[Intel-gfx] [PATCH 1/1] drm/i915/guc: s/intel_guc_fw_upload/intel_guc_init_hw/

[Michal Wajdeczko]

On Thu, 01 Mar 2018 11:28:03 +0100, Sagar Arun Kamble  
<sagar.a.kamble at intel.com> wrote:

>
>
> On 3/1/2018 3:36 PM, Michal Wajdeczko wrote:
>> On Thu, 01 Mar 2018 09:18:18 +0100, Sagar Arun Kamble  
>> <sagar.a.kamble at intel.com> wrote:
>>
>>> GuC and HuC get loaded from intel_uc_init_hw. HuC load function is
>>> named intel_huc_init_hw, however GuC load function is still named in
>>> old style as intel_guc_fw_upload. Update it and the function doc. for
>>> both functions.
>>> Move of GuC load function's def. & decl. to intel_guc.c|h seems  
>>> necessary
>>> as it is more about core GuC functionality and not so much about fw  
>>> itself.
>>> This can be done in later patch if needed.
>>>
>>
>> Function intel_guc_fw_upload() was named this way on purpose to follow
>> object-verb naming pattern, where our object is GuC FW (hence file name
>> intel_guc_fw.*)
>>
>> There was a plan to unify this approach with HuC but in the opposite  
>> way:
>> by moving HuC firmware selection code to intel_huc_fw.* but since only
>> one function will be left in intel_huc.c this action was deferred.
>>
> Thanks for background on this.
>> Note that there will be nothing wrong to call fw_upload functions from
>> our uc_init_hw function:
>>
>> intel_uc_init_hw()
>>   intel_uc_reset()
>>   intel_huc_fw_upload()
> Will just do HuC name change (s/intel_huc_init_hw/intel_huc_fw_upload/)  
> and comments update. HuC related move can be done later.
> Is that ok?

Hmm, I've mixed feelings, as on one hand, this small step will unify
fw_upload calls, but at the same time it will break object-verb pattern
in intel_huc.* files ... so maybe we should do it only right?

>> intel_guc_fw_upload()
>>   intel_guc_enable_comm()
>>   intel_huc_auth()
>>
>>
>> /Michal