
-		}
-	}
+	if (IS_ERR(ptr))
+		return ERR_CAST(ptr);
 
 	memcpy(ptr, &fw->data[4], fw->size - 4);
 
@@ -372,8 +358,7 @@ static int a5xx_zap_shader_init(struct msm_gpu *gpu)
 {
 	static bool loaded;
 	struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu);
-	struct a5xx_gpu *a5xx_gpu = to_a5xx_gpu(adreno_gpu);
-	struct platform_device *pdev = a5xx_gpu->pdev;
+	struct platform_device *pdev = gpu->pdev;
 	int ret;
 
 	/*
@@ -410,6 +395,7 @@ static int a5xx_zap_shader_init(struct msm_gpu *gpu)
 	  A5XX_RBBM_INT_0_MASK_RBBM_ETS_MS_TIMEOUT | \
 	  A5XX_RBBM_INT_0_MASK_RBBM_ATB_ASYNC_OVERFLOW | \
 	  A5XX_RBBM_INT_0_MASK_CP_HW_ERROR | \
+	  A5XX_RBBM_INT_0_MASK_MISC_HANG_DETECT | \
 	  A5XX_RBBM_INT_0_MASK_CP_CACHE_FLUSH_TS | \
 	  A5XX_RBBM_INT_0_MASK_UCHE_OOB_ACCESS | \
 	  A5XX_RBBM_INT_0_MASK_GPMU_VOLTAGE_DROOP)
@@ -812,6 +798,27 @@ static void a5xx_gpmu_err_irq(struct msm_gpu *gpu)
 	dev_err_ratelimited(gpu->dev->dev, "GPMU | voltage droop\n");
 }
 
+static void a5xx_fault_detect_irq(struct msm_gpu *gpu)
+{
+	struct drm_device *dev = gpu->dev;
+	struct msm_drm_private *priv = dev->dev_private;
+
+	dev_err(dev->dev, "gpu fault fence %x status %8.8X rb %4.4x/%4.4x ib1 %16.16llX/%4.4x ib2 %16.16llX/%4.4x\n",
+		gpu->funcs->last_fence(gpu),
+		gpu_read(gpu, REG_A5XX_RBBM_STATUS),
+		gpu_read(gpu, REG_A5XX_CP_RB_RPTR),
+		gpu_read(gpu, REG_A5XX_CP_RB_WPTR),
+		gpu_read64(gpu, REG_A5XX_CP_IB1_BASE, REG_A5XX_CP_IB1_BASE_HI),
+		gpu_read(gpu, REG_A5XX_CP_IB1_BUFSZ),
+		gpu_read64(gpu, REG_A5XX_CP_IB2_BASE, REG_A5XX_CP_IB2_BASE_HI),
+		gpu_read(gpu, REG_A5XX_CP_IB2_BUFSZ));
+
+	/* Turn off the hangcheck timer to keep it from bothering us */
+	del_timer(&gpu->hangcheck_timer);
+
+	queue_work(priv->wq, &gpu->recover_work);
+}
+
 #define RBBM_ERROR_MASK \
 	(A5XX_RBBM_INT_0_MASK_RBBM_AHB_ERROR | \
 	A5XX_RBBM_INT_0_MASK_RBBM_TRANSFER_TIMEOUT | \
@@ -838,6 +845,9 @@ static irqreturn_t a5xx_irq(struct msm_gpu *gpu)
 	if (status & A5XX_RBBM_INT_0_MASK_CP_HW_ERROR)
 		a5xx_cp_err_irq(gpu);
 
+	if (status & A5XX_RBBM_INT_0_MASK_MISC_HANG_DETECT)
+		a5xx_fault_detect_irq(gpu);
+
 	if (status & A5XX_RBBM_INT_0_MASK_UCHE_OOB_ACCESS)
 		a5xx_uche_err_irq(gpu);
 
@@ -1015,7 +1025,6 @@ struct msm_gpu *a5xx_gpu_init(struct drm_device *dev)
 	adreno_gpu = &a5xx_gpu->base;
 	gpu = &adreno_gpu->base;
 
-	a5xx_gpu->pdev = pdev;
 	adreno_gpu->registers = a5xx_registers;
 	adreno_gpu->reg_offsets = a5xx_register_offsets;
 
diff --git a/drivers/gpu/drm/msm/adreno/a5xx_gpu.h b/drivers/gpu/drm/msm/adreno/a5xx_gpu.h
index 1137092241d5..e94451685bf8 100644
--- a/drivers/gpu/drm/msm/adreno/a5xx_gpu.h
+++ b/drivers/gpu/drm/msm/adreno/a5xx_gpu.h
@@ -23,7 +23,6 @@
 
 struct a5xx_gpu {
 	struct adreno_gpu base;
-	struct platform_device *pdev;
 
 	struct drm_gem_object *pm4_bo;
 	uint64_t pm4_iova;
diff --git a/drivers/gpu/drm/msm/adreno/a5xx_power.c b/drivers/gpu/drm/msm/adreno/a5xx_power.c
index 87af6eea0483..04aab1dcae2b 100644
--- a/drivers/gpu/drm/msm/adreno/a5xx_power.c
+++ b/drivers/gpu/drm/msm/adreno/a5xx_power.c
@@ -294,16 +294,10 @@ void a5xx_gpmu_ucode_init(struct msm_gpu *gpu)
 	 */
 	bosize = (cmds_size + (cmds_size / TYPE4_MAX_PAYLOAD) + 1) << 2;
 
-	a5xx_gpu->gpmu_bo = msm_gem_new_locked(drm, bosize, MSM_BO_UNCACHED);
-	if (IS_ERR(a5xx_gpu->gpmu_bo))
-		goto err;
-
-	if (msm_gem_get_iova(a5xx_gpu->gpmu_bo, gpu->aspace,
-			&a5xx_gpu->gpmu_iova))
-		goto err;
-
-	ptr = msm_gem_get_vaddr(a5xx_gpu->gpmu_bo);
-	if (!ptr)
+	ptr = msm_gem_kernel_new_locked(drm, bosize,
+		MSM_BO_UNCACHED | MSM_BO_GPU_READONLY, gpu->aspace,
+		&a5xx_gpu->gpmu_bo, &a5xx_gpu->gpmu_iova);
+	if (IS_ERR(ptr))
 		goto err;
 
 	while (cmds_size > 0) {
diff --git a/drivers/gpu/drm/msm/adreno/adreno_gpu.c b/drivers/gpu/drm/msm/adreno/adreno_gpu.c
index 7414c6bbd582..c8b4ac254bb5 100644
--- a/drivers/gpu/drm/msm/adreno/adreno_gpu.c
+++ b/drivers/gpu/drm/msm/adreno/adreno_gpu.c
@@ -337,11 +337,6 @@ void adreno_wait_ring(struct msm_gpu *gpu, uint32_t ndwords)
 		DRM_ERROR("%s: timeout waiting for ringbuffer space\n", gpu->name);
 }
 
-static const char *iommu_ports[] = {
-		"gfx3d_user", "gfx3d_priv",
-		"gfx3d1_user", "gfx3d1_priv",
-};
-
 int adreno_gpu_init(struct drm_device *drm, struct platform_device *pdev,
 		struct adreno_gpu *adreno_gpu, const struct adreno_gpu_funcs *funcs)
 {
@@ -373,15 +368,15 @@ int adreno_gpu_init(struct drm_device *drm, struct platform_device *pdev,
 
 	adreno_gpu_config.ringsz = RB_SIZE;
 
+	pm_runtime_set_autosuspend_delay(&pdev->dev, DRM_MSM_INACTIVE_PERIOD);
+	pm_runtime_use_autosuspend(&pdev->dev);
+	pm_runtime_enable(&pdev->dev);
+
 	ret = msm_gpu_init(drm, pdev, &adreno_gpu->base, &funcs->base,
 			adreno_gpu->info->name, &adreno_gpu_config);
 	if (ret)
 		return ret;
 
-	pm_runtime_set_autosuspend_delay(&pdev->dev, DRM_MSM_INACTIVE_PERIOD);
-	pm_runtime_use_autosuspend(&pdev->dev);
-	pm_runtime_enable(&pdev->dev);
-
 	ret = request_firmware(&adreno_gpu->pm4, adreno_gpu->info->pm4fw, drm->dev);
 	if (ret) {
 		dev_err(drm->dev, "failed to load %s PM4 firmware: %d\n",
@@ -396,37 +391,17 @@ int adreno_gpu_init(struct drm_device *drm, struct platform_device *pdev,
 		return ret;
 	}
 
-	if (gpu->aspace && gpu->aspace->mmu) {
-		struct msm_mmu *mmu = gpu->aspace->mmu;
-		ret = mmu->funcs->attach(mmu, iommu_ports,
-				ARRAY_SIZE(iommu_ports));
-		if (ret)
-			return ret;
-	}
+	adreno_gpu->memptrs = msm_gem_kernel_new(drm,
+		sizeof(*adreno_gpu->memptrs), MSM_BO_UNCACHED, gpu->aspace,
+		&adreno_gpu->memptrs_bo, &adreno_gpu->memptrs_iova);
 
-	adreno_gpu->memptrs_bo = msm_gem_new(drm, sizeof(*adreno_gpu->memptrs),
-			MSM_BO_UNCACHED);
-	if (IS_ERR(adreno_gpu->memptrs_bo)) {
-		ret = PTR_ERR(adreno_gpu->memptrs_bo);
-		adreno_gpu->memptrs_bo = NULL;
-		dev_err(drm->dev, "could not allocate memptrs: %d\n", ret);
-		return ret;
-	}
-
-	adreno_gpu->memptrs = msm_gem_get_vaddr(adreno_gpu->memptrs_bo);
 	if (IS_ERR(adreno_gpu->memptrs)) {
-		dev_err(drm->dev, "could not vmap memptrs\n");
-		return -ENOMEM;
-	}
-
-	ret = msm_gem_get_iova(adreno_gpu->memptrs_bo, gpu->aspace,
-			&adreno_gpu->memptrs_iova);
-	if (ret) {
-		dev_err(drm->dev, "could not map memptrs: %d\n", ret);
-		return ret;
+		ret = PTR_ERR(adreno_gpu->memptrs);
+		adreno_gpu->memptrs = NULL;
+		dev_err(drm->dev, "could not allocate memptrs: %d\n", ret);
 	}
 
