drm/i915: add power monitoring support

Add power monitoring support to the i915 driver for use by the IPS
driver.  Export the available power info to the IPS driver through a few
new inter-driver hooks.  When used together, the IPS driver and this
patch can significantly increase graphics performance on Ironlake class
chips.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
[anholt: Fixed 32-bit compile.  stupid obfuscating div_u64()]
Signed-off-by: Eric Anholt <eric@anholt.net>

drivers/gpu/drm/i915/i915_debugfs.c

@@ -491,11 +491,14 @@ static int i915_cur_delayinfo(struct seq_file *m, void *unused)
 	struct drm_device *dev = node->minor->dev;
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	u16 rgvswctl = I915_READ16(MEMSWCTL);
+	u16 rgvstat = I915_READ16(MEMSTAT_ILK);
 
-	seq_printf(m, "Last command: 0x%01x\n", (rgvswctl >> 13) & 0x3);
-	seq_printf(m, "Command status: %d\n", (rgvswctl >> 12) & 1);
-	seq_printf(m, "P%d DELAY 0x%02x\n", (rgvswctl >> 8) & 0xf,
-		   rgvswctl & 0x3f);
+	seq_printf(m, "Requested P-state: %d\n", (rgvswctl >> 8) & 0xf);
+	seq_printf(m, "Requested VID: %d\n", rgvswctl & 0x3f);
+	seq_printf(m, "Current VID: %d\n", (rgvstat & MEMSTAT_VID_MASK) >>
+		   MEMSTAT_VID_SHIFT);
+	seq_printf(m, "Current P-state: %d\n",
+		   (rgvstat & MEMSTAT_PSTATE_MASK) >> MEMSTAT_PSTATE_SHIFT);
 
 	return 0;
 }
@@ -510,7 +513,8 @@ static int i915_delayfreq_table(struct seq_file *m, void *unused)
 
 	for (i = 0; i < 16; i++) {
 		delayfreq = I915_READ(PXVFREQ_BASE + i * 4);
-		seq_printf(m, "P%02dVIDFREQ: 0x%08x\n", i, delayfreq);
+		seq_printf(m, "P%02dVIDFREQ: 0x%08x (VID: %d)\n", i, delayfreq,
+			   (delayfreq & PXVFREQ_PX_MASK) >> PXVFREQ_PX_SHIFT);
 	}
 
 	return 0;
@@ -543,6 +547,8 @@ static int i915_drpc_info(struct seq_file *m, void *unused)
 	struct drm_device *dev = node->minor->dev;
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	u32 rgvmodectl = I915_READ(MEMMODECTL);
+	u32 rstdbyctl = I915_READ(MCHBAR_RENDER_STANDBY);
+	u16 crstandvid = I915_READ16(CRSTANDVID);
 
 	seq_printf(m, "HD boost: %s\n", (rgvmodectl & MEMMODE_BOOST_EN) ?
 		   "yes" : "no");
@@ -557,9 +563,13 @@ static int i915_drpc_info(struct seq_file *m, void *unused)
 		   rgvmodectl & MEMMODE_RCLK_GATE ? "yes" : "no");
 	seq_printf(m, "Starting frequency: P%d\n",
 		   (rgvmodectl & MEMMODE_FSTART_MASK) >> MEMMODE_FSTART_SHIFT);
-	seq_printf(m, "Max frequency: P%d\n",
+	seq_printf(m, "Max P-state: P%d\n",
 		   (rgvmodectl & MEMMODE_FMAX_MASK) >> MEMMODE_FMAX_SHIFT);
-	seq_printf(m, "Min frequency: P%d\n", (rgvmodectl & MEMMODE_FMIN_MASK));
+	seq_printf(m, "Min P-state: P%d\n", (rgvmodectl & MEMMODE_FMIN_MASK));
+	seq_printf(m, "RS1 VID: %d\n", (crstandvid & 0x3f));
+	seq_printf(m, "RS2 VID: %d\n", ((crstandvid >> 8) & 0x3f));
+	seq_printf(m, "Render standby enabled: %s\n",
+		   (rstdbyctl & RCX_SW_EXIT) ? "no" : "yes");
 
 	return 0;
 }
@@ -623,6 +633,36 @@ static int i915_sr_status(struct seq_file *m, void *unused)
 	return 0;
 }
 
