[PATCH] drm/msm/dpu: ensure device suspend happens during PM sleep

[PATCH] drm/msm/dpu: ensure device suspend happens during PM sleep

[Kalyan Thota]

"The PM core always increments the runtime usage counter
before calling the ->suspend() callback and decrements it
after calling the ->resume() callback"

DPU and DSI are managed as runtime devices. When
suspend is triggered, PM core adds a refcount on all the
devices and calls device suspend, since usage count is
already incremented, runtime suspend was not getting called
and it kept the clocks on which resulted in target not
entering into XO shutdown.

Add changes to manage runtime devices during pm sleep.

Signed-off-by: Kalyan Thota <kalyan_t@codeaurora.org>
---
 drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c | 41 +++++++++++++++++++++++++++++++++
 drivers/gpu/drm/msm/dsi/dsi.c           |  7 ++++++
 drivers/gpu/drm/msm/msm_drv.c           | 14 +++++++++++
 drivers/gpu/drm/msm/msm_kms.h           |  2 ++
 4 files changed, 64 insertions(+)

diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
index cb08faf..6e103d5 100644
--- a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
+++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
@@ -26,6 +26,7 @@
 #include "dpu_encoder.h"
 #include "dpu_plane.h"
 #include "dpu_crtc.h"
+#include "dsi.h"
 
 #define CREATE_TRACE_POINTS
 #include "dpu_trace.h"
@@ -250,6 +251,37 @@ static void dpu_kms_disable_commit(struct msm_kms *kms)
 	pm_runtime_put_sync(&dpu_kms->pdev->dev);
 }
 
+static void _dpu_kms_disable_dpu(struct msm_kms *kms)
+{
+	struct drm_device *dev;
+	struct msm_drm_private *priv;
+	struct dpu_kms *dpu_kms;
+	int i = 0;
+	struct msm_dsi *dsi;
+
+	dpu_kms = to_dpu_kms(kms);
+	dev = dpu_kms->dev;
+	if (!dev) {
+		DPU_ERROR("invalid device\n");
+		return;
+	}
+
+	priv = dev->dev_private;
+	if (!priv) {
+		DPU_ERROR("invalid private data\n");
+		return;
+	}
+
+	dpu_kms_disable_commit(kms);
+
+	for (i = 0; i < ARRAY_SIZE(priv->dsi); i++) {
+		if (!priv->dsi[i])
+			continue;
+		dsi = priv->dsi[i];
+		pm_runtime_put_sync(&dsi->pdev->dev);
+	}
+}
+
 static ktime_t dpu_kms_vsync_time(struct msm_kms *kms, struct drm_crtc *crtc)
 {
 	struct drm_encoder *encoder;
@@ -683,6 +715,7 @@ static void dpu_irq_uninstall(struct msm_kms *kms)
 #ifdef CONFIG_DEBUG_FS
 	.debugfs_init    = dpu_kms_debugfs_init,
 #endif
+	.disable_dpu = _dpu_kms_disable_dpu,
 };
 
 static void _dpu_kms_mmu_destroy(struct dpu_kms *dpu_kms)
@@ -1053,7 +1086,15 @@ static int __maybe_unused dpu_runtime_resume(struct device *dev)
 	return rc;
 }
 
+
+static int __maybe_unused dpu_pm_suspend_late(struct device *dev)
+{
+	pm_runtime_get_noresume(dev);
+	return 0;
+}
+
 static const struct dev_pm_ops dpu_pm_ops = {
+	SET_LATE_SYSTEM_SLEEP_PM_OPS(dpu_pm_suspend_late, NULL)
 	SET_RUNTIME_PM_OPS(dpu_runtime_suspend, dpu_runtime_resume, NULL)
 };
 
diff --git a/drivers/gpu/drm/msm/dsi/dsi.c b/drivers/gpu/drm/msm/dsi/dsi.c
index 55ea4bc2..3d3740e 100644
--- a/drivers/gpu/drm/msm/dsi/dsi.c
+++ b/drivers/gpu/drm/msm/dsi/dsi.c
@@ -154,12 +154,19 @@ static int dsi_dev_remove(struct platform_device *pdev)
 	return 0;
 }
 
+static int __maybe_unused dsi_pm_suspend_late(struct device *dev)
+{
+	pm_runtime_get_noresume(dev);
+	return 0;
+}
+
 static const struct of_device_id dt_match[] = {
 	{ .compatible = "qcom,mdss-dsi-ctrl" },
 	{}
 };
 
 static const struct dev_pm_ops dsi_pm_ops = {
+	SET_LATE_SYSTEM_SLEEP_PM_OPS(dsi_pm_suspend_late, NULL)
 	SET_RUNTIME_PM_OPS(msm_dsi_runtime_suspend, msm_dsi_runtime_resume, NULL)
 };
 
diff --git a/drivers/gpu/drm/msm/msm_drv.c b/drivers/gpu/drm/msm/msm_drv.c
index e4b750b..12ec1c6 100644
--- a/drivers/gpu/drm/msm/msm_drv.c
+++ b/drivers/gpu/drm/msm/msm_drv.c
@@ -1038,6 +1038,7 @@ static int msm_pm_suspend(struct device *dev)
 {
 	struct drm_device *ddev = dev_get_drvdata(dev);
 	struct msm_drm_private *priv = ddev->dev_private;
+	struct msm_kms *kms = priv->kms;
 
 	if (WARN_ON(priv->pm_state))
 		drm_atomic_state_put(priv->pm_state);
@@ -1049,6 +1050,11 @@ static int msm_pm_suspend(struct device *dev)
 		return ret;
 	}
 
+	if (kms->funcs->disable_dpu)
+		kms->funcs->disable_dpu(kms);
+
+	pm_runtime_put_sync(dev);
+
 	return 0;
 }
 
@@ -1067,6 +1073,13 @@ static int msm_pm_resume(struct device *dev)
 
 	return ret;
 }
+
+static int msm_pm_suspend_late(struct device *dev)
+{
+	pm_runtime_get_noresume(dev);
+	return 0;
+}
+
 #endif
 
 #ifdef CONFIG_PM
@@ -1100,6 +1113,7 @@ static int msm_runtime_resume(struct device *dev)
 #endif
 
 static const struct dev_pm_ops msm_pm_ops = {
+	SET_LATE_SYSTEM_SLEEP_PM_OPS(msm_pm_suspend_late, NULL)
 	SET_SYSTEM_SLEEP_PM_OPS(msm_pm_suspend, msm_pm_resume)
 	SET_RUNTIME_PM_OPS(msm_runtime_suspend, msm_runtime_resume, NULL)
 };
diff --git a/drivers/gpu/drm/msm/msm_kms.h b/drivers/gpu/drm/msm/msm_kms.h
index 1cbef6b..c73a89b 100644
--- a/drivers/gpu/drm/msm/msm_kms.h
+++ b/drivers/gpu/drm/msm/msm_kms.h
@@ -126,6 +126,8 @@ struct msm_kms_funcs {
 	/* debugfs: */
 	int (*debugfs_init)(struct msm_kms *kms, struct drm_minor *minor);
 #endif
+	void (*disable_dpu)(struct msm_kms *kms);
+
 };
 
 struct msm_kms;
-- 
1.9.1

Re: [PATCH] drm/msm/dpu: ensure device suspend happens during PM sleep

[Rob Clark]

On Mon, Mar 16, 2020 at 4:05 AM Kalyan Thota <kalyan_t@codeaurora.org> wrote:
>
> "The PM core always increments the runtime usage counter
> before calling the ->suspend() callback and decrements it
> after calling the ->resume() callback"
>
> DPU and DSI are managed as runtime devices. When
> suspend is triggered, PM core adds a refcount on all the
> devices and calls device suspend, since usage count is
> already incremented, runtime suspend was not getting called
> and it kept the clocks on which resulted in target not
> entering into XO shutdown.
>
> Add changes to manage runtime devices during pm sleep.
>
> Signed-off-by: Kalyan Thota <kalyan_t@codeaurora.org>
> ---
>  drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c | 41 +++++++++++++++++++++++++++++++++
>  drivers/gpu/drm/msm/dsi/dsi.c           |  7 ++++++
>  drivers/gpu/drm/msm/msm_drv.c           | 14 +++++++++++
>  drivers/gpu/drm/msm/msm_kms.h           |  2 ++
>  4 files changed, 64 insertions(+)
>
> diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
> index cb08faf..6e103d5 100644
> --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
> +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
> @@ -26,6 +26,7 @@
>  #include "dpu_encoder.h"
>  #include "dpu_plane.h"
>  #include "dpu_crtc.h"
> +#include "dsi.h"
>
>  #define CREATE_TRACE_POINTS
>  #include "dpu_trace.h"
> @@ -250,6 +251,37 @@ static void dpu_kms_disable_commit(struct msm_kms *kms)
>         pm_runtime_put_sync(&dpu_kms->pdev->dev);
>  }
>
> +static void _dpu_kms_disable_dpu(struct msm_kms *kms)
> +{
> +       struct drm_device *dev;
> +       struct msm_drm_private *priv;
> +       struct dpu_kms *dpu_kms;
> +       int i = 0;
> +       struct msm_dsi *dsi;
> +
> +       dpu_kms = to_dpu_kms(kms);
> +       dev = dpu_kms->dev;
> +       if (!dev) {
> +               DPU_ERROR("invalid device\n");
> +               return;
> +       }
> +
> +       priv = dev->dev_private;
> +       if (!priv) {
> +               DPU_ERROR("invalid private data\n");
> +               return;
> +       }

