drm/i915: Add self-refresh support on Sandybridge

Add the support of memory self-refresh on Sandybridge, which is now
support 3 levels of watermarks and the source of the latency values
for watermarks has changed.

On Sandybridge, the LP0 WM value is not hardcoded any more. All the
latency value is now should be extracted from MCHBAR SSKPD register.
And the MCHBAR base address is changed, too.

For the WM values, if any calculated watermark values is larger than
the maximum value that can be programmed into the associated watermark
register, that watermark must be disabled.

Signed-off-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
[ickle: remove duplicate compute routines and fixup for checkpatch]
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

3 files changed, 334 insertions, 18 deletions

diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
index 22821994b35..864e75d762e 100644
--- a/drivers/gpu/drm/i915/i915_debugfs.c
+++ b/drivers/gpu/drm/i915/i915_debugfs.c
@@ -933,7 +933,7 @@ static int i915_sr_status(struct seq_file *m, void *unused)
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	bool sr_enabled = false;
 
-	if (IS_GEN5(dev))
+	if (HAS_PCH_SPLIT(dev))
 		sr_enabled = I915_READ(WM1_LP_ILK) & WM1_LP_SR_EN;
 	else if (IS_CRESTLINE(dev) || IS_I945G(dev) || IS_I945GM(dev))
 		sr_enabled = I915_READ(FW_BLC_SELF) & FW_BLC_SELF_EN;
diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
index b284c13fd63..61ef98db9ff 100644
--- a/drivers/gpu/drm/i915/i915_reg.h
+++ b/drivers/gpu/drm/i915/i915_reg.h
@@ -596,6 +596,8 @@
 
 #define ILK_DISPLAY_CHICKEN1	0x42000
 #define   ILK_FBCQ_DIS		(1<<22)
+#define   ILK_PABSTRETCH_DIS 	(1<<21)
+
 
 /*
  * GPIO regs
@@ -955,6 +957,8 @@
  */
 #define MCHBAR_MIRROR_BASE	0x10000
 
+#define MCHBAR_MIRROR_BASE_SNB	0x140000
+
 /** 915-945 and GM965 MCH register controlling DRAM channel access */
 #define DCC			0x10200
 #define DCC_ADDRESSING_MODE_SINGLE_CHANNEL		(0 << 0)
@@ -2346,6 +2350,40 @@
 
 #define ILK_FIFO_LINE_SIZE	64
 
+/* define the WM info on Sandybridge */
+#define SNB_DISPLAY_FIFO	128
+#define SNB_DISPLAY_MAXWM	0x7f	/* bit 16:22 */
+#define SNB_DISPLAY_DFTWM	8
+#define SNB_CURSOR_FIFO		32
+#define SNB_CURSOR_MAXWM	0x1f	/* bit 4:0 */
+#define SNB_CURSOR_DFTWM	8
+
+#define SNB_DISPLAY_SR_FIFO	512
+#define SNB_DISPLAY_MAX_SRWM	0x1ff	/* bit 16:8 */
+#define SNB_DISPLAY_DFT_SRWM	0x3f
+#define SNB_CURSOR_SR_FIFO	64
+#define SNB_CURSOR_MAX_SRWM	0x3f	/* bit 5:0 */
+#define SNB_CURSOR_DFT_SRWM	8
+
+#define SNB_FBC_MAX_SRWM	0xf	/* bit 23:20 */
+
+#define SNB_FIFO_LINE_SIZE	64
+
+
+/* the address where we get all kinds of latency value */
+#define SSKPD			0x5d10
+#define SSKPD_WM_MASK		0x3f
+#define SSKPD_WM0_SHIFT		0
+#define SSKPD_WM1_SHIFT		8
+#define SSKPD_WM2_SHIFT		16
+#define SSKPD_WM3_SHIFT		24
+
+#define SNB_LATENCY(shift)	(I915_READ(MCHBAR_MIRROR_BASE_SNB + SSKPD) >> (shift) & SSKPD_WM_MASK)
+#define SNB_READ_WM0_LATENCY()		SNB_LATENCY(SSKPD_WM0_SHIFT)
+#define SNB_READ_WM1_LATENCY()		SNB_LATENCY(SSKPD_WM1_SHIFT)
+#define SNB_READ_WM2_LATENCY()		SNB_LATENCY(SSKPD_WM2_SHIFT)
+#define SNB_READ_WM3_LATENCY()		SNB_LATENCY(SSKPD_WM3_SHIFT)
+
 /*
  * The two pipe frame counter registers are not synchronized, so
  * reading a stable value is somewhat tricky. The following code
@@ -2651,6 +2689,8 @@
 #define  ILK_VSDPFD_FULL	(1<<21)
 #define ILK_DSPCLK_GATE		0x42020
 #define  ILK_DPARB_CLK_GATE	(1<<5)
+#define  ILK_DPFD_CLK_GATE	(1<<7)
+
 /* According to spec this bit 7/8/9 of 0x42020 should be set to enable FBC */
 #define   ILK_CLK_FBC		(1<<7)
 #define   ILK_DPFC_DIS1		(1<<8)
diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index 840bfc3ae71..eaf2bc6b537 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -2845,6 +2845,39 @@ static struct intel_watermark_params ironlake_cursor_srwm_info = {
 	ILK_FIFO_LINE_SIZE
 };
 