-	return 0;
+	return ret;
 }
 
 void adreno_gpu_cleanup(struct adreno_gpu *adreno_gpu)
@@ -446,10 +421,4 @@ void adreno_gpu_cleanup(struct adreno_gpu *adreno_gpu)
 	release_firmware(adreno_gpu->pfp);
 
 	msm_gpu_cleanup(gpu);
-
-	if (gpu->aspace) {
-		gpu->aspace->mmu->funcs->detach(gpu->aspace->mmu,
-			iommu_ports, ARRAY_SIZE(iommu_ports));
-		msm_gem_address_space_put(gpu->aspace);
-	}
 }
diff --git a/drivers/gpu/drm/msm/dsi/dsi.c b/drivers/gpu/drm/msm/dsi/dsi.c
index 311c1c1e7d6c..98742d7af6dc 100644
--- a/drivers/gpu/drm/msm/dsi/dsi.c
+++ b/drivers/gpu/drm/msm/dsi/dsi.c
@@ -161,12 +161,17 @@ static const struct of_device_id dt_match[] = {
 	{}
 };
 
+static const struct dev_pm_ops dsi_pm_ops = {
+	SET_RUNTIME_PM_OPS(msm_dsi_runtime_suspend, msm_dsi_runtime_resume, NULL)
+};
+
 static struct platform_driver dsi_driver = {
 	.probe = dsi_dev_probe,
 	.remove = dsi_dev_remove,
 	.driver = {
 		.name = "msm_dsi",
 		.of_match_table = dt_match,
+		.pm = &dsi_pm_ops,
 	},
 };
 
diff --git a/drivers/gpu/drm/msm/dsi/dsi.h b/drivers/gpu/drm/msm/dsi/dsi.h
index 9e6017387efb..2302046197a8 100644
--- a/drivers/gpu/drm/msm/dsi/dsi.h
+++ b/drivers/gpu/drm/msm/dsi/dsi.h
@@ -179,6 +179,8 @@ void msm_dsi_host_destroy(struct mipi_dsi_host *host);
 int msm_dsi_host_modeset_init(struct mipi_dsi_host *host,
 					struct drm_device *dev);
 int msm_dsi_host_init(struct msm_dsi *msm_dsi);
+int msm_dsi_runtime_suspend(struct device *dev);
+int msm_dsi_runtime_resume(struct device *dev);
 
 /* dsi phy */
 struct msm_dsi_phy;
diff --git a/drivers/gpu/drm/msm/dsi/dsi_host.c b/drivers/gpu/drm/msm/dsi/dsi_host.c
index c7b612c3d771..dbb31a014419 100644
--- a/drivers/gpu/drm/msm/dsi/dsi_host.c
+++ b/drivers/gpu/drm/msm/dsi/dsi_host.c
@@ -135,7 +135,6 @@ struct msm_dsi_host {
 	struct completion video_comp;
 	struct mutex dev_mutex;
 	struct mutex cmd_mutex;
-	struct mutex clk_mutex;
 	spinlock_t intr_lock; /* Protect interrupt ctrl register */
 
 	u32 err_work_state;
@@ -221,6 +220,8 @@ static const struct msm_dsi_cfg_handler *dsi_get_config(
 		goto put_gdsc;
 	}
 
+	pm_runtime_get_sync(dev);
+
 	ret = regulator_enable(gdsc_reg);
 	if (ret) {
 		pr_err("%s: unable to enable gdsc\n", __func__);
@@ -247,6 +248,7 @@ disable_clks:
 	clk_disable_unprepare(ahb_clk);
 disable_gdsc:
 	regulator_disable(gdsc_reg);
+	pm_runtime_put_autosuspend(dev);
 put_clk:
 	clk_put(ahb_clk);
 put_gdsc:
@@ -455,6 +457,34 @@ static void dsi_bus_clk_disable(struct msm_dsi_host *msm_host)
 		clk_disable_unprepare(msm_host->bus_clks[i]);
 }
 
+int msm_dsi_runtime_suspend(struct device *dev)
+{
+	struct platform_device *pdev = to_platform_device(dev);
+	struct msm_dsi *msm_dsi = platform_get_drvdata(pdev);
+	struct mipi_dsi_host *host = msm_dsi->host;
+	struct msm_dsi_host *msm_host = to_msm_dsi_host(host);
+
+	if (!msm_host->cfg_hnd)
+		return 0;
+
+	dsi_bus_clk_disable(msm_host);
+
+	return 0;
+}
+
+int msm_dsi_runtime_resume(struct device *dev)
+{
+	struct platform_device *pdev = to_platform_device(dev);
+	struct msm_dsi *msm_dsi = platform_get_drvdata(pdev);
+	struct mipi_dsi_host *host = msm_dsi->host;
+	struct msm_dsi_host *msm_host = to_msm_dsi_host(host);
+
+	if (!msm_host->cfg_hnd)
+		return 0;
+
+	return dsi_bus_clk_enable(msm_host);
+}
+
 static int dsi_link_clk_enable_6g(struct msm_dsi_host *msm_host)
 {
 	int ret;
@@ -596,35 +626,6 @@ static void dsi_link_clk_disable(struct msm_dsi_host *msm_host)
 	}
 }
 
-static int dsi_clk_ctrl(struct msm_dsi_host *msm_host, bool enable)
-{
-	int ret = 0;
-
-	mutex_lock(&msm_host->clk_mutex);
-	if (enable) {
-		ret = dsi_bus_clk_enable(msm_host);
-		if (ret) {
-			pr_err("%s: Can not enable bus clk, %d\n",
-				__func__, ret);
-			goto unlock_ret;
-		}
-		ret = dsi_link_clk_enable(msm_host);
-		if (ret) {
-			pr_err("%s: Can not enable link clk, %d\n",
-				__func__, ret);
-			dsi_bus_clk_disable(msm_host);
-			goto unlock_ret;
-		}
-	} else {
-		dsi_link_clk_disable(msm_host);
-		dsi_bus_clk_disable(msm_host);
-	}
-
-unlock_ret:
-	mutex_unlock(&msm_host->clk_mutex);
-	return ret;
-}
-
 static int dsi_calc_clk_rate(struct msm_dsi_host *msm_host)
 {
 	struct drm_display_mode *mode = msm_host->mode;
@@ -1699,6 +1700,7 @@ int msm_dsi_host_init(struct msm_dsi *msm_dsi)
 	}
 
 	msm_host->pdev = pdev;
+	msm_dsi->host = &msm_host->base;
 
 	ret = dsi_host_parse_dt(msm_host);
 	if (ret) {
@@ -1713,6 +1715,8 @@ int msm_dsi_host_init(struct msm_dsi *msm_dsi)
 		goto fail;
 	}
 
+	pm_runtime_enable(&pdev->dev);
+
 	msm_host->cfg_hnd = dsi_get_config(msm_host);
 	if (!msm_host->cfg_hnd) {
 		ret = -EINVAL;
@@ -1753,7 +1757,6 @@ int msm_dsi_host_init(struct msm_dsi *msm_dsi)
 	init_completion(&msm_host->video_comp);
 	mutex_init(&msm_host->dev_mutex);
 	mutex_init(&msm_host->cmd_mutex);
-	mutex_init(&msm_host->clk_mutex);
 	spin_lock_init(&msm_host->intr_lock);
 
 	/* setup workqueue */
@@ -1761,7 +1764,6 @@ int msm_dsi_host_init(struct msm_dsi *msm_dsi)
 	INIT_WORK(&msm_host->err_work, dsi_err_worker);
 	INIT_WORK(&msm_host->hpd_work, dsi_hpd_worker);
 
-	msm_dsi->host = &msm_host->base;
 	msm_dsi->id = msm_host->id;
 
 	DBG("Dsi Host %d initialized", msm_host->id);
@@ -1783,9 +1785,10 @@ void msm_dsi_host_destroy(struct mipi_dsi_host *host)
 		msm_host->workqueue = NULL;
 	}
 
-	mutex_destroy(&msm_host->clk_mutex);
 	mutex_destroy(&msm_host->cmd_mutex);
 	mutex_destroy(&msm_host->dev_mutex);
+
+	pm_runtime_disable(&msm_host->pdev->dev);
 }
 
 int msm_dsi_host_modeset_init(struct mipi_dsi_host *host,
@@ -1881,7 +1884,8 @@ int msm_dsi_host_xfer_prepare(struct mipi_dsi_host *host,
 	 * mdss interrupt is generated in mdp core clock domain
 	 * mdp clock need to be enabled to receive dsi interrupt
 	 */
-	dsi_clk_ctrl(msm_host, 1);
+	pm_runtime_get_sync(&msm_host->pdev->dev);
+	dsi_link_clk_enable(msm_host);
 
 	/* TODO: vote for bus bandwidth */
 
@@ -1911,7 +1915,8 @@ void msm_dsi_host_xfer_restore(struct mipi_dsi_host *host,
 
 	/* TODO: unvote for bus bandwidth */
 
-	dsi_clk_ctrl(msm_host, 0);
+	dsi_link_clk_disable(msm_host);
+	pm_runtime_put_autosuspend(&msm_host->pdev->dev);
 }
 
 int msm_dsi_host_cmd_tx(struct mipi_dsi_host *host,
@@ -2160,8 +2165,11 @@ int msm_dsi_host_enable(struct mipi_dsi_host *host)
 	 * and only turned on before MDP START.
 	 * This part of code should be enabled once mdp driver support it.
 	 */
-	/* if (msm_panel->mode == MSM_DSI_CMD_MODE)
-		dsi_clk_ctrl(msm_host, 0); */
+	/* if (msm_panel->mode == MSM_DSI_CMD_MODE) {
+	 *	dsi_link_clk_disable(msm_host);
+	 *	pm_runtime_put_autosuspend(&msm_host->pdev->dev);
+	 * }
+	 */
 
 	return 0;
 }
@@ -2217,9 +2225,11 @@ int msm_dsi_host_power_on(struct mipi_dsi_host *host,
 		goto unlock_ret;
 	}
 
-	ret = dsi_clk_ctrl(msm_host, 1);
+	pm_runtime_get_sync(&msm_host->pdev->dev);
+	ret = dsi_link_clk_enable(msm_host);
 	if (ret) {
-		pr_err("%s: failed to enable clocks. ret=%d\n", __func__, ret);
+		pr_err("%s: failed to enable link clocks. ret=%d\n",
+		       __func__, ret);
 		goto fail_disable_reg;
 	}
 
@@ -2243,7 +2253,8 @@ int msm_dsi_host_power_on(struct mipi_dsi_host *host,
 	return 0;
 
 fail_disable_clk:
-	dsi_clk_ctrl(msm_host, 0);
+	dsi_link_clk_disable(msm_host);
+	pm_runtime_put_autosuspend(&msm_host->pdev->dev);
 fail_disable_reg:
 	dsi_host_regulator_disable(msm_host);
 unlock_ret:
@@ -2268,7 +2279,8 @@ int msm_dsi_host_power_off(struct mipi_dsi_host *host)
 
 	pinctrl_pm_select_sleep_state(&msm_host->pdev->dev);
 
