Commit 5df099e8 authored by Jay Cornwall's avatar Jay Cornwall Committed by Alex Deucher
Browse files

drm/amdkfd: Add wavefront context save state retrieval ioctl 



Wavefront context save data is of interest to userspace clients for
debugging static wavefront state. The MQD contains two parameters
required to parse the control stack and the control stack itself
is kept in the MQD from gfx9 onwards.

Add an ioctl to fetch the context save area and control stack offsets
and to copy the control stack to a userspace address if it is kept in
the MQD.

Signed-off-by: default avatarJay Cornwall <Jay.Cornwall@amd.com>
Reviewed-by: default avatarFelix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: default avatarFelix Kuehling <Felix.Kuehling@amd.com>
Acked-by: default avatarAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: default avatarAlex Deucher <alexander.deucher@amd.com>
parent 5ade6c9c

drivers/gpu/drm/amd/amdkfd/kfd_chardev.c

+21 −0
Original line number Diff line number Diff line
@@ -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)

drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c

+37 −0
Original line number Diff line number Diff line
@@ -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);

drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h

+8 −0
Original line number Diff line number Diff line
@@ -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 {

drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.h

+8 −0
Original line number Diff line number Diff line
@@ -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

drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c

+23 −0
Original line number Diff line number Diff line
@@ -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

CRA Git | Maintained and supported by SUSTech CRA and CCSE