Skip to contentSkip to conversation

Optimize use of DBuf slices

PATCH In review Mahesh Kumar 4 participants 10 messages Apr 5, 2018 – Apr 25, 2018

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

cover Optimize use of DBuf slices

Patches in this series were originally part of series:
https://patchwork.freedesktop.org/series/36993/

Reposting it here after rebase

Mahesh Kumar (3):
drm/i915/icl: track dbuf slice-2 status
drm/i915/icl: Enable 2nd DBuf slice only when needed drm/i915/icl: update ddb entry start/end mask during hw ddb readout

 drivers/gpu/drm/i915/i915_drv.h         |  1 +
 drivers/gpu/drm/i915/i915_reg.h         |  3 ++
 drivers/gpu/drm/i915/intel_display.c    | 15 ++++++
 drivers/gpu/drm/i915/intel_drv.h        |  6 +++
 drivers/gpu/drm/i915/intel_pm.c         | 95 +++++++++++++++++++++++++++------
 drivers/gpu/drm/i915/intel_runtime_pm.c | 69 +++++++++++++++++++-----
 6 files changed, 160 insertions(+), 29 deletions(-)
Signature · 6 lines
2.16.2

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#1 lore raw

PATCH 1/3 drm/i915/icl: track dbuf slice-2 status

This patch adds support to start tracking status of DBUF slices. This is foundation to introduce support for enabling/disabling second DBUF slice dynamically for ICL.

---
 drivers/gpu/drm/i915/i915_drv.h         |  1 +
 drivers/gpu/drm/i915/intel_display.c    |  5 +++++
 drivers/gpu/drm/i915/intel_pm.c         | 20 ++++++++++++++++++++
 drivers/gpu/drm/i915/intel_runtime_pm.c |  4 ++++
 4 files changed, 30 insertions(+)
drivers/gpu/drm/i915/i915_drv.h +1 -0
index 5373b171bb96..e51da42297d8 100644
@@ -1184,6 +1184,7 @@ static inline bool skl_ddb_entry_equal(const struct skl_ddb_entry *e1,
 struct skl_ddb_allocation {
 	struct skl_ddb_entry plane[I915_MAX_PIPES][I915_MAX_PLANES]; /* packed/uv */
 	struct skl_ddb_entry y_plane[I915_MAX_PIPES][I915_MAX_PLANES];
+	uint8_t enabled_slices; /* GEN11 has configurable 2 slices */
 };
 
 struct skl_wm_values {
drivers/gpu/drm/i915/intel_display.c +5 -0
index 415fb8cf2cf4..96a1e6a7f69e 100644
@@ -11385,6 +11385,11 @@ static void verify_wm_state(struct drm_crtc *crtc,
 	skl_ddb_get_hw_state(dev_priv, &hw_ddb);
 	sw_ddb = &dev_priv->wm.skl_hw.ddb;
 
+	if (INTEL_GEN(dev_priv) >= 11)
+		if (hw_ddb.enabled_slices != sw_ddb->enabled_slices)
+			DRM_ERROR("mismatch in DBUF Slices (expected %u, got %u)\n",
+				  sw_ddb->enabled_slices,
+				  hw_ddb.enabled_slices);
 	/* planes */
 	for_each_universal_plane(dev_priv, pipe, plane) {
 		hw_plane_wm = &hw_wm.planes[plane];
drivers/gpu/drm/i915/intel_pm.c +20 -0
index 19e82aaa9863..3ff37d5ba353 100644
@@ -3567,6 +3567,23 @@ bool ilk_disable_lp_wm(struct drm_device *dev)
 	return _ilk_disable_lp_wm(dev_priv, WM_DIRTY_LP_ALL);
 }
 
+static uint8_t intel_enabled_dbuf_slices_num(struct drm_i915_private *dev_priv)
+{
+	uint8_t enabled_slices;
+
+	/* Slice 1 will always be enabled */
+	enabled_slices = 1;
+
+	/* Gen prior to GEN11 have only one DBuf slice */
+	if (INTEL_GEN(dev_priv) < 11)
+		return enabled_slices;
+
+	if (I915_READ(DBUF_CTL_S2) & DBUF_POWER_STATE)
+		enabled_slices++;
+
+	return enabled_slices;
+}
+
 /*
  * FIXME: We still don't have the proper code detect if we need to apply the WA,
  * so assume we'll always need it in order to avoid underruns.
@@ -3832,6 +3849,8 @@ void skl_ddb_get_hw_state(struct drm_i915_private *dev_priv,
 
 	memset(ddb, 0, sizeof(*ddb));
 
+	ddb->enabled_slices = intel_enabled_dbuf_slices_num(dev_priv);
+
 	for_each_intel_crtc(&dev_priv->drm, crtc) {
 		enum intel_display_power_domain power_domain;
 		enum plane_id plane_id;
@@ -5050,6 +5069,7 @@ skl_copy_wm_for_pipe(struct skl_wm_values *dst,
 	       sizeof(dst->ddb.y_plane[pipe]));
 	memcpy(dst->ddb.plane[pipe], src->ddb.plane[pipe],
 	       sizeof(dst->ddb.plane[pipe]));
+	dst->ddb.enabled_slices = src->ddb.enabled_slices;
 }
 
 static void
drivers/gpu/drm/i915/intel_runtime_pm.c +4 -0
index 53ea564f971e..58be542d660b 100644
@@ -2664,6 +2664,8 @@ static void icl_dbuf_enable(struct drm_i915_private *dev_priv)
 	if (!(I915_READ(DBUF_CTL_S1) & DBUF_POWER_STATE) ||
 	    !(I915_READ(DBUF_CTL_S2) & DBUF_POWER_STATE))
 		DRM_ERROR("DBuf power enable timeout\n");
+	else
+		dev_priv->wm.skl_hw.ddb.enabled_slices = 2;
 }
 
 static void icl_dbuf_disable(struct drm_i915_private *dev_priv)
@@ -2677,6 +2679,8 @@ static void icl_dbuf_disable(struct drm_i915_private *dev_priv)
 	if ((I915_READ(DBUF_CTL_S1) & DBUF_POWER_STATE) ||
 	    (I915_READ(DBUF_CTL_S2) & DBUF_POWER_STATE))
 		DRM_ERROR("DBuf power disable timeout!\n");
+	else
+		dev_priv->wm.skl_hw.ddb.enabled_slices = 0;
 }
 
 static void icl_mbus_init(struct drm_i915_private *dev_priv)
Signature · 6 lines
2.16.2

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
maintainer
#2 lore raw

Re: [PATCH 1/3] drm/i915/icl: track dbuf slice-2 status

Quoted 18 lines · Mahesh Kumar

This patch adds support to start tracking status of DBUF slices. This is foundation to introduce support for enabling/disabling second DBUF slice dynamically for ICL.