-	dsi_clk_ctrl(msm_host, 0);
+	dsi_link_clk_disable(msm_host);
+	pm_runtime_put_autosuspend(&msm_host->pdev->dev);
 
 	dsi_host_regulator_disable(msm_host);
 
diff --git a/drivers/gpu/drm/msm/dsi/phy/dsi_phy.c b/drivers/gpu/drm/msm/dsi/phy/dsi_phy.c
index 0c2eb9c9a1fc..7c9bf91bc22b 100644
--- a/drivers/gpu/drm/msm/dsi/phy/dsi_phy.c
+++ b/drivers/gpu/drm/msm/dsi/phy/dsi_phy.c
@@ -373,7 +373,7 @@ static int dsi_phy_enable_resource(struct msm_dsi_phy *phy)
 static void dsi_phy_disable_resource(struct msm_dsi_phy *phy)
 {
 	clk_disable_unprepare(phy->ahb_clk);
-	pm_runtime_put_sync(&phy->pdev->dev);
+	pm_runtime_put_autosuspend(&phy->pdev->dev);
 }
 
 static const struct of_device_id dsi_phy_dt_match[] = {
diff --git a/drivers/gpu/drm/msm/hdmi/hdmi.c b/drivers/gpu/drm/msm/hdmi/hdmi.c
index a968cad509c2..17e069a133a4 100644
--- a/drivers/gpu/drm/msm/hdmi/hdmi.c
+++ b/drivers/gpu/drm/msm/hdmi/hdmi.c
@@ -239,6 +239,8 @@ static struct hdmi *msm_hdmi_init(struct platform_device *pdev)
 		hdmi->pwr_clks[i] = clk;
 	}
 
+	pm_runtime_enable(&pdev->dev);
+
 	hdmi->workq = alloc_ordered_workqueue("msm_hdmi", 0);
 
 	hdmi->i2c = msm_hdmi_i2c_init(hdmi);
diff --git a/drivers/gpu/drm/msm/hdmi/hdmi_bridge.c b/drivers/gpu/drm/msm/hdmi/hdmi_bridge.c
index 13ac822dee5d..7e357077ed26 100644
--- a/drivers/gpu/drm/msm/hdmi/hdmi_bridge.c
+++ b/drivers/gpu/drm/msm/hdmi/hdmi_bridge.c
@@ -35,6 +35,8 @@ static void msm_hdmi_power_on(struct drm_bridge *bridge)
 	const struct hdmi_platform_config *config = hdmi->config;
 	int i, ret;
 
+	pm_runtime_get_sync(&hdmi->pdev->dev);
+
 	for (i = 0; i < config->pwr_reg_cnt; i++) {
 		ret = regulator_enable(hdmi->pwr_regs[i]);
 		if (ret) {
@@ -84,6 +86,8 @@ static void power_off(struct drm_bridge *bridge)
 					config->pwr_reg_names[i], ret);
 		}
 	}
+
+	pm_runtime_put_autosuspend(&hdmi->pdev->dev);
 }
 
 #define AVI_IFRAME_LINE_NUMBER 1
diff --git a/drivers/gpu/drm/msm/hdmi/hdmi_connector.c b/drivers/gpu/drm/msm/hdmi/hdmi_connector.c
index 71536d9c7fe8..c0848dfedd50 100644
--- a/drivers/gpu/drm/msm/hdmi/hdmi_connector.c
+++ b/drivers/gpu/drm/msm/hdmi/hdmi_connector.c
@@ -137,6 +137,36 @@ err:
 	return ret;
 }
 
+static void enable_hpd_clocks(struct hdmi *hdmi, bool enable)
+{
+	const struct hdmi_platform_config *config = hdmi->config;
+	struct device *dev = &hdmi->pdev->dev;
+	int i, ret;
+
+	if (enable) {
+		for (i = 0; i < config->hpd_clk_cnt; i++) {
+			if (config->hpd_freq && config->hpd_freq[i]) {
+				ret = clk_set_rate(hdmi->hpd_clks[i],
+						   config->hpd_freq[i]);
+				if (ret)
+					dev_warn(dev,
+						 "failed to set clk %s (%d)\n",
+						 config->hpd_clk_names[i], ret);
+			}
+
+			ret = clk_prepare_enable(hdmi->hpd_clks[i]);
+			if (ret) {
+				dev_err(dev,
+					"failed to enable hpd clk: %s (%d)\n",
+					config->hpd_clk_names[i], ret);
+			}
+		}
+	} else {
+		for (i = config->hpd_clk_cnt - 1; i >= 0; i--)
+			clk_disable_unprepare(hdmi->hpd_clks[i]);
+	}
+}
+
 static int hpd_enable(struct hdmi_connector *hdmi_connector)
 {
 	struct hdmi *hdmi = hdmi_connector->hdmi;
@@ -167,22 +197,8 @@ static int hpd_enable(struct hdmi_connector *hdmi_connector)
 		goto fail;
 	}
 
-	for (i = 0; i < config->hpd_clk_cnt; i++) {
-		if (config->hpd_freq && config->hpd_freq[i]) {
-			ret = clk_set_rate(hdmi->hpd_clks[i],
-					config->hpd_freq[i]);
-			if (ret)
-				dev_warn(dev, "failed to set clk %s (%d)\n",
-						config->hpd_clk_names[i], ret);
-		}
-
-		ret = clk_prepare_enable(hdmi->hpd_clks[i]);
-		if (ret) {
-			dev_err(dev, "failed to enable hpd clk: %s (%d)\n",
-					config->hpd_clk_names[i], ret);
-			goto fail;
-		}
-	}
+	pm_runtime_get_sync(dev);
+	enable_hpd_clocks(hdmi, true);
 
 	msm_hdmi_set_mode(hdmi, false);
 	msm_hdmi_phy_reset(hdmi);
@@ -225,8 +241,8 @@ static void hdp_disable(struct hdmi_connector *hdmi_connector)
 
 	msm_hdmi_set_mode(hdmi, false);
 
-	for (i = 0; i < config->hpd_clk_cnt; i++)
-		clk_disable_unprepare(hdmi->hpd_clks[i]);
+	enable_hpd_clocks(hdmi, false);
+	pm_runtime_put_autosuspend(dev);
 
 	ret = gpio_config(hdmi, false);
 	if (ret)
@@ -285,7 +301,16 @@ void msm_hdmi_connector_irq(struct drm_connector *connector)
 
 static enum drm_connector_status detect_reg(struct hdmi *hdmi)
 {
-	uint32_t hpd_int_status = hdmi_read(hdmi, REG_HDMI_HPD_INT_STATUS);
+	uint32_t hpd_int_status;
+
+	pm_runtime_get_sync(&hdmi->pdev->dev);
+	enable_hpd_clocks(hdmi, true);
+
+	hpd_int_status = hdmi_read(hdmi, REG_HDMI_HPD_INT_STATUS);
+
+	enable_hpd_clocks(hdmi, false);
+	pm_runtime_put_autosuspend(&hdmi->pdev->dev);
+
 	return (hpd_int_status & HDMI_HPD_INT_STATUS_CABLE_DETECTED) ?
 			connector_status_connected : connector_status_disconnected;
 }
diff --git a/drivers/gpu/drm/msm/mdp/mdp5/mdp5_cmd_encoder.c b/drivers/gpu/drm/msm/mdp/mdp5/mdp5_cmd_encoder.c
index aa7402e03f67..60790df91bfa 100644
--- a/drivers/gpu/drm/msm/mdp/mdp5/mdp5_cmd_encoder.c
+++ b/drivers/gpu/drm/msm/mdp/mdp5/mdp5_cmd_encoder.c
@@ -192,6 +192,7 @@ int mdp5_cmd_encoder_set_split_display(struct drm_encoder *encoder,
 {
 	struct mdp5_encoder *mdp5_cmd_enc = to_mdp5_encoder(encoder);
 	struct mdp5_kms *mdp5_kms;
+	struct device *dev;
 	int intf_num;
 	u32 data = 0;
 
@@ -214,14 +215,16 @@ int mdp5_cmd_encoder_set_split_display(struct drm_encoder *encoder,
 	/* Smart Panel, Sync mode */
 	data |= MDP5_SPLIT_DPL_UPPER_SMART_PANEL;
 
+	dev = &mdp5_kms->pdev->dev;
+
 	/* Make sure clocks are on when connectors calling this function. */
-	mdp5_enable(mdp5_kms);
+	pm_runtime_get_sync(dev);
 	mdp5_write(mdp5_kms, REG_MDP5_SPLIT_DPL_UPPER, data);
 
 	mdp5_write(mdp5_kms, REG_MDP5_SPLIT_DPL_LOWER,
 		   MDP5_SPLIT_DPL_LOWER_SMART_PANEL);
 	mdp5_write(mdp5_kms, REG_MDP5_SPLIT_DPL_EN, 1);
-	mdp5_disable(mdp5_kms);
+	pm_runtime_put_autosuspend(dev);
 
 	return 0;
 }
diff --git a/drivers/gpu/drm/msm/mdp/mdp5/mdp5_crtc.c b/drivers/gpu/drm/msm/mdp/mdp5/mdp5_crtc.c
index 3a81e26629c7..6fcb58ab718c 100644
--- a/drivers/gpu/drm/msm/mdp/mdp5/mdp5_crtc.c
+++ b/drivers/gpu/drm/msm/mdp/mdp5/mdp5_crtc.c
@@ -415,6 +415,7 @@ static void mdp5_crtc_atomic_disable(struct drm_crtc *crtc,
 	struct mdp5_crtc *mdp5_crtc = to_mdp5_crtc(crtc);
 	struct mdp5_crtc_state *mdp5_cstate = to_mdp5_crtc_state(crtc->state);
 	struct mdp5_kms *mdp5_kms = get_kms(crtc);
+	struct device *dev = &mdp5_kms->pdev->dev;
 
 	DBG("%s", crtc->name);
 
@@ -425,7 +426,7 @@ static void mdp5_crtc_atomic_disable(struct drm_crtc *crtc,
 		mdp_irq_unregister(&mdp5_kms->base, &mdp5_crtc->pp_done);
 
 	mdp_irq_unregister(&mdp5_kms->base, &mdp5_crtc->err);
-	mdp5_disable(mdp5_kms);
+	pm_runtime_put_autosuspend(dev);
 
 	mdp5_crtc->enabled = false;
 }
@@ -436,13 +437,17 @@ static void mdp5_crtc_atomic_enable(struct drm_crtc *crtc,
 	struct mdp5_crtc *mdp5_crtc = to_mdp5_crtc(crtc);
 	struct mdp5_crtc_state *mdp5_cstate = to_mdp5_crtc_state(crtc->state);
 	struct mdp5_kms *mdp5_kms = get_kms(crtc);
+	struct device *dev = &mdp5_kms->pdev->dev;
 
