@@ -1053,7 +1053,8 @@ void i915_request_add(struct i915_request *request)
local_bh_disable();
rcu_read_lock(); /* RCU serialisation for set-wedged protection */
if (engine->schedule) {
- struct i915_sched_attr attr = request->gem_context->sched;
+ struct i915_gem_context *ctx = request->gem_context;
+ struct i915_sched_attr attr = ctx->sched;
/*
* Boost priorities to new clients (new request flows).
@@ -1064,6 +1065,9 @@ void i915_request_add(struct i915_request *request)
if (!prev || i915_request_completed(prev))
attr.priority |= I915_PRIORITY_NEWCLIENT;
+ if (intel_engine_queue_is_empty(engine))
+ attr.priority |= I915_PRIORITY_STALL;
+
engine->schedule(request, &attr);
}
rcu_read_unlock();
@@ -19,13 +19,14 @@ enum {
I915_PRIORITY_INVALID = INT_MIN
};
-#define I915_USER_PRIORITY_SHIFT 1
+#define I915_USER_PRIORITY_SHIFT 2
#define I915_USER_PRIORITY(x) ((x) << I915_USER_PRIORITY_SHIFT)
#define I915_PRIORITY_COUNT BIT(I915_USER_PRIORITY_SHIFT)
#define I915_PRIORITY_MASK (-I915_PRIORITY_COUNT)
-#define I915_PRIORITY_NEWCLIENT BIT(0)
+#define I915_PRIORITY_NEWCLIENT BIT(1)
+#define I915_PRIORITY_STALL BIT(0)
struct i915_sched_attr {
/**
@@ -1069,6 +1069,12 @@ gen8_emit_ggtt_write(u32 *cs, u32 value, u32 gtt_offset)
return cs;
}
+static inline bool
+intel_engine_queue_is_empty(const struct intel_engine_cs *engine)
+{
+ return RB_EMPTY_ROOT(&engine->execlists.queue.rb_root);
+}
+
bool intel_engine_is_idle(struct intel_engine_cs *engine);
bool intel_engines_are_idle(struct drm_i915_private *dev_priv);
In order to maximise concurrency between engines, if we queue a request to a current idle ring, reorder its dependencies to execute that request as early as possible and ideally improve occupancy of multiple engines simultaneously. Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> --- drivers/gpu/drm/i915/i915_request.c | 6 +++++- drivers/gpu/drm/i915/i915_scheduler.h | 5 +++-- drivers/gpu/drm/i915/intel_ringbuffer.h | 6 ++++++ 3 files changed, 14 insertions(+), 3 deletions(-)