diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
index 758398bdb39b..14d5b5fa822d 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
@@ -447,6 +447,24 @@ static int kfd_ioctl_set_cu_mask(struct file *filp, struct kfd_process *p,
 	return retval;
 }
 
+static int kfd_ioctl_get_queue_wave_state(struct file *filep,
+					  struct kfd_process *p, void *data)
+{
+	struct kfd_ioctl_get_queue_wave_state_args *args = data;
+	int r;
+
+	mutex_lock(&p->mutex);
+
+	r = pqm_get_wave_state(&p->pqm, args->queue_id,
+			       (void __user *)args->ctl_stack_address,
+			       &args->ctl_stack_used_size,
+			       &args->save_area_used_size);
+
+	mutex_unlock(&p->mutex);
+
+	return r;
+}
+
 static int kfd_ioctl_set_memory_policy(struct file *filep,
 					struct kfd_process *p, void *data)
 {
@@ -1615,6 +1633,9 @@ static const struct amdkfd_ioctl_desc amdkfd_ioctls[] = {
 	AMDKFD_IOCTL_DEF(AMDKFD_IOC_SET_CU_MASK,
 			kfd_ioctl_set_cu_mask, 0),
 
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_GET_QUEUE_WAVE_STATE,
+			kfd_ioctl_get_queue_wave_state, 0)
+
 };
 
 #define AMDKFD_CORE_IOCTL_COUNT	ARRAY_SIZE(amdkfd_ioctls)
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
index ec0d62a16e53..408888911361 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
@@ -1528,6 +1528,41 @@ static int process_termination_nocpsch(struct device_queue_manager *dqm,
 	return retval;
 }
 
+static int get_wave_state(struct device_queue_manager *dqm,
+			  struct queue *q,
+			  void __user *ctl_stack,
+			  u32 *ctl_stack_used_size,
+			  u32 *save_area_used_size)
+{
+	struct mqd_manager *mqd;
+	int r;
+
+	dqm_lock(dqm);
+
+	if (q->properties.type != KFD_QUEUE_TYPE_COMPUTE ||
+	    q->properties.is_active || !q->device->cwsr_enabled) {
+		r = -EINVAL;
+		goto dqm_unlock;
+	}
+
+	mqd = dqm->ops.get_mqd_manager(dqm, KFD_MQD_TYPE_COMPUTE);
+	if (!mqd) {
+		r = -ENOMEM;
+		goto dqm_unlock;
+	}
+
+	if (!mqd->get_wave_state) {
+		r = -EINVAL;
+		goto dqm_unlock;
+	}
+
+	r = mqd->get_wave_state(mqd, q->mqd, ctl_stack, ctl_stack_used_size,
+				save_area_used_size);
+
+dqm_unlock:
+	dqm_unlock(dqm);
+	return r;
+}
 
 static int process_termination_cpsch(struct device_queue_manager *dqm,
 		struct qcm_process_device *qpd)
@@ -1649,6 +1684,7 @@ struct device_queue_manager *device_queue_manager_init(struct kfd_dev *dev)
 		dqm->ops.process_termination = process_termination_cpsch;
 		dqm->ops.evict_process_queues = evict_process_queues_cpsch;
 		dqm->ops.restore_process_queues = restore_process_queues_cpsch;
+		dqm->ops.get_wave_state = get_wave_state;
 		break;
 	case KFD_SCHED_POLICY_NO_HWS:
 		/* initialize dqm for no cp scheduling */
@@ -1668,6 +1704,7 @@ struct device_queue_manager *device_queue_manager_init(struct kfd_dev *dev)
 		dqm->ops.evict_process_queues = evict_process_queues_nocpsch;
 		dqm->ops.restore_process_queues =
 			restore_process_queues_nocpsch;
+		dqm->ops.get_wave_state = get_wave_state;
 		break;
 	default:
 		pr_err("Invalid scheduling policy %d\n", dqm->sched_policy);
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h
index 00da3169a004..e7bd19d09845 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h
@@ -82,6 +82,8 @@ struct device_process_node {
  *
  * @restore_process_queues: Restore all evicted queues queues of a process
  *
+ * @get_wave_state: Retrieves context save state and optionally copies the
+ * control stack, if kept in the MQD, to the given userspace address.
  */
 
