[Intel-gfx] [PATCH v2 06/11] drm/i915/scheduler: Record all dependencies upon request construction

[Chris Wilson]

On Thu, Nov 10, 2016 at 11:54:51AM +0000, Tvrtko Ursulin wrote:
> 
> On 10/11/2016 10:55, Chris Wilson wrote:
> >On Thu, Nov 10, 2016 at 10:44:44AM +0000, Tvrtko Ursulin wrote:
> >>
> >>On 08/11/2016 12:20, Chris Wilson wrote:
> >>>On Mon, Nov 07, 2016 at 01:59:45PM +0000, Chris Wilson wrote:
> >>>>The scheduler needs to know the dependencies of each request for the
> >>>>lifetime of the request, as it may choose to reschedule the requests at
> >>>>any time and must ensure the dependency tree is not broken. This is in
> >>>>additional to using the fence to only allow execution after all
> >>>>dependencies have been completed.
> >>>>
> >>>>One option was to extend the fence to support the bidirectional
> >>>>dependency tracking required by the scheduler. However the mismatch in
> >>>>lifetimes between the submit fence and the request essentially meant
> >>>>that we had to build a completely separate struct (and we could not
> >>>>simply reuse the existing waitqueue in the fence for one half of the
> >>>>dependency tracking). The extra dependency tracking simply did not mesh
> >>>>well with the fence, and keeping it separate both keeps the fence
> >>>>implementation simpler and allows us to extend the dependency tracking
> >>>>into a priority tree (whilst maintaining support for reordering the
> >>>>tree).
> >>>>
> >>>>To avoid the additional allocations and list manipulations, the use of
> >>>>the priotree is disabled when there are no schedulers to use it.
> >>>>
> >>>>v2: Create a dedicated slab for i915_dependency.
> >>>>   Rename the lists.
> >>>
> >>>Sod. I've squashed the priority sort into this as a rebase calamity.
> >>
> >>Waiting for the other patches to get looked at before
> >>untangle/respin or just forgot? :)
> >
> >Just imagine the split ;) Think of the intel_lrc.c as a separate patch.
> 
> It is a little bit more than that, but more importantly, is your
> plan now to just keep it squashed? I can pre-review it squashed, but
> I think it would be very desirable to eventually split it up again.

I've unsplit it, but I didn't resend the series again because I didn't
think it was that important, resending just this pair would mess up CI.
The changed code is just the DFS which is where I would appreciate
comments.
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre