On Fri, 09 Feb 2018, Chris Wilson <ch...@chris-wilson.co.uk> wrote:
> Quoting Chris Wilson (2018-02-09 11:29:08)
>> Move the storage of the runtime discovery of EDRAM to intel_driver_caps
>> so that we include it in all the relevant error state dumping and
>> debugging.
>> Signed-off-by: Chris Wilson <ch...@chris-wilson.co.uk>
>> ---
>> Or do we want to keep driver_caps only for driver related info, and move
>> this sort of runtime discovery to intel_runtime_info? With the intention
>> of killing off all remaining mkwrite_intel_info()
> I think I prefer a new intel_runtime_info, then we have
> static intel_device_info,
> write-once (hopefully!) intel_runtime_info,
> mutable intel_driver_caps

Yeah, I'd like to see mkwrite_intel_info() go.

I suppose struct drm_i915_private would then have a const pointer field
to device info? That's one painful change. I think better to go with
changing everything to use INTEL_INFO() (or intel_info()?) first. Auch.


Jani Nikula, Intel Open Source Technology Center
Intel-gfx mailing list

Reply via email to