@@ -344,6 +344,9 @@ static int __intel_context_active(struct i915_active *active)
{
struct intel_context *ce = container_of(active, typeof(*ce), active);
+ CE_TRACE(ce, "active\n");
+ ce->saturated = 0;
+
intel_context_get(ce);
/* everything should already be activated by intel_context_pre_pin() */
@@ -109,6 +109,8 @@ struct intel_context {
} lrc;
u32 tag; /* cookie passed to HW to track this context on submission */
+ intel_engine_mask_t saturated; /* submitting semaphores too late? */
+
/** stats: Context GPU engine busyness tracking. */
struct intel_context_stats {
u64 active;
@@ -253,8 +253,6 @@ static int __engine_park(struct intel_wakeref *wf)
struct intel_engine_cs *engine =
container_of(wf, typeof(*engine), wakeref);
- engine->saturated = 0;
-
/*
* If one and only one request is completed between pm events,
* we know that we are inside the kernel context and it is
@@ -303,8 +303,6 @@ struct intel_engine_cs {
struct intel_context *kernel_context; /* pinned */
- intel_engine_mask_t saturated; /* submitting semaphores too late? */
-
struct {
struct delayed_work work;
struct i915_request *systole;
@@ -3413,21 +3413,6 @@ intel_execlists_create_virtual(struct intel_engine_cs **siblings,
ve->base.instance = I915_ENGINE_CLASS_INVALID_VIRTUAL;
ve->base.uabi_instance = I915_ENGINE_CLASS_INVALID_VIRTUAL;
- /*
- * The decision on whether to submit a request using semaphores
- * depends on the saturated state of the engine. We only compute
- * this during HW submission of the request, and we need for this
- * state to be globally applied to all requests being submitted
- * to this engine. Virtual engines encompass more than one physical
- * engine and so we cannot accurately tell in advance if one of those
- * engines is already saturated and so cannot afford to use a semaphore
- * and be pessimized in priority for doing so -- if we are the only
- * context using semaphores after all other clients have stopped, we
- * will be starved on the saturated system. Such a global switch for
- * semaphores is less than ideal, but alas is the current compromise.
- */
- ve->base.saturated = ALL_ENGINES;
-
snprintf(ve->base.name, sizeof(ve->base.name), "virtual");
intel_engine_init_execlists(&ve->base);
@@ -516,7 +516,7 @@ bool __i915_request_submit(struct i915_request *request)
*/
if (request->sched.semaphores &&
i915_sw_fence_signaled(&request->semaphore))
- engine->saturated |= request->sched.semaphores;
+ request->context->saturated |= request->sched.semaphores;
engine->emit_fini_breadcrumb(request,
request->ring->vaddr + request->postfix);
@@ -977,7 +977,7 @@ already_busywaiting(struct i915_request *rq)
*
* See the are-we-too-late? check in __i915_request_submit().
*/
- return rq->sched.semaphores | READ_ONCE(rq->engine->saturated);
+ return rq->sched.semaphores | READ_ONCE(rq->context->saturated);
}
static int
@@ -1071,7 +1071,7 @@ emit_semaphore_wait(struct i915_request *to,
if (__emit_semaphore_wait(to, from, from->fence.seqno))
goto await_fence;
- to->sched.semaphores |= mask;
+ to->sched.semaphores |= mask & ~to->engine->mask;
wait = &to->semaphore;
await_fence:
When we introduced the saturated workload detection to tell us to back off from semaphore usage [semaphores have a noticeable impact on contended bus cycles with the CPU for some heavy workloads], we first introduced it as a per-context tracker. This allows individual contexts to try and optimise their own usage, but we found that with the local tracking and the no-semaphore boosting, the first context to disable semaphores got a massive priority boost and so would starve the rest and all new contexts (as they started with semaphores enabled and lower priority). Hence we moved the saturated workload detection to the engine, and a consequence had to disable semaphores on virtual engines. Now that we do not have semaphore priority boosting, and try to fairly schedule irrespective of semaphore usage, we can move the tracking back to the context and virtual engines can now utilise the faster inter-engine synchronisation. If we see that any context fairs to use the semaphore, because the system is oversubscribed and was busy doing something else instead of spinning on the semaphore, we disable further usage of semaphores with that context until it idles again. This should restrict the semaphores to lightly utilised system where the latency between requests is more noticeable, and curtail the bus-contention from checking for signaled semaphores. References: 44d89409a12e ("drm/i915: Make the semaphore saturation mask global") Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> --- drivers/gpu/drm/i915/gt/intel_context.c | 3 +++ drivers/gpu/drm/i915/gt/intel_context_types.h | 2 ++ drivers/gpu/drm/i915/gt/intel_engine_pm.c | 2 -- drivers/gpu/drm/i915/gt/intel_engine_types.h | 2 -- .../gpu/drm/i915/gt/intel_execlists_submission.c | 15 --------------- drivers/gpu/drm/i915/i915_request.c | 6 +++--- 6 files changed, 8 insertions(+), 22 deletions(-)