+static struct intel_watermark_params sandybridge_display_wm_info = {
+	SNB_DISPLAY_FIFO,
+	SNB_DISPLAY_MAXWM,
+	SNB_DISPLAY_DFTWM,
+	2,
+	SNB_FIFO_LINE_SIZE
+};
+
+static struct intel_watermark_params sandybridge_cursor_wm_info = {
+	SNB_CURSOR_FIFO,
+	SNB_CURSOR_MAXWM,
+	SNB_CURSOR_DFTWM,
+	2,
+	SNB_FIFO_LINE_SIZE
+};
+
+static struct intel_watermark_params sandybridge_display_srwm_info = {
+	SNB_DISPLAY_SR_FIFO,
+	SNB_DISPLAY_MAX_SRWM,
+	SNB_DISPLAY_DFT_SRWM,
+	2,
+	SNB_FIFO_LINE_SIZE
+};
+
+static struct intel_watermark_params sandybridge_cursor_srwm_info = {
+	SNB_CURSOR_SR_FIFO,
+	SNB_CURSOR_MAX_SRWM,
+	SNB_CURSOR_DFT_SRWM,
+	2,
+	SNB_FIFO_LINE_SIZE
+};
+
+
 /**
  * intel_calculate_wm - calculate watermark level
  * @clock_in_khz: pixel clock
@@ -3378,6 +3411,10 @@ static void i830_update_wm(struct drm_device *dev, int planea_clock, int unused,
 
 static bool ironlake_compute_wm0(struct drm_device *dev,
 				 int pipe,
+				 const struct intel_watermark_params *display,
+				 int display_latency,
+				 const struct intel_watermark_params *cursor,
+				 int cursor_latency,
 				 int *plane_wm,
 				 int *cursor_wm)
 {
@@ -3395,22 +3432,20 @@ static bool ironlake_compute_wm0(struct drm_device *dev,
 	pixel_size = crtc->fb->bits_per_pixel / 8;
 
 	/* Use the small buffer method to calculate plane watermark */
-	entries = ((clock * pixel_size / 1000) * ILK_LP0_PLANE_LATENCY) / 1000;
-	entries = DIV_ROUND_UP(entries,
-			       ironlake_display_wm_info.cacheline_size);
-	*plane_wm = entries + ironlake_display_wm_info.guard_size;
-	if (*plane_wm > (int)ironlake_display_wm_info.max_wm)
-		*plane_wm = ironlake_display_wm_info.max_wm;
+	entries = ((clock * pixel_size / 1000) * display_latency * 100) / 1000;
+	entries = DIV_ROUND_UP(entries, display->cacheline_size);
+	*plane_wm = entries + display->guard_size;
+	if (*plane_wm > (int)display->max_wm)
+		*plane_wm = display->max_wm;
 
 	/* Use the large buffer method to calculate cursor watermark */
 	line_time_us = ((htotal * 1000) / clock);
-	line_count = (ILK_LP0_CURSOR_LATENCY / line_time_us + 1000) / 1000;
+	line_count = (cursor_latency * 100 / line_time_us + 1000) / 1000;
 	entries = line_count * 64 * pixel_size;
-	entries = DIV_ROUND_UP(entries,
-			       ironlake_cursor_wm_info.cacheline_size);
-	*cursor_wm = entries + ironlake_cursor_wm_info.guard_size;
-	if (*cursor_wm > ironlake_cursor_wm_info.max_wm)
-		*cursor_wm = ironlake_cursor_wm_info.max_wm;
+	entries = DIV_ROUND_UP(entries, cursor->cacheline_size);
+	*cursor_wm = entries + cursor->guard_size;
+	if (*cursor_wm > (int)cursor->max_wm)
+		*cursor_wm = (int)cursor->max_wm;
 
 	return true;
 }
@@ -3425,7 +3460,12 @@ static void ironlake_update_wm(struct drm_device *dev,
 	int tmp;
 
 	enabled = 0;
-	if (ironlake_compute_wm0(dev, 0, &plane_wm, &cursor_wm)) {
+	if (ironlake_compute_wm0(dev, 0,
+				 &ironlake_display_wm_info,
+				 ILK_LP0_PLANE_LATENCY,
+				 &ironlake_cursor_wm_info,
+				 ILK_LP0_CURSOR_LATENCY,
+				 &plane_wm, &cursor_wm)) {
 		I915_WRITE(WM0_PIPEA_ILK,
 			   (plane_wm << WM0_PIPE_PLANE_SHIFT) | cursor_wm);
 		DRM_DEBUG_KMS("FIFO watermarks For pipe A -"
@@ -3434,7 +3474,12 @@ static void ironlake_update_wm(struct drm_device *dev,
 		enabled++;
 	}
 
-	if (ironlake_compute_wm0(dev, 1, &plane_wm, &cursor_wm)) {
+	if (ironlake_compute_wm0(dev, 1,
+				 &ironlake_display_wm_info,
+				 ILK_LP0_PLANE_LATENCY,
+				 &ironlake_cursor_wm_info,
+				 ILK_LP0_CURSOR_LATENCY,
+				 &plane_wm, &cursor_wm)) {
 		I915_WRITE(WM0_PIPEB_ILK,
 			   (plane_wm << WM0_PIPE_PLANE_SHIFT) | cursor_wm);
 		DRM_DEBUG_KMS("FIFO watermarks For pipe B -"
@@ -3500,6 +3545,197 @@ static void ironlake_update_wm(struct drm_device *dev,
 	/* XXX setup WM2 and WM3 */
 }
 
+/*
+ * Check the wm result.
+ *
+ * If any calculated watermark values is larger than the maximum value that
+ * can be programmed into the associated watermark register, that watermark
+ * must be disabled.
+ *
+ * Also return true if all of those watermark values is 0, which is set by
+ * sandybridge_compute_srwm, to indicate the latency is ZERO.
+ */
+static bool sandybridge_check_srwm(struct drm_device *dev, int level,
+				   int fbc_wm, int display_wm, int cursor_wm)
+{
+	struct drm_i915_private *dev_priv = dev->dev_private;
+
+	DRM_DEBUG_KMS("watermark %d: display plane %d, fbc lines %d,"
+		      " cursor %d\n", level, display_wm, fbc_wm, cursor_wm);
+
+	if (fbc_wm > SNB_FBC_MAX_SRWM) {
+		DRM_DEBUG_KMS("fbc watermark(%d) is too large(%d), disabling wm%d+\n",
+				fbc_wm, SNB_FBC_MAX_SRWM, level);
+
+		/* fbc has it's own way to disable FBC WM */
+		I915_WRITE(DISP_ARB_CTL,
+			   I915_READ(DISP_ARB_CTL) | DISP_FBC_WM_DIS);
+		return false;
+	}
+
+	if (display_wm > SNB_DISPLAY_MAX_SRWM) {
+		DRM_DEBUG_KMS("display watermark(%d) is too large(%d), disabling wm%d+\n",
+				display_wm, SNB_DISPLAY_MAX_SRWM, level);
+		return false;
+	}
+
+	if (cursor_wm > SNB_CURSOR_MAX_SRWM) {
+		DRM_DEBUG_KMS("cursor watermark(%d) is too large(%d), disabling wm%d+\n",
+				cursor_wm, SNB_CURSOR_MAX_SRWM, level);
+		return false;
+	}
+
+	if (!(fbc_wm || display_wm || cursor_wm)) {
+		DRM_DEBUG_KMS("latency %d is 0, disabling wm%d+\n", level, level);
+		return false;
+	}
+
+	return true;
+}
+
+/*
+ * Compute watermark values of WM[1-3],
+ */
+static bool sandybridge_compute_srwm(struct drm_device *dev, int level,
+				     int hdisplay, int htotal, int pixel_size,
+				     int clock, int latency_ns, int *fbc_wm,
+				     int *display_wm, int *cursor_wm)
+{
+
+	unsigned long line_time_us;
+	int small, large;
+	int entries;
+	int line_count, line_size;
+
+	if (!latency_ns) {
+		*fbc_wm = *display_wm = *cursor_wm = 0;
+		return false;
+	}
+
+	line_time_us = (htotal * 1000) / clock;
+	line_count = (latency_ns / line_time_us + 1000) / 1000;
+	line_size = hdisplay * pixel_size;
+
+	/* Use the minimum of the small and large buffer method for primary */
+	small = ((clock * pixel_size / 1000) * latency_ns) / 1000;
+	large = line_count * line_size;
+
+	entries = DIV_ROUND_UP(min(small, large),
+				sandybridge_display_srwm_info.cacheline_size);
+	*display_wm = entries + sandybridge_display_srwm_info.guard_size;
+
+	/*
+	 * Spec said:
+	 * FBC WM = ((Final Primary WM * 64) / number of bytes per line) + 2
+	 */
+	*fbc_wm = DIV_ROUND_UP(*display_wm * 64, line_size) + 2;
+
+	/* calculate the self-refresh watermark for display cursor */
+	entries = line_count * pixel_size * 64;
+	entries = DIV_ROUND_UP(entries,
+			       sandybridge_cursor_srwm_info.cacheline_size);
+	*cursor_wm = entries + sandybridge_cursor_srwm_info.guard_size;
+
+	return sandybridge_check_srwm(dev, level,
+				      *fbc_wm, *display_wm, *cursor_wm);
+}
+
+static void sandybridge_update_wm(struct drm_device *dev,
+			       int planea_clock, int planeb_clock,
+			       int hdisplay, int htotal,
+			       int pixel_size)
+{
+	struct drm_i915_private *dev_priv = dev->dev_private;
+	int latency = SNB_READ_WM0_LATENCY();
+	int fbc_wm, plane_wm, cursor_wm, enabled;
+	int clock;
+
+	enabled = 0;
+	if (ironlake_compute_wm0(dev, 0,
+				 &sandybridge_display_wm_info, latency,
+				 &sandybridge_cursor_wm_info, latency,
+				 &plane_wm, &cursor_wm)) {
+		I915_WRITE(WM0_PIPEA_ILK,
+			   (plane_wm << WM0_PIPE_PLANE_SHIFT) | cursor_wm);
+		DRM_DEBUG_KMS("FIFO watermarks For pipe A -"
+			      " plane %d, " "cursor: %d\n",
+			      plane_wm, cursor_wm);
+		enabled++;
+	}
+
+	if (ironlake_compute_wm0(dev, 1,
+				 &sandybridge_display_wm_info, latency,
+				 &sandybridge_cursor_wm_info, latency,
+				 &plane_wm, &cursor_wm)) {
+		I915_WRITE(WM0_PIPEB_ILK,
+			   (plane_wm << WM0_PIPE_PLANE_SHIFT) | cursor_wm);
+		DRM_DEBUG_KMS("FIFO watermarks For pipe B -"
+			      " plane %d, cursor: %d\n",
+			      plane_wm, cursor_wm);
+		enabled++;
+	}
+
+	/*
+	 * Calculate and update the self-refresh watermark only when one
+	 * display plane is used.
+	 *
+	 * SNB support 3 levels of watermark.
+	 *
+	 * WM1/WM2/WM2 watermarks have to be enabled in the ascending order,
+	 * and disabled in the descending order
+	 *
+	 */
+	I915_WRITE(WM3_LP_ILK, 0);
+	I915_WRITE(WM2_LP_ILK, 0);
+	I915_WRITE(WM1_LP_ILK, 0);
+
+	if (enabled != 1)
+		return;
+
+	clock = planea_clock ? planea_clock : planeb_clock;
+
+	/* WM1 */
+	if (!sandybridge_compute_srwm(dev, 1, hdisplay, htotal, pixel_size,
+				      clock, SNB_READ_WM1_LATENCY() * 500,
+				      &fbc_wm, &plane_wm, &cursor_wm))
+		return;
+
+	I915_WRITE(WM1_LP_ILK,
+		   WM1_LP_SR_EN |
+		   (SNB_READ_WM1_LATENCY() << WM1_LP_LATENCY_SHIFT) |
+		   (fbc_wm << WM1_LP_FBC_SHIFT) |
+		   (plane_wm << WM1_LP_SR_SHIFT) |
+		   cursor_wm);
+
+	/* WM2 */
+	if (!sandybridge_compute_srwm(dev, 2,
+				      hdisplay, htotal, pixel_size,
+				      clock, SNB_READ_WM2_LATENCY() * 500,
+				      &fbc_wm, &plane_wm, &cursor_wm))
+		return;
+
+	I915_WRITE(WM2_LP_ILK,
+		   WM2_LP_EN |
+		   (SNB_READ_WM2_LATENCY() << WM1_LP_LATENCY_SHIFT) |
+		   (fbc_wm << WM1_LP_FBC_SHIFT) |
+		   (plane_wm << WM1_LP_SR_SHIFT) |
+		   cursor_wm);
+
+	/* WM3 */
+	if (!sandybridge_compute_srwm(dev, 3,
+				      hdisplay, htotal, pixel_size,
+				      clock, SNB_READ_WM3_LATENCY() * 500,
+				      &fbc_wm, &plane_wm, &cursor_wm))
+		return;
+
+	I915_WRITE(WM3_LP_ILK,
+		   WM3_LP_EN |
+		   (SNB_READ_WM3_LATENCY() << WM1_LP_LATENCY_SHIFT) |
+		   (fbc_wm << WM1_LP_FBC_SHIFT) |
+		   (plane_wm << WM1_LP_SR_SHIFT) |
+		   cursor_wm);
+}
+
 /**
  * intel_update_watermarks - update FIFO watermark values based on current modes
  *
@@ -5975,9 +6211,9 @@ void intel_enable_clock_gating(struct drm_device *dev)
 			I915_WRITE(DISP_ARB_CTL,
 					(I915_READ(DISP_ARB_CTL) |
 						DISP_FBC_WM_DIS));
-		I915_WRITE(WM3_LP_ILK, 0);
-		I915_WRITE(WM2_LP_ILK, 0);
-		I915_WRITE(WM1_LP_ILK, 0);
+			I915_WRITE(WM3_LP_ILK, 0);
+			I915_WRITE(WM2_LP_ILK, 0);
+			I915_WRITE(WM1_LP_ILK, 0);
 		}
 		/*
 		 * Based on the document from hardware guys the following bits
@@ -6010,6 +6246,38 @@ void intel_enable_clock_gating(struct drm_device *dev)
 				   _3D_CHICKEN2_WM_READ_PIPELINED);
 		}
 
+		if (IS_GEN6(dev)) {
+			I915_WRITE(WM3_LP_ILK, 0);
+			I915_WRITE(WM2_LP_ILK, 0);
+			I915_WRITE(WM1_LP_ILK, 0);
+
+			/*
+			 * According to the spec the following bits should be
+			 * set in order to enable memory self-refresh and fbc:
+			 * The bit21 and bit22 of 0x42000
+			 * The bit21 and bit22 of 0x42004
+			 * The bit5 and bit7 of 0x42020
+			 * The bit14 of 0x70180
+			 * The bit14 of 0x71180
+			 */
+			I915_WRITE(ILK_DISPLAY_CHICKEN1,
+				   I915_READ(ILK_DISPLAY_CHICKEN1) |
+				   ILK_FBCQ_DIS | ILK_PABSTRETCH_DIS);
+			I915_WRITE(ILK_DISPLAY_CHICKEN2,
+				   I915_READ(ILK_DISPLAY_CHICKEN2) |
+				   ILK_DPARB_GATE | ILK_VSDPFD_FULL);
+			I915_WRITE(ILK_DSPCLK_GATE,
+				   I915_READ(ILK_DSPCLK_GATE) |
+				   ILK_DPARB_CLK_GATE  |
+				   ILK_DPFD_CLK_GATE);
+
+			I915_WRITE(DSPACNTR,
+				   I915_READ(DSPACNTR) |
+				   DISPPLANE_TRICKLE_FEED_DISABLE);
+			I915_WRITE(DSPBCNTR,
+				   I915_READ(DSPBCNTR) |
+				   DISPPLANE_TRICKLE_FEED_DISABLE);
+		}
 	} else if (IS_G4X(dev)) {
 		uint32_t dspclk_gate;
 		I915_WRITE(RENCLK_GATE_D1, 0);
@@ -6176,6 +6444,14 @@ static void intel_init_display(struct drm_device *dev)
 					      "Disable CxSR\n");
 				dev_priv->display.update_wm = NULL;
 			}
+		} else if (IS_GEN6(dev)) {
+			if (SNB_READ_WM0_LATENCY()) {
+				dev_priv->display.update_wm = sandybridge_update_wm;
+			} else {
+				DRM_DEBUG_KMS("Failed to read display plane latency. "
+					      "Disable CxSR\n");
+				dev_priv->display.update_wm = NULL;
+			}
 		} else
 			dev_priv->display.update_wm = NULL;
 	} else if (IS_PINEVIEW(dev)) {