drm/amd/pm: fix race in power state check before mutex lock

The power state check in amdgpu_dpm_set_powergating_by_smu() is done
before acquiring the pm mutex, leading to a race condition where:
1. Thread A checks state and thinks no change is needed
2. Thread B acquires mutex and modifies the state
3. Thread A returns without updating state, causing inconsistency

Fix this by moving the mutex lock before the power state check,
ensuring atomicity of the state check and modification.

Fixes: 6ee27ee27ba8 ("drm/amd/pm: avoid duplicate powergate/ungate setting")
Signed-off-by: Yang Wang <[e-mail]>
Reviewed-by: Kenneth Feng <[e-mail]>
Signed-off-by: Alex Deucher <[e-mail]>
(cherry picked from commit 7a3fbdfd19ec5992c0fc2d0bd83888644f5f2f38)

drivers/gpu/drm/amd/pm/amdgpu_dpm.c

···
 	enum ip_power_state pwr_state = gate ? POWER_STATE_OFF : POWER_STATE_ON;
 	bool is_vcn = block_type == AMD_IP_BLOCK_TYPE_VCN;
 
+	mutex_lock(&adev->pm.mutex);
+
 	if (atomic_read(&adev->pm.pwr_state[block_type]) == pwr_state &&
 			(!is_vcn || adev->vcn.num_vcn_inst == 1)) {
 		dev_dbg(adev->dev, "IP block%d already in the target %s state!",
 				block_type, gate ? "gate" : "ungate");
-		return 0;
+		goto out_unlock;
 	}
-
-	mutex_lock(&adev->pm.mutex);
 
 	switch (block_type) {
 	case AMD_IP_BLOCK_TYPE_UVD:
···
 	if (!ret)
 		atomic_set(&adev->pm.pwr_state[block_type], pwr_state);
 
+out_unlock:
 	mutex_unlock(&adev->pm.mutex);
 
 	return ret;