the !dev and !priv checks can be dropped.. these aren't things a user
should hit, and if I screw somethign up in development and hit that
case, I'd rather see a stack trace

otherwise, I think it looks reasonable

BR,
-R

> +
> +       dpu_kms_disable_commit(kms);
> +
> +       for (i = 0; i < ARRAY_SIZE(priv->dsi); i++) {
> +               if (!priv->dsi[i])
> +                       continue;
> +               dsi = priv->dsi[i];
> +               pm_runtime_put_sync(&dsi->pdev->dev);
> +       }
> +}
> +
>  static ktime_t dpu_kms_vsync_time(struct msm_kms *kms, struct drm_crtc *crtc)
>  {
>         struct drm_encoder *encoder;
> @@ -683,6 +715,7 @@ static void dpu_irq_uninstall(struct msm_kms *kms)
>  #ifdef CONFIG_DEBUG_FS
>         .debugfs_init    = dpu_kms_debugfs_init,
>  #endif
> +       .disable_dpu = _dpu_kms_disable_dpu,
>  };
>
>  static void _dpu_kms_mmu_destroy(struct dpu_kms *dpu_kms)
> @@ -1053,7 +1086,15 @@ static int __maybe_unused dpu_runtime_resume(struct device *dev)
>         return rc;
>  }
>
> +
> +static int __maybe_unused dpu_pm_suspend_late(struct device *dev)
> +{
> +       pm_runtime_get_noresume(dev);
> +       return 0;
> +}
> +
>  static const struct dev_pm_ops dpu_pm_ops = {
> +       SET_LATE_SYSTEM_SLEEP_PM_OPS(dpu_pm_suspend_late, NULL)
>         SET_RUNTIME_PM_OPS(dpu_runtime_suspend, dpu_runtime_resume, NULL)
>  };
>
> diff --git a/drivers/gpu/drm/msm/dsi/dsi.c b/drivers/gpu/drm/msm/dsi/dsi.c
> index 55ea4bc2..3d3740e 100644
> --- a/drivers/gpu/drm/msm/dsi/dsi.c
> +++ b/drivers/gpu/drm/msm/dsi/dsi.c
> @@ -154,12 +154,19 @@ static int dsi_dev_remove(struct platform_device *pdev)
>         return 0;
>  }
>
> +static int __maybe_unused dsi_pm_suspend_late(struct device *dev)
> +{
> +       pm_runtime_get_noresume(dev);
> +       return 0;
> +}
> +
>  static const struct of_device_id dt_match[] = {
>         { .compatible = "qcom,mdss-dsi-ctrl" },
>         {}
>  };
>
>  static const struct dev_pm_ops dsi_pm_ops = {
> +       SET_LATE_SYSTEM_SLEEP_PM_OPS(dsi_pm_suspend_late, NULL)
>         SET_RUNTIME_PM_OPS(msm_dsi_runtime_suspend, msm_dsi_runtime_resume, NULL)
>  };
>
> diff --git a/drivers/gpu/drm/msm/msm_drv.c b/drivers/gpu/drm/msm/msm_drv.c
> index e4b750b..12ec1c6 100644
> --- a/drivers/gpu/drm/msm/msm_drv.c
> +++ b/drivers/gpu/drm/msm/msm_drv.c
> @@ -1038,6 +1038,7 @@ static int msm_pm_suspend(struct device *dev)
>  {
>         struct drm_device *ddev = dev_get_drvdata(dev);
>         struct msm_drm_private *priv = ddev->dev_private;
> +       struct msm_kms *kms = priv->kms;
>
>         if (WARN_ON(priv->pm_state))
>                 drm_atomic_state_put(priv->pm_state);
> @@ -1049,6 +1050,11 @@ static int msm_pm_suspend(struct device *dev)
>                 return ret;
>         }
>
> +       if (kms->funcs->disable_dpu)
> +               kms->funcs->disable_dpu(kms);
> +
> +       pm_runtime_put_sync(dev);
> +
>         return 0;
>  }
>
> @@ -1067,6 +1073,13 @@ static int msm_pm_resume(struct device *dev)
>
>         return ret;
>  }
> +
> +static int msm_pm_suspend_late(struct device *dev)
> +{
> +       pm_runtime_get_noresume(dev);
> +       return 0;
> +}
> +
>  #endif
>
>  #ifdef CONFIG_PM
> @@ -1100,6 +1113,7 @@ static int msm_runtime_resume(struct device *dev)
>  #endif
>
>  static const struct dev_pm_ops msm_pm_ops = {
> +       SET_LATE_SYSTEM_SLEEP_PM_OPS(msm_pm_suspend_late, NULL)
>         SET_SYSTEM_SLEEP_PM_OPS(msm_pm_suspend, msm_pm_resume)
>         SET_RUNTIME_PM_OPS(msm_runtime_suspend, msm_runtime_resume, NULL)
>  };
> diff --git a/drivers/gpu/drm/msm/msm_kms.h b/drivers/gpu/drm/msm/msm_kms.h
> index 1cbef6b..c73a89b 100644
> --- a/drivers/gpu/drm/msm/msm_kms.h
> +++ b/drivers/gpu/drm/msm/msm_kms.h
> @@ -126,6 +126,8 @@ struct msm_kms_funcs {
>         /* debugfs: */
>         int (*debugfs_init)(struct msm_kms *kms, struct drm_minor *minor);
>  #endif
> +       void (*disable_dpu)(struct msm_kms *kms);
> +
>  };
>
>  struct msm_kms;
> --
> 1.9.1
>

Re: [PATCH] drm/msm/dpu: ensure device suspend happens during PM sleep

[Doug Anderson]

