[Intel-gfx] [PATCH 3/8] drm/i915/execlists: Suppress redundant preemption

[Chris Wilson]

Quoting Tvrtko Ursulin (2019-01-29 12:26:46)
> 
> On 29/01/2019 11:21, Chris Wilson wrote:
> > Quoting Tvrtko Ursulin (2019-01-29 11:15:35)
> >>
> >> On 29/01/2019 08:58, Chris Wilson wrote:
> >>> On unwinding the active request we give it a small (limited to internal
> >>> priority levels) boost to prevent it from being gazumped a second time.
> >>> However, this means that it can be promoted to above the request that
> >>> triggered the preemption request, causing a preempt-to-idle cycle for no
> >>
> >> How can it go higher? Only because I915_PRIORITY_NEWCLIENT is higher
> >> than I915_PRIORITY_WAIT? In that case would re-ordering those help? I
> >> don't remember if there was a reason for the current order.
> > 
> > Yes. There's more reasons later (media-bench is quite insistent about
> > this ;).
> > 
> > You were quite adamant in insisting that WAIT be the most minor of
> > boosts :)
> 
> I was afraid that might have been the case. :) So we are firmly into 
> tweaking the scheduler for particulars clients territory.. and that 
> before we even have time-slicing!

What's coming up is a significantly different kettle of fish, I'm sure
you will agree when you start thinking about submitting dependent jobs
in parallel.
 
> In my mind it probably still makes sense that WAIT is lowest bump.

I agree. I like having WAIT, but I concur that it shouldn't be trivially
abusable.
 
> How about a we add a new smallest internal dedicated level for preempted 
> contexts? Because I don't like how this patch would intermingle with 
> either WAIT or NEWCLIENT.
> 
> #define I915_PRIORITY_PREEMPTED ((u8)BIT(0)) /* Must be lowest */
> #define I915_PRIORITY_WAIT      ((u8)BIT(1))
> #define I915_PRIORITY_NEWCLIENT ((u8)BIT(2))
> 
> ?

No, the preempted client already goes to the front of their respective
bucket which is the same as the end of the one above.

The trick that comes later is what happens if we want to add a level
above NEWCLIENT that we wish to promote on preemption? Which is the
consideration for adding semaphores.
-Chris