 	DBG("%s", crtc->name);
 
 	if (WARN_ON(mdp5_crtc->enabled))
 		return;
 
-	mdp5_enable(mdp5_kms);
+	pm_runtime_get_sync(dev);
+
+	mdp5_crtc_mode_set_nofb(crtc);
+
 	mdp_irq_register(&mdp5_kms->base, &mdp5_crtc->err);
 
 	if (mdp5_cstate->cmd_mode)
@@ -533,7 +538,7 @@ static bool is_fullscreen(struct drm_crtc_state *cstate,
 		((pstate->crtc_y + pstate->crtc_h) >= cstate->mode.vdisplay);
 }
 
-enum mdp_mixer_stage_id get_start_stage(struct drm_crtc *crtc,
+static enum mdp_mixer_stage_id get_start_stage(struct drm_crtc *crtc,
 					struct drm_crtc_state *new_crtc_state,
 					struct drm_plane_state *bpstate)
 {
@@ -727,6 +732,7 @@ static int mdp5_crtc_cursor_set(struct drm_crtc *crtc,
 	struct mdp5_pipeline *pipeline = &mdp5_cstate->pipeline;
 	struct drm_device *dev = crtc->dev;
 	struct mdp5_kms *mdp5_kms = get_kms(crtc);
+	struct platform_device *pdev = mdp5_kms->pdev;
 	struct msm_kms *kms = &mdp5_kms->base.base;
 	struct drm_gem_object *cursor_bo, *old_bo = NULL;
 	uint32_t blendcfg, stride;
@@ -755,7 +761,7 @@ static int mdp5_crtc_cursor_set(struct drm_crtc *crtc,
 	if (!handle) {
 		DBG("Cursor off");
 		cursor_enable = false;
-		mdp5_enable(mdp5_kms);
+		pm_runtime_get_sync(&pdev->dev);
 		goto set_cursor;
 	}
 
@@ -770,6 +776,8 @@ static int mdp5_crtc_cursor_set(struct drm_crtc *crtc,
 	lm = mdp5_cstate->pipeline.mixer->lm;
 	stride = width * drm_format_plane_cpp(DRM_FORMAT_ARGB8888, 0);
 
+	pm_runtime_get_sync(&pdev->dev);
+
 	spin_lock_irqsave(&mdp5_crtc->cursor.lock, flags);
 	old_bo = mdp5_crtc->cursor.scanout_bo;
 
@@ -779,8 +787,6 @@ static int mdp5_crtc_cursor_set(struct drm_crtc *crtc,
 
 	get_roi(crtc, &roi_w, &roi_h);
 
-	mdp5_enable(mdp5_kms);
-
 	mdp5_write(mdp5_kms, REG_MDP5_LM_CURSOR_STRIDE(lm), stride);
 	mdp5_write(mdp5_kms, REG_MDP5_LM_CURSOR_FORMAT(lm),
 			MDP5_LM_CURSOR_FORMAT_FORMAT(CURSOR_FMT_ARGB8888));
@@ -798,6 +804,8 @@ static int mdp5_crtc_cursor_set(struct drm_crtc *crtc,
 
 	spin_unlock_irqrestore(&mdp5_crtc->cursor.lock, flags);
 
+	pm_runtime_put_autosuspend(&pdev->dev);
+
 set_cursor:
 	ret = mdp5_ctl_set_cursor(ctl, pipeline, 0, cursor_enable);
 	if (ret) {
@@ -809,7 +817,7 @@ set_cursor:
 	crtc_flush(crtc, flush_mask);
 
 end:
-	mdp5_disable(mdp5_kms);
+	pm_runtime_put_autosuspend(&pdev->dev);
 	if (old_bo) {
 		drm_flip_work_queue(&mdp5_crtc->unref_cursor_work, old_bo);
 		/* enable vblank to complete cursor work: */
@@ -842,7 +850,7 @@ static int mdp5_crtc_cursor_move(struct drm_crtc *crtc, int x, int y)
 
 	get_roi(crtc, &roi_w, &roi_h);
 
-	mdp5_enable(mdp5_kms);
+	pm_runtime_get_sync(&mdp5_kms->pdev->dev);
 
 	spin_lock_irqsave(&mdp5_crtc->cursor.lock, flags);
 	mdp5_write(mdp5_kms, REG_MDP5_LM_CURSOR_SIZE(lm),
@@ -855,7 +863,7 @@ static int mdp5_crtc_cursor_move(struct drm_crtc *crtc, int x, int y)
 
 	crtc_flush(crtc, flush_mask);
 
-	mdp5_disable(mdp5_kms);
+	pm_runtime_put_autosuspend(&mdp5_kms->pdev->dev);
 
 	return 0;
 }
diff --git a/drivers/gpu/drm/msm/mdp/mdp5/mdp5_encoder.c b/drivers/gpu/drm/msm/mdp/mdp5/mdp5_encoder.c
index 70bef51245af..5b851380d3f2 100644
--- a/drivers/gpu/drm/msm/mdp/mdp5/mdp5_encoder.c
+++ b/drivers/gpu/drm/msm/mdp/mdp5/mdp5_encoder.c
@@ -297,6 +297,10 @@ static void mdp5_encoder_enable(struct drm_encoder *encoder)
 {
 	struct mdp5_encoder *mdp5_encoder = to_mdp5_encoder(encoder);
 	struct mdp5_interface *intf = mdp5_encoder->intf;
+	/* this isn't right I think */
+	struct drm_crtc_state *cstate = encoder->crtc->state;
+
+	mdp5_encoder_mode_set(encoder, &cstate->mode, &cstate->adjusted_mode);
 
 	if (intf->mode == MDP5_INTF_DSI_MODE_COMMAND)
 		mdp5_cmd_encoder_enable(encoder);
@@ -320,7 +324,6 @@ static int mdp5_encoder_atomic_check(struct drm_encoder *encoder,
 }
 
 static const struct drm_encoder_helper_funcs mdp5_encoder_helper_funcs = {
-	.mode_set = mdp5_encoder_mode_set,
 	.disable = mdp5_encoder_disable,
 	.enable = mdp5_encoder_enable,
 	.atomic_check = mdp5_encoder_atomic_check,
@@ -350,6 +353,7 @@ int mdp5_vid_encoder_set_split_display(struct drm_encoder *encoder,
 	struct mdp5_encoder *mdp5_encoder = to_mdp5_encoder(encoder);
 	struct mdp5_encoder *mdp5_slave_enc = to_mdp5_encoder(slave_encoder);
 	struct mdp5_kms *mdp5_kms;
+	struct device *dev;
 	int intf_num;
 	u32 data = 0;
 
@@ -369,8 +373,10 @@ int mdp5_vid_encoder_set_split_display(struct drm_encoder *encoder,
 	else
 		return -EINVAL;
 
+	dev = &mdp5_kms->pdev->dev;
 	/* Make sure clocks are on when connectors calling this function. */
-	mdp5_enable(mdp5_kms);
+	pm_runtime_get_sync(dev);
+
 	/* Dumb Panel, Sync mode */
 	mdp5_write(mdp5_kms, REG_MDP5_SPLIT_DPL_UPPER, 0);
 	mdp5_write(mdp5_kms, REG_MDP5_SPLIT_DPL_LOWER, data);
@@ -378,7 +384,7 @@ int mdp5_vid_encoder_set_split_display(struct drm_encoder *encoder,
 
 	mdp5_ctl_pair(mdp5_encoder->ctl, mdp5_slave_enc->ctl, true);
 
-	mdp5_disable(mdp5_kms);
+	pm_runtime_put_autosuspend(dev);
 
 	return 0;
 }
diff --git a/drivers/gpu/drm/msm/mdp/mdp5/mdp5_irq.c b/drivers/gpu/drm/msm/mdp/mdp5/mdp5_irq.c
index 3ce8b9dec9c1..bb5deb00c899 100644
--- a/drivers/gpu/drm/msm/mdp/mdp5/mdp5_irq.c
+++ b/drivers/gpu/drm/msm/mdp/mdp5/mdp5_irq.c
@@ -49,16 +49,19 @@ static void mdp5_irq_error_handler(struct mdp_irq *irq, uint32_t irqstatus)
 void mdp5_irq_preinstall(struct msm_kms *kms)
 {
 	struct mdp5_kms *mdp5_kms = to_mdp5_kms(to_mdp_kms(kms));
-	mdp5_enable(mdp5_kms);
+	struct device *dev = &mdp5_kms->pdev->dev;
+
+	pm_runtime_get_sync(dev);
 	mdp5_write(mdp5_kms, REG_MDP5_INTR_CLEAR, 0xffffffff);
 	mdp5_write(mdp5_kms, REG_MDP5_INTR_EN, 0x00000000);
-	mdp5_disable(mdp5_kms);
+	pm_runtime_put_autosuspend(dev);
 }
 
 int mdp5_irq_postinstall(struct msm_kms *kms)
 {
 	struct mdp_kms *mdp_kms = to_mdp_kms(kms);
 	struct mdp5_kms *mdp5_kms = to_mdp5_kms(mdp_kms);
+	struct device *dev = &mdp5_kms->pdev->dev;
 	struct mdp_irq *error_handler = &mdp5_kms->error_handler;
 
 	error_handler->irq = mdp5_irq_error_handler;
@@ -67,9 +70,9 @@ int mdp5_irq_postinstall(struct msm_kms *kms)
 			MDP5_IRQ_INTF2_UNDER_RUN |
 			MDP5_IRQ_INTF3_UNDER_RUN;
 
-	mdp5_enable(mdp5_kms);
+	pm_runtime_get_sync(dev);
 	mdp_irq_register(mdp_kms, error_handler);
-	mdp5_disable(mdp5_kms);
+	pm_runtime_put_autosuspend(dev);
 
 	return 0;
 }
@@ -77,9 +80,11 @@ int mdp5_irq_postinstall(struct msm_kms *kms)
 void mdp5_irq_uninstall(struct msm_kms *kms)
 {
 	struct mdp5_kms *mdp5_kms = to_mdp5_kms(to_mdp_kms(kms));
-	mdp5_enable(mdp5_kms);
+	struct device *dev = &mdp5_kms->pdev->dev;
+
+	pm_runtime_get_sync(dev);
 	mdp5_write(mdp5_kms, REG_MDP5_INTR_EN, 0x00000000);
-	mdp5_disable(mdp5_kms);
+	pm_runtime_put_autosuspend(dev);
 }
 
 irqreturn_t mdp5_irq(struct msm_kms *kms)
@@ -109,11 +114,12 @@ irqreturn_t mdp5_irq(struct msm_kms *kms)
 int mdp5_enable_vblank(struct msm_kms *kms, struct drm_crtc *crtc)
 {
 	struct mdp5_kms *mdp5_kms = to_mdp5_kms(to_mdp_kms(kms));
+	struct device *dev = &mdp5_kms->pdev->dev;
 
