Re: [Intel-gfx] [RESEND] drm/i915: hide mkwrite_device_info() better

[Jani Nikula <jani.nikula@intel.com>]

On Thu, 13 Apr 2023, Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> wrote:
> On 11/04/2023 11:56, Jani Nikula wrote:
>> The goal has been to just make device info a pointer to static const
>> data, i.e. the static const structs in i915_pci.c. See [1]. However,
>> there were issues with intel_device_info_runtime_init() clearing the
>> display sub-struct of device info on the !HAS_DISPLAY() path, which
>> consequently disables a lot of display functionality, like it
>> should. Looks like we'd have to cover all those paths, and maybe
>> sprinkle HAS_DISPLAY() checks in them, which we haven't gotten around
>> to.
>> 
>> In the mean time, hide mkwrite_device_info() better within
>> intel_device_info.c by adding a intel_device_info_driver_create() for
>> the very early initialization of the device info and initial runtime
>> info. This also lets us declutter i915_drv.h a bit, and stops promoting
>> mkwrite_device_info() as something that could be used.
>> 
>> [1] https://lore.kernel.org/r/a0422f0a8ac055f65b7922bcd3119b180a41e79e.1655712106.git.jani.nikula@intel.com
>> 
>> Signed-off-by: Jani Nikula <jani.nikula@intel.com>
>> ---
>>   drivers/gpu/drm/i915/i915_driver.c       | 12 ++--------
>>   drivers/gpu/drm/i915/i915_drv.h          |  7 ------
>>   drivers/gpu/drm/i915/intel_device_info.c | 29 ++++++++++++++++++++++++
>>   drivers/gpu/drm/i915/intel_device_info.h |  2 ++
>>   4 files changed, 33 insertions(+), 17 deletions(-)
>> 
>> diff --git a/drivers/gpu/drm/i915/i915_driver.c b/drivers/gpu/drm/i915/i915_driver.c
>> index 93fdc40d724f..2980ccdef6cd 100644
>> --- a/drivers/gpu/drm/i915/i915_driver.c
>> +++ b/drivers/gpu/drm/i915/i915_driver.c
>> @@ -720,8 +720,6 @@ i915_driver_create(struct pci_dev *pdev, const struct pci_device_id *ent)
>>   {
>>   	const struct intel_device_info *match_info =
>>   		(struct intel_device_info *)ent->driver_data;
>> -	struct intel_device_info *device_info;
>> -	struct intel_runtime_info *runtime;
>>   	struct drm_i915_private *i915;
>>   
>>   	i915 = devm_drm_dev_alloc(&pdev->dev, &i915_drm_driver,
>> @@ -734,14 +732,8 @@ i915_driver_create(struct pci_dev *pdev, const struct pci_device_id *ent)
>>   	/* Device parameters start as a copy of module parameters. */
>>   	i915_params_copy(&i915->params, &i915_modparams);
>>   
>> -	/* Setup the write-once "constant" device info */
>> -	device_info = mkwrite_device_info(i915);
>> -	memcpy(device_info, match_info, sizeof(*device_info));
>> -
>> -	/* Initialize initial runtime info from static const data and pdev. */
>> -	runtime = RUNTIME_INFO(i915);
>> -	memcpy(runtime, &INTEL_INFO(i915)->__runtime, sizeof(*runtime));
>> -	runtime->device_id = pdev->device;
>> +	/* Set up device info and initial runtime info. */
>> +	intel_device_info_driver_create(i915, pdev->device, match_info);
>>   
>>   	return i915;
>>   }
>> diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
>> index e771fdc3099c..fe7eeafe9cff 100644
>> --- a/drivers/gpu/drm/i915/i915_drv.h
>> +++ b/drivers/gpu/drm/i915/i915_drv.h
>> @@ -931,11 +931,4 @@ IS_SUBPLATFORM(const struct drm_i915_private *i915,
