[Intel-gfx] Polymorphic to_i915()

Dave Gordon david.s.gordon at intel.com
Wed Apr 20 14:29:30 UTC 2016


On 20/04/16 13:57, Daniel Vetter wrote:
> On Mon, Apr 18, 2016 at 12:18:23PM +0300, Jani Nikula wrote:
>> On Fri, 15 Apr 2016, Chris Wilson <chris at chris-wilson.co.uk> wrote:
>>> Final canvas for opinions for using a magic macro to reduce typing in
>>> the common operation of getting our drm_i915_private from the object.
>>>
>>> 	21 files changed, 333 insertions(+), 392 deletions(-)
>>>
>>> Not to mention the ease it makes for later patches to reduce the pointer
>>> dance.
>>
>> I've expressed my reservations about this the last time.
>>
>> My compromise proposal is this: let's add the to_i915()
>> "superconvenience macro", but let's not embed that into other
>> macros. Instead, move away from convenience macros in them, explicitly
>> requiring dev_priv.
>>
>> This would make just one macro special, and would keep the rest less
>> surprising and "C-like". We already need dev_priv all over the place, so
>> I don't think having a local variable or an explicit to_i915() is a big
>> burden.
>
> Not much more to add, but I'm not strongly opinionated here really. But I
> do think that a trick of this magnitude needs much more enthusiastic
> support from a bunch of people before we can merge it.
> -Daniel

How about taking the just smallest of these patches (items 1/6 and 4/6) 
now, dropping 2 entirely, and leaving the more intrusive 3, 5, and 6 for 
later, maybe at a time when there's not too much code churn?

.Dave.



More information about the Intel-gfx mailing list