Re: [Intel-gfx] [PATCH 05/10] drm/i915: Track runtime spent in unreachable intel_contexts

From: Chris Wilson <chris@chris-wilson.co.uk>
To: Intel-gfx@lists.freedesktop.org,
	Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Subject: Re: [Intel-gfx] [PATCH 05/10] drm/i915: Track runtime spent in unreachable intel_contexts
Date: Wed, 18 Mar 2020 14:32:52 +0000	[thread overview]
Message-ID: <158454197252.6873.7587021909076621666@build.alporthouse.com> (raw)
In-Reply-To: <3565cd32-5dd8-0d45-ee55-fb4fb51aa7bd@linux.intel.com>

Quoting Tvrtko Ursulin (2020-03-18 14:13:14)
> 
> On 18/03/2020 13:55, Chris Wilson wrote:
> > Quoting Tvrtko Ursulin (2020-03-18 12:11:34)
> >> From: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
> >>
> >> As contexts are abandoned we want to remember how much GPU time they used
> >> (per class) so later we can used it for smarter purposes.
> >>
> >> Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
> >> ---
> >>   drivers/gpu/drm/i915/gem/i915_gem_context.c       | 13 ++++++++++++-
> >>   drivers/gpu/drm/i915/gem/i915_gem_context_types.h |  5 +++++
> >>   2 files changed, 17 insertions(+), 1 deletion(-)
> >>
> >> diff --git a/drivers/gpu/drm/i915/gem/i915_gem_context.c b/drivers/gpu/drm/i915/gem/i915_gem_context.c
> >> index 7c119a3a2cbd..5edf79ed6247 100644
> >> --- a/drivers/gpu/drm/i915/gem/i915_gem_context.c
> >> +++ b/drivers/gpu/drm/i915/gem/i915_gem_context.c
> >> @@ -257,7 +257,19 @@ static void free_engines_rcu(struct rcu_head *rcu)
> >>   {
> >>          struct i915_gem_engines *engines =
> >>                  container_of(rcu, struct i915_gem_engines, rcu);
> >> +       struct i915_gem_context *ctx = engines->ctx;
> >> +       struct i915_gem_engines_iter it;
> >> +       struct intel_context *ce;
> >> +
> >> +       /* Transfer accumulated runtime to the parent GEM context. */
> >> +       for_each_gem_engine(ce, engines, it) {
> >> +               unsigned int class = ce->engine->uabi_class;
> >>   
> >> +               GEM_BUG_ON(class >= ARRAY_SIZE(ctx->past_runtime));
> >> +               atomic64_add(ce->runtime.total, &ctx->past_runtime[class]);
> > 
> > Hmm, there's an odd situation where the free_engines_rcu could fire
> > before we do the final schedule_out of the context.
> > 
> > GEM_BUG_ON(intel_context_inflight(ce)) to see if that's being too
> > paranoid.
> 
> Deja vu.. have I forgotten to move this into intel_context_free while 
> the purpose of keeping ce->gem_context valid was exactly to enable that.

I would rather not have it in gt/ if possible. The delay should be
nominal at worst.
-Chris
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx