drm/amdgpu/pm: return an error during GPU reset or suspend (v2)

Return an error for sysfs and debugfs power interfaces during
gpu reset and suspend.  Prevents access to the hw while it may
be in an unusable state.

v2: squash in fix to drop suspend check

Acked-by: Nirmoy Das <nirmoy.das@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

1 files changed, 171 insertions, 0 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c
index d7646cbce346..775e389c9a13 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c
@@ -163,6 +163,9 @@ static ssize_t amdgpu_get_power_dpm_state(struct device *dev,
 	enum amd_pm_state_type pm;
 	int ret;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = pm_runtime_get_sync(ddev->dev);
 	if (ret < 0)
 		return ret;
@@ -196,6 +199,9 @@ static ssize_t amdgpu_set_power_dpm_state(struct device *dev,
 	enum amd_pm_state_type  state;
 	int ret;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	if (strncmp("battery", buf, strlen("battery")) == 0)
 		state = POWER_STATE_TYPE_BATTERY;
 	else if (strncmp("balanced", buf, strlen("balanced")) == 0)
@@ -297,6 +303,9 @@ static ssize_t amdgpu_get_power_dpm_force_performance_level(struct device *dev,
 	enum amd_dpm_forced_level level = 0xff;
 	int ret;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = pm_runtime_get_sync(ddev->dev);
 	if (ret < 0)
 		return ret;
@@ -334,6 +343,9 @@ static ssize_t amdgpu_set_power_dpm_force_performance_level(struct device *dev,
 	enum amd_dpm_forced_level current_level = 0xff;
 	int ret = 0;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	if (strncmp("low", buf, strlen("low")) == 0) {
 		level = AMD_DPM_FORCED_LEVEL_LOW;
 	} else if (strncmp("high", buf, strlen("high")) == 0) {
@@ -433,6 +445,9 @@ static ssize_t amdgpu_get_pp_num_states(struct device *dev,
 	struct pp_states_info data;
 	int i, buf_len, ret;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = pm_runtime_get_sync(ddev->dev);
 	if (ret < 0)
 		return ret;
@@ -472,6 +487,9 @@ static ssize_t amdgpu_get_pp_cur_state(struct device *dev,
 	enum amd_pm_state_type pm = 0;
 	int i = 0, ret = 0;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = pm_runtime_get_sync(ddev->dev);
 	if (ret < 0)
 		return ret;
@@ -508,6 +526,9 @@ static ssize_t amdgpu_get_pp_force_state(struct device *dev,
 	struct drm_device *ddev = dev_get_drvdata(dev);
 	struct amdgpu_device *adev = ddev->dev_private;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	if (adev->pp_force_state_enabled)
 		return amdgpu_get_pp_cur_state(dev, attr, buf);
 	else
@@ -525,6 +546,9 @@ static ssize_t amdgpu_set_pp_force_state(struct device *dev,
 	unsigned long idx;
 	int ret;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	if (strlen(buf) == 1)
 		adev->pp_force_state_enabled = false;
 	else if (is_support_sw_smu(adev))
@@ -580,6 +604,9 @@ static ssize_t amdgpu_get_pp_table(struct device *dev,
 	char *table = NULL;
 	int size, ret;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = pm_runtime_get_sync(ddev->dev);
 	if (ret < 0)
 		return ret;
@@ -619,6 +646,9 @@ static ssize_t amdgpu_set_pp_table(struct device *dev,
 	struct amdgpu_device *adev = ddev->dev_private;
 	int ret = 0;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = pm_runtime_get_sync(ddev->dev);
 	if (ret < 0)
 		return ret;
@@ -721,6 +751,9 @@ static ssize_t amdgpu_set_pp_od_clk_voltage(struct device *dev,
 	const char delimiter[3] = {' ', '\n', '\0'};
 	uint32_t type;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	if (count > 127)
 		return -EINVAL;
 
@@ -810,6 +843,9 @@ static ssize_t amdgpu_get_pp_od_clk_voltage(struct device *dev,
 	ssize_t size;
 	int ret;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = pm_runtime_get_sync(ddev->dev);
 	if (ret < 0)
 		return ret;
@@ -859,6 +895,9 @@ static ssize_t amdgpu_set_pp_features(struct device *dev,
 	uint64_t featuremask;
 	int ret;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = kstrtou64(buf, 0, &featuremask);
 	if (ret)
 		return -EINVAL;
@@ -899,6 +938,9 @@ static ssize_t amdgpu_get_pp_features(struct device *dev,
 	ssize_t size;
 	int ret;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = pm_runtime_get_sync(ddev->dev);
 	if (ret < 0)
 		return ret;
@@ -955,6 +997,9 @@ static ssize_t amdgpu_get_pp_dpm_sclk(struct device *dev,
 	ssize_t size;
 	int ret;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = pm_runtime_get_sync(ddev->dev);
 	if (ret < 0)
 		return ret;
@@ -1018,6 +1063,9 @@ static ssize_t amdgpu_set_pp_dpm_sclk(struct device *dev,
 	int ret;
 	uint32_t mask = 0;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = amdgpu_read_mask(buf, count, &mask);
 	if (ret)
 		return ret;
@@ -1049,6 +1097,9 @@ static ssize_t amdgpu_get_pp_dpm_mclk(struct device *dev,
 	ssize_t size;
 	int ret;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = pm_runtime_get_sync(ddev->dev);
 	if (ret < 0)
 		return ret;
@@ -1076,6 +1127,9 @@ static ssize_t amdgpu_set_pp_dpm_mclk(struct device *dev,
 	uint32_t mask = 0;
 	int ret;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = amdgpu_read_mask(buf, count, &mask);
 	if (ret)
 		return ret;
@@ -1107,6 +1161,9 @@ static ssize_t amdgpu_get_pp_dpm_socclk(struct device *dev,
 	ssize_t size;
 	int ret;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = pm_runtime_get_sync(ddev->dev);
 	if (ret < 0)
 		return ret;
@@ -1134,6 +1191,9 @@ static ssize_t amdgpu_set_pp_dpm_socclk(struct device *dev,
 	int ret;
 	uint32_t mask = 0;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = amdgpu_read_mask(buf, count, &mask);
 	if (ret)
 		return ret;
@@ -1167,6 +1227,9 @@ static ssize_t amdgpu_get_pp_dpm_fclk(struct device *dev,
 	ssize_t size;
 	int ret;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = pm_runtime_get_sync(ddev->dev);
 	if (ret < 0)
 		return ret;
@@ -1194,6 +1257,9 @@ static ssize_t amdgpu_set_pp_dpm_fclk(struct device *dev,
 	int ret;
 	uint32_t mask = 0;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = amdgpu_read_mask(buf, count, &mask);
 	if (ret)
 		return ret;
@@ -1227,6 +1293,9 @@ static ssize_t amdgpu_get_pp_dpm_dcefclk(struct device *dev,
 	ssize_t size;
 	int ret;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = pm_runtime_get_sync(ddev->dev);
 	if (ret < 0)
 		return ret;
@@ -1254,6 +1323,9 @@ static ssize_t amdgpu_set_pp_dpm_dcefclk(struct device *dev,
 	int ret;
 	uint32_t mask = 0;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = amdgpu_read_mask(buf, count, &mask);
 	if (ret)
 		return ret;
@@ -1287,6 +1359,9 @@ static ssize_t amdgpu_get_pp_dpm_pcie(struct device *dev,
 	ssize_t size;
 	int ret;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = pm_runtime_get_sync(ddev->dev);
 	if (ret < 0)
 		return ret;
@@ -1314,6 +1389,9 @@ static ssize_t amdgpu_set_pp_dpm_pcie(struct device *dev,
 	int ret;
 	uint32_t mask = 0;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = amdgpu_read_mask(buf, count, &mask);
 	if (ret)
 		return ret;
@@ -1347,6 +1425,9 @@ static ssize_t amdgpu_get_pp_sclk_od(struct device *dev,
 	uint32_t value = 0;
 	int ret;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = pm_runtime_get_sync(ddev->dev);
 	if (ret < 0)
 		return ret;
@@ -1372,6 +1453,9 @@ static ssize_t amdgpu_set_pp_sclk_od(struct device *dev,
 	int ret;
 	long int value;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = kstrtol(buf, 0, &value);
 