 struct device_queue_manager_ops {
@@ -137,6 +139,12 @@ struct device_queue_manager_ops {
 				    struct qcm_process_device *qpd);
 	int (*restore_process_queues)(struct device_queue_manager *dqm,
 				      struct qcm_process_device *qpd);
+
+	int	(*get_wave_state)(struct device_queue_manager *dqm,
+				  struct queue *q,
+				  void __user *ctl_stack,
+				  u32 *ctl_stack_used_size,
+				  u32 *save_area_used_size);
 };
 
 struct device_queue_manager_asic_ops {
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.h b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.h
index 4e84052d4e21..f8261313ae7b 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.h
@@ -43,6 +43,9 @@
  *
  * @is_occupied: Checks if the relevant HQD slot is occupied.
  *
+ * @get_wave_state: Retrieves context save state and optionally copies the
+ * control stack, if kept in the MQD, to the given userspace address.
+ *
  * @mqd_mutex: Mqd manager mutex.
  *
  * @dev: The kfd device structure coupled with this module.
@@ -85,6 +88,11 @@ struct mqd_manager {
 				uint64_t queue_address,	uint32_t pipe_id,
 				uint32_t queue_id);
 
+	int	(*get_wave_state)(struct mqd_manager *mm, void *mqd,
+				  void __user *ctl_stack,
+				  u32 *ctl_stack_used_size,
+				  u32 *save_area_used_size);
+
 #if defined(CONFIG_DEBUG_FS)
 	int	(*debugfs_show_mqd)(struct seq_file *m, void *data);
 #endif
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c
index 0cedb37cf513..f381c1cb27bd 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c
@@ -266,6 +266,28 @@ static bool is_occupied(struct mqd_manager *mm, void *mqd,
 		pipe_id, queue_id);
 }
 
+static int get_wave_state(struct mqd_manager *mm, void *mqd,
+			  void __user *ctl_stack,
+			  u32 *ctl_stack_used_size,
+			  u32 *save_area_used_size)
+{
+	struct v9_mqd *m;
+
+	/* Control stack is located one page after MQD. */
+	void *mqd_ctl_stack = (void *)((uintptr_t)mqd + PAGE_SIZE);
+
+	m = get_mqd(mqd);
+
+	*ctl_stack_used_size = m->cp_hqd_cntl_stack_size -
+		m->cp_hqd_cntl_stack_offset;
+	*save_area_used_size = m->cp_hqd_wg_state_offset;
+
+	if (copy_to_user(ctl_stack, mqd_ctl_stack, m->cp_hqd_cntl_stack_size))
+		return -EFAULT;
+
+	return 0;
+}
+
 static int init_mqd_hiq(struct mqd_manager *mm, void **mqd,
 			struct kfd_mem_obj **mqd_mem_obj, uint64_t *gart_addr,
 			struct queue_properties *q)
@@ -435,6 +457,7 @@ struct mqd_manager *mqd_manager_init_v9(enum KFD_MQD_TYPE type,
 		mqd->update_mqd = update_mqd;
 		mqd->destroy_mqd = destroy_mqd;
 		mqd->is_occupied = is_occupied;
+		mqd->get_wave_state = get_wave_state;
 #if defined(CONFIG_DEBUG_FS)
 		mqd->debugfs_show_mqd = debugfs_show_mqd;
 #endif
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_vi.c b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_vi.c
index b81fda3754da..6469b3456f00 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_vi.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_vi.c
@@ -269,6 +269,28 @@ static bool is_occupied(struct mqd_manager *mm, void *mqd,
 		pipe_id, queue_id);
 }
 
+static int get_wave_state(struct mqd_manager *mm, void *mqd,
+			  void __user *ctl_stack,
+			  u32 *ctl_stack_used_size,
+			  u32 *save_area_used_size)
+{
+	struct vi_mqd *m;
+
+	m = get_mqd(mqd);
+
+	*ctl_stack_used_size = m->cp_hqd_cntl_stack_size -
+		m->cp_hqd_cntl_stack_offset;
+	*save_area_used_size = m->cp_hqd_wg_state_offset -
+		m->cp_hqd_cntl_stack_size;
+
+	/* Control stack is not copied to user mode for GFXv8 because
+	 * it's part of the context save area that is already
+	 * accessible to user mode
+	 */
+
+	return 0;
+}
+
 static int init_mqd_hiq(struct mqd_manager *mm, void **mqd,
 			struct kfd_mem_obj **mqd_mem_obj, uint64_t *gart_addr,
 			struct queue_properties *q)
@@ -436,6 +458,7 @@ struct mqd_manager *mqd_manager_init_vi(enum KFD_MQD_TYPE type,
 		mqd->update_mqd = update_mqd;
 		mqd->destroy_mqd = destroy_mqd;
 		mqd->is_occupied = is_occupied;
+		mqd->get_wave_state = get_wave_state;
 #if defined(CONFIG_DEBUG_FS)
 		mqd->debugfs_show_mqd = debugfs_show_mqd;
 #endif
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
index 6f3a5bd489bd..968098bf76dc 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
@@ -862,6 +862,11 @@ int pqm_set_cu_mask(struct process_queue_manager *pqm, unsigned int qid,
 			struct queue_properties *p);
 struct kernel_queue *pqm_get_kernel_queue(struct process_queue_manager *pqm,
 						unsigned int qid);
+int pqm_get_wave_state(struct process_queue_manager *pqm,
+		       unsigned int qid,
+		       void __user *ctl_stack,
+		       u32 *ctl_stack_used_size,
+		       u32 *save_area_used_size);
 
 int amdkfd_fence_wait_timeout(unsigned int *fence_addr,
 				unsigned int fence_value,
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c
index c8cad9c078ae..fcaaf93681ac 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c
@@ -408,6 +408,28 @@ struct kernel_queue *pqm_get_kernel_queue(
 	return NULL;
 }
 
+int pqm_get_wave_state(struct process_queue_manager *pqm,
+		       unsigned int qid,
+		       void __user *ctl_stack,
+		       u32 *ctl_stack_used_size,
+		       u32 *save_area_used_size)
+{
+	struct process_queue_node *pqn;
+
+	pqn = get_queue_by_qid(pqm, qid);
+	if (!pqn) {
+		pr_debug("amdkfd: No queue %d exists for operation\n",
+			 qid);
+		return -EFAULT;
+	}
+
+	return pqn->q->device->dqm->ops.get_wave_state(pqn->q->device->dqm,
+						       pqn->q,
+						       ctl_stack,
+						       ctl_stack_used_size,
+						       save_area_used_size);
+}
+
 #if defined(CONFIG_DEBUG_FS)
 
 int pqm_debugfs_mqds(struct seq_file *m, void *data)
diff --git a/include/uapi/linux/kfd_ioctl.h b/include/uapi/linux/kfd_ioctl.h
index 01674b56e14f..f5ff8a76e208 100644
--- a/include/uapi/linux/kfd_ioctl.h
+++ b/include/uapi/linux/kfd_ioctl.h
@@ -82,6 +82,14 @@ struct kfd_ioctl_set_cu_mask_args {
 	__u64 cu_mask_ptr;		/* to KFD */
 };
 
+struct kfd_ioctl_get_queue_wave_state_args {
+	uint64_t ctl_stack_address;	/* to KFD */
+	uint32_t ctl_stack_used_size;	/* from KFD */
+	uint32_t save_area_used_size;	/* from KFD */
+	uint32_t queue_id;		/* to KFD */
+	uint32_t pad;
+};
+
 /* For kfd_ioctl_set_memory_policy_args.default_policy and alternate_policy */
 #define KFD_IOC_CACHE_POLICY_COHERENT 0
 #define KFD_IOC_CACHE_POLICY_NONCOHERENT 1
@@ -475,7 +483,10 @@ struct kfd_ioctl_unmap_memory_from_gpu_args {
 #define AMDKFD_IOC_SET_CU_MASK		\
 		AMDKFD_IOW(0x1A, struct kfd_ioctl_set_cu_mask_args)
 
+#define AMDKFD_IOC_GET_QUEUE_WAVE_STATE		\
+		AMDKFD_IOWR(0x1B, struct kfd_ioctl_get_queue_wave_state_args)
+
 #define AMDKFD_COMMAND_START		0x01
-#define AMDKFD_COMMAND_END		0x1B
+#define AMDKFD_COMMAND_END		0x1C
 
 #endif