-	mdp5_enable(mdp5_kms);
+	pm_runtime_get_sync(dev);
 	mdp_update_vblank_mask(to_mdp_kms(kms),
 			mdp5_crtc_vblank(crtc), true);
-	mdp5_disable(mdp5_kms);
+	pm_runtime_put_autosuspend(dev);
 
 	return 0;
 }
@@ -121,9 +127,10 @@ int mdp5_enable_vblank(struct msm_kms *kms, struct drm_crtc *crtc)
 void mdp5_disable_vblank(struct msm_kms *kms, struct drm_crtc *crtc)
 {
 	struct mdp5_kms *mdp5_kms = to_mdp5_kms(to_mdp_kms(kms));
+	struct device *dev = &mdp5_kms->pdev->dev;
 
-	mdp5_enable(mdp5_kms);
+	pm_runtime_get_sync(dev);
 	mdp_update_vblank_mask(to_mdp_kms(kms),
 			mdp5_crtc_vblank(crtc), false);
-	mdp5_disable(mdp5_kms);
+	pm_runtime_put_autosuspend(dev);
 }
diff --git a/drivers/gpu/drm/msm/mdp/mdp5/mdp5_kms.c b/drivers/gpu/drm/msm/mdp/mdp5/mdp5_kms.c
index 1c603aef3c59..f7c0698fec40 100644
--- a/drivers/gpu/drm/msm/mdp/mdp5/mdp5_kms.c
+++ b/drivers/gpu/drm/msm/mdp/mdp5/mdp5_kms.c
@@ -30,11 +30,10 @@ static const char *iommu_ports[] = {
 static int mdp5_hw_init(struct msm_kms *kms)
 {
 	struct mdp5_kms *mdp5_kms = to_mdp5_kms(to_mdp_kms(kms));
-	struct platform_device *pdev = mdp5_kms->pdev;
+	struct device *dev = &mdp5_kms->pdev->dev;
 	unsigned long flags;
 
-	pm_runtime_get_sync(&pdev->dev);
-	mdp5_enable(mdp5_kms);
+	pm_runtime_get_sync(dev);
 
 	/* Magic unknown register writes:
 	 *
@@ -66,8 +65,7 @@ static int mdp5_hw_init(struct msm_kms *kms)
 
 	mdp5_ctlm_hw_reset(mdp5_kms->ctlm);
 
-	mdp5_disable(mdp5_kms);
-	pm_runtime_put_sync(&pdev->dev);
+	pm_runtime_put_sync(dev);
 
 	return 0;
 }
@@ -111,8 +109,9 @@ static void mdp5_swap_state(struct msm_kms *kms, struct drm_atomic_state *state)
 static void mdp5_prepare_commit(struct msm_kms *kms, struct drm_atomic_state *state)
 {
 	struct mdp5_kms *mdp5_kms = to_mdp5_kms(to_mdp_kms(kms));
+	struct device *dev = &mdp5_kms->pdev->dev;
 
-	mdp5_enable(mdp5_kms);
+	pm_runtime_get_sync(dev);
 
 	if (mdp5_kms->smp)
 		mdp5_smp_prepare_commit(mdp5_kms->smp, &mdp5_kms->state->smp);
@@ -121,11 +120,12 @@ static void mdp5_prepare_commit(struct msm_kms *kms, struct drm_atomic_state *st
 static void mdp5_complete_commit(struct msm_kms *kms, struct drm_atomic_state *state)
 {
 	struct mdp5_kms *mdp5_kms = to_mdp5_kms(to_mdp_kms(kms));
+	struct device *dev = &mdp5_kms->pdev->dev;
 
 	if (mdp5_kms->smp)
 		mdp5_smp_complete_commit(mdp5_kms->smp, &mdp5_kms->state->smp);
 
-	mdp5_disable(mdp5_kms);
+	pm_runtime_put_autosuspend(dev);
 }
 
 static void mdp5_wait_for_crtc_commit_done(struct msm_kms *kms,
@@ -249,6 +249,9 @@ int mdp5_disable(struct mdp5_kms *mdp5_kms)
 {
 	DBG("");
 
+	mdp5_kms->enable_count--;
+	WARN_ON(mdp5_kms->enable_count < 0);
+
 	clk_disable_unprepare(mdp5_kms->ahb_clk);
 	clk_disable_unprepare(mdp5_kms->axi_clk);
 	clk_disable_unprepare(mdp5_kms->core_clk);
@@ -262,6 +265,8 @@ int mdp5_enable(struct mdp5_kms *mdp5_kms)
 {
 	DBG("");
 
+	mdp5_kms->enable_count++;
+
 	clk_prepare_enable(mdp5_kms->ahb_clk);
 	clk_prepare_enable(mdp5_kms->axi_clk);
 	clk_prepare_enable(mdp5_kms->core_clk);
@@ -486,11 +491,12 @@ fail:
 static void read_mdp_hw_revision(struct mdp5_kms *mdp5_kms,
 				 u32 *major, u32 *minor)
 {
+	struct device *dev = &mdp5_kms->pdev->dev;
 	u32 version;
 
-	mdp5_enable(mdp5_kms);
+	pm_runtime_get_sync(dev);
 	version = mdp5_read(mdp5_kms, REG_MDP5_HW_VERSION);
-	mdp5_disable(mdp5_kms);
+	pm_runtime_put_autosuspend(dev);
 
 	*major = FIELD(version, MDP5_HW_VERSION_MAJOR);
 	*minor = FIELD(version, MDP5_HW_VERSION_MINOR);
@@ -643,7 +649,7 @@ struct msm_kms *mdp5_kms_init(struct drm_device *dev)
 	 * have left things on, in which case we'll start getting faults if
 	 * we don't disable):
 	 */
-	mdp5_enable(mdp5_kms);
+	pm_runtime_get_sync(&pdev->dev);
 	for (i = 0; i < MDP5_INTF_NUM_MAX; i++) {
 		if (mdp5_cfg_intf_is_virtual(config->hw->intf.connect[i]) ||
 		    !config->hw->intf.base[i])
@@ -652,7 +658,6 @@ struct msm_kms *mdp5_kms_init(struct drm_device *dev)
 
 		mdp5_write(mdp5_kms, REG_MDP5_INTF_FRAME_LINE_COUNT_EN(i), 0x3);
 	}
-	mdp5_disable(mdp5_kms);
 	mdelay(16);
 
 	if (config->platform.iommu) {
@@ -678,6 +683,8 @@ struct msm_kms *mdp5_kms_init(struct drm_device *dev)
 		aspace = NULL;;
 	}
 
+	pm_runtime_put_autosuspend(&pdev->dev);
+
 	ret = modeset_init(mdp5_kms);
 	if (ret) {
 		dev_err(&pdev->dev, "modeset_init failed: %d\n", ret);
@@ -1005,6 +1012,30 @@ static int mdp5_dev_remove(struct platform_device *pdev)
 	return 0;
 }
 
+static __maybe_unused int mdp5_runtime_suspend(struct device *dev)
+{
+	struct platform_device *pdev = to_platform_device(dev);
+	struct mdp5_kms *mdp5_kms = platform_get_drvdata(pdev);
+
+	DBG("");
+
+	return mdp5_disable(mdp5_kms);
+}
+
+static __maybe_unused int mdp5_runtime_resume(struct device *dev)
+{
+	struct platform_device *pdev = to_platform_device(dev);
+	struct mdp5_kms *mdp5_kms = platform_get_drvdata(pdev);
+
+	DBG("");
+
+	return mdp5_enable(mdp5_kms);
+}
+
+static const struct dev_pm_ops mdp5_pm_ops = {
+	SET_RUNTIME_PM_OPS(mdp5_runtime_suspend, mdp5_runtime_resume, NULL)
+};
+
 static const struct of_device_id mdp5_dt_match[] = {
 	{ .compatible = "qcom,mdp5", },
 	/* to support downstream DT files */
@@ -1019,6 +1050,7 @@ static struct platform_driver mdp5_driver = {
 	.driver = {
 		.name = "msm_mdp",
 		.of_match_table = mdp5_dt_match,
+		.pm = &mdp5_pm_ops,
 	},
 };
 
diff --git a/drivers/gpu/drm/msm/mdp/mdp5/mdp5_kms.h b/drivers/gpu/drm/msm/mdp/mdp5/mdp5_kms.h
index 17caa0e8c8ae..9b3fe01089d1 100644
--- a/drivers/gpu/drm/msm/mdp/mdp5/mdp5_kms.h
+++ b/drivers/gpu/drm/msm/mdp/mdp5/mdp5_kms.h
@@ -76,6 +76,8 @@ struct mdp5_kms {
 	bool rpm_enabled;
 
 	struct mdp_irq error_handler;
+
+	int enable_count;
 };
 #define to_mdp5_kms(x) container_of(x, struct mdp5_kms, base)
 
@@ -167,11 +169,13 @@ struct mdp5_encoder {
 
 static inline void mdp5_write(struct mdp5_kms *mdp5_kms, u32 reg, u32 data)
 {
+	WARN_ON(mdp5_kms->enable_count <= 0);
 	msm_writel(data, mdp5_kms->mmio + reg);
 }
 
 static inline u32 mdp5_read(struct mdp5_kms *mdp5_kms, u32 reg)
 {
+	WARN_ON(mdp5_kms->enable_count <= 0);
 	return msm_readl(mdp5_kms->mmio + reg);
 }
 
@@ -255,9 +259,6 @@ static inline uint32_t lm2ppdone(struct mdp5_hw_mixer *mixer)
 	return MDP5_IRQ_PING_PONG_0_DONE << mixer->pp;
 }
 
-int mdp5_disable(struct mdp5_kms *mdp5_kms);
-int mdp5_enable(struct mdp5_kms *mdp5_kms);
-
 void mdp5_set_irqmask(struct mdp_kms *mdp_kms, uint32_t irqmask,
 		uint32_t old_irqmask);
 void mdp5_irq_preinstall(struct msm_kms *kms);
diff --git a/drivers/gpu/drm/msm/mdp/mdp5/mdp5_mdss.c b/drivers/gpu/drm/msm/mdp/mdp5/mdp5_mdss.c
index 9c34d7824988..f2a0db7a8a03 100644
--- a/drivers/gpu/drm/msm/mdp/mdp5/mdp5_mdss.c
+++ b/drivers/gpu/drm/msm/mdp/mdp5/mdp5_mdss.c
@@ -31,6 +31,10 @@ struct msm_mdss {
 
 	struct regulator *vdd;
 
+	struct clk *ahb_clk;
+	struct clk *axi_clk;
+	struct clk *vsync_clk;
+
 	struct {
 		volatile unsigned long enabled_mask;
 		struct irq_domain *domain;
@@ -140,6 +144,51 @@ static int mdss_irq_domain_init(struct msm_mdss *mdss)
 	return 0;
 }
 
