[Intel-gfx] [PATCH 14/40] drm/i915: Load balancing across a virtual engine

[Tvrtko Ursulin]

On 08/05/2019 12:17, Chris Wilson wrote:
> Quoting Tvrtko Ursulin (2019-05-08 11:29:34)
>>
>> On 08/05/2019 09:06, Chris Wilson wrote:
>>> +static int live_virtual_engine(void *arg)
>>> +{
>>> +     struct drm_i915_private *i915 = arg;
>>> +     struct intel_engine_cs *siblings[MAX_ENGINE_INSTANCE + 1];
>>> +     struct intel_engine_cs *engine;
>>> +     enum intel_engine_id id;
>>> +     unsigned int class, inst;
>>> +     int err = -ENODEV;
>>> +
>>> +     if (USES_GUC_SUBMISSION(i915))
>>> +             return 0;
>>> +
>>> +     mutex_lock(&i915->drm.struct_mutex);
>>> +
>>> +     for_each_engine(engine, i915, id) {
>>> +             err = nop_virtual_engine(i915, &engine, 1, 1, 0);
>>> +             if (err) {
>>> +                     pr_err("Failed to wrap engine %s: err=%d\n",
>>> +                            engine->name, err);
>>> +                     goto out_unlock;
>>> +             }
>>> +     }
>>> +
>>> +     for (class = 0; class <= MAX_ENGINE_CLASS; class++) {
>>> +             int nsibling, n;
>>> +
>>> +             nsibling = 0;
>>> +             for (inst = 0; inst <= MAX_ENGINE_INSTANCE; inst++) {
>>> +                     if (!i915->engine_class[class][inst])
>>> +                             break;
>>
>> I previous review I said I think this should be continue instead of
>> break so vcs0 + vcs2 skus can also be tested.
> 
> Completely missed that, sorry.
> 
>>> +
>>> +                     siblings[nsibling++] = i915->engine_class[class][inst];
>>> +             }
>>> +             if (nsibling < 2)
>>> +                     continue;
>>
>> And also that single engine VE could be tested just as well, unless I am
>> missing something.
> 
> There's no such thing as single engine VE. The current design requires
> that this type of struct virtual_engine encompasses more than one engine
> (failing that we break the single request scheduling, although might be
> able to lift that with timeslicing but the early results were not
> favourable); the single engine being a regular intel_context instance.

Yeah my bad, the auto-magic replacement with physical engine happens one 
level higher than what this selftest is operating on.

Regards,

Tvrtko