drm/etnaviv: add etnaviv cooling device

Each Vivante GPU contains a clock divider which can divide the GPU clock
by 2^n, which can lower the power dissipation from the GPU.  It has been
suggested that the GC600 on Dove is responsible for 20-30% of the power
dissipation from the SoC, so lowering the GPU clock rate provides a way
to throttle the power dissiptation, and reduce the temperature when the
SoC gets hot.

This patch hooks the Etnaviv driver into the kernel's thermal management
to allow the GPUs to be throttled when necessary, allowing a reduction in
GPU clock rate from /1 to /64 in power of 2 steps.
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
Reviewed-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>

drivers/gpu/drm/etnaviv/etnaviv_gpu.c

@@ -18,6 +18,7 @@
 #include <linux/dma-fence.h>
 #include <linux/moduleparam.h>
 #include <linux/of_device.h>
+#include <linux/thermal.h>
 
 #include "etnaviv_cmdbuf.h"
 #include "etnaviv_dump.h"
@@ -409,6 +410,17 @@ static void etnaviv_gpu_load_clock(struct etnaviv_gpu *gpu, u32 clock)
 	gpu_write(gpu, VIVS_HI_CLOCK_CONTROL, clock);
 }
 
+static void etnaviv_gpu_update_clock(struct etnaviv_gpu *gpu)
+{
+	unsigned int fscale = 1 << (6 - gpu->freq_scale);
+	u32 clock;
+
+	clock = VIVS_HI_CLOCK_CONTROL_DISABLE_DEBUG_REGISTERS |
+		VIVS_HI_CLOCK_CONTROL_FSCALE_VAL(fscale);
+
+	etnaviv_gpu_load_clock(gpu, clock);
+}
+
 static int etnaviv_hw_reset(struct etnaviv_gpu *gpu)
 {
 	u32 control, idle;
@@ -426,11 +438,10 @@ static int etnaviv_hw_reset(struct etnaviv_gpu *gpu)
 	timeout = jiffies + msecs_to_jiffies(1000);
 
 	while (time_is_after_jiffies(timeout)) {
-		control = VIVS_HI_CLOCK_CONTROL_DISABLE_DEBUG_REGISTERS |
-			  VIVS_HI_CLOCK_CONTROL_FSCALE_VAL(0x40);
-
 		/* enable clock */
-		etnaviv_gpu_load_clock(gpu, control);
+		etnaviv_gpu_update_clock(gpu);
+
+		control = gpu_read(gpu, VIVS_HI_CLOCK_CONTROL);
 
 		/* Wait for stable clock.  Vivante's code waited for 1ms */
 		usleep_range(1000, 10000);
@@ -490,11 +501,7 @@ static int etnaviv_hw_reset(struct etnaviv_gpu *gpu)
 	}
 
 	/* We rely on the GPU running, so program the clock */
-	control = VIVS_HI_CLOCK_CONTROL_DISABLE_DEBUG_REGISTERS |
-		  VIVS_HI_CLOCK_CONTROL_FSCALE_VAL(0x40);
-
-	/* enable clock */
-	etnaviv_gpu_load_clock(gpu, control);
+	etnaviv_gpu_update_clock(gpu);
 
 	return 0;
 }
@@ -1532,17 +1539,13 @@ static int etnaviv_gpu_hw_suspend(struct etnaviv_gpu *gpu)
 #ifdef CONFIG_PM
 static int etnaviv_gpu_hw_resume(struct etnaviv_gpu *gpu)
 {
-	u32 clock;
 	int ret;
 
 	ret = mutex_lock_killable(&gpu->lock);
 	if (ret)
 		return ret;
 
-	clock = VIVS_HI_CLOCK_CONTROL_DISABLE_DEBUG_REGISTERS |
-		VIVS_HI_CLOCK_CONTROL_FSCALE_VAL(0x40);
-
-	etnaviv_gpu_load_clock(gpu, clock);
+	etnaviv_gpu_update_clock(gpu);
 	etnaviv_gpu_hw_init(gpu);
 
 	gpu->switch_context = true;
@@ -1554,6 +1557,47 @@ static int etnaviv_gpu_hw_resume(struct etnaviv_gpu *gpu)
 }
 #endif
 
+static int
+etnaviv_gpu_cooling_get_max_state(struct thermal_cooling_device *cdev,
+				  unsigned long *state)
+{
+	*state = 6;
+
+	return 0;
+}
+
+static int
+etnaviv_gpu_cooling_get_cur_state(struct thermal_cooling_device *cdev,
+				  unsigned long *state)
+{
+	struct etnaviv_gpu *gpu = cdev->devdata;
+
+	*state = gpu->freq_scale;
+
+	return 0;
+}
+
+static int
+etnaviv_gpu_cooling_set_cur_state(struct thermal_cooling_device *cdev,
+				  unsigned long state)
+{
+	struct etnaviv_gpu *gpu = cdev->devdata;
+
+	mutex_lock(&gpu->lock);
+	gpu->freq_scale = state;
+	if (!pm_runtime_suspended(gpu->dev))
+		etnaviv_gpu_update_clock(gpu);
+	mutex_unlock(&gpu->lock);
+
+	return 0;
+}
+
+static struct thermal_cooling_device_ops cooling_ops = {
+	.get_max_state = etnaviv_gpu_cooling_get_max_state,
+	.get_cur_state = etnaviv_gpu_cooling_get_cur_state,
+	.set_cur_state = etnaviv_gpu_cooling_set_cur_state,
+};
+
 static int etnaviv_gpu_bind(struct device *dev, struct device *master,
 	void *data)
 {
@@ -1562,13 +1606,20 @@ static int etnaviv_gpu_bind(struct device *dev, struct device *master,
 	struct etnaviv_gpu *gpu = dev_get_drvdata(dev);
 	int ret;
 
+	gpu->cooling = thermal_of_cooling_device_register(dev->of_node,
+				(char *)dev_name(dev), gpu, &cooling_ops);
+	if (IS_ERR(gpu->cooling))
+		return PTR_ERR(gpu->cooling);
+
 #ifdef CONFIG_PM
 	ret = pm_runtime_get_sync(gpu->dev);
 #else
 	ret = etnaviv_gpu_clk_enable(gpu);
 #endif
-	if (ret < 0)
+	if (ret < 0) {
+		thermal_cooling_device_unregister(gpu->cooling);
 		return ret;
+	}
 
 	gpu->drm = drm;
 	gpu->fence_context = dma_fence_context_alloc(1);
@@ -1622,6 +1673,9 @@ static void etnaviv_gpu_unbind(struct device *dev, struct device *master,
 	}
 
 	gpu->drm = NULL;
+
+	thermal_cooling_device_unregister(gpu->cooling);
+	gpu->cooling = NULL;
 }
 
 static const struct component_ops gpu_ops = {

drivers/gpu/drm/etnaviv/etnaviv_gpu.h

@@ -97,6 +97,7 @@ struct etnaviv_cmdbuf;
 
 struct etnaviv_gpu {
 	struct drm_device *drm;
+	struct thermal_cooling_device *cooling;
 	struct device *dev;
 	struct mutex lock;
 	struct etnaviv_chip_identity identity;
@@ -150,6 +151,7 @@ struct etnaviv_gpu {
 	u32 hangcheck_fence;
 	u32 hangcheck_dma_addr;
 	struct work_struct recover_work;
+	unsigned int freq_scale;
 };
 
 static inline void gpu_write(struct etnaviv_gpu *gpu, u32 reg, u32 data)