Hi,

On Mon, Mar 16, 2020 at 4:06 AM Kalyan Thota <kalyan_t@codeaurora.org> wrote:
>
> "The PM core always increments the runtime usage counter
> before calling the ->suspend() callback and decrements it
> after calling the ->resume() callback"
>
> DPU and DSI are managed as runtime devices. When
> suspend is triggered, PM core adds a refcount on all the
> devices and calls device suspend, since usage count is
> already incremented, runtime suspend was not getting called
> and it kept the clocks on which resulted in target not
> entering into XO shutdown.
>
> Add changes to manage runtime devices during pm sleep.
>
> Signed-off-by: Kalyan Thota <kalyan_t@codeaurora.org>
> ---
>  drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c | 41 +++++++++++++++++++++++++++++++++
>  drivers/gpu/drm/msm/dsi/dsi.c           |  7 ++++++
>  drivers/gpu/drm/msm/msm_drv.c           | 14 +++++++++++
>  drivers/gpu/drm/msm/msm_kms.h           |  2 ++
>  4 files changed, 64 insertions(+)
>
> diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
> index cb08faf..6e103d5 100644
> --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
> +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
> @@ -26,6 +26,7 @@
>  #include "dpu_encoder.h"
>  #include "dpu_plane.h"
>  #include "dpu_crtc.h"
> +#include "dsi.h"
>
>  #define CREATE_TRACE_POINTS
>  #include "dpu_trace.h"
> @@ -250,6 +251,37 @@ static void dpu_kms_disable_commit(struct msm_kms *kms)
>         pm_runtime_put_sync(&dpu_kms->pdev->dev);
>  }
>
> +static void _dpu_kms_disable_dpu(struct msm_kms *kms)
> +{
> +       struct drm_device *dev;
> +       struct msm_drm_private *priv;
> +       struct dpu_kms *dpu_kms;
> +       int i = 0;
> +       struct msm_dsi *dsi;
> +
> +       dpu_kms = to_dpu_kms(kms);
> +       dev = dpu_kms->dev;
> +       if (!dev) {
> +               DPU_ERROR("invalid device\n");
> +               return;
> +       }
> +
> +       priv = dev->dev_private;
> +       if (!priv) {
> +               DPU_ERROR("invalid private data\n");
> +               return;
> +       }
> +
> +       dpu_kms_disable_commit(kms);
> +
> +       for (i = 0; i < ARRAY_SIZE(priv->dsi); i++) {
> +               if (!priv->dsi[i])
> +                       continue;
> +               dsi = priv->dsi[i];
> +               pm_runtime_put_sync(&dsi->pdev->dev);
> +       }
> +}
> +
>  static ktime_t dpu_kms_vsync_time(struct msm_kms *kms, struct drm_crtc *crtc)
>  {
>         struct drm_encoder *encoder;
> @@ -683,6 +715,7 @@ static void dpu_irq_uninstall(struct msm_kms *kms)
>  #ifdef CONFIG_DEBUG_FS
>         .debugfs_init    = dpu_kms_debugfs_init,
>  #endif
> +       .disable_dpu = _dpu_kms_disable_dpu,
>  };
>
>  static void _dpu_kms_mmu_destroy(struct dpu_kms *dpu_kms)
> @@ -1053,7 +1086,15 @@ static int __maybe_unused dpu_runtime_resume(struct device *dev)
>         return rc;
>  }
>
> +
> +static int __maybe_unused dpu_pm_suspend_late(struct device *dev)
> +{
> +       pm_runtime_get_noresume(dev);
> +       return 0;
> +}
> +
>  static const struct dev_pm_ops dpu_pm_ops = {
> +       SET_LATE_SYSTEM_SLEEP_PM_OPS(dpu_pm_suspend_late, NULL)
>         SET_RUNTIME_PM_OPS(dpu_runtime_suspend, dpu_runtime_resume, NULL)
>  };
>
> diff --git a/drivers/gpu/drm/msm/dsi/dsi.c b/drivers/gpu/drm/msm/dsi/dsi.c
> index 55ea4bc2..3d3740e 100644
> --- a/drivers/gpu/drm/msm/dsi/dsi.c
> +++ b/drivers/gpu/drm/msm/dsi/dsi.c
> @@ -154,12 +154,19 @@ static int dsi_dev_remove(struct platform_device *pdev)
>         return 0;
>  }
>
> +static int __maybe_unused dsi_pm_suspend_late(struct device *dev)
> +{
> +       pm_runtime_get_noresume(dev);
> +       return 0;
> +}
> +
>  static const struct of_device_id dt_match[] = {
>         { .compatible = "qcom,mdss-dsi-ctrl" },
>         {}
>  };
>
>  static const struct dev_pm_ops dsi_pm_ops = {
> +       SET_LATE_SYSTEM_SLEEP_PM_OPS(dsi_pm_suspend_late, NULL)
>         SET_RUNTIME_PM_OPS(msm_dsi_runtime_suspend, msm_dsi_runtime_resume, NULL)
>  };
>
> diff --git a/drivers/gpu/drm/msm/msm_drv.c b/drivers/gpu/drm/msm/msm_drv.c
> index e4b750b..12ec1c6 100644
> --- a/drivers/gpu/drm/msm/msm_drv.c
> +++ b/drivers/gpu/drm/msm/msm_drv.c
> @@ -1038,6 +1038,7 @@ static int msm_pm_suspend(struct device *dev)
>  {
>         struct drm_device *ddev = dev_get_drvdata(dev);
>         struct msm_drm_private *priv = ddev->dev_private;
> +       struct msm_kms *kms = priv->kms;
>
>         if (WARN_ON(priv->pm_state))
>                 drm_atomic_state_put(priv->pm_state);
> @@ -1049,6 +1050,11 @@ static int msm_pm_suspend(struct device *dev)
>                 return ret;
>         }
>
> +       if (kms->funcs->disable_dpu)
> +               kms->funcs->disable_dpu(kms);
> +
> +       pm_runtime_put_sync(dev);
> +
>         return 0;
>  }
>
> @@ -1067,6 +1073,13 @@ static int msm_pm_resume(struct device *dev)
>
>         return ret;
>  }
> +
> +static int msm_pm_suspend_late(struct device *dev)
> +{
> +       pm_runtime_get_noresume(dev);
> +       return 0;
> +}
> +
>  #endif
>
>  #ifdef CONFIG_PM
> @@ -1100,6 +1113,7 @@ static int msm_runtime_resume(struct device *dev)
>  #endif
>
>  static const struct dev_pm_ops msm_pm_ops = {
> +       SET_LATE_SYSTEM_SLEEP_PM_OPS(msm_pm_suspend_late, NULL)
>         SET_SYSTEM_SLEEP_PM_OPS(msm_pm_suspend, msm_pm_resume)
>         SET_RUNTIME_PM_OPS(msm_runtime_suspend, msm_runtime_resume, NULL)
>  };
> diff --git a/drivers/gpu/drm/msm/msm_kms.h b/drivers/gpu/drm/msm/msm_kms.h
> index 1cbef6b..c73a89b 100644
> --- a/drivers/gpu/drm/msm/msm_kms.h
> +++ b/drivers/gpu/drm/msm/msm_kms.h
> @@ -126,6 +126,8 @@ struct msm_kms_funcs {
>         /* debugfs: */
>         int (*debugfs_init)(struct msm_kms *kms, struct drm_minor *minor);
>  #endif
> +       void (*disable_dpu)(struct msm_kms *kms);
> +

I'm by no means an expert on any of this, but it seems awfully
asymmetric to me and that feels like you're going to end up with bugs.

* Is it possible that disable_dpu() will be called more than once
during normal operation, like if the screen goes off?  Will your
counts be off then?

* What happens if suspend is aborted partway through (by getting a
wakeup even as you're suspending, for instance)?  In such a case some
of the normal suspend calls will be called but "suspend_late" won't be
called.  Does that mess up your counting?


From your description, it sure seems like this part of the
runtime_pm.rst doc is relevant to you:

> Namely, if a system suspend .prepare() callback returns a positive
> number for a device, that indicates to the PM core that the device
> appears to be runtime-suspended and its state is fine, so it may
> be left in runtime suspend provided that all of its descendants
> are also left in runtime suspend.

Did I misunderstand and this isn't what you want?  Looking a bit
further, maybe the right thing is to use the "SMART_SUSPEND" flag?

I'll also note that sometimes when people just want to be Runtime PM
managed that I see this pattern:

SET_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend, pm_runtime_force_resume)


-Doug

[PATCH] drm/msm/dpu: ensure device suspend happens during PM sleep

[Kalyan Thota]

"The PM core always increments the runtime usage counter
before calling the ->suspend() callback and decrements it
after calling the ->resume() callback"

DPU and DSI are managed as runtime devices. When
suspend is triggered, PM core adds a refcount on all the
devices and calls device suspend, since usage count is
already incremented, runtime suspend was not getting called
and it kept the clocks on which resulted in target not
entering into XO shutdown.

Add changes to manage runtime devices during pm sleep.

Changes in v1:
 - Remove unnecessary checks in the function
     _dpu_kms_disable_dpu (Rob Clark).

Signed-off-by: Kalyan Thota <kalyan_t@codeaurora.org>
---
 drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c | 28 ++++++++++++++++++++++++++++
 drivers/gpu/drm/msm/dsi/dsi.c           |  7 +++++++
 drivers/gpu/drm/msm/msm_drv.c           | 14 ++++++++++++++
 drivers/gpu/drm/msm/msm_kms.h           |  2 ++
 4 files changed, 51 insertions(+)

diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
index ce19f1d..c3e8287 100644
--- a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
+++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
@@ -26,6 +26,7 @@
 #include "dpu_encoder.h"
 #include "dpu_plane.h"
 #include "dpu_crtc.h"
+#include "dsi.h"
 
 #define CREATE_TRACE_POINTS
 #include "dpu_trace.h"
@@ -325,6 +326,24 @@ static void dpu_kms_disable_commit(struct msm_kms *kms)
 	pm_runtime_put_sync(&dpu_kms->pdev->dev);
 }
 
+static void _dpu_kms_disable_dpu(struct msm_kms *kms)
+{
+	struct dpu_kms *dpu_kms = to_dpu_kms(kms);
+	struct drm_device *dev = dpu_kms->dev;
+	struct msm_drm_private *priv = dev->dev_private;
+	struct msm_dsi *dsi;
+	int i;
+
+	dpu_kms_disable_commit(kms);
+
+	for (i = 0; i < ARRAY_SIZE(priv->dsi); i++) {
+		if (!priv->dsi[i])
+			continue;
+		dsi = priv->dsi[i];
+		pm_runtime_put_sync(&dsi->pdev->dev);
+	}
+}
+
 static ktime_t dpu_kms_vsync_time(struct msm_kms *kms, struct drm_crtc *crtc)
 {
 	struct drm_encoder *encoder;
@@ -751,6 +770,7 @@ static void dpu_irq_uninstall(struct msm_kms *kms)
 #ifdef CONFIG_DEBUG_FS
 	.debugfs_init    = dpu_kms_debugfs_init,
 #endif
+	.disable_dpu = _dpu_kms_disable_dpu,
 };
 
 static void _dpu_kms_mmu_destroy(struct dpu_kms *dpu_kms)
@@ -1121,7 +1141,15 @@ static int __maybe_unused dpu_runtime_resume(struct device *dev)
 	return rc;
 }
 
+
+static int __maybe_unused dpu_pm_suspend_late(struct device *dev)
+{
+	pm_runtime_get_noresume(dev);
+	return 0;
+}
+
 static const struct dev_pm_ops dpu_pm_ops = {
+	SET_LATE_SYSTEM_SLEEP_PM_OPS(dpu_pm_suspend_late, NULL)
 	SET_RUNTIME_PM_OPS(dpu_runtime_suspend, dpu_runtime_resume, NULL)
 };
 
diff --git a/drivers/gpu/drm/msm/dsi/dsi.c b/drivers/gpu/drm/msm/dsi/dsi.c
index 55ea4bc2..3d3740e 100644
--- a/drivers/gpu/drm/msm/dsi/dsi.c
+++ b/drivers/gpu/drm/msm/dsi/dsi.c
@@ -154,12 +154,19 @@ static int dsi_dev_remove(struct platform_device *pdev)
 	return 0;
 }
 