+static int i915_emon_status(struct seq_file *m, void *unused)
+{
+	struct drm_info_node *node = (struct drm_info_node *) m->private;
+	struct drm_device *dev = node->minor->dev;
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	unsigned long temp, chipset, gfx;
+
+	temp = i915_mch_val(dev_priv);
+	chipset = i915_chipset_val(dev_priv);
+	gfx = i915_gfx_val(dev_priv);
+
+	seq_printf(m, "GMCH temp: %ld\n", temp);
+	seq_printf(m, "Chipset power: %ld\n", chipset);
+	seq_printf(m, "GFX power: %ld\n", gfx);
+	seq_printf(m, "Total power: %ld\n", chipset + gfx);
+
+	return 0;
+}
+
+static int i915_gfxec(struct seq_file *m, void *unused)
+{
+	struct drm_info_node *node = (struct drm_info_node *) m->private;
+	struct drm_device *dev = node->minor->dev;
+	drm_i915_private_t *dev_priv = dev->dev_private;
+
+	seq_printf(m, "GFXEC: %ld\n", (unsigned long)I915_READ(0x112f4));
+
+	return 0;
+}
+
 static int
 i915_wedged_open(struct inode *inode,
 		 struct file *filp)
@@ -745,6 +785,8 @@ static struct drm_info_list i915_debugfs_list[] = {
 	{"i915_delayfreq_table", i915_delayfreq_table, 0},
 	{"i915_inttoext_table", i915_inttoext_table, 0},
 	{"i915_drpc_info", i915_drpc_info, 0},
+	{"i915_emon_status", i915_emon_status, 0},
+	{"i915_gfxec", i915_gfxec, 0},
 	{"i915_fbc_status", i915_fbc_status, 0},
 	{"i915_sr_status", i915_sr_status, 0},
 };

drivers/gpu/drm/i915/i915_drv.h

@@ -619,6 +619,18 @@ typedef struct drm_i915_private {
 	u8 cur_delay;
 	u8 min_delay;
 	u8 max_delay;
+	u8 fmax;
+	u8 fstart;
+
+ 	u64 last_count1;
+ 	unsigned long last_time1;
+ 	u64 last_count2;
+ 	struct timespec last_time2;
+ 	unsigned long gfx_power;
+ 	int c_m;
+ 	int r_t;
+ 	u8 corr;
+	spinlock_t *mchdev_lock;
 
 	enum no_fbc_reason no_fbc_reason;
 
@@ -802,6 +814,11 @@ extern int i915_emit_box(struct drm_device *dev,
 			 struct drm_clip_rect *boxes,
 			 int i, int DR1, int DR4);
 extern int i965_reset(struct drm_device *dev, u8 flags);
+extern unsigned long i915_chipset_val(struct drm_i915_private *dev_priv);
+extern unsigned long i915_mch_val(struct drm_i915_private *dev_priv);
+extern unsigned long i915_gfx_val(struct drm_i915_private *dev_priv);
+extern void i915_update_gfx_val(struct drm_i915_private *dev_priv);
+
 
 /* i915_irq.c */
 void i915_hangcheck_elapsed(unsigned long data);
@@ -1005,7 +1022,7 @@ extern void g4x_disable_fbc(struct drm_device *dev);
 extern void intel_disable_fbc(struct drm_device *dev);
 extern void intel_enable_fbc(struct drm_crtc *crtc, unsigned long interval);
 extern bool intel_fbc_enabled(struct drm_device *dev);
-
+extern bool ironlake_set_drps(struct drm_device *dev, u8 val);
 extern void intel_detect_pch (struct drm_device *dev);
 extern int intel_trans_dp_port_sel (struct drm_crtc *crtc);
 
@@ -1030,6 +1047,7 @@ extern int intel_trans_dp_port_sel (struct drm_crtc *crtc);
 #define I915_WRITE64(reg, val)	writeq(val, dev_priv->regs + (reg))
 #define I915_READ64(reg)	readq(dev_priv->regs + (reg))
 #define POSTING_READ(reg)	(void)I915_READ(reg)
+#define POSTING_READ16(reg)	(void)I915_READ16(reg)
 
 #define I915_VERBOSE 0
 

drivers/gpu/drm/i915/i915_irq.c

@@ -278,10 +278,9 @@ static void i915_handle_rps_change(struct drm_device *dev)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	u32 busy_up, busy_down, max_avg, min_avg;
-	u16 rgvswctl;
 	u8 new_delay = dev_priv->cur_delay;
 
-	I915_WRITE(MEMINTRSTS, I915_READ(MEMINTRSTS) & ~MEMINT_EVAL_CHG);
+	I915_WRITE16(MEMINTRSTS, MEMINT_EVAL_CHG);
 	busy_up = I915_READ(RCPREVBSYTUPAVG);
 	busy_down = I915_READ(RCPREVBSYTDNAVG);
 	max_avg = I915_READ(RCBMAXAVG);
@@ -300,27 +299,8 @@ static void i915_handle_rps_change(struct drm_device *dev)
 			new_delay = dev_priv->min_delay;
 	}
 