---
 drivers/gpu/drm/i915/i915_drv.h         |  1 +
 drivers/gpu/drm/i915/intel_display.c    |  5 +++++
 drivers/gpu/drm/i915/intel_pm.c         | 20 ++++++++++++++++++++
 drivers/gpu/drm/i915/intel_runtime_pm.c |  4 ++++
 4 files changed, 30 insertions(+)
drivers/gpu/drm/i915/i915_drv.h +0 -0
index 5373b171bb96..e51da42297d8 100644

drivers/gpu/drm/i915/i915_drv.h static inline bool skl_ddb_entry_equal(const struct skl_ddb_entry *e1, · line 1187

Quoted 4 lines
drivers/gpu/drm/i915/i915_drv.h +1 -0
@@ -1184,6 +1184,7 @@ static inline bool skl_ddb_entry_equal(const struct skl_ddb_entry *e1,
 struct skl_ddb_allocation {
 	struct skl_ddb_entry plane[I915_MAX_PIPES][I915_MAX_PLANES]; /* packed/uv */
 	struct skl_ddb_entry y_plane[I915_MAX_PIPES][I915_MAX_PLANES];
+	uint8_t enabled_slices; /* GEN11 has configurable 2 slices */

Please prefer kernel types u8/u16/u32/u64 over uint8_t and friends. It's fine to use the stdint types when the context uses them, but please let's not add new ones.

BR,
Jani.

Quoted 86 lines
drivers/gpu/drm/i915/i915_drv.h +0 -0
 };
 
 struct skl_wm_values {
drivers/gpu/drm/i915/intel_display.c +5 -0
index 415fb8cf2cf4..96a1e6a7f69e 100644
@@ -11385,6 +11385,11 @@ static void verify_wm_state(struct drm_crtc *crtc,
 	skl_ddb_get_hw_state(dev_priv, &hw_ddb);
 	sw_ddb = &dev_priv->wm.skl_hw.ddb;
 
+	if (INTEL_GEN(dev_priv) >= 11)
+		if (hw_ddb.enabled_slices != sw_ddb->enabled_slices)
+			DRM_ERROR("mismatch in DBUF Slices (expected %u, got %u)\n",
+				  sw_ddb->enabled_slices,
+				  hw_ddb.enabled_slices);
 	/* planes */
 	for_each_universal_plane(dev_priv, pipe, plane) {
 		hw_plane_wm = &hw_wm.planes[plane];
drivers/gpu/drm/i915/intel_pm.c +20 -0
index 19e82aaa9863..3ff37d5ba353 100644
@@ -3567,6 +3567,23 @@ bool ilk_disable_lp_wm(struct drm_device *dev)
 	return _ilk_disable_lp_wm(dev_priv, WM_DIRTY_LP_ALL);
 }
 
+static uint8_t intel_enabled_dbuf_slices_num(struct drm_i915_private *dev_priv)
+{
+	uint8_t enabled_slices;
+
+	/* Slice 1 will always be enabled */
+	enabled_slices = 1;
+
+	/* Gen prior to GEN11 have only one DBuf slice */
+	if (INTEL_GEN(dev_priv) < 11)
+		return enabled_slices;
+
+	if (I915_READ(DBUF_CTL_S2) & DBUF_POWER_STATE)
+		enabled_slices++;
+
+	return enabled_slices;
+}
+
 /*
  * FIXME: We still don't have the proper code detect if we need to apply the WA,
  * so assume we'll always need it in order to avoid underruns.
@@ -3832,6 +3849,8 @@ void skl_ddb_get_hw_state(struct drm_i915_private *dev_priv,
 
 	memset(ddb, 0, sizeof(*ddb));
 
+	ddb->enabled_slices = intel_enabled_dbuf_slices_num(dev_priv);
+
 	for_each_intel_crtc(&dev_priv->drm, crtc) {
 		enum intel_display_power_domain power_domain;
 		enum plane_id plane_id;
@@ -5050,6 +5069,7 @@ skl_copy_wm_for_pipe(struct skl_wm_values *dst,
 	       sizeof(dst->ddb.y_plane[pipe]));
 	memcpy(dst->ddb.plane[pipe], src->ddb.plane[pipe],
 	       sizeof(dst->ddb.plane[pipe]));
+	dst->ddb.enabled_slices = src->ddb.enabled_slices;
 }
 
 static void
drivers/gpu/drm/i915/intel_runtime_pm.c +4 -0
index 53ea564f971e..58be542d660b 100644
@@ -2664,6 +2664,8 @@ static void icl_dbuf_enable(struct drm_i915_private *dev_priv)
 	if (!(I915_READ(DBUF_CTL_S1) & DBUF_POWER_STATE) ||
 	    !(I915_READ(DBUF_CTL_S2) & DBUF_POWER_STATE))
 		DRM_ERROR("DBuf power enable timeout\n");
+	else
+		dev_priv->wm.skl_hw.ddb.enabled_slices = 2;
 }
 
 static void icl_dbuf_disable(struct drm_i915_private *dev_priv)
@@ -2677,6 +2679,8 @@ static void icl_dbuf_disable(struct drm_i915_private *dev_priv)
 	if ((I915_READ(DBUF_CTL_S1) & DBUF_POWER_STATE) ||
 	    (I915_READ(DBUF_CTL_S2) & DBUF_POWER_STATE))
 		DRM_ERROR("DBuf power disable timeout!\n");
+	else
+		dev_priv->wm.skl_hw.ddb.enabled_slices = 0;
 }
 
 static void icl_mbus_init(struct drm_i915_private *dev_priv)
Signature · 5 lines
Jani Nikula, Intel Open Source Technology Center
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#3 lore raw

Re: [PATCH 1/3] drm/i915/icl: track dbuf slice-2 status

Hi,

Quoted 14 lines · Jani Nikula
Quoted 13 lines · Mahesh Kumar

This patch adds support to start tracking status of DBUF slices. This is foundation to introduce support for enabling/disabling second DBUF slice dynamically for ICL.

