On Sun, 23 Jun 2019 at 11:36, Sam Ravnborg <sam@xxxxxxxxxxxx> wrote:
> -int mga_driver_fence_wait(struct drm_device *dev, unsigned int *sequence)
> +void mga_driver_fence_wait(struct drm_device *dev, unsigned int *sequence)
> {
> drm_mga_private_t *dev_priv = (drm_mga_private_t *) dev->dev_private;
> unsigned int cur_fence;
> - int ret = 0;
>
> /* Assume that the user has missed the current sequence number
> * by about a day rather than she wants to wait for years
> * using fences.
> */
> - DRM_WAIT_ON(ret, dev_priv->fence_queue, 3 * HZ,
> + wait_event_timeout(dev_priv->fence_queue,
> (((cur_fence = atomic_read(&dev_priv->last_fence_retired))
> - - *sequence) <= (1 << 23)));
> + - *sequence) <= (1 << 23)),
> + msecs_to_jiffies(3000));
>
> *sequence = cur_fence;
> -
> - return ret;
> }
>
Most of the patch is a trivial substitution. This piece is not though :-\
For the future, please keep mechanical and functional changes in
separate patches.
Thanks
Emil
_______________________________________________
dri-devel mailing list
dri-devel@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/dri-devel
