[PATCH] drm/amdkfd: Move set_trap_handler out of dqm->ops

[Felix Kuehling]

Am 2021-03-04 um 10:12 p.m. schrieb Jay Cornwall:
> Trap handler is set per-process per-device and is unrelated
> to queue management.
>
> Move implementation closer to TMA setup code.
>
> Signed-off-by: Jay Cornwall <jay.cornwall at amd.com>

Reviewed-by: Felix Kuehling <Felix.Kuehling at amd.com>


> ---
>  drivers/gpu/drm/amd/amdkfd/kfd_chardev.c      |  6 +----
>  .../drm/amd/amdkfd/kfd_device_queue_manager.c | 22 -------------------
>  .../drm/amd/amdkfd/kfd_device_queue_manager.h |  5 -----
>  drivers/gpu/drm/amd/amdkfd/kfd_priv.h         |  4 ++++
>  drivers/gpu/drm/amd/amdkfd/kfd_process.c      | 19 ++++++++++++++++
>  5 files changed, 24 insertions(+), 32 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
> index 8cc51cec988a..6802c616e10e 100644
> --- a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
> +++ b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
> @@ -556,11 +556,7 @@ static int kfd_ioctl_set_trap_handler(struct file *filep,
>  		goto out;
>  	}
>  
> -	if (dev->dqm->ops.set_trap_handler(dev->dqm,
> -					&pdd->qpd,
> -					args->tba_addr,
> -					args->tma_addr))
> -		err = -EINVAL;
> +	kfd_process_set_trap_handler(&pdd->qpd, args->tba_addr, args->tma_addr);
>  
>  out:
>  	mutex_unlock(&p->mutex);
> 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 c37e9c4b1fb4..6bb778f24441 100644
> --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
> +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
> @@ -1596,26 +1596,6 @@ static bool set_cache_memory_policy(struct device_queue_manager *dqm,
>  	return retval;
>  }
>  
> -static int set_trap_handler(struct device_queue_manager *dqm,
> -				struct qcm_process_device *qpd,
> -				uint64_t tba_addr,
> -				uint64_t tma_addr)
> -{
> -	uint64_t *tma;
> -
> -	if (dqm->dev->cwsr_enabled) {
> -		/* Jump from CWSR trap handler to user trap */
> -		tma = (uint64_t *)(qpd->cwsr_kaddr + KFD_CWSR_TMA_OFFSET);
> -		tma[0] = tba_addr;
> -		tma[1] = tma_addr;
> -	} else {
> -		qpd->tba_addr = tba_addr;
> -		qpd->tma_addr = tma_addr;
> -	}
> -
> -	return 0;
> -}
> -
>  static int process_termination_nocpsch(struct device_queue_manager *dqm,
>  		struct qcm_process_device *qpd)
>  {
> @@ -1859,7 +1839,6 @@ struct device_queue_manager *device_queue_manager_init(struct kfd_dev *dev)
>  		dqm->ops.create_kernel_queue = create_kernel_queue_cpsch;
>  		dqm->ops.destroy_kernel_queue = destroy_kernel_queue_cpsch;
>  		dqm->ops.set_cache_memory_policy = set_cache_memory_policy;
> -		dqm->ops.set_trap_handler = set_trap_handler;
>  		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;
> @@ -1878,7 +1857,6 @@ struct device_queue_manager *device_queue_manager_init(struct kfd_dev *dev)
>  		dqm->ops.initialize = initialize_nocpsch;
>  		dqm->ops.uninitialize = uninitialize;
>  		dqm->ops.set_cache_memory_policy = set_cache_memory_policy;
> -		dqm->ops.set_trap_handler = set_trap_handler;
>  		dqm->ops.process_termination = process_termination_nocpsch;
>  		dqm->ops.evict_process_queues = evict_process_queues_nocpsch;
>  		dqm->ops.restore_process_queues =
> 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 16262e5d93f5..aee033b1d148 100644
> --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h
> +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h
> @@ -121,11 +121,6 @@ struct device_queue_manager_ops {
>  					   void __user *alternate_aperture_base,
>  					   uint64_t alternate_aperture_size);
>  
> -	int	(*set_trap_handler)(struct device_queue_manager *dqm,
> -				    struct qcm_process_device *qpd,
> -				    uint64_t tba_addr,
> -				    uint64_t tma_addr);
> -
>  	int (*process_termination)(struct device_queue_manager *dqm,
>  			struct qcm_process_device *qpd);
>  
> diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
> index e2ebd5a1d4de..8f839154bf1f 100644
> --- a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
> +++ b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
> @@ -944,6 +944,10 @@ bool interrupt_is_wanted(struct kfd_dev *dev,
>  /* amdkfd Apertures */
>  int kfd_init_apertures(struct kfd_process *process);
>  
> +void kfd_process_set_trap_handler(struct qcm_process_device *qpd,
> +				  uint64_t tba_addr,
> +				  uint64_t tma_addr);
> +
>  /* Queue Context Management */
>  int init_queue(struct queue **q, const struct queue_properties *properties);
>  void uninit_queue(struct queue *q);
> diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
> index a4d7682289bb..ff3e76450b66 100644
> --- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c
> +++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
> @@ -1162,6 +1162,25 @@ static int kfd_process_device_init_cwsr_dgpu(struct kfd_process_device *pdd)
>  	return 0;
>  }
>  
> +void kfd_process_set_trap_handler(struct qcm_process_device *qpd,
> +				  uint64_t tba_addr,
> +				  uint64_t tma_addr)
> +{
> +	if (qpd->cwsr_kaddr) {
> +		/* KFD trap handler is bound, record as second-level TBA/TMA
> +		 * in first-level TMA. First-level trap will jump to second.
> +		 */
> +		uint64_t *tma =
> +			(uint64_t *)(qpd->cwsr_kaddr + KFD_CWSR_TMA_OFFSET);
> +		tma[0] = tba_addr;
> +		tma[1] = tma_addr;
> +	} else {
> +		/* No trap handler bound, bind as first-level TBA/TMA. */
> +		qpd->tba_addr = tba_addr;
> +		qpd->tma_addr = tma_addr;
> +	}
> +}
> +
>  /*
>   * On return the kfd_process is fully operational and will be freed when the
>   * mm is released