+int msm_mdss_enable(struct msm_mdss *mdss)
+{
+	DBG("");
+
+	clk_prepare_enable(mdss->ahb_clk);
+	if (mdss->axi_clk)
+		clk_prepare_enable(mdss->axi_clk);
+	if (mdss->vsync_clk)
+		clk_prepare_enable(mdss->vsync_clk);
+
+	return 0;
+}
+
+int msm_mdss_disable(struct msm_mdss *mdss)
+{
+	DBG("");
+
+	if (mdss->vsync_clk)
+		clk_disable_unprepare(mdss->vsync_clk);
+	if (mdss->axi_clk)
+		clk_disable_unprepare(mdss->axi_clk);
+	clk_disable_unprepare(mdss->ahb_clk);
+
+	return 0;
+}
+
+static int msm_mdss_get_clocks(struct msm_mdss *mdss)
+{
+	struct platform_device *pdev = to_platform_device(mdss->dev->dev);
+
+	mdss->ahb_clk = msm_clk_get(pdev, "iface");
+	if (IS_ERR(mdss->ahb_clk))
+		mdss->ahb_clk = NULL;
+
+	mdss->axi_clk = msm_clk_get(pdev, "bus");
+	if (IS_ERR(mdss->axi_clk))
+		mdss->axi_clk = NULL;
+
+	mdss->vsync_clk = msm_clk_get(pdev, "vsync");
+	if (IS_ERR(mdss->vsync_clk))
+		mdss->vsync_clk = NULL;
+
+	return 0;
+}
+
 void msm_mdss_destroy(struct drm_device *dev)
 {
 	struct msm_drm_private *priv = dev->dev_private;
@@ -153,8 +202,6 @@ void msm_mdss_destroy(struct drm_device *dev)
 
 	regulator_disable(mdss->vdd);
 
-	pm_runtime_put_sync(dev->dev);
-
 	pm_runtime_disable(dev->dev);
 }
 
@@ -190,6 +237,12 @@ int msm_mdss_init(struct drm_device *dev)
 		goto fail;
 	}
 
+	ret = msm_mdss_get_clocks(mdss);
+	if (ret) {
+		dev_err(dev->dev, "failed to get clocks: %d\n", ret);
+		goto fail;
+	}
+
 	/* Regulator to enable GDSCs in downstream kernels */
 	mdss->vdd = devm_regulator_get(dev->dev, "vdd");
 	if (IS_ERR(mdss->vdd)) {
@@ -221,12 +274,6 @@ int msm_mdss_init(struct drm_device *dev)
 
 	pm_runtime_enable(dev->dev);
 
-	/*
-	 * TODO: This is needed as the MDSS GDSC is only tied to MDSS's power
-	 * domain. Remove this once runtime PM is adapted for all the devices.
-	 */
-	pm_runtime_get_sync(dev->dev);
-
 	return 0;
 fail_irq:
 	regulator_disable(mdss->vdd);
diff --git a/drivers/gpu/drm/msm/mdp/mdp5/mdp5_smp.c b/drivers/gpu/drm/msm/mdp/mdp5/mdp5_smp.c
index 58f712d37e7f..ae4983d9d0a5 100644
--- a/drivers/gpu/drm/msm/mdp/mdp5/mdp5_smp.c
+++ b/drivers/gpu/drm/msm/mdp/mdp5/mdp5_smp.c
@@ -28,6 +28,13 @@ struct mdp5_smp {
 
 	int blk_cnt;
 	int blk_size;
+
+	/* register cache */
+	u32 alloc_w[22];
+	u32 alloc_r[22];
+	u32 pipe_reqprio_fifo_wm0[SSPP_MAX];
+	u32 pipe_reqprio_fifo_wm1[SSPP_MAX];
+	u32 pipe_reqprio_fifo_wm2[SSPP_MAX];
 };
 
 static inline
@@ -98,16 +105,15 @@ static int smp_request_block(struct mdp5_smp *smp,
 static void set_fifo_thresholds(struct mdp5_smp *smp,
 		enum mdp5_pipe pipe, int nblks)
 {
-	struct mdp5_kms *mdp5_kms = get_kms(smp);
 	u32 smp_entries_per_blk = smp->blk_size / (128 / BITS_PER_BYTE);
 	u32 val;
 
 	/* 1/4 of SMP pool that is being fetched */
 	val = (nblks * smp_entries_per_blk) / 4;
 
-	mdp5_write(mdp5_kms, REG_MDP5_PIPE_REQPRIO_FIFO_WM_0(pipe), val * 1);
-	mdp5_write(mdp5_kms, REG_MDP5_PIPE_REQPRIO_FIFO_WM_1(pipe), val * 2);
-	mdp5_write(mdp5_kms, REG_MDP5_PIPE_REQPRIO_FIFO_WM_2(pipe), val * 3);
+	smp->pipe_reqprio_fifo_wm0[pipe] = val * 1;
+	smp->pipe_reqprio_fifo_wm1[pipe] = val * 2;
+	smp->pipe_reqprio_fifo_wm2[pipe] = val * 3;
 }
 
 /*
@@ -222,7 +228,6 @@ void mdp5_smp_release(struct mdp5_smp *smp, struct mdp5_smp_state *state,
 static unsigned update_smp_state(struct mdp5_smp *smp,
 		u32 cid, mdp5_smp_state_t *assigned)
 {
-	struct mdp5_kms *mdp5_kms = get_kms(smp);
 	int cnt = smp->blk_cnt;
 	unsigned nblks = 0;
 	u32 blk, val;
@@ -231,7 +236,7 @@ static unsigned update_smp_state(struct mdp5_smp *smp,
 		int idx = blk / 3;
 		int fld = blk % 3;
 
-		val = mdp5_read(mdp5_kms, REG_MDP5_SMP_ALLOC_W_REG(idx));
+		val = smp->alloc_w[idx];
 
 		switch (fld) {
 		case 0:
@@ -248,8 +253,8 @@ static unsigned update_smp_state(struct mdp5_smp *smp,
 			break;
 		}
 
-		mdp5_write(mdp5_kms, REG_MDP5_SMP_ALLOC_W_REG(idx), val);
-		mdp5_write(mdp5_kms, REG_MDP5_SMP_ALLOC_R_REG(idx), val);
+		smp->alloc_w[idx] = val;
+		smp->alloc_r[idx] = val;
 
 		nblks++;
 	}
@@ -257,6 +262,39 @@ static unsigned update_smp_state(struct mdp5_smp *smp,
 	return nblks;
 }
 
+static void write_smp_alloc_regs(struct mdp5_smp *smp)
+{
+	struct mdp5_kms *mdp5_kms = get_kms(smp);
+	int i, num_regs;
+
+	num_regs = smp->blk_cnt / 3 + 1;
+
+	for (i = 0; i < num_regs; i++) {
+		mdp5_write(mdp5_kms, REG_MDP5_SMP_ALLOC_W_REG(i),
+			   smp->alloc_w[i]);
+		mdp5_write(mdp5_kms, REG_MDP5_SMP_ALLOC_R_REG(i),
+			   smp->alloc_r[i]);
+	}
+}
+
+static void write_smp_fifo_regs(struct mdp5_smp *smp)
+{
+	struct mdp5_kms *mdp5_kms = get_kms(smp);
+	int i;
+
+	for (i = 0; i < mdp5_kms->num_hwpipes; i++) {
+		struct mdp5_hw_pipe *hwpipe = mdp5_kms->hwpipes[i];
+		enum mdp5_pipe pipe = hwpipe->pipe;
+
+		mdp5_write(mdp5_kms, REG_MDP5_PIPE_REQPRIO_FIFO_WM_0(pipe),
+			   smp->pipe_reqprio_fifo_wm0[pipe]);
+		mdp5_write(mdp5_kms, REG_MDP5_PIPE_REQPRIO_FIFO_WM_1(pipe),
+			   smp->pipe_reqprio_fifo_wm1[pipe]);
+		mdp5_write(mdp5_kms, REG_MDP5_PIPE_REQPRIO_FIFO_WM_2(pipe),
+			   smp->pipe_reqprio_fifo_wm2[pipe]);
+	}
+}
+
 void mdp5_smp_prepare_commit(struct mdp5_smp *smp, struct mdp5_smp_state *state)
 {
 	enum mdp5_pipe pipe;
@@ -277,6 +315,9 @@ void mdp5_smp_prepare_commit(struct mdp5_smp *smp, struct mdp5_smp_state *state)
 		set_fifo_thresholds(smp, pipe, nblks);
 	}
 
+	write_smp_alloc_regs(smp);
+	write_smp_fifo_regs(smp);
+
 	state->assigned = 0;
 }
 
@@ -289,6 +330,8 @@ void mdp5_smp_complete_commit(struct mdp5_smp *smp, struct mdp5_smp_state *state
 		set_fifo_thresholds(smp, pipe, 0);
 	}
 
+	write_smp_fifo_regs(smp);
+
 	state->released = 0;
 }
 
diff --git a/drivers/gpu/drm/msm/msm_drv.c b/drivers/gpu/drm/msm/msm_drv.c
index b0129e7b29e3..606df7bea97b 100644
--- a/drivers/gpu/drm/msm/msm_drv.c
+++ b/drivers/gpu/drm/msm/msm_drv.c
@@ -73,6 +73,10 @@ bool dumpstate = false;
 MODULE_PARM_DESC(dumpstate, "Dump KMS state on errors");
 module_param(dumpstate, bool, 0600);
 
+static bool modeset = true;
+MODULE_PARM_DESC(modeset, "Use kernel modesetting [KMS] (1=on (default), 0=disable)");
+module_param(modeset, bool, 0600);
+
 /*
  * Util/helpers:
  */
@@ -878,8 +882,37 @@ static int msm_pm_resume(struct device *dev)
 }
 #endif
 
+#ifdef CONFIG_PM
+static int msm_runtime_suspend(struct device *dev)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct msm_drm_private *priv = ddev->dev_private;
+
+	DBG("");
+
+	if (priv->mdss)
+		return msm_mdss_disable(priv->mdss);
+
+	return 0;
+}
+
+static int msm_runtime_resume(struct device *dev)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct msm_drm_private *priv = ddev->dev_private;
+
+	DBG("");
+
+	if (priv->mdss)
+		return msm_mdss_enable(priv->mdss);
+
+	return 0;
+}
+#endif
+
 static const struct dev_pm_ops msm_pm_ops = {
 	SET_SYSTEM_SLEEP_PM_OPS(msm_pm_suspend, msm_pm_resume)
+	SET_RUNTIME_PM_OPS(msm_runtime_suspend, msm_runtime_resume, NULL)
 };
 