 	if (ret)
@@ -1410,6 +1494,9 @@ static ssize_t amdgpu_get_pp_mclk_od(struct device *dev,
 	uint32_t value = 0;
 	int ret;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = pm_runtime_get_sync(ddev->dev);
 	if (ret < 0)
 		return ret;
@@ -1435,6 +1522,9 @@ static ssize_t amdgpu_set_pp_mclk_od(struct device *dev,
 	int ret;
 	long int value;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = kstrtol(buf, 0, &value);
 
 	if (ret)
@@ -1493,6 +1583,9 @@ static ssize_t amdgpu_get_pp_power_profile_mode(struct device *dev,
 	ssize_t size;
 	int ret;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = pm_runtime_get_sync(ddev->dev);
 	if (ret < 0)
 		return ret;
@@ -1528,6 +1621,9 @@ static ssize_t amdgpu_set_pp_power_profile_mode(struct device *dev,
 	long int profile_mode = 0;
 	const char delimiter[3] = {' ', '\n', '\0'};
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	tmp[0] = *(buf);
 	tmp[1] = '\0';
 	ret = kstrtol(tmp, 0, &profile_mode);
@@ -1587,6 +1683,9 @@ static ssize_t amdgpu_get_gpu_busy_percent(struct device *dev,
 	struct amdgpu_device *adev = ddev->dev_private;
 	int r, value, size = sizeof(value);
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	r = pm_runtime_get_sync(ddev->dev);
 	if (r < 0)
 		return r;
@@ -1620,6 +1719,9 @@ static ssize_t amdgpu_get_mem_busy_percent(struct device *dev,
 	struct amdgpu_device *adev = ddev->dev_private;
 	int r, value, size = sizeof(value);
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	r = pm_runtime_get_sync(ddev->dev);
 	if (r < 0)
 		return r;
@@ -1658,6 +1760,9 @@ static ssize_t amdgpu_get_pcie_bw(struct device *dev,
 	uint64_t count0 = 0, count1 = 0;
 	int ret;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	if (adev->flags & AMD_IS_APU)
 		return -ENODATA;
 
@@ -1694,6 +1799,9 @@ static ssize_t amdgpu_get_unique_id(struct device *dev,
 	struct drm_device *ddev = dev_get_drvdata(dev);
 	struct amdgpu_device *adev = ddev->dev_private;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	if (adev->unique_id)
 		return snprintf(buf, PAGE_SIZE, "%016llx\n", adev->unique_id);
 
@@ -1888,6 +1996,9 @@ static ssize_t amdgpu_hwmon_show_temp(struct device *dev,
 	int channel = to_sensor_dev_attr(attr)->index;
 	int r, temp = 0, size = sizeof(temp);
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	if (channel >= PP_TEMP_MAX)
 		return -EINVAL;
 