+static int __maybe_unused dsi_pm_suspend_late(struct device *dev)
+{
+	pm_runtime_get_noresume(dev);
+	return 0;
+}
+
 static const struct of_device_id dt_match[] = {
 	{ .compatible = "qcom,mdss-dsi-ctrl" },
 	{}
 };
 
 static const struct dev_pm_ops dsi_pm_ops = {
+	SET_LATE_SYSTEM_SLEEP_PM_OPS(dsi_pm_suspend_late, NULL)
 	SET_RUNTIME_PM_OPS(msm_dsi_runtime_suspend, msm_dsi_runtime_resume, NULL)
 };
 
diff --git a/drivers/gpu/drm/msm/msm_drv.c b/drivers/gpu/drm/msm/msm_drv.c
index 7d985f8..7451ae0 100644
--- a/drivers/gpu/drm/msm/msm_drv.c
+++ b/drivers/gpu/drm/msm/msm_drv.c
@@ -1040,6 +1040,7 @@ static int msm_pm_suspend(struct device *dev)
 {
 	struct drm_device *ddev = dev_get_drvdata(dev);
 	struct msm_drm_private *priv = ddev->dev_private;
+	struct msm_kms *kms = priv->kms;
 
 	if (WARN_ON(priv->pm_state))
 		drm_atomic_state_put(priv->pm_state);
@@ -1051,6 +1052,11 @@ static int msm_pm_suspend(struct device *dev)
 		return ret;
 	}
 
+	if (kms->funcs->disable_dpu)
+		kms->funcs->disable_dpu(kms);
+
+	pm_runtime_put_sync(dev);
+
 	return 0;
 }
 
@@ -1069,6 +1075,13 @@ static int msm_pm_resume(struct device *dev)
 
 	return ret;
 }
+
+static int msm_pm_suspend_late(struct device *dev)
+{
+	pm_runtime_get_noresume(dev);
+	return 0;
+}
+
 #endif
 
 #ifdef CONFIG_PM
@@ -1102,6 +1115,7 @@ static int msm_runtime_resume(struct device *dev)
 #endif
 
 static const struct dev_pm_ops msm_pm_ops = {
+	SET_LATE_SYSTEM_SLEEP_PM_OPS(msm_pm_suspend_late, NULL)
 	SET_SYSTEM_SLEEP_PM_OPS(msm_pm_suspend, msm_pm_resume)
 	SET_RUNTIME_PM_OPS(msm_runtime_suspend, msm_runtime_resume, NULL)
 };
diff --git a/drivers/gpu/drm/msm/msm_kms.h b/drivers/gpu/drm/msm/msm_kms.h
index 1cbef6b..c73a89b 100644
--- a/drivers/gpu/drm/msm/msm_kms.h
+++ b/drivers/gpu/drm/msm/msm_kms.h
@@ -126,6 +126,8 @@ struct msm_kms_funcs {
 	/* debugfs: */
 	int (*debugfs_init)(struct msm_kms *kms, struct drm_minor *minor);
 #endif
+	void (*disable_dpu)(struct msm_kms *kms);
+
 };
 
 struct msm_kms;
-- 
1.9.1

Re: [PATCH] drm/msm/dpu: ensure device suspend happens during PM sleep

[Doug Anderson]

Hi,

On Sun, Mar 22, 2020 at 11:14 PM Kalyan Thota <kalyan_t@codeaurora.org> wrote:
>
> "The PM core always increments the runtime usage counter
> before calling the ->suspend() callback and decrements it
> after calling the ->resume() callback"
>
> DPU and DSI are managed as runtime devices. When
> suspend is triggered, PM core adds a refcount on all the
> devices and calls device suspend, since usage count is
> already incremented, runtime suspend was not getting called
> and it kept the clocks on which resulted in target not
> entering into XO shutdown.
>
> Add changes to manage runtime devices during pm sleep.
>
> Changes in v1:
>  - Remove unnecessary checks in the function
>      _dpu_kms_disable_dpu (Rob Clark).

I'm wondering what happened with my feedback on v1, AKA:

https://lore.kernel.org/r/CAD=FV=VxzEV40g+ieuEN+7o=34+wM8MHO8o7T5zA1Yosx7SVWg@mail.gmail.com

Maybe you didn't see it?  ...or if you or Rob think I'm way off base
(always possible) then please tell me so.

Thanks!

-Doug

Re: [PATCH] drm/msm/dpu: ensure device suspend happens during PM sleep

[Rob Clark]

On Tue, Mar 24, 2020 at 7:35 AM Doug Anderson <dianders@chromium.org> wrote:
>
> Hi,
>
> On Sun, Mar 22, 2020 at 11:14 PM Kalyan Thota <kalyan_t@codeaurora.org> wrote:
> >
> > "The PM core always increments the runtime usage counter
> > before calling the ->suspend() callback and decrements it
> > after calling the ->resume() callback"
> >
> > DPU and DSI are managed as runtime devices. When
> > suspend is triggered, PM core adds a refcount on all the
> > devices and calls device suspend, since usage count is
> > already incremented, runtime suspend was not getting called
> > and it kept the clocks on which resulted in target not
> > entering into XO shutdown.
> >
> > Add changes to manage runtime devices during pm sleep.
> >
> > Changes in v1:
> >  - Remove unnecessary checks in the function
> >      _dpu_kms_disable_dpu (Rob Clark).
>
> I'm wondering what happened with my feedback on v1, AKA:
>
> https://lore.kernel.org/r/CAD=FV=VxzEV40g+ieuEN+7o=34+wM8MHO8o7T5zA1Yosx7SVWg@mail.gmail.com
>
> Maybe you didn't see it?  ...or if you or Rob think I'm way off base
> (always possible) then please tell me so.
>

At least w/ the current patch, disable_dpu should not be called for
screen-off (although I'd hope if all the screens are off the device
would suspend).  But I won't claim to be a pm expert.. so not really
sure if this is the best approach or not.  I don't think our
arrangement of sub-devices under a parent is completely abnormal, so
it does feel like there should be a simpler solution..

BR,
-R

Re: [PATCH] drm/msm/dpu: ensure device suspend happens during PM sleep

[Doug Anderson]

Hi,

On Wed, Mar 25, 2020 at 8:40 AM Rob Clark <robdclark@gmail.com> wrote:
>
> On Tue, Mar 24, 2020 at 7:35 AM Doug Anderson <dianders@chromium.org> wrote:
> >
> > Hi,
> >
> > On Sun, Mar 22, 2020 at 11:14 PM Kalyan Thota <kalyan_t@codeaurora.org> wrote:
> > >
> > > "The PM core always increments the runtime usage counter
> > > before calling the ->suspend() callback and decrements it
> > > after calling the ->resume() callback"
> > >
> > > DPU and DSI are managed as runtime devices. When
> > > suspend is triggered, PM core adds a refcount on all the
> > > devices and calls device suspend, since usage count is
> > > already incremented, runtime suspend was not getting called
> > > and it kept the clocks on which resulted in target not
> > > entering into XO shutdown.
> > >
> > > Add changes to manage runtime devices during pm sleep.
> > >
> > > Changes in v1:
> > >  - Remove unnecessary checks in the function
> > >      _dpu_kms_disable_dpu (Rob Clark).
> >
> > I'm wondering what happened with my feedback on v1, AKA:
> >
> > https://lore.kernel.org/r/CAD=FV=VxzEV40g+ieuEN+7o=34+wM8MHO8o7T5zA1Yosx7SVWg@mail.gmail.com
> >
> > Maybe you didn't see it?  ...or if you or Rob think I'm way off base
> > (always possible) then please tell me so.
> >
>
> At least w/ the current patch, disable_dpu should not be called for
> screen-off (although I'd hope if all the screens are off the device
> would suspend).

OK, that's good.

> But I won't claim to be a pm expert.. so not really
> sure if this is the best approach or not.  I don't think our
> arrangement of sub-devices under a parent is completely abnormal, so
> it does feel like there should be a simpler solution..

I think the other arguments about asymmetry are still valid and I've
fixed bugs around this type of thing in the past.  For instance, see
commit f7ccbed656f7 ("drm/rockchip: Suspend DP late").


-Doug

Re: [PATCH] drm/msm/dpu: ensure device suspend happens during PM sleep

[kalyan_t]

On 2020-03-25 21:20, Doug Anderson wrote:
> Hi,
> 
> On Wed, Mar 25, 2020 at 8:40 AM Rob Clark <robdclark@gmail.com> wrote:
>> 
>> On Tue, Mar 24, 2020 at 7:35 AM Doug Anderson <dianders@chromium.org> 
>> wrote:
>> >
>> > Hi,
>> >
>> > On Sun, Mar 22, 2020 at 11:14 PM Kalyan Thota <kalyan_t@codeaurora.org> wrote:
>> > >
>> > > "The PM core always increments the runtime usage counter
>> > > before calling the ->suspend() callback and decrements it
>> > > after calling the ->resume() callback"
>> > >
>> > > DPU and DSI are managed as runtime devices. When
>> > > suspend is triggered, PM core adds a refcount on all the
>> > > devices and calls device suspend, since usage count is
>> > > already incremented, runtime suspend was not getting called
>> > > and it kept the clocks on which resulted in target not
>> > > entering into XO shutdown.
>> > >
>> > > Add changes to manage runtime devices during pm sleep.
>> > >
>> > > Changes in v1:
>> > >  - Remove unnecessary checks in the function
>> > >      _dpu_kms_disable_dpu (Rob Clark).
>> >
>> > I'm wondering what happened with my feedback on v1, AKA:
>> >
>> > https://lore.kernel.org/r/CAD=FV=VxzEV40g+ieuEN+7o=34+wM8MHO8o7T5zA1Yosx7SVWg@mail.gmail.com
>> >
>> > Maybe you didn't see it?  ...or if you or Rob think I'm way off base
>> > (always possible) then please tell me so.
>> >
-- I didn't notice your comments earlier. Apologies !!

