Re: [PATCH 5/5] drm/i915: Flatten engine init control flow

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Thu, Nov 20, 2014 at 08:10:27AM +0000, Chris Wilson wrote:
> On Thu, Nov 20, 2014 at 12:33:08AM +0100, Daniel Vetter wrote:
> > Now that sanity prevails and we have the clean split between software
> > init and starting the engines we can drop all the "have we allocate
> > this struct already?" nonsense.
> 
> Actually, I was hoping for something completely different when you sad
> flattening the init...

I presume you've hoped for more ;-) Essentially this is just the legacy
ringbuffer blueprint, applying to execlist left as an exercise for the
reader. And it's just a very small patch.

> Compare with
> http://cgit.freedesktop.org/~ickle/linux-2.6/tree/drivers/gpu/drm/i915/intel_ringbuffer.c#n1983
> 
> There is a lot of duplicated code in the ringbuffer setup, and the code
> movement here wants to be split differently so that the sequence is
> identical to execlists. (The only difference is that we only use one
> ring with legacy, rather than one ring in each context [per-engine].)
> Making sure the logic is foolproof for both is esssential.

Imo we first need to beat execlist into shape (with small patches, not by
rewriting it all). Once stuff settles we can take another look at whether
unification makes sense, but for now I want a big piece of insulation
between execlist and stuff we actually depend on. Or at least stuff Dave
might rip my head off over.

So for now I still want to embrace the duplication.
-Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
http://lists.freedesktop.org/mailman/listinfo/intel-gfx





[Index of Archives]     [Linux USB Devel]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]
  Powered by Linux