[Intel-gfx] [RFC 01/14] drm/i915: Make i915_check_and_clear_faults take uncore

[Tvrtko Ursulin]

On 11/06/2019 09:52, Chris Wilson wrote:
> Quoting Tvrtko Ursulin (2019-06-11 09:35:07)
>>
>> On 10/06/2019 17:26, Chris Wilson wrote:
>>> Quoting Tvrtko Ursulin (2019-06-10 16:54:06)
>>>> From: Tvrtko Ursulin <tvrtko.ursulin at intel.com>
>>>>
>>>> Continuing the conversion and elimination of implicit dev_priv.
>>>>
>>>> Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin at intel.com>
>>>> Suggested-by: Rodrigo Vivi <rodrigo.vivi at intel.com>
>>>> ---
>>>>    drivers/gpu/drm/i915/gt/intel_engine_cs.c |  2 +-
>>>>    drivers/gpu/drm/i915/gt/intel_reset.c     | 28 ++++++++++++-----------
>>>>    drivers/gpu/drm/i915/gt/intel_reset.h     |  2 +-
>>>>    drivers/gpu/drm/i915/i915_drv.c           |  2 +-
>>>>    drivers/gpu/drm/i915/i915_gem_gtt.c       |  4 ++--
>>>>    5 files changed, 20 insertions(+), 18 deletions(-)
>>>>
>>>> diff --git a/drivers/gpu/drm/i915/gt/intel_engine_cs.c b/drivers/gpu/drm/i915/gt/intel_engine_cs.c
>>>> index c0d986db5a75..a046e8dccc96 100644
>>>> --- a/drivers/gpu/drm/i915/gt/intel_engine_cs.c
>>>> +++ b/drivers/gpu/drm/i915/gt/intel_engine_cs.c
>>>> @@ -453,7 +453,7 @@ int intel_engines_init_mmio(struct drm_i915_private *i915)
>>>>    
>>>>           RUNTIME_INFO(i915)->num_engines = hweight32(mask);
>>>>    
>>>> -       i915_check_and_clear_faults(i915);
>>>> +       i915_check_and_clear_faults(&i915->uncore);
>>>
>>> This name is still setting off red flags for me, but I have to confess
>>> that staring at it, passing uncore does make sense.
>>
>> Rename to intel_uncore_check_and_clear_faults?
>>
>> Or move later in the series as intel_gt_check_and_clear_faults?
> 
> I think I prefer the latter option, intel_gt_check_and_clear_faults.

Yep agreed.

Any comments on the intel_gt.c the series added?

And the end result in i915_gem_init(_hw)?

>>> I just wish we have per-engines faults everywhere and this could be
>>> reduced to passing engine.
>>>
>>> Hmm, this I guess we will just have to revisit in the near future as we
>>> may get the opportunity to put these regs under more scrutiny.
>>>
>>>>    
>>>>           intel_setup_engine_capabilities(i915);
>>>>    
>>>> diff --git a/drivers/gpu/drm/i915/gt/intel_reset.c b/drivers/gpu/drm/i915/gt/intel_reset.c
>>>> index 60d24110af80..13471916559b 100644
>>>> --- a/drivers/gpu/drm/i915/gt/intel_reset.c
>>>> +++ b/drivers/gpu/drm/i915/gt/intel_reset.c
>>>> @@ -1166,10 +1166,10 @@ static void gen8_clear_engine_error_register(struct intel_engine_cs *engine)
>>>>           GEN6_RING_FAULT_REG_POSTING_READ(engine);
>>>>    }
>>>>    
>>>> -static void clear_error_registers(struct drm_i915_private *i915,
>>>> +static void clear_error_registers(struct intel_uncore *uncore,
>>>>                                     intel_engine_mask_t engine_mask)
>>>>    {
>>>> -       struct intel_uncore *uncore = &i915->uncore;
>>>> +       struct drm_i915_private *i915 = uncore_to_i915(uncore);
>>>
>>> Grr, I should have objected to uncore_to_i915() loudly from the
>>> beginning
>>>
>>> What's done is done,
>>
>> Is it too late already? Shouldn't be. My thinking was the implementation
>> can easily be changed if/when backpointer is added (instead of
>> container_of). But if you would prefer we start without a helper, but
>> with a direct access to backpointer straight away that is fine by me.
> 
> I'm optimistic that we can land a split display/gt intel_uncore early
> and so the churn is in the not too distant future.

Okay but that doesn't explicitly answer whether you prefer I just drop 
all the XXX_to_YYY wrappers in favour of using direct pointer dereferences.

You are also in favour of replacing engine->i915 with engine->gt 
straight away?

Regards,

Tvrtko