-	DRM_DEBUG("rps change requested: %d -> %d\n",
-		  dev_priv->cur_delay, new_delay);
-
-	rgvswctl = I915_READ(MEMSWCTL);
-	if (rgvswctl & MEMCTL_CMD_STS) {
-		DRM_ERROR("gpu busy, RCS change rejected\n");
-		return; /* still busy with another command */
-	}
-
-	/* Program the new state */
-	rgvswctl = (MEMCTL_CMD_CHFREQ << MEMCTL_CMD_SHIFT) |
-		(new_delay << MEMCTL_FREQ_SHIFT) | MEMCTL_SFCAVM;
-	I915_WRITE(MEMSWCTL, rgvswctl);
-	POSTING_READ(MEMSWCTL);
-
-	rgvswctl |= MEMCTL_CMD_STS;
-	I915_WRITE(MEMSWCTL, rgvswctl);
-
-	dev_priv->cur_delay = new_delay;
-
-	DRM_DEBUG("rps changed\n");
+	if (ironlake_set_drps(dev, new_delay))
+		dev_priv->cur_delay = new_delay;
 
 	return;
 }
@@ -392,7 +372,7 @@ irqreturn_t ironlake_irq_handler(struct drm_device *dev)
 	}
 
 	if (de_iir & DE_PCU_EVENT) {
-		I915_WRITE(MEMINTRSTS, I915_READ(MEMINTRSTS));
+		I915_WRITE16(MEMINTRSTS, I915_READ(MEMINTRSTS));
 		i915_handle_rps_change(dev);
 	}
 

drivers/gpu/drm/i915/i915_reg.h

@@ -838,6 +838,12 @@
 #define CLKCFG_MEM_800					(3 << 4)
 #define CLKCFG_MEM_MASK					(7 << 4)
 
+#define TR1			0x11006
+#define TSFS			0x11020
+#define   TSFS_SLOPE_MASK	0x0000ff00
+#define   TSFS_SLOPE_SHIFT	8
+#define   TSFS_INTR_MASK	0x000000ff
+
 #define CRSTANDVID		0x11100
 #define PXVFREQ_BASE		0x11110 /* P[0-15]VIDFREQ (0x1114c) (Ironlake) */
 #define   PXVFREQ_PX_MASK	0x7f000000
@@ -976,6 +982,41 @@
 #define   MEMSTAT_SRC_CTL_STDBY 3
 #define RCPREVBSYTUPAVG		0x113b8
 #define RCPREVBSYTDNAVG		0x113bc
+#define SDEW			0x1124c
+#define CSIEW0			0x11250
+#define CSIEW1			0x11254
+#define CSIEW2			0x11258
+#define PEW			0x1125c
+#define DEW			0x11270
+#define MCHAFE			0x112c0
+#define CSIEC			0x112e0
+#define DMIEC			0x112e4
+#define DDREC			0x112e8
+#define PEG0EC			0x112ec
+#define PEG1EC			0x112f0
+#define GFXEC			0x112f4
+#define RPPREVBSYTUPAVG		0x113b8
+#define RPPREVBSYTDNAVG		0x113bc
+#define ECR			0x11600
+#define   ECR_GPFE		(1<<31)
+#define   ECR_IMONE		(1<<30)
+#define   ECR_CAP_MASK		0x0000001f /* Event range, 0-31 */
+#define OGW0			0x11608
+#define OGW1			0x1160c
+#define EG0			0x11610
+#define EG1			0x11614
+#define EG2			0x11618
+#define EG3			0x1161c
+#define EG4			0x11620
+#define EG5			0x11624
+#define EG6			0x11628
+#define EG7			0x1162c
+#define PXW			0x11664
+#define PXWL			0x11680
+#define LCFUSE02		0x116c0
+#define   LCFUSE_HIV_MASK	0x000000ff
+#define CSIPLL0			0x12c10
+#define DDRMPLL1		0X12c20
 #define PEG_BAND_GAP_DATA	0x14d68
 
 /*

drivers/gpu/drm/i915/intel_display.c

@@ -4459,6 +4459,8 @@ static void intel_idle_update(struct work_struct *work)
 
 	mutex_lock(&dev->struct_mutex);
 
+	i915_update_gfx_val(dev_priv);
+
 	if (IS_I945G(dev) || IS_I945GM(dev)) {
 		DRM_DEBUG_DRIVER("enable memory self refresh on 945\n");
 		I915_WRITE(FW_BLC_SELF, FW_BLC_SELF_EN_MASK | FW_BLC_SELF_EN);
@@ -5045,10 +5047,32 @@ intel_alloc_power_context(struct drm_device *dev)
 	return NULL;
 }
 
+bool ironlake_set_drps(struct drm_device *dev, u8 val)
+{
+	struct drm_i915_private *dev_priv = dev->dev_private;
+	u16 rgvswctl;
+
+	rgvswctl = I915_READ16(MEMSWCTL);
+	if (rgvswctl & MEMCTL_CMD_STS) {
+		DRM_DEBUG("gpu busy, RCS change rejected\n");
+		return false; /* still busy with another command */
+	}
+
+	rgvswctl = (MEMCTL_CMD_CHFREQ << MEMCTL_CMD_SHIFT) |
+		(val << MEMCTL_FREQ_SHIFT) | MEMCTL_SFCAVM;
+	I915_WRITE16(MEMSWCTL, rgvswctl);
+	POSTING_READ16(MEMSWCTL);
+
+	rgvswctl |= MEMCTL_CMD_STS;
+	I915_WRITE16(MEMSWCTL, rgvswctl);
+
+	return true;
+}
+
 void ironlake_enable_drps(struct drm_device *dev)
 {
 	struct drm_i915_private *dev_priv = dev->dev_private;
-	u32 rgvmodectl = I915_READ(MEMMODECTL), rgvswctl;
+	u32 rgvmodectl = I915_READ(MEMMODECTL);
 	u8 fmax, fmin, fstart, vstart;
 	int i = 0;
 
@@ -5067,13 +5091,21 @@ void ironlake_enable_drps(struct drm_device *dev)
 	fmin = (rgvmodectl & MEMMODE_FMIN_MASK);
 	fstart = (rgvmodectl & MEMMODE_FSTART_MASK) >>
 		MEMMODE_FSTART_SHIFT;
+	fstart = fmax;
+
 	vstart = (I915_READ(PXVFREQ_BASE + (fstart * 4)) & PXVFREQ_PX_MASK) >>
 		PXVFREQ_PX_SHIFT;
 
-	dev_priv->max_delay = fstart; /* can't go to fmax w/o IPS */
+	dev_priv->fmax = fstart; /* IPS callback will increase this */
+	dev_priv->fstart = fstart;
+
+	dev_priv->max_delay = fmax;
 	dev_priv->min_delay = fmin;
 	dev_priv->cur_delay = fstart;
 
+	DRM_DEBUG_DRIVER("fmax: %d, fmin: %d, fstart: %d\n", fmax, fmin,
+			 fstart);
+
 	I915_WRITE(MEMINTREN, MEMINT_CX_SUPR_EN | MEMINT_EVAL_CHG_EN);
 
 	/*
@@ -5095,20 +5127,19 @@ void ironlake_enable_drps(struct drm_device *dev)
 	}
 	msleep(1);
 
-	rgvswctl = (MEMCTL_CMD_CHFREQ << MEMCTL_CMD_SHIFT) |
-		(fstart << MEMCTL_FREQ_SHIFT) | MEMCTL_SFCAVM;
-	I915_WRITE(MEMSWCTL, rgvswctl);
-	POSTING_READ(MEMSWCTL);
+	ironlake_set_drps(dev, fstart);
 
-	rgvswctl |= MEMCTL_CMD_STS;
-	I915_WRITE(MEMSWCTL, rgvswctl);
+	dev_priv->last_count1 = I915_READ(0x112e4) + I915_READ(0x112e8) +
+		I915_READ(0x112e0);
+	dev_priv->last_time1 = jiffies_to_msecs(jiffies);
+	dev_priv->last_count2 = I915_READ(0x112f4);
+	getrawmonotonic(&dev_priv->last_time2);
 }
 
 void ironlake_disable_drps(struct drm_device *dev)
 {
 	struct drm_i915_private *dev_priv = dev->dev_private;
-	u32 rgvswctl;
-	u8 fstart;
+	u16 rgvswctl = I915_READ16(MEMSWCTL);
 
 	/* Ack interrupts, disable EFC interrupt */
 	I915_WRITE(MEMINTREN, I915_READ(MEMINTREN) & ~MEMINT_EVAL_CHG_EN);
@@ -5118,11 +5149,7 @@ void ironlake_disable_drps(struct drm_device *dev)
 	I915_WRITE(DEIMR, I915_READ(DEIMR) | DE_PCU_EVENT);
 