---
  drivers/gpu/drm/i915/i915_drv.h         |  1 +
  drivers/gpu/drm/i915/intel_display.c    |  5 +++++
  drivers/gpu/drm/i915/intel_pm.c         | 20 ++++++++++++++++++++
  drivers/gpu/drm/i915/intel_runtime_pm.c |  4 ++++
  4 files changed, 30 insertions(+)
Quoted 12 lines

drivers/gpu/drm/i915/i915_drv.h static inline bool skl_ddb_entry_equal(const struct skl_ddb_entry *e1, · line 1187

Quoted 9 lines
drivers/gpu/drm/i915/i915_drv.h +1 -0
index 5373b171bb96..e51da42297d8 100644
@@ -1184,6 +1184,7 @@ static inline bool skl_ddb_entry_equal(const struct skl_ddb_entry *e1,
  struct skl_ddb_allocation {
  	struct skl_ddb_entry plane[I915_MAX_PIPES][I915_MAX_PLANES]; /* packed/uv */
  	struct skl_ddb_entry y_plane[I915_MAX_PIPES][I915_MAX_PLANES];
+	uint8_t enabled_slices; /* GEN11 has configurable 2 slices */

Please prefer kernel types u8/u16/u32/u64 over uint8_t and friends. It's fine to use the stdint types when the context uses them, but please let's not add new ones.

ok, sure will update in new version
-Mahesh

Quoted 90 lines

BR,
Jani.

Quoted 85 lines
drivers/gpu/drm/i915/i915_drv.h +0 -0
  };
 
  struct skl_wm_values {
drivers/gpu/drm/i915/intel_display.c +5 -0
index 415fb8cf2cf4..96a1e6a7f69e 100644
@@ -11385,6 +11385,11 @@ static void verify_wm_state(struct drm_crtc *crtc,
  	skl_ddb_get_hw_state(dev_priv, &hw_ddb);
  	sw_ddb = &dev_priv->wm.skl_hw.ddb;
 
+	if (INTEL_GEN(dev_priv) >= 11)
+		if (hw_ddb.enabled_slices != sw_ddb->enabled_slices)
+			DRM_ERROR("mismatch in DBUF Slices (expected %u, got %u)\n",
+				  sw_ddb->enabled_slices,
+				  hw_ddb.enabled_slices);
  	/* planes */
  	for_each_universal_plane(dev_priv, pipe, plane) {
  		hw_plane_wm = &hw_wm.planes[plane];
drivers/gpu/drm/i915/intel_pm.c +20 -0
index 19e82aaa9863..3ff37d5ba353 100644
@@ -3567,6 +3567,23 @@ bool ilk_disable_lp_wm(struct drm_device *dev)
  	return _ilk_disable_lp_wm(dev_priv, WM_DIRTY_LP_ALL);
  }
 
+static uint8_t intel_enabled_dbuf_slices_num(struct drm_i915_private *dev_priv)
+{
+	uint8_t enabled_slices;
+
+	/* Slice 1 will always be enabled */
+	enabled_slices = 1;
+
+	/* Gen prior to GEN11 have only one DBuf slice */
+	if (INTEL_GEN(dev_priv) < 11)
+		return enabled_slices;
+
+	if (I915_READ(DBUF_CTL_S2) & DBUF_POWER_STATE)
+		enabled_slices++;
+
+	return enabled_slices;
+}
+
  /*
   * FIXME: We still don't have the proper code detect if we need to apply the WA,
   * so assume we'll always need it in order to avoid underruns.
@@ -3832,6 +3849,8 @@ void skl_ddb_get_hw_state(struct drm_i915_private *dev_priv,
 
  	memset(ddb, 0, sizeof(*ddb));
 
+	ddb->enabled_slices = intel_enabled_dbuf_slices_num(dev_priv);
+
  	for_each_intel_crtc(&dev_priv->drm, crtc) {
  		enum intel_display_power_domain power_domain;
  		enum plane_id plane_id;
@@ -5050,6 +5069,7 @@ skl_copy_wm_for_pipe(struct skl_wm_values *dst,
  	       sizeof(dst->ddb.y_plane[pipe]));
  	memcpy(dst->ddb.plane[pipe], src->ddb.plane[pipe],
  	       sizeof(dst->ddb.plane[pipe]));
+	dst->ddb.enabled_slices = src->ddb.enabled_slices;
  }
 
  static void
drivers/gpu/drm/i915/intel_runtime_pm.c +4 -0
index 53ea564f971e..58be542d660b 100644
@@ -2664,6 +2664,8 @@ static void icl_dbuf_enable(struct drm_i915_private *dev_priv)
  	if (!(I915_READ(DBUF_CTL_S1) & DBUF_POWER_STATE) ||
  	    !(I915_READ(DBUF_CTL_S2) & DBUF_POWER_STATE))
  		DRM_ERROR("DBuf power enable timeout\n");
+	else
+		dev_priv->wm.skl_hw.ddb.enabled_slices = 2;
  }
 
  static void icl_dbuf_disable(struct drm_i915_private *dev_priv)
@@ -2677,6 +2679,8 @@ static void icl_dbuf_disable(struct drm_i915_private *dev_priv)
  	if ((I915_READ(DBUF_CTL_S1) & DBUF_POWER_STATE) ||
  	    (I915_READ(DBUF_CTL_S2) & DBUF_POWER_STATE))
  		DRM_ERROR("DBuf power disable timeout!\n");
+	else
+		dev_priv->wm.skl_hw.ddb.enabled_slices = 0;
  }
 