>>   #define HAS_LMEMBAR_SMEM_STOLEN(i915) (!HAS_LMEM(i915) && \
>>   				       GRAPHICS_VER_FULL(i915) >= IP_VER(12, 70))
>>   
>> -/* intel_device_info.c */
>> -static inline struct intel_device_info *
>> -mkwrite_device_info(struct drm_i915_private *dev_priv)
>> -{
>> -	return (struct intel_device_info *)INTEL_INFO(dev_priv);
>> -}
>> -
>>   #endif
>> diff --git a/drivers/gpu/drm/i915/intel_device_info.c b/drivers/gpu/drm/i915/intel_device_info.c
>> index fc5cd14adfcc..4e23be2995bf 100644
>> --- a/drivers/gpu/drm/i915/intel_device_info.c
>> +++ b/drivers/gpu/drm/i915/intel_device_info.c
>> @@ -381,6 +381,13 @@ void intel_device_info_runtime_init_early(struct drm_i915_private *i915)
>>   	intel_device_info_subplatform_init(i915);
>>   }
>>   
>> +/* FIXME: Remove this, and make device info a const pointer to rodata. */
>> +static struct intel_device_info *
>> +mkwrite_device_info(struct drm_i915_private *i915)
>> +{
>> +	return (struct intel_device_info *)INTEL_INFO(i915);
>> +}
>> +
>>   /**
>>    * intel_device_info_runtime_init - initialize runtime info
>>    * @dev_priv: the i915 device
>> @@ -548,6 +555,28 @@ void intel_device_info_runtime_init(struct drm_i915_private *dev_priv)
>>   		dev_priv->drm.driver_features &= ~DRIVER_ATOMIC;
>>   }
>>   
>> +/*
>> + * Set up device info and initial runtime info at driver create.
>> + *
>> + * Note: i915 is only an allocated blob of memory at this point.
>> + */
>> +void intel_device_info_driver_create(struct drm_i915_private *i915,
>> +				     u16 device_id,
>> +				     const struct intel_device_info *match_info)
>> +{
>> +	struct intel_device_info *info;
>> +	struct intel_runtime_info *runtime;
>> +
>> +	/* Setup the write-once "constant" device info */
>> +	info = mkwrite_device_info(i915);
>> +	memcpy(info, match_info, sizeof(*info));
>> +
>> +	/* Initialize initial runtime info from static const data and pdev. */
>> +	runtime = RUNTIME_INFO(i915);
>> +	memcpy(runtime, &INTEL_INFO(i915)->__runtime, sizeof(*runtime));
>> +	runtime->device_id = device_id;
>> +}
>> +
>>   void intel_driver_caps_print(const struct intel_driver_caps *caps,
>>   			     struct drm_printer *p)
>>   {
>> diff --git a/drivers/gpu/drm/i915/intel_device_info.h b/drivers/gpu/drm/i915/intel_device_info.h
>> index 080a4557899b..f032f2500f50 100644
>> --- a/drivers/gpu/drm/i915/intel_device_info.h
>> +++ b/drivers/gpu/drm/i915/intel_device_info.h
>> @@ -317,6 +317,8 @@ struct intel_driver_caps {
>>   
>>   const char *intel_platform_name(enum intel_platform platform);
>>   
>> +void intel_device_info_driver_create(struct drm_i915_private *i915, u16 device_id,
>> +				     const struct intel_device_info *match_info);
>
> Match_info maybe reads a bit odd when "exported" but I have no superior 
> suggestions either (const? template? pci?). Hiding seems a good plan so:
>
> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>

Naming is hard, pushed as-is, thanks for the review.

BR,
Jani.

>
> Regards,
>
> Tvrtko
>
>>   void intel_device_info_runtime_init_early(struct drm_i915_private *dev_priv);
>>   void intel_device_info_runtime_init(struct drm_i915_private *dev_priv);
>>   

-- 
Jani Nikula, Intel Open Source Graphics Center