[PATCH] amdgpu/pm/legacy: fix suspend/resume issues

[Alex Deucher]

From: "chr[]" <chris at rudorff.com>

resume and irq handler happily races in set_power_state()

* amdgpu_legacy_dpm_compute_clocks() needs lock
* protect irq work handler
* fix dpm_enabled usage

v2: fix clang build, integrate Lijo's comments (Alex)

Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/2524
Fixes: 3712e7a49459 ("drm/amd/pm: unified lock protections in amdgpu_dpm.c")
Tested-by: Maciej S. Szmigiero <mail at maciej.szmigiero.name> # on Oland PRO
Signed-off-by: chr[] <chris at rudorff.com>
Signed-off-by: Alex Deucher <alexander.deucher at amd.com>
---
 drivers/gpu/drm/amd/pm/legacy-dpm/kv_dpm.c    | 25 +++++++++++++-----
 .../gpu/drm/amd/pm/legacy-dpm/legacy_dpm.c    |  8 ++++--
 drivers/gpu/drm/amd/pm/legacy-dpm/si_dpm.c    | 26 ++++++++++++++-----
 3 files changed, 45 insertions(+), 14 deletions(-)

diff --git a/drivers/gpu/drm/amd/pm/legacy-dpm/kv_dpm.c b/drivers/gpu/drm/amd/pm/legacy-dpm/kv_dpm.c
index 67a8e22b1126d..e237ea1185a71 100644
--- a/drivers/gpu/drm/amd/pm/legacy-dpm/kv_dpm.c
+++ b/drivers/gpu/drm/amd/pm/legacy-dpm/kv_dpm.c
@@ -3042,6 +3042,7 @@ static int kv_dpm_hw_init(struct amdgpu_ip_block *ip_block)
 	if (!amdgpu_dpm)
 		return 0;
 
+	mutex_lock(&adev->pm.mutex);
 	kv_dpm_setup_asic(adev);
 	ret = kv_dpm_enable(adev);
 	if (ret)
@@ -3049,6 +3050,8 @@ static int kv_dpm_hw_init(struct amdgpu_ip_block *ip_block)
 	else
 		adev->pm.dpm_enabled = true;
 	amdgpu_legacy_dpm_compute_clocks(adev);
+	mutex_unlock(&adev->pm.mutex);
+
 	return ret;
 }
 
@@ -3066,32 +3069,42 @@ static int kv_dpm_suspend(struct amdgpu_ip_block *ip_block)
 {
 	struct amdgpu_device *adev = ip_block->adev;
 
+	cancel_work_sync(&adev->pm.dpm.thermal.work);
+
 	if (adev->pm.dpm_enabled) {
+		mutex_lock(&adev->pm.mutex);
+		adev->pm.dpm_enabled = false;
 		/* disable dpm */
 		kv_dpm_disable(adev);
 		/* reset the power state */
 		adev->pm.dpm.current_ps = adev->pm.dpm.requested_ps = adev->pm.dpm.boot_ps;
+		mutex_unlock(&adev->pm.mutex);
 	}
 	return 0;
 }
 
 static int kv_dpm_resume(struct amdgpu_ip_block *ip_block)
 {
-	int ret;
+	int ret = 0;
 	struct amdgpu_device *adev = ip_block->adev;
 
-	if (adev->pm.dpm_enabled) {
+	if (!amdgpu_dpm)
+		return 0;
+
+	if (!adev->pm.dpm_enabled) {
+		mutex_lock(&adev->pm.mutex);
 		/* asic init will reset to the boot state */
 		kv_dpm_setup_asic(adev);
 		ret = kv_dpm_enable(adev);
-		if (ret)
+		if (ret) {
 			adev->pm.dpm_enabled = false;
-		else
+		} else {
 			adev->pm.dpm_enabled = true;
-		if (adev->pm.dpm_enabled)
 			amdgpu_legacy_dpm_compute_clocks(adev);
+		}
+		mutex_unlock(&adev->pm.mutex);
 	}
-	return 0;
+	return ret;
 }
 
 static bool kv_dpm_is_idle(void *handle)
diff --git a/drivers/gpu/drm/amd/pm/legacy-dpm/legacy_dpm.c b/drivers/gpu/drm/amd/pm/legacy-dpm/legacy_dpm.c
index e861355ebd75b..c7518b13e7879 100644
--- a/drivers/gpu/drm/amd/pm/legacy-dpm/legacy_dpm.c
+++ b/drivers/gpu/drm/amd/pm/legacy-dpm/legacy_dpm.c
@@ -1009,9 +1009,12 @@ void amdgpu_dpm_thermal_work_handler(struct work_struct *work)
 	enum amd_pm_state_type dpm_state = POWER_STATE_TYPE_INTERNAL_THERMAL;
 	int temp, size = sizeof(temp);
 
-	if (!adev->pm.dpm_enabled)
-		return;
+	mutex_lock(&adev->pm.mutex);
 
+	if (!adev->pm.dpm_enabled) {
+		mutex_unlock(&adev->pm.mutex);
+		return;
+	}
 	if (!pp_funcs->read_sensor(adev->powerplay.pp_handle,
 				   AMDGPU_PP_SENSOR_GPU_TEMP,
 				   (void *)&temp,
@@ -1033,4 +1036,5 @@ void amdgpu_dpm_thermal_work_handler(struct work_struct *work)
 	adev->pm.dpm.state = dpm_state;
 
 	amdgpu_legacy_dpm_compute_clocks(adev->powerplay.pp_handle);
+	mutex_unlock(&adev->pm.mutex);
 }
diff --git a/drivers/gpu/drm/amd/pm/legacy-dpm/si_dpm.c b/drivers/gpu/drm/amd/pm/legacy-dpm/si_dpm.c
index a87dcf0974bc1..d6dfe2599ebea 100644
--- a/drivers/gpu/drm/amd/pm/legacy-dpm/si_dpm.c
+++ b/drivers/gpu/drm/amd/pm/legacy-dpm/si_dpm.c
@@ -7786,6 +7786,7 @@ static int si_dpm_hw_init(struct amdgpu_ip_block *ip_block)
 	if (!amdgpu_dpm)
 		return 0;
 
+	mutex_lock(&adev->pm.mutex);
 	si_dpm_setup_asic(adev);
 	ret = si_dpm_enable(adev);
 	if (ret)
@@ -7793,6 +7794,7 @@ static int si_dpm_hw_init(struct amdgpu_ip_block *ip_block)
 	else
 		adev->pm.dpm_enabled = true;
 	amdgpu_legacy_dpm_compute_clocks(adev);
+	mutex_unlock(&adev->pm.mutex);
 	return ret;
 }
 
@@ -7810,32 +7812,44 @@ static int si_dpm_suspend(struct amdgpu_ip_block *ip_block)
 {
 	struct amdgpu_device *adev = ip_block->adev;
 
+	cancel_work_sync(&adev->pm.dpm.thermal.work);
+
 	if (adev->pm.dpm_enabled) {
+		mutex_lock(&adev->pm.mutex);
+		adev->pm.dpm_enabled = false;
 		/* disable dpm */
 		si_dpm_disable(adev);
 		/* reset the power state */
 		adev->pm.dpm.current_ps = adev->pm.dpm.requested_ps = adev->pm.dpm.boot_ps;
+		mutex_unlock(&adev->pm.mutex);
 	}
+
 	return 0;
 }
 
 static int si_dpm_resume(struct amdgpu_ip_block *ip_block)
 {
-	int ret;
+	int ret = 0;
 	struct amdgpu_device *adev = ip_block->adev;
 
-	if (adev->pm.dpm_enabled) {
+	if (!amdgpu_dpm)
+		return 0;
+
+	if (!adev->pm.dpm_enabled) {
 		/* asic init will reset to the boot state */
+		mutex_lock(&adev->pm.mutex);
 		si_dpm_setup_asic(adev);
 		ret = si_dpm_enable(adev);
-		if (ret)
+		if (ret) {
 			adev->pm.dpm_enabled = false;
-		else
+		} else {
 			adev->pm.dpm_enabled = true;
-		if (adev->pm.dpm_enabled)
 			amdgpu_legacy_dpm_compute_clocks(adev);
+		}
+		mutex_unlock(&adev->pm.mutex);
 	}
-	return 0;
+
+	return ret;
 }
 
 static bool si_dpm_is_idle(void *handle)
-- 
2.48.1