Quoted 1 line

drivers/gpu/drm/i915/intel_runtime_pm.c static void icl_dbuf_disable(struct drm_i915_private *dev_priv) · line 2686

Quoted 1 line
drivers/gpu/drm/i915/intel_runtime_pm.c +0 -0
  static void icl_mbus_init(struct drm_i915_private *dev_priv)

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#4 lore raw

PATCH 2/3 drm/i915/icl: Enable 2nd DBuf slice only when needed

In reply to Mahesh Kumar

ICL has two slices of DBuf, each slice of size 1024 blocks. We should not always enable slice-2. It should be enabled only if display total required BW is > 12GBps OR more than 1 pipes are enabled.

Changes since V1:
- typecast total_data_rate to u64 before multiplication to solve any possible overflow (Rodrigo)
- fix where skl_wm_get_hw_state was memsetting ddb, resulting enabled_slices to become zero
- Fix the logic of calculating ddb_size
Changes since V2:
- If no-crtc is part of commit required_slices will have value "0", don't try to disable DBuf slice.
Changes since V3:
- Create a generic helper to enable/disable slice
- don't return early if total_data_rate is 0, it may be cursor only commit, or atomic modeset without any plane.

---
 drivers/gpu/drm/i915/intel_display.c    | 10 +++++
 drivers/gpu/drm/i915/intel_drv.h        |  6 +++
 drivers/gpu/drm/i915/intel_pm.c         | 57 +++++++++++++++++++++++------
 drivers/gpu/drm/i915/intel_runtime_pm.c | 65 ++++++++++++++++++++++++++-------
 4 files changed, 113 insertions(+), 25 deletions(-)
drivers/gpu/drm/i915/intel_display.c +10 -0
index 96a1e6a7f69e..bc795307657a 100644
@@ -12196,6 +12196,8 @@ static void skl_update_crtcs(struct drm_atomic_state *state)
 	bool progress;
 	enum pipe pipe;
 	int i;
+	uint8_t hw_enabled_slices = dev_priv->wm.skl_hw.ddb.enabled_slices;
+	uint8_t required_slices = intel_state->wm_results.ddb.enabled_slices;
 
 	const struct skl_ddb_entry *entries[I915_MAX_PIPES] = {};
 
@@ -12204,6 +12206,10 @@ static void skl_update_crtcs(struct drm_atomic_state *state)
 		if (new_crtc_state->active)
 			entries[i] = &to_intel_crtc_state(old_crtc_state)->wm.skl.ddb;
 
+	/* If 2nd DBuf slice required, enable it here */
+	if (INTEL_GEN(dev_priv) >= 11 && required_slices > hw_enabled_slices)
+		icl_dbuf_slices_update(dev_priv, required_slices);
+
 	/*
 	 * Whenever the number of active pipes changes, we need to make sure we
 	 * update the pipes in the right order so that their ddb allocations
@@ -12254,6 +12260,10 @@ static void skl_update_crtcs(struct drm_atomic_state *state)
 			progress = true;
 		}
 	} while (progress);
+
+	/* If 2nd DBuf slice is no more required disable it */
+	if (INTEL_GEN(dev_priv) >= 11 && required_slices < hw_enabled_slices)
+		icl_dbuf_slices_update(dev_priv, required_slices);
 }
 
 static void intel_atomic_helper_free_state(struct drm_i915_private *dev_priv)
drivers/gpu/drm/i915/intel_drv.h +6 -0
index d1452fd2a58d..473a1ad7f32b 100644
@@ -140,6 +140,10 @@
 #define KHz(x) (1000 * (x))
 #define MHz(x) KHz(1000 * (x))
 
+#define KBps(x) (1000 * (x))
+#define MBps(x) KBps(1000 * (x))
+#define GBps(x) ((uint64_t) 1000 * MBps((x)))
+
 /*
  * Display related stuff
  */
@@ -1914,6 +1918,8 @@ bool intel_display_power_get_if_enabled(struct drm_i915_private *dev_priv,
 					enum intel_display_power_domain domain);
 void intel_display_power_put(struct drm_i915_private *dev_priv,
 			     enum intel_display_power_domain domain);
+void icl_dbuf_slices_update(struct drm_i915_private *dev_priv,
+			    uint8_t req_slices);
 
 static inline void
 assert_rpm_device_not_suspended(struct drm_i915_private *dev_priv)
drivers/gpu/drm/i915/intel_pm.c +46 -11

90 lines not on this page · raw message

drivers/gpu/drm/i915/intel_runtime_pm.c +51 -14

83 lines not on this page · raw message

Signature · 6 lines
2.16.2

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
maintainer Reviewed-by on v1 2/3
#5 lore raw

Re: [PATCH 2/3] drm/i915/icl: Enable 2nd DBuf slice only when needed

Quoted 31 lines · Mahesh Kumar

ICL has two slices of DBuf, each slice of size 1024 blocks. We should not always enable slice-2. It should be enabled only if display total required BW is > 12GBps OR more than 1 pipes are enabled.

Changes since V1:
- typecast total_data_rate to u64 before multiplication to solve any possible overflow (Rodrigo)
- fix where skl_wm_get_hw_state was memsetting ddb, resulting enabled_slices to become zero
- Fix the logic of calculating ddb_size
Changes since V2:
- If no-crtc is part of commit required_slices will have value "0", don't try to disable DBuf slice.
Changes since V3:
- Create a generic helper to enable/disable slice
- don't return early if total_data_rate is 0, it may be cursor only commit, or atomic modeset without any plane.

