[PATCH 9/9] drm/ttm: make ttm_bo_get internal

Matthew Brost matthew.brost at intel.com
Tue Jul 16 21:37:29 UTC 2024


On Tue, Jul 16, 2024 at 02:35:19PM +0200, Christian König wrote:
> Prevent drivers from using this directly.
> 

This is a good change. Early on in Xe, our reference counting for BOs
was flawed (and incorrect) due to confusion between GEM ref count and
TTM ref count.

Is there any way we can just eliminate this and use GEM ref count? I'm
still confused about why we seemingly have two reference counts for the
same object.

Matt

> Signed-off-by: Christian König <christian.koenig at amd.com>
> ---
>  drivers/gpu/drm/ttm/ttm_bo_internal.h | 10 ++++++++++
>  include/drm/ttm/ttm_bo.h              | 10 ----------
>  2 files changed, 10 insertions(+), 10 deletions(-)
> 
> diff --git a/drivers/gpu/drm/ttm/ttm_bo_internal.h b/drivers/gpu/drm/ttm/ttm_bo_internal.h
> index 6a7305efd778..9d8b747a34db 100644
> --- a/drivers/gpu/drm/ttm/ttm_bo_internal.h
> +++ b/drivers/gpu/drm/ttm/ttm_bo_internal.h
> @@ -27,6 +27,16 @@
>  
>  #include <drm/ttm/ttm_bo.h>
>  
> +/**
> + * ttm_bo_get - reference a struct ttm_buffer_object
> + *
> + * @bo: The buffer object.
> + */
> +static inline void ttm_bo_get(struct ttm_buffer_object *bo)
> +{
> +	kref_get(&bo->kref);
> +}
> +
>  /**
>   * ttm_bo_get_unless_zero - reference a struct ttm_buffer_object unless
>   * its refcount has already reached zero.
> diff --git a/include/drm/ttm/ttm_bo.h b/include/drm/ttm/ttm_bo.h
> index 31ec7fd34eeb..8c1577d8793c 100644
> --- a/include/drm/ttm/ttm_bo.h
> +++ b/include/drm/ttm/ttm_bo.h
> @@ -229,16 +229,6 @@ struct ttm_lru_walk {
>  s64 ttm_lru_walk_for_evict(struct ttm_lru_walk *walk, struct ttm_device *bdev,
>  			   struct ttm_resource_manager *man, s64 target);
>  
> -/**
> - * ttm_bo_get - reference a struct ttm_buffer_object
> - *
> - * @bo: The buffer object.
> - */
> -static inline void ttm_bo_get(struct ttm_buffer_object *bo)
> -{
> -	kref_get(&bo->kref);
> -}
> -
>  /**
>   * ttm_bo_reserve:
>   *
> -- 
> 2.34.1
> 


More information about the dri-devel mailing list