>> 
>> At least w/ the current patch, disable_dpu should not be called for
>> screen-off (although I'd hope if all the screens are off the device
>> would suspend).
> 
> OK, that's good.

-- Rob has answered it, with current change disable_dpu will only be 
called during pm_suspend.
> 
>> But I won't claim to be a pm expert.. so not really
>> sure if this is the best approach or not.  I don't think our
>> arrangement of sub-devices under a parent is completely abnormal, so
>> it does feel like there should be a simpler solution..
> 
> I think the other arguments about asymmetry are still valid and I've
> fixed bugs around this type of thing in the past.  For instance, see
> commit f7ccbed656f7 ("drm/rockchip: Suspend DP late").
> 

* What happens if suspend is aborted partway through (by getting a
wakeup even as you're suspending, for instance)?  In such a case some
of the normal suspend calls will be called but "suspend_late" won't be
called.  Does that mess up your counting?

-- I understand this concern, i'll explore a bit more on how to handle 
"failed to suspend","early awake"
cases (to restore the usage_count) since suspend_late wont be called.

*From your description, it sure seems like this part of the
runtime_pm.rst doc is relevant to you:

Did I misunderstand and this isn't what you want?  Looking a bit
further, maybe the right thing is to use the "SMART_SUSPEND" flag?

-- if you notice in the device_prepare 
(https://elixir.bootlin.com/linux/latest/source/drivers/base/power/main.c#L1913)
there is a pm_runtime_get_noresume at L1931, which will increment the 
usagecount before triggering client prepare call, hence implementing 
prepare wont fetch us much.

This appears to be more for the cases when device is runtime suspended 
and suspend followed later
"one example usecase that i can think of, is screen timeout after that 
suspend is triggered"

currently the problem i am looking at is that
	PM Core does +1 in device prepare
		DPU driver does -1 in suspend
		DPU driver does +1 in suspend late  ( look for right place )
	PM core does -1 in device complete

i'll get back after exploring a bit.

> 
> -Doug

[PATCH] drm/msm/dpu: ensure device suspend happens during PM sleep

[Kalyan Thota]

"The PM core always increments the runtime usage counter
before calling the ->suspend() callback and decrements it
after calling the ->resume() callback"

DPU and DSI are managed as runtime devices. When
suspend is triggered, PM core adds a refcount on all the
devices and calls device suspend, since usage count is
already incremented, runtime suspend was not getting called
and it kept the clocks on which resulted in target not
entering into XO shutdown.

Add changes to manage runtime devices during pm sleep.

Changes in v1:
 - Remove unnecessary checks in the function
   _dpu_kms_disable_dpu (Rob Clark).

Changes in v2:
 - Avoid using suspend_late to reset the usagecount
   as suspend_late might not be called during suspend
   call failures (Doug).

Signed-off-by: Kalyan Thota <kalyan_t@codeaurora.org>
---
 drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c | 33 +++++++++++++++++++++++++++++++++
 drivers/gpu/drm/msm/msm_drv.c           |  4 ++++
 drivers/gpu/drm/msm/msm_kms.h           |  2 ++
 3 files changed, 39 insertions(+)

diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
index ce19f1d..2343cbd 100644
--- a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
+++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
@@ -26,6 +26,7 @@
 #include "dpu_encoder.h"
 #include "dpu_plane.h"
 #include "dpu_crtc.h"
+#include "dsi.h"
 
 #define CREATE_TRACE_POINTS
 #include "dpu_trace.h"
@@ -325,6 +326,37 @@ static void dpu_kms_disable_commit(struct msm_kms *kms)
 	pm_runtime_put_sync(&dpu_kms->pdev->dev);
 }
 
+static void _dpu_kms_disable_dpu(struct msm_kms *kms)
+{
+	struct dpu_kms *dpu_kms = to_dpu_kms(kms);
+	struct drm_device *dev = dpu_kms->dev;
+	struct msm_drm_private *priv = dev->dev_private;
+	struct msm_dsi *dsi;
+	int i;
+
+	dpu_kms_disable_commit(kms);
+
+	for (i = 0; i < ARRAY_SIZE(priv->dsi); i++) {
+		if (!priv->dsi[i])
+			continue;
+		dsi = priv->dsi[i];
+		pm_runtime_put_sync(&dsi->pdev->dev);
+	}
+	pm_runtime_put_sync(dev->dev);
+
+	/* Increment the usagecount without triggering a resume */
+	pm_runtime_get_noresume(dev->dev);
+
+	pm_runtime_get_noresume(&dpu_kms->pdev->dev);
+
+	for (i = 0; i < ARRAY_SIZE(priv->dsi); i++) {
+		if (!priv->dsi[i])
+			continue;
+		dsi = priv->dsi[i];
+		pm_runtime_get_noresume(&dsi->pdev->dev);
+	}
+}
+
 static ktime_t dpu_kms_vsync_time(struct msm_kms *kms, struct drm_crtc *crtc)
 {
 	struct drm_encoder *encoder;
@@ -751,6 +783,7 @@ static void dpu_irq_uninstall(struct msm_kms *kms)
 #ifdef CONFIG_DEBUG_FS
 	.debugfs_init    = dpu_kms_debugfs_init,
 #endif
+	.disable_dpu = _dpu_kms_disable_dpu,
 };
 
 static void _dpu_kms_mmu_destroy(struct dpu_kms *dpu_kms)
diff --git a/drivers/gpu/drm/msm/msm_drv.c b/drivers/gpu/drm/msm/msm_drv.c
index 7d985f8..bb11e00 100644
--- a/drivers/gpu/drm/msm/msm_drv.c
+++ b/drivers/gpu/drm/msm/msm_drv.c
@@ -1040,6 +1040,7 @@ static int msm_pm_suspend(struct device *dev)
 {
 	struct drm_device *ddev = dev_get_drvdata(dev);
 	struct msm_drm_private *priv = ddev->dev_private;
+	struct msm_kms *kms = priv->kms;
 
 	if (WARN_ON(priv->pm_state))
 		drm_atomic_state_put(priv->pm_state);
@@ -1051,6 +1052,9 @@ static int msm_pm_suspend(struct device *dev)
 		return ret;
 	}
 
+	if (kms->funcs->disable_dpu)
+		kms->funcs->disable_dpu(kms);
+
 	return 0;
 }
 
diff --git a/drivers/gpu/drm/msm/msm_kms.h b/drivers/gpu/drm/msm/msm_kms.h
index 1cbef6b..c73a89b 100644
--- a/drivers/gpu/drm/msm/msm_kms.h
+++ b/drivers/gpu/drm/msm/msm_kms.h
@@ -126,6 +126,8 @@ struct msm_kms_funcs {
 	/* debugfs: */
 	int (*debugfs_init)(struct msm_kms *kms, struct drm_minor *minor);
 #endif
+	void (*disable_dpu)(struct msm_kms *kms);
+
 };
 
 struct msm_kms;
-- 
1.9.1

Re: [PATCH] drm/msm/dpu: ensure device suspend happens during PM sleep

[Doug Anderson]

Hi,

On Mon, Mar 30, 2020 at 2:04 AM Kalyan Thota <kalyan_t@codeaurora.org> wrote:
>
> "The PM core always increments the runtime usage counter
> before calling the ->suspend() callback and decrements it
> after calling the ->resume() callback"
>
> DPU and DSI are managed as runtime devices. When
> suspend is triggered, PM core adds a refcount on all the
> devices and calls device suspend, since usage count is
> already incremented, runtime suspend was not getting called
> and it kept the clocks on which resulted in target not
> entering into XO shutdown.
>
> Add changes to manage runtime devices during pm sleep.
>
> Changes in v1:
>  - Remove unnecessary checks in the function
>    _dpu_kms_disable_dpu (Rob Clark).
>
> Changes in v2:
>  - Avoid using suspend_late to reset the usagecount
>    as suspend_late might not be called during suspend
>    call failures (Doug).
>
> Signed-off-by: Kalyan Thota <kalyan_t@codeaurora.org>
> ---
>  drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c | 33 +++++++++++++++++++++++++++++++++
>  drivers/gpu/drm/msm/msm_drv.c           |  4 ++++
>  drivers/gpu/drm/msm/msm_kms.h           |  2 ++
>  3 files changed, 39 insertions(+)

I am still 100% baffled by your patch and I never did quite understand
your response to my previous comments [1].  I think you're saying that
the problem you were facing is that if you call "suspend" but never
called "runtime_suspend" that the device stays active.  Is that right?
 If that's true, did you try something like this suggestion I made?

