[Intel-gfx] [PATCH v3 04/63] drm/i915: Pin timeline map after first timeline pin, v3.

[Thomas Hellström (Intel)]

On 10/27/20 5:25 PM, Thomas Hellström (Intel) wrote:
>
>>>> +
>>>> +    if (WARN_ON(!i915_gem_object_trylock(tl->hwsp_ggtt->obj)))
>>>> +        return -EBUSY;
>>> I think we should either annotate this properly as an isolated lock, 
>>> or allow a silent -EBUSY.
>> This is done in a controlled selftest where we mock the entire i915 
>> device, so normally this can't happen. :)
>
> But if it does, we get erratic BAT failures, and if it indeed doesn't 
> happen, then an annotated isolated lock would guarantee that.

...but if we can't do that since we're actually not locking isolated, I 
guess we have to leave it as is. (I presume I can't make you add /* Here 
we should have needed a i915_do_ww() */ :D)

/Thomas

>
> /Thomas
>