 /*
@@ -1103,6 +1136,9 @@ static struct platform_driver msm_platform_driver = {
 
 static int __init msm_drm_register(void)
 {
+	if (!modeset)
+		return -EINVAL;
+
 	DBG("init");
 	msm_mdp_register();
 	msm_dsi_register();
diff --git a/drivers/gpu/drm/msm/msm_drv.h b/drivers/gpu/drm/msm/msm_drv.h
index fc8d24f7c084..5e8109c07560 100644
--- a/drivers/gpu/drm/msm/msm_drv.h
+++ b/drivers/gpu/drm/msm/msm_drv.h
@@ -55,8 +55,6 @@ struct msm_fence_cb;
 struct msm_gem_address_space;
 struct msm_gem_vma;
 
-#define NUM_DOMAINS 2    /* one for KMS, then one per gpu core (?) */
-
 struct msm_file_private {
 	/* currently we don't do anything useful with this.. but when
 	 * per-context address spaces are supported we'd keep track of
@@ -237,6 +235,12 @@ struct drm_gem_object *msm_gem_new(struct drm_device *dev,
 		uint32_t size, uint32_t flags);
 struct drm_gem_object *msm_gem_new_locked(struct drm_device *dev,
 		uint32_t size, uint32_t flags);
+void *msm_gem_kernel_new(struct drm_device *dev, uint32_t size,
+		uint32_t flags, struct msm_gem_address_space *aspace,
+		struct drm_gem_object **bo, uint64_t *iova);
+void *msm_gem_kernel_new_locked(struct drm_device *dev, uint32_t size,
+		uint32_t flags, struct msm_gem_address_space *aspace,
+		struct drm_gem_object **bo, uint64_t *iova);
 struct drm_gem_object *msm_gem_import(struct drm_device *dev,
 		struct dma_buf *dmabuf, struct sg_table *sgt);
 
@@ -248,10 +252,10 @@ uint32_t msm_framebuffer_iova(struct drm_framebuffer *fb,
 		struct msm_gem_address_space *aspace, int plane);
 struct drm_gem_object *msm_framebuffer_bo(struct drm_framebuffer *fb, int plane);
 const struct msm_format *msm_framebuffer_format(struct drm_framebuffer *fb);
-struct drm_framebuffer *msm_framebuffer_init(struct drm_device *dev,
-		const struct drm_mode_fb_cmd2 *mode_cmd, struct drm_gem_object **bos);
 struct drm_framebuffer *msm_framebuffer_create(struct drm_device *dev,
 		struct drm_file *file, const struct drm_mode_fb_cmd2 *mode_cmd);
+struct drm_framebuffer * msm_alloc_stolen_fb(struct drm_device *dev,
+		int w, int h, int p, uint32_t format);
 
 struct drm_fb_helper *msm_fbdev_init(struct drm_device *dev);
 void msm_fbdev_free(struct drm_device *dev);
diff --git a/drivers/gpu/drm/msm/msm_fb.c b/drivers/gpu/drm/msm/msm_fb.c
index 6ecb7b170316..fc175e724ad6 100644
--- a/drivers/gpu/drm/msm/msm_fb.c
+++ b/drivers/gpu/drm/msm/msm_fb.c
@@ -20,6 +20,7 @@
 
 #include "msm_drv.h"
 #include "msm_kms.h"
+#include "msm_gem.h"
 
 struct msm_framebuffer {
 	struct drm_framebuffer base;
@@ -28,6 +29,8 @@ struct msm_framebuffer {
 };
 #define to_msm_framebuffer(x) container_of(x, struct msm_framebuffer, base)
 
+static struct drm_framebuffer *msm_framebuffer_init(struct drm_device *dev,
+		const struct drm_mode_fb_cmd2 *mode_cmd, struct drm_gem_object **bos);
 
 static int msm_framebuffer_create_handle(struct drm_framebuffer *fb,
 		struct drm_file *file_priv,
@@ -161,7 +164,7 @@ out_unref:
 	return ERR_PTR(ret);
 }
 
-struct drm_framebuffer *msm_framebuffer_init(struct drm_device *dev,
+static struct drm_framebuffer *msm_framebuffer_init(struct drm_device *dev,
 		const struct drm_mode_fb_cmd2 *mode_cmd, struct drm_gem_object **bos)
 {
 	struct msm_drm_private *priv = dev->dev_private;
@@ -237,3 +240,43 @@ fail:
 
 	return ERR_PTR(ret);
 }
+
+struct drm_framebuffer *
+msm_alloc_stolen_fb(struct drm_device *dev, int w, int h, int p, uint32_t format)
+{
+	struct drm_mode_fb_cmd2 mode_cmd = {
+		.pixel_format = format,
+		.width = w,
+		.height = h,
+		.pitches = { p },
+	};
+	struct drm_gem_object *bo;
+	struct drm_framebuffer *fb;
+	int size;
+
+	/* allocate backing bo */
+	size = mode_cmd.pitches[0] * mode_cmd.height;
+	DBG("allocating %d bytes for fb %d", size, dev->primary->index);
+	bo = msm_gem_new(dev, size, MSM_BO_SCANOUT | MSM_BO_WC | MSM_BO_STOLEN);
+	if (IS_ERR(bo)) {
+		dev_warn(dev->dev, "could not allocate stolen bo\n");
+		/* try regular bo: */
+		bo = msm_gem_new(dev, size, MSM_BO_SCANOUT | MSM_BO_WC);
+	}
+	if (IS_ERR(bo)) {
+		dev_err(dev->dev, "failed to allocate buffer object\n");
+		return ERR_CAST(bo);
+	}
+
+	fb = msm_framebuffer_init(dev, &mode_cmd, &bo);
+	if (IS_ERR(fb)) {
+		dev_err(dev->dev, "failed to allocate fb\n");
+		/* note: if fb creation failed, we can't rely on fb destroy
+		 * to unref the bo:
+		 */
+		drm_gem_object_unreference_unlocked(bo);
+		return ERR_CAST(fb);
+	}
+
+	return fb;
+}
diff --git a/drivers/gpu/drm/msm/msm_fbdev.c b/drivers/gpu/drm/msm/msm_fbdev.c
index 9c00fedfc741..c178563fcd4d 100644
--- a/drivers/gpu/drm/msm/msm_fbdev.c
+++ b/drivers/gpu/drm/msm/msm_fbdev.c
@@ -19,7 +19,6 @@
 #include <drm/drm_fb_helper.h>
 
 #include "msm_drv.h"
-#include "msm_gem.h"
 #include "msm_kms.h"
 
 extern int msm_gem_mmap_obj(struct drm_gem_object *obj,
@@ -35,7 +34,6 @@ static int msm_fbdev_mmap(struct fb_info *info, struct vm_area_struct *vma);
 struct msm_fbdev {
 	struct drm_fb_helper base;
 	struct drm_framebuffer *fb;
-	struct drm_gem_object *bo;
 };
 
 static struct fb_ops msm_fb_ops = {
@@ -57,16 +55,16 @@ static int msm_fbdev_mmap(struct fb_info *info, struct vm_area_struct *vma)
 {
 	struct drm_fb_helper *helper = (struct drm_fb_helper *)info->par;
 	struct msm_fbdev *fbdev = to_msm_fbdev(helper);
-	struct drm_gem_object *drm_obj = fbdev->bo;
+	struct drm_gem_object *bo = msm_framebuffer_bo(fbdev->fb, 0);
 	int ret = 0;
 
-	ret = drm_gem_mmap_obj(drm_obj, drm_obj->size, vma);
+	ret = drm_gem_mmap_obj(bo, bo->size, vma);
 	if (ret) {
 		pr_err("%s:drm_gem_mmap_obj fail\n", __func__);
 		return ret;
 	}
 
-	return msm_gem_mmap_obj(drm_obj, vma);
+	return msm_gem_mmap_obj(bo, vma);
 }
 
 static int msm_fbdev_create(struct drm_fb_helper *helper,
@@ -76,47 +74,30 @@ static int msm_fbdev_create(struct drm_fb_helper *helper,
 	struct drm_device *dev = helper->dev;
 	struct msm_drm_private *priv = dev->dev_private;
 	struct drm_framebuffer *fb = NULL;
+	struct drm_gem_object *bo;
 	struct fb_info *fbi = NULL;
-	struct drm_mode_fb_cmd2 mode_cmd = {0};
 	uint64_t paddr;
-	int ret, size;
+	uint32_t format;
+	int ret, pitch;
+
+	format = drm_mode_legacy_fb_format(sizes->surface_bpp, sizes->surface_depth);
 
 	DBG("create fbdev: %dx%d@%d (%dx%d)", sizes->surface_width,
 			sizes->surface_height, sizes->surface_bpp,
 			sizes->fb_width, sizes->fb_height);
 
-	mode_cmd.pixel_format = drm_mode_legacy_fb_format(sizes->surface_bpp,
-			sizes->surface_depth);
-
-	mode_cmd.width = sizes->surface_width;
-	mode_cmd.height = sizes->surface_height;
-
-	mode_cmd.pitches[0] = align_pitch(
-			mode_cmd.width, sizes->surface_bpp);
+	pitch = align_pitch(sizes->surface_width, sizes->surface_bpp);
+	fb = msm_alloc_stolen_fb(dev, sizes->surface_width,
+			sizes->surface_height, pitch, format);
 
-	/* allocate backing bo */
-	size = mode_cmd.pitches[0] * mode_cmd.height;
-	DBG("allocating %d bytes for fb %d", size, dev->primary->index);
-	fbdev->bo = msm_gem_new(dev, size, MSM_BO_SCANOUT |
-			MSM_BO_WC | MSM_BO_STOLEN);
-	if (IS_ERR(fbdev->bo)) {
-		ret = PTR_ERR(fbdev->bo);
-		fbdev->bo = NULL;
-		dev_err(dev->dev, "failed to allocate buffer object: %d\n", ret);
-		goto fail;
-	}
-
-	fb = msm_framebuffer_init(dev, &mode_cmd, &fbdev->bo);
 	if (IS_ERR(fb)) {
 		dev_err(dev->dev, "failed to allocate fb\n");
-		/* note: if fb creation failed, we can't rely on fb destroy
-		 * to unref the bo:
-		 */
-		drm_gem_object_unreference_unlocked(fbdev->bo);
 		ret = PTR_ERR(fb);
 		goto fail;
 	}
 
+	bo = msm_framebuffer_bo(fb, 0);
+
 	mutex_lock(&dev->struct_mutex);
 
