[Intel-gfx] [PATCH 5/6] drm/i915: Split i915_gem_timeline into individual timelines

[Chris Wilson]

Quoting Tvrtko Ursulin (2018-04-23 13:33:04)
> 
> On 23/04/2018 11:13, Chris Wilson wrote:
> > We need to move to a more flexible timeline that doesn't assume one
> > fence context per engine, and so allow for a single timeline to be used
> > across a combination of engines. This means that preallocating a fence
> > context per engine is now a hindrance, and so we want to introduce the
> > singular timeline. From the code perspective, this has the notable
> > advantage of clearing up a lot of mirky semantics and some clumsy
> > pointer chasing.
> > 
> > By splitting the timeline up into a single entity rather than an array
> > of per-engine timelines, we can realise the goal of the previous patch
> > of tracking the timeline alongside the ring.
> 
> Isn't single fence context and a single seqno space breaking the ABI? 
> Submissions from a context are now serialized across all engines. I am 
> thinking about await and dependency created in __i915_add_request to 
> timeline->last_request.

It's still one per engine, the ABI shouldn't have changed unless I've
screwed up. I now I wrote some tests to assert the independence... But
probably only have those for the new ABI we were discussing.

I guess I should make sure gem_exec_schedule is covering it.
-Chris