 	/* Go back to the starting frequency */
-	fstart = (I915_READ(MEMMODECTL) & MEMMODE_FSTART_MASK) >>
-		MEMMODE_FSTART_SHIFT;
-	rgvswctl = (MEMCTL_CMD_CHFREQ << MEMCTL_CMD_SHIFT) |
-		(fstart << MEMCTL_FREQ_SHIFT) | MEMCTL_SFCAVM;
-	I915_WRITE(MEMSWCTL, rgvswctl);
+	ironlake_set_drps(dev, dev_priv->fstart);
 	msleep(1);
 	rgvswctl |= MEMCTL_CMD_STS;
 	I915_WRITE(MEMSWCTL, rgvswctl);
@@ -5130,6 +5157,92 @@ void ironlake_disable_drps(struct drm_device *dev)
 
 }
 
+static unsigned long intel_pxfreq(u32 vidfreq)
+{
+	unsigned long freq;
+	int div = (vidfreq & 0x3f0000) >> 16;
+	int post = (vidfreq & 0x3000) >> 12;
+	int pre = (vidfreq & 0x7);
+
+	if (!pre)
+		return 0;
+
+	freq = ((div * 133333) / ((1<<post) * pre));
+
+	return freq;
+}
+
+void intel_init_emon(struct drm_device *dev)
+{
+	struct drm_i915_private *dev_priv = dev->dev_private;
+	u32 lcfuse;
+	u8 pxw[16];
+	int i;
+
+	/* Disable to program */
+	I915_WRITE(ECR, 0);
+	POSTING_READ(ECR);
+
+	/* Program energy weights for various events */
+	I915_WRITE(SDEW, 0x15040d00);
+	I915_WRITE(CSIEW0, 0x007f0000);
+	I915_WRITE(CSIEW1, 0x1e220004);
+	I915_WRITE(CSIEW2, 0x04000004);
+
+	for (i = 0; i < 5; i++)
+		I915_WRITE(PEW + (i * 4), 0);
+	for (i = 0; i < 3; i++)
+		I915_WRITE(DEW + (i * 4), 0);
+
+	/* Program P-state weights to account for frequency power adjustment */
+	for (i = 0; i < 16; i++) {
+		u32 pxvidfreq = I915_READ(PXVFREQ_BASE + (i * 4));
+		unsigned long freq = intel_pxfreq(pxvidfreq);
+		unsigned long vid = (pxvidfreq & PXVFREQ_PX_MASK) >>
+			PXVFREQ_PX_SHIFT;
+		unsigned long val;
+
+		val = vid * vid;
+		val *= (freq / 1000);
+		val *= 255;
+		val /= (127*127*900);
+		if (val > 0xff)
+			DRM_ERROR("bad pxval: %ld\n", val);
+		pxw[i] = val;
+	}
+	/* Render standby states get 0 weight */
+	pxw[14] = 0;
+	pxw[15] = 0;
+
+	for (i = 0; i < 4; i++) {
+		u32 val = (pxw[i*4] << 24) | (pxw[(i*4)+1] << 16) |
+			(pxw[(i*4)+2] << 8) | (pxw[(i*4)+3]);
+		I915_WRITE(PXW + (i * 4), val);
+	}
+
+	/* Adjust magic regs to magic values (more experimental results) */
+	I915_WRITE(OGW0, 0);
+	I915_WRITE(OGW1, 0);
+	I915_WRITE(EG0, 0x00007f00);
+	I915_WRITE(EG1, 0x0000000e);
+	I915_WRITE(EG2, 0x000e0000);
+	I915_WRITE(EG3, 0x68000300);
+	I915_WRITE(EG4, 0x42000000);
+	I915_WRITE(EG5, 0x00140031);
+	I915_WRITE(EG6, 0);
+	I915_WRITE(EG7, 0);
+
+	for (i = 0; i < 8; i++)
+		I915_WRITE(PXWL + (i * 4), 0);
+
+	/* Enable PMON + select events */
+	I915_WRITE(ECR, 0x80000019);
+
+	lcfuse = I915_READ(LCFUSE02);
+
+	dev_priv->corr = (lcfuse & LCFUSE_HIV_MASK);
+}
+
 void intel_init_clock_gating(struct drm_device *dev)
 {
 	struct drm_i915_private *dev_priv = dev->dev_private;
@@ -5376,8 +5489,10 @@ void intel_modeset_init(struct drm_device *dev)
 
 	intel_init_clock_gating(dev);
 
-	if (IS_IRONLAKE_M(dev))
+	if (IS_IRONLAKE_M(dev)) {
 		ironlake_enable_drps(dev);
+		intel_init_emon(dev);
+	}
 
 	INIT_WORK(&dev_priv->idle_work, intel_idle_update);
 	setup_timer(&dev_priv->idle_timer, intel_gpu_idle_timer,