---
 drivers/gpu/drm/i915/intel_display.c    | 10 +++++
 drivers/gpu/drm/i915/intel_drv.h        |  6 +++
 drivers/gpu/drm/i915/intel_pm.c         | 57 +++++++++++++++++++++++------
 drivers/gpu/drm/i915/intel_runtime_pm.c | 65 ++++++++++++++++++++++++++-------
 4 files changed, 113 insertions(+), 25 deletions(-)
drivers/gpu/drm/i915/intel_display.c +0 -0
index 96a1e6a7f69e..bc795307657a 100644

drivers/gpu/drm/i915/intel_display.c static void skl_update_crtcs(struct drm_atomic_state *state) · line 12200

Quoted 5 lines
drivers/gpu/drm/i915/intel_display.c +2 -0
@@ -12196,6 +12196,8 @@ static void skl_update_crtcs(struct drm_atomic_state *state)
 	bool progress;
 	enum pipe pipe;
 	int i;
+	uint8_t hw_enabled_slices = dev_priv->wm.skl_hw.ddb.enabled_slices;
+	uint8_t required_slices = intel_state->wm_results.ddb.enabled_slices;

in many parts this enabled_slices meaning required or calculated confused me a bit, but I don't have better suggestion and the code seems right:

Quoted 243 lines
drivers/gpu/drm/i915/intel_display.c +8 -0
 	const struct skl_ddb_entry *entries[I915_MAX_PIPES] = {};
 
@@ -12204,6 +12206,10 @@ static void skl_update_crtcs(struct drm_atomic_state *state)
 		if (new_crtc_state->active)
 			entries[i] = &to_intel_crtc_state(old_crtc_state)->wm.skl.ddb;
 
+	/* If 2nd DBuf slice required, enable it here */
+	if (INTEL_GEN(dev_priv) >= 11 && required_slices > hw_enabled_slices)
+		icl_dbuf_slices_update(dev_priv, required_slices);
+
 	/*
 	 * Whenever the number of active pipes changes, we need to make sure we
 	 * update the pipes in the right order so that their ddb allocations
@@ -12254,6 +12260,10 @@ static void skl_update_crtcs(struct drm_atomic_state *state)
 			progress = true;
 		}
 	} while (progress);
+
+	/* If 2nd DBuf slice is no more required disable it */
+	if (INTEL_GEN(dev_priv) >= 11 && required_slices < hw_enabled_slices)
+		icl_dbuf_slices_update(dev_priv, required_slices);
 }
 
 static void intel_atomic_helper_free_state(struct drm_i915_private *dev_priv)
drivers/gpu/drm/i915/intel_drv.h +6 -0
index d1452fd2a58d..473a1ad7f32b 100644
@@ -140,6 +140,10 @@
 #define KHz(x) (1000 * (x))
 #define MHz(x) KHz(1000 * (x))
 
+#define KBps(x) (1000 * (x))
+#define MBps(x) KBps(1000 * (x))
+#define GBps(x) ((uint64_t) 1000 * MBps((x)))
+
 /*
  * Display related stuff
  */
@@ -1914,6 +1918,8 @@ bool intel_display_power_get_if_enabled(struct drm_i915_private *dev_priv,
 					enum intel_display_power_domain domain);
 void intel_display_power_put(struct drm_i915_private *dev_priv,
 			     enum intel_display_power_domain domain);
+void icl_dbuf_slices_update(struct drm_i915_private *dev_priv,
+			    uint8_t req_slices);
 
 static inline void
 assert_rpm_device_not_suspended(struct drm_i915_private *dev_priv)
drivers/gpu/drm/i915/intel_pm.c +46 -11
index 3ff37d5ba353..caa29f949335 100644
@@ -3771,9 +3771,42 @@ bool intel_can_enable_sagv(struct drm_atomic_state *state)
 	return true;
 }
 
+static unsigned int intel_get_ddb_size(struct drm_i915_private *dev_priv,
+				       const struct intel_crtc_state *cstate,
+				       const unsigned int total_data_rate,
+				       const int num_active,
+				       struct skl_ddb_allocation *ddb)
+{
+	const struct drm_display_mode *adjusted_mode;
+	u64 total_data_bw;
+	u16 ddb_size = INTEL_INFO(dev_priv)->ddb_size;
+
+	WARN_ON(ddb_size == 0);
+
+	if (INTEL_GEN(dev_priv) < 11)
+		return ddb_size - 4; /* 4 blocks for bypass path allocation */
+
+	adjusted_mode = &cstate->base.adjusted_mode;
+	total_data_bw = (u64)total_data_rate * drm_mode_vrefresh(adjusted_mode);
+
+	/*
+	 * 12GB/s is maximum BW supported by single DBuf slice.
+	 */
+	if (total_data_bw >= GBps(12) || num_active > 1)
+		ddb->enabled_slices = 2;
+	else {
+		ddb->enabled_slices = 1;
+		ddb_size /= 2;
+	}
+
+	return ddb_size;
+}
+
 static void
 skl_ddb_get_pipe_allocation_limits(struct drm_device *dev,
 				   const struct intel_crtc_state *cstate,
+				   const unsigned int total_data_rate,
+				   struct skl_ddb_allocation *ddb,
 				   struct skl_ddb_entry *alloc, /* out */
 				   int *num_active /* out */)
 {
@@ -3796,11 +3829,8 @@ skl_ddb_get_pipe_allocation_limits(struct drm_device *dev,
 	else
 		*num_active = hweight32(dev_priv->active_crtcs);
 
-	ddb_size = INTEL_INFO(dev_priv)->ddb_size;
-	WARN_ON(ddb_size == 0);
-
-	if (INTEL_GEN(dev_priv) < 11)
-		ddb_size -= 4; /* 4 blocks for bypass path allocation */
+	ddb_size = intel_get_ddb_size(dev_priv, cstate, total_data_rate,
+				      *num_active, ddb);
 
 	/*
 	 * If the state doesn't change the active CRTC's, then there's
@@ -4239,7 +4269,11 @@ skl_allocate_pipe_ddb(struct intel_crtc_state *cstate,
 		return 0;
 	}
 
