Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1634856 > unrolled thread

Re: [PATCH v2 4/5] drm/etnaviv: Reuse dma_fence_release.

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: [PATCH v2 4/5] drm/etnaviv: Reuse dma_fence_release. Lucas Stach <l.stach@pengutronix.de> - 2017-05-03 12:30 +0200
    Re: [PATCH v2 4/5] drm/etnaviv: Reuse dma_fence_release. Eric Anholt <eric@anholt.net> - 2017-05-03 18:10 +0200

#1634856 — Re: [PATCH v2 4/5] drm/etnaviv: Reuse dma_fence_release.

Hi Eric,

Am Mittwoch, den 12.04.2017, 12:12 -0700 schrieb Eric Anholt:
> If we follow the typical pattern of the base class being the first
> member, we can use the default dma_fence_free function.

Sorry, I don't like this change.
While it provides a bit of code simplification, it also bakes the
implicit assumption into the code that dma_fence is the first struct
member without any checks to validate this assumption. I would at least
expect a build bug assert, but then the current bit of code to keep this
explicit and self-documenting is probably just as good.

Regards,
Lucas

> Signed-off-by: Eric Anholt <eric@anholt.net>
> Cc: Lucas Stach <l.stach@pengutronix.de>
> Cc: Russell King <linux+etnaviv@armlinux.org.uk>
> Cc: Christian Gmeiner <christian.gmeiner@gmail.com>
> Cc: etnaviv@lists.freedesktop.org
> ---
>  drivers/gpu/drm/etnaviv/etnaviv_gpu.c | 11 ++---------
>  1 file changed, 2 insertions(+), 9 deletions(-)
> 
> diff --git a/drivers/gpu/drm/etnaviv/etnaviv_gpu.c b/drivers/gpu/drm/etnaviv/etnaviv_gpu.c
> index da48819ff2e6..0d26ca56e94b 100644
> --- a/drivers/gpu/drm/etnaviv/etnaviv_gpu.c
> +++ b/drivers/gpu/drm/etnaviv/etnaviv_gpu.c
> @@ -998,8 +998,8 @@ static void hangcheck_disable(struct etnaviv_gpu *gpu)
>  
>  /* fence object management */
>  struct etnaviv_fence {
> -	struct etnaviv_gpu *gpu;
>  	struct dma_fence base;
> +	struct etnaviv_gpu *gpu;
>  };
>  
>  static inline struct etnaviv_fence *to_etnaviv_fence(struct dma_fence *fence)
> @@ -1031,20 +1031,13 @@ static bool etnaviv_fence_signaled(struct dma_fence *fence)
>  	return fence_completed(f->gpu, f->base.seqno);
>  }
>  
> -static void etnaviv_fence_release(struct dma_fence *fence)
> -{
> -	struct etnaviv_fence *f = to_etnaviv_fence(fence);
> -
> -	kfree_rcu(f, base.rcu);
> -}
> -
>  static const struct dma_fence_ops etnaviv_fence_ops = {
>  	.get_driver_name = etnaviv_fence_get_driver_name,
>  	.get_timeline_name = etnaviv_fence_get_timeline_name,
>  	.enable_signaling = etnaviv_fence_enable_signaling,
>  	.signaled = etnaviv_fence_signaled,
>  	.wait = dma_fence_default_wait,
> -	.release = etnaviv_fence_release,
> +	.release = dma_fence_free,
>  };
>  
>  static struct dma_fence *etnaviv_gpu_fence_alloc(struct etnaviv_gpu *gpu)

[toc] | [next] | [standalone]

#1635113

[Multipart message — attachments visible in raw view] — view raw

Lucas Stach <l.stach@pengutronix.de> writes:

> Hi Eric,
>
> Am Mittwoch, den 12.04.2017, 12:12 -0700 schrieb Eric Anholt:
>> If we follow the typical pattern of the base class being the first
>> member, we can use the default dma_fence_free function.
>
> Sorry, I don't like this change.
> While it provides a bit of code simplification, it also bakes the
> implicit assumption into the code that dma_fence is the first struct
> member without any checks to validate this assumption. I would at least
> expect a build bug assert, but then the current bit of code to keep this
> explicit and self-documenting is probably just as good.

Fine with me.  Please make sure you grab patch 2, though.

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web