 	/*
@@ -124,7 +105,7 @@ static int msm_fbdev_create(struct drm_fb_helper *helper,
 	 * in panic (ie. lock-safe, etc) we could avoid pinning the
 	 * buffer now:
 	 */
-	ret = msm_gem_get_iova(fbdev->bo, priv->kms->aspace, &paddr);
+	ret = msm_gem_get_iova(bo, priv->kms->aspace, &paddr);
 	if (ret) {
 		dev_err(dev->dev, "failed to get buffer obj iova: %d\n", ret);
 		goto fail_unlock;
@@ -152,14 +133,14 @@ static int msm_fbdev_create(struct drm_fb_helper *helper,
 
 	dev->mode_config.fb_base = paddr;
 
-	fbi->screen_base = msm_gem_get_vaddr(fbdev->bo);
+	fbi->screen_base = msm_gem_get_vaddr(bo);
 	if (IS_ERR(fbi->screen_base)) {
 		ret = PTR_ERR(fbi->screen_base);
 		goto fail_unlock;
 	}
-	fbi->screen_size = fbdev->bo->size;
+	fbi->screen_size = bo->size;
 	fbi->fix.smem_start = paddr;
-	fbi->fix.smem_len = fbdev->bo->size;
+	fbi->fix.smem_len = bo->size;
 
 	DBG("par=%p, %dx%d", fbi->par, fbi->var.xres, fbi->var.yres);
 	DBG("allocated %dx%d fb", fbdev->fb->width, fbdev->fb->height);
@@ -241,7 +222,9 @@ void msm_fbdev_free(struct drm_device *dev)
 
 	/* this will free the backing object */
 	if (fbdev->fb) {
-		msm_gem_put_vaddr(fbdev->bo);
+		struct drm_gem_object *bo =
+			msm_framebuffer_bo(fbdev->fb, 0);
+		msm_gem_put_vaddr(bo);
 		drm_framebuffer_remove(fbdev->fb);
 	}
 
diff --git a/drivers/gpu/drm/msm/msm_gem.c b/drivers/gpu/drm/msm/msm_gem.c
index a0c60e738db8..f15821a0d900 100644
--- a/drivers/gpu/drm/msm/msm_gem.c
+++ b/drivers/gpu/drm/msm/msm_gem.c
@@ -1024,3 +1024,49 @@ fail:
 	drm_gem_object_unreference_unlocked(obj);
 	return ERR_PTR(ret);
 }
+
+static void *_msm_gem_kernel_new(struct drm_device *dev, uint32_t size,
+		uint32_t flags, struct msm_gem_address_space *aspace,
+		struct drm_gem_object **bo, uint64_t *iova, bool locked)
+{
+	void *vaddr;
+	struct drm_gem_object *obj = _msm_gem_new(dev, size, flags, locked);
+	int ret;
+
+	if (IS_ERR(obj))
+		return ERR_CAST(obj);
+
+	if (iova) {
+		ret = msm_gem_get_iova(obj, aspace, iova);
+		if (ret) {
+			drm_gem_object_unreference(obj);
+			return ERR_PTR(ret);
+		}
+	}
+
+	vaddr = msm_gem_get_vaddr(obj);
+	if (!vaddr) {
+		msm_gem_put_iova(obj, aspace);
+		drm_gem_object_unreference(obj);
+		return ERR_PTR(-ENOMEM);
+	}
+
+	if (bo)
+		*bo = obj;
+
+	return vaddr;
+}
+
+void *msm_gem_kernel_new(struct drm_device *dev, uint32_t size,
+		uint32_t flags, struct msm_gem_address_space *aspace,
+		struct drm_gem_object **bo, uint64_t *iova)
+{
+	return _msm_gem_kernel_new(dev, size, flags, aspace, bo, iova, false);
+}
+
+void *msm_gem_kernel_new_locked(struct drm_device *dev, uint32_t size,
+		uint32_t flags, struct msm_gem_address_space *aspace,
+		struct drm_gem_object **bo, uint64_t *iova)
+{
+	return _msm_gem_kernel_new(dev, size, flags, aspace, bo, iova, true);
+}
diff --git a/drivers/gpu/drm/msm/msm_gpu.c b/drivers/gpu/drm/msm/msm_gpu.c
index 9f3dbc236ab3..ffbff27600e0 100644
--- a/drivers/gpu/drm/msm/msm_gpu.c
+++ b/drivers/gpu/drm/msm/msm_gpu.c
@@ -562,11 +562,49 @@ static int get_clocks(struct platform_device *pdev, struct msm_gpu *gpu)
 	return 0;
 }
 
+static struct msm_gem_address_space *
+msm_gpu_create_address_space(struct msm_gpu *gpu, struct platform_device *pdev,
+		uint64_t va_start, uint64_t va_end)
+{
+	struct iommu_domain *iommu;
+	struct msm_gem_address_space *aspace;
+	int ret;
+
+	/*
+	 * Setup IOMMU.. eventually we will (I think) do this once per context
+	 * and have separate page tables per context.  For now, to keep things
+	 * simple and to get something working, just use a single address space:
+	 */
+	iommu = iommu_domain_alloc(&platform_bus_type);
+	if (!iommu)
+		return NULL;
+
+	iommu->geometry.aperture_start = va_start;
+	iommu->geometry.aperture_end = va_end;
+
+	dev_info(gpu->dev->dev, "%s: using IOMMU\n", gpu->name);
+
+	aspace = msm_gem_address_space_create(&pdev->dev, iommu, "gpu");
+	if (IS_ERR(aspace)) {
+		dev_err(gpu->dev->dev, "failed to init iommu: %ld\n",
+			PTR_ERR(aspace));
+		iommu_domain_free(iommu);
+		return ERR_CAST(aspace);
+	}
+
+	ret = aspace->mmu->funcs->attach(aspace->mmu, NULL, 0);
+	if (ret) {
+		msm_gem_address_space_put(aspace);
+		return ERR_PTR(ret);
+	}
+
+	return aspace;
+}
+
 int msm_gpu_init(struct drm_device *drm, struct platform_device *pdev,
 		struct msm_gpu *gpu, const struct msm_gpu_funcs *funcs,
 		const char *name, struct msm_gpu_config *config)
 {
-	struct iommu_domain *iommu;
 	int ret;
 
 	if (WARN_ON(gpu->num_perfcntrs > ARRAY_SIZE(gpu->last_cntrs)))
@@ -636,28 +674,19 @@ int msm_gpu_init(struct drm_device *drm, struct platform_device *pdev,
 	if (IS_ERR(gpu->gpu_cx))
 		gpu->gpu_cx = NULL;
 
-	/* Setup IOMMU.. eventually we will (I think) do this once per context
-	 * and have separate page tables per context.  For now, to keep things
-	 * simple and to get something working, just use a single address space:
-	 */
-	iommu = iommu_domain_alloc(&platform_bus_type);
-	if (iommu) {
-		iommu->geometry.aperture_start = config->va_start;
-		iommu->geometry.aperture_end = config->va_end;
-
-		dev_info(drm->dev, "%s: using IOMMU\n", name);
-		gpu->aspace = msm_gem_address_space_create(&pdev->dev,
-				iommu, "gpu");
-		if (IS_ERR(gpu->aspace)) {
-			ret = PTR_ERR(gpu->aspace);
-			dev_err(drm->dev, "failed to init iommu: %d\n", ret);
-			gpu->aspace = NULL;
-			iommu_domain_free(iommu);
-			goto fail;
-		}
+	gpu->pdev = pdev;
+	platform_set_drvdata(pdev, gpu);
+
+	bs_init(gpu);
 
-	} else {
+	gpu->aspace = msm_gpu_create_address_space(gpu, pdev,
+		config->va_start, config->va_end);
+
+	if (gpu->aspace == NULL)
 		dev_info(drm->dev, "%s: no IOMMU, fallback to VRAM carveout!\n", name);
+	else if (IS_ERR(gpu->aspace)) {
+		ret = PTR_ERR(gpu->aspace);
+		goto fail;
 	}
 
 	/* Create ringbuffer: */
@@ -669,14 +698,10 @@ int msm_gpu_init(struct drm_device *drm, struct platform_device *pdev,
 		goto fail;
 	}
 
-	gpu->pdev = pdev;
-	platform_set_drvdata(pdev, gpu);
-
-	bs_init(gpu);
-
 	return 0;
 
 fail:
+	platform_set_drvdata(pdev, NULL);
 	return ret;
 }
 
@@ -693,7 +718,9 @@ void msm_gpu_cleanup(struct msm_gpu *gpu)
 			msm_gem_put_iova(gpu->rb->bo, gpu->aspace);
 		msm_ringbuffer_destroy(gpu->rb);
 	}
-
-	if (gpu->fctx)
-		msm_fence_context_free(gpu->fctx);
+	if (gpu->aspace) {
+		gpu->aspace->mmu->funcs->detach(gpu->aspace->mmu,
+			NULL, 0);
+		msm_gem_address_space_put(gpu->aspace);
+	}
 }
diff --git a/drivers/gpu/drm/msm/msm_kms.h b/drivers/gpu/drm/msm/msm_kms.h
index a8f2ba5e5f07..17d5824417ad 100644
--- a/drivers/gpu/drm/msm/msm_kms.h
+++ b/drivers/gpu/drm/msm/msm_kms.h
@@ -99,5 +99,7 @@ struct msm_kms *mdp4_kms_init(struct drm_device *dev);
 struct msm_kms *mdp5_kms_init(struct drm_device *dev);
 int msm_mdss_init(struct drm_device *dev);
 void msm_mdss_destroy(struct drm_device *dev);
+int msm_mdss_enable(struct msm_mdss *mdss);
+int msm_mdss_disable(struct msm_mdss *mdss);
 
 #endif /* __MSM_KMS_H__ */
diff --git a/drivers/gpu/drm/msm/msm_ringbuffer.c b/drivers/gpu/drm/msm/msm_ringbuffer.c
index 791bca3c6a9c..bf065a540130 100644
--- a/drivers/gpu/drm/msm/msm_ringbuffer.c
+++ b/drivers/gpu/drm/msm/msm_ringbuffer.c
@@ -33,16 +33,14 @@ struct msm_ringbuffer *msm_ringbuffer_new(struct msm_gpu *gpu, int size)
 	}
 
 	ring->gpu = gpu;
-	ring->bo = msm_gem_new(gpu->dev, size, MSM_BO_WC);
-	if (IS_ERR(ring->bo)) {
-		ret = PTR_ERR(ring->bo);
-		ring->bo = NULL;
-		goto fail;
-	}
 
-	ring->start = msm_gem_get_vaddr(ring->bo);
+	/* Pass NULL for the iova pointer - we will map it later */
+	ring->start = msm_gem_kernel_new(gpu->dev, size, MSM_BO_WC,
+		gpu->aspace, &ring->bo, NULL);
+
 	if (IS_ERR(ring->start)) {
 		ret = PTR_ERR(ring->start);
+		ring->start = 0;
 		goto fail;
 	}
 	ring->end   = ring->start + (size / 4);