-	skl_ddb_get_pipe_allocation_limits(dev, cstate, alloc, &num_active);
+	total_data_rate = skl_get_total_relative_data_rate(cstate,
+							   plane_data_rate,
+							   plane_y_data_rate);
+	skl_ddb_get_pipe_allocation_limits(dev, cstate, total_data_rate, ddb,
+					   alloc, &num_active);
 	alloc_size = skl_ddb_entry_size(alloc);
 	if (alloc_size == 0)
 		return 0;
@@ -4274,9 +4308,6 @@ skl_allocate_pipe_ddb(struct intel_crtc_state *cstate,
 	 *
 	 * FIXME: we may not allocate every single block here.
 	 */
-	total_data_rate = skl_get_total_relative_data_rate(cstate,
-							   plane_data_rate,
-							   plane_y_data_rate);
 	if (total_data_rate == 0)
 		return 0;
 
@@ -5382,8 +5413,12 @@ void skl_wm_get_hw_state(struct drm_device *dev)
 		/* Fully recompute DDB on first atomic commit */
 		dev_priv->wm.distrust_bios_wm = true;
 	} else {
-		/* Easy/common case; just sanitize DDB now if everything off */
-		memset(ddb, 0, sizeof(*ddb));
+		/*
+		 * Easy/common case; just sanitize DDB now if everything off
+		 * Keep dbuf slice info intact
+		 */
+		memset(ddb->plane, 0, sizeof(ddb->plane));
+		memset(ddb->y_plane, 0, sizeof(ddb->y_plane));
 	}
 }
 
drivers/gpu/drm/i915/intel_runtime_pm.c +51 -14
index 58be542d660b..56cdec06e6b2 100644
@@ -2627,32 +2627,69 @@ static void intel_power_domains_sync_hw(struct drm_i915_private *dev_priv)
 	mutex_unlock(&power_domains->lock);
 }
 
-static void gen9_dbuf_enable(struct drm_i915_private *dev_priv)
+static inline
+bool intel_dbuf_slice_set(struct drm_i915_private *dev_priv,
+			  i915_reg_t reg, bool enable)
 {
-	I915_WRITE(DBUF_CTL, I915_READ(DBUF_CTL) | DBUF_POWER_REQUEST);
-	POSTING_READ(DBUF_CTL);
+	u32 val, status;
 
+	val = I915_READ(reg);
+	val = enable ? (val | DBUF_POWER_REQUEST) : (val & ~DBUF_POWER_REQUEST);
+	I915_WRITE(reg, val);
+	POSTING_READ(reg);
 	udelay(10);
 
-	if (!(I915_READ(DBUF_CTL) & DBUF_POWER_STATE))
-		DRM_ERROR("DBuf power enable timeout\n");
+	status = I915_READ(reg) & DBUF_POWER_STATE;
+	if ((enable && !status) || (!enable && status)) {
+		DRM_ERROR("DBus power %s timeout!\n",
+					enable ? "enable" : "disable");
+		return false;
+	}
+	return true;
+}
+
+static void gen9_dbuf_enable(struct drm_i915_private *dev_priv)
+{
+	intel_dbuf_slice_set(dev_priv, DBUF_CTL, true);
 }
 
 static void gen9_dbuf_disable(struct drm_i915_private *dev_priv)
 {
-	I915_WRITE(DBUF_CTL, I915_READ(DBUF_CTL) & ~DBUF_POWER_REQUEST);
-	POSTING_READ(DBUF_CTL);
+	intel_dbuf_slice_set(dev_priv, DBUF_CTL, false);
+}
 
-	udelay(10);
+static uint8_t intel_dbuf_max_slices(struct drm_i915_private *dev_priv)
+{
+	if (INTEL_GEN(dev_priv) < 11)
+		return 1;
+	return 2;
+}
 
-	if (I915_READ(DBUF_CTL) & DBUF_POWER_STATE)
-		DRM_ERROR("DBuf power disable timeout!\n");
+void icl_dbuf_slices_update(struct drm_i915_private *dev_priv,
+			    uint8_t req_slices)
+{
+	uint8_t hw_enabled_slices = dev_priv->wm.skl_hw.ddb.enabled_slices;
+	u32 val;
+	bool ret;
+
+	if (req_slices > intel_dbuf_max_slices(dev_priv)) {
+		DRM_ERROR("Invalid number of dbuf slices requested\n");
+		return;
+	}
+
+	if (req_slices == hw_enabled_slices || req_slices == 0)
+		return;
+
+	val = I915_READ(DBUF_CTL_S2);
+	if (req_slices > hw_enabled_slices)
+		ret = intel_dbuf_slice_set(dev_priv, DBUF_CTL_S2, true);
+	else
+		ret = intel_dbuf_slice_set(dev_priv, DBUF_CTL_S2, false);
+
+	if (ret)
+		dev_priv->wm.skl_hw.ddb.enabled_slices = req_slices;
 }
 
-/*
- * TODO: we shouldn't always enable DBUF_CTL_S2, we should only enable it when
- * needed and keep it disabled as much as possible.
- */
 static void icl_dbuf_enable(struct drm_i915_private *dev_priv)
 {
 	I915_WRITE(DBUF_CTL_S1, I915_READ(DBUF_CTL_S1) | DBUF_POWER_REQUEST);
Signature · 6 lines
2.16.2

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#6 lore raw

PATCH 3/3 drm/i915/icl: update ddb entry start/end mask during hw ddb readout

In reply to Mahesh Kumar

Gen11/ICL onward ddb entry start/end mask is increased from 10 bits to 11 bits. This patch make changes to use proper mask for ICL+ during hardware ddb value readout.