SET_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend, pm_runtime_force_resume)


> diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
> index ce19f1d..2343cbd 100644
> --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
> +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
> @@ -26,6 +26,7 @@
>  #include "dpu_encoder.h"
>  #include "dpu_plane.h"
>  #include "dpu_crtc.h"
> +#include "dsi.h"
>
>  #define CREATE_TRACE_POINTS
>  #include "dpu_trace.h"
> @@ -325,6 +326,37 @@ static void dpu_kms_disable_commit(struct msm_kms *kms)
>         pm_runtime_put_sync(&dpu_kms->pdev->dev);
>  }
>
> +static void _dpu_kms_disable_dpu(struct msm_kms *kms)
> +{
> +       struct dpu_kms *dpu_kms = to_dpu_kms(kms);
> +       struct drm_device *dev = dpu_kms->dev;
> +       struct msm_drm_private *priv = dev->dev_private;
> +       struct msm_dsi *dsi;
> +       int i;
> +
> +       dpu_kms_disable_commit(kms);
> +
> +       for (i = 0; i < ARRAY_SIZE(priv->dsi); i++) {
> +               if (!priv->dsi[i])
> +                       continue;
> +               dsi = priv->dsi[i];
> +               pm_runtime_put_sync(&dsi->pdev->dev);
> +       }
> +       pm_runtime_put_sync(dev->dev);
> +
> +       /* Increment the usagecount without triggering a resume */
> +       pm_runtime_get_noresume(dev->dev);
> +
> +       pm_runtime_get_noresume(&dpu_kms->pdev->dev);
> +
> +       for (i = 0; i < ARRAY_SIZE(priv->dsi); i++) {
> +               if (!priv->dsi[i])
> +                       continue;
> +               dsi = priv->dsi[i];
> +               pm_runtime_get_noresume(&dsi->pdev->dev);
> +       }
> +}

My pm_runtime knowledge is pretty weak sometimes, but the above
function looks crazy.  Maybe it's just me not understanding, but can
you please summarize what you're trying to accomplish?

-Doug

[1] https://lore.kernel.org/r/114130f68c494f83303c51157e2c5bfa@codeaurora.org

[PATCH] drm/msm/dpu: ensure device suspend happens during PM sleep

[Kalyan Thota]

"The PM core always increments the runtime usage counter
before calling the ->suspend() callback and decrements it
after calling the ->resume() callback"

DPU and DSI are managed as runtime devices. When
suspend is triggered, PM core adds a refcount on all the
devices and calls device suspend, since usage count is
already incremented, runtime suspend was not getting called
and it kept the clocks on which resulted in target not
entering into XO shutdown.

Add changes to force suspend on runtime devices during pm sleep.

Changes in v1:
 - Remove unnecessary checks in the function
    _dpu_kms_disable_dpu (Rob Clark).

Changes in v2:
 - Avoid using suspend_late to reset the usagecount
   as suspend_late might not be called during suspend
   call failures (Doug).

Changes in v3:
 - Use force suspend instead of managing device usage_count
   via runtime put and get API's to trigger callbacks (Doug).

Signed-off-by: Kalyan Thota <kalyan_t@codeaurora.org>
---
 drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c | 2 ++
 drivers/gpu/drm/msm/dsi/dsi.c           | 2 ++
 drivers/gpu/drm/msm/msm_drv.c           | 4 ++++
 3 files changed, 8 insertions(+)

diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
index ce19f1d..b886d9d 100644
--- a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
+++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
@@ -1123,6 +1123,8 @@ static int __maybe_unused dpu_runtime_resume(struct device *dev)
 
 static const struct dev_pm_ops dpu_pm_ops = {
 	SET_RUNTIME_PM_OPS(dpu_runtime_suspend, dpu_runtime_resume, NULL)
+	SET_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend,
+				pm_runtime_force_resume)
 };
 
 static const struct of_device_id dpu_dt_match[] = {
diff --git a/drivers/gpu/drm/msm/dsi/dsi.c b/drivers/gpu/drm/msm/dsi/dsi.c
index 55ea4bc2..62704885 100644
--- a/drivers/gpu/drm/msm/dsi/dsi.c
+++ b/drivers/gpu/drm/msm/dsi/dsi.c
@@ -161,6 +161,8 @@ static int dsi_dev_remove(struct platform_device *pdev)
 
 static const struct dev_pm_ops dsi_pm_ops = {
 	SET_RUNTIME_PM_OPS(msm_dsi_runtime_suspend, msm_dsi_runtime_resume, NULL)
+	SET_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend,
+				pm_runtime_force_resume)
 };
 
 static struct platform_driver dsi_driver = {
diff --git a/drivers/gpu/drm/msm/msm_drv.c b/drivers/gpu/drm/msm/msm_drv.c
index 7d985f8..2b8c99c 100644
--- a/drivers/gpu/drm/msm/msm_drv.c
+++ b/drivers/gpu/drm/msm/msm_drv.c
@@ -1051,6 +1051,8 @@ static int msm_pm_suspend(struct device *dev)
 		return ret;
 	}
 
+	pm_runtime_force_suspend(dev);
+
 	return 0;
 }
 
@@ -1063,6 +1065,8 @@ static int msm_pm_resume(struct device *dev)
 	if (WARN_ON(!priv->pm_state))
 		return -ENOENT;
 
+	pm_runtime_force_resume(dev);
+
 	ret = drm_atomic_helper_resume(ddev, priv->pm_state);
 	if (!ret)
 		priv->pm_state = NULL;
-- 
1.9.1

Re: [PATCH] drm/msm/dpu: ensure device suspend happens during PM sleep

[Doug Anderson]

Hi,

On Tue, Mar 31, 2020 at 6:58 AM Kalyan Thota <kalyan_t@codeaurora.org> wrote:
>
> "The PM core always increments the runtime usage counter
> before calling the ->suspend() callback and decrements it
> after calling the ->resume() callback"
>
> DPU and DSI are managed as runtime devices. When
> suspend is triggered, PM core adds a refcount on all the
> devices and calls device suspend, since usage count is
> already incremented, runtime suspend was not getting called
> and it kept the clocks on which resulted in target not
> entering into XO shutdown.
>
> Add changes to force suspend on runtime devices during pm sleep.
>
> Changes in v1:
>  - Remove unnecessary checks in the function
>     _dpu_kms_disable_dpu (Rob Clark).
>
> Changes in v2:
>  - Avoid using suspend_late to reset the usagecount
>    as suspend_late might not be called during suspend
>    call failures (Doug).
>
> Changes in v3:
>  - Use force suspend instead of managing device usage_count
>    via runtime put and get API's to trigger callbacks (Doug).
>
> Signed-off-by: Kalyan Thota <kalyan_t@codeaurora.org>
> ---
>  drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c | 2 ++
>  drivers/gpu/drm/msm/dsi/dsi.c           | 2 ++
>  drivers/gpu/drm/msm/msm_drv.c           | 4 ++++
>  3 files changed, 8 insertions(+)

This looks much saner to me.  Thanks!  I assume it still works fine
for you?  I'm still no expert on how all the pieces of DRM drivers
work together, but at least there's not a bunch of strange fiddling
with pm_runtime state and hopefully it will avoid weird corner
cases...


> diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
> index ce19f1d..b886d9d 100644
> --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
> +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
> @@ -1123,6 +1123,8 @@ static int __maybe_unused dpu_runtime_resume(struct device *dev)
>
>  static const struct dev_pm_ops dpu_pm_ops = {
>         SET_RUNTIME_PM_OPS(dpu_runtime_suspend, dpu_runtime_resume, NULL)
> +       SET_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend,
> +                               pm_runtime_force_resume)
>  };
>
>  static const struct of_device_id dpu_dt_match[] = {
> diff --git a/drivers/gpu/drm/msm/dsi/dsi.c b/drivers/gpu/drm/msm/dsi/dsi.c
> index 55ea4bc2..62704885 100644
> --- a/drivers/gpu/drm/msm/dsi/dsi.c
> +++ b/drivers/gpu/drm/msm/dsi/dsi.c
> @@ -161,6 +161,8 @@ static int dsi_dev_remove(struct platform_device *pdev)
>
>  static const struct dev_pm_ops dsi_pm_ops = {
>         SET_RUNTIME_PM_OPS(msm_dsi_runtime_suspend, msm_dsi_runtime_resume, NULL)
> +       SET_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend,
> +                               pm_runtime_force_resume)
>  };
>
>  static struct platform_driver dsi_driver = {
> diff --git a/drivers/gpu/drm/msm/msm_drv.c b/drivers/gpu/drm/msm/msm_drv.c
> index 7d985f8..2b8c99c 100644
> --- a/drivers/gpu/drm/msm/msm_drv.c
> +++ b/drivers/gpu/drm/msm/msm_drv.c
> @@ -1051,6 +1051,8 @@ static int msm_pm_suspend(struct device *dev)
>                 return ret;
>         }
>
> +       pm_runtime_force_suspend(dev);

nit: check return value of pm_runtime_force_suspend()?