@@ -2019,6 +2130,9 @@ static ssize_t amdgpu_hwmon_get_pwm1_enable(struct device *dev,
 	u32 pwm_mode = 0;
 	int ret;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = pm_runtime_get_sync(adev->ddev->dev);
 	if (ret < 0)
 		return ret;
@@ -2050,6 +2164,9 @@ static ssize_t amdgpu_hwmon_set_pwm1_enable(struct device *dev,
 	int err, ret;
 	int value;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	err = kstrtoint(buf, 10, &value);
 	if (err)
 		return err;
@@ -2099,6 +2216,9 @@ static ssize_t amdgpu_hwmon_set_pwm1(struct device *dev,
 	u32 value;
 	u32 pwm_mode;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	err = pm_runtime_get_sync(adev->ddev->dev);
 	if (err < 0)
 		return err;
@@ -2148,6 +2268,9 @@ static ssize_t amdgpu_hwmon_get_pwm1(struct device *dev,
 	int err;
 	u32 speed = 0;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	err = pm_runtime_get_sync(adev->ddev->dev);
 	if (err < 0)
 		return err;
@@ -2178,6 +2301,9 @@ static ssize_t amdgpu_hwmon_get_fan1_input(struct device *dev,
 	int err;
 	u32 speed = 0;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	err = pm_runtime_get_sync(adev->ddev->dev);
 	if (err < 0)
 		return err;
@@ -2207,6 +2333,9 @@ static ssize_t amdgpu_hwmon_get_fan1_min(struct device *dev,
 	u32 size = sizeof(min_rpm);
 	int r;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	r = pm_runtime_get_sync(adev->ddev->dev);
 	if (r < 0)
 		return r;
@@ -2232,6 +2361,9 @@ static ssize_t amdgpu_hwmon_get_fan1_max(struct device *dev,
 	u32 size = sizeof(max_rpm);
 	int r;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	r = pm_runtime_get_sync(adev->ddev->dev);
 	if (r < 0)
 		return r;
@@ -2256,6 +2388,9 @@ static ssize_t amdgpu_hwmon_get_fan1_target(struct device *dev,
 	int err;
 	u32 rpm = 0;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	err = pm_runtime_get_sync(adev->ddev->dev);
 	if (err < 0)
 		return err;
@@ -2285,6 +2420,9 @@ static ssize_t amdgpu_hwmon_set_fan1_target(struct device *dev,
 	u32 value;
 	u32 pwm_mode;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	err = pm_runtime_get_sync(adev->ddev->dev);
 	if (err < 0)
 		return err;
@@ -2331,6 +2469,9 @@ static ssize_t amdgpu_hwmon_get_fan1_enable(struct device *dev,
 	u32 pwm_mode = 0;
 	int ret;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	ret = pm_runtime_get_sync(adev->ddev->dev);
 	if (ret < 0)
 		return ret;
@@ -2363,6 +2504,9 @@ static ssize_t amdgpu_hwmon_set_fan1_enable(struct device *dev,
 	int value;
 	u32 pwm_mode;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	err = kstrtoint(buf, 10, &value);
 	if (err)
 		return err;
@@ -2403,6 +2547,9 @@ static ssize_t amdgpu_hwmon_show_vddgfx(struct device *dev,
 	u32 vddgfx;
 	int r, size = sizeof(vddgfx);
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	r = pm_runtime_get_sync(adev->ddev->dev);
 	if (r < 0)
 		return r;
@@ -2435,6 +2582,9 @@ static ssize_t amdgpu_hwmon_show_vddnb(struct device *dev,
 	u32 vddnb;
 	int r, size = sizeof(vddnb);
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	/* only APUs have vddnb */
 	if  (!(adev->flags & AMD_IS_APU))
 		return -EINVAL;
@@ -2472,6 +2622,9 @@ static ssize_t amdgpu_hwmon_show_power_avg(struct device *dev,
 	int r, size = sizeof(u32);
 	unsigned uw;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	r = pm_runtime_get_sync(adev->ddev->dev);
 	if (r < 0)
 		return r;
@@ -2508,6 +2661,9 @@ static ssize_t amdgpu_hwmon_show_power_cap_max(struct device *dev,
 	ssize_t size;
 	int r;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	r = pm_runtime_get_sync(adev->ddev->dev);
 	if (r < 0)
 		return r;
@@ -2537,6 +2693,9 @@ static ssize_t amdgpu_hwmon_show_power_cap(struct device *dev,
 	ssize_t size;
 	int r;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	r = pm_runtime_get_sync(adev->ddev->dev);
 	if (r < 0)
 		return r;
@@ -2567,6 +2726,9 @@ static ssize_t amdgpu_hwmon_set_power_cap(struct device *dev,
 	int err;
 	u32 value;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	if (amdgpu_sriov_vf(adev))
 		return -EINVAL;
 
@@ -2605,6 +2767,9 @@ static ssize_t amdgpu_hwmon_show_sclk(struct device *dev,
 	uint32_t sclk;
 	int r, size = sizeof(sclk);
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	r = pm_runtime_get_sync(adev->ddev->dev);
 	if (r < 0)
 		return r;
@@ -2637,6 +2802,9 @@ static ssize_t amdgpu_hwmon_show_mclk(struct device *dev,
 	uint32_t mclk;
 	int r, size = sizeof(mclk);
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	r = pm_runtime_get_sync(adev->ddev->dev);
 	if (r < 0)
 		return r;
@@ -3497,6 +3665,9 @@ static int amdgpu_debugfs_pm_info(struct seq_file *m, void *data)
 	u32 flags = 0;
 	int r;
 
+	if (adev->in_gpu_reset)
+		return -EPERM;
+
 	r = pm_runtime_get_sync(dev->dev);
 	if (r < 0)
 		return r;