[Intel-gfx] [PATCH 09/11] drm/i915: Release vma when the handle is closed

[Tvrtko Ursulin]

On 17/12/15 14:21, Chris Wilson wrote:
> On Thu, Dec 17, 2015 at 01:46:58PM +0000, Tvrtko Ursulin wrote:
>>>   	list_for_each_entry_safe(vma, next, &obj->vma_list, obj_link) {
>>> -		int ret;
>>> -
>>>   		vma->pin_count = 0;
>>> -		ret = i915_vma_unbind(vma);
>>> -		if (WARN_ON(ret == -ERESTARTSYS)) {
>>> -			bool was_interruptible;
>>> -
>>> -			was_interruptible = dev_priv->mm.interruptible;
>>> -			dev_priv->mm.interruptible = false;
>>> -
>>> -			WARN_ON(i915_vma_unbind(vma));
>>> -
>>> -			dev_priv->mm.interruptible = was_interruptible;
>>> -		}
>>> +		i915_vma_close(vma);
>>
>> In what circumstances can there be any VMAs still left unclosed at
>> this point? I thought i915_gem_close_object would had closed them
>> all.
>
>
> diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
> index 688162703070..edfa5ebc4e77 100644
> --- a/drivers/gpu/drm/i915/i915_gem.c
> +++ b/drivers/gpu/drm/i915/i915_gem.c
> @@ -3857,7 +3857,14 @@ void i915_gem_free_object(struct drm_gem_object *gem_obj)
>
>          trace_i915_gem_object_destroy(obj);
>
> +       /* All file-owned VMA should have been released by this point (through
> +        * i915_gem_close_object). However, the object may also be bound into
> +        * the global GTT (e.g. older GPUs without per-process support, or
> +        * for direct access through the GTT either for the user or for
> +        * scanout). Those VMA still need to unbound now.
> +        */
>          list_for_each_entry_safe(vma, next, &obj->vma_list, obj_link) {
> +               RQ_BUG_ON(!i915_is_ggtt(vma->vm));
>                  RQ_BUG_ON(vma->active);
>                  vma->pin_count = 0;
>                  i915_vma_close(vma);
>

Ah yes, I've missed that detail. Very good to have it in a comment (and 
assert).

Regards,

Tvrtko