> +
>         return 0;
>  }
>
> @@ -1063,6 +1065,8 @@ static int msm_pm_resume(struct device *dev)
>         if (WARN_ON(!priv->pm_state))
>                 return -ENOENT;
>
> +       pm_runtime_force_resume(dev);

nit: check return value of pm_runtime_force_resume()?


-Doug

Re: [PATCH] drm/msm/dpu: ensure device suspend happens during PM sleep

[kalyan_t]

On 2020-03-31 21:30, Doug Anderson wrote:
> Hi,
> 
> On Tue, Mar 31, 2020 at 6:58 AM Kalyan Thota <kalyan_t@codeaurora.org> 
> wrote:
>> 
>> "The PM core always increments the runtime usage counter
>> before calling the ->suspend() callback and decrements it
>> after calling the ->resume() callback"
>> 
>> DPU and DSI are managed as runtime devices. When
>> suspend is triggered, PM core adds a refcount on all the
>> devices and calls device suspend, since usage count is
>> already incremented, runtime suspend was not getting called
>> and it kept the clocks on which resulted in target not
>> entering into XO shutdown.
>> 
>> Add changes to force suspend on runtime devices during pm sleep.
>> 
>> Changes in v1:
>>  - Remove unnecessary checks in the function
>>     _dpu_kms_disable_dpu (Rob Clark).
>> 
>> Changes in v2:
>>  - Avoid using suspend_late to reset the usagecount
>>    as suspend_late might not be called during suspend
>>    call failures (Doug).
>> 
>> Changes in v3:
>>  - Use force suspend instead of managing device usage_count
>>    via runtime put and get API's to trigger callbacks (Doug).
>> 
>> Signed-off-by: Kalyan Thota <kalyan_t@codeaurora.org>
>> ---
>>  drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c | 2 ++
>>  drivers/gpu/drm/msm/dsi/dsi.c           | 2 ++
>>  drivers/gpu/drm/msm/msm_drv.c           | 4 ++++
>>  3 files changed, 8 insertions(+)
> 
> This looks much saner to me.  Thanks!  I assume it still works fine
> for you?  I'm still no expert on how all the pieces of DRM drivers
> work together, but at least there's not a bunch of strange fiddling
> with pm_runtime state and hopefully it will avoid weird corner
> cases...
> 
--- Yes, verified the change on trogdor device, and display can suspend 
with the change.
> 
>> diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c 
>> b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
>> index ce19f1d..b886d9d 100644
>> --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
>> +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
>> @@ -1123,6 +1123,8 @@ static int __maybe_unused 
>> dpu_runtime_resume(struct device *dev)
>> 
>>  static const struct dev_pm_ops dpu_pm_ops = {
>>         SET_RUNTIME_PM_OPS(dpu_runtime_suspend, dpu_runtime_resume, 
>> NULL)
>> +       SET_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend,
>> +                               pm_runtime_force_resume)
>>  };
>> 
>>  static const struct of_device_id dpu_dt_match[] = {
>> diff --git a/drivers/gpu/drm/msm/dsi/dsi.c 
>> b/drivers/gpu/drm/msm/dsi/dsi.c
>> index 55ea4bc2..62704885 100644
>> --- a/drivers/gpu/drm/msm/dsi/dsi.c
>> +++ b/drivers/gpu/drm/msm/dsi/dsi.c
>> @@ -161,6 +161,8 @@ static int dsi_dev_remove(struct platform_device 
>> *pdev)
>> 
>>  static const struct dev_pm_ops dsi_pm_ops = {
>>         SET_RUNTIME_PM_OPS(msm_dsi_runtime_suspend, 
>> msm_dsi_runtime_resume, NULL)
>> +       SET_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend,
>> +                               pm_runtime_force_resume)
>>  };
>> 
>>  static struct platform_driver dsi_driver = {
>> diff --git a/drivers/gpu/drm/msm/msm_drv.c 
>> b/drivers/gpu/drm/msm/msm_drv.c
>> index 7d985f8..2b8c99c 100644
>> --- a/drivers/gpu/drm/msm/msm_drv.c
>> +++ b/drivers/gpu/drm/msm/msm_drv.c
>> @@ -1051,6 +1051,8 @@ static int msm_pm_suspend(struct device *dev)
>>                 return ret;
>>         }
>> 
>> +       pm_runtime_force_suspend(dev);
> 
> nit: check return value of pm_runtime_force_suspend()?
> 
> 
>> +
>>         return 0;
>>  }
>> 
>> @@ -1063,6 +1065,8 @@ static int msm_pm_resume(struct device *dev)
>>         if (WARN_ON(!priv->pm_state))
>>                 return -ENOENT;
>> 
>> +       pm_runtime_force_resume(dev);
> 
> nit: check return value of pm_runtime_force_resume()?
> 
> 
> -Doug

[PATCH] drm/msm/dpu: ensure device suspend happens during PM sleep

[Kalyan Thota]

"The PM core always increments the runtime usage counter
before calling the ->suspend() callback and decrements it
after calling the ->resume() callback"

DPU and DSI are managed as runtime devices. When
suspend is triggered, PM core adds a refcount on all the
devices and calls device suspend, since usage count is
already incremented, runtime suspend was not getting called
and it kept the clocks on which resulted in target not
entering into XO shutdown.

Add changes to force suspend on runtime devices during pm sleep.

Changes in v1:
 - Remove unnecessary checks in the function
    _dpu_kms_disable_dpu (Rob Clark).

Changes in v2:
 - Avoid using suspend_late to reset the usagecount
   as suspend_late might not be called during suspend
   call failures (Doug).

Changes in v3:
 - Use force suspend instead of managing device usage_count
   via runtime put and get API's to trigger callbacks (Doug).

Changes in v4:
 - Check the return values of pm_runtime_force_suspend and
   pm_runtime_force_resume API's and pass appropriately (Doug).

Signed-off-by: Kalyan Thota <kalyan_t@codeaurora.org>
---
 drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c |  2 ++
 drivers/gpu/drm/msm/dsi/dsi.c           |  2 ++
 drivers/gpu/drm/msm/msm_drv.c           | 14 +++++++++++++-
 3 files changed, 17 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
index ce19f1d..b886d9d 100644
--- a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
+++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
@@ -1123,6 +1123,8 @@ static int __maybe_unused dpu_runtime_resume(struct device *dev)
 
 static const struct dev_pm_ops dpu_pm_ops = {
 	SET_RUNTIME_PM_OPS(dpu_runtime_suspend, dpu_runtime_resume, NULL)
+	SET_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend,
+				pm_runtime_force_resume)
 };
 
 static const struct of_device_id dpu_dt_match[] = {
diff --git a/drivers/gpu/drm/msm/dsi/dsi.c b/drivers/gpu/drm/msm/dsi/dsi.c
index 55ea4bc2..62704885 100644
--- a/drivers/gpu/drm/msm/dsi/dsi.c
+++ b/drivers/gpu/drm/msm/dsi/dsi.c
@@ -161,6 +161,8 @@ static int dsi_dev_remove(struct platform_device *pdev)
 
 static const struct dev_pm_ops dsi_pm_ops = {
 	SET_RUNTIME_PM_OPS(msm_dsi_runtime_suspend, msm_dsi_runtime_resume, NULL)
+	SET_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend,
+				pm_runtime_force_resume)
 };
 
 static struct platform_driver dsi_driver = {
diff --git a/drivers/gpu/drm/msm/msm_drv.c b/drivers/gpu/drm/msm/msm_drv.c
index 7d985f8..4b93fc1 100644
--- a/drivers/gpu/drm/msm/msm_drv.c
+++ b/drivers/gpu/drm/msm/msm_drv.c
@@ -1040,6 +1040,7 @@ static int msm_pm_suspend(struct device *dev)
 {
 	struct drm_device *ddev = dev_get_drvdata(dev);
 	struct msm_drm_private *priv = ddev->dev_private;
+	int ret, rc;
 
 	if (WARN_ON(priv->pm_state))
 		drm_atomic_state_put(priv->pm_state);
@@ -1051,7 +1052,14 @@ static int msm_pm_suspend(struct device *dev)
 		return ret;
 	}
 
-	return 0;
+	ret = pm_runtime_force_suspend(dev);
+	if (ret) {
+		rc = drm_atomic_helper_resume(ddev, priv->pm_state);
+		if (!rc)
+			priv->pm_state = NULL;
+	}
+
+	return ret;
 }
 
 static int msm_pm_resume(struct device *dev)
@@ -1063,6 +1071,10 @@ static int msm_pm_resume(struct device *dev)
 	if (WARN_ON(!priv->pm_state))
 		return -ENOENT;
 
+	ret = pm_runtime_force_resume(dev);
+	if (ret)
+		return ret;
+
 	ret = drm_atomic_helper_resume(ddev, priv->pm_state);
 	if (!ret)
 		priv->pm_state = NULL;
-- 
1.9.1

Re: [PATCH] drm/msm/dpu: ensure device suspend happens during PM sleep

[Doug Anderson]

Hi,

On Fri, Apr 17, 2020 at 12:13 AM Kalyan Thota <kalyan_t@codeaurora.org> wrote:
>
> "The PM core always increments the runtime usage counter
> before calling the ->suspend() callback and decrements it
> after calling the ->resume() callback"
>
> DPU and DSI are managed as runtime devices. When
> suspend is triggered, PM core adds a refcount on all the
> devices and calls device suspend, since usage count is
> already incremented, runtime suspend was not getting called
> and it kept the clocks on which resulted in target not
> entering into XO shutdown.
>
> Add changes to force suspend on runtime devices during pm sleep.
>
> Changes in v1:
>  - Remove unnecessary checks in the function
>     _dpu_kms_disable_dpu (Rob Clark).
>
> Changes in v2:
>  - Avoid using suspend_late to reset the usagecount
>    as suspend_late might not be called during suspend
>    call failures (Doug).
>
> Changes in v3:
>  - Use force suspend instead of managing device usage_count
>    via runtime put and get API's to trigger callbacks (Doug).
>
> Changes in v4:
>  - Check the return values of pm_runtime_force_suspend and
>    pm_runtime_force_resume API's and pass appropriately (Doug).
>
> Signed-off-by: Kalyan Thota <kalyan_t@codeaurora.org>
> ---
>  drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c |  2 ++
>  drivers/gpu/drm/msm/dsi/dsi.c           |  2 ++
>  drivers/gpu/drm/msm/msm_drv.c           | 14 +++++++++++++-
>  3 files changed, 17 insertions(+), 1 deletion(-)

I am most certainly not an expert in this code, but as far as I can
tell it looks sane.  Thus:

Reviewed-by: Douglas Anderson <dianders@chromium.org>

-Doug

[PATCH] drm/msm/dpu: ensure device suspend happens during PM sleep

[Kalyan Thota]

"The PM core always increments the runtime usage counter
before calling the ->suspend() callback and decrements it
after calling the ->resume() callback"

DPU and DSI are managed as runtime devices. When
suspend is triggered, PM core adds a refcount on all the
devices and calls device suspend, since usage count is
already incremented, runtime suspend was not getting called
and it kept the clocks on which resulted in target not
entering into XO shutdown.

Add changes to force suspend on runtime devices during pm sleep.

Changes in v1:
 - Remove unnecessary checks in the function
    _dpu_kms_disable_dpu (Rob Clark).

Changes in v2:
 - Avoid using suspend_late to reset the usagecount
   as suspend_late might not be called during suspend
   call failures (Doug).

Changes in v3:
 - Use force suspend instead of managing device usage_count
   via runtime put and get API's to trigger callbacks (Doug).

Changes in v4:
 - Check the return values of pm_runtime_force_suspend and
   pm_runtime_force_resume API's and pass appropriately (Doug).

Changes in v5:
 - With v4 patch, test cycle has uncovered issues in device resume.

   On bubs: cmd tx failures were seen as SW is sending panel off
   commands when the dsi resources are turned off.

   Upon suspend, DRM driver will issue a NULL composition to the
   dpu, followed by turning off all the HW blocks.

   v5 changes will serialize the NULL commit and resource unwinding
   by handling them under PM prepare and PM complete phases there by
   ensuring that clks are on when panel off commands are being
   processed.

Signed-off-by: Kalyan Thota <kalyan_t@codeaurora.org>
---
 drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c |  2 ++
 drivers/gpu/drm/msm/dsi/dsi.c           |  2 ++
 drivers/gpu/drm/msm/msm_drv.c           | 20 ++++++++++++++++----
 3 files changed, 20 insertions(+), 4 deletions(-)

diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
index ce19f1d..b886d9d 100644
--- a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
+++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
@@ -1123,6 +1123,8 @@ static int __maybe_unused dpu_runtime_resume(struct device *dev)
 
 static const struct dev_pm_ops dpu_pm_ops = {
 	SET_RUNTIME_PM_OPS(dpu_runtime_suspend, dpu_runtime_resume, NULL)
+	SET_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend,
+				pm_runtime_force_resume)
 };
 
 static const struct of_device_id dpu_dt_match[] = {
diff --git a/drivers/gpu/drm/msm/dsi/dsi.c b/drivers/gpu/drm/msm/dsi/dsi.c
index 55ea4bc2..62704885 100644
--- a/drivers/gpu/drm/msm/dsi/dsi.c
+++ b/drivers/gpu/drm/msm/dsi/dsi.c
@@ -161,6 +161,8 @@ static int dsi_dev_remove(struct platform_device *pdev)
 
 static const struct dev_pm_ops dsi_pm_ops = {
 	SET_RUNTIME_PM_OPS(msm_dsi_runtime_suspend, msm_dsi_runtime_resume, NULL)
+	SET_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend,
+				pm_runtime_force_resume)
 };
 
 static struct platform_driver dsi_driver = {
diff --git a/drivers/gpu/drm/msm/msm_drv.c b/drivers/gpu/drm/msm/msm_drv.c
index 7d985f8..47d492b 100644
--- a/drivers/gpu/drm/msm/msm_drv.c
+++ b/drivers/gpu/drm/msm/msm_drv.c
@@ -1036,7 +1036,7 @@ static int msm_ioctl_submitqueue_close(struct drm_device *dev, void *data,
 };
 
 #ifdef CONFIG_PM_SLEEP
-static int msm_pm_suspend(struct device *dev)
+static int msm_pm_prepare(struct device *dev)
 {
 	struct drm_device *ddev = dev_get_drvdata(dev);
 	struct msm_drm_private *priv = ddev->dev_private;
@@ -1054,18 +1054,28 @@ static int msm_pm_suspend(struct device *dev)
 	return 0;
 }
 
-static int msm_pm_resume(struct device *dev)
+static void msm_pm_complete(struct device *dev)
 {
 	struct drm_device *ddev = dev_get_drvdata(dev);
 	struct msm_drm_private *priv = ddev->dev_private;
 	int ret;
 
 	if (WARN_ON(!priv->pm_state))
-		return -ENOENT;
+		return;
 
 	ret = drm_atomic_helper_resume(ddev, priv->pm_state);
 	if (!ret)
 		priv->pm_state = NULL;
+}
+
+static int msm_pm_resume(struct device *dev)
+{
+	int ret;
+
+	/* unwind runtime_disable called by force suspend */
+	pm_runtime_enable(dev);
+
+	ret = pm_runtime_resume(dev);
 
 	return ret;
 }
@@ -1102,8 +1112,10 @@ static int msm_runtime_resume(struct device *dev)
 #endif
 
 static const struct dev_pm_ops msm_pm_ops = {
-	SET_SYSTEM_SLEEP_PM_OPS(msm_pm_suspend, msm_pm_resume)
+	SET_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend, msm_pm_resume)
 	SET_RUNTIME_PM_OPS(msm_runtime_suspend, msm_runtime_resume, NULL)
+	.prepare = msm_pm_prepare,
+	.complete = msm_pm_complete,
 };
 
 /*
-- 
1.9.1

Re: [PATCH] drm/msm/dpu: ensure device suspend happens during PM sleep

[Doug Anderson]

Hi,

On Fri, May 1, 2020 at 6:31 AM Kalyan Thota <kalyan_t@codeaurora.org> wrote:
>
> "The PM core always increments the runtime usage counter
> before calling the ->suspend() callback and decrements it
> after calling the ->resume() callback"
>
> DPU and DSI are managed as runtime devices. When
> suspend is triggered, PM core adds a refcount on all the
> devices and calls device suspend, since usage count is
> already incremented, runtime suspend was not getting called
> and it kept the clocks on which resulted in target not
> entering into XO shutdown.
>
> Add changes to force suspend on runtime devices during pm sleep.
>
> Changes in v1:
>  - Remove unnecessary checks in the function
>     _dpu_kms_disable_dpu (Rob Clark).
>
> Changes in v2:
>  - Avoid using suspend_late to reset the usagecount
>    as suspend_late might not be called during suspend
>    call failures (Doug).
>
> Changes in v3:
>  - Use force suspend instead of managing device usage_count
>    via runtime put and get API's to trigger callbacks (Doug).
>
> Changes in v4:
>  - Check the return values of pm_runtime_force_suspend and
>    pm_runtime_force_resume API's and pass appropriately (Doug).
>
> Changes in v5:

Can you please put the version number properly in your subject?  It's
really hard to tell one version of your patch from another.


>  - With v4 patch, test cycle has uncovered issues in device resume.
>
>    On bubs: cmd tx failures were seen as SW is sending panel off
>    commands when the dsi resources are turned off.
>
>    Upon suspend, DRM driver will issue a NULL composition to the
>    dpu, followed by turning off all the HW blocks.
>
>    v5 changes will serialize the NULL commit and resource unwinding
>    by handling them under PM prepare and PM complete phases there by
>    ensuring that clks are on when panel off commands are being
>    processed.

I'm still most definitely not an expert in how all the DRM pieces all
hook up together, but the solution you have in this patch seems wrong
to me.  As far as I can tell the "prepare" state isn't supposed to be
actually doing the suspend work and here that's exactly what you're
doing.  I think you should find a different solution to ensure
ordering is correct.

-Doug