Changes since V1:
- Use _MASK & _SHIFT macro (James)

---
 drivers/gpu/drm/i915/i915_reg.h |  3 +++
 drivers/gpu/drm/i915/intel_pm.c | 18 ++++++++++++++----
 2 files changed, 17 insertions(+), 4 deletions(-)
drivers/gpu/drm/i915/i915_reg.h +3 -0
index 176dca6554f4..e3a6c535617d 100644
@@ -6459,6 +6459,9 @@ enum {
 
 #define _PLANE_BUF_CFG_1_B			0x7127c
 #define _PLANE_BUF_CFG_2_B			0x7137c
+#define  SKL_DDB_ENTRY_MASK			0x3FF
+#define  ICL_DDB_ENTRY_MASK			0x7FF
+#define  DDB_ENTRY_END_SHIFT			16
 #define _PLANE_BUF_CFG_1(pipe)	\
 	_PIPE(pipe, _PLANE_BUF_CFG_1_A, _PLANE_BUF_CFG_1_B)
 #define _PLANE_BUF_CFG_2(pipe)	\
drivers/gpu/drm/i915/intel_pm.c +14 -4
index caa29f949335..98e91f4a5ab4 100644
@@ -3864,10 +3864,18 @@ static unsigned int skl_cursor_allocation(int num_active)
 	return 8;
 }
 
-static void skl_ddb_entry_init_from_hw(struct skl_ddb_entry *entry, u32 reg)
+static void skl_ddb_entry_init_from_hw(struct drm_i915_private *dev_priv,
+				       struct skl_ddb_entry *entry, u32 reg)
 {
-	entry->start = reg & 0x3ff;
-	entry->end = (reg >> 16) & 0x3ff;
+	uint16_t mask;
+
+	if (INTEL_GEN(dev_priv) >= 11)
+		mask = ICL_DDB_ENTRY_MASK;
+	else
+		mask = SKL_DDB_ENTRY_MASK;
+	entry->start = reg & mask;
+	entry->end = (reg >> DDB_ENTRY_END_SHIFT) & mask;
+
 	if (entry->end)
 		entry->end += 1;
 }
@@ -3898,7 +3906,9 @@ void skl_ddb_get_hw_state(struct drm_i915_private *dev_priv,
 			else
 				val = I915_READ(CUR_BUF_CFG(pipe));
 
-			skl_ddb_entry_init_from_hw(&ddb->plane[pipe][plane_id], val);
+			skl_ddb_entry_init_from_hw(dev_priv,
+						   &ddb->plane[pipe][plane_id],
+						   val);
 		}
 
 		intel_display_power_put(dev_priv, power_domain);
Signature · 6 lines
2.16.2

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
maintainer Reviewed-by on v1 3/3
#7 lore raw

Re: [PATCH 3/3] drm/i915/icl: update ddb entry start/end mask during hw ddb readout

Quoted 8 lines · Mahesh Kumar

Gen11/ICL onward ddb entry start/end mask is increased from 10 bits to 11 bits. This patch make changes to use proper mask for ICL+ during hardware ddb value readout.

Changes since V1:
- Use _MASK & _SHIFT macro (James)

Quoted 63 lines
---
 drivers/gpu/drm/i915/i915_reg.h |  3 +++
 drivers/gpu/drm/i915/intel_pm.c | 18 ++++++++++++++----
 2 files changed, 17 insertions(+), 4 deletions(-)
drivers/gpu/drm/i915/i915_reg.h +3 -0
index 176dca6554f4..e3a6c535617d 100644
@@ -6459,6 +6459,9 @@ enum {
 
 #define _PLANE_BUF_CFG_1_B			0x7127c
 #define _PLANE_BUF_CFG_2_B			0x7137c
+#define  SKL_DDB_ENTRY_MASK			0x3FF
+#define  ICL_DDB_ENTRY_MASK			0x7FF
+#define  DDB_ENTRY_END_SHIFT			16
 #define _PLANE_BUF_CFG_1(pipe)	\
 	_PIPE(pipe, _PLANE_BUF_CFG_1_A, _PLANE_BUF_CFG_1_B)
 #define _PLANE_BUF_CFG_2(pipe)	\
drivers/gpu/drm/i915/intel_pm.c +14 -4
index caa29f949335..98e91f4a5ab4 100644
@@ -3864,10 +3864,18 @@ static unsigned int skl_cursor_allocation(int num_active)
 	return 8;
 }
 
-static void skl_ddb_entry_init_from_hw(struct skl_ddb_entry *entry, u32 reg)
+static void skl_ddb_entry_init_from_hw(struct drm_i915_private *dev_priv,
+				       struct skl_ddb_entry *entry, u32 reg)
 {
-	entry->start = reg & 0x3ff;
-	entry->end = (reg >> 16) & 0x3ff;
+	uint16_t mask;
+
+	if (INTEL_GEN(dev_priv) >= 11)
+		mask = ICL_DDB_ENTRY_MASK;
+	else
+		mask = SKL_DDB_ENTRY_MASK;
+	entry->start = reg & mask;
+	entry->end = (reg >> DDB_ENTRY_END_SHIFT) & mask;
+
 	if (entry->end)
 		entry->end += 1;
 }
@@ -3898,7 +3906,9 @@ void skl_ddb_get_hw_state(struct drm_i915_private *dev_priv,
 			else
 				val = I915_READ(CUR_BUF_CFG(pipe));
 
-			skl_ddb_entry_init_from_hw(&ddb->plane[pipe][plane_id], val);
+			skl_ddb_entry_init_from_hw(dev_priv,
+						   &ddb->plane[pipe][plane_id],
+						   val);
 		}
 
 		intel_display_power_put(dev_priv, power_domain);
Signature · 6 lines
2.16.2

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#8 lore raw

✗ Fi.CI.CHECKPATCH: warning for Optimize use of DBuf slices

In reply to Mahesh Kumar
== Series Details ==

== Series Details ==

Series: Optimize use of DBuf slices
URL : https://patchwork.freedesktop.org/series/41180/
State : warning

== Summary ==

$ dim checkpatch origin/drm-tip
e2b3a175155d drm/i915/icl: track dbuf slice-2 status
5cc032e8d384 drm/i915/icl: Enable 2nd DBuf slice only when needed
-:71: CHECK:SPACING: No space is necessary after a cast
#71: FILE: drivers/gpu/drm/i915/intel_drv.h:145:
+#define GBps(x) ((uint64_t) 1000 * MBps((x)))

-:114: CHECK:BRACES: braces {} should be used on all arms of this statement #114: FILE: drivers/gpu/drm/i915/intel_pm.c:3795:
+ if (total_data_bw >= GBps(12) || num_active > 1)
[...]
+ else {
[...]

-:116: CHECK:BRACES: Unbalanced braces around else statement #116: FILE: drivers/gpu/drm/i915/intel_pm.c:3797:
+ else {

-:212: CHECK:PARENTHESIS_ALIGNMENT: Alignment should match open parenthesis #212: FILE: drivers/gpu/drm/i915/intel_runtime_pm.c:2645:
+ DRM_ERROR("DBus power %s timeout!\n",
+ enable ? "enable" : "disable");

total: 0 errors, 0 warnings, 4 checks, 219 lines checked 587921dcf1da drm/i915/icl: update ddb entry start/end mask during hw ddb readout

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#9 lore raw

✗ Fi.CI.BAT: warning for Optimize use of DBuf slices

In reply to Mahesh Kumar
== Series Details ==

== Series Details ==

Series: Optimize use of DBuf slices
URL : https://patchwork.freedesktop.org/series/41180/
State : warning

== Summary ==

Series 41180v1 Optimize use of DBuf slices
https://patchwork.freedesktop.org/api/1.0/series/41180/revisions/1/mbox/

---- Possible new issues:

Test kms_force_connector_basic:

        Subgroup prune-stale-modes:
                pass       -> SKIP       (fi-snb-2600)

---- Known issues:

Test gem_exec_suspend:

        Subgroup basic-s3:
                dmesg-warn -> PASS       (fi-glk-j4005) fdo#103359

fdo#103359 https://bugs.freedesktop.org/show_bug.cgi?id=103359

fi-bdw-5557u     total:285  pass:264  dwarn:0   dfail:0   fail:0   skip:21  time:431s
fi-bdw-gvtdvm    total:285  pass:261  dwarn:0   dfail:0   fail:0   skip:24  time:437s
fi-blb-e6850     total:285  pass:220  dwarn:1   dfail:0   fail:0   skip:64  time:381s
fi-bsw-n3050     total:285  pass:239  dwarn:0   dfail:0   fail:0   skip:46  time:538s
fi-bwr-2160      total:285  pass:180  dwarn:0   dfail:0   fail:0   skip:105 time:296s
fi-bxt-dsi       total:285  pass:255  dwarn:0   dfail:0   fail:0   skip:30  time:515s
fi-bxt-j4205     total:285  pass:256  dwarn:0   dfail:0   fail:0   skip:29  time:522s
fi-byt-j1900     total:285  pass:250  dwarn:0   dfail:0   fail:0   skip:35  time:521s
fi-byt-n2820     total:285  pass:246  dwarn:0   dfail:0   fail:0   skip:39  time:508s
fi-cfl-8700k     total:285  pass:257  dwarn:0   dfail:0   fail:0   skip:28  time:415s
fi-cfl-s3        total:285  pass:259  dwarn:0   dfail:0   fail:0   skip:26  time:560s
fi-cfl-u         total:285  pass:259  dwarn:0   dfail:0   fail:0   skip:26  time:508s
fi-elk-e7500     total:285  pass:226  dwarn:0   dfail:0   fail:0   skip:59  time:424s
fi-gdg-551       total:285  pass:176  dwarn:0   dfail:0   fail:1   skip:108 time:316s
fi-glk-1         total:285  pass:257  dwarn:0   dfail:0   fail:0   skip:28  time:535s
fi-glk-j4005     total:285  pass:256  dwarn:0   dfail:0   fail:0   skip:29  time:483s
fi-hsw-4770      total:285  pass:258  dwarn:0   dfail:0   fail:0   skip:27  time:405s
fi-ilk-650       total:285  pass:225  dwarn:0   dfail:0   fail:0   skip:60  time:421s
fi-ivb-3520m     total:285  pass:256  dwarn:0   dfail:0   fail:0   skip:29  time:462s
fi-ivb-3770      total:285  pass:252  dwarn:0   dfail:0   fail:0   skip:33  time:433s
fi-kbl-7500u     total:285  pass:260  dwarn:1   dfail:0   fail:0   skip:24  time:472s
fi-kbl-7567u     total:285  pass:265  dwarn:0   dfail:0   fail:0   skip:20  time:460s
fi-kbl-r         total:285  pass:258  dwarn:0   dfail:0   fail:0   skip:27  time:511s
fi-skl-6260u     total:285  pass:265  dwarn:0   dfail:0   fail:0   skip:20  time:443s
fi-skl-6600u     total:285  pass:258  dwarn:0   dfail:0   fail:0   skip:27  time:534s
fi-skl-6700k2    total:285  pass:261  dwarn:0   dfail:0   fail:0   skip:24  time:499s
fi-skl-6770hq    total:285  pass:265  dwarn:0   dfail:0   fail:0   skip:20  time:510s
fi-skl-guc       total:285  pass:257  dwarn:0   dfail:0   fail:0   skip:28  time:428s
fi-skl-gvtdvm    total:285  pass:262  dwarn:0   dfail:0   fail:0   skip:23  time:446s
fi-snb-2520m     total:285  pass:245  dwarn:0   dfail:0   fail:0   skip:40  time:591s
fi-snb-2600      total:285  pass:244  dwarn:0   dfail:0   fail:0   skip:41  time:398s

Blacklisted hosts:
fi-cnl-psr total:285 pass:256 dwarn:3 dfail:0 fail:0 skip:26 time:516s

e29a10513429cca404e9847a399efbdbb4bdd4bf drm-tip: 2018y-04m-04d-20h-47m-24s UTC integration manifest 587921dcf1da drm/i915/icl: update ddb entry start/end mask during hw ddb readout 5cc032e8d384 drm/i915/icl: Enable 2nd DBuf slice only when needed
e2b3a175155d drm/i915/icl: track dbuf slice-